linux-block.git
2 weeks agomailmap: add Jarkko's employer email address
Jarkko Sakkinen [Fri, 23 May 2025 12:11:04 +0000 (15:11 +0300)]
mailmap: add Jarkko's employer email address

Add the current employer email address to mailmap.

Link: https://lkml.kernel.org/r/20250523121105.15850-1-jarkko@kernel.org
Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org>
Cc: Alexander Sverdlin <alexander.sverdlin@gmail.com>
Cc: Antonio Quartulli <antonio@openvpn.net>
Cc: Carlos Bilbao <carlos.bilbao@kernel.org>
Cc: Kees Cook <kees@kernel.org>
Cc: Simon Wunderlich <sw@simonwunderlich.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 weeks agomm: fix copy_vma() error handling for hugetlb mappings
Ricardo Cañuelo Navarro [Fri, 23 May 2025 12:19:10 +0000 (14:19 +0200)]
mm: fix copy_vma() error handling for hugetlb mappings

If, during a mremap() operation for a hugetlb-backed memory mapping,
copy_vma() fails after the source vma has been duplicated and opened (ie.
vma_link() fails), the error is handled by closing the new vma.  This
updates the hugetlbfs reservation counter of the reservation map which at
this point is referenced by both the source vma and the new copy.  As a
result, once the new vma has been freed and copy_vma() returns, the
reservation counter for the source vma will be incorrect.

This patch addresses this corner case by clearing the hugetlb private page
reservation reference for the new vma and decrementing the reference
before closing the vma, so that vma_close() won't update the reservation
counter.  This is also what copy_vma_and_data() does with the source vma
if copy_vma() succeeds, so a helper function has been added to do the
fixup in both functions.

The issue was reported by a private syzbot instance and can be reproduced
using the C reproducer in [1].  It's also a possible duplicate of public
syzbot report [2].  The WARNING report is:

============================================================
page_counter underflow: -1024 nr_pages=1024
WARNING: CPU: 0 PID: 3287 at mm/page_counter.c:61 page_counter_cancel+0xf6/0x120
Modules linked in:
CPU: 0 UID: 0 PID: 3287 Comm: repro__WARNING_ Not tainted 6.15.0-rc7+ #54 NONE
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.16.3-2-gc13ff2cd-prebuilt.qemu.org 04/01/2014
RIP: 0010:page_counter_cancel+0xf6/0x120
Code: ff 5b 41 5e 41 5f 5d c3 cc cc cc cc e8 f3 4f 8f ff c6 05 64 01 27 06 01 48 c7 c7 60 15 f8 85 48 89 de 4c 89 fa e8 2a a7 51 ff <0f> 0b e9 66 ff ff ff 44 89 f9 80 e1 07 38 c1 7c 9d 4c 81
RSP: 0018:ffffc900025df6a0 EFLAGS: 00010246
RAX: 2edfc409ebb44e00 RBX: fffffffffffffc00 RCX: ffff8880155f0000
RDX: 0000000000000000 RSI: 0000000000000001 RDI: 0000000000000000
RBP: dffffc0000000000 R08: ffffffff81c4a23c R09: 1ffff1100330482a
R10: dffffc0000000000 R11: ffffed100330482b R12: 0000000000000000
R13: ffff888058a882c0 R14: ffff888058a882c0 R15: 0000000000000400
FS:  0000000000000000(0000) GS:ffff88808fc53000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00000000004b33e0 CR3: 00000000076d6000 CR4: 00000000000006f0
Call Trace:
 <TASK>
 page_counter_uncharge+0x33/0x80
 hugetlb_cgroup_uncharge_counter+0xcb/0x120
 hugetlb_vm_op_close+0x579/0x960
 ? __pfx_hugetlb_vm_op_close+0x10/0x10
 remove_vma+0x88/0x130
 exit_mmap+0x71e/0xe00
 ? __pfx_exit_mmap+0x10/0x10
 ? __mutex_unlock_slowpath+0x22e/0x7f0
 ? __pfx_exit_aio+0x10/0x10
 ? __up_read+0x256/0x690
 ? uprobe_clear_state+0x274/0x290
 ? mm_update_next_owner+0xa9/0x810
 __mmput+0xc9/0x370
 exit_mm+0x203/0x2f0
 ? __pfx_exit_mm+0x10/0x10
 ? taskstats_exit+0x32b/0xa60
 do_exit+0x921/0x2740
 ? do_raw_spin_lock+0x155/0x3b0
 ? __pfx_do_exit+0x10/0x10
 ? __pfx_do_raw_spin_lock+0x10/0x10
 ? _raw_spin_lock_irq+0xc5/0x100
 do_group_exit+0x20c/0x2c0
 get_signal+0x168c/0x1720
 ? __pfx_get_signal+0x10/0x10
 ? schedule+0x165/0x360
 arch_do_signal_or_restart+0x8e/0x7d0
 ? __pfx_arch_do_signal_or_restart+0x10/0x10
 ? __pfx___se_sys_futex+0x10/0x10
 syscall_exit_to_user_mode+0xb8/0x2c0
 do_syscall_64+0x75/0x120
 entry_SYSCALL_64_after_hwframe+0x76/0x7e
RIP: 0033:0x422dcd
Code: Unable to access opcode bytes at 0x422da3.
RSP: 002b:00007ff266cdb208 EFLAGS: 00000246 ORIG_RAX: 00000000000000ca
RAX: 0000000000000001 RBX: 00007ff266cdbcdc RCX: 0000000000422dcd
RDX: 00000000000f4240 RSI: 0000000000000081 RDI: 00000000004c7bec
RBP: 00007ff266cdb220 R08: 203a6362696c6720 R09: 203a6362696c6720
R10: 0000200000c00000 R11: 0000000000000246 R12: ffffffffffffffd0
R13: 0000000000000002 R14: 00007ffe1cb5f520 R15: 00007ff266cbb000
 </TASK>
============================================================

Link: https://lkml.kernel.org/r/20250523-warning_in_page_counter_cancel-v2-1-b6df1a8cfefd@igalia.com
Link: https://people.igalia.com/rcn/kernel_logs/20250422__WARNING_in_page_counter_cancel__repro.c
Link: https://lore.kernel.org/all/67000a50.050a0220.49194.048d.GAE@google.com/
Signed-off-by: Ricardo Cañuelo Navarro <rcn@igalia.com>
Suggested-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Florent Revest <revest@google.com>
Cc: Jann Horn <jannh@google.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 weeks agomemcg: always call cond_resched() after fn()
Breno Leitao [Fri, 23 May 2025 17:21:06 +0000 (10:21 -0700)]
memcg: always call cond_resched() after fn()

I am seeing soft lockup on certain machine types when a cgroup OOMs.  This
is happening because killing the process in certain machine might be very
slow, which causes the soft lockup and RCU stalls.  This happens usually
when the cgroup has MANY processes and memory.oom.group is set.

Example I am seeing in real production:

       [462012.244552] Memory cgroup out of memory: Killed process 3370438 (crosvm) ....
       ....
       [462037.318059] Memory cgroup out of memory: Killed process 4171372 (adb) ....
       [462037.348314] watchdog: BUG: soft lockup - CPU#64 stuck for 26s! [stat_manager-ag:1618982]
       ....

Quick look at why this is so slow, it seems to be related to serial flush
for certain machine types.  For all the crashes I saw, the target CPU was
at console_flush_all().

In the case above, there are thousands of processes in the cgroup, and it
is soft locking up before it reaches the 1024 limit in the code (which
would call the cond_resched()).  So, cond_resched() in 1024 blocks is not
sufficient.

Remove the counter-based conditional rescheduling logic and call
cond_resched() unconditionally after each task iteration, after fn() is
called.  This avoids the lockup independently of how slow fn() is.

Link: https://lkml.kernel.org/r/20250523-memcg_fix-v1-1-ad3eafb60477@debian.org
Fixes: ade81479c7dd ("memcg: fix soft lockup in the OOM process")
Signed-off-by: Breno Leitao <leitao@debian.org>
Suggested-by: Rik van Riel <riel@surriel.com>
Acked-by: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Michael van der Westhuizen <rmikey@meta.com>
Cc: Usama Arif <usamaarif642@gmail.com>
Cc: Pavel Begunkov <asml.silence@gmail.com>
Cc: Chen Ridong <chenridong@huawei.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 weeks agomm/hugetlb: fix kernel NULL pointer dereference when replacing free hugetlb folios
Ge Yang [Thu, 22 May 2025 03:22:17 +0000 (11:22 +0800)]
mm/hugetlb: fix kernel NULL pointer dereference when replacing free hugetlb folios

A kernel crash was observed when replacing free hugetlb folios:

BUG: kernel NULL pointer dereference, address: 0000000000000028
PGD 0 P4D 0
Oops: Oops: 0000 [#1] SMP NOPTI
CPU: 28 UID: 0 PID: 29639 Comm: test_cma.sh Tainted 6.15.0-rc6-zp #41 PREEMPT(voluntary)
RIP: 0010:alloc_and_dissolve_hugetlb_folio+0x1d/0x1f0
RSP: 0018:ffffc9000b30fa90 EFLAGS: 00010286
RAX: 0000000000000000 RBX: 0000000000342cca RCX: ffffea0043000000
RDX: ffffc9000b30fb08 RSI: ffffea0043000000 RDI: 0000000000000000
RBP: ffffc9000b30fb20 R08: 0000000000001000 R09: 0000000000000000
R10: ffff88886f92eb00 R11: 0000000000000000 R12: ffffea0043000000
R13: 0000000000000000 R14: 00000000010c0200 R15: 0000000000000004
FS:  00007fcda5f14740(0000) GS:ffff8888ec1d8000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000000028 CR3: 0000000391402000 CR4: 0000000000350ef0
Call Trace:
<TASK>
 replace_free_hugepage_folios+0xb6/0x100
 alloc_contig_range_noprof+0x18a/0x590
 ? srso_return_thunk+0x5/0x5f
 ? down_read+0x12/0xa0
 ? srso_return_thunk+0x5/0x5f
 cma_range_alloc.constprop.0+0x131/0x290
 __cma_alloc+0xcf/0x2c0
 cma_alloc_write+0x43/0xb0
 simple_attr_write_xsigned.constprop.0.isra.0+0xb2/0x110
 debugfs_attr_write+0x46/0x70
 full_proxy_write+0x62/0xa0
 vfs_write+0xf8/0x420
 ? srso_return_thunk+0x5/0x5f
 ? filp_flush+0x86/0xa0
 ? srso_return_thunk+0x5/0x5f
 ? filp_close+0x1f/0x30
 ? srso_return_thunk+0x5/0x5f
 ? do_dup2+0xaf/0x160
 ? srso_return_thunk+0x5/0x5f
 ksys_write+0x65/0xe0
 do_syscall_64+0x64/0x170
 entry_SYSCALL_64_after_hwframe+0x76/0x7e

There is a potential race between __update_and_free_hugetlb_folio() and
replace_free_hugepage_folios():

CPU1                              CPU2
__update_and_free_hugetlb_folio   replace_free_hugepage_folios
                                    folio_test_hugetlb(folio)
                                    -- It's still hugetlb folio.

  __folio_clear_hugetlb(folio)
  hugetlb_free_folio(folio)
                                    h = folio_hstate(folio)
                                    -- Here, h is NULL pointer

When the above race condition occurs, folio_hstate(folio) returns NULL,
and subsequent access to this NULL pointer will cause the system to crash.
To resolve this issue, execute folio_hstate(folio) under the protection
of the hugetlb_lock lock, ensuring that folio_hstate(folio) does not
return NULL.

Link: https://lkml.kernel.org/r/1747884137-26685-1-git-send-email-yangge1116@126.com
Fixes: 04f13d241b8b ("mm: replace free hugepage folios after migration")
Signed-off-by: Ge Yang <yangge1116@126.com>
Reviewed-by: Muchun Song <muchun.song@linux.dev>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Barry Song <21cnbao@gmail.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 weeks agomm: vmalloc: only zero-init on vrealloc shrink
Kees Cook [Thu, 15 May 2025 21:42:16 +0000 (14:42 -0700)]
mm: vmalloc: only zero-init on vrealloc shrink

The common case is to grow reallocations, and since init_on_alloc will
have already zeroed the whole allocation, we only need to zero when
shrinking the allocation.

Link: https://lkml.kernel.org/r/20250515214217.619685-2-kees@kernel.org
Fixes: a0309faf1cb0 ("mm: vmalloc: support more granular vrealloc() sizing")
Signed-off-by: Kees Cook <kees@kernel.org>
Tested-by: Pawan Gupta <pawan.kumar.gupta@linux.intel.com>
Cc: Danilo Krummrich <dakr@kernel.org>
Cc: Eduard Zingerman <eddyz87@gmail.com>
Cc: "Erhard F." <erhard_f@mailbox.org>
Cc: Shung-Hsi Yu <shung-hsi.yu@suse.com>
Cc: "Uladzislau Rezki (Sony)" <urezki@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 weeks agomm: vmalloc: actually use the in-place vrealloc region
Kees Cook [Thu, 15 May 2025 21:42:15 +0000 (14:42 -0700)]
mm: vmalloc: actually use the in-place vrealloc region

Patch series "mm: vmalloc: Actually use the in-place vrealloc region".

This fixes a performance regression[1] with vrealloc()[1].

The refactoring to not build a new vmalloc region only actually worked
when shrinking.  Actually return the resized area when it grows.  Ugh.

Link: https://lkml.kernel.org/r/20250515214217.619685-1-kees@kernel.org
Fixes: a0309faf1cb0 ("mm: vmalloc: support more granular vrealloc() sizing")
Signed-off-by: Kees Cook <kees@kernel.org>
Reported-by: Shung-Hsi Yu <shung-hsi.yu@suse.com>
Closes: https://lore.kernel.org/all/20250515-bpf-verifier-slowdown-vwo2meju4cgp2su5ckj@6gi6ssxbnfqg [1]
Tested-by: Eduard Zingerman <eddyz87@gmail.com>
Tested-by: Pawan Gupta <pawan.kumar.gupta@linux.intel.com>
Tested-by: Shung-Hsi Yu <shung-hsi.yu@suse.com>
Reviewed-by: "Uladzislau Rezki (Sony)" <urezki@gmail.com>
Reviewed-by: Danilo Krummrich <dakr@kernel.org>
Cc: "Erhard F." <erhard_f@mailbox.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 weeks agoalloc_tag: allocate percpu counters for module tags dynamically
Suren Baghdasaryan [Sat, 17 May 2025 00:07:39 +0000 (17:07 -0700)]
alloc_tag: allocate percpu counters for module tags dynamically

When a module gets unloaded it checks whether any of its tags are still in
use and if so, we keep the memory containing module's allocation tags
alive until all tags are unused.  However percpu counters referenced by
the tags are freed by free_module().  This will lead to UAF if the memory
allocated by a module is accessed after module was unloaded.

