bfd09d203f
The existing logic was simply to flip syscall entry/return state when a syscall trap was seen, and even then only with active 'catch syscall'. That can get out of sync if 'catch syscall' is toggled at odd times. This patch updates the entry/return state for all syscall traps, regardless of catching state, and also updates known syscall state for other kinds of traps. Almost all PTRACE_EVENT stops are delivered from the middle of a syscall, so this can act like an entry. Every other kind of ptrace stop is only delivered outside of syscall event pairs, so marking them ignored ensures the next syscall trap looks like an entry. Three new test scenarios are added to catch-syscall.exp: - Disable 'catch syscall' from an entry to deliberately miss the return event, then re-enable to make sure a new entry is recognized. - Enable 'catch syscall' for the first time from a vfork event, which is a PTRACE_EVENT_VFORK in the middle of the syscall. Make sure the next syscall event is recognized as the return. - Make sure entry and return are recognized for an ENOSYS syscall. This is to defeat a common x86 hack that uses the pre-filled ENOSYS return value as a sign of being on the entry side. gdb/ChangeLog: 2015-10-19 Josh Stone <jistone@redhat.com> * linux-nat.c (linux_handle_syscall_trap): Always update entry/ return state, even when not actively catching syscalls at all. (linux_handle_extended_wait): Mark syscall_state like an entry. (wait_lwp): Set syscall_state ignored for other traps. (linux_nat_filter_event): Likewise. gdb/testsuite/ChangeLog: 2015-10-19 Josh Stone <jistone@redhat.com> * gdb.base/catch-syscall.c: Include <sched.h>. (unknown_syscall): New variable. (main): Trigger a vfork and an unknown syscall. * gdb.base/catch-syscall.exp (vfork_syscalls): New variable. (unknown_syscall_number): Likewise. (check_call_to_syscall): Accept an optional syscall pattern. (check_return_from_syscall): Likewise. (check_continue): Likewise. (test_catch_syscall_without_args): Check for vfork and ENOSYS. (test_catch_syscall_skipping_return): New test toggling off 'catch syscall' to step over the syscall return, then toggling back on. (test_catch_syscall_mid_vfork): New test turning on 'catch syscall' during a PTRACE_EVENT_VFORK stop, in the middle of a vfork syscall. (do_syscall_tests): Call test_catch_syscall_without_args and test_catch_syscall_mid_vfork. (test_catch_syscall_without_args_noxml): Check for vfork and ENOSYS. (fill_all_syscalls_numbers): Initialize unknown_syscall_number.
62 lines
1.6 KiB
C
62 lines
1.6 KiB
C
/* This file is used to test the 'catch syscall' feature on GDB.
|
|
|
|
Please, if you are going to edit this file DO NOT change the syscalls
|
|
being called (nor the order of them). If you really must do this, then
|
|
take a look at catch-syscall.exp and modify there too.
|
|
|
|
Written by Sergio Durigan Junior <sergiodj@linux.vnet.ibm.com>
|
|
September, 2008 */
|
|
|
|
#include <unistd.h>
|
|
#include <sys/syscall.h>
|
|
#include <fcntl.h>
|
|
#include <sys/stat.h>
|
|
#include <sched.h>
|
|
|
|
/* These are the syscalls numbers used by the test. */
|
|
|
|
int close_syscall = SYS_close;
|
|
int chroot_syscall = SYS_chroot;
|
|
/* GDB had a bug where it couldn't catch syscall number 0 (PR 16297).
|
|
In most GNU/Linux architectures, syscall number 0 is
|
|
restart_syscall, which can't be called from userspace. However,
|
|
the "read" syscall is zero on x86_64. */
|
|
int read_syscall = SYS_read;
|
|
#ifdef SYS_pipe
|
|
int pipe_syscall = SYS_pipe;
|
|
#else
|
|
int pipe2_syscall = SYS_pipe2;
|
|
#endif
|
|
int write_syscall = SYS_write;
|
|
int unknown_syscall = 123456789;
|
|
int exit_group_syscall = SYS_exit_group;
|
|
|
|
int
|
|
main (void)
|
|
{
|
|
int fd[2];
|
|
char buf1[2] = "a";
|
|
char buf2[2];
|
|
|
|
/* A close() with a wrong argument. We are only
|
|
interested in the syscall. */
|
|
close (-1);
|
|
|
|
chroot (".");
|
|
|
|
pipe (fd);
|
|
|
|
write (fd[1], buf1, sizeof (buf1));
|
|
read (fd[0], buf2, sizeof (buf2));
|
|
|
|
/* Test vfork-event interactions. Child exits immediately.
|
|
(Plain fork won't work on no-mmu kernel configurations.) */
|
|
if (vfork () == 0)
|
|
_exit (0);
|
|
|
|
/* Trigger an intentional ENOSYS. */
|
|
syscall (unknown_syscall);
|
|
|
|
/* The last syscall. Do not change this. */
|
|
_exit (0);
|
|
}
|