2004-02-28 15:38:04 +00:00
|
|
|
/* Native-dependent code for the i386.
|
|
|
|
|
2014-01-01 03:54:24 +00:00
|
|
|
Copyright (C) 2001-2014 Free Software Foundation, Inc.
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
|
|
|
|
This file is part of GDB.
|
|
|
|
|
|
|
|
This program is free software; you can redistribute it and/or modify
|
|
|
|
it under the terms of the GNU General Public License as published by
|
2007-08-23 18:08:50 +00:00
|
|
|
the Free Software Foundation; either version 3 of the License, or
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
(at your option) any later version.
|
|
|
|
|
|
|
|
This program is distributed in the hope that it will be useful,
|
|
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
GNU General Public License for more details.
|
|
|
|
|
|
|
|
You should have received a copy of the GNU General Public License
|
2007-08-23 18:08:50 +00:00
|
|
|
along with this program. If not, see <http://www.gnu.org/licenses/>. */
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
|
|
|
|
#include "defs.h"
|
2012-11-07 20:10:13 +00:00
|
|
|
#include "i386-nat.h"
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
#include "gdbcmd.h"
|
2012-01-24 13:46:55 +00:00
|
|
|
#include "inferior.h"
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
|
2004-02-28 15:38:04 +00:00
|
|
|
/* Support for hardware watchpoints and breakpoints using the i386
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
debug registers.
|
|
|
|
|
|
|
|
This provides several functions for inserting and removing
|
2004-02-28 15:38:04 +00:00
|
|
|
hardware-assisted breakpoints and watchpoints, testing if one or
|
|
|
|
more of the watchpoints triggered and at what address, checking
|
|
|
|
whether a given region can be watched, etc.
|
|
|
|
|
|
|
|
The functions below implement debug registers sharing by reference
|
|
|
|
counts, and allow to watch regions up to 16 bytes long. */
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
|
2014-06-11 14:30:50 +00:00
|
|
|
/* Whether or not to print the mirrored debug registers. */
|
2014-06-19 09:53:12 +00:00
|
|
|
int debug_hw_points;
|
2014-06-17 10:32:58 +00:00
|
|
|
|
2014-06-17 10:32:26 +00:00
|
|
|
/* Low-level function vector. */
|
2009-05-14 09:37:00 +00:00
|
|
|
struct i386_dr_low_type i386_dr_low;
|
|
|
|
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
/* Per-process data. We don't bind this to a per-inferior registry
|
|
|
|
because of targets like x86 GNU/Linux that need to keep track of
|
|
|
|
processes that aren't bound to any inferior (e.g., fork children,
|
|
|
|
checkpoints). */
|
2011-07-22 16:58:32 +00:00
|
|
|
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
struct i386_process_info
|
2011-07-22 16:58:32 +00:00
|
|
|
{
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
/* Linked list. */
|
|
|
|
struct i386_process_info *next;
|
2011-07-22 16:58:32 +00:00
|
|
|
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
/* The process identifier. */
|
|
|
|
pid_t pid;
|
2012-01-24 13:46:55 +00:00
|
|
|
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
/* Copy of i386 hardware debug registers. */
|
2012-01-24 13:46:55 +00:00
|
|
|
struct i386_debug_reg_state state;
|
|
|
|
};
|
|
|
|
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
static struct i386_process_info *i386_process_list = NULL;
|
2012-01-24 13:49:58 +00:00
|
|
|
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
/* Find process data for process PID. */
|
|
|
|
|
|
|
|
static struct i386_process_info *
|
|
|
|
i386_find_process_pid (pid_t pid)
|
2012-01-24 13:49:58 +00:00
|
|
|
{
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
struct i386_process_info *proc;
|
|
|
|
|
|
|
|
for (proc = i386_process_list; proc; proc = proc->next)
|
|
|
|
if (proc->pid == pid)
|
|
|
|
return proc;
|
2012-01-24 13:49:58 +00:00
|
|
|
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
return NULL;
|
2012-01-24 13:49:58 +00:00
|
|
|
}
|
|
|
|
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
/* Add process data for process PID. Returns newly allocated info
|
|
|
|
object. */
|
2012-01-24 13:46:55 +00:00
|
|
|
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
static struct i386_process_info *
|
|
|
|
i386_add_process (pid_t pid)
|
2012-01-24 13:46:55 +00:00
|
|
|
{
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
struct i386_process_info *proc;
|
2012-01-24 13:49:58 +00:00
|
|
|
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
proc = xcalloc (1, sizeof (*proc));
|
|
|
|
proc->pid = pid;
|
2012-01-24 13:46:55 +00:00
|
|
|
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
proc->next = i386_process_list;
|
|
|
|
i386_process_list = proc;
|
2012-01-24 13:46:55 +00:00
|
|
|
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
return proc;
|
|
|
|
}
|
2012-01-24 13:46:55 +00:00
|
|
|
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
/* Get data specific info for process PID, creating it if necessary.
|
|
|
|
Never returns NULL. */
|
2012-01-24 13:46:55 +00:00
|
|
|
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
static struct i386_process_info *
|
|
|
|
i386_process_info_get (pid_t pid)
|
|
|
|
{
|
|
|
|
struct i386_process_info *proc;
|
|
|
|
|
|
|
|
proc = i386_find_process_pid (pid);
|
|
|
|
if (proc == NULL)
|
|
|
|
proc = i386_add_process (pid);
|
2012-01-24 13:46:55 +00:00
|
|
|
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
return proc;
|
2012-01-24 13:46:55 +00:00
|
|
|
}
|
|
|
|
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
/* Get debug registers state for process PID. */
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
|
gdb/
2011-12-14 Pedro Alves <pedro@codesourcery.com>
PR threads/10729
* linux-nat.c (linux_nat_new_thread): Change parameter to an lwp
pointer.
(linux_nat_prepare_to_resume): New global.
(lwp_free): New.
(purge_lwp_list): Use it.
(add_lwp): Call linux_nat_new_thread even on the first LWP.
Adjust to interface change.
(delete_lwp): Call lwp_free instead of xfree.
(detach_callback, linux_nat_detach, resume_lwp, linux_nat_resume)
(linux_handle_syscall_trap, linux_handle_extended_wait)
(linux_nat_filter_event, resume_stopped_resumed_lwps): Call
linux_nat_prepare_to_resume before resuming.
(linux_stop_lwp): New.
(linux_nat_set_new_thread): Adjust.
(linux_nat_set_prepare_to_resume): New.
* linux-nat.h (struct arch_lwp_info): Forward declare.
(struct lwp_info) <arch_private>: New field.
(linux_stop_lwp): Declare.
(linux_nat_set_new_thread): Adjust.
(linux_nat_set_prepare_to_resume): New.
* i386-nat.c (DR_NADDR, DR_STATUS, DR_CONTROL)
(struct i386_debug_reg_state): Move to i386-nat.h.
(dr_mirror): Comment.
(i386_debug_reg_state): New.
(i386_update_inferior_debug_regs): Simplify.
(i386_stopped_data_address): Use the debug register state from the
inferior, not from the local cache.
* i386-nat.h (struct i386_dr_low_type): Delete reset_addr and
unset_status fields. New get_addr and get_control fields.
(DR_FIRSTADDR, DR_LASTADDR, DR_CONTROL): Moved from i386-nat.c.
(DR_NADDR, DR_STATUS): New.
(struct i386_debug_reg_state): Moved from i386-nat.c.
* amd64-linux-nat.c (struct arch_lwp_info): New.
(amd64_linux_dr): Delete global.
(amd64_linux_dr_get_addr): New.
(amd64_linux_dr_get_control): New.
(amd64_linux_dr_unset_status): Delete.
(amd64_linux_dr_set_addr): Reimplement.
(amd64_linux_dr_reset_addr): Delete.
(update_debug_registers_callback): New.
(amd64_linux_dr_set_control): Reimplement.
(amd64_linux_dr_set_addr): Reimplement.
(amd64_linux_prepare_to_resume): New.
(amd64_linux_new_thread): Change parameter to an lwp pointer.
Reimplement.
(_initialize_amd64_linux_nat): No longer install
i386_dr_low.reset_addr and i386_dr_low.unset_status. Install
amd64_linux_dr_get_control as i386_dr_low.get_control. Install
amd64_linux_dr_get_addr as i386_dr_low.get_addr. Install
amd64_linux_prepare_to_resume.
* i386-linux-nat.c (DR_FIRSTADDR, DR_LASTADDR, DR_STATUS)
(DR_CONTROL): Delete.
(struct arch_lwp_info): New.
(i386_linux_dr): Delete global.
(i386_linux_dr_set_control): Reimplement.
(i386_linux_dr_get_addr): New.
(i386_linux_dr_set_addr): Reimplement.
(i386_linux_dr_get_control): New.
(update_debug_registers_callback): New.
(i386_linux_dr_unset_status): Delete.
(i386_linux_dr_set_addr): Reimplement.
(i386_linux_prepare_to_resume): New.
(i386_linux_new_thread): Change parameter to an lwp pointer.
Reimplement.
(_initialize_i386_linux_nat): No longer install
i386_dr_low.reset_addr and i386_dr_low.unset_status. Install
i386_linux_dr_get_control as i386_dr_low.get_control. Install
i386_linux_dr_get_addr as i386_dr_low.get_addr. Install
i386_linux_prepare_to_resume.
* arm-linux-nat.c (arm_linux_new_thread): Change parameter to an
lwp pointer. Adjust.
* ia64-linux-nat.c (ia64_linux_new_thread): Likewise.
* mips-linux-nat.c (mips_linux_new_thread): Likewise.
* ppc-linux-nat.c (ppc_linux_new_thread): Likewise.
* s390-nat.c (s390_fix_watch_points): Likewise.
* i386-darwin-nat.c (DR_FIRSTADDR, DR_LASTADDR, DR_STATUS)
(DR_CONTROL): Delete.
(i386_darwin_dr_reset_addr): Delete.
(i386_darwin_dr_get_addr): New.
(i386_darwin_dr_get_control): New.
* go32-nat.c
(go32_get_dr7, go32_get_dr): New.
(init_go32_ops): No longer install i386_dr_low.reset_addr.
Install go32_get_dr7 as i386_dr_low.get_control. Install
go32_get_dr as i386_dr_low.get_addr.
* i386bsd-nat.c (i386bsd_dr_get): New.
(i386bsd_dr_reset_addr): Delete.
(i386bsd_dr_get_addr): New.