To fix this we allocate percpu counters for module allocation tags
dynamically and we keep it alive for tags which are still in use after
module unloading.  This also removes the requirement of a larger
PERCPU_MODULE_RESERVE when memory allocation profiling is enabled because
percpu memory for counters does not need to be reserved anymore.

Link: https://lkml.kernel.org/r/20250517000739.5930-1-surenb@google.com
Fixes: 0db6f8d7820a ("alloc_tag: load module tags into separate contiguous memory")
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Reported-by: David Wang <00107082@163.com>
Closes: https://lore.kernel.org/all/20250516131246.6244-1-00107082@163.com/
Tested-by: David Wang <00107082@163.com>
Cc: Christoph Lameter (Ampere) <cl@gentwo.org>
Cc: Dennis Zhou <dennis@kernel.org>
Cc: Kent Overstreet <kent.overstreet@linux.dev>
Cc: Pasha Tatashin <pasha.tatashin@soleen.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 weeks agomodule: release codetag section when module load fails
David Wang [Mon, 19 May 2025 16:38:23 +0000 (00:38 +0800)]
module: release codetag section when module load fails

When module load fails after memory for codetag section is ready, codetag
section memory will not be properly released.  This causes memory leak,
and if next module load happens to get the same module address, codetag
may pick the uninitialized section when manipulating tags during module
unload, and leads to "unable to handle page fault" BUG.

Link: https://lkml.kernel.org/r/20250519163823.7540-1-00107082@163.com
Fixes: 0db6f8d7820a ("alloc_tag: load module tags into separate contiguous memory")
Closes: https://lore.kernel.org/all/20250516131246.6244-1-00107082@163.com/
Signed-off-by: David Wang <00107082@163.com>
Acked-by: Suren Baghdasaryan <surenb@google.com>
Cc: Petr Pavlu <petr.pavlu@suse.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 weeks agomm/cma: make detection of highmem_start more robust
Mike Rapoport (Microsoft) [Mon, 19 May 2025 17:18:05 +0000 (20:18 +0300)]
mm/cma: make detection of highmem_start more robust

Pratyush Yadav reports the following crash:

    ------------[ cut here ]------------
    kernel BUG at arch/x86/mm/physaddr.c:23!
    ception 0x06 IP 10:ffffffff812ebbf8 error 0 cr2 0xffff88903ffff000
    CPU: 0 UID: 0 PID: 0 Comm: swapper Not tainted 6.15.0-rc6+ #231 PREEMPT(undef)
    Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Arch Linux 1.16.3-1-1 04/01/2014
    RIP: 0010:__phys_addr+0x58/0x60
    Code: 01 48 89 c2 48 d3 ea 48 85 d2 75 05 e9 91 52 cf 00 0f 0b 48 3d ff ff ff 1f 77 0f 48 8b 05 20 54 55 01 48 01 d0 e9 78 52 cf 00 <0f> 0b 90 0f 1f 44 00 00 90 90 90 90 90 90 90 90 90 90 90 90 90 90
    RSP: 0000:ffffffff82803dd8 EFLAGS: 00010006 ORIG_RAX: 0000000000000000
    RAX: 000000007fffffff RBX: 00000000ffffffff RCX: 0000000000000000
    RDX: 000000007fffffff RSI: 0000000280000000 RDI: ffffffffffffffff
    RBP: ffffffff82803e68 R08: 0000000000000000 R09: 0000000000000000
    R10: ffffffff83153180 R11: ffffffff82803e48 R12: ffffffff83c9aed0
    R13: 0000000000000000 R14: 0000001040000000 R15: 0000000000000000
    FS:  0000000000000000(0000) GS:0000000000000000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: ffff88903ffff000 CR3: 0000000002838000 CR4: 00000000000000b0
    Call Trace:
     <TASK>
     ? __cma_declare_contiguous_nid+0x6e/0x340
     ? cma_declare_contiguous_nid+0x33/0x70
     ? dma_contiguous_reserve_area+0x2f/0x70
     ? setup_arch+0x6f1/0x870
     ? start_kernel+0x52/0x4b0
     ? x86_64_start_reservations+0x29/0x30
     ? x86_64_start_kernel+0x7c/0x80
     ? common_startup_64+0x13e/0x141

  The reason is that __cma_declare_contiguous_nid() does:

          highmem_start = __pa(high_memory - 1) + 1;

  If dma_contiguous_reserve_area() (or any other CMA declaration) is
  called before free_area_init(), high_memory is uninitialized. Without
  CONFIG_DEBUG_VIRTUAL, it will likely work but use the wrong value for
  highmem_start.

The issue occurs because commit e120d1bc12da ("arch, mm: set high_memory
in free_area_init()") moved initialization of high_memory after the call
to dma_contiguous_reserve() -> __cma_declare_contiguous_nid() on several
architectures.

In the case CONFIG_HIGHMEM is enabled, some architectures that actually
support HIGHMEM (arm, powerpc and x86) have initialization of high_memory
before a possible call to __cma_declare_contiguous_nid() and some
initialized high_memory late anyway (arc, csky, microblase, mips, sparc,
xtensa) even before the commit e120d1bc12da so they are fine with using
uninitialized value of high_memory.

And in the case CONFIG_HIGHMEM is disabled high_memory essentially becomes
the first address after memory end, so instead of relying on high_memory
to calculate highmem_start use memblock_end_of_DRAM() and eliminate the
dependency of CMA area creation on high_memory in majority of
configurations.

Link: https://lkml.kernel.org/r/20250519171805.1288393-1-rppt@kernel.org
Fixes: e120d1bc12da ("arch, mm: set high_memory in free_area_init()")
Signed-off-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
Reported-by: Pratyush Yadav <ptyadav@amazon.de>
Tested-by: Pratyush Yadav <ptyadav@amazon.de>
Tested-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
3 weeks agoMAINTAINERS: add mm memory policy section
Lorenzo Stoakes [Thu, 15 May 2025 19:13:58 +0000 (20:13 +0100)]
MAINTAINERS: add mm memory policy section

As part of the ongoing efforts to sub-divide memory management
maintainership and reviewership, establish a section for memory policy and
migration and add appropriate maintainers and reviewers.

[lorenzo.stoakes@oracle.com: add Ying as reviewer]
Link: https://lkml.kernel.org/r/ed6f0fc2-5608-4eea-b1be-07e3e19be263@lucifer.local
Link: https://lkml.kernel.org/r/20250515191358.205684-1-lorenzo.stoakes@oracle.com
Signed-off-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Reviewed-by: Rakie Kim <rakie.kim@sk.com>
Acked-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Joshua Hahn <joshua.hahnjy@gmail.com>
Acked-by: David Hildenbrand <david@redhat.com>
Acked-by: Zi Yan <ziy@nvidia.com>
Acked-by: Huang Ying <ying.huang@linux.alibaba.com>
Acked-by: Byungchul Park <byungchul@sk.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Gregory Price <gourry@gourry.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
3 weeks agoMAINTAINERS: add mm ksm section
Lorenzo Stoakes [Thu, 15 May 2025 19:04:04 +0000 (20:04 +0100)]
MAINTAINERS: add mm ksm section

As part of the ongoing efforts to sub-divide memory management
maintainership and reviewership, establish a section for Kernel Samepage
Merging (KSM) and add appropriate maintainers and reviewers.

Link: https://lkml.kernel.org/r/20250515190404.203596-1-lorenzo.stoakes@oracle.com
Signed-off-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Acked-by: David Hildenbrand <david@redhat.com>
Acked-by: Chengming Zhou <chengming.zhou@linux.dev>
Acked-by: Xu Xin <xu.xin16@zte.com.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
3 weeks agokasan: avoid sleepable page allocation from atomic context
Alexander Gordeev [Thu, 15 May 2025 13:55:38 +0000 (15:55 +0200)]
kasan: avoid sleepable page allocation from atomic context

apply_to_pte_range() enters the lazy MMU mode and then invokes
kasan_populate_vmalloc_pte() callback on each page table walk iteration.
However, the callback can go into sleep when trying to allocate a single
page, e.g.  if an architecutre disables preemption on lazy MMU mode enter.

On s390 if make arch_enter_lazy_mmu_mode() -> preempt_enable() and
arch_leave_lazy_mmu_mode() -> preempt_disable(), such crash occurs:

[    0.663336] BUG: sleeping function called from invalid context at ./include/linux/sched/mm.h:321
[    0.663348] in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 2, name: kthreadd
[    0.663358] preempt_count: 1, expected: 0
[    0.663366] RCU nest depth: 0, expected: 0
[    0.663375] no locks held by kthreadd/2.
[    0.663383] Preemption disabled at:
[    0.663386] [<0002f3284cbb4eda>] apply_to_pte_range+0xfa/0x4a0
[    0.663405] CPU: 0 UID: 0 PID: 2 Comm: kthreadd Not tainted 6.15.0-rc5-gcc-kasan-00043-gd76bb1ebb558-dirty #162 PREEMPT
[    0.663408] Hardware name: IBM 3931 A01 701 (KVM/Linux)
[    0.663409] Call Trace:
[    0.663410]  [<0002f3284c385f58>] dump_stack_lvl+0xe8/0x140
[    0.663413]  [<0002f3284c507b9e>] __might_resched+0x66e/0x700
[    0.663415]  [<0002f3284cc4f6c0>] __alloc_frozen_pages_noprof+0x370/0x4b0
[    0.663419]  [<0002f3284ccc73c0>] alloc_pages_mpol+0x1a0/0x4a0
[    0.663421]  [<0002f3284ccc8518>] alloc_frozen_pages_noprof+0x88/0xc0
[    0.663424]  [<0002f3284ccc8572>] alloc_pages_noprof+0x22/0x120
[    0.663427]  [<0002f3284cc341ac>] get_free_pages_noprof+0x2c/0xc0
[    0.663429]  [<0002f3284cceba70>] kasan_populate_vmalloc_pte+0x50/0x120
[    0.663433]  [<0002f3284cbb4ef8>] apply_to_pte_range+0x118/0x4a0
[    0.663435]  [<0002f3284cbc7c14>] apply_to_pmd_range+0x194/0x3e0
[    0.663437]  [<0002f3284cbc99be>] __apply_to_page_range+0x2fe/0x7a0
[    0.663440]  [<0002f3284cbc9e88>] apply_to_page_range+0x28/0x40
[    0.663442]  [<0002f3284ccebf12>] kasan_populate_vmalloc+0x82/0xa0
[    0.663445]  [<0002f3284cc1578c>] alloc_vmap_area+0x34c/0xc10
[    0.663448]  [<0002f3284cc1c2a6>] __get_vm_area_node+0x186/0x2a0
[    0.663451]  [<0002f3284cc1e696>] __vmalloc_node_range_noprof+0x116/0x310
[    0.663454]  [<0002f3284cc1d950>] __vmalloc_node_noprof+0xd0/0x110
[    0.663457]  [<0002f3284c454b88>] alloc_thread_stack_node+0xf8/0x330
[    0.663460]  [<0002f3284c458d56>] dup_task_struct+0x66/0x4d0
[    0.663463]  [<0002f3284c45be90>] copy_process+0x280/0x4b90
[    0.663465]  [<0002f3284c460940>] kernel_clone+0xd0/0x4b0
[    0.663467]  [<0002f3284c46115e>] kernel_thread+0xbe/0xe0
[    0.663469]  [<0002f3284c4e440e>] kthreadd+0x50e/0x7f0
[    0.663472]  [<0002f3284c38c04a>] __ret_from_fork+0x8a/0xf0
[    0.663475]  [<0002f3284ed57ff2>] ret_from_fork+0xa/0x38

Instead of allocating single pages per-PTE, bulk-allocate the shadow
memory prior to applying kasan_populate_vmalloc_pte() callback on a page
range.

Link: https://lkml.kernel.org/r/c61d3560297c93ed044f0b1af085610353a06a58.1747316918.git.agordeev@linux.ibm.com
Fixes: 3c5c3cfb9ef4 ("kasan: support backing vmalloc space with real shadow memory")
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
Suggested-by: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Reviewed-by: Harry Yoo <harry.yoo@oracle.com>
Cc: Daniel Axtens <dja@axtens.net>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
3 weeks agohighmem: add folio_test_partial_kmap()
Matthew Wilcox (Oracle) [Wed, 14 May 2025 17:06:02 +0000 (18:06 +0100)]
highmem: add folio_test_partial_kmap()

In commit c749d9b7ebbc ("iov_iter: fix copy_page_from_iter_atomic() if
KMAP_LOCAL_FORCE_MAP"), Hugh correctly noted that if KMAP_LOCAL_FORCE_MAP
is enabled, we must limit ourselves to PAGE_SIZE bytes per call to
kmap_local().  The same problem exists in memcpy_from_folio(),
memcpy_to_folio(), folio_zero_tail(), folio_fill_tail() and
memcpy_from_file_folio(), so add folio_test_partial_kmap() to do this more
succinctly.

Link: https://lkml.kernel.org/r/20250514170607.3000994-2-willy@infradead.org
Fixes: 00cdf76012ab ("mm: add memcpy_from_file_folio()")
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Hugh Dickins <hughd@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
3 weeks agoMAINTAINERS: add hung-task detector section
Lance Yang [Tue, 13 May 2025 05:22:34 +0000 (13:22 +0800)]
MAINTAINERS: add hung-task detector section

The hung-task detector is missing in MAINTAINERS.  While it's been quiet
recently, I'm actively working on it and volunteering to review patches.
Adding this section will make it easier for contributors to know who to
contact.

Link: https://lkml.kernel.org/r/20250513052234.46463-1-lance.yang@linux.dev
Signed-off-by: Lance Yang <lance.yang@linux.dev>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
3 weeks agotaskstats: fix struct taskstats breaks backward compatibility since version 15
Wang Yaxin [Sat, 10 May 2025 07:54:13 +0000 (15:54 +0800)]
taskstats: fix struct taskstats breaks backward compatibility since version 15

Problem
========
commit 658eb5ab916d ("delayacct: add delay max to record delay peak")
  - adding more fields
commit f65c64f311ee ("delayacct: add delay min to record delay peak")
  - adding more fields
commit b016d0873777 ("taskstats: modify taskstats version")
 - version bump to 15

Since version 15 (TASKSTATS_VERSION=15) the new layout of the structure
adds fields in the middle of the structure, rendering all old software
incompatible with newer kernels and software compiled against the new
kernel headers incompatible with older kernels.

