This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH] Remove support for "rtld_" prefix on solib-svr4 probes
- From: Pedro Alves <palves at redhat dot com>
- To: Sergio Durigan Junior <sergiodj at redhat dot com>
- Cc: GDB Patches <gdb-patches at sourceware dot org>, Gary Benson <gbenson at redhat dot com>
- Date: Thu, 25 Sep 2014 22:13:29 +0100
- Subject: Re: [PATCH] Remove support for "rtld_" prefix on solib-svr4 probes
- Authentication-results: sourceware.org; auth=none
- References: <1411581801-19126-1-git-send-email-sergiodj at redhat dot com> <5423F08B dot 1040409 at redhat dot com> <87bnq3h1gf dot fsf_-_ at redhat dot com>
On 09/25/2014 09:47 PM, Sergio Durigan Junior wrote:
> Thanks for the review.
>
> On Thursday, September 25 2014, Pedro Alves wrote:
>
>> Well, AFAICS, upstream GDB still supports F17's probes. See
>> svr4_create_solib_event_breakpoints:
>>
>> memset (probes, 0, sizeof (probes));
>> for (i = 0; i < NUM_PROBES; i++)
>> {
>> const char *name = probe_info[i].name;
>> struct probe *p;
>> char buf[32];
>>
>> /* Fedora 17 and Red Hat Enterprise Linux 6.2-6.4
>> shipped with an early version of the probes code in
>> which the probes' names were prefixed with "rtld_"
>> and the "map_failed" probe did not exist. The
>> locations of the probes are otherwise the same, so
>> we check for probes with prefixed names if probes
>> with unprefixed names are not present. */
>> if (with_prefix)
>> {
>> xsnprintf (buf, sizeof (buf), "rtld_%s", name);
>> name = buf;
>> }
>>
>> probes[i]
>
> Indeed it does, thanks for catching this.
>
>> So it seems to me the test should cope with both variants.
>
> Or maybe we should simplify this code and remove this support.
>
> Really, Fedora 17 was EOL'ed more than 1 year ago:
>
> <https://lists.fedoraproject.org/pipermail/announce/2013-July/003177.html>
>
> And we are already on Fedora 20, moving towards Fedora 21. Also, this
> code was needed because a patch present in Fedora 17's glibc, so I think
> it is fair to leave this to be handled by Fedora GDB if needed (but it
> won't be, because the upstream glibc patches already made into Fedora
> too).
There's RHEL (at least, per the comment) 6.4 too, which isn't EOL'ed,
though. It's reasonable to expect that people may still want to
build/test upstream gdb on those?
Thanks,
Pedro Alves