(i386bsd_dr_get_status): Use i386bsd_dr_get.
(i386bsd_dr_get_control): New.
* i386bsd-nat.h (i386bsd_dr_reset_addr): Delete.
(i386bsd_dr_get_addr): New.
(i386bsd_dr_get_control): New.
* i386fbsd-nat.c (_initialize_i386fbsd_nat): No longer install
i386_dr_low.reset_addr and i386_dr_low.unset_status. Install
i386bsd_dr_get_control as i386_dr_low.get_control. Install
i386bsd_dr_get_addr as i386_dr_low.get_addr.
* windows-nat.c (init_windows_ops): No longer install
i386_dr_low.reset_addr and i386_dr_low.unset_status. Install
cygwin_get_dr7 as i386_dr_low.get_control. Install cygwin_get_dr
as i386_dr_low.get_addr.
(cygwin_get_dr): New.
(cygwin_get_dr7): New.
gdb/testsuite/
2011-12-14 Pedro Alves <pedro@codesourcery.com>
PR threads/10729
* gdb.mi/watch-nonstop.c: New file.
* gdb.mi/mi-watch-nonstop.exp: New file.
2011-12-14 17:20:32 +00:00
|
|
|
struct i386_debug_reg_state *
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
i386_debug_reg_state (pid_t pid)
|
gdb/
2011-12-14 Pedro Alves <pedro@codesourcery.com>
PR threads/10729
* linux-nat.c (linux_nat_new_thread): Change parameter to an lwp
pointer.
(linux_nat_prepare_to_resume): New global.
(lwp_free): New.
(purge_lwp_list): Use it.
(add_lwp): Call linux_nat_new_thread even on the first LWP.
Adjust to interface change.
(delete_lwp): Call lwp_free instead of xfree.
(detach_callback, linux_nat_detach, resume_lwp, linux_nat_resume)
(linux_handle_syscall_trap, linux_handle_extended_wait)
(linux_nat_filter_event, resume_stopped_resumed_lwps): Call
linux_nat_prepare_to_resume before resuming.
(linux_stop_lwp): New.
(linux_nat_set_new_thread): Adjust.
(linux_nat_set_prepare_to_resume): New.
* linux-nat.h (struct arch_lwp_info): Forward declare.
(struct lwp_info) <arch_private>: New field.
(linux_stop_lwp): Declare.
(linux_nat_set_new_thread): Adjust.
(linux_nat_set_prepare_to_resume): New.
* i386-nat.c (DR_NADDR, DR_STATUS, DR_CONTROL)
(struct i386_debug_reg_state): Move to i386-nat.h.
(dr_mirror): Comment.
(i386_debug_reg_state): New.
(i386_update_inferior_debug_regs): Simplify.
(i386_stopped_data_address): Use the debug register state from the
inferior, not from the local cache.
* i386-nat.h (struct i386_dr_low_type): Delete reset_addr and
unset_status fields. New get_addr and get_control fields.
(DR_FIRSTADDR, DR_LASTADDR, DR_CONTROL): Moved from i386-nat.c.
(DR_NADDR, DR_STATUS): New.
(struct i386_debug_reg_state): Moved from i386-nat.c.
* amd64-linux-nat.c (struct arch_lwp_info): New.
(amd64_linux_dr): Delete global.
(amd64_linux_dr_get_addr): New.
(amd64_linux_dr_get_control): New.
(amd64_linux_dr_unset_status): Delete.
(amd64_linux_dr_set_addr): Reimplement.
(amd64_linux_dr_reset_addr): Delete.
(update_debug_registers_callback): New.
(amd64_linux_dr_set_control): Reimplement.
(amd64_linux_dr_set_addr): Reimplement.
(amd64_linux_prepare_to_resume): New.
(amd64_linux_new_thread): Change parameter to an lwp pointer.
Reimplement.
(_initialize_amd64_linux_nat): No longer install
i386_dr_low.reset_addr and i386_dr_low.unset_status. Install
amd64_linux_dr_get_control as i386_dr_low.get_control. Install
amd64_linux_dr_get_addr as i386_dr_low.get_addr. Install
amd64_linux_prepare_to_resume.
* i386-linux-nat.c (DR_FIRSTADDR, DR_LASTADDR, DR_STATUS)
(DR_CONTROL): Delete.
(struct arch_lwp_info): New.
(i386_linux_dr): Delete global.
(i386_linux_dr_set_control): Reimplement.
(i386_linux_dr_get_addr): New.
(i386_linux_dr_set_addr): Reimplement.
(i386_linux_dr_get_control): New.
(update_debug_registers_callback): New.
(i386_linux_dr_unset_status): Delete.
(i386_linux_dr_set_addr): Reimplement.
(i386_linux_prepare_to_resume): New.
(i386_linux_new_thread): Change parameter to an lwp pointer.
Reimplement.
(_initialize_i386_linux_nat): No longer install
i386_dr_low.reset_addr and i386_dr_low.unset_status. Install
i386_linux_dr_get_control as i386_dr_low.get_control. Install
i386_linux_dr_get_addr as i386_dr_low.get_addr. Install
i386_linux_prepare_to_resume.
* arm-linux-nat.c (arm_linux_new_thread): Change parameter to an
lwp pointer. Adjust.
* ia64-linux-nat.c (ia64_linux_new_thread): Likewise.
* mips-linux-nat.c (mips_linux_new_thread): Likewise.
* ppc-linux-nat.c (ppc_linux_new_thread): Likewise.
* s390-nat.c (s390_fix_watch_points): Likewise.
* i386-darwin-nat.c (DR_FIRSTADDR, DR_LASTADDR, DR_STATUS)
(DR_CONTROL): Delete.
(i386_darwin_dr_reset_addr): Delete.
(i386_darwin_dr_get_addr): New.
(i386_darwin_dr_get_control): New.
* go32-nat.c
(go32_get_dr7, go32_get_dr): New.
(init_go32_ops): No longer install i386_dr_low.reset_addr.
Install go32_get_dr7 as i386_dr_low.get_control. Install
go32_get_dr as i386_dr_low.get_addr.
* i386bsd-nat.c (i386bsd_dr_get): New.
(i386bsd_dr_reset_addr): Delete.
(i386bsd_dr_get_addr): New.
(i386bsd_dr_get_status): Use i386bsd_dr_get.
(i386bsd_dr_get_control): New.
* i386bsd-nat.h (i386bsd_dr_reset_addr): Delete.
(i386bsd_dr_get_addr): New.
(i386bsd_dr_get_control): New.
* i386fbsd-nat.c (_initialize_i386fbsd_nat): No longer install
i386_dr_low.reset_addr and i386_dr_low.unset_status. Install
i386bsd_dr_get_control as i386_dr_low.get_control. Install
i386bsd_dr_get_addr as i386_dr_low.get_addr.
* windows-nat.c (init_windows_ops): No longer install
i386_dr_low.reset_addr and i386_dr_low.unset_status. Install
cygwin_get_dr7 as i386_dr_low.get_control. Install cygwin_get_dr
as i386_dr_low.get_addr.
(cygwin_get_dr): New.
(cygwin_get_dr7): New.
gdb/testsuite/
2011-12-14 Pedro Alves <pedro@codesourcery.com>
PR threads/10729
* gdb.mi/watch-nonstop.c: New file.
* gdb.mi/mi-watch-nonstop.exp: New file.
2011-12-14 17:20:32 +00:00
|
|
|
{
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
return &i386_process_info_get (pid)->state;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* See declaration in i386-nat.h. */
|
|
|
|
|
|
|
|
void
|
|
|
|
i386_forget_process (pid_t pid)
|
|
|
|
{
|
|
|
|
struct i386_process_info *proc, **proc_link;
|
|
|
|
|
|
|
|
proc = i386_process_list;
|
|
|
|
proc_link = &i386_process_list;
|
|
|
|
|
|
|
|
while (proc != NULL)
|
|
|
|
{
|
|
|
|
if (proc->pid == pid)
|
|
|
|
{
|
|
|
|
*proc_link = proc->next;
|
|
|
|
|
|
|
|
xfree (proc);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
proc_link = &proc->next;
|
|
|
|
proc = *proc_link;
|
|
|
|
}
|
gdb/
2011-12-14 Pedro Alves <pedro@codesourcery.com>
PR threads/10729
* linux-nat.c (linux_nat_new_thread): Change parameter to an lwp
pointer.
(linux_nat_prepare_to_resume): New global.
(lwp_free): New.
(purge_lwp_list): Use it.
(add_lwp): Call linux_nat_new_thread even on the first LWP.
Adjust to interface change.
(delete_lwp): Call lwp_free instead of xfree.
(detach_callback, linux_nat_detach, resume_lwp, linux_nat_resume)
(linux_handle_syscall_trap, linux_handle_extended_wait)
(linux_nat_filter_event, resume_stopped_resumed_lwps): Call
linux_nat_prepare_to_resume before resuming.
(linux_stop_lwp): New.
(linux_nat_set_new_thread): Adjust.
(linux_nat_set_prepare_to_resume): New.
* linux-nat.h (struct arch_lwp_info): Forward declare.
(struct lwp_info) <arch_private>: New field.
(linux_stop_lwp): Declare.
(linux_nat_set_new_thread): Adjust.
(linux_nat_set_prepare_to_resume): New.
* i386-nat.c (DR_NADDR, DR_STATUS, DR_CONTROL)
(struct i386_debug_reg_state): Move to i386-nat.h.
(dr_mirror): Comment.
(i386_debug_reg_state): New.
(i386_update_inferior_debug_regs): Simplify.
(i386_stopped_data_address): Use the debug register state from the
inferior, not from the local cache.
* i386-nat.h (struct i386_dr_low_type): Delete reset_addr and
unset_status fields. New get_addr and get_control fields.
(DR_FIRSTADDR, DR_LASTADDR, DR_CONTROL): Moved from i386-nat.c.
(DR_NADDR, DR_STATUS): New.
(struct i386_debug_reg_state): Moved from i386-nat.c.
* amd64-linux-nat.c (struct arch_lwp_info): New.
(amd64_linux_dr): Delete global.
(amd64_linux_dr_get_addr): New.
(amd64_linux_dr_get_control): New.
(amd64_linux_dr_unset_status): Delete.
(amd64_linux_dr_set_addr): Reimplement.
(amd64_linux_dr_reset_addr): Delete.
(update_debug_registers_callback): New.
(amd64_linux_dr_set_control): Reimplement.
(amd64_linux_dr_set_addr): Reimplement.
