Re: [PATCH 6.12 000/262] 6.12.53-rc1 review
From: Ilya Leoshkevich
Date: Tue Oct 14 2025 - 10:46:16 EST
On Tue, 2025-10-14 at 19:38 +0530, Naresh Kamboju wrote:
> On Tue, 14 Oct 2025 at 16:56, Naresh Kamboju
> <naresh.kamboju@xxxxxxxxxx> wrote:
> >
> > On Mon, 13 Oct 2025 at 20:38, Greg Kroah-Hartman
> > <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
> > >
> > > This is the start of the stable review cycle for the 6.12.53
> > > release.
> > > There are 262 patches in this series, all will be posted as a
> > > response
> > > to this one. If anyone has any issues with these being applied,
> > > please
> > > let me know.
> > >
> > > Responses should be made by Wed, 15 Oct 2025 14:42:41 +0000.
> > > Anything received after that time might be too late.
> > >
> > > The whole patch series can be found in one patch at:
> > >
> > > https://www.kernel.org/pub/linux/kernel/v6.x/stable-review/patch-6.12.53-rc1.gz
> > > or in the git tree and branch at:
> > >
> > > git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-
> > > stable-rc.git linux-6.12.y
> > > and the diffstat can be found below.
> > >
> > > thanks,
> > >
> > > greg k-h
> >
> > The S390 defconfig builds failed on the Linux stable-rc 6.12.53-rc1
> > and 6.6.112-rc1 tag build due to following build warnings / errors
> > with gcc and clang toolchains.
> >
> > Also seen on 6.6.112-rc1.
> >
> > * s390, build
> > - clang-21-defconfig
> > - clang-nightly-defconfig
> > - clang-nightly-lkftconfig-hardening
> > - clang-nightly-lkftconfig-lto-full
> > - clang-nightly-lkftconfig-lto-thing
> > - gcc-14-allmodconfig
> > - gcc-14-defconfig
> > - gcc-14-lkftconfig-hardening
> > - gcc-8-defconfig-fe40093d
> > - gcc-8-lkftconfig-hardening
> > - korg-clang-21-lkftconfig-hardening
> > - korg-clang-21-lkftconfig-lto-full
> > - korg-clang-21-lkftconfig-lto-thing
> >
> > First seen on 6.12.53-rc1
> > Good: v6.12.52
> > Bad: 6.12.53-rc1 also seen on 6.6.112-rc1
> >
> > Regression Analysis:
> > - New regression? yes
> > - Reproducibility? yes
> >
> > Build regressions: arch/s390/net/bpf_jit_comp.c:1813:49: error:
> > 'struct bpf_jit' has no member named 'frame_off'
> >
> > Reported-by: Linux Kernel Functional Testing <lkft@xxxxxxxxxx>
> >
> > # Build error
> > arch/s390/net/bpf_jit_comp.c: In function 'bpf_jit_insn':
> > arch/s390/net/bpf_jit_comp.c:1813:49: error: 'struct bpf_jit' has
> > no
> > member named 'frame_off'
> > 1813 | _EMIT6(0xd203f000 | (jit->frame_off
> > +
> > | ^~
> > arch/s390/net/bpf_jit_comp.c:211:55: note: in definition of macro
> > '_EMIT6'
> > 211 | *(u32 *) (jit->prg_buf + jit->prg) =
> > (op1); \
> > | ^~~
> > include/linux/stddef.h:16:33: error: invalid use of undefined type
> > 'struct prog_frame'
> > 16 | #define offsetof(TYPE, MEMBER) __builtin_offsetof(TYPE,
> > MEMBER)
> > | ^~~~~~~~~~~~~~~~~~
> > arch/s390/net/bpf_jit_comp.c:211:55: note: in definition of macro
> > '_EMIT6'
> > 211 | *(u32 *) (jit->prg_buf + jit->prg) =
> > (op1); \
> > | ^~~
> > arch/s390/net/bpf_jit_comp.c:1814:46: note: in expansion of macro
> > 'offsetof'
> > 1814 |
