[Bug remote/20396] New: async_remote_interrupt_twice firing too quickly

["sstallion at gmail dot com" <sourceware-bugzilla at sourceware dot org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=20396

            Bug ID: 20396
           Summary: async_remote_interrupt_twice firing too quickly
           Product: gdb
           Version: 7.10.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: remote
          Assignee: unassigned at sourceware dot org
          Reporter: sstallion at gmail dot com
  Target Milestone: ---

Recent support for the ^C^C interrupt sequence in GDB using a remote target is
broken. From the best I've been able to tell, a single ^C sequence results in
two SIGINT signals being delivered back to back, which causes GDB to issue a
diagnostic message that the target is no longer responding. Reverting back to
the previous behavior of not installing a signal handler for
async_remote_interrupt_twice fixes the problem, however I am at a loss as to
why two signals are being delivered for a single interrupt event.

I've verified that the remote target is receiving the interrupt packet and is
responding with a correct stop message.

Console session as noted below:

% gdb ...
Breakpoint 1, main () at main.c:16
16          for (;;);
(gdb) c
Continuing.
^CThe target is not responding to interrupt requests.
Stop debugging it? (y or n) n
[New Thread 2]
[New Thread 3]
...

This is a non-critical bug and has been reported elsewhere, most notably in
Launchpad for the ARM GCC distribution:
https://bugs.launchpad.net/gcc-arm-embedded/+bug/1594341

I've duplicated the problem using OpenOCD master, and GDB 7.10.1 and 7.11.1
built from source on OS X. The above mentioned bug report appears to have been
reported on linux. Using the same setup with version 7.8.1 (before the change
in signal handling) yielded no problems.

-- 
You are receiving this mail because:
You are on the CC list for the bug.