(amd64_linux_prepare_to_resume): New.
(amd64_linux_new_thread): Change parameter to an lwp pointer.
Reimplement.
(_initialize_amd64_linux_nat): No longer install
i386_dr_low.reset_addr and i386_dr_low.unset_status. Install
amd64_linux_dr_get_control as i386_dr_low.get_control. Install
amd64_linux_dr_get_addr as i386_dr_low.get_addr. Install
amd64_linux_prepare_to_resume.
* i386-linux-nat.c (DR_FIRSTADDR, DR_LASTADDR, DR_STATUS)
(DR_CONTROL): Delete.
(struct arch_lwp_info): New.
(i386_linux_dr): Delete global.
(i386_linux_dr_set_control): Reimplement.
(i386_linux_dr_get_addr): New.
(i386_linux_dr_set_addr): Reimplement.
(i386_linux_dr_get_control): New.
(update_debug_registers_callback): New.
(i386_linux_dr_unset_status): Delete.
(i386_linux_dr_set_addr): Reimplement.
(i386_linux_prepare_to_resume): New.
(i386_linux_new_thread): Change parameter to an lwp pointer.
Reimplement.
(_initialize_i386_linux_nat): No longer install
i386_dr_low.reset_addr and i386_dr_low.unset_status. Install
i386_linux_dr_get_control as i386_dr_low.get_control. Install
i386_linux_dr_get_addr as i386_dr_low.get_addr. Install
i386_linux_prepare_to_resume.
* arm-linux-nat.c (arm_linux_new_thread): Change parameter to an
lwp pointer. Adjust.
* ia64-linux-nat.c (ia64_linux_new_thread): Likewise.
* mips-linux-nat.c (mips_linux_new_thread): Likewise.
* ppc-linux-nat.c (ppc_linux_new_thread): Likewise.
* s390-nat.c (s390_fix_watch_points): Likewise.
* i386-darwin-nat.c (DR_FIRSTADDR, DR_LASTADDR, DR_STATUS)
(DR_CONTROL): Delete.
(i386_darwin_dr_reset_addr): Delete.
(i386_darwin_dr_get_addr): New.
(i386_darwin_dr_get_control): New.
* go32-nat.c
(go32_get_dr7, go32_get_dr): New.
(init_go32_ops): No longer install i386_dr_low.reset_addr.
Install go32_get_dr7 as i386_dr_low.get_control. Install
go32_get_dr as i386_dr_low.get_addr.
* i386bsd-nat.c (i386bsd_dr_get): New.
(i386bsd_dr_reset_addr): Delete.
(i386bsd_dr_get_addr): New.
(i386bsd_dr_get_status): Use i386bsd_dr_get.
(i386bsd_dr_get_control): New.
* i386bsd-nat.h (i386bsd_dr_reset_addr): Delete.
(i386bsd_dr_get_addr): New.
(i386bsd_dr_get_control): New.
* i386fbsd-nat.c (_initialize_i386fbsd_nat): No longer install
i386_dr_low.reset_addr and i386_dr_low.unset_status. Install
i386bsd_dr_get_control as i386_dr_low.get_control. Install
i386bsd_dr_get_addr as i386_dr_low.get_addr.
* windows-nat.c (init_windows_ops): No longer install
i386_dr_low.reset_addr and i386_dr_low.unset_status. Install
cygwin_get_dr7 as i386_dr_low.get_control. Install cygwin_get_dr
as i386_dr_low.get_addr.
(cygwin_get_dr): New.
(cygwin_get_dr7): New.
gdb/testsuite/
2011-12-14 Pedro Alves <pedro@codesourcery.com>
PR threads/10729
* gdb.mi/watch-nonstop.c: New file.
* gdb.mi/mi-watch-nonstop.exp: New file.
2011-12-14 17:20:32 +00:00
|
|
|
}
|
|
|
|
|
2004-02-28 15:38:04 +00:00
|
|
|
/* Clear the reference counts and forget everything we knew about the
|
|
|
|
debug registers. */
|
|
|
|
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
void
|
|
|
|
i386_cleanup_dregs (void)
|
|
|
|
{
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
/* Starting from scratch has the same effect. */
|
|
|
|
i386_forget_process (ptid_get_pid (inferior_ptid));
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Insert a watchpoint to watch a memory region which starts at
|
|
|
|
address ADDR and whose length is LEN bytes. Watch memory accesses
|
|
|
|
of the type TYPE. Return 0 on success, -1 on failure. */
|
2004-02-28 15:38:04 +00:00
|
|
|
|
2009-05-14 09:37:00 +00:00
|
|
|
static int
|
2013-12-18 04:29:31 +00:00
|
|
|
i386_insert_watchpoint (struct target_ops *self,
|
|
|
|
CORE_ADDR addr, int len, int type,
|
2010-07-07 16:15:18 +00:00
|
|
|
struct expression *cond)
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
{
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
struct i386_debug_reg_state *state
|
|
|
|
= i386_debug_reg_state (ptid_get_pid (inferior_ptid));
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
|
2014-06-19 09:53:12 +00:00
|
|
|
return i386_dr_insert_watchpoint (state, type, addr, len);
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Remove a watchpoint that watched the memory region which starts at
|
|
|
|
address ADDR, whose length is LEN bytes, and for accesses of the
|
|
|
|
type TYPE. Return 0 on success, -1 on failure. */
|
2009-05-14 09:37:00 +00:00
|
|
|
static int
|
2013-12-18 04:29:17 +00:00
|
|
|
i386_remove_watchpoint (struct target_ops *self,
|
|
|
|
CORE_ADDR addr, int len, int type,
|
2010-07-07 16:15:18 +00:00
|
|
|
struct expression *cond)
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
{
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
struct i386_debug_reg_state *state
|
|
|
|
= i386_debug_reg_state (ptid_get_pid (inferior_ptid));
|
2011-07-22 16:58:32 +00:00
|
|
|
|
2014-06-19 09:53:12 +00:00
|
|
|
return i386_dr_remove_watchpoint (state, type, addr, len);
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Return non-zero if we can watch a memory region that starts at
|
|
|
|
address ADDR and whose length is LEN bytes. */
|
2004-02-28 15:38:04 +00:00
|
|
|
|
2009-05-14 09:37:00 +00:00
|
|
|
static int
|
2013-12-18 04:29:45 +00:00
|
|
|
i386_region_ok_for_watchpoint (struct target_ops *self,
|
|
|
|
CORE_ADDR addr, int len)
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
{
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
struct i386_debug_reg_state *state
|
|
|
|
= i386_debug_reg_state (ptid_get_pid (inferior_ptid));
|
2004-02-28 15:38:04 +00:00
|
|
|
|
2014-06-19 09:53:12 +00:00
|
|
|
return i386_dr_region_ok_for_watchpoint (state, addr, len);
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
}
|
|
|
|
|
2014-06-17 10:32:26 +00:00
|
|
|
/* If the inferior has some break/watchpoint that triggered, set the
|
|
|
|
address associated with that break/watchpoint and return non-zero.
|
2004-10-08 17:30:48 +00:00
|
|
|
Otherwise, return zero. */
|
2004-02-28 15:38:04 +00:00
|
|
|
|
2009-05-14 09:37:00 +00:00
|
|
|
static int
|
2008-03-01 04:39:36 +00:00
|
|
|
i386_stopped_data_address (struct target_ops *ops, CORE_ADDR *addr_p)
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
{
|
[native x86 GNU/Linux] Access debug register mirror from the corresponding process.
While reviewing the native AArch64 patch, I noticed a problem:
On 02/06/2013 08:46 PM, Pedro Alves wrote:
>
>> > +static void
>> > +aarch64_linux_prepare_to_resume (struct lwp_info *lwp)
>> > +{
>> > + struct arch_lwp_info *info = lwp->arch_private;
>> > +
>> > + /* NULL means this is the main thread still going through the shell,
>> > + or, no watchpoint has been set yet. In that case, there's
>> > + nothing to do. */
>> > + if (info == NULL)
>> > + return;
>> > +
>> > + if (DR_HAS_CHANGED (info->dr_changed_bp)
>> > + || DR_HAS_CHANGED (info->dr_changed_wp))
>> > + {
>> > + int tid = GET_LWP (lwp->ptid);
>> > + struct aarch64_debug_reg_state *state = aarch64_get_debug_reg_state ();
> Hmm. This is always fetching the debug_reg_state of
> the current inferior, but may not be the inferior of lwp.
> I see the same bug on x86. Sorry about that. I'll fix it.
A natural fix would be to make xxx_get_debug_reg_state take an
inferior argument, but that doesn't work because of the case where we
detach breakpoints/watchpoints from the child fork, at a time there's
no inferior for the child fork at all. We do a nasty hack in
i386_inferior_data_get, but that relies on all callers pointing the
current inferior to the correct inferior, which isn't actually being
done by all callers, and I don't think we want to enforce that -- deep
in the bowls of linux-nat.c, there are many cases we resume lwps
behind the scenes, and it's be better to not have that code rely on
global state (as it doesn't today).
The fix is to decouple the watchpoints code from inferiors, making it
track target processes instead. This way, we can freely keep track of
the watchpoint mirrors for these processes behind the core's back.
Checkpoints also play dirty tricks with swapping the process behind
the inferior, so they get special treatment too in the patch (which
just amounts to calling a new hook). Instead of the old hack in
i386_inferior_data_get, where we returned a copy of the current
inferior's debug registers mirror, as soon as we detect a fork in the
target, we copy the debug register mirror from the parent to the child
process.
I don't have an old kernel handy to test, but I stepped through gdb doing
the watchpoint removal in the fork child in the watchpoint-fork test
seeing that the debug registers end up cleared in the child.
I didn't find the need for linux_nat_iterate_watchpoint_lwps. If
we use plain iterate_over_lwps instead, what happens is that
when removing watchpoints, that iterate_over_lwps doesn't actually
iterate over anything, since the fork child is not added to the
lwp list until later, at detach time, in linux_child_follow_fork.
And if we don't iterate over that lwp, we don't mark its debug
registers as needing update. But linux_child_follow_fork takes
care of doing that explicitly:
child_lp = add_lwp (inferior_ptid);
child_lp->stopped = 1;
child_lp->last_resume_kind = resume_stop;
make_cleanup (delete_lwp_cleanup, child_lp);
/* CHILD_LP has new PID, therefore linux_nat_new_thread is not called for it.
See i386_inferior_data_get for the Linux kernel specifics.
