[patch] Fix target-async SIGTTOU stop (PR 12260)

[patch] Fix target-async SIGTTOU stop (PR 12260)

[Jan Kratochvil]

Hi,

it is difficult to play with `set target-async on' as it usually SIGTTOU stops
at various places.
http://sourceware.org/bugzilla/show_bug.cgi?id=12260

$ ./gdb -nx -ex 'set target-async on' -ex start ./gdb
[...]
Starting program: .../gdb/gdb 
[1]+  Stopped                 ./gdb -nx -ex 'set target-async on' -ex start ./gdb

If it has a regression it IMO only means there is missing
target_terminal_inferior call at some other place.

No regressions on {x86_64,x86_64-m32,i686}-fedora16pre-linux-gnu.  But I was
unable to reproduce the problem under DejaGnu so the regression test may not
be meaningful.

Not going to check it in without a review.


Thanks,
Jan


gdb/
2011-08-23  Jan Kratochvil  <jan.kratochvil@redhat.com>

	PR 12260
	* event-top.c (cli_command_loop): Call target_terminal_ours.

--- a/gdb/event-top.c
+++ b/gdb/event-top.c
@@ -193,6 +193,8 @@ cli_command_loop (void)
       char *a_prompt;
       char *gdb_prompt = get_prompt (0);
 
+      target_terminal_ours ();
+
       /* Tell readline what the prompt to display is and what function
          it will need to call after a whole line is read.  This also
          displays the first prompt.  */

Re: [patch] Fix target-async SIGTTOU stop (PR 12260)

[Matt Rice]

On Tue, Aug 23, 2011 at 12:31 PM, Jan Kratochvil
<jan.kratochvil@redhat.com> wrote:
> Hi,
>
> it is difficult to play with `set target-async on' as it usually SIGTTOU stops
> at various places.
> http://sourceware.org/bugzilla/show_bug.cgi?id=12260
>
> $ ./gdb -nx -ex 'set target-async on' -ex start ./gdb
> [...]
> Starting program: .../gdb/gdb
> [1]+  Stopped                 ./gdb -nx -ex 'set target-async on' -ex start ./gdb
>
> If it has a regression it IMO only means there is missing
> target_terminal_inferior call at some other place.
>
> No regressions on {x86_64,x86_64-m32,i686}-fedora16pre-linux-gnu.  But I was
> unable to reproduce the problem under DejaGnu so the regression test may not
> be meaningful.
>
> Not going to check it in without a review.
>

Not really a review, just thought i'd mention this is also fixed by
the following patch
http://sourceware.org/ml/gdb-patches/2011-08/msg00235.html

this also fixes the original report of PR 10720 which seems to be a
dupe, not sure about the latter reports in 10720.
the test cases i added do not cover this either.

to quote from elsewhere in that thread
matt> more information on PR 10720, which I'd had random difficulties
matt> reproducing, I managed to figure out why, the process being attached
matt> to must be on the same tty as the gdb process to reproduce it.
matt> Haven't been able to reproduce that in the testsuite though.

my guess is that deja-gnu doesn't meet this requirement to be able to
reproduce it.

Re: [patch] Fix target-async SIGTTOU stop (PR 12260)

[Pedro Alves]

On Wednesday 24 August 2011 06:24:36, Matt Rice wrote:
> On Tue, Aug 23, 2011 at 12:31 PM, Jan Kratochvil
> <jan.kratochvil@redhat.com> wrote:
> > Hi,
> >
> > it is difficult to play with `set target-async on' as it usually SIGTTOU stops
> > at various places.
> > http://sourceware.org/bugzilla/show_bug.cgi?id=12260
> >
> > $ ./gdb -nx -ex 'set target-async on' -ex start ./gdb
> > [...]
> > Starting program: .../gdb/gdb
> > [1]+  Stopped                 ./gdb -nx -ex 'set target-async on' -ex start ./gdb
> >
> > If it has a regression it IMO only means there is missing
> > target_terminal_inferior call at some other place.
> >
> > No regressions on {x86_64,x86_64-m32,i686}-fedora16pre-linux-gnu.  But I was
> > unable to reproduce the problem under DejaGnu so the regression test may not
> > be meaningful.
> >
> > Not going to check it in without a review.
> >
> 
> Not really a review, just thought i'd mention this is also fixed by
> the following patch
> http://sourceware.org/ml/gdb-patches/2011-08/msg00235.html