> > offsetof(struct prog_frame,
> > | ^~~~~~~~
> > include/linux/stddef.h:16:33: error: invalid use of undefined type
> > 'struct prog_frame'
> > 16 | #define offsetof(TYPE, MEMBER) __builtin_offsetof(TYPE,
> > MEMBER)
> > | ^~~~~~~~~~~~~~~~~~
> > arch/s390/net/bpf_jit_comp.c:212:59: note: in definition of macro
> > '_EMIT6'
> > 212 | *(u16 *) (jit->prg_buf + jit->prg + 4) =
> > (op2); \
> > |
> > ^~~
> > arch/s390/net/bpf_jit_comp.c:1816:41: note: in expansion of macro
> > 'offsetof'
> > 1816 | 0xf000 | offsetof(struct
> > prog_frame,
> > | ^~~~~~~~
> > arch/s390/net/bpf_jit_comp.c: In function
> > '__arch_prepare_bpf_trampoline':
> > include/linux/stddef.h:16:33: error: invalid use of undefined type
> > 'struct prog_frame'
> > 16 | #define offsetof(TYPE, MEMBER) __builtin_offsetof(TYPE,
> > MEMBER)
> > | ^~~~~~~~~~~~~~~~~~
> > arch/s390/net/bpf_jit_comp.c:212:59: note: in definition of macro
> > '_EMIT6'
> > 212 | *(u16 *) (jit->prg_buf + jit->prg + 4) =
> > (op2); \
> > |
> > ^~~
> > arch/s390/net/bpf_jit_comp.c:2813:33: note: in expansion of macro
> > 'offsetof'
> > 2813 | 0xf000 | offsetof(struct prog_frame,
> > tail_call_cnt));
> > | ^~~~~~~~
> > make[5]: *** [scripts/Makefile.build:229:
> > arch/s390/net/bpf_jit_comp.o] Error 1
> >
> > The git blame is pointing to,
> > $ git blame -L 1813 arch/s390/net/bpf_jit_comp.c
> > 162513d7d81487 (Ilya Leoshkevich) _EMIT6(0xd203f000 | (jit-
> > >frame_off +
> >
> > Commit pointing to,
> > s390/bpf: Write back tail call counter for BPF_PSEUDO_CALL
> > [ Upstream commit c861a6b147137d10b5ff88a2c492ba376cd1b8b0 ]
>
> Anders bisected reported regressions and also suggested the missing
> patches.
>
> Ilya Leoshkevich,
> Is it a good idea to backport / cherry pick these two patches on the
> 6.12 branch ?
>
> b2268d550d20 ("s390/bpf: Centralize frame offset calculations")
> e26d523edf2a ("s390/bpf: Describe the frame using a struct instead of
> constants")
Thank you for the report and the investigation!
I think it would be a good idea to backport these.
Both are NFC changes that went into v6.17 and there were no complaints.
For v6.6 we also need this one (also NFC):
67aed27bcd46 ("s390/bpf: Change seen_reg to a mask")
> > ## Build
> > * kernel: 6.12.53-rc1
> > * git:
> > https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable-rc.git
> > * git commit: 7e50c0945b4ab1d4019f9905f6cf5350082c6a84
> > * git describe: v6.12.52-263-g7e50c0945b4a
> > * test details:
> > https://qa-reports.linaro.org/lkft/linux-stable-rc-linux-6.12.y/build/v6.12.52-263-g7e50c0945b4a
> >
> > ## Test Regressions (compared to v6.12.50-47-gf7ad21173a19)
> > * s390, build
> > - clang-21-defconfig
> > - clang-nightly-defconfig
> > - clang-nightly-lkftconfig-hardening
> > - clang-nightly-lkftconfig-lto-full
> > - clang-nightly-lkftconfig-lto-thing
> > - gcc-14-allmodconfig
> > - gcc-14-defconfig
> > - gcc-14-lkftconfig-hardening
> > - gcc-8-defconfig-fe40093d
> > - gcc-8-lkftconfig-hardening
> > - korg-clang-21-lkftconfig-hardening
> > - korg-clang-21-lkftconfig-lto-full
> > - korg-clang-21-lkftconfig-lto-thing
>
> - Naresh