[Bug cli/17300] crash in non-stop mode with continue -a & (readline_callback_read_char() called with no handler!)

["palves at redhat dot com" <sourceware-bugzilla@sourceware.org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=17300

--- Comment #14 from Pedro Alves <palves at redhat dot com> ---
> There is no point to repeat "continue -a&" on subsequent <RET> when the 
> program is already running.

The command is repeated, but given it resumes all threads that are stopped, it
has no effect if all threads are already running.  It repeats in the sense that
if you have:

 (gdb) continue -a&
 (gdb) <RET>

(nothing happens)

 * some thread stops *
 (gdb) <RET>

Then this last <RET> repeats the c -a& and resumes the thread that stopped.

This is _no_ different from explicitly issuing the command rather than
just typing <RET>:

 (gdb) continue -a&
 (gdb) continue -a&

(nothing happens)

 * some thread stops *
 (gdb) continue -a&

Then this last c -a& resumes the thread that stopped.

And in turn, is different from commands that are documented to truly not
repeat, like, e.g., attach, detach, display, signal, target etc.  E.g.:

 (gdb) attach 1
 Attaching to process 1
 ptrace: Operation not permitted.
 (gdb) <RET>
 (gdb) <RET>
 (gdb) <RET>
 (gdb) <RET>

Those RET's don't really execute any command (related, I think GDB should say
something like "no command to repeat", or some such, rather than be quiet, for
truly non-repeat commands).

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