Ensure linux_nat_prepare_to_resume will reset the hardware debug
registers. It is done by the linux_nat_new_thread call, which is
being skipped in add_lwp above for the first lwp of a pid. */
gdb_assert (num_lwps (GET_PID (child_lp->ptid)) == 1);
if (linux_nat_new_thread != NULL)
linux_nat_new_thread (child_lp);
if (linux_nat_prepare_to_resume != NULL)
linux_nat_prepare_to_resume (child_lp);
ptrace (PTRACE_DETACH, child_pid, 0, 0);
so unless I'm missing something (quite possible) it ends up all
the same. But, the !detach-on-fork, and the "follow-fork child" paths
should also call linux_nat_new_thread, and they don't presently. It
seems to me in those cases we're not clearing debug regs correctly
when that's needed. Instead of copying that bit that works around
add_lwp bypassing the linux_nat_new_thread call, I thought it'd
be better to add an add_initial_lwp call to be used in the case we
really need to bypass linux_nat_new_thread, and make
add_lwp always call linux_nat_new_thread.
i386_cleanup_dregs is rewritten to forget about the current process
debug mirrors, which takes cares of other i386 ports. Only a couple
of extra tweaks here and there were needed, as some targets wheren't
actually calling i386_cleanup_dregs.
Tested on Fedora 17 x86_64 -m64/-m32.
GDBserver already fetches the i386_debug_reg_state from the right
process, and, it doesn't handle forks at all, so no fix is needed over
there.
gdb/
2013-02-13 Pedro Alves <palves@redhat.com>
* amd64-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(amd64_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(amd64_linux_new_fork): New function.
(_initialize_amd64_linux_nat): Install amd64_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-linux-nat.c (update_debug_registers_callback):
Update comment.
(amd64_linux_dr_set_control, amd64_linux_dr_set_addr): Use
iterate_over_lwps.
(i386_linux_prepare_to_resume): Pass the lwp's pid to
i386_debug_reg_state.
(i386_linux_new_fork): New function.
(_initialize_i386_linux_nat): Install i386_linux_new_fork as
linux_nat_new_fork hook, and i386_forget_process as
linux_nat_forget_process hook.
* i386-nat.c (i386_init_dregs): Delete.
(i386_inferior_data, struct i386_inferior_data):
Delete.
(struct i386_process_info): New.
(i386_process_list): New global.
(i386_find_process_pid, i386_add_process, i386_process_info_get):
New functions.
(i386_inferior_data_get): Delete.
(i386_process_info_get): New function.
(i386_debug_reg_state): New parameter 'pid'. Reimplement.
(i386_forget_process): New function.
(i386_cleanup_dregs): Rewrite.
(i386_update_inferior_debug_regs, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Adjust to pass the current process id
to i386_debug_reg_state.
(i386_use_watchpoints): Don't register inferior data.
* i386-nat.h (i386_debug_reg_state): Add new 'pid' parameter, and
adjust comment.
(i386_forget_process): Declare.
* linux-fork.c (delete_fork): Call linux_nat_forget_process.
* linux-nat.c (linux_nat_new_fork, linux_nat_forget_process_hook):
New static globals.
(linux_child_follow_fork): Don't call linux_nat_new_thread here.
(add_initial_lwp): New, factored out from ...
(add_lwp): ... this. Don't check the number of lwps before
calling linux_nat_new_thread.
(linux_nat_iterate_watchpoint_lwps): Delete.
(linux_nat_attach): Use add_initial_lwp instead of add_lwp.
(linux_handle_extended_wait): Call the linux_nat_new_fork hook on
forks and vforks.
(linux_nat_wait_1): Use add_initial_lwp instead of add_lwp for the
initial lwp.
(linux_nat_kill, linux_nat_mourn_inferior): Call
linux_nat_forget_process.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New functions.
* linux-nat.h (linux_nat_iterate_watchpoint_lwps_ftype): Delete
type.
(linux_nat_iterate_watchpoint_lwps): Delete declaration.
(linux_nat_new_fork_ftype, linux_nat_forget_process_ftype): New
types.
(linux_nat_set_new_fork, linux_nat_set_forget_process)
(linux_nat_forget_process): New declarations.
* amd64fbsd-nat.c (super_mourn_inferior): New global.
(amd64fbsd_mourn_inferior): New function.
(_initialize_amd64fbsd_nat): Override to_mourn_inferior.
* windows-nat.c (windows_detach): Call i386_cleanup_dregs.
2013-02-13 14:59:49 +00:00
|
|
|
struct i386_debug_reg_state *state
|
|
|
|
= i386_debug_reg_state (ptid_get_pid (inferior_ptid));
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
|
2014-06-19 09:53:12 +00:00
|
|
|
return i386_dr_stopped_data_address (state, addr_p);
|
2004-10-08 17:30:48 +00:00
|
|
|
}
|
|
|
|
|
2014-06-17 10:32:26 +00:00
|
|
|
/* Return non-zero if the inferior has some watchpoint that triggered.
|
|
|
|
Otherwise return zero. */
|
|
|
|
|
2009-05-14 09:37:00 +00:00
|
|
|
static int
|
add "this" pointers to more target APIs
A subsequent pass introduces delegation helper functions to the target
API. This delegation is much cleaner if the target_ops pointer is
directly available at delegation time.
This patch adds the "this" pointer to various to_* methods for this
purpose.
This updates a number of ports which I am unable to test. Please give
them a look-over. Any possible problem here is trivial, though, as
all that is required is adding an argument to a function.
2014-02-19 Tom Tromey <tromey@redhat.com>
* aarch64-linux-nat.c (aarch64_linux_stopped_by_watchpoint):
Add 'ops' argument.
* arm-linux-nat.c (arm_linux_stopped_by_watchpoint): Add
'ops' argument.
* i386-nat.c (i386_stopped_by_watchpoint): Add 'ops' argument.
* ia64-linux-nat.c (ia64_linux_stopped_by_watchpoint): Add
'ops' argument.
* inf-ttrace.c (inf_ttrace_stopped_by_watchpoint): Add 'ops'
argument.
* linux-nat.c (save_sigtrap): Update.
(linux_nat_stopped_by_watchpoint, linux_nat_is_async_p)
(linux_nat_can_async_p, linux_nat_async): Add 'ops' argument.
(linux_nat_close): Update.
* mips-linux-nat.c (mips_linux_stopped_by_watchpoint): Add 'ops'
argument.
* ppc-linux-nat.c (ppc_linux_stopped_by_watchpoint): Add 'ops'
argument.
* procfs.c (procfs_stopped_by_watchpoint): Add 'ops' argument.
* record-full.c (record_full_beneath_to_stopped_by_watchpoint)
(record_full_beneath_to_async, tmp_to_stopped_by_watchpoint)
(tmp_to_async): Add 'ops' argument.
(record_full_stopped_by_watchpoint, record_full_async)
(record_full_can_async_p, record_full_is_async_p): Add 'ops'
argument.
* remote-m32r-sdi.c (m32r_insert_breakpoint, m32r_remove_breakpoint)
(m32r_stopped_by_watchpoint): Add 'ops' argument.
* remote-mips.c (mips_stopped_by_watchpoint): Add 'ops' argument.
* remote.c (remote_stopped_by_watchpoint_p, remote_can_async_p)
(remote_is_async_p, remote_async): Add 'ops' argument.
(remote_stopped_data_address): Update.
* s390-nat.c (s390_stopped_by_watchpoint): Add 'ops' argument.
* target.c (update_current_target)
(find_default_can_async_p, find_default_is_async_p): Update.
(init_dummy_target): Update.
(debug_to_stopped_by_watchpoint): Add 'ops' argument.
* target.h (struct target_ops) <to_stopped_by_watchpoint,
to_can_async_p, to_is_async_p, to_async>: Add 'ops' argument.
(target_can_async_p, target_is_async_p, target_async)
(target_stopped_by_watchpoint): Update.
2013-07-30 16:36:07 +00:00
|
|
|
i386_stopped_by_watchpoint (struct target_ops *ops)
|
2004-10-08 17:30:48 +00:00
|
|
|
{
|
|
|
|
CORE_ADDR addr = 0;
|
add "this" pointers to more target APIs
A subsequent pass introduces delegation helper functions to the target
API. This delegation is much cleaner if the target_ops pointer is
directly available at delegation time.
This patch adds the "this" pointer to various to_* methods for this
purpose.
This updates a number of ports which I am unable to test. Please give
them a look-over. Any possible problem here is trivial, though, as
all that is required is adding an argument to a function.
2014-02-19 Tom Tromey <tromey@redhat.com>
* aarch64-linux-nat.c (aarch64_linux_stopped_by_watchpoint):
Add 'ops' argument.
* arm-linux-nat.c (arm_linux_stopped_by_watchpoint): Add
'ops' argument.
* i386-nat.c (i386_stopped_by_watchpoint): Add 'ops' argument.
* ia64-linux-nat.c (ia64_linux_stopped_by_watchpoint): Add
'ops' argument.
* inf-ttrace.c (inf_ttrace_stopped_by_watchpoint): Add 'ops'
argument.
* linux-nat.c (save_sigtrap): Update.
(linux_nat_stopped_by_watchpoint, linux_nat_is_async_p)
(linux_nat_can_async_p, linux_nat_async): Add 'ops' argument.
(linux_nat_close): Update.
* mips-linux-nat.c (mips_linux_stopped_by_watchpoint): Add 'ops'
argument.
* ppc-linux-nat.c (ppc_linux_stopped_by_watchpoint): Add 'ops'
argument.
* procfs.c (procfs_stopped_by_watchpoint): Add 'ops' argument.
* record-full.c (record_full_beneath_to_stopped_by_watchpoint)
(record_full_beneath_to_async, tmp_to_stopped_by_watchpoint)
(tmp_to_async): Add 'ops' argument.
(record_full_stopped_by_watchpoint, record_full_async)
(record_full_can_async_p, record_full_is_async_p): Add 'ops'
argument.
* remote-m32r-sdi.c (m32r_insert_breakpoint, m32r_remove_breakpoint)
(m32r_stopped_by_watchpoint): Add 'ops' argument.
* remote-mips.c (mips_stopped_by_watchpoint): Add 'ops' argument.
* remote.c (remote_stopped_by_watchpoint_p, remote_can_async_p)
(remote_is_async_p, remote_async): Add 'ops' argument.
(remote_stopped_data_address): Update.
* s390-nat.c (s390_stopped_by_watchpoint): Add 'ops' argument.