Solution
=========
move delay max and delay min to the end of taskstat, and bump
the version to 16 after the change

[wang.yaxin@zte.com.cn: adjust indentation]
Link: https://lkml.kernel.org/r/202505192131489882NSciXV4EGd8zzjLuwoOK@zte.com.cn
Link: https://lkml.kernel.org/r/20250510155413259V4JNRXxukdDgzsaL0Fo6a@zte.com.cn
Fixes: f65c64f311ee ("delayacct: add delay min to record delay peak")
Signed-off-by: Wang Yaxin <wang.yaxin@zte.com.cn>
Signed-off-by: xu xin <xu.xin16@zte.com.cn>
Signed-off-by: Kun Jiang <jiang.kun2@zte.com.cn>
Reviewed-by: Yang Yang <yang.yang29@zte.com.cn>
Cc: Balbir Singh <bsingharora@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
3 weeks agomm/truncate: fix out-of-bounds when doing a right-aligned split
Zhang Yi [Mon, 12 May 2025 06:28:25 +0000 (14:28 +0800)]
mm/truncate: fix out-of-bounds when doing a right-aligned split

When performing a right split on a folio, the split_at2 may point to a
not-present page if the offset + length equals the original folio size,
which will trigger the following error:

 BUG: unable to handle page fault for address: ffffea0006000008
 #PF: supervisor read access in kernel mode
 #PF: error_code(0x0000) - not-present page
 PGD 143ffb9067 P4D 143ffb9067 PUD 143ffb8067 PMD 0
 Oops: Oops: 0000 [#1] SMP PTI
 CPU: 0 UID: 0 PID: 502640 Comm: fsx Not tainted 6.15.0-rc3-gc6156189fc6b #889 PR
 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.3-2.fc40 04/01/4
 RIP: 0010:truncate_inode_partial_folio+0x208/0x620
 Code: ff 03 48 01 da e8 78 7e 13 00 48 83 05 10 b5 5a 0c 01 85 c0 0f 85 1c 02 001
 RSP: 0018:ffffc90005bafab0 EFLAGS: 00010286
 RAX: 0000000000000000 RBX: ffffea0005ffff00 RCX: 0000000000000002
 RDX: 000000000000000c RSI: 0000000000013975 RDI: ffffc90005bafa30
 RBP: ffffea0006000000 R08: 0000000000000000 R09: 00000000000009bf
 R10: 00000000000007e0 R11: 0000000000000000 R12: 0000000000001633
 R13: 0000000000000000 R14: ffffea0005ffff00 R15: fffffffffffffffe
 FS:  00007f9f9a161740(0000) GS:ffff8894971fd000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: ffffea0006000008 CR3: 000000017c2ae000 CR4: 00000000000006f0
 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
 Call Trace:
  <TASK>
  truncate_inode_pages_range+0x226/0x720
  truncate_pagecache+0x57/0x90
  ...

Fix this issue by skipping the split if truncation aligns with the folio
size, make sure the split page number lies within the folio.

Link: https://lkml.kernel.org/r/20250512062825.3533342-1-yi.zhang@huaweicloud.com
Fixes: 7460b470a131 ("mm/truncate: use folio_split() in truncate operation")
Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
Reviewed-by: Zi Yan <ziy@nvidia.com>
Cc: ErKun Yang <yangerkun@huawei.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
3 weeks agoMAINTAINERS: add mm reclaim section
Lorenzo Stoakes [Mon, 12 May 2025 14:31:22 +0000 (15:31 +0100)]
MAINTAINERS: add mm reclaim section

In furtherance of ongoing efforts to ensure people are aware of who
de-facto maintains/has an interest in specific parts of mm, as well trying
to avoid get_maintainers.pl listing only Andrew and the mailing list for
mm files - establish a reclaim memory management section and add relevant
maintainers/reviewers.

This is a key part of memory management so sensibly deserves its own
section.

This encompasses both 'classical' reclaim and MGLRU and thus reflects this
in the reviewers from both, as well as those who have contributed
specifically on the memcg side of things.

Link: https://lkml.kernel.org/r/20250512143122.87740-1-lorenzo.stoakes@oracle.com
Signed-off-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Acked-by: Qi Zheng <zhengqi.arch@bytedance.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Shakeel Butt <shakeel.butt@linux.dev>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
3 weeks agoMAINTAINERS: update page allocator section
Lorenzo Stoakes [Mon, 12 May 2025 14:46:03 +0000 (15:46 +0100)]
MAINTAINERS: update page allocator section

Make Vlastimil maintainer of this section (with thanks to Vlastimil for
agreeing to this!) and add page isolation files for which this section
seem most appropriate.

We may wish to, in future, refactor/rename some of these files to more
logically fit what is actually being performed, but for the time being
this seems the most sensible place.

Additionally, fix the alphabetical ordering of files.

Link: https://lkml.kernel.org/r/20250512144603.90379-1-lorenzo.stoakes@oracle.com
Signed-off-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Zi Yan <ziy@nvidia.com>
Cc: Brendan Jackman <jackmanb@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
3 weeks agomm: fix VM_UFFD_MINOR == VM_SHADOW_STACK on USERFAULTFD=y && ARM64_GCS=y
Florent Revest [Wed, 7 May 2025 13:09:57 +0000 (15:09 +0200)]
mm: fix VM_UFFD_MINOR == VM_SHADOW_STACK on USERFAULTFD=y && ARM64_GCS=y

On configs with CONFIG_ARM64_GCS=y, VM_SHADOW_STACK is bit 38.  On configs
with CONFIG_HAVE_ARCH_USERFAULTFD_MINOR=y (selected by CONFIG_ARM64 when
CONFIG_USERFAULTFD=y), VM_UFFD_MINOR is _also_ bit 38.

This bit being shared by two different VMA flags could lead to all sorts
of unintended behaviors.  Presumably, a process could maybe call into
userfaultfd in a way that disables the shadow stack vma flag.  I can't
think of any attack where this would help (presumably, if an attacker
tries to disable shadow stacks, they are trying to hijack control flow so
can't arbitrarily call into userfaultfd yet anyway) but this still feels
somewhat scary.

Link: https://lkml.kernel.org/r/20250507131000.1204175-2-revest@chromium.org
Fixes: ae80e1629aea ("mm: Define VM_SHADOW_STACK for arm64 when we support GCS")
Signed-off-by: Florent Revest <revest@chromium.org>
Reviewed-by: Mark Brown <broonie@kernel.org>
Cc: Borislav Betkov <bp@alien8.de>
Cc: Brendan Jackman <jackmanb@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Florent Revest <revest@chromium.org>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Thiago Jung Bauermann <thiago.bauermann@linaro.org>
Cc: Thomas Gleinxer <tglx@linutronix.de>
Cc: Will Deacon <will@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
3 weeks agomm: mmap: map MAP_STACK to VM_NOHUGEPAGE only if THP is enabled
Ignacio Moreno Gonzalez [Wed, 7 May 2025 13:28:06 +0000 (15:28 +0200)]
mm: mmap: map MAP_STACK to VM_NOHUGEPAGE only if THP is enabled

commit c4608d1bf7c6 ("mm: mmap: map MAP_STACK to VM_NOHUGEPAGE") maps the
mmap option MAP_STACK to VM_NOHUGEPAGE.  This is also done if
CONFIG_TRANSPARENT_HUGEPAGE is not defined.  But in that case, the
VM_NOHUGEPAGE does not make sense.

I discovered this issue when trying to use the tool CRIU to checkpoint and
restore a container.  Our running kernel is compiled without
CONFIG_TRANSPARENT_HUGEPAGE.  CRIU parses the output of /proc/<pid>/smaps
and saves the "nh" flag.  When trying to restore the container, CRIU fails
to restore the "nh" mappings, since madvise() MADV_NOHUGEPAGE always
returns an error because CONFIG_TRANSPARENT_HUGEPAGE is not defined.

Link: https://lkml.kernel.org/r/20250507-map-map_stack-to-vm_nohugepage-only-if-thp-is-enabled-v5-1-c6c38cfefd6e@kuka.com
Fixes: c4608d1bf7c6 ("mm: mmap: map MAP_STACK to VM_NOHUGEPAGE")
Signed-off-by: Ignacio Moreno Gonzalez <Ignacio.MorenoGonzalez@kuka.com>
Acked-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Reviewed-by: Yang Shi <yang@os.amperecomputing.com>
Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
3 weeks agoMAINTAINERS: add myself as vmalloc co-maintainer
Uladzislau Rezki (Sony) [Wed, 7 May 2025 15:02:57 +0000 (17:02 +0200)]
MAINTAINERS: add myself as vmalloc co-maintainer

I have been working on the vmalloc code for several years, contributing to
improvements and fixes.  Add myself as co-maintainer ("M") alongside
Andrew Morton.

Link: https://lkml.kernel.org/r/20250507150257.61485-1-urezki@gmail.com
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Acked-by: David Hildenbrand <david@redhat.com>
Acked-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Acked-by: Baoquan He <bhe@redhat.com>
Cc: Christop Hellwig <hch@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
3 weeks agomm/page_alloc.c: avoid infinite retries caused by cpuset race
Tianyang Zhang [Wed, 16 Apr 2025 08:24:05 +0000 (16:24 +0800)]
mm/page_alloc.c: avoid infinite retries caused by cpuset race

__alloc_pages_slowpath has no change detection for ac->nodemask in the
part of retry path, while cpuset can modify it in parallel.  For some
processes that set mempolicy as MPOL_BIND, this results ac->nodemask
changes, and then the should_reclaim_retry will judge based on the latest
nodemask and jump to retry, while the get_page_from_freelist only
traverses the zonelist from ac->preferred_zoneref, which selected by a
expired nodemask and may cause infinite retries in some cases

cpu 64:
__alloc_pages_slowpath {
        /* ..... */
retry:
        /* ac->nodemask = 0x1, ac->preferred->zone->nid = 1 */
        if (alloc_flags & ALLOC_KSWAPD)
                wake_all_kswapds(order, gfp_mask, ac);
        /* cpu 1:
        cpuset_write_resmask
            update_nodemask
                update_nodemasks_hier
                    update_tasks_nodemask
                        mpol_rebind_task
                         mpol_rebind_policy
                          mpol_rebind_nodemask
// mempolicy->nodes has been modified,
// which ac->nodemask point to

        */
        /* ac->nodemask = 0x3, ac->preferred->zone->nid = 1 */
        if (should_reclaim_retry(gfp_mask, order, ac, alloc_flags,
                                 did_some_progress > 0, &no_progress_loops))
                goto retry;
}

Simultaneously starting multiple cpuset01 from LTP can quickly reproduce
this issue on a multi node server when the maximum memory pressure is
reached and the swap is enabled

Link: https://lkml.kernel.org/r/20250416082405.20988-1-zhangtianyang@loongson.cn
Fixes: c33d6c06f60f ("mm, page_alloc: avoid looking up the first zone in a zonelist twice")
Signed-off-by: Tianyang Zhang <zhangtianyang@loongson.cn>
Reviewed-by: Suren Baghdasaryan <surenb@google.com>
Reviewed-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Brendan Jackman <jackmanb@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Zi Yan <ziy@nvidia.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
4 weeks agomm: userfaultfd: correct dirty flags set for both present and swap pte
Barry Song [Thu, 8 May 2025 22:09:12 +0000 (10:09 +1200)]
mm: userfaultfd: correct dirty flags set for both present and swap pte

As David pointed out, what truly matters for mremap and userfaultfd move
operations is the soft dirty bit.  The current comment and
implementation—which always sets the dirty bit for present PTEs and
fails to set the soft dirty bit for swap PTEs—are incorrect.  This could
break features like Checkpoint-Restore in Userspace (CRIU).

This patch updates the behavior to correctly set the soft dirty bit for
both present and swap PTEs in accordance with mremap.

Link: https://lkml.kernel.org/r/20250508220912.7275-1-21cnbao@gmail.com
Fixes: adef440691ba ("userfaultfd: UFFDIO_MOVE uABI")
Signed-off-by: Barry Song <v-songbaohua@oppo.com>
Reported-by: David Hildenbrand <david@redhat.com>
Closes: https://lore.kernel.org/linux-mm/02f14ee1-923f-47e3-a994-4950afb9afcc@redhat.com/
Acked-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Suren Baghdasaryan <surenb@google.com>
Cc: Lokesh Gidra <lokeshgidra@google.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
4 weeks agozsmalloc: don't underflow size calculation in zs_obj_write()
Sergey Senozhatsky [Wed, 7 May 2025 05:42:24 +0000 (14:42 +0900)]
zsmalloc: don't underflow size calculation in zs_obj_write()

Do not mix class->size and object size during offsets/sizes calculation in
zs_obj_write().  Size classes can merge into clusters, based on
objects-per-zspage and pages-per-zspage characteristics, so some size
classes can store objects smaller than class->size.  This becomes
problematic when object size is much smaller than class->size.  zsmalloc
can falsely decide that object spans two physical pages, because a larger
class->size value is used for that check, while the actual object is much
smaller and fits the free space of the first physical page, so there is
nothing to write to the second page and memcpy() size calculation
underflows.

 Unable to handle kernel paging request at virtual address ffffc00081ff4000
 pc : __memcpy+0x10/0x24
 lr : zs_obj_write+0x1b0/0x1d0 [zsmalloc]
 Call trace:
  __memcpy+0x10/0x24 (P)
  zram_write_page+0x150/0x4fc [zram]
  zram_submit_bio+0x5e0/0x6a4 [zram]
  __submit_bio+0x168/0x220
  submit_bio_noacct_nocheck+0x128/0x2c8
  submit_bio_noacct+0x19c/0x2f8

This is mostly seen on system with larger page-sizes, because size class
cluters of such systems hold wider size ranges than on 4K PAGE_SIZE
systems.

Assume a 16K PAGE_SIZE system, a write of 820 bytes object to a 864-bytes
size class at offset 15560.  15560 + 864 is more than 16384 so zsmalloc
attempts to memcpy() it to two physical pages.  However, 16384 - 15560 =
824 which is more than 820, so the object in fact doesn't span two
physical pages, and there is no data to write to the second physical page.

We always know the exact size in bytes of the object that we are about to
write (store), so use it instead of class->size.

Link: https://lkml.kernel.org/r/20250507054312.4135983-1-senozhatsky@chromium.org
Fixes: 44f76413496e ("zsmalloc: introduce new object mapping API")
Signed-off-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Reported-by: Igor Belousov <igor.b@beldev.am>
Tested-by: Igor Belousov <igor.b@beldev.am>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
4 weeks agomm/page_alloc: fix race condition in unaccepted memory handling
Kirill A. Shutemov [Tue, 6 May 2025 13:32:07 +0000 (16:32 +0300)]
mm/page_alloc: fix race condition in unaccepted memory handling

The page allocator tracks the number of zones that have unaccepted memory
using static_branch_enc/dec() and uses that static branch in hot paths to
determine if it needs to deal with unaccepted memory.

Borislav and Thomas pointed out that the tracking is racy: operations on
static_branch are not serialized against adding/removing unaccepted pages
to/from the zone.

