Elena Zannoni <ezannoni@kwikemart.cygnus.com>,
David Taylor <taylor@texas.cygnus.com>, and
Edith Epstein <eepstein@sophia.cygnus.com> as part of a project to
merge in changes by HP; HP did not create ChangeLog entries.
* dis-buf.c (generic_strcat_address): new function.
* hppa-dis.c: Changes to improve hppa disassembly.
Changed formatting in : reg_names, fp_reg_names,control_reg,
New variables : sign_extension_names, deposit_names, conversion_names
float_test_names, compare_cond_names_double, add_cond_names_double,
logical_cond_names_double, unit_cond_names_double,
branch_push_pop_names, saturation_names, shift_names, mix_names,
New Macros : GET_COMPL_O, GET_PUSH_POP,MERGED_REG
Move some definitions to libhppa.h: GET_FIELD, GET_BIT
(fput_const): renamed as fput_hex_const
(print_insn_hppa):
- use the macros fputs_filtered and
fput_decimal_const whenever possible; calls to sign_extend require
2 params -- add a missing second param of 0.
- Some new code ifdefed for LOCAL_ONLY, all related to figuring out
architecture version number of current machine. HP folks are
trying to handle situation where the target program was compiled
for PA 1.x (32-bit), but is running on a PA 2.0 machine and
visa versa.
- added new cases : 'g', 'B', 'm'
- added cases specifically for PA 2.0
- changed the following cases : '"', 'n', 'N', 'p', 'Z',
- calls to fput_const become calls to fput_hex_const
Zannoni.
* hpux-core.c (hpux_core_core_file_p): this function now
understands HPUX 10.30 thread info. Yes, the thread stacks can be
found in the corefile! Also, in hpux_core_core_file_p, if we
encounter a section with an unknown type, don't punt. Instead,
just skip it. Also, count the number of sections of known type
that we encounter. If we encounter at least one good one, then
we'll declare that the file is a core file. If we encounter any
unknown ones but some known ones, then we'll issue a warning (but
still declare it to be a core file). Also, correctly decide when a
file is not a core.
(PROC_INFO_HAS_THREAD_ID): define if CORE_ANON_SHMEM defined.
(CORE_ANON_SHMEM): define if not defined.
(hpux_core_struct): add members lwpid and user_tid.
(core_kernel_thread_id, core_user_thread_id): new macros.
(make_bfd_asection): use bfd_alloc to allocate room for a copy of
the name before storing it in the bfd.
(hpux_core_core_file_p): handle threads.
* libhppa.h (GET_FIELD): protect against redefinition.
(HPPA_WIDE): define.
(GET_FIELD): define.
(GET_BIT): define.
(MASK): define.
(CATENATE): define.
(ELEVEN): define.
(sign_extend): redefine.
(assemble_6): define.
(assemble_12): rewrite.
(assemble_16): define.
(assemble_16a): define.
(assemble_17): rewrite.
(assemble_22): define.
* config/i386/xm-cygwin.h: Remove REQUEST_QUIT definition.
* config/powerpc/xm-cygwin.h: Ditto.
PR 17664.
With this define in place, you can't type "q" to get out of GDB's pager
under Cygwin.
the stack frame. Their offset from the previous stack frame is in
fdata.gpr_offset and fdata.fpr_offset, not fdata.offset.
(gdb.base/return.exp)
* config/rs6000/tm-rs6000.h: Doc fixes.
(objdump_sprintf): Merge both versions into one. Increase buffer
size as needed to avoid overflow.
(disassemble_bytes): Change buf from 1000 bytes to 50. Change
initialization and use of sfile to match changes to
objdump_sprintf.
* configure: Regenerate.
* sim-main.h: Protect against multiple inclusion.
Don't include cgen-scache.h,cgen-cpu.h,cgen-trace.h,cpuall.h.
Done by cgen-sim.h now.
* tconfig.in (SIM_HAVE_MODEL): Delete, moved to cgen-types.h.
* cpuall.h: Regenerate.
* cpu.h,decode.c,sem-switch.c,sem.c: Regenerate.
* mloop.in (extract16): Make static inline again.