* target.c (update_current_target)
(find_default_can_async_p, find_default_is_async_p): Update.
(init_dummy_target): Update.
(debug_to_stopped_by_watchpoint): Add 'ops' argument.
* target.h (struct target_ops) <to_stopped_by_watchpoint,
to_can_async_p, to_is_async_p, to_async>: Add 'ops' argument.
(target_can_async_p, target_is_async_p, target_async)
(target_stopped_by_watchpoint): Update.
2013-07-30 16:36:07 +00:00
|
|
|
return i386_stopped_data_address (ops, &addr);
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
}
|
|
|
|
|
gdb/
* breakpoint.c (deprecated_read_memory_nobpt): Update to use
shadow_len.
(insert_bp_location, reattach_breakpoints, remove_breakpoint)
(delete_breakpoint): Update calls to changed methods.
(deprecated_insert_raw_breakpoint, deprecated_remove_raw_breakpoint)
(single_step_breakpoints, insert_single_step_breakpoint)
(remove_single_step_breakpoints): New.
* breakpoint.h (struct bp_target_info): New.
(struct bp_location): Replace shadow_contents with
target_info and overlay_target_info.
(deprecated_insert_raw_breakpoint, deprecated_remove_raw_breakpoint)
(insert_single_step_breakpoint, remove_single_step_breakpoints): New
prototypes.
* gdbarch.sh: Forward declare struct bp_target_info in gdbarch.h.
(memory_insert_breakpoint, memory_remove_breakpoint): Update second
argument.
* mem-break.c (default_memory_insert_breakpoint): Update. Set
placed_address, placed_size, and shadow_len.
(default_memory_remove_breakpoint): Update. Don't use
BREAKPOINT_FROM_PC.
(memory_insert_breakpoint, memory_remove_breakpoint): Update.
* target.c (update_current_target): Update prototypes for changed
functions.
(debug_to_insert_breakpoint, debug_to_remove_breakpoint)
(debug_to_insert_hw_breakpoint, debug_to_remove_hw_breakpoint):
Update.
* target.h: Forward declare struct bp_target_info.
(struct target_ops): Use a bp_target_info argument for
to_insert_breakpoint, to_remove_breakpoint,
to_insert_hw_breakpoint, and to_remove_hw_breakpoint.
(target_insert_breakpoint, target_remove_breakpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint)
(memory_insert_breakpoint, memory_remove_breakpoint)
(default_memory_insert_breakpoint, default_memory_remove_breakpoint):
Update.
* config/i386/nm-i386.h: Forward declare struct bp_target_info.
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): Update.
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Likewise.
* gdbarch.c, gdbarch.h: Regenerated.
* alpha-tdep.c (alpha_software_single_step): Use
insert_single_step_breakpoint and remove_single_step_breakpoints.
Remove unused statics.
* arm-tdep.c (arm_software_single_step): Likewise. Add a note.
* cris-tdep.c (cris_software_single_step): Likewise.
* mips-tdep.c (mips_software_single_step): Likewise.
* rs6000-tdep.c (rs6000_software_single_step): Likewise.
* sparc-tdep.c (sparc_software_single_step): Likewise.
* wince.c (struct thread_info_struct): Remove step_prev.
(undoSStep): Use remove_single_step_breakpoints.
(wince_software_single_step): Use insert_single_step_breakpoint.
* corelow.c (ignore): Remove unneeded prototype. Update arguments.
* exec.c (ignore): Likewise.
* sol-thread.c (ignore): Likewise.
* procfs.c (dbx_link_shadow_contents): Delete.
(dbx_link_bpt): New.
(procfs_mourn_inferior): Remove it if necessary.
(remove_dbx_link_breakpoint): Use it.
(insert_dbx_link_bpt_in_file): Set it.
(procfs_init_inferior): Don't update dbx_link_bpt_addr.
* rs6000-nat.c (exec_one_dummy_insn): Use
deprecated_insert_raw_breakpoint and
deprecated_remove_raw_breakpoint.
* solib-irix.c (shadow_contents, breakpoint_addr): Delete.
(base_breakpoint): New.
(disable_break): Use it.
(enable_break): Set it.
* i386-nat.c (i386_insert_hw_breakpoint, i386_remove_hw_breakpoint):
Update.
* ia64-tdep.c (ia64_memory_insert_breakpoint)
(ia64_memory_remove_breakpoint): Likewise.
* m32r-tdep.c (m32r_memory_insert_breakpoint)
(m32r_memory_remove_breakpoint): Likewise.
* monitor.c (monitor_insert_breakpoint, monitor_remove_breakpoint):
Likewise. Remove unnecessary prototypes. Use placed_address
and placed_size. Removed useless read from memory.
* nto-procfs.c (procfs_insert_breakpoint)
(procfs_remove_breakpoint, procfs_insert_hw_breakpoint)
(procfs_remove_hw_breakpoint): Update.
* ocd.c (ocd_insert_breakpoint, ocd_remove_breakpoint): Likewise.
* ocd.h (ocd_insert_breakpoint, ocd_remove_breakpoint): Likewise.
* ppc-linux-tdep.c (ppc_linux_memory_remove_breakpoint): Likewise.
* ppc-tdep.h (ppc_linux_memory_remove_breakpoint): Likewise.
* remote-e7000.c (e7000_insert_breakpoint)
(e7000_remove_breakpoint): Likewise.
* remote-m32r-sdi.c (m32r_insert_breakpoint)
(m32r_remove_breakpoint): Likewise.
* remote-mips.c (mips_insert_breakpoint)
(mips_remove_breakpoint): Likewise.
* remote-rdp.c (remote_rdp_insert_breakpoint)
(remote_rdp_remove_breakpoint): Likewise.
(rdp_step): Use deprecated_insert_raw_breakpoint and
deprecated_remove_raw_breakpoint.
* remote-sds.c (sds_insert_breakpoint, sds_remove_breakpoint):
Update.
* remote-sim.c (gdbsim_insert_breakpoint, gdbsim_remove_breakpoint):
Delete.
(init_gdbsim_ops): Use memory_insert_breakpoint and
memory_remove_breakpoint.
* remote-st.c (st2000_insert_breakpoint)
(st2000_remove_breakpoint): Update. Remove unused
BREAKPOINT_FROM_PC.
* remote.c (remote_insert_breakpoint, remote_remove_breakpoint):
Update. Use placed_address and placed_size.
(remote_insert_hw_breakpoint, remote_remove_hw_breakpoint): Likewise.
gdb/doc/
* gdbint.texinfo (x86 Watchpoints, Target Conditionals): Update insert
and remove breakpoint prototypes.
(Watchpoints): Move description of target_insert_hw_breakpoint and
target_remove_hw_breakpoint ...
(Breakpoints): ... to here. Document target_insert_breakpoint and
target_remove_breakpoint.
2006-04-18 19:20:08 +00:00
|
|
|
/* Insert a hardware-assisted breakpoint at BP_TGT->placed_address.
|
|
|
|
Return 0 on success, EBUSY on failure. */
|
2014-06-19 09:50:20 +00:00
|
|
|
|
2009-05-14 09:37:00 +00:00
|
|
|
static int
|
2013-12-18 04:28:50 +00:00
|
|
|
i386_insert_hw_breakpoint (struct target_ops *self, struct gdbarch *gdbarch,
|
* breakpoint.h (struct breakpoint): New member GDBARCH.
* breakpoint.c: Include "arch-utils.h".
(set_raw_breakpoint_without_location): Add GDBARCH parameter.
Use it to set breakpoint architecture.
(set_raw_breakpoint): Add GDBARCH parameter.
(create_internal_breakpoint): Likewise.
(create_catchpoint): Likewise.
(create_fork_vfork_event_catchpoint): Likewise.
(create_breakpoint): Likewise.
(create_breakpoints): Likewise.
(break_command_really): Likewise.
(create_ada_exception_breakpoint): Likewise.
Update local callers to pass architecture:
(create_internal_breakpoint): Update.
(create_overlay_event_breakpoint): Update.
(create_longjmp_master_breakpoint): Update.
(create_thread_event_breakpoint): Update.
(create_solib_event_breakpoint): Update.
(create_catchpoint): Update.
(create_fork_vfork_event_catchpoint): Update.
(set_momentary_breakpoint): Update.
(clone_momentary_breakpoint): Update.
(create_breakpoint): Update.
(create_breakpoints): Update.
(break_command_really): Update.
(break_command_1): Update.
(set_breakpoint): Update.
(watch_command_1): Update.
(catch_fork_command_1): Update.
(catch_exec_commnd_1): Update.
(handle_gnu_v3_exceptions): Update.
(create_ada_exception_breakpoint): Update.
(catch_ada_exception_command): Update.
(catch_assert_command): Update.
(trace_command): Update.
* breakpoint.h (struct bp_location): New member GDBARCH.
* breakpoint.c (get_sal_arch): New function.
(set_raw_breakpoint): Set location architecture.
(add_location_to_breakpoint): Likewise.
(clone_momentary_breakpoint): Likewise.
(watch_command_1): Likewise.
(update_watchpoint): Likewise.
(bp_loc_is_permanent): Use location architecture instead of
current_gdbarch.
(adjust_breakpoint_address): Add GDBARCH parameter; use it
instead of current_gdbarch.
Update callers of adjust_breakpoint_address to pass
breakpoint location architecture:
(set_raw_breakpoint): Update.
(watch_command_1): Update.
* tracepoint.c: (collect_symbol): Add GDBARCH parameter, use instead
of current_gdbarch.
(add_local_symbols): Add GDBARCH parameter. Pass to collect_symbol.
(encode_actions): Pass tracepoint architecture to add_local_symbols
(encode_actions): Use tracepoint architecture instead of
current_gdbarch. Pass it to add_local_symbols and collect_symbol.
* breakpoint.h (struct breakpoint_ops): Replace last_addr parameter
of print_one callback with last_loc.
* breakpoint.c (print_one_breakpoint_location): Replace last_addr
parameter with last_loc.
(print_one_breakpoint): Likewise.
(do_captured_breakpoint_query): Update call.
(breakpoint_1): Pass last_loc instead of last_addr to
print_one_breakpoint. Pass last location architecture instead of
current_gdbarch to set_next_address.
Update all implementations of the print_one callback:
* breakpoint.c (print_one_catch_fork): Update.
(print_one_catch_vfork): Update.
(print_one_catch_exec): Update.
(print_one_exception_catchpoint): Update.
