Re: [patch] [7.6.1] Fix argv[0] symlink regression (PR 15415)

[Yufeng Zhang <Yufeng.Zhang@arm.com>]

On 09/04/13 17:53, Jan Kratochvil wrote:
> On Wed, 04 Sep 2013 18:46:07 +0200, Yufeng Zhang wrote:
>> On 08/26/13 19:21, Jan Kratochvil wrote:
>>> gdb/testsuite/
>>> 2013-08-26  Jan Kratochvil<jan.kratochvil@redhat.com>
>>>
>>>          PR gdb/15415
>>>          * gdb.base/argv0-symlink.c: New file.
>>>          * gdb.base/argv0-symlink.exp: New file.
>>
>> I wonder if the tests shall be skipped in the remote environment
>> where gdb has no control over argv[0].
>
> The testcase runs gdbserver (even in gdbserver mode) so it has control over
> argv[0].  The testcase PASSes for me both with gdbserver and with gdbserver in
> extended mode running on localhost (Fedora Rawhide x86_64).
>
> If one runs gdbserver on remote host with different filesystem I believe the
> testcase still should work as I test there only the latest filename component
> (and latest directory component).
>
> Could you post your FAIL gdb.log or do you have just theoretical objectives?

I am running the test on a simulator implementing the gdb remote stub. 
  I think it is related with the way the simulator is spawned; it seems 
like only the file name 'argv0-symlink-filelink' is passed to the simulator:

spawn <simulator listening on 9784> argv0-symlink-filelink
target remote localhost:9784
Remote debugging using localhost:9784
0x0000000000400180 in _start ()
(gdb) continue
Continuing.

Breakpoint 1, main (argc=1, argv=0x20414fe0) at 
/work/src/binutils/gdb/testsuite/gdb.base/argv0-symlink.c:21
21        return 0;
(gdb) print argv[0]
$1 = 0x412fa8 "argv0-symlink-filelink"
(gdb) FAIL: gdb.base/argv0-symlink.exp: kept file symbolic link name


I'll check my test config.

Thanks,
Yufeng