Sanity checks inside static_branch machinery detects it:

WARNING: CPU: 0 PID: 10 at kernel/jump_label.c:276 __static_key_slow_dec_cpuslocked+0x8e/0xa0

The comment around the WARN() explains the problem:

/*
 * Warn about the '-1' case though; since that means a
 * decrement is concurrent with a first (0->1) increment. IOW
 * people are trying to disable something that wasn't yet fully
 * enabled. This suggests an ordering problem on the user side.
 */

The effect of this static_branch optimization is only visible on
microbenchmark.

Instead of adding more complexity around it, remove it altogether.

Link: https://lkml.kernel.org/r/20250506133207.1009676-1-kirill.shutemov@linux.intel.com
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Fixes: dcdfdd40fa82 ("mm: Add support for unaccepted memory")
Link: https://lore.kernel.org/all/20250506092445.GBaBnVXXyvnazly6iF@fat_crate.local
Reported-by: Borislav Petkov <bp@alien8.de>
Tested-by: Borislav Petkov (AMD) <bp@alien8.de>
Reported-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Brendan Jackman <jackmanb@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: <stable@vger.kernel.org> [6.5+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
4 weeks agomm/page_alloc: ensure try_alloc_pages() plays well with unaccepted memory
Kirill A. Shutemov [Tue, 6 May 2025 11:25:08 +0000 (14:25 +0300)]
mm/page_alloc: ensure try_alloc_pages() plays well with unaccepted memory

try_alloc_pages() will not attempt to allocate memory if the system has
*any* unaccepted memory.  Memory is accepted as needed and can remain in
the system indefinitely, causing the interface to always fail.

Rather than immediately giving up, attempt to use already accepted memory
on free lists.

Pass 'alloc_flags' to cond_accept_memory() and do not accept new memory
for ALLOC_TRYLOCK requests.

Found via code inspection - only BPF uses this at present and the
runtime effects are unclear.

Link: https://lkml.kernel.org/r/20250506112509.905147-2-kirill.shutemov@linux.intel.com
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Fixes: 97769a53f117 ("mm, bpf: Introduce try_alloc_pages() for opportunistic page allocation")
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Brendan Jackman <jackmanb@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
4 weeks agoMAINTAINERS: add mm GUP section
Lorenzo Stoakes [Tue, 6 May 2025 17:36:01 +0000 (18:36 +0100)]
MAINTAINERS: add mm GUP section

As part of the ongoing efforts to sub-divide memory management
maintainership and reviewership, establish a section for GUP (Get User
Pages) support and add appropriate maintainers and reviewers.

Link: https://lkml.kernel.org/r/20250506173601.97562-1-lorenzo.stoakes@oracle.com
Signed-off-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Reviewed-by: John Hubbard <jhubbard@nvidia.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Cc: Peter Xu <peterx@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
4 weeks agomm/codetag: move tag retrieval back upfront in __free_pages()
David Wang [Mon, 5 May 2025 19:30:34 +0000 (03:30 +0800)]
mm/codetag: move tag retrieval back upfront in __free_pages()

Commit 51ff4d7486f0 ("mm: avoid extra mem_alloc_profiling_enabled()
 checks") introduces a possible use-after-free scenario, when page
is non-compound, page[0] could be released by other thread right
after put_page_testzero failed in current thread, pgalloc_tag_sub_pages
afterwards would manipulate an invalid page for accounting remaining
pages:

[timeline]   [thread1]                     [thread2]
  |          alloc_page non-compound
  V
  |                                        get_page, rf counter inc
  V
  |          in ___free_pages
  |          put_page_testzero fails
  V
  |                                        put_page, page released
  V
  |          in ___free_pages,
  |          pgalloc_tag_sub_pages
  |          manipulate an invalid page
  V

Restore __free_pages() to its state before, retrieve alloc tag
beforehand.

Link: https://lkml.kernel.org/r/20250505193034.91682-1-00107082@163.com
Fixes: 51ff4d7486f0 ("mm: avoid extra mem_alloc_profiling_enabled() checks")
Signed-off-by: David Wang <00107082@163.com>
Acked-by: Suren Baghdasaryan <surenb@google.com>
Reviewed-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Brendan Jackman <jackmanb@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
4 weeks agomm/memory: fix mapcount / refcount sanity check for mTHP reuse
Kairui Song [Fri, 25 Apr 2025 07:43:25 +0000 (15:43 +0800)]
mm/memory: fix mapcount / refcount sanity check for mTHP reuse

The following WARNING was triggered during swap stress test with mTHP
enabled:

[ 6609.335758] ------------[ cut here ]------------
[ 6609.337758] WARNING: CPU: 82 PID: 755116 at mm/memory.c:3794 do_wp_page+0x1084/0x10e0
[ 6609.340922] Modules linked in: zram virtiofs
[ 6609.342699] CPU: 82 UID: 0 PID: 755116 Comm: sh Kdump: loaded Not tainted 6.15.0-rc1+ #1429 PREEMPT(voluntary)
[ 6609.347620] Hardware name: Red Hat KVM/RHEL-AV, BIOS 0.0.0 02/06/2015
[ 6609.349909] RIP: 0010:do_wp_page+0x1084/0x10e0
[ 6609.351532] Code: ff ff 48 c7 c6 80 ba 49 82 4c 89 ef e8 95 fd fe ff 0f 0b bd f5 ff ff ff e9 43 fb ff ff 41 83 a9 bc 12 00 00 01 e9 5c fb ff ff <0f> 0b e9 a6 fc ff ff 65 ff 00 f0 48 0f b
a 6d 00 1f 0f 83 82 fc ff
[ 6609.357959] RSP: 0000:ffffc90002273d40 EFLAGS: 00010287
[ 6609.359915] RAX: 000000000000000f RBX: 0000000000000000 RCX: 000fffffffe00000
[ 6609.362606] RDX: 0000000000000010 RSI: 000055a119ac1000 RDI: ffffea000ae6ec00
[ 6609.365143] RBP: ffffea000ae6ec68 R08: 84000002b9bb1025 R09: 000055a119ab6000
[ 6609.367569] R10: ffff8881caa2ad80 R11: 0000000000000000 R12: ffff8881caa2ad80
[ 6609.370255] R13: ffffea000ae6ec00 R14: 000055a119ac1c9c R15: ffffc90002273dd8
[ 6609.373007] FS:  00007f08e467f740(0000) GS:ffff88a07c214000(0000) knlGS:0000000000000000
[ 6609.375999] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 6609.377946] CR2: 000055a119ac1c9c CR3: 00000001adfd6005 CR4: 0000000000770eb0
[ 6609.380376] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 6609.382853] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[ 6609.385216] PKRU: 55555554
[ 6609.386141] Call Trace:
[ 6609.387017]  <TASK>
[ 6609.387718]  ? ___pte_offset_map+0x1b/0x110
[ 6609.389056]  __handle_mm_fault+0xa51/0xf00
[ 6609.390363]  ? exc_page_fault+0x6a/0x140
[ 6609.391629]  handle_mm_fault+0x13d/0x360
[ 6609.392856]  do_user_addr_fault+0x2f2/0x7f0
[ 6609.394160]  ? sigprocmask+0x77/0xa0
[ 6609.395375]  exc_page_fault+0x6a/0x140
[ 6609.396735]  asm_exc_page_fault+0x26/0x30
[ 6609.398224] RIP: 0033:0x55a1050bc18b
[ 6609.399567] Code: 8b 3f 4d 85 ff 74 40 41 39 5f 18 75 f2 49 8b 7f 08 44 38 27 75 e9 4c 89 c6 4c 89 45 c8 e8 bd 83 fa ff 4c 8b 45 c8 85 c0 75 d5 <41> 83 47 1c 01 48 83 c4 28 4c 89 f8 5b 4
1 5c 41 5d 41 5e 41 5f 5d
[ 6609.405971] RSP: 002b:00007ffcf5f37d90 EFLAGS: 00010246
[ 6609.407737] RAX: 0000000000000000 RBX: 00000000182768fa RCX: 0000000000000000
[ 6609.410151] RDX: 00000000000000fa RSI: 000055a105175c7b RDI: 000055a119ac1c60
[ 6609.412606] RBP: 00007ffcf5f37de0 R08: 000055a105175c7b R09: 0000000000000000
[ 6609.414998] R10: 000000004d2dfb5a R11: 0000000000000246 R12: 0000000000000050
[ 6609.417193] R13: 00000000000000fa R14: 000055a119abaf60 R15: 000055a119ac1c80
[ 6609.419268]  </TASK>
[ 6609.419928] ---[ end trace 0000000000000000 ]---

The WARN_ON here is simply incorrect.  The refcount here must be at least
the mapcount, not the opposite.  Each mapcount must have a corresponding
refcount, but the refcount may increase if other components grab the
folio, which is acceptable.  Meanwhile, having a mapcount larger than
refcount is a real problem.

So fix the WARN_ON condition.

Link: https://lkml.kernel.org/r/20250425074325.61833-1-ryncsn@gmail.com
Fixes: 1da190f4d0a6 ("mm: Copy-on-Write (COW) reuse support for PTE-mapped THP")
Signed-off-by: Kairui Song <kasong@tencent.com>
Reported-by: Kairui Song <kasong@tencent.com>
Closes: https://lore.kernel.org/all/CAMgjq7D+ea3eg9gRCVvRnto3Sv3_H3WVhupX4e=k8T5QAfBHbw@mail.gmail.com/
Suggested-by: David Hildenbrand <david@redhat.com>
Acked-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
4 weeks agokernel/fork: only call untrack_pfn_clear() on VMAs duplicated for fork()
David Hildenbrand [Tue, 22 Apr 2025 14:49:42 +0000 (16:49 +0200)]
kernel/fork: only call untrack_pfn_clear() on VMAs duplicated for fork()

Not intuitive, but vm_area_dup() located in kernel/fork.c is not only used
for duplicating VMAs during fork(), but also for duplicating VMAs when
splitting VMAs or when mremap()'ing them.

VM_PFNMAP mappings can at least get ordinarily mremap()'ed (no change in
size) and apparently also shrunk during mremap(), which implies
duplicating the VMA in __split_vma() first.

In case of ordinary mremap() (no change in size), we first duplicate the
VMA in copy_vma_and_data()->copy_vma() to then call untrack_pfn_clear() on
the old VMA: we effectively move the VM_PAT reservation.  So the
untrack_pfn_clear() call on the new VMA duplicating is wrong in that
context.

Splitting of VMAs seems problematic, because we don't duplicate/adjust the
reservation when splitting the VMA.  Instead, in memtype_erase() -- called
during zapping/munmap -- we shrink a reservation in case only the end
address matches: Assume we split a VMA into A and B, both would share a
reservation until B is unmapped.

So when unmapping B, the reservation would be updated to cover only A.
When unmapping A, we would properly remove the now-shrunk reservation.
That scenario describes the mremap() shrinking (old_size > new_size),
where we split + unmap B, and the untrack_pfn_clear() on the new VMA when
is wrong.

What if we manage to split a VM_PFNMAP VMA into A and B and unmap A first?
It would be broken because we would never free the reservation.  Likely,
there are ways to trigger such a VMA split outside of mremap().

Affecting other VMA duplication was not intended, vm_area_dup() being used
outside of kernel/fork.c was an oversight.  So let's fix that for; how to
handle VMA splits better should be investigated separately.

With a simple reproducer that uses mprotect() to split such a VMA I can
trigger

x86/PAT: pat_mremap:26448 freeing invalid memtype [mem 0x00000000-0x00000fff]

Link: https://lkml.kernel.org/r/20250422144942.2871395-1-david@redhat.com
Fixes: dc84bc2aba85 ("x86/mm/pat: Fix VM_PAT handling when fork() fails in copy_page_range()")
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Rik van Riel <riel@surriel.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
4 weeks agomm: hugetlb: fix incorrect fallback for subpool
Wupeng Ma [Thu, 10 Apr 2025 06:26:33 +0000 (14:26 +0800)]
mm: hugetlb: fix incorrect fallback for subpool

During our testing with hugetlb subpool enabled, we observe that
hstate->resv_huge_pages may underflow into negative values.  Root cause
analysis reveals a race condition in subpool reservation fallback handling
as follow:

hugetlb_reserve_pages()
    /* Attempt subpool reservation */
    gbl_reserve = hugepage_subpool_get_pages(spool, chg);

    /* Global reservation may fail after subpool allocation */
    if (hugetlb_acct_memory(h, gbl_reserve) < 0)
        goto out_put_pages;

out_put_pages:
    /* This incorrectly restores reservation to subpool */
    hugepage_subpool_put_pages(spool, chg);

When hugetlb_acct_memory() fails after subpool allocation, the current
implementation over-commits subpool reservations by returning the full
'chg' value instead of the actual allocated 'gbl_reserve' amount.  This
discrepancy propagates to global reservations during subsequent releases,
eventually causing resv_huge_pages underflow.

This problem can be trigger easily with the following steps:
1. reverse hugepage for hugeltb allocation
2. mount hugetlbfs with min_size to enable hugetlb subpool
3. alloc hugepages with two task(make sure the second will fail due to
   insufficient amount of hugepages)
4. with for a few seconds and repeat step 3 which will make
   hstate->resv_huge_pages to go below zero.

To fix this problem, return corrent amount of pages to subpool during the
fallback after hugepage_subpool_get_pages is called.

Link: https://lkml.kernel.org/r/20250410062633.3102457-1-mawupeng1@huawei.com
Fixes: 1c5ecae3a93f ("hugetlbfs: add minimum size accounting to subpools")
Signed-off-by: Wupeng Ma <mawupeng1@huawei.com>
Tested-by: Joshua Hahn <joshua.hahnjy@gmail.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Cc: David Hildenbrand <david@redhat.com>
Cc: Ma Wupeng <mawupeng1@huawei.com>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
4 weeks agoLinux 6.15-rc6 v6.15-rc6
Linus Torvalds [Sun, 11 May 2025 21:54:11 +0000 (14:54 -0700)]
Linux 6.15-rc6

4 weeks agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Sun, 11 May 2025 18:30:13 +0000 (11:30 -0700)]
Merge tag 'for-linus' of git://git./virt/kvm/kvm

Pull KVM fixes from Paolo Bonzini:
 "ARM:

   - Avoid use of uninitialized memcache pointer in user_mem_abort()

   - Always set HCR_EL2.xMO bits when running in VHE, allowing
     interrupts to be taken while TGE=0 and fixing an ugly bug on
     AmpereOne that occurs when taking an interrupt while clearing the
     xMO bits (AC03_CPU_36)

   - Prevent VMMs from hiding support for AArch64 at any EL virtualized
     by KVM

   - Save/restore the host value for HCRX_EL2 instead of restoring an
     incorrect fixed value

   - Make host_stage2_set_owner_locked() check that the entire requested
     range is memory rather than just the first page

  RISC-V:

   - Add missing reset of smstateen CSRs

  x86:

   - Forcibly leave SMM on SHUTDOWN interception on AMD CPUs to avoid
     causing problems due to KVM stuffing INIT on SHUTDOWN (KVM needs to
     sanitize the VMCB as its state is undefined after SHUTDOWN,
     emulating INIT is the least awful choice).