* ada-lang.c (print_one_exception): Update.
(print_one_catch_exception): Update.
(print_one_catch_exception_unhandled): Update.
(print_one_catch_assert): Update.
* breakpoint.c (print_one_breakpoint_location): Add PRINT_ADDRESS_BITS
parameter. Use it instead of gdbarch_addr_bit (current_gdbarch).
(print_one_breakpoint): Add PRINT_ADDRESS_BITS parameter and pass it
to print_one_breakpoint_location.
(breakpoint_address_bits): New function.
(do_captured_breakpoint_query): Compute number of address bits to print
and pass it to print_one_breakpoint.
(breakpoint_1): Likewise. Use it instead of current_gdbarch.
* breakpoint.h (create_thread_event_breakpoint): Add GDBARCH.
* breakpoint.c (create_thread_event_breakpoint): Likewise.
Update callers to create_thread_event_breakpoint:
* aix-thread.c (pd_enable): Update.
* linux-thread-db.c (enable_thread_event): Update.
* breakpoint.h (create_solib_event_breakpoint): Add GDBARCH.
* breakpoint.c (create_solib_event_breakpoint): Likewise.
Update callers to create_solib_event_breakpoint:
* solib-frv.c (enable_break, enable_break2): Update.
* solib-pa64.c (pa64_solib_create_inferior_hook): Update.
* solib-som.c (som_solib_create_inferior_hook): Update.
* solib-darwin.c (darwin_solib_create_inferior_hook): Update.
* solib-svr4.c (enable_break): Update.
* breakpoint.h (insert_single_step_breakpoint): Add GDBARCH.
* breakpoint.c (insert_single_step_breakpoint): Likewise.
Update callers to insert_single_step_breakpoint:
* alpha-tdep.c (alpha_software_single_step): Update.
* arm-linux-tdep.c (arm_linux_software_single_step): Update.
* arm-tdep.c (arm_software_single_step): Update.
* cris-tdep.c (cris_software_single_step): Update.
* rs6000-aix-tdep.c (rs6000_software_single_step): Update.
* rs6000-tdep.c (ppc_deal_with_atomic_sequence): Update.
* sparc-tdep.c (sparc_software_single_step): Update.
* spu-tdep.c (spu_software_single_step): Update.
* mips-tdep.c (deal_with_atomic_sequence): Add GDBARCH parameter.
Pass it to insert_single_step_breakpoint.
(mips_software_single_step): Pass architecture to
deal_with_atomic_sequence and insert_single_step_breakpoint.
* breakpoint.h (deprecated_insert_raw_breakpoint): Add GDBARCH.
(deprecated_remove_raw_breakpoint): Likewise.
* breakpoint.c (deprecated_insert_raw_breakpoint): Add GDBARCH.
(deprecated_remove_raw_breakpoint): Likewise.
Update callers to deprecated_insert_raw_breakpoint and
deprecated_remove_raw_breakpoint:
* breakpoint.c (single_step_gdbarch): New static variable.
(insert_single_step_breakpoint): Pass GDBARCH parameter to
deprecated_insert_raw_breakpoint. Store it in single_step_gdbarch.
(remove_single_step_breakpoints): Pass architecture stored in
single_step_gdbarch to deprecated_remove_raw_breakpoint.
* rs6000-nat.c (exec_one_dummy_insn): Update.
* solib-irix.c (enable_break, disable_break): Update.
* procfs.c (procfs_mourn_inferior): Update.
(remove_dbx_link_breakpoint): Update.
* breakpoint.h (set_breakpoint): Add GDBARCH parameter.
(set_momentary_breakpoint, set_momentary_breakpoint_at_pc): Likewise.
* breakpoint.c (set_breakpoint): Add GDBARCH parameter.
(set_momentary_breakpoint, set_momentary_breakpoint_at_pc): Likewise.
Update callers to set_breakpoint, set_momentary_breakpoint and
set_momentary_breakpoint_at_pc:
* breakpoint.c (set_momentary_breakpoint_at_pc): Update.
(until_break_command): Update.
* infcall.c (call_function_by_hand): Update.
* infcmd.c (finish_backward, finish_forward): Update.
* infrun.c (insert_step_resume_breakpoint_at_sal): Add GDBARCH
parameter. Pass it to set_momentary_breakpoint.
(insert_longjmp_resume_breakpoint): Add GDBARCH parameter.
Pass it to set_momentary_breakpoint_at_pc.
(handle_inferior_event): Update.
(insert_step_resume_breakpoint_at_frame): Update.
(insert_step_resume_breakpoint_at_caller): Update..
* mi/mi-cmd-break.c: Include "arch-utils.h".
(mi_cmd_break_insert): Update.
* target.h (struct target_ops): Add GDBARCH parameter to
to_insert_breakpoint, to_remove_breakpoint, to_insert_hw_breakpoint,
and to_remove_hw_breakpoint members.
(target_insert_breakpoint, target_remove_breakpoint,
target_insert_hw_breakpoint, target_remove_hw_breakpoint): Add GDBARCH
parameter, pass to target routine.
(memory_remove_breakpoint, memory_insert_breakpoint): Add GDBARCH
parameter.
* target.c (debug_to_insert_breakpoint, debug_to_remove_breakpoint,
debug_to_insert_hw_breakpoint, debug_to_remove_hw_breakpoint): Add
GDBARCH parameter, pass to target routine.
(update_current_target): Update function signature.
* breakpoint.c (insert_bp_location, remove_breakpoint,
deprecated_insert_raw_breakpoint, deprecated_remove_raw_breakpoint):
Pass architecture to target_ routines.
Update all implementations of the target breakpoint routines to take
GDBARCH parameter and use it instead of GDBARCH as appropriate:
* corelow.c (ignore): Update.
* exec.c (ignore): Update.
* mem-break.c (memory_insert_breakpoint): Update.
(memory_remove_breakpoint): Update.
* monitor.c (monitor_insert_breakpoint): Update.
(monitor_remove_breakpoint): Update.
* record.c (record_insert_breakpoint): Update.
(record_beneath_to_insert_breakpoint): Update.
(record_remove_breakpoint): Update.
(record_beneath_to_remove_breakpoint): Update.
* remote.c (remote_insert_breakpoint): Update.
(remote_remove_breakpoint): Update.
(remote_insert_hw_breakpoint): Update.
(remote_remove_hw_breakpoint): Update.
* remote-m32r-sdi.c (m32r_insert_breakpoint): Update.
(m32r_remove_breakpoint): Update.
* remote-mips.c (mips_insert_breakpoint): Update.
(mips_remove_breakpoint): Update.
* i386-nat.c (i386_insert_hw_breakpoint): Update.
(i386_remove_hw_breakpoint): Update.
* nto-procfs.c (procfs_insert_breakpoint): Update.
(procfs_remove_breakpoint): Update.
(procfs_insert_hw_breakpoint): Update.
(procfs_remove_hw_breakpoint): Update.
doc/ChangeLog:
* gdbint.texi (Examples of Use of @code{ui_out} functions):
Update example code extrated from breakpoint.c.
2009-07-02 17:12:28 +00:00
|
|
|
struct bp_target_info *bp_tgt)
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
{
|
2014-06-19 09:50:20 +00:00
|
|
|
return i386_insert_watchpoint (self, bp_tgt->placed_address, 1,
|
|
|
|
hw_execute, NULL) ? EBUSY : 0;
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
}
|
|
|
|
|
gdb/
* breakpoint.c (deprecated_read_memory_nobpt): Update to use
shadow_len.
(insert_bp_location, reattach_breakpoints, remove_breakpoint)
(delete_breakpoint): Update calls to changed methods.
(deprecated_insert_raw_breakpoint, deprecated_remove_raw_breakpoint)
(single_step_breakpoints, insert_single_step_breakpoint)
(remove_single_step_breakpoints): New.
* breakpoint.h (struct bp_target_info): New.
(struct bp_location): Replace shadow_contents with
target_info and overlay_target_info.
(deprecated_insert_raw_breakpoint, deprecated_remove_raw_breakpoint)
(insert_single_step_breakpoint, remove_single_step_breakpoints): New
prototypes.
* gdbarch.sh: Forward declare struct bp_target_info in gdbarch.h.
(memory_insert_breakpoint, memory_remove_breakpoint): Update second
argument.
* mem-break.c (default_memory_insert_breakpoint): Update. Set
placed_address, placed_size, and shadow_len.
(default_memory_remove_breakpoint): Update. Don't use
BREAKPOINT_FROM_PC.
(memory_insert_breakpoint, memory_remove_breakpoint): Update.
* target.c (update_current_target): Update prototypes for changed
functions.
(debug_to_insert_breakpoint, debug_to_remove_breakpoint)
(debug_to_insert_hw_breakpoint, debug_to_remove_hw_breakpoint):
Update.
* target.h: Forward declare struct bp_target_info.
(struct target_ops): Use a bp_target_info argument for
to_insert_breakpoint, to_remove_breakpoint,
to_insert_hw_breakpoint, and to_remove_hw_breakpoint.
(target_insert_breakpoint, target_remove_breakpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint)
(memory_insert_breakpoint, memory_remove_breakpoint)
(default_memory_insert_breakpoint, default_memory_remove_breakpoint):
Update.
* config/i386/nm-i386.h: Forward declare struct bp_target_info.
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): Update.
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Likewise.
* gdbarch.c, gdbarch.h: Regenerated.
* alpha-tdep.c (alpha_software_single_step): Use
insert_single_step_breakpoint and remove_single_step_breakpoints.
Remove unused statics.
* arm-tdep.c (arm_software_single_step): Likewise. Add a note.
* cris-tdep.c (cris_software_single_step): Likewise.
* mips-tdep.c (mips_software_single_step): Likewise.
* rs6000-tdep.c (rs6000_software_single_step): Likewise.
* sparc-tdep.c (sparc_software_single_step): Likewise.
* wince.c (struct thread_info_struct): Remove step_prev.
(undoSStep): Use remove_single_step_breakpoints.
(wince_software_single_step): Use insert_single_step_breakpoint.
* corelow.c (ignore): Remove unneeded prototype. Update arguments.
* exec.c (ignore): Likewise.
* sol-thread.c (ignore): Likewise.
* procfs.c (dbx_link_shadow_contents): Delete.
(dbx_link_bpt): New.
(procfs_mourn_inferior): Remove it if necessary.
(remove_dbx_link_breakpoint): Use it.
(insert_dbx_link_bpt_in_file): Set it.
