[GIT PULL] sparc updates for v7.4

[GIT PULL] sparc updates for v7.4

[Andreas Larsson]

Hi Linus,

The following changes since commit cee9395acd8043be0644b25c34bfa86623f2b935:

  Linux 7.3-rc1 (2026-08-30 13:34:40 -0700)

are available in the Git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/alarsson/linux-sparc.git tags/sparc-for-7.4-tag1

for you to fetch changes up to 6ea75a63132d17ac3f62a37ff1b893ff7965b962:

  sparc64: flush vmalloc ranges from the D-cache on map and unmap (2026-10-28 11:02:33 +0100)

----------------------------------------------------------------
This includes the following changes related to sparc for v7.4:

- Allow loading sparc32 kernels on Sun machines away from start of RAM

- Fix timer comparator problem for sparc64 resulting in RCU stalls

- Increase kernel thread stack size for sparc64

- Fix bug in fault handling during window fill for sparc64

- Fix bug in vmalloc fault handling for sparc32

- Fix fault address reporting bugs for both sparc32 and sparc64

- Use membanks for early memory setup, removing legacy sp_banks construction for
  sparc32

- Add seccomp filter support for sparc64 and seccomp support for sparc32

- Fix problem with nested tick interrupts and IPIs leading to stack overflow for
  sparc32

- Fix missing cache flushes as well as optimise away redundant cache flushes for
  sparc64

- Clean up dead/unnecessary code and obsolete comments

----------------------------------------------------------------
Imre Kaloz (4):
      sparc32: keep PIL raised on the SMP ticker and LEON IPI exit paths
      sparc64: skip the trap block in xcall_deliver() when there is no target
      sparc64: flush only the aliased page, not its whole folio
      sparc64: flush vmalloc ranges from the D-cache on map and unmap

Ludwig Rydberg (5):
      sparc32: Split common syscall_trace() to syscall_trace_{enter|leave}()
      sparc32: add seccomp filter support
      selftests/seccomp: add sparc32 support
      sparc32: Avoid NULL pointer dereference in fault handler
      sparc32: Fix pgd lookup in vmalloc fault handling

Magnus Lindholm (9):
      sparc32: honour phys_base in the viking cache flush routines
      sparc32: derive phys_base from the PAGE_OFFSET mapping
      sparc32: advertise relocatable kernel with HdrS 0x0300
      sparc32: use memblock when mapping the kernel
      sparc32: use memblock to find available system memory
      sparc32: populate memblock from the PROM memory map
      sparc32: drop unused valid address bitmap
      sparc32: move early memory setup to setup_arch
      sparc32: drop sp_banks

Manuel Ebner (1):
      sparc: NMI: Remove obsolete comment

Sang-Heon Jeon (1):
      sparc32: remove redundant memset in srmmu_nocache_init()

Stian Halseth (6):
      sparc64: remove dead THREAD_SIZE branches for non-8K pages
      sparc64: restore %asi in user_rtt_fill_fixup_common
      sparc64: use the fault address for si_addr on window fixup faults
      sparc64: decide the TSB huge-page window fixup before the bank switch
      sparc64: add seccomp filter support
      selftests/seccomp: add sparc64 support

Tony Rodriguez (2):
      sparc64: Fix comparator problem with timer interrupts
      sparc64: increase kernel thread stack size to 32K

 arch/sparc/Kconfig                            |   2 +-
 arch/sparc/include/asm/cacheflush_64.h        |   9 +-
 arch/sparc/include/asm/highmem.h              |   3 -
 arch/sparc/include/asm/leon.h                 |   2 +-
 arch/sparc/include/asm/page_32.h              |  16 ---
 arch/sparc/include/asm/pgtable_32.h           |   3 +-
 arch/sparc/include/asm/pgtable_64.h           |   8 +-
 arch/sparc/include/asm/pgtsrmmu.h             |   1 -
 arch/sparc/include/asm/seccomp.h              |  15 +++
 arch/sparc/include/asm/syscall.h              |  11 +-
 arch/sparc/include/asm/thread_info_32.h       |   2 +
 arch/sparc/include/asm/thread_info_64.h       |  15 +--
 arch/sparc/kernel/entry.S                     |  42 ++++---
 arch/sparc/kernel/entry.h                     |   2 +-
 arch/sparc/kernel/head_32.S                   |   2 +-
 arch/sparc/kernel/kernel.h                    |   3 +-
 arch/sparc/kernel/nmi.c                       |   2 +-
 arch/sparc/kernel/ptrace_32.c                 |  26 ++--
 arch/sparc/kernel/ptrace_64.c                 |  28 +++--
 arch/sparc/kernel/setup_32.c                  |  91 +++++++++++---
 arch/sparc/kernel/smp_64.c                    |  51 ++++++--
 arch/sparc/kernel/syscalls.S                  |  28 ++++-
 arch/sparc/kernel/time_64.c                   |   6 +-
 arch/sparc/kernel/tsb.S                       |  24 +++-
 arch/sparc/kernel/urtt_fill.S                 |  10 ++
 arch/sparc/mm/fault_32.c                      |  32 ++---
 arch/sparc/mm/fault_64.c                      |   9 ++
 arch/sparc/mm/init_32.c                       | 167 +-------------------------
 arch/sparc/mm/init_64.c                       |  65 ++++++++++
 arch/sparc/mm/leon_mm.c                       |   1 +
 arch/sparc/mm/srmmu.c                         |  70 ++++-------
 arch/sparc/mm/tlb.c                           |   2 +-
 arch/sparc/mm/ultra.S                         |  25 ++++
 arch/sparc/mm/viking.S                        |   6 +
 arch/sparc/prom/memory.c                      |  51 ++------
 tools/testing/selftests/seccomp/seccomp_bpf.c |  52 ++++++++
 36 files changed, 481 insertions(+), 401 deletions(-)