   - Track the valid sync/dirty fields in kvm_run as a u64 to ensure KVM
     KVM doesn't goof a sanity check in the future.

   - Free obsolete roots when (re)loading the MMU to fix a bug where
     pre-faulting memory can get stuck due to always encountering a
     stale root.

   - When dumping GHCB state, use KVM's snapshot instead of the raw GHCB
     page to print state, so that KVM doesn't print stale/wrong
     information.

   - When changing memory attributes (e.g. shared <=> private), add
     potential hugepage ranges to the mmu_invalidate_range_{start,end}
     set so that KVM doesn't create a shared/private hugepage when the
     the corresponding attributes will become mixed (the attributes are
     commited *after* KVM finishes the invalidation).

   - Rework the SRSO mitigation to enable BP_SPEC_REDUCE only when KVM
     has at least one active VM. Effectively BP_SPEC_REDUCE when KVM is
     loaded led to very measurable performance regressions for non-KVM
     workloads"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: SVM: Set/clear SRSO's BP_SPEC_REDUCE on 0 <=> 1 VM count transitions
  KVM: arm64: Fix memory check in host_stage2_set_owner_locked()
  KVM: arm64: Kill HCRX_HOST_FLAGS
  KVM: arm64: Properly save/restore HCRX_EL2
  KVM: arm64: selftest: Don't try to disable AArch64 support
  KVM: arm64: Prevent userspace from disabling AArch64 support at any virtualisable EL
  KVM: arm64: Force HCR_EL2.xMO to 1 at all times in VHE mode
  KVM: arm64: Fix uninitialized memcache pointer in user_mem_abort()
  KVM: x86/mmu: Prevent installing hugepages when mem attributes are changing
  KVM: SVM: Update dump_ghcb() to use the GHCB snapshot fields
  KVM: RISC-V: reset smstateen CSRs
  KVM: x86/mmu: Check and free obsolete roots in kvm_mmu_reload()
  KVM: x86: Check that the high 32bits are clear in kvm_arch_vcpu_ioctl_run()
  KVM: SVM: Forcibly leave SMM mode on SHUTDOWN interception

4 weeks agoMerge tag 'mips-fixes_6.15_1' of git://git.kernel.org/pub/scm/linux/kernel/git/mips...
Linus Torvalds [Sun, 11 May 2025 18:19:52 +0000 (11:19 -0700)]
Merge tag 'mips-fixes_6.15_1' of git://git./linux/kernel/git/mips/linux

Pull MIPS fixes from Thomas Bogendoerfer:

 - Fix delayed timers

 - Fix NULL pointer deref

 - Fix wrong range check

* tag 'mips-fixes_6.15_1' of git://git.kernel.org/pub/scm/linux/kernel/git/mips/linux:
  MIPS: Fix MAX_REG_OFFSET
  MIPS: CPS: Fix potential NULL pointer dereferences in cps_prepare_cpus()
  MIPS: rename rollback_handler with skipover_handler
  MIPS: Move r4k_wait() to .cpuidle.text section
  MIPS: Fix idle VS timer enqueue

4 weeks agoMerge tag 'x86-urgent-2025-05-11' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 11 May 2025 18:08:55 +0000 (11:08 -0700)]
Merge tag 'x86-urgent-2025-05-11' of git://git./linux/kernel/git/tip/tip

Pull x86 fix from Ingo Molnar:
 "Fix a boot regression on very old x86 CPUs without CPUID support"

* tag 'x86-urgent-2025-05-11' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/microcode: Consolidate the loader enablement checking

4 weeks agoMerge tag 'timers-urgent-2025-05-11' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 11 May 2025 17:33:25 +0000 (10:33 -0700)]
Merge tag 'timers-urgent-2025-05-11' of git://git./linux/kernel/git/tip/tip

Pull misc timers fixes from Ingo Molnar:

 - Fix time keeping bugs in CLOCK_MONOTONIC_COARSE clocks

 - Work around absolute relocations into vDSO code that GCC erroneously
   emits in certain arm64 build environments

 - Fix a false positive lockdep warning in the i8253 clocksource driver

* tag 'timers-urgent-2025-05-11' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  clocksource/i8253: Use raw_spinlock_irqsave() in clockevent_i8253_disable()
  arm64: vdso: Work around invalid absolute relocations from GCC
  timekeeping: Prevent coarse clocks going backwards

4 weeks agoMerge tag 'input-for-v6.15-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 11 May 2025 17:29:29 +0000 (10:29 -0700)]
Merge tag 'input-for-v6.15-rc5' of git://git./linux/kernel/git/dtor/input

Pull input fixes from Dmitry Torokhov:

 - Synaptics touchpad on multiple laptops (Dynabook Portege X30L-G,
   Dynabook Portege X30-D, TUXEDO InfinityBook Pro 14 v5, Dell Precision
   M3800, HP Elitebook 850 G1) switched from PS/2 to SMBus mode

 - a number of new controllers added to xpad driver: HORI Drum
   controller, PowerA Fusion Pro 4, PowerA MOGA XP-Ultra controller,
   8BitDo Ultimate 2 Wireless Controller, 8BitDo Ultimate 3-mode
   Controller, Hyperkin DuchesS Xbox One controller

 - fixes to xpad driver to properly handle Mad Catz JOYTECH NEO SE
   Advanced and PDP Mirror's Edge Official controllers

 - fixes to xpad driver to properly handle "Share" button on some
   controllers

 - a fix for device initialization timing and for waking up the
   controller in cyttsp5 driver

 - a fix for hisi_powerkey driver to properly wake up from s2idle state

 - other assorted cleanups and fixes

* tag 'input-for-v6.15-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: xpad - fix xpad_device sorting
  Input: xpad - add support for several more controllers
  Input: xpad - fix Share button on Xbox One controllers
  Input: xpad - fix two controller table values
  Input: hisi_powerkey - enable system-wakeup for s2idle
  Input: synaptics - enable InterTouch on Dell Precision M3800
  Input: synaptics - enable InterTouch on TUXEDO InfinityBook Pro 14 v5
  Input: synaptics - enable InterTouch on Dynabook Portege X30L-G
  Input: synaptics - enable InterTouch on Dynabook Portege X30-D
  Input: synaptics - enable SMBus for HP Elitebook 850 G1
  Input: mtk-pmic-keys - fix possible null pointer dereference
  Input: xpad - add support for 8BitDo Ultimate 2 Wireless Controller
  Input: cyttsp5 - fix power control issue on wakeup
  MAINTAINERS: .mailmap: update Mattijs Korpershoek's email address
  dt-bindings: mediatek,mt6779-keypad: Update Mattijs' email address
  Input: stmpe-ts - use module alias instead of device table
  Input: cyttsp5 - ensure minimum reset pulse width
  Input: sparcspkr - avoid unannotated fall-through
  input/joystick: magellan: Mark __nonstring look-up table

4 weeks agoMerge tag 'fixes-2025-05-11' of git://git.kernel.org/pub/scm/linux/kernel/git/rppt...
Linus Torvalds [Sun, 11 May 2025 17:23:53 +0000 (10:23 -0700)]
Merge tag 'fixes-2025-05-11' of git://git./linux/kernel/git/rppt/memblock

Pull memblock fixes from Mike Rapoport:

 - Mark set_high_memory() as __init to fix section mismatch

 - Accept memory allocated in memblock_double_array() to mitigate crash
   of SNP guests

* tag 'fixes-2025-05-11' of git://git.kernel.org/pub/scm/linux/kernel/git/rppt/memblock:
  memblock: Accept allocated memory before use in memblock_double_array()
  mm,mm_init: Mark set_high_memory as __init

4 weeks agoInput: xpad - fix xpad_device sorting
Vicki Pfau [Sun, 11 May 2025 06:06:34 +0000 (23:06 -0700)]
Input: xpad - fix xpad_device sorting

A recent commit put one entry in the wrong place. This just moves it to the
right place.

Signed-off-by: Vicki Pfau <vi@endrift.com>
Link: https://lore.kernel.org/r/20250328234345.989761-5-vi@endrift.com
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
4 weeks agoInput: xpad - add support for several more controllers
Vicki Pfau [Sun, 11 May 2025 06:00:10 +0000 (23:00 -0700)]
Input: xpad - add support for several more controllers

This adds support for several new controllers, all of which include
Share buttons:

- HORI Drum controller
- PowerA Fusion Pro 4
- 8BitDo Ultimate 3-mode Controller
- Hyperkin DuchesS Xbox One controller
- PowerA MOGA XP-Ultra controller

Signed-off-by: Vicki Pfau <vi@endrift.com>
Link: https://lore.kernel.org/r/20250328234345.989761-4-vi@endrift.com
Cc: stable@vger.kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
4 weeks agoInput: xpad - fix Share button on Xbox One controllers
Vicki Pfau [Sun, 11 May 2025 05:59:25 +0000 (22:59 -0700)]
Input: xpad - fix Share button on Xbox One controllers

The Share button, if present, is always one of two offsets from the end of the
file, depending on the presence of a specific interface. As we lack parsing for
the identify packet we can't automatically determine the presence of that
interface, but we can hardcode which of these offsets is correct for a given
controller.

More controllers are probably fixable by adding the MAP_SHARE_BUTTON in the
future, but for now I only added the ones that I have the ability to test
directly.

Signed-off-by: Vicki Pfau <vi@endrift.com>
Link: https://lore.kernel.org/r/20250328234345.989761-2-vi@endrift.com
Cc: stable@vger.kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
4 weeks agoInput: xpad - fix two controller table values
Vicki Pfau [Fri, 28 Mar 2025 23:43:36 +0000 (16:43 -0700)]
Input: xpad - fix two controller table values

Two controllers -- Mad Catz JOYTECH NEO SE Advanced and PDP Mirror's
Edge Official -- were missing the value of the mapping field, and thus
wouldn't detect properly.

Signed-off-by: Vicki Pfau <vi@endrift.com>
Link: https://lore.kernel.org/r/20250328234345.989761-1-vi@endrift.com
Fixes: 540602a43ae5 ("Input: xpad - add a few new VID/PID combinations")
Fixes: 3492321e2e60 ("Input: xpad - add multiple supported devices")
Cc: stable@vger.kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
4 weeks agoInput: hisi_powerkey - enable system-wakeup for s2idle
Ulf Hansson [Thu, 6 Mar 2025 11:50:21 +0000 (12:50 +0100)]
Input: hisi_powerkey - enable system-wakeup for s2idle

To wake up the system from s2idle when pressing the power-button, let's
convert from using pm_wakeup_event() to pm_wakeup_dev_event(), as it allows
us to specify the "hard" in-parameter, which needs to be set for s2idle.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Link: https://lore.kernel.org/r/20250306115021.797426-1-ulf.hansson@linaro.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
4 weeks agoMerge tag 'mm-hotfixes-stable-2025-05-10-14-23' of git://git.kernel.org/pub/scm/linux...
Linus Torvalds [Sat, 10 May 2025 22:50:56 +0000 (15:50 -0700)]
Merge tag 'mm-hotfixes-stable-2025-05-10-14-23' of git://git./linux/kernel/git/akpm/mm

Pull misc hotfixes from Andrew Morton:
 "22 hotfixes. 13 are cc:stable and the remainder address post-6.14
  issues or aren't considered necessary for -stable kernels.

  About half are for MM. Five OCFS2 fixes and a few MAINTAINERS updates"

* tag 'mm-hotfixes-stable-2025-05-10-14-23' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm: (22 commits)
  mm: fix folio_pte_batch() on XEN PV
  nilfs2: fix deadlock warnings caused by lock dependency in init_nilfs()
  mm/hugetlb: copy the CMA flag when demoting
  mm, swap: fix false warning for large allocation with !THP_SWAP
  selftests/mm: fix a build failure on powerpc
  selftests/mm: fix build break when compiling pkey_util.c
  mm: vmalloc: support more granular vrealloc() sizing
  tools/testing/selftests: fix guard region test tmpfs assumption
  ocfs2: stop quota recovery before disabling quotas
  ocfs2: implement handshaking with ocfs2 recovery thread
  ocfs2: switch osb->disable_recovery to enum
  mailmap: map Uwe's BayLibre addresses to a single one
  MAINTAINERS: add mm THP section
  mm/userfaultfd: fix uninitialized output field for -EAGAIN race
  selftests/mm: compaction_test: support platform with huge mount of memory
  MAINTAINERS: add core mm section
  ocfs2: fix panic in failed foilio allocation
  mm/huge_memory: fix dereferencing invalid pmd migration entry
  MAINTAINERS: add reverse mapping section
  x86: disable image size check for test builds
  ...