Simplify with call to @cpu@_fill_argbuf,@cpu@_fill_argbuf_tp.
(extract32): Ditto.
Simplify with call to @cpu@_fill_argbuf,@cpu@_fill_argbuf_tp.
(execute): Test ARGBUF_PROFILE_P before profiling.
Update calls to TRACE_INSN_INIT,TRACE_INSN_FINI.
* cpux.h,decodex.c,modelx.c,semx-switch.c: Regenerate.
* mloopx.in: Rewrite.
* cgen-sim.h: Simple header that includes others.
* sim-arange.c: New file.
* sim-arange.h: New file.
* sim-basics.h: Include it.
* Make-common.in (SIM_NEW_COMMON_OBJS): Add sim-arange.o.
(sim-arange.o): Add rule for.
* sim-cpu.h (sim_cpu_msg_prefix): Add prototype.
(sim_io_eprintf_cpu): Add prototype.
* sim-inline.h (HAVE_INLINE): Define if GNUC.
(INLINE2): New macro.
(EXTERN_INLINE): New macro.
* sim-module.c (sim_post_argv_init): Initialize cpu backlink
before calling module init fns.
* sim-profile.h (OPTION_PROFILE_*): Move into enum.
(profile_init): New function.
(profile_options): New option --profile-range.
(profile_option_handler): Handle --profile-range.
(profile_print_insn): Qualify address range specific section titles.
(profile_print_addr_ranges): New function.
(profile_info): Print address ranges if specified.
(profile_install): Set profile_init init fn.
* sim-profile.h (PROFILE_DATA): New member `range'.
* sim-trace.c (trace_init): New function.
(trace_options): New option --trace-range.
(trace_option_handler): Handle --trace-range.
(trace_install): Set trace_init init fn.
* sim-trace.h (TRACE_DATA): New member `range'.
* sim-utils.c (sim_cpu_msg_prefix): New function.
(sim_io_eprintf_cpu): New function.
* cgen-engine.h (PC_IN_TRACE_RANGE_P): New macro.
(PC_IN_PROFILE_RANGE_P): New macro.
* cgen-trace.c (trace_insn_init): Set current_insn to NULL.
(trace_insn_fini): New arg abuf. All callers updated.
Exit early if trace_insn not called. Check ARGBUF_PROFILE_P before
printing cycle counts.
* cgen-trace.h (trace_insn_fini): Update prototype.
(TRACE_RESULT_P): New macro.
(TRACE_INSN_INIT,TRACE_INSN_FINI): New arg abuf. All callers updated.
(TRACE_INSN): Check ARGBUF_TRACE_P.
(TRACE_EXTRACT,TRACE_RESULT): New arg abuf. All callers updated.
* cgen-types.h (SIM_INLINE): Delete.
(SIM_HAVE_MODEL,SIM_HAVE_ADDR_RANGE): Define.
* cgen-utils.c: Don't include cgen-engine.h
* genmloop.sh (@cpu@_fill_argbuf): New function.
(@cpu@_fill_argbuf_tp): New function.
(@cpu@_emit_before,@cpu@_emit_after): New functions.
(@cpu@_pbb_begin): Prefix cti_sc,insn_count with '_'.
(SET_CTI_VPC,SET_INSN_COUNT): Update.
(@cpu@_pbb_before): Check ARGBUF_PROFILE_P before calling
doing profiling. Update call to TRACE_INSN_INIT,TRACE_INSN_FINI.
(@cpu@_pbb_after): Check ARGBUF_PROFILE_P before calling
doing profiling. Update call to TRACE_INSN_FINI.
* monitor.c (monitor_read_memory): Zero out pattern buffers
before calling re_search.
(parse_register_dump): Ditto.
PR 18049. This bug had existed erratically since I upgraded to
the new gnu-regex.c this last summer. The problem is mostly in
parse_register_dump; the allocated structure has some random values
in it and there is a flag set in the register_pattern structure by the
gnu-regex library which indicates that the values in the re_registers
should be trusted.
If those arbitrary contents aren't zero, gnu-regex tries to run realloc
on them and we get a core dump on some hosts for some targets when the
moon is just right.