I was going to look at Jan's patch, and I'm a bit confused on the state
of things, and the state of Matt's patch --- there was a follow up, which
had a problem.  Is the msg00235.html version the final one instead?

> this also fixes the original report of PR 10720 which seems to be a
> dupe, not sure about the latter reports in 10720.
> the test cases i added do not cover this either.

By "this" you mean Jan's or yours?

> 
> to quote from elsewhere in that thread
> matt> more information on PR 10720, which I'd had random difficulties
> matt> reproducing, I managed to figure out why, the process being attached
> matt> to must be on the same tty as the gdb process to reproduce it.
> matt> Haven't been able to reproduce that in the testsuite though.
> 
> my guess is that deja-gnu doesn't meet this requirement to be able to
> reproduce it.
> 

-- 
Pedro Alves

Re: [patch] Fix target-async SIGTTOU stop (PR 12260)

[Matt Rice]

On Mon, Aug 29, 2011 at 8:17 AM, Pedro Alves <pedro@codesourcery.com> wrote:
> On Wednesday 24 August 2011 06:24:36, Matt Rice wrote:
>> On Tue, Aug 23, 2011 at 12:31 PM, Jan Kratochvil
>> <jan.kratochvil@redhat.com> wrote:
>> > Hi,
>> >
>> > it is difficult to play with `set target-async on' as it usually SIGTTOU stops
>> > at various places.
>> > http://sourceware.org/bugzilla/show_bug.cgi?id=12260
>> >
>> > $ ./gdb -nx -ex 'set target-async on' -ex start ./gdb
>> > [...]
>> > Starting program: .../gdb/gdb
>> > [1]+  Stopped                 ./gdb -nx -ex 'set target-async on' -ex start ./gdb
>> >
>> > If it has a regression it IMO only means there is missing
>> > target_terminal_inferior call at some other place.
>> >
>> > No regressions on {x86_64,x86_64-m32,i686}-fedora16pre-linux-gnu.  But I was
>> > unable to reproduce the problem under DejaGnu so the regression test may not
>> > be meaningful.
>> >
>> > Not going to check it in without a review.
>> >
>>
>> Not really a review, just thought i'd mention this is also fixed by
>> the following patch
>> http://sourceware.org/ml/gdb-patches/2011-08/msg00235.html
>
> I was going to look at Jan's patch, and I'm a bit confused on the state
> of things, and the state of Matt's patch --- there was a follow up, which
> had a problem.  Is the msg00235.html version the final one instead?

Yes, I apologize for the confusion. the follow up was intended to be
functionally equivalent
to msg00235, but getting rid of the weird empty prompt case.
Turns out the weird empty prompt case is required, A follow up to
00235 without simply
adding comments about the empty prompt case would probably have been prudent.

>> this also fixes the original report of PR 10720 which seems to be a
>> dupe, not sure about the latter reports in 10720.
>> the test cases i added do not cover this either.
>
> By "this" you mean Jan's or yours?

either really.

cancel/resolved: [patch] Fix target-async SIGTTOU stop (PR 12260)

[Jan Kratochvil]

On Tue, 23 Aug 2011 21:31:56 +0200, Jan Kratochvil wrote:
> $ ./gdb -nx -ex 'set target-async on' -ex start ./gdb
> [...]
> Starting program: .../gdb/gdb 
> [1]+  Stopped                 ./gdb -nx -ex 'set target-async on' -ex start ./gdb

This got fixed by Matt Rice:
	http://sourceware.org/bugzilla/show_bug.cgi?id=10720


Regards,
Jan