This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH 1/4] __fdelt_chk: Removed range check
- From: KOSAKI Motohiro <kosaki dot motohiro at gmail dot com>
- To: Florian Weimer <fweimer at redhat dot com>
- Cc: libc-alpha <libc-alpha at sourceware dot org>, "libc-ports at sourceware dot org" <libc-ports at sourceware dot org>
- Date: Fri, 12 Apr 2013 16:28:07 -0400
- Subject: Re: [PATCH 1/4] __fdelt_chk: Removed range check
- References: <1365744803-19197-1-git-send-email-kosaki dot motohiro at gmail dot com> <1365744803-19197-2-git-send-email-kosaki dot motohiro at gmail dot com> <5167BADC dot 3090101 at redhat dot com>
On Fri, Apr 12, 2013 at 3:42 AM, Florian Weimer <fweimer@redhat.com> wrote:
> On 04/12/2013 07:33 AM, KOSAKI Motohiro wrote:
>>
>> +strong_alias (__fdelt_nochk, __fdelt_chk)
>> +strong_alias (__fdelt_nochk, __fdelt_warn)
>
>
> This change (which disables checking for existing compiled binaries) seems
> the wrong thing to do to me.
>
> I tend to agree that it might make sense to make fd_set fortification
> optional, but it should be enabled by default. Could you please change your
> patch so that it performs the checking by default, and preserves checking
> for applications which were compiled against pre-2.18 versions?
Could you clarify why you think so? Because I proposed exact same approach
at first and Carlos disagreed it because he pointed out an application
shouldn't
crash when post-2.18 application link with pre-2.18 libraries. Now I
agreed him and
then I changed my patch.
OK, I don't think it's impossible. However, I need to understand more.
At least Carlos
explained why and it seems reasonable request to me.
> By the way, if you see crashes with Qt, we have a patch which replaces
> select with poll (qt-4.8-poll.patch in Fedora). We tried to upstream it,
> but no luck so far.
Thanks. That's definitely forward step. Even though I don't think
select() is only used from Qt.