5 weeks agoMerge tag 'driver-core-6.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 10 May 2025 16:53:11 +0000 (09:53 -0700)]
Merge tag 'driver-core-6.15-rc6' of git://git./linux/kernel/git/driver-core/driver-core

Pull driver core fix from Greg KH:
 "Here is a single driver core fix for a regression for platform devices
  that is a regression from a change that went into 6.15-rc1 that
  affected Pixel devices. It has been in linux-next for over a week with
  no reported problems"

* tag 'driver-core-6.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/driver-core/driver-core:
  platform: Fix race condition during DMA configure at IOMMU probe time

5 weeks agoMerge tag 'usb-6.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Linus Torvalds [Sat, 10 May 2025 16:18:05 +0000 (09:18 -0700)]
Merge tag 'usb-6.15-rc6' of git://git./linux/kernel/git/gregkh/usb

Pull USB fixes from Greg KH:
 "Here are some small USB driver fixes for 6.15-rc6. Included in here
  are:

   - typec driver fixes

   - usbtmc ioctl fixes

   - xhci driver fixes

   - cdnsp driver fixes

   - some gadget driver fixes

  Nothing really major, just all little stuff that people have reported
  being issues. All of these have been in linux-next this week with no
  reported issues"

* tag 'usb-6.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
  xhci: dbc: Avoid event polling busyloop if pending rx transfers are inactive.
  usb: xhci: Don't trust the EP Context cycle bit when moving HW dequeue
  usb: usbtmc: Fix erroneous generic_read ioctl return
  usb: usbtmc: Fix erroneous wait_srq ioctl return
  usb: usbtmc: Fix erroneous get_stb ioctl error returns
  usb: typec: tcpm: delay SNK_TRY_WAIT_DEBOUNCE to SRC_TRYWAIT transition
  USB: usbtmc: use interruptible sleep in usbtmc_read
  usb: cdnsp: fix L1 resume issue for RTL_REVISION_NEW_LPM version
  usb: typec: ucsi: displayport: Fix NULL pointer access
  usb: typec: ucsi: displayport: Fix deadlock
  usb: misc: onboard_usb_dev: fix support for Cypress HX3 hubs
  usb: uhci-platform: Make the clock really optional
  usb: dwc3: gadget: Make gadget_wakeup asynchronous
  usb: gadget: Use get_status callback to set remote wakeup capability
  usb: gadget: f_ecm: Add get_status callback
  usb: host: tegra: Prevent host controller crash when OTG port is used
  usb: cdnsp: Fix issue with resuming from L1
  usb: gadget: tegra-xudc: ACK ST_RC after clearing CTRL_RUN

5 weeks agoMerge tag 'staging-6.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Sat, 10 May 2025 16:08:19 +0000 (09:08 -0700)]
Merge tag 'staging-6.15-rc6' of git://git./linux/kernel/git/gregkh/staging

Pull staging driver fixes from Greg KH:
 "Here are three small staging driver fixes for 6.15-rc6. These are:

   - bcm2835-camera driver fix

   - two axis-fifo driver fixes

  All of these have been in linux-next for a few weeks with no reported
  issues"

* tag 'staging-6.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
  staging: axis-fifo: Remove hardware resets for user errors
  staging: axis-fifo: Correct handling of tx_fifo_depth for size validation
  staging: bcm2835-camera: Initialise dev in v4l2_dev

5 weeks agoMerge tag 'char-misc-6.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregk...
Linus Torvalds [Sat, 10 May 2025 15:55:15 +0000 (08:55 -0700)]
Merge tag 'char-misc-6.15-rc6' of git://git./linux/kernel/git/gregkh/char-misc

Pull char/misc/IIO driver fixes from Greg KH:
 "Here are a bunch of small driver fixes (mostly all IIO) for 6.15-rc6.
  Included in here are:

   - loads of tiny IIO driver fixes for reported issues

   - hyperv driver fix for a much-reported and worked on sysfs ring
     buffer creation bug

  All of these have been in linux-next for over a week (the IIO ones for
  many weeks now), with no reported issues"

* tag 'char-misc-6.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc: (30 commits)
  Drivers: hv: Make the sysfs node size for the ring buffer dynamic
  uio_hv_generic: Fix sysfs creation path for ring buffer
  iio: adis16201: Correct inclinometer channel resolution
  iio: adc: ad7606: fix serial register access
  iio: pressure: mprls0025pa: use aligned_s64 for timestamp
  iio: imu: adis16550: align buffers for timestamp
  staging: iio: adc: ad7816: Correct conditional logic for store mode
  iio: adc: ad7266: Fix potential timestamp alignment issue.
  iio: adc: ad7768-1: Fix insufficient alignment of timestamp.
  iio: adc: dln2: Use aligned_s64 for timestamp
  iio: accel: adxl355: Make timestamp 64-bit aligned using aligned_s64
  iio: temp: maxim-thermocouple: Fix potential lack of DMA safe buffer.
  iio: chemical: pms7003: use aligned_s64 for timestamp
  iio: chemical: sps30: use aligned_s64 for timestamp
  iio: imu: inv_mpu6050: align buffer for timestamp
  iio: imu: st_lsm6dsx: Fix wakeup source leaks on device unbind
  iio: adc: qcom-spmi-iadc: Fix wakeup source leaks on device unbind
  iio: accel: fxls8962af: Fix wakeup source leaks on device unbind
  iio: adc: ad7380: fix event threshold shift
  iio: hid-sensor-prox: Fix incorrect OFFSET calculation
  ...

5 weeks agoMerge tag 'i2c-for-6.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa...
Linus Torvalds [Sat, 10 May 2025 15:52:41 +0000 (08:52 -0700)]
Merge tag 'i2c-for-6.15-rc6' of git://git./linux/kernel/git/wsa/linux

Pull i2c fixes from Wolfram Sang:

 - omap: use correct function to read from device tree

 - MAINTAINERS: remove Seth from ISMT maintainership

* tag 'i2c-for-6.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  MAINTAINERS: Remove entry for Seth Heasley
  i2c: omap: fix deprecated of_property_read_bool() use

5 weeks agoMerge tag 'for-linus-6.15a-rc6-tag' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 10 May 2025 15:44:36 +0000 (08:44 -0700)]
Merge tag 'for-linus-6.15a-rc6-tag' of git://git./linux/kernel/git/xen/tip

Pull xen fixes from Juergen Gross:

 - A fix for the xenbus driver allowing to use a PVH Dom0 with
   Xenstore running in another domain

 - A fix for the xenbus driver addressing a rare race condition
   resulting in NULL dereferences and other problems

 - A fix for the xen-swiotlb driver fixing a problem seen on Arm
   platforms

* tag 'for-linus-6.15a-rc6-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip:
  xenbus: Use kref to track req lifetime
  xenbus: Allow PVH dom0 a non-local xenstore
  xen: swiotlb: Use swiotlb bouncing if kmalloc allocation demands it

5 weeks agoMerge tag 'pull-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Sat, 10 May 2025 15:36:07 +0000 (08:36 -0700)]
Merge tag 'pull-fixes' of git://git./linux/kernel/git/viro/vfs

Pull mount fixes from Al Viro:
 "A couple of races around legalize_mnt vs umount (both fairly old and
  hard to hit) plus two bugs in move_mount(2) - both around 'move
  detached subtree in place' logics"

* tag 'pull-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  fix IS_MNT_PROPAGATING uses
  do_move_mount(): don't leak MNTNS_PROPAGATING on failures
  do_umount(): add missing barrier before refcount checks in sync case
  __legitimize_mnt(): check for MNT_SYNC_UMOUNT should be under mount_lock

5 weeks agoMerge tag 'kvm-x86-fixes-6.15-rcN' of https://github.com/kvm-x86/linux into HEAD
Paolo Bonzini [Sat, 10 May 2025 15:11:06 +0000 (11:11 -0400)]
Merge tag 'kvm-x86-fixes-6.15-rcN' of https://github.com/kvm-x86/linux into HEAD

KVM x86 fixes for 6.15-rcN

 - Forcibly leave SMM on SHUTDOWN interception on AMD CPUs to avoid causing
   problems due to KVM stuffing INIT on SHUTDOWN (KVM needs to sanitize the
   VMCB as its state is undefined after SHUTDOWN, emulating INIT is the
   least awful choice).

 - Track the valid sync/dirty fields in kvm_run as a u64 to ensure KVM
   KVM doesn't goof a sanity check in the future.

 - Free obsolete roots when (re)loading the MMU to fix a bug where
   pre-faulting memory can get stuck due to always encountering a stale
   root.

 - When dumping GHCB state, use KVM's snapshot instead of the raw GHCB page
   to print state, so that KVM doesn't print stale/wrong information.

 - When changing memory attributes (e.g. shared <=> private), add potential
   hugepage ranges to the mmu_invalidate_range_{start,end} set so that KVM
   doesn't create a shared/private hugepage when the the corresponding
   attributes will become mixed (the attributes are commited *after* KVM
   finishes the invalidation).

 - Rework the SRSO mitigation to enable BP_SPEC_REDUCE only when KVM has at
   least one active VM.  Effectively BP_SPEC_REDUCE when KVM is loaded led
   to very measurable performance regressions for non-KVM workloads.

5 weeks agoMerge tag 'kvmarm-fixes-6.15-3' of https://git.kernel.org/pub/scm/linux/kernel/git...
Paolo Bonzini [Sat, 10 May 2025 15:10:02 +0000 (11:10 -0400)]
Merge tag 'kvmarm-fixes-6.15-3' of https://git./linux/kernel/git/kvmarm/kvmarm into HEAD

KVM/arm64 fixes for 6.15, round #3

 - Avoid use of uninitialized memcache pointer in user_mem_abort()

 - Always set HCR_EL2.xMO bits when running in VHE, allowing interrupts
   to be taken while TGE=0 and fixing an ugly bug on AmpereOne that
   occurs when taking an interrupt while clearing the xMO bits
   (AC03_CPU_36)

 - Prevent VMMs from hiding support for AArch64 at any EL virtualized by
   KVM

 - Save/restore the host value for HCRX_EL2 instead of restoring an
   incorrect fixed value

 - Make host_stage2_set_owner_locked() check that the entire requested
   range is memory rather than just the first page

5 weeks agoMerge tag 'kvm-riscv-fixes-6.15-1' of https://github.com/kvm-riscv/linux into HEAD
Paolo Bonzini [Sat, 10 May 2025 15:09:26 +0000 (11:09 -0400)]
Merge tag 'kvm-riscv-fixes-6.15-1' of https://github.com/kvm-riscv/linux into HEAD

KVM/riscv fixes for 6.15, take #1

- Add missing reset of smstateen CSRs

5 weeks agoMerge tag 'i2c-host-fixes-6.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel...
Wolfram Sang [Sat, 10 May 2025 09:41:13 +0000 (11:41 +0200)]
Merge tag 'i2c-host-fixes-6.15-rc6' of git://git./linux/kernel/git/andi.shyti/linux into i2c/for-current

i2c-host-fixes for v6.15-rc6

- omap: use correct function to read from device tree
- MAINTAINERS: remove Seth from ISMT maintainership

5 weeks agoMerge tag '6.15-rc5-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Fri, 9 May 2025 23:45:21 +0000 (16:45 -0700)]
Merge tag '6.15-rc5-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6

Pull smb client fixes from Steve French:

 - Fix dentry leak which can cause umount crash

 - Add warning for parse contexts error on compounded operation

* tag '6.15-rc5-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6:
  smb: client: Avoid race in open_cached_dir with lease breaks
  smb3 client: warn when parse contexts returns error on compounded operation

5 weeks agofix IS_MNT_PROPAGATING uses
Al Viro [Thu, 8 May 2025 19:35:51 +0000 (15:35 -0400)]
fix IS_MNT_PROPAGATING uses

propagate_mnt() does not attach anything to mounts created during
propagate_mnt() itself.  What's more, anything on ->mnt_slave_list
of such new mount must also be new, so we don't need to even look
there.

When move_mount() had been introduced, we've got an additional
class of mounts to skip - if we are moving from anon namespace,
we do not want to propagate to mounts we are moving (i.e. all
mounts in that anon namespace).

Unfortunately, the part about "everything on their ->mnt_slave_list
will also be ignorable" is not true - if we have propagation graph
A -> B -> C
and do OPEN_TREE_CLONE open_tree() of B, we get
A -> [B <-> B'] -> C
as propagation graph, where B' is a clone of B in our detached tree.
Making B private will result in
A -> B' -> C
C still gets propagation from A, as it would after making B private
if we hadn't done that open_tree(), but now the propagation goes
through B'.  Trying to move_mount() our detached tree on subdirectory
in A should have
* moved B' on that subdirectory in A
* skipped the corresponding subdirectory in B' itself
* copied B' on the corresponding subdirectory in C.
As it is, the logics in propagation_next() and friends ends up
skipping propagation into C, since it doesn't consider anything
downstream of B'.

IOW, walking the propagation graph should only skip the ->mnt_slave_list
of new mounts; the only places where the check for "in that one
anon namespace" are applicable are propagate_one() (where we should
treat that as the same kind of thing as "mountpoint we are looking
at is not visible in the mount we are looking at") and
propagation_would_overmount().  The latter is better dealt with
in the caller (can_move_mount_beneath()); on the first call of
propagation_would_overmount() the test is always false, on the
second it is always true in "move from anon namespace" case and
always false in "move within our namespace" one, so it's easier
to just use check_mnt() before bothering with the second call and
be done with that.

Fixes: 064fe6e233e8 ("mount: handle mount propagation for detached mount trees")
Reviewed-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
5 weeks agodo_move_mount(): don't leak MNTNS_PROPAGATING on failures
Al Viro [Tue, 29 Apr 2025 01:43:23 +0000 (21:43 -0400)]
do_move_mount(): don't leak MNTNS_PROPAGATING on failures

as it is, a failed move_mount(2) from anon namespace breaks
all further propagation into that namespace, including normal
mounts in non-anon namespaces that would otherwise propagate
there.

Fixes: 064fe6e233e8 ("mount: handle mount propagation for detached mount trees")
Reviewed-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
5 weeks agodo_umount(): add missing barrier before refcount checks in sync case
Al Viro [Tue, 29 Apr 2025 03:56:14 +0000 (23:56 -0400)]
do_umount(): add missing barrier before refcount checks in sync case

do_umount() analogue of the race fixed in 119e1ef80ecf "fix
__legitimize_mnt()/mntput() race".  Here we want to make sure that
if __legitimize_mnt() doesn't notice our lock_mount_hash(), we will
notice their refcount increment.  Harder to hit than mntput_no_expire()
one, fortunately, and consequences are milder (sync umount acting
like umount -l on a rare race with RCU pathwalk hitting at just the
wrong time instead of use-after-free galore mntput_no_expire()
counterpart used to be hit).  Still a bug...

Fixes: 48a066e72d97 ("RCU'd vfsmounts")
Reviewed-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
5 weeks ago__legitimize_mnt(): check for MNT_SYNC_UMOUNT should be under mount_lock
Al Viro [Sun, 27 Apr 2025 19:41:51 +0000 (15:41 -0400)]
__legitimize_mnt(): check for MNT_SYNC_UMOUNT should be under mount_lock

... or we risk stealing final mntput from sync umount - raising mnt_count
after umount(2) has verified that victim is not busy, but before it
has set MNT_SYNC_UMOUNT; in that case __legitimize_mnt() doesn't see
that it's safe to quietly undo mnt_count increment and leaves dropping
the reference to caller, where it'll be a full-blown mntput().

Check under mount_lock is needed; leaving the current one done before
taking that makes no sense - it's nowhere near common enough to bother
with.