Thanks,
Andreas

Re: [GIT PULL] sparc updates for v7.4

[Linus Torvalds]

On Wed, 28 Oct 2026 at 03:44, Andreas Larsson <andreas@gaisler.com> wrote:
>
> This includes the following changes related to sparc for v7.4:

Send them for 7.5

This is the second week, and the commits are all apparently basically
just a couple of hours before your pull request.

The merge window is for things that were ready before the merge
window, not for hurried last-minute updates.

               Linus

Re: [GIT PULL] sparc updates for v7.4

[John Paul Adrian Glaubitz]

Hi Linus,

On Wed, 2026-10-28 at 07:03 -1000, Linus Torvalds wrote:
> On Wed, 28 Oct 2026 at 03:44, Andreas Larsson <andreas@gaisler.com> wrote:
> > 
> > This includes the following changes related to sparc for v7.4:
> 
> Send them for 7.5
> 
> This is the second week, and the commits are all apparently basically
> just a couple of hours before your pull request.

I think this was actually accidental. Andreas had most patches already
merged before the merge window opened as my local checkout shows:

glaubitz@node54:/data/home/glaubitz/linux> git log --graph --pretty=format:'%C(auto)%h%d (%cr) %cn <%ce> %s' | head -n20
* a9e88d617a25 (HEAD, alarsson/for-next) (3 weeks ago) Andreas Larsson <andreas@gaisler.com> sparc32: remove redundant memset in srmmu_nocache_init()
* 077e15cf545a (3 weeks ago) Andreas Larsson <andreas@gaisler.com> selftests/seccomp: add sparc64 support
* 60b5c9eda5a3 (3 weeks ago) Andreas Larsson <andreas@gaisler.com> sparc64: add seccomp filter support
* 9840f43c3d09 (3 weeks ago) Andreas Larsson <andreas@gaisler.com> sparc: NMI: Remove obsolete comment
* e77e845f7226 (3 weeks ago) Andreas Larsson <andreas@gaisler.com> sparc32: drop sp_banks
* 0a8f5903b456 (3 weeks ago) Andreas Larsson <andreas@gaisler.com> sparc32: move early memory setup to setup_arch
* 91c897e01883 (3 weeks ago) Andreas Larsson <andreas@gaisler.com> sparc32: drop unused valid address bitmap
* 6b2510b9d8f7 (3 weeks ago) Andreas Larsson <andreas@gaisler.com> sparc32: populate memblock from the PROM memory map
* a33ec04570f3 (3 weeks ago) Andreas Larsson <andreas@gaisler.com> sparc32: use memblock to find available system memory
* ccdc278b61c2 (3 weeks ago) Andreas Larsson <andreas@gaisler.com> sparc32: use memblock when mapping the kernel
* 22a749706e3b (3 weeks ago) Andreas Larsson <andreas@gaisler.com> sparc64: decide the TSB huge-page window fixup before the bank switch
* f1c0029fd6bf (3 weeks ago) Andreas Larsson <andreas@gaisler.com> sparc64: use the fault address for si_addr on window fixup faults
* 80fd9c765ddc (3 weeks ago) Andreas Larsson <andreas@gaisler.com> sparc64: restore %asi in user_rtt_fill_fixup_common
* 1ea502a48efc (3 weeks ago) Andreas Larsson <andreas@gaisler.com> sparc64: remove dead THREAD_SIZE branches for non-8K pages
* ca0a9905074e (3 weeks ago) Andreas Larsson <andreas@gaisler.com> sparc64: increase kernel thread stack size to 32K
* 6982cb1c57c8 (3 weeks ago) Andreas Larsson <andreas@gaisler.com> sparc64: Fix comparator problem with timer interrupts
* 01d609628750 (3 weeks ago) Andreas Larsson <andreas@gaisler.com> sparc32: advertise relocatable kernel with HdrS 0x0300
* 7c0a0dda3830 (3 weeks ago) Andreas Larsson <andreas@gaisler.com> sparc32: derive phys_base from the PAGE_OFFSET mapping
* 494e0351b740 (3 weeks ago) Andreas Larsson <andreas@gaisler.com> sparc32: honour phys_base in the viking cache flush routines
* cee9395acd80 (tag: v7.3-rc1, alarsson/for-linus) (8 weeks ago) Linus Torvalds <torvalds@linux-foundation.org> Linux 7.3-rc1
glaubitz@node54:/data/home/glaubitz/linux>

