Monday Patch Queue Review update (2024-01-29)

Monday Patch Queue Review update (2024-01-29)

[Carlos O'Donell]

Most recent meeting status is always here:
https://sourceware.org/glibc/wiki/PatchworkReviewMeetings#Update

Meeting: 2024-01-29 @ 0900h EST5EDT

Video/Audio: https://bbb.linuxfoundation.org/room/adm-alk-1uu-7fu

IRC: #glibc on OFTC.

Review new patches and restart review at the top.

 * State NEW delegate NOBODY is at 457 patches.
 * Carlos's SLI at 269 days average patch age in queue and 124257 accumulated patch days.
 * Reviewing blockers for glibc 2.39
 * Carlos: I wanted to answer the question that Andreas asked on IRC which was: "Can we but the release branch early?" The answer is Yes, if we feel we're ready we can do that.
 * Andreas: Only the s390x issue that we discovered late with --disable-multi-arch.
 * Carlos: Yes, 2.39 will just go out the door with this quirk which is that you must have multiarch enabled. Using --disable-multi-arch will break s390x.
 * Advisory format updates and release process updates are a release blocker
  * Carlos: I need to review this today, and will review ~1pm EDT.
 * Carlos new added test results since the 22nd:
  * tst-align3 failures on 32-bit ARM.
  * Update https://sourceware.org/glibc/wiki/Testing/Tests/elf/tst-align3 with failure?
 * Andreas: Added loongarch failures. There are a lot of them.
 * Carlos: I wouldn't hold up the release for the failures seen on one specific Loongarch system e.g. gcc401.
 * https://portal.cfarm.net/tickets/ should be used for Compile Farm tickets.
 * The new failures for s390 FAIL: gmon/tst-gmon-pie, FAIL: gmon/tst-gmon-pie-gprof, are both gcc bugs.
 * Carlos: The 32-bit i686 failures on the old boxes look interesting.
 * Florian: Possible IFUNC selector bug.
 * Andreas: Will capture kernel dmesg to see which instruction causes the SIGILL.
 * Andreas: This is not a regression present since 2.38.
 * Ended meeting early. Carlos will review advisory discussion blocker.

-- 
Cheers,
Carlos.

Re: Monday Patch Queue Review update (2024-01-29)

[Xi Ruoyao]

On Mon, 2024-01-29 at 09:49 -0500, Carlos O'Donell wrote:
>  * Andreas: Added loongarch failures. There are a lot of them.
>  * Carlos: I wouldn't hold up the release for the failures seen on one
> specific Loongarch system e.g. gcc401.

These seem caused by the Debian-specific "multiarch" directory layout:

xry111@gcc401:~/git-repos/glibc-build$ make test t=misc/tst-gettid-kill
(... ...)
libgcc_s.so.1 must be installed for pthread_cancel to work
make[2]: Leaving directory '/home/xry111/git-repos/glibc/misc'
FAIL: misc/tst-gettid-kill
original exit status 1
Didn't expect signal from child: got `Aborted'

libgcc_s.so.1 is in the "multiarch directory" /usr/lib/loongarch64-
linux-gnu.  With --libdir=/usr/lib/loongarch64-linux-gnu and
libc_cv_slibdir=/usr/lib/loongarch64-linux-gnu this test passes.

Now running make check -j32...

-- 
Xi Ruoyao <xry111@xry111.site>
School of Aerospace Science and Technology, Xidian University

Re: Monday Patch Queue Review update (2024-01-29)

[Xi Ruoyao]

On Mon, 2024-01-29 at 23:54 +0800, Xi Ruoyao wrote:
> On Mon, 2024-01-29 at 09:49 -0500, Carlos O'Donell wrote:
> >  * Andreas: Added loongarch failures. There are a lot of them.
> >  * Carlos: I wouldn't hold up the release for the failures seen on one
> > specific Loongarch system e.g. gcc401.
> 
> These seem caused by the Debian-specific "multiarch" directory layout:
> 
> xry111@gcc401:~/git-repos/glibc-build$ make test t=misc/tst-gettid-kill
> (... ...)
> libgcc_s.so.1 must be installed for pthread_cancel to work
> make[2]: Leaving directory '/home/xry111/git-repos/glibc/misc'
> FAIL: misc/tst-gettid-kill
> original exit status 1
> Didn't expect signal from child: got `Aborted'
> 
> libgcc_s.so.1 is in the "multiarch directory" /usr/lib/loongarch64-
> linux-gnu.  With --libdir=/usr/lib/loongarch64-linux-gnu and
> libc_cv_slibdir=/usr/lib/loongarch64-linux-gnu this test passes.
> 
> Now running make check -j32...

Only one test failure with those:

FAIL: misc/tst-glibcsyscalls
original exit status 1
info: glibc syscall 'cachestat' not known to kernel
info: glibc syscall 'futex_requeue' not known to kernel
info: glibc syscall 'fchmodat2' not known to kernel
info: glibc syscall 'futex_wake' not known to kernel
info: glibc syscall 'map_shadow_stack' not known to kernel
info: glibc syscall 'futex_wait' not known to kernel
error: kernel syscall 'fstat' (80) not known to glibc
error: kernel syscall 'newfstatat' (79) not known to glibc
info: glibc tables are based on kernel version 6.7
info: installed kernel headers are version 6.4

I suppose there is something wrong with Debian-shipped kernel headers
(Debian on LoongArch is still WIP).