Reviewed-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
5 weeks agoMerge tag 'rust-fixes-6.15-2' of git://git.kernel.org/pub/scm/linux/kernel/git/ojeda...
Linus Torvalds [Fri, 9 May 2025 21:06:34 +0000 (14:06 -0700)]
Merge tag 'rust-fixes-6.15-2' of git://git./linux/kernel/git/ojeda/linux

Pull rust fixes from Miguel Ojeda:

 - Make CFI_AUTO_DEFAULT depend on !RUST or Rust >= 1.88.0

 - Clean Rust (and Clippy) lints for the upcoming Rust 1.87.0 and 1.88.0
   releases

 - Clean objtool warning for the upcoming Rust 1.87.0 release by adding
   one more noreturn function

* tag 'rust-fixes-6.15-2' of git://git.kernel.org/pub/scm/linux/kernel/git/ojeda/linux:
  x86/Kconfig: make CFI_AUTO_DEFAULT depend on !RUST or Rust >= 1.88
  rust: clean Rust 1.88.0's `clippy::uninlined_format_args` lint
  rust: clean Rust 1.88.0's warning about `clippy::disallowed_macros` configuration
  rust: clean Rust 1.88.0's `unnecessary_transmutes` lint
  rust: allow Rust 1.87.0's `clippy::ptr_eq` lint
  objtool/rust: add one more `noreturn` Rust function for Rust 1.87.0

5 weeks agoMerge tag 'drm-fixes-2025-05-10' of https://gitlab.freedesktop.org/drm/kernel
Linus Torvalds [Fri, 9 May 2025 19:41:34 +0000 (12:41 -0700)]
Merge tag 'drm-fixes-2025-05-10' of https://gitlab.freedesktop.org/drm/kernel

Pull drm fixes from Dave Airlie:
 "Weekly drm fixes, bit bigger than last week, but overall amdgpu/xe
  with some ivpu bits and a random few fixes, and dropping the
  ttm_backup struct which wrapped struct file and was recently
  frowned at.

  drm:
   - Fix overflow when generating wedged event

  ttm:
   - Fix documentation
   - Remove struct ttm_backup

  panel:
   - simple: Fix timings for AUO G101EVN010

  amdgpu:
   - DC FP fixes
   - Freesync fix
   - DMUB AUX fixes
   - VCN fix
   - Hibernation fixes
   - HDP fixes

  xe:
   - Prevent PF queue overflow
   - Hold all forcewake during mocs test
   - Remove GSC flush on reset path
   - Fix forcewake put on error path
   - Fix runtime warning when building without svm

  i915:
   - Fix oops on resume after disconnecting DP MST sinks during suspend
   - Fix SPLC num_waiters refcounting

  ivpu:
   - Increase timeouts
   - Fix deadlock in cmdq ioctl
   - Unlock mutices in correct order

  v3d:
   - Avoid memory leak in job handling"

* tag 'drm-fixes-2025-05-10' of https://gitlab.freedesktop.org/drm/kernel: (32 commits)
  drm/i915/dp: Fix determining SST/MST mode during MTP TU state computation
  drm/xe: Add config control for svm flush work
  drm/xe: Release force wake first then runtime power
  drm/xe/gsc: do not flush the GSC worker from the reset path
  drm/xe/tests/mocs: Hold XE_FORCEWAKE_ALL for LNCF regs
  drm/xe: Add page queue multiplier
  drm/amdgpu/hdp7: use memcfg register to post the write for HDP flush
  drm/amdgpu/hdp6: use memcfg register to post the write for HDP flush
  drm/amdgpu/hdp5.2: use memcfg register to post the write for HDP flush
  drm/amdgpu/hdp5: use memcfg register to post the write for HDP flush
  drm/amdgpu/hdp4: use memcfg register to post the write for HDP flush
  drm/amdgpu: fix pm notifier handling
  Revert "drm/amd: Stop evicting resources on APUs in suspend"
  drm/amdgpu/vcn: using separate VCN1_AON_SOC offset
  drm/amd/display: Fix wrong handling for AUX_DEFER case
  drm/amd/display: Copy AUX read reply data whenever length > 0
  drm/amd/display: Remove incorrect checking in dmub aux handler
  drm/amd/display: Fix the checking condition in dmub aux handling
  drm/amd/display: Shift DMUB AUX reply command if necessary
  drm/amd/display: Call FP Protect Before Mode Programming/Mode Support
  ...

5 weeks agoMerge tag 'drm-intel-fixes-2025-05-09' of https://gitlab.freedesktop.org/drm/i915...
Dave Airlie [Fri, 9 May 2025 19:07:17 +0000 (05:07 +1000)]
Merge tag 'drm-intel-fixes-2025-05-09' of https://gitlab.freedesktop.org/drm/i915/kernel into drm-fixes

drm/i915 fixes for v6.15-rc6:
- Fix oops on resume after disconnecting DP MST sinks during suspend
- Fix SPLC num_waiters refcounting

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Jani Nikula <jani.nikula@intel.com>
Link: https://lore.kernel.org/r/87tt5umeaw.fsf@intel.com
5 weeks agoMerge tag 'drm-xe-fixes-2025-05-09' of https://gitlab.freedesktop.org/drm/xe/kernel...
Dave Airlie [Fri, 9 May 2025 19:02:38 +0000 (05:02 +1000)]
Merge tag 'drm-xe-fixes-2025-05-09' of https://gitlab.freedesktop.org/drm/xe/kernel into drm-fixes

Driver Changes:
- Prevent PF queue overflow
- Hold all forcewake during mocs test
- Remove GSC flush on reset path
- Fix forcewake put on error path
- Fix runtime warning when building without svm

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/jffqa56f2zp4i5ztz677cdspgxhnw7qfop3dd3l2epykfpfvza@q2nw6wapsphz
5 weeks agoMerge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Linus Torvalds [Fri, 9 May 2025 18:30:26 +0000 (11:30 -0700)]
Merge tag 'arm64-fixes' of git://git./linux/kernel/git/arm64/linux

Pull arm64 fix from Catalin Marinas:
 "Move the arm64_use_ng_mappings variable from the .bss to the .data
  section as it is accessed very early during boot with the MMU off and
  before the .bss has been initialised.

  This could lead to incorrect idmap page table"

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  arm64: cpufeature: Move arm64_use_ng_mappings to the .data section to prevent wrong idmap generation

5 weeks agoMerge tag 'riscv-for-linus-6.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 9 May 2025 18:17:50 +0000 (11:17 -0700)]
Merge tag 'riscv-for-linus-6.15-rc6' of git://git./linux/kernel/git/riscv/linux

Pull RISC-V fixes from Palmer Dabbelt:

 - The compressed half-word misaligned access instructions (c.lhu, c.lh,
   and c.sh) from the Zcb extension are now properly emulated

 - A series of fixes to properly emulate permissions while handling
   userspace misaligned accesses

 - A pair of fixes for PR_GET_TAGGED_ADDR_CTRL to avoid accessing the
   envcfg CSR on systems that don't support that CSR, and to report
   those failures up to userspace

 - The .rela.dyn section is no longer stripped from vmlinux, as it is
   necessary to relocate the kernel under some conditions (including
   kexec)

* tag 'riscv-for-linus-6.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/riscv/linux:
  riscv: Disallow PR_GET_TAGGED_ADDR_CTRL without Supm
  scripts: Do not strip .rela.dyn section
  riscv: Fix kernel crash due to PR_SET_TAGGED_ADDR_CTRL
  riscv: misaligned: use get_user() instead of __get_user()
  riscv: misaligned: enable IRQs while handling misaligned accesses
  riscv: misaligned: factorize trap handling
  riscv: misaligned: Add handling for ZCB instructions

5 weeks agoMerge tag 'block-6.15-20250509' of git://git.kernel.dk/linux
Linus Torvalds [Fri, 9 May 2025 17:34:50 +0000 (10:34 -0700)]
Merge tag 'block-6.15-20250509' of git://git.kernel.dk/linux

Pull block fixes from Jens Axboe:

 - Fix for a regression in this series for loop and read/write iterator
   handling

 - zone append block update tweak

 - remove a broken IO priority test

 - NVMe pull request via Christoph:
      - unblock ctrl state transition for firmware update (Daniel
        Wagner)

* tag 'block-6.15-20250509' of git://git.kernel.dk/linux:
  block: remove test of incorrect io priority level
  nvme: unblock ctrl state transition for firmware update
  block: only update request sector if needed
  loop: Add sanity check for read/write_iter

5 weeks agoMerge tag 'io_uring-6.15-20250509' of git://git.kernel.dk/linux
Linus Torvalds [Fri, 9 May 2025 16:26:46 +0000 (09:26 -0700)]
Merge tag 'io_uring-6.15-20250509' of git://git.kernel.dk/linux

Pull io_uring fixes from Jens Axboe:

 - Fix for linked timeouts arming and firing wrt prep and issue of the
   request being managed by the linked timeout

 - Fix for a CQE ordering issue between requests with multishot and
   using the same buffer group. This is a dumbed down version for this
   release and for stable, it'll get improved for v6.16

 - Tweak the SQPOLL submit batch size. A previous commit made SQPOLL
   manage its own task_work and chose a tiny batch size, bump it from 8
   to 32 to fix a performance regression due to that

* tag 'io_uring-6.15-20250509' of git://git.kernel.dk/linux:
  io_uring/sqpoll: Increase task_work submission batch size
  io_uring: ensure deferred completions are flushed for multishot
  io_uring: always arm linked timeouts prior to issue

5 weeks agoMerge tag 'modules-6.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/modules...
Linus Torvalds [Fri, 9 May 2025 16:09:49 +0000 (09:09 -0700)]
Merge tag 'modules-6.15-rc6' of git://git./linux/kernel/git/modules/linux

Pull modules fix from Petr Pavlu:
 "A single fix to prevent use of an uninitialized completion pointer
  when releasing a module_kobject in specific situations.

  This addresses a latent bug exposed by commit f95bbfe18512 ("drivers:
  base: handle module_kobject creation")"

* tag 'modules-6.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/modules/linux:
  module: ensure that kobject_put() is safe for module type kobjects

5 weeks agox86/mm: Eliminate window where TLB flushes may be inadvertently skipped
Dave Hansen [Thu, 8 May 2025 22:41:32 +0000 (15:41 -0700)]
x86/mm: Eliminate window where TLB flushes may be inadvertently skipped

tl;dr: There is a window in the mm switching code where the new CR3 is
set and the CPU should be getting TLB flushes for the new mm.  But
should_flush_tlb() has a bug and suppresses the flush.  Fix it by
widening the window where should_flush_tlb() sends an IPI.

Long Version:

=== History ===

There were a few things leading up to this.

First, updating mm_cpumask() was observed to be too expensive, so it was
made lazier.  But being lazy caused too many unnecessary IPIs to CPUs
due to the now-lazy mm_cpumask().  So code was added to cull
mm_cpumask() periodically[2].  But that culling was a bit too aggressive
and skipped sending TLB flushes to CPUs that need them.  So here we are
again.

=== Problem ===

The too-aggressive code in should_flush_tlb() strikes in this window:

// Turn on IPIs for this CPU/mm combination, but only
// if should_flush_tlb() agrees:
cpumask_set_cpu(cpu, mm_cpumask(next));

next_tlb_gen = atomic64_read(&next->context.tlb_gen);
choose_new_asid(next, next_tlb_gen, &new_asid, &need_flush);
load_new_mm_cr3(need_flush);
// ^ After 'need_flush' is set to false, IPIs *MUST*
// be sent to this CPU and not be ignored.

        this_cpu_write(cpu_tlbstate.loaded_mm, next);
// ^ Not until this point does should_flush_tlb()
// become true!

should_flush_tlb() will suppress TLB flushes between load_new_mm_cr3()
and writing to 'loaded_mm', which is a window where they should not be
suppressed.  Whoops.

=== Solution ===

Thankfully, the fuzzy "just about to write CR3" window is already marked
with loaded_mm==LOADED_MM_SWITCHING.  Simply checking for that state in
should_flush_tlb() is sufficient to ensure that the CPU is targeted with
an IPI.

This will cause more TLB flush IPIs.  But the window is relatively small
and I do not expect this to cause any kind of measurable performance
impact.

Update the comment where LOADED_MM_SWITCHING is written since it grew
yet another user.

Peter Z also raised a concern that should_flush_tlb() might not observe
'loaded_mm' and 'is_lazy' in the same order that switch_mm_irqs_off()
writes them.  Add a barrier to ensure that they are observed in the
order they are written.

Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Acked-by: Rik van Riel <riel@surriel.com>
Link: https://lore.kernel.org/oe-lkp/202411282207.6bd28eae-lkp@intel.com/
Fixes: 6db2526c1d69 ("x86/mm/tlb: Only trim the mm_cpumask once a second") [2]
Reported-by: Stephen Dolan <sdolan@janestreet.com>
Cc: stable@vger.kernel.org
Acked-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 weeks agoio_uring/sqpoll: Increase task_work submission batch size io_uring-6.15-20250509
Gabriel Krisman Bertazi [Thu, 8 May 2025 18:12:03 +0000 (14:12 -0400)]
io_uring/sqpoll: Increase task_work submission batch size

Our QA team reported a 10%-23%, throughput reduction on an io_uring
sqpoll testcase doing IO to a null_blk, that I traced back to a
reduction of the device submission queue depth utilization. It turns out
that, after commit af5d68f8892f ("io_uring/sqpoll: manage task_work
privately"), we capped the number of task_work entries that can be
completed from a single spin of sqpoll to only 8 entries, before the
sqpoll goes around to (potentially) sleep.  While this cap doesn't drive
the submission side directly, it impacts the completion behavior, which
affects the number of IO queued by fio per sqpoll cycle on the
submission side, and io_uring ends up seeing less ios per sqpoll cycle.
As a result, block layer plugging is less effective, and we see more
time spent inside the block layer in profilings charts, and increased
submission latency measured by fio.

There are other places that have increased overhead once sqpoll sleeps
more often, such as the sqpoll utilization calculation.  But, in this
microbenchmark, those were not representative enough in perf charts, and
their removal didn't yield measurable changes in throughput.  The major
overhead comes from the fact we plug less, and less often, when submitting
to the block layer.

My benchmark is:

fio --ioengine=io_uring --direct=1 --iodepth=128 --runtime=300 --bs=4k \
    --invalidate=1 --time_based  --ramp_time=10 --group_reporting=1 \
    --filename=/dev/nullb0 --name=RandomReads-direct-nullb-sqpoll-4k-1 \
    --rw=randread --numjobs=1 --sqthread_poll

In one machine, tested on top of Linux 6.15-rc1, we have the following
baseline:
  READ: bw=4994MiB/s (5236MB/s), 4994MiB/s-4994MiB/s (5236MB/s-5236MB/s), io=439GiB (471GB), run=90001-90001msec

With this patch:
  READ: bw=5762MiB/s (6042MB/s), 5762MiB/s-5762MiB/s (6042MB/s-6042MB/s), io=506GiB (544GB), run=90001-90001msec

which is a 15% improvement in measured bandwidth.  The average
submission latency is noticeably lowered too.  As measured by
fio:

Baseline:
   lat (usec): min=20, max=241, avg=99.81, stdev=3.38
Patched:
   lat (usec): min=26, max=226, avg=86.48, stdev=4.82

If we look at blktrace, we can also see the plugging behavior is
improved. In the baseline, we end up limited to plugging 8 requests in
the block layer regardless of the device queue depth size, while after
patching we can drive more io, and we manage to utilize the full device
queue.