It seems that Andreas rebased his tree today and therefore bumped all
the commit dates.  I unfortunately don't have a newer pull of Andreas'
tree, so I'm missing some commits.

Andreas, do you happen to have a local backup from before?

Adrian


-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer
`. `'   Physicist
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913

Re: [GIT PULL] sparc updates for v7.4

[John Paul Adrian Glaubitz]

Hi Linus,

On Wed, 2026-10-28 at 21:50 +0100, John Paul Adrian Glaubitz wrote:
> It seems that Andreas rebased his tree today and therefore bumped all
> the commit dates.  I unfortunately don't have a newer pull of Andreas'
> tree, so I'm missing some commits.
> 
> Andreas, do you happen to have a local backup from before?

Short follow-up question: If Andreas doesn't happen to have a backup,
is there maybe a daily cron job that backs up git.kernel.org which
could be used to retrieve the original tree?

Adrian

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer
`. `'   Physicist
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913

Re: [GIT PULL] sparc updates for v7.4

[Andreas Larsson]

On 2026-10-28 18:03, Linus Torvalds wrote:
> On Wed, 28 Oct 2026 at 03:44, Andreas Larsson <andreas@gaisler.com> wrote:
>>
>> This includes the following changes related to sparc for v7.4:
> 
> Send them for 7.5
> 
> This is the second week, and the commits are all apparently basically
> just a couple of hours before your pull request.
> 
> The merge window is for things that were ready before the merge
> window, not for hurried last-minute updates.
Hi Linus,

I got a late Tested-by for one of the patches, "sparc64: increase kernel
thread stack size to 32K" that I picked up already 2026-09-27, that I
inserted just before preparing the pull request. Sorry about that. Apart
from that added Tested-by, all changes has been on my for-next since
2026-10-23, and most much longer than that, and with that included in
linux-next.

Below follows a pull request for my original for-next before the added
Tested-by. I hope you can take either my original pull request with my
explanation of the one below if you rather want to have commit dates
that better represents when they were picked up.


The following changes since commit cee9395acd8043be0644b25c34bfa86623f2b935:

  Linux 7.3-rc1 (2026-08-30 13:34:40 -0700)

are available in the Git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/alarsson/linux-sparc.git tags/sparc-for-7.4-tag2

for you to fetch changes up to d957b367d560c4a74637d400ca4e0e63bbdfbcd1:

  sparc64: flush vmalloc ranges from the D-cache on map and unmap (2026-10-23 11:40:02 +0200)

----------------------------------------------------------------
This includes the following changes related to sparc for v7.4:

- Allow loading sparc32 kernels on Sun machines away from start of RAM

- Fix timer comparator problem for sparc64 resulting in RCU stalls

- Increase kernel thread stack size for sparc64

- Fix bug in fault handling during window fill for sparc64

- Fix bug in vmalloc fault handling for sparc32

- Fix fault address reporting bugs for both sparc32 and sparc64

- Use membanks for early memory setup, removing legacy sp_banks construction for
  sparc32

- Add seccomp filter support for sparc64 and seccomp support for sparc32

- Fix problem with nested tick interrupts and IPIs leading to stack overflow for
  sparc32

- Fix missing cache flushes as well as optimise away redundant cache flushes for
  sparc64

- Clean up dead/unnecessary code and obsolete comments

----------------------------------------------------------------
Imre Kaloz (4):
      sparc32: keep PIL raised on the SMP ticker and LEON IPI exit paths
      sparc64: skip the trap block in xcall_deliver() when there is no target
      sparc64: flush only the aliased page, not its whole folio
      sparc64: flush vmalloc ranges from the D-cache on map and unmap

