On 2020-07-06 3:02 p.m., Pedro Alves wrote:
> This patch series fixes PR 26199, a GDB 10 blocker.
> 
> I discussed how to fix this with Simon, and we came to the conclusion
> that we can fix it by removing code.  Easy.  :-) That's the last patch
> in the series.
> 
> Well, not so easy, actually... :-/
> 
> Doing that alone caused gdb.multi/multi-target.exp to regress.  And
> the reason was that the fix for PR 261299 made that testcase trip on a
> few latent bugs...  One of those bugs also caused a similar 100% cpu
> busy loop.
> 
> And then, while fixing those, I added a new test scenario to
> gdb.multi/multi-target.exp to exercise the TARGET_WAITKIND_NO_RESUMED
> handling fixes in this series.  That new test requires sending Ctrl-C
> to GDB after the test is done with, in order to cleanly kill gdbserver
> via "monitor exit".  But, that Ctrl-C didn't work, due to an issue
> with GDB's terminal handling, GDB would just hang...
> 
> That's all fixed by this series.
> 
> Pedro Alves (6):
>   Fix spurious unhandled remote %Stop notifications
>   Fix latent bug in target_pass_ctrlc
>   Avoid constant stream of TARGET_WAITKIND_NO_RESUMED
>   Fix handle_no_resumed w/ multiple targets
>   Make handle_no_resumed transfer terminal
>   Testcase for previous handle_no_resumed fixes
> 
> Simon Marchi (1):
>   Fix GDB busy loop when interrupting non-stop program (PR 26199)
> 
>  gdb/infrun.c                             | 116 +++++++++++++++++++++----------
>  gdb/remote.c                             |  15 +++-
>  gdb/target.c                             |   2 +-
>  gdb/testsuite/gdb.multi/multi-target.c   |   4 +-
>  gdb/testsuite/gdb.multi/multi-target.exp |  76 ++++++++++++++++++++
>  5 files changed, 173 insertions(+), 40 deletions(-)
> 
> 
> base-commit: ad8464f799a4c96c7ab8bdfec3f95846cf54f9b0
> -- 
> 2.14.5
> 

I am not sure why I didn't see this earlier, but gdb.multi/multi-target.exp
fails for me with my ASan-enabled build, using current master.  There is
a use after free.  Do you see the same?  I've attached the ASan log.

Simon