-- 
Xi Ruoyao <xry111@xry111.site>
School of Aerospace Science and Technology, Xidian University

Re: Monday Patch Queue Review update (2024-01-29)

[Andreas K. Huettel]

[-- Attachment #1: Type: text/plain, Size: 2194 bytes --]

Am Montag, 29. Januar 2024, 17:35:33 CET schrieb Xi Ruoyao:
> On Mon, 2024-01-29 at 23:54 +0800, Xi Ruoyao wrote:
> > On Mon, 2024-01-29 at 09:49 -0500, Carlos O'Donell wrote:
> > >  * Andreas: Added loongarch failures. There are a lot of them.
> > >  * Carlos: I wouldn't hold up the release for the failures seen on one
> > > specific Loongarch system e.g. gcc401.
> > 
> > These seem caused by the Debian-specific "multiarch" directory layout:
> > 
> > xry111@gcc401:~/git-repos/glibc-build$ make test t=misc/tst-gettid-kill
> > (... ...)
> > libgcc_s.so.1 must be installed for pthread_cancel to work
> > make[2]: Leaving directory '/home/xry111/git-repos/glibc/misc'
> > FAIL: misc/tst-gettid-kill
> > original exit status 1
> > Didn't expect signal from child: got `Aborted'
> > 
> > libgcc_s.so.1 is in the "multiarch directory" /usr/lib/loongarch64-
> > linux-gnu.  With --libdir=/usr/lib/loongarch64-linux-gnu and
> > libc_cv_slibdir=/usr/lib/loongarch64-linux-gnu this test passes.

Ah. I see.

> > 
> > Now running make check -j32...
> 
> Only one test failure with those:

OK thanks a lot, much better.

> FAIL: misc/tst-glibcsyscalls
> original exit status 1
> info: glibc syscall 'cachestat' not known to kernel
> info: glibc syscall 'futex_requeue' not known to kernel
> info: glibc syscall 'fchmodat2' not known to kernel
> info: glibc syscall 'futex_wake' not known to kernel
> info: glibc syscall 'map_shadow_stack' not known to kernel
> info: glibc syscall 'futex_wait' not known to kernel
> error: kernel syscall 'fstat' (80) not known to glibc
> error: kernel syscall 'newfstatat' (79) not known to glibc
> info: glibc tables are based on kernel version 6.7
> info: installed kernel headers are version 6.4
> 
> I suppose there is something wrong with Debian-shipped kernel headers
> (Debian on LoongArch is still WIP).

That's also a 6.1 kernel on the machine, so ancient for loong ...

> 
> 


-- 
PD Dr. Andreas K. Huettel
Institute for Experimental and Applied Physics
University of Regensburg
93040 Regensburg
Germany

tel. +49 151 241 67748 (mobile)
tel. +49 941 943 1618 (office)
e-mail andreas.huettel@ur.de
https://www.akhuettel.de/
https://www.akhuettel.de/group/

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 981 bytes --]

Re: Monday Patch Queue Review update (2024-01-29)

[Xi Ruoyao]

On Mon, 2024-01-29 at 17:41 +0100, Andreas K. Huettel wrote:
> Am Montag, 29. Januar 2024, 17:35:33 CET schrieb Xi Ruoyao:
> > On Mon, 2024-01-29 at 23:54 +0800, Xi Ruoyao wrote:
> > > On Mon, 2024-01-29 at 09:49 -0500, Carlos O'Donell wrote:
> > > >  * Andreas: Added loongarch failures. There are a lot of them.
> > > >  * Carlos: I wouldn't hold up the release for the failures seen on one
> > > > specific Loongarch system e.g. gcc401.
> > > 
> > > These seem caused by the Debian-specific "multiarch" directory layout:
> > > 
> > > xry111@gcc401:~/git-repos/glibc-build$ make test t=misc/tst-gettid-kill
> > > (... ...)
> > > libgcc_s.so.1 must be installed for pthread_cancel to work
> > > make[2]: Leaving directory '/home/xry111/git-repos/glibc/misc'
> > > FAIL: misc/tst-gettid-kill
> > > original exit status 1
> > > Didn't expect signal from child: got `Aborted'
> > > 
> > > libgcc_s.so.1 is in the "multiarch directory" /usr/lib/loongarch64-
> > > linux-gnu.  With --libdir=/usr/lib/loongarch64-linux-gnu and
> > > libc_cv_slibdir=/usr/lib/loongarch64-linux-gnu this test passes.
> 
> Ah. I see.
> 
> > > 
> > > Now running make check -j32...
> > 
> > Only one test failure with those:
> 
> OK thanks a lot, much better.
> 
> > FAIL: misc/tst-glibcsyscalls
> > original exit status 1
> > info: glibc syscall 'cachestat' not known to kernel
> > info: glibc syscall 'futex_requeue' not known to kernel
> > info: glibc syscall 'fchmodat2' not known to kernel
> > info: glibc syscall 'futex_wake' not known to kernel
> > info: glibc syscall 'map_shadow_stack' not known to kernel
> > info: glibc syscall 'futex_wait' not known to kernel
> > error: kernel syscall 'fstat' (80) not known to glibc
> > error: kernel syscall 'newfstatat' (79) not known to glibc
> > info: glibc tables are based on kernel version 6.7
> > info: installed kernel headers are version 6.4
> > 
> > I suppose there is something wrong with Debian-shipped kernel headers
> > (Debian on LoongArch is still WIP).
> 
> That's also a 6.1 kernel on the machine, so ancient for loong ...

All passed after I added --with-headers=$HOME/linux-6.7.2/usr
($HOME/linux-6.7.2 is extracted from linux-6.7.2.tar.xz and I've ran
"make headers" in it).

-- 
Xi Ruoyao <xry111@xry111.site>
School of Aerospace Science and Technology, Xidian University