Ludwig Rydberg (5):
      sparc32: Split common syscall_trace() to syscall_trace_{enter|leave}()
      sparc32: add seccomp filter support
      selftests/seccomp: add sparc32 support
      sparc32: Avoid NULL pointer dereference in fault handler
      sparc32: Fix pgd lookup in vmalloc fault handling

Magnus Lindholm (9):
      sparc32: honour phys_base in the viking cache flush routines
      sparc32: derive phys_base from the PAGE_OFFSET mapping
      sparc32: advertise relocatable kernel with HdrS 0x0300
      sparc32: use memblock when mapping the kernel
      sparc32: use memblock to find available system memory
      sparc32: populate memblock from the PROM memory map
      sparc32: drop unused valid address bitmap
      sparc32: move early memory setup to setup_arch
      sparc32: drop sp_banks

Manuel Ebner (1):
      sparc: NMI: Remove obsolete comment

Sang-Heon Jeon (1):
      sparc32: remove redundant memset in srmmu_nocache_init()

Stian Halseth (6):
      sparc64: remove dead THREAD_SIZE branches for non-8K pages
      sparc64: restore %asi in user_rtt_fill_fixup_common
      sparc64: use the fault address for si_addr on window fixup faults
      sparc64: decide the TSB huge-page window fixup before the bank switch
      sparc64: add seccomp filter support
      selftests/seccomp: add sparc64 support

Tony Rodriguez (2):
      sparc64: Fix comparator problem with timer interrupts
      sparc64: increase kernel thread stack size to 32K

 arch/sparc/Kconfig                            |   2 +-
 arch/sparc/include/asm/cacheflush_64.h        |   9 +-
 arch/sparc/include/asm/highmem.h              |   3 -
 arch/sparc/include/asm/leon.h                 |   2 +-
 arch/sparc/include/asm/page_32.h              |  16 ---
 arch/sparc/include/asm/pgtable_32.h           |   3 +-
 arch/sparc/include/asm/pgtable_64.h           |   8 +-
 arch/sparc/include/asm/pgtsrmmu.h             |   1 -
 arch/sparc/include/asm/seccomp.h              |  15 +++
 arch/sparc/include/asm/syscall.h              |  11 +-
 arch/sparc/include/asm/thread_info_32.h       |   2 +
 arch/sparc/include/asm/thread_info_64.h       |  15 +--
 arch/sparc/kernel/entry.S                     |  42 ++++---
 arch/sparc/kernel/entry.h                     |   2 +-
 arch/sparc/kernel/head_32.S                   |   2 +-
 arch/sparc/kernel/kernel.h                    |   3 +-
 arch/sparc/kernel/nmi.c                       |   2 +-
 arch/sparc/kernel/ptrace_32.c                 |  26 ++--
 arch/sparc/kernel/ptrace_64.c                 |  28 +++--
 arch/sparc/kernel/setup_32.c                  |  91 +++++++++++---
 arch/sparc/kernel/smp_64.c                    |  51 ++++++--
 arch/sparc/kernel/syscalls.S                  |  28 ++++-
 arch/sparc/kernel/time_64.c                   |   6 +-
 arch/sparc/kernel/tsb.S                       |  24 +++-
 arch/sparc/kernel/urtt_fill.S                 |  10 ++
 arch/sparc/mm/fault_32.c                      |  32 ++---
 arch/sparc/mm/fault_64.c                      |   9 ++
 arch/sparc/mm/init_32.c                       | 167 +-------------------------
 arch/sparc/mm/init_64.c                       |  65 ++++++++++
 arch/sparc/mm/leon_mm.c                       |   1 +
 arch/sparc/mm/srmmu.c                         |  70 ++++-------
 arch/sparc/mm/tlb.c                           |   2 +-
 arch/sparc/mm/ultra.S                         |  25 ++++
 arch/sparc/mm/viking.S                        |   6 +
 arch/sparc/prom/memory.c                      |  51 ++------
 tools/testing/selftests/seccomp/seccomp_bpf.c |  52 ++++++++
 36 files changed, 481 insertions(+), 401 deletions(-)

Thanks,
Andreas

Re: [GIT PULL] sparc updates for v7.4

[John Paul Adrian Glaubitz]

Hi Andreas,

On Thu, 2026-10-29 at 08:03 +0100, Andreas Larsson wrote:
> I got a late Tested-by for one of the patches, "sparc64: increase kernel
> thread stack size to 32K" that I picked up already 2026-09-27, that I
> inserted just before preparing the pull request. Sorry about that. Apart
> from that added Tested-by, all changes has been on my for-next since
> 2026-10-23, and most much longer than that, and with that included in
> linux-next.

