|
From: Nicholas N. <nj...@ca...> - 2004-02-22 17:23:00
|
Hi, It looks like upcoming glibc 2.3.3 will use a different version of pthreadtypes.h. This caused 69616 for Gentoo users who get a cutting-edge version, which breaks Valgrind. Note that this bug has 165 votes; the next highest one has 28 votes. I wrote a patch that seems to fix the problem. Basically, we were using the LinuxThreads pthread structures, so when glibc changed the structures, things broke. I've just introduced a level of indirection -- the patch uses the LinuxThreads structure layouts, but gives them a different name, and all the original types get cast to these differently-named types before doing anything. But I'm not totally certain about it. For example, for some structure fields I've used slightly different types to the LinuxThread structures. And I'm not very familiar with the pthreads code in general. I'd appreciate someone else looking at the patch. Thanks. N |
|
From: Nicholas N. <nj...@ca...> - 2004-02-23 14:08:44
|
On Sun, 22 Feb 2004, Nicholas Nethercote wrote: > > It looks like upcoming glibc 2.3.3 will use a different version of > pthreadtypes.h. This caused 69616 for Gentoo users who get a cutting-edge > version, which breaks Valgrind. Note that this bug has 165 votes; the > next highest one has 28 votes. > > I wrote a patch that seems to fix the problem. Basically, we were using > [...] > I'd appreciate someone else looking at the patch. Thanks. I didn't make it clear that the patch is in Bugzilla, see bugs.kde.org/attachment.cgi?id=4618&action=view N |
|
From: Julian S. <js...@ac...> - 2004-02-28 12:34:55
|
My $0.02 is to commit it. If it turns out to be a Patch of Mass Destruction, the nightly scripts will soon tell us, and we can back it out. J On Monday 23 February 2004 14:00, Nicholas Nethercote wrote: > On Sun, 22 Feb 2004, Nicholas Nethercote wrote: > > It looks like upcoming glibc 2.3.3 will use a different version of > > pthreadtypes.h. This caused 69616 for Gentoo users who get a > > cutting-edge version, which breaks Valgrind. Note that this bug has 165 > > votes; the next highest one has 28 votes. > > > > I wrote a patch that seems to fix the problem. Basically, we were using > > [...] > > I'd appreciate someone else looking at the patch. Thanks. > > I didn't make it clear that the patch is in Bugzilla, see > bugs.kde.org/attachment.cgi?id=4618&action=view > > N > > > ------------------------------------------------------- > SF.Net is sponsored by: Speed Start Your Linux Apps Now. > Build and deploy apps & Web services for Linux with > a free DVD software kit from IBM. Click Now! > http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click > _______________________________________________ > Valgrind-developers mailing list > Val...@li... > https://lists.sourceforge.net/lists/listinfo/valgrind-developers |
|
From: Tom H. <th...@cy...> - 2004-02-23 14:59:02
|
In message <Pin...@re...>
Nicholas Nethercote <nj...@ca...> wrote:
> It looks like upcoming glibc 2.3.3 will use a different version of
> pthreadtypes.h. This caused 69616 for Gentoo users who get a cutting-edge
> version, which breaks Valgrind. Note that this bug has 165 votes; the
> next highest one has 28 votes.
>
> I wrote a patch that seems to fix the problem. Basically, we were using
> the LinuxThreads pthread structures, so when glibc changed the structures,
> things broke. I've just introduced a level of indirection -- the patch
> uses the LinuxThreads structure layouts, but gives them a different name,
> and all the original types get cast to these differently-named types
> before doing anything.
>
> But I'm not totally certain about it. For example, for some structure
> fields I've used slightly different types to the LinuxThread structures.
> And I'm not very familiar with the pthreads code in general. I'd
> appreciate someone else looking at the patch. Thanks.
I'm not hugely familiar with LinuxThreads or NPTL but the changes look
right to me, and the structures all seem to match sufficiently to work
correctly.
Tom
--
Tom Hughes (th...@cy...)
Software Engineer, Cyberscience Corporation
http://www.cyberscience.com/
|