stepping is disturbed by setjmp/longjmp | try/catch in other threads
At https://sourceware.org/ml/gdb-patches/2015-08/msg00097.html, Joel
observed that trying to next/step a program on GNU/Linux sometimes
results in the following failed assertion:
% gdb -q .obj/gprof/main
(gdb) start
(gdb) n
(gdb) step
[...]/infrun.c:2391: internal-error:
resume: Assertion `sig != GDB_SIGNAL_0' failed.
What happened is that, during the "next" operation, GDB hit a
longjmp/exception/step-resume breakpoint but failed to see that this
breakpoint was set for a different thread than the one being stepped.
Joel's detailed analysis follows:
More precisely, at the end of the "start" command, we are stopped at
the start of function Main in main.adb; there are 4 threads in total,
and we are in the main thread (which is thread 1):
(gdb) info thread
Id Target Id Frame
4 Thread 0xb7a56ba0 (LWP 28379) 0xffffe410 in __kernel_vsyscall ()
3 Thread 0xb7c5aba0 (LWP 28378) 0xffffe410 in __kernel_vsyscall ()
2 Thread 0xb7e5eba0 (LWP 28377) 0xffffe410 in __kernel_vsyscall ()
* 1 Thread 0xb7ea18c0 (LWP 28370) main () at /[...]/main.adb:57
All the logs below reference Thread ID/LWP, but it'll be easier to
talk about the threads by GDB thread number. For instance, thread 1
is LWP 28370 while thread 3 is LWP 28378. So, the explanations below
translate the LWPs into thread numbers.
Back to what happens while we are trying to "next' our program:
(gdb) n
infrun: clear_proceed_status_thread (Thread 0xb7a56ba0 (LWP 28379))
infrun: clear_proceed_status_thread (Thread 0xb7c5aba0 (LWP 28378))
infrun: clear_proceed_status_thread (Thread 0xb7e5eba0 (LWP 28377))
infrun: clear_proceed_status_thread (Thread 0xb7ea18c0 (LWP 28370))
infrun: proceed (addr=0xffffffff, signal=GDB_SIGNAL_DEFAULT)
infrun: resume (step=1, signal=GDB_SIGNAL_0), trap_expected=0, current thread [Thread 0xb7ea18c0 (LWP 28370)] at 0x805451e
infrun: target_wait (-1.0.0, status) =
infrun: 28370.28370.0 [Thread 0xb7ea18c0 (LWP 28370)],
infrun: status->kind = stopped, signal = GDB_SIGNAL_TRAP
infrun: TARGET_WAITKIND_STOPPED
infrun: stop_pc = 0x8054523
We've resumed thread 1 (LWP 28370), and received in return a signal
that the same thread stopped slightly further. It's still in the
range of instructions for the line of source we started the "next"
from, as evidenced by the following trace...
infrun: stepping inside range [0x805451e-0x8054531]
... and thus, we decide to continue stepping the same thread:
infrun: resume (step=1, signal=GDB_SIGNAL_0), trap_expected=0, current thread [Thread 0xb7ea18c0 (LWP 28370)] at 0x8054523
infrun: prepare_to_wait
That's when we get an event from a different thread (thread 3)...
infrun: target_wait (-1.0.0, status) =
infrun: 28370.28378.0 [Thread 0xb7c5aba0 (LWP 28378)],
infrun: status->kind = stopped, signal = GDB_SIGNAL_TRAP
infrun: TARGET_WAITKIND_STOPPED
infrun: stop_pc = 0x80782d0
infrun: context switch
infrun: Switching context from Thread 0xb7ea18c0 (LWP 28370) to Thread 0xb7c5aba0 (LWP 28378)
... which we find to be at the address where we set a breakpoint on
"the unwinder debug hook" (namely "_Unwind_DebugHook"). But GDB fails
to notice that the breakpoint was inserted for thread 1 only, and so
decides to handle it as...
infrun: BPSTAT_WHAT_SET_LONGJMP_RESUME
... and inserts a breakpoint at the corresponding resume address, as
evidenced by this the next log:
infrun: exception resume at
80542a2
That breakpoint seems innocent right now, but will play a role fairly
quickly. But for now, GDB has inserted the exception-resume
breakpoint, and needs to single-step thread 3 past the breakpoint it
just hit. Thus, it temporarily disables the exception breakpoint, and
requests a step of that thread:
infrun: skipping breakpoint: stepping past insn at: 0x80782d0
infrun: skipping breakpoint: stepping past insn at: 0x80782d0
infrun: skipping breakpoint: stepping past insn at: 0x80782d0
infrun: resume (step=1, signal=GDB_SIGNAL_0), trap_expected=1, current thread [Thread 0xb7c5aba0 (LWP 28378)] at 0x80782d0
infrun: prepare_to_wait
We then get a notification, still from thread 3, that it's now past
that breakpoint...