If you really want to modify an old commit without touching the date, you
can use this git rebase command:

git -c rebase.instructionFormat='%s%nexec GIT_COMMITTER_DATE="%cD" \
       GIT_AUTHOR_DATE="%aD" git commit --amend' rebase $COMMIT_ID

> Below follows a pull request for my original for-next before the added
> Tested-by. I hope you can take either my original pull request with my
> explanation of the one below if you rather want to have commit dates
> that better represents when they were picked up.

Can you push this to your for-linus branch?

Adrian

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer
`. `'   Physicist
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913

Re: [GIT PULL] sparc updates for v7.4

[Andreas Larsson]

On 2026-10-29 09:05, John Paul Adrian Glaubitz wrote:
> Hi Andreas,
> 
> On Thu, 2026-10-29 at 08:03 +0100, Andreas Larsson wrote:
>> I got a late Tested-by for one of the patches, "sparc64: increase kernel
>> thread stack size to 32K" that I picked up already 2026-09-27, that I
>> inserted just before preparing the pull request. Sorry about that. Apart
>> from that added Tested-by, all changes has been on my for-next since
>> 2026-10-23, and most much longer than that, and with that included in
>> linux-next.
> 
> If you really want to modify an old commit without touching the date, you
> can use this git rebase command:
> ...

I'm happy if my sparc-for-7.4-tag2 tag, that points to the actual commit
that has been included by linux-next, is merged as is, without the added
Tested-by. No need to complicate things with additional options at this
point I think.


>> Below follows a pull request for my original for-next before the added
>> Tested-by. I hope you can take either my original pull request with my
>> explanation of the one below if you rather want to have commit dates
>> that better represents when they were picked up.
> 
> Can you push this to your for-linus branch?

My for-next branch is back where it was, i.e. where sparc-for-7.4-tag2
is now pointing.

Cheers,
Andreas

Re: [GIT PULL] sparc updates for v7.4

[Linus Torvalds]

On Wed, 28 Oct 2026 at 22:06, John Paul Adrian Glaubitz
<glaubitz@physik.fu-berlin.de> wrote:
>
> If you really want to modify an old commit without touching the date, you
> can use this git rebase command:

STOP THIS DAMMIT!

You are now telling people to LIE about what they do.

No. The fix is to NOT REBASE POINTLESSLY.

And no, adding a "tested-by" is simply not a good enough reason to
rebase long after-the-fact.

So stop doing bad things. Bad things include:
 - lying about commit times
 - rebasing mindlessly
 - sending me pull requests for stuff that hasn't been in linux-next

And if you have a *major* and actual real reason to rebase (completely
broken history that needs to be fixed), it should be mentioned and
explained.

Not hidden away by lying about the commit times.

John, wow I don't want to pull anything from you, since you clearly
have this completely broken mentality where it's ok to lie about
rebases to hide the fact that you do it.

     Linus

Re: [GIT PULL] sparc updates for v7.4

[John Paul Adrian Glaubitz]

Hi Linus,

On Thu, 2026-10-29 at 12:49 -1000, Linus Torvalds wrote:
> John, wow I don't want to pull anything from you, since you clearly
> have this completely broken mentality where it's ok to lie about
> rebases to hide the fact that you do it.

I have not said that one should lie about rebases and the commit history
of one's tree. I was merely trying to help Andreas fix his tree so it
can be pulled.

I'm sorry if I left the impression that it's okay to manipulate a git
history that was not my intention. I'm already going through a shitty
situation in my personal life at the moment and I'm not really prepared
to just take another hit like this.

Adrian

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer
`. `'   Physicist
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913

Re: [GIT PULL] sparc updates for v7.4

[Andreas Larsson]

On 2026-10-29 23:49, Linus Torvalds wrote:
> No. The fix is to NOT REBASE POINTLESSLY.
> 
> And no, adding a "tested-by" is simply not a good enough reason to
> rebase long after-the-fact.
Understood. This was my mistake. I was trying to do the right thing and
clearly did not. Sorry about that.

Just to be clear, the followup pull request for sparc-for-7.4-tag2 I did
in this thread is for what has actually been in linux-next. Each commit
there was pushed to my for-next the same day as I committed it.

> And if you have a *major* and actual real reason to rebase (completely
> broken history that needs to be fixed), it should be mentioned and
> explained.

Understood. Sorry for the mess.

Best regards,
Andreas