Re: [patch] Fix racy FAILs due to "read1" [Re: [PATCH] add -s option to make -break-insert support dprintf]

[Hui Zhu <teawater@gmail.com>]

On Tue, May 21, 2013 at 4:10 PM, Jan Kratochvil
<jan.kratochvil@redhat.com> wrote:
> On Tue, 21 May 2013 06:24:16 +0200, Hui Zhu wrote:
>> Checked in http://sourceware.org/ml/gdb-cvs/2013-05/msg00193.html
>
> It has random FAILs during normal run:
>
> +PASS: gdb.mi/mi-dprintf.exp: fprintf: mi 1st dprintf
> +FAIL: gdb.mi/mi-dprintf.exp: fprintf: mi 1st dprintf stop (timeout)
> +PASS: gdb.mi/mi-dprintf.exp: fprintf: mi 2nd dprintf
>
> and when I tested it with:
>         reproducer for races of expect incomplete reads
>         http://sourceware.org/bugzilla/show_bug.cgi?id=12649
>
> it FAILs reproducibly, could you review the proposed fix?

This patch fixed the issue in my part.

Thanks,
Hui

>
>
> Thanks,
> Jan
>
>
> gdb/testsuite/
> 2013-05-21  Jan Kratochvil  <jan.kratochvil@redhat.com>
>
>         PR testsuite/12649
>         * gdb.mi/mi-dprintf.exp (mi_continue_dprintf): Fix expect strings for
>         racy matches.
>
> diff --git a/gdb/testsuite/gdb.mi/mi-dprintf.exp b/gdb/testsuite/gdb.mi/mi-dprintf.exp
> index 457f332..ea8b3a8 100644
> --- a/gdb/testsuite/gdb.mi/mi-dprintf.exp
> +++ b/gdb/testsuite/gdb.mi/mi-dprintf.exp
> @@ -68,7 +68,7 @@ proc mi_continue_dprintf {args} {
>         mi_run_cmd
>         set msg "mi 1st dprintf"
>         gdb_expect {
> -            -re ".*At foo entry.*arg=1234, g=1234.*" {
> +            -re ".*At foo entry.*arg=1234, g=1234" {
>                 pass $msg
>             }
>             -re ".*$mi_gdb_prompt$" {
> @@ -83,7 +83,7 @@ proc mi_continue_dprintf {args} {
>         set msg "mi 2nd dprintf"
>         mi_send_resuming_command "exec-continue" "$msg continue"
>         gdb_expect {
> -            -re ".*At foo entry.*arg=1235, g=2222.*" {
> +            -re ".*At foo entry.*arg=1235, g=2222.*$mi_gdb_prompt$" {
>                 pass $msg
>             }
>             -re ".*$mi_gdb_prompt$" {