infrun: prepare_to_wait
infrun: target_wait (-1.0.0, status) =
infrun: 28370.28378.0 [Thread 0xb7c5aba0 (LWP 28378)],
infrun: status->kind = stopped, signal = GDB_SIGNAL_TRAP
infrun: TARGET_WAITKIND_STOPPED
infrun: stop_pc = 0x8078424
... so we can resume what we were doing before, which is single-stepping
thread 1 until we get to a new line of code:
infrun: switching back to stepped thread
infrun: Switching context from Thread 0xb7c5aba0 (LWP 28378) to Thread 0xb7ea18c0 (LWP 28370)
infrun: expected thread still hasn't advanced
infrun: resume (step=1, signal=GDB_SIGNAL_0), trap_expected=0, current thread [Thread 0xb7ea18c0 (LWP 28370)] at 0x8054523
The "resume" log above shows that we're resuming thread 1 from where
we left off (0x8054523). We get one more stop at 0x8054529, which is
still inside our stepping range so we go again. That's when we get
the following event, from thread 3:
infrun: prepare_to_wait
infrun: target_wait (-1.0.0, status) =
infrun: 28370.28378.0 [Thread 0xb7c5aba0 (LWP 28378)],
infrun: status->kind = stopped, signal = GDB_SIGNAL_TRAP
infrun: TARGET_WAITKIND_STOPPED
infrun: stop_pc = 0x80542a2
Now the stop_pc address is interesting, because it's the address of
"exception resume" breakpoint...
infrun: context switch
infrun: Switching context from Thread 0xb7ea18c0 (LWP 28370) to Thread 0xb7c5aba0 (LWP 28378)
infrun: BPSTAT_WHAT_CLEAR_LONGJMP_RESUME
... and since that location is at a different line of code, this is
where it decides the "next" operation should stop:
infrun: stop_waiting
[Switching to Thread 0xb7c5aba0 (LWP 28378)]
0x080542a2 in inte_tache_rt.ttache_rt (
<_task>=0x80968ec <inte_tache_rt_inst.tache2>)
at /[...]/inte_tache_rt.adb:54
54 end loop;
However, what GDB should have noticed earlier that the exception
breakpoint we hit was for a different thread, thus should have
single-stepped that thread out of the breakpoint _without_ inserting
the exception-return breakpoint, and then resumed the single-stepping
of the initial thread (thread 1) until that thread stepped out of its
stepping range.
This is what this patch does, and after applying it, GDB now correctly
stops on the next line of code.
The patch adds a C++ test that exercises this, both for setjmp/longjmp
and exception breakpoints. With an unpatched GDB it shows:
(gdb) next
[Switching to Thread 22445.22455]
thread_try_catch (arg=0x0) at /home/pedro/gdb/mygit/build/../src/gdb/testsuite/gdb.threads/next-other-thr-longjmp.c:59
59 catch (...)
(gdb) FAIL: gdb.threads/next-other-thr-longjmp.exp: next to line 1
next
/home/pedro/gdb/mygit/build/../src/gdb/infrun.c:4865: internal-error: process_event_stop_test: Assertion `ecs->event_thread->control.exception_resume_breakpoint != NULL' fa
iled.
A problem internal to GDB has been detected,
further debugging may prove unreliable.
Quit this debugging session? (y or n) FAIL: gdb.threads/next-other-thr-longjmp.exp: next to line 2 (GDB internal error)
Resyncing due to internal error.
n
Tested on x86_64-linux, no regressions.
gdb/ChangeLog:
2015-08-05 Pedro Alves <palves@redhat.com>
Joel Brobecker <brobecker@adacore.com>
* breakpoint.c (bpstat_what) <bp_longjmp, bp_longjmp_call_dummy>
<bp_exception, bp_longjmp_resume, bp_exception_resume>: Handle the
case where BS->STOP is not set.
gdb/testsuite/ChangeLog:
2015-08-05 Pedro Alves <palves@redhat.com>
* gdb.threads/next-while-other-thread-longjmps.c: New file.
* gdb.threads/next-while-other-thread-longjmps.exp: New file.