(procfs_init_inferior): Don't update dbx_link_bpt_addr.
* rs6000-nat.c (exec_one_dummy_insn): Use
deprecated_insert_raw_breakpoint and
deprecated_remove_raw_breakpoint.
* solib-irix.c (shadow_contents, breakpoint_addr): Delete.
(base_breakpoint): New.
(disable_break): Use it.
(enable_break): Set it.
* i386-nat.c (i386_insert_hw_breakpoint, i386_remove_hw_breakpoint):
Update.
* ia64-tdep.c (ia64_memory_insert_breakpoint)
(ia64_memory_remove_breakpoint): Likewise.
* m32r-tdep.c (m32r_memory_insert_breakpoint)
(m32r_memory_remove_breakpoint): Likewise.
* monitor.c (monitor_insert_breakpoint, monitor_remove_breakpoint):
Likewise. Remove unnecessary prototypes. Use placed_address
and placed_size. Removed useless read from memory.
* nto-procfs.c (procfs_insert_breakpoint)
(procfs_remove_breakpoint, procfs_insert_hw_breakpoint)
(procfs_remove_hw_breakpoint): Update.
* ocd.c (ocd_insert_breakpoint, ocd_remove_breakpoint): Likewise.
* ocd.h (ocd_insert_breakpoint, ocd_remove_breakpoint): Likewise.
* ppc-linux-tdep.c (ppc_linux_memory_remove_breakpoint): Likewise.
* ppc-tdep.h (ppc_linux_memory_remove_breakpoint): Likewise.
* remote-e7000.c (e7000_insert_breakpoint)
(e7000_remove_breakpoint): Likewise.
* remote-m32r-sdi.c (m32r_insert_breakpoint)
(m32r_remove_breakpoint): Likewise.
* remote-mips.c (mips_insert_breakpoint)
(mips_remove_breakpoint): Likewise.
* remote-rdp.c (remote_rdp_insert_breakpoint)
(remote_rdp_remove_breakpoint): Likewise.
(rdp_step): Use deprecated_insert_raw_breakpoint and
deprecated_remove_raw_breakpoint.
* remote-sds.c (sds_insert_breakpoint, sds_remove_breakpoint):
Update.
* remote-sim.c (gdbsim_insert_breakpoint, gdbsim_remove_breakpoint):
Delete.
(init_gdbsim_ops): Use memory_insert_breakpoint and
memory_remove_breakpoint.
* remote-st.c (st2000_insert_breakpoint)
(st2000_remove_breakpoint): Update. Remove unused
BREAKPOINT_FROM_PC.
* remote.c (remote_insert_breakpoint, remote_remove_breakpoint):
Update. Use placed_address and placed_size.
(remote_insert_hw_breakpoint, remote_remove_hw_breakpoint): Likewise.
gdb/doc/
* gdbint.texinfo (x86 Watchpoints, Target Conditionals): Update insert
and remove breakpoint prototypes.
(Watchpoints): Move description of target_insert_hw_breakpoint and
target_remove_hw_breakpoint ...
(Breakpoints): ... to here. Document target_insert_breakpoint and
target_remove_breakpoint.
2006-04-18 19:20:08 +00:00
|
|
|
/* Remove a hardware-assisted breakpoint at BP_TGT->placed_address.
|
|
|
|
Return 0 on success, -1 on failure. */
|
2004-02-28 15:38:04 +00:00
|
|
|
|
2009-05-14 09:37:00 +00:00
|
|
|
static int
|
2013-12-18 04:29:03 +00:00
|
|
|
i386_remove_hw_breakpoint (struct target_ops *self, struct gdbarch *gdbarch,
|
* breakpoint.h (struct breakpoint): New member GDBARCH.
* breakpoint.c: Include "arch-utils.h".
(set_raw_breakpoint_without_location): Add GDBARCH parameter.
Use it to set breakpoint architecture.
(set_raw_breakpoint): Add GDBARCH parameter.
(create_internal_breakpoint): Likewise.
(create_catchpoint): Likewise.
(create_fork_vfork_event_catchpoint): Likewise.
(create_breakpoint): Likewise.
(create_breakpoints): Likewise.
(break_command_really): Likewise.
(create_ada_exception_breakpoint): Likewise.
Update local callers to pass architecture:
(create_internal_breakpoint): Update.
(create_overlay_event_breakpoint): Update.
(create_longjmp_master_breakpoint): Update.
(create_thread_event_breakpoint): Update.
(create_solib_event_breakpoint): Update.
(create_catchpoint): Update.
(create_fork_vfork_event_catchpoint): Update.
(set_momentary_breakpoint): Update.
(clone_momentary_breakpoint): Update.
(create_breakpoint): Update.
(create_breakpoints): Update.
(break_command_really): Update.
(break_command_1): Update.
(set_breakpoint): Update.
(watch_command_1): Update.
(catch_fork_command_1): Update.
(catch_exec_commnd_1): Update.
(handle_gnu_v3_exceptions): Update.
(create_ada_exception_breakpoint): Update.
(catch_ada_exception_command): Update.
(catch_assert_command): Update.
(trace_command): Update.
* breakpoint.h (struct bp_location): New member GDBARCH.
* breakpoint.c (get_sal_arch): New function.
(set_raw_breakpoint): Set location architecture.
(add_location_to_breakpoint): Likewise.
(clone_momentary_breakpoint): Likewise.
(watch_command_1): Likewise.
(update_watchpoint): Likewise.
(bp_loc_is_permanent): Use location architecture instead of
current_gdbarch.
(adjust_breakpoint_address): Add GDBARCH parameter; use it
instead of current_gdbarch.
Update callers of adjust_breakpoint_address to pass
breakpoint location architecture:
(set_raw_breakpoint): Update.
(watch_command_1): Update.
* tracepoint.c: (collect_symbol): Add GDBARCH parameter, use instead
of current_gdbarch.
(add_local_symbols): Add GDBARCH parameter. Pass to collect_symbol.
(encode_actions): Pass tracepoint architecture to add_local_symbols
(encode_actions): Use tracepoint architecture instead of
current_gdbarch. Pass it to add_local_symbols and collect_symbol.
* breakpoint.h (struct breakpoint_ops): Replace last_addr parameter
of print_one callback with last_loc.
* breakpoint.c (print_one_breakpoint_location): Replace last_addr
parameter with last_loc.
(print_one_breakpoint): Likewise.
(do_captured_breakpoint_query): Update call.
(breakpoint_1): Pass last_loc instead of last_addr to
print_one_breakpoint. Pass last location architecture instead of
current_gdbarch to set_next_address.
Update all implementations of the print_one callback:
* breakpoint.c (print_one_catch_fork): Update.
(print_one_catch_vfork): Update.
(print_one_catch_exec): Update.
(print_one_exception_catchpoint): Update.
* ada-lang.c (print_one_exception): Update.
(print_one_catch_exception): Update.
(print_one_catch_exception_unhandled): Update.
(print_one_catch_assert): Update.
* breakpoint.c (print_one_breakpoint_location): Add PRINT_ADDRESS_BITS
parameter. Use it instead of gdbarch_addr_bit (current_gdbarch).
(print_one_breakpoint): Add PRINT_ADDRESS_BITS parameter and pass it
to print_one_breakpoint_location.
(breakpoint_address_bits): New function.
(do_captured_breakpoint_query): Compute number of address bits to print
and pass it to print_one_breakpoint.
(breakpoint_1): Likewise. Use it instead of current_gdbarch.
* breakpoint.h (create_thread_event_breakpoint): Add GDBARCH.
* breakpoint.c (create_thread_event_breakpoint): Likewise.
Update callers to create_thread_event_breakpoint:
* aix-thread.c (pd_enable): Update.
* linux-thread-db.c (enable_thread_event): Update.
* breakpoint.h (create_solib_event_breakpoint): Add GDBARCH.
* breakpoint.c (create_solib_event_breakpoint): Likewise.
Update callers to create_solib_event_breakpoint:
* solib-frv.c (enable_break, enable_break2): Update.
* solib-pa64.c (pa64_solib_create_inferior_hook): Update.
* solib-som.c (som_solib_create_inferior_hook): Update.
* solib-darwin.c (darwin_solib_create_inferior_hook): Update.
* solib-svr4.c (enable_break): Update.
* breakpoint.h (insert_single_step_breakpoint): Add GDBARCH.
* breakpoint.c (insert_single_step_breakpoint): Likewise.
Update callers to insert_single_step_breakpoint:
* alpha-tdep.c (alpha_software_single_step): Update.
* arm-linux-tdep.c (arm_linux_software_single_step): Update.
* arm-tdep.c (arm_software_single_step): Update.
* cris-tdep.c (cris_software_single_step): Update.
* rs6000-aix-tdep.c (rs6000_software_single_step): Update.
* rs6000-tdep.c (ppc_deal_with_atomic_sequence): Update.
* sparc-tdep.c (sparc_software_single_step): Update.
* spu-tdep.c (spu_software_single_step): Update.
* mips-tdep.c (deal_with_atomic_sequence): Add GDBARCH parameter.
Pass it to insert_single_step_breakpoint.
(mips_software_single_step): Pass architecture to
deal_with_atomic_sequence and insert_single_step_breakpoint.
* breakpoint.h (deprecated_insert_raw_breakpoint): Add GDBARCH.
(deprecated_remove_raw_breakpoint): Likewise.
* breakpoint.c (deprecated_insert_raw_breakpoint): Add GDBARCH.
(deprecated_remove_raw_breakpoint): Likewise.
Update callers to deprecated_insert_raw_breakpoint and
deprecated_remove_raw_breakpoint:
* breakpoint.c (single_step_gdbarch): New static variable.
(insert_single_step_breakpoint): Pass GDBARCH parameter to
deprecated_insert_raw_breakpoint. Store it in single_step_gdbarch.
(remove_single_step_breakpoints): Pass architecture stored in
single_step_gdbarch to deprecated_remove_raw_breakpoint.
* rs6000-nat.c (exec_one_dummy_insn): Update.
* solib-irix.c (enable_break, disable_break): Update.
* procfs.c (procfs_mourn_inferior): Update.
(remove_dbx_link_breakpoint): Update.
* breakpoint.h (set_breakpoint): Add GDBARCH parameter.
(set_momentary_breakpoint, set_momentary_breakpoint_at_pc): Likewise.
* breakpoint.c (set_breakpoint): Add GDBARCH parameter.
(set_momentary_breakpoint, set_momentary_breakpoint_at_pc): Likewise.
Update callers to set_breakpoint, set_momentary_breakpoint and
set_momentary_breakpoint_at_pc:
* breakpoint.c (set_momentary_breakpoint_at_pc): Update.
(until_break_command): Update.
* infcall.c (call_function_by_hand): Update.
* infcmd.c (finish_backward, finish_forward): Update.
* infrun.c (insert_step_resume_breakpoint_at_sal): Add GDBARCH
parameter. Pass it to set_momentary_breakpoint.
(insert_longjmp_resume_breakpoint): Add GDBARCH parameter.
Pass it to set_momentary_breakpoint_at_pc.
(handle_inferior_event): Update.
(insert_step_resume_breakpoint_at_frame): Update.
(insert_step_resume_breakpoint_at_caller): Update..
* mi/mi-cmd-break.c: Include "arch-utils.h".
(mi_cmd_break_insert): Update.
* target.h (struct target_ops): Add GDBARCH parameter to
to_insert_breakpoint, to_remove_breakpoint, to_insert_hw_breakpoint,
and to_remove_hw_breakpoint members.
(target_insert_breakpoint, target_remove_breakpoint,
target_insert_hw_breakpoint, target_remove_hw_breakpoint): Add GDBARCH
parameter, pass to target routine.
(memory_remove_breakpoint, memory_insert_breakpoint): Add GDBARCH
parameter.
* target.c (debug_to_insert_breakpoint, debug_to_remove_breakpoint,
debug_to_insert_hw_breakpoint, debug_to_remove_hw_breakpoint): Add
GDBARCH parameter, pass to target routine.
(update_current_target): Update function signature.
* breakpoint.c (insert_bp_location, remove_breakpoint,
deprecated_insert_raw_breakpoint, deprecated_remove_raw_breakpoint):
Pass architecture to target_ routines.
Update all implementations of the target breakpoint routines to take
GDBARCH parameter and use it instead of GDBARCH as appropriate:
* corelow.c (ignore): Update.
* exec.c (ignore): Update.
* mem-break.c (memory_insert_breakpoint): Update.
(memory_remove_breakpoint): Update.
* monitor.c (monitor_insert_breakpoint): Update.
(monitor_remove_breakpoint): Update.
* record.c (record_insert_breakpoint): Update.
(record_beneath_to_insert_breakpoint): Update.
(record_remove_breakpoint): Update.
(record_beneath_to_remove_breakpoint): Update.
* remote.c (remote_insert_breakpoint): Update.
(remote_remove_breakpoint): Update.
(remote_insert_hw_breakpoint): Update.
(remote_remove_hw_breakpoint): Update.
* remote-m32r-sdi.c (m32r_insert_breakpoint): Update.
(m32r_remove_breakpoint): Update.
* remote-mips.c (mips_insert_breakpoint): Update.
(mips_remove_breakpoint): Update.
* i386-nat.c (i386_insert_hw_breakpoint): Update.
(i386_remove_hw_breakpoint): Update.
* nto-procfs.c (procfs_insert_breakpoint): Update.
(procfs_remove_breakpoint): Update.
(procfs_insert_hw_breakpoint): Update.
(procfs_remove_hw_breakpoint): Update.
doc/ChangeLog:
* gdbint.texi (Examples of Use of @code{ui_out} functions):
Update example code extrated from breakpoint.c.
2009-07-02 17:12:28 +00:00
|
|
|
struct bp_target_info *bp_tgt)
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
{
|
2014-06-19 09:50:20 +00:00
|
|
|
return i386_remove_watchpoint (self, bp_tgt->placed_address, 1,
|
|
|
|
hw_execute, NULL);
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
}
|
|
|
|
|
2008-03-01 04:39:36 +00:00
|
|
|
/* Returns the number of hardware watchpoints of type TYPE that we can
|
|
|
|
set. Value is positive if we can set CNT watchpoints, zero if
|
|
|
|
setting watchpoints of type TYPE is not supported, and negative if
|
|
|
|
CNT is more than the maximum number of watchpoints of type TYPE
|
|
|
|
that we can support. TYPE is one of bp_hardware_watchpoint,
|
|
|
|
bp_read_watchpoint, bp_write_watchpoint, or bp_hardware_breakpoint.
|
|
|
|
CNT is the number of such watchpoints used so far (including this
|
|
|
|
one). OTHERTYPE is non-zero if other types of watchpoints are
|
|
|
|
currently enabled.
|
|
|
|
|
|
|
|
We always return 1 here because we don't have enough information
|
|
|
|
about possible overlap of addresses that they want to watch. As an
|
|
|
|
extreme example, consider the case where all the watchpoints watch
|
|
|
|
the same address and the same region length: then we can handle a
|
|
|
|
virtually unlimited number of watchpoints, due to debug register
|
|
|
|
sharing implemented via reference counts in i386-nat.c. */
|
|
|
|
|
|
|
|
static int
|
2013-12-18 04:28:37 +00:00
|
|
|
i386_can_use_hw_breakpoint (struct target_ops *self,
|
|
|
|
int type, int cnt, int othertype)
|
2008-03-01 04:39:36 +00:00
|
|
|
{
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2009-05-14 09:37:00 +00:00
|
|
|
static void
|
|
|
|
add_show_debug_regs_command (void)
|
|
|
|
{
|
|
|
|
/* A maintenance command to enable printing the internal DRi mirror
|
|
|
|
variables. */
|
|
|
|
add_setshow_boolean_cmd ("show-debug-regs", class_maintenance,
|
2014-06-11 14:30:50 +00:00
|
|
|
&debug_hw_points, _("\
|
2009-05-14 09:37:00 +00:00
|
|
|
Set whether to show variables that mirror the x86 debug registers."), _("\
|
|
|
|
Show whether to show variables that mirror the x86 debug registers."), _("\
|
|
|
|
Use \"on\" to enable, \"off\" to disable.\n\
|
|
|
|
If enabled, the debug registers values are shown when GDB inserts\n\
|
|
|
|
or removes a hardware breakpoint or watchpoint, and when the inferior\n\
|
|
|
|
triggers a breakpoint or watchpoint."),
|
|
|
|
NULL,
|
|
|
|
NULL,
|
|
|
|
&maintenance_set_cmdlist,
|
|
|
|
&maintenance_show_cmdlist);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* There are only two global functions left. */
|
|
|
|
|
2008-03-01 04:39:36 +00:00
|
|
|
void
|
|
|
|
i386_use_watchpoints (struct target_ops *t)
|
|
|
|
{
|
|
|
|
/* After a watchpoint trap, the PC points to the instruction after the
|
|
|
|
one that caused the trap. Therefore we don't need to step over it.
|
|
|
|
But we do need to reset the status register to avoid another trap. */
|
|
|
|
t->to_have_continuable_watchpoint = 1;
|
|
|
|
|
|
|
|
t->to_can_use_hw_breakpoint = i386_can_use_hw_breakpoint;
|
|
|
|
t->to_region_ok_for_hw_watchpoint = i386_region_ok_for_watchpoint;
|
|
|
|
t->to_stopped_by_watchpoint = i386_stopped_by_watchpoint;
|
|
|
|
t->to_stopped_data_address = i386_stopped_data_address;
|
|
|
|
t->to_insert_watchpoint = i386_insert_watchpoint;
|
|
|
|
t->to_remove_watchpoint = i386_remove_watchpoint;
|
|
|
|
t->to_insert_hw_breakpoint = i386_insert_hw_breakpoint;
|
|
|
|
t->to_remove_hw_breakpoint = i386_remove_hw_breakpoint;
|
|
|
|
}
|
|
|
|
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
void
|
2009-05-14 09:37:00 +00:00
|
|
|
i386_set_debug_register_length (int len)
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
{
|
2009-05-14 09:37:00 +00:00
|
|
|
/* This function should be called only once for each native target. */
|
|
|
|
gdb_assert (i386_dr_low.debug_register_length == 0);
|
|
|
|
gdb_assert (len == 4 || len == 8);
|
|
|
|
i386_dr_low.debug_register_length = len;
|
|
|
|
add_show_debug_regs_command ();
|
Unified support for hardware breakpoints and watchpoints on
x86 targets:
* config/i386/nm-i386.h: New file.
* config/i386/nm-i386.h: (i386_cleanup_dregs)
(i386_insert_watchpoint, i386_remove_watchpoint)
(i386_region_ok_for_watchpoint, i386_stopped_by_hwbp)
(i386_stopped_data_address, i386_insert_hw_breakpoint)
(i386_remove_hw_breakpoint): Declare prototypes.
[I386_USE_GENERIC_WATCHPOINTS] (TARGET_CAN_USE_HARDWARE_WATCHPOINT):
Define if not already defined.
(TARGET_REGION_OK_FOR_HW_WATCHPOINT, HAVE_CONTINUABLE_WATCHPOINT)
(STOPPED_BY_WATCHPOINT, target_stopped_data_address)
(target_insert_watchpoint, target_remove_watchpoint)
(target_insert_hw_breakpoint, target_remove_hw_breakpoint): Define
to call the appropriate i386_* functions.
* i386-nat.c: New file.
* i386-nat.c (I386_DR_CONTROL_MASK, I386_DR_LOCAL_ENABLE)
(I386_DR_GLOBAL_ENABLE, I386_DR_DISABLE, I386_DR_SET_RW_LEN)
(I386_DR_GET_RW_LEN, I386_DR_WATCH_HIT): New macros.
(dr_mirror, dr_status_mirror, dr_control_mirror, dr_ref_count)
(maint_show_dr): New variables.
(i386_cleanup_dregs, i386_show_dr, i386_length_and_rw_bits)
(i386_insert_aligned_watchpoint, i386_remove_aligned_watchpoint)
(i386_handle_nonaligned_watchpoint, i386_insert_watchpoint)
(i386_remove_watchpoint, i386_region_ok_for_watchpoint)
(i386_stopped_data_address, i386_stopped_by_hwbp)
(i386_insert_hw_breakpoint, i386_remove_hw_breakpoint): New
functions.
(_initialize_i386_nat): New function.
[I386_USE_GENERIC_WATCHPOINTS]: Add new maint command
`show-debug-regs', sets maint_show_dr to non-zero value and
activates debugging print-outs in functions which insert, remove,
and test watchpoints and hardware breakpoints.
* Makefile.in (i386-nat.o): New target.
(ALLDEPFILES): Add i386-nat.o.
2001-03-21 11:36:58 +00:00
|
|
|
}
|