In the baseline, after a stabilization phase, an ordinary submission
looks like:
  254,0    1    49942     0.016028795  5977  U   N [iou-sqp-5976] 7

After patching, I see consistently more requests per unplug.
  254,0    1     4996     0.001432872  3145  U   N [iou-sqp-3144] 32

Ideally, the cap size would at least be the deep enough to fill the
device queue, but we can't predict that behavior, or assume all IO goes
to a single device, and thus can't guess the ideal batch size.  We also
don't want to let the tw run unbounded, though I'm not sure it would
really be a problem.  Instead, let's just give it a more sensible value
that will allow for more efficient batching.  I've tested with different
cap values, and initially proposed to increase the cap to 1024.  Jens
argued it is too big of a bump and I observed that, with 32, I'm no
longer able to observe this bottleneck in any of my machines.

Fixes: af5d68f8892f ("io_uring/sqpoll: manage task_work privately")
Signed-off-by: Gabriel Krisman Bertazi <krisman@suse.de>
Link: https://lore.kernel.org/r/20250508181203.3785544-1-krisman@suse.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 weeks agodrm/i915/dp: Fix determining SST/MST mode during MTP TU state computation
Imre Deak [Wed, 7 May 2025 15:19:53 +0000 (18:19 +0300)]
drm/i915/dp: Fix determining SST/MST mode during MTP TU state computation

Determining the SST/MST mode during state computation must be done based
on the output type stored in the CRTC state, which in turn is set once
based on the modeset connector's SST vs. MST type and will not change as
long as the connector is using the CRTC. OTOH the MST mode indicated by
the given connector's intel_dp::is_mst flag can change independently of
the above output type, based on what sink is at any moment plugged to
the connector.

Fix the state computation accordingly.

Cc: Jani Nikula <jani.nikula@intel.com>
Fixes: f6971d7427c2 ("drm/i915/mst: adapt intel_dp_mtp_tu_compute_config() for 128b/132b SST")
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/4607
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://lore.kernel.org/r/20250507151953.251846-1-imre.deak@intel.com
(cherry picked from commit 0f45696ddb2b901fbf15cb8d2e89767be481d59f)
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
5 weeks agomemblock: Accept allocated memory before use in memblock_double_array()
Tom Lendacky [Thu, 8 May 2025 17:24:10 +0000 (12:24 -0500)]
memblock: Accept allocated memory before use in memblock_double_array()

When increasing the array size in memblock_double_array() and the slab
is not yet available, a call to memblock_find_in_range() is used to
reserve/allocate memory. However, the range returned may not have been
accepted, which can result in a crash when booting an SNP guest:

  RIP: 0010:memcpy_orig+0x68/0x130
  Code: ...
  RSP: 0000:ffffffff9cc03ce8 EFLAGS: 00010006
  RAX: ff11001ff83e5000 RBX: 0000000000000000 RCX: fffffffffffff000
  RDX: 0000000000000bc0 RSI: ffffffff9dba8860 RDI: ff11001ff83e5c00
  RBP: 0000000000002000 R08: 0000000000000000 R09: 0000000000002000
  R10: 000000207fffe000 R11: 0000040000000000 R12: ffffffff9d06ef78
  R13: ff11001ff83e5000 R14: ffffffff9dba7c60 R15: 0000000000000c00
  memblock_double_array+0xff/0x310
  memblock_add_range+0x1fb/0x2f0
  memblock_reserve+0x4f/0xa0
  memblock_alloc_range_nid+0xac/0x130
  memblock_alloc_internal+0x53/0xc0
  memblock_alloc_try_nid+0x3d/0xa0
  swiotlb_init_remap+0x149/0x2f0
  mem_init+0xb/0xb0
  mm_core_init+0x8f/0x350
  start_kernel+0x17e/0x5d0
  x86_64_start_reservations+0x14/0x30
  x86_64_start_kernel+0x92/0xa0
  secondary_startup_64_no_verify+0x194/0x19b

Mitigate this by calling accept_memory() on the memory range returned
before the slab is available.

Prior to v6.12, the accept_memory() interface used a 'start' and 'end'
parameter instead of 'start' and 'size', therefore the accept_memory()
call must be adjusted to specify 'start + size' for 'end' when applying
to kernels prior to v6.12.

Cc: stable@vger.kernel.org # see patch description, needs adjustments for <= 6.11
Fixes: dcdfdd40fa82 ("mm: Add support for unaccepted memory")
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Link: https://lore.kernel.org/r/da1ac73bf4ded761e21b4e4bb5178382a580cd73.1746725050.git.thomas.lendacky@amd.com
Signed-off-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
5 weeks agoMerge tag 'amd-drm-fixes-6.15-2025-05-08' of https://gitlab.freedesktop.org/agd5f...
Dave Airlie [Fri, 9 May 2025 01:10:41 +0000 (11:10 +1000)]
Merge tag 'amd-drm-fixes-6.15-2025-05-08' of https://gitlab.freedesktop.org/agd5f/linux into drm-fixes

amd-drm-fixes-6.15-2025-05-08:

amdgpu:
- DC FP fixes
- Freesync fix
- DMUB AUX fixes
- VCN fix
- Hibernation fixes
- HDP fixes

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://lore.kernel.org/r/20250508194102.3242372-1-alexander.deucher@amd.com
5 weeks agoMerge tag 'drm-misc-fixes-2025-05-08' of https://gitlab.freedesktop.org/drm/misc...
Dave Airlie [Thu, 8 May 2025 22:51:57 +0000 (08:51 +1000)]
Merge tag 'drm-misc-fixes-2025-05-08' of https://gitlab.freedesktop.org/drm/misc/kernel into drm-fixes

Short summary of fixes pull:

drm:
- Fix overflow when generating wedged event

ivpu:
- Increate timeouts
- Fix deadlock in cmdq ioctl
- Unlock mutices in correct order

panel:
- simple: Fix timings for AUO G101EVN010

ttm:
- Fix documentation
- Remove struct ttm_backup

v3d:
- Avoid memory leak in job handling

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://lore.kernel.org/r/20250508104939.GA76697@2a02-2454-fd5e-fd00-c110-cbf2-6528-c5be.dyn6.pyur.net
5 weeks agoMerge tag 'bcachefs-2025-05-08' of git://evilpiepirate.org/bcachefs
Linus Torvalds [Thu, 8 May 2025 21:28:49 +0000 (14:28 -0700)]
Merge tag 'bcachefs-2025-05-08' of git://evilpiepirate.org/bcachefs

Pull bcachefs fixes from Kent Overstreet:

 - Some fixes to help with filesystem analysis: ensure superblock
   error count gets written if we go ERO, don't discard the journal
   aggressively (so it's available for list_journal -a)

 - Fix lost wakeup on arm causing us to get stuck when reading btree
   nodes

 - Fix fsck failing to exit on ctrl-c

 - An additional fix for filesystems with misaligned bucket sizes: we
   now ensure that allocations are properly aligned

 - Setting background target but not promote target will now leave that
   data cached on the foreground target, as it used to

 - Revert a change to when we allocate the VFS superblock, this was done
   for implementing blk_holder_ops but ended up not being needed, and
   allocating a superblock and not setting SB_BORN while we do recovery
   caused sync() calls and other things to hang

 - Assorted fixes for harmless error messages that caused concern to
   users

* tag 'bcachefs-2025-05-08' of git://evilpiepirate.org/bcachefs:
  bcachefs: Don't aggressively discard the journal
  bcachefs: Ensure superblock gets written when we go ERO
  bcachefs: Filter out harmless EROFS error messages
  bcachefs: journal_shutdown is EROFS, not EIO
  bcachefs: Call bch2_fs_start before getting vfs superblock
  bcachefs: fix hung task timeout in journal read
  bcachefs: Add missing barriers before wake_up_bit()
  bcachefs: Ensure proper write alignment
  bcachefs: Improve want_cached_ptr()
  bcachefs: thread_with_stdio: fix spinning instead of exiting

5 weeks agodrm/xe: Add config control for svm flush work
Shuicheng Lin [Fri, 2 May 2025 17:00:52 +0000 (17:00 +0000)]
drm/xe: Add config control for svm flush work

Without CONFIG_DRM_XE_GPUSVM set, GPU SVM is not initialized thus below
warning pops. Refine the flush work code to be controlled by the config
to avoid below warning:
"
[  453.132028] ------------[ cut here ]------------
[  453.132527] WARNING: CPU: 9 PID: 4491 at kernel/workqueue.c:4205 __flush_work+0x379/0x3a0
[  453.133355] Modules linked in: xe drm_ttm_helper ttm gpu_sched drm_buddy drm_suballoc_helper drm_gpuvm drm_exec
[  453.134352] CPU: 9 UID: 0 PID: 4491 Comm: xe_exec_mix_mod Tainted: G     U  W           6.15.0-rc3+ #7 PREEMPT(full)
[  453.135405] Tainted: [U]=USER, [W]=WARN
...
[  453.136921] RIP: 0010:__flush_work+0x379/0x3a0
[  453.137417] Code: 8b 45 00 48 8b 55 08 89 c7 48 c1 e8 04 83 e7 08 83 e0 0f 83 cf 02 89 c6 48 0f ba 6d 00 03 e9 d5 fe ff ff 0f 0b e9 db fd ff ff <0f> 0b 45 31 e4 e9 d1 fd ff ff 0f 0b e9 03 ff ff ff 0f 0b e9 d6 fe
[  453.139250] RSP: 0018:ffffc90000c67b18 EFLAGS: 00010246
[  453.139782] RAX: 0000000000000000 RBX: ffff888108a24000 RCX: 0000000000002000
[  453.140521] RDX: 0000000000000001 RSI: 0000000000000000 RDI: ffff8881016d61c8
[  453.141253] RBP: ffff8881016d61c8 R08: 0000000000000000 R09: 0000000000000000
[  453.141985] R10: 0000000000000000 R11: 0000000008a24000 R12: 0000000000000001
[  453.142709] R13: 0000000000000002 R14: 0000000000000000 R15: ffff888107db8c00
[  453.143450] FS:  00007f44853d4c80(0000) GS:ffff8882f469b000(0000) knlGS:0000000000000000
[  453.144276] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  453.144853] CR2: 00007f4487629228 CR3: 00000001016aa000 CR4: 00000000000406f0
[  453.145594] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[  453.146320] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[  453.147061] Call Trace:
[  453.147336]  <TASK>
[  453.147579]  ? tick_nohz_tick_stopped+0xd/0x30
[  453.148067]  ? xas_load+0x9/0xb0
[  453.148435]  ? xa_load+0x6f/0xb0
[  453.148781]  __xe_vm_bind_ioctl+0xbd5/0x1500 [xe]
[  453.149338]  ? dev_printk_emit+0x48/0x70
[  453.149762]  ? _dev_printk+0x57/0x80
[  453.150148]  ? drm_ioctl+0x17c/0x440
[  453.150544]  ? __drm_dev_vprintk+0x36/0x90
[  453.150983]  ? __pfx_xe_vm_bind_ioctl+0x10/0x10 [xe]
[  453.151575]  ? drm_ioctl_kernel+0x9f/0xf0
[  453.151998]  ? __pfx_xe_vm_bind_ioctl+0x10/0x10 [xe]
[  453.152560]  drm_ioctl_kernel+0x9f/0xf0
[  453.152968]  drm_ioctl+0x20f/0x440
[  453.153332]  ? __pfx_xe_vm_bind_ioctl+0x10/0x10 [xe]
[  453.153893]  ? ioctl_has_perm.constprop.0.isra.0+0xae/0x100
[  453.154489]  ? memory_bm_test_bit+0x5/0x60
[  453.154935]  xe_drm_ioctl+0x47/0x70 [xe]
[  453.155419]  __x64_sys_ioctl+0x8d/0xc0
[  453.155824]  do_syscall_64+0x47/0x110
[  453.156228]  entry_SYSCALL_64_after_hwframe+0x76/0x7e
"

v2 (Matt):
    refine commit message to have more details
    add Fixes tag
    move the code to xe_svm.h which already have the config
    remove a blank line per codestyle suggestion

Fixes: 63f6e480d115 ("drm/xe: Add SVM garbage collector")
Cc: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Shuicheng Lin <shuicheng.lin@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20250502170052.1787973-1-shuicheng.lin@intel.com
(cherry picked from commit 9d80698bcd97a5ad1088bcbb055e73fd068895e2)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
5 weeks agodrm/xe: Release force wake first then runtime power
Shuicheng Lin [Wed, 7 May 2025 02:23:02 +0000 (02:23 +0000)]
drm/xe: Release force wake first then runtime power

xe_force_wake_get() is dependent on xe_pm_runtime_get(), so for
the release path, xe_force_wake_put() should be called first then
xe_pm_runtime_put().
Combine the error path and normal path together with goto.

Fixes: 85d547608ef5 ("drm/xe/xe_gt_debugfs: Update handling of xe_force_wake_get return")
Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Shuicheng Lin <shuicheng.lin@intel.com>
Reviewed-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Link: https://lore.kernel.org/r/20250507022302.2187527-1-shuicheng.lin@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
(cherry picked from commit 432cd94efdca06296cc5e76d673546f58aa90ee1)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
5 weeks agodrm/xe/gsc: do not flush the GSC worker from the reset path
Daniele Ceraolo Spurio [Fri, 2 May 2025 15:51:04 +0000 (08:51 -0700)]
drm/xe/gsc: do not flush the GSC worker from the reset path

The workqueue used for the reset worker is marked as WQ_MEM_RECLAIM,
while the GSC one isn't (and can't be as we need to do memory
allocations in the gsc worker). Therefore, we can't flush the latter
from the former.

The reason why we had such a flush was to avoid interrupting either
the GSC FW load or in progress GSC proxy operations. GSC proxy
operations fall into 2 categories:

1) GSC proxy init: this only happens once immediately after GSC FW load
   and does not support being interrupted. The only way to recover from
   an interruption of the proxy init is to do an FLR and re-load the GSC.

2) GSC proxy request: this can happen in response to a request that
   the driver sends to the GSC. If this is interrupted, the GSC FW will
   timeout and the driver request will be failed, but overall the GSC
   will keep working fine.

Flushing the work allowed us to avoid interruption in both cases (unless
the hang came from the GSC engine itself, in which case we're toast
anyway). However, a failure on a proxy request is tolerable if we're in
a scenario where we're triggering a GT reset (i.e., something is already
gone pretty wrong), so what we really need to avoid is interrupting
the init flow, which we can do by polling on the register that reports
when the proxy init is complete (as that ensure us that all the load and
init operations have been completed).

Note that during suspend we still want to do a flush of the worker to
make sure it completes any operations involving the HW before the power
is cut.

v2: fix spelling in commit msg, rename waiter function (Julia)

Fixes: dd0e89e5edc2 ("drm/xe/gsc: GSC FW load")
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/4830
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: John Harrison <John.C.Harrison@Intel.com>
Cc: Alan Previn <alan.previn.teres.alexis@intel.com>
Cc: <stable@vger.kernel.org> # v6.8+
Reviewed-by: Julia Filipchuk <julia.filipchuk@intel.com>
Link: https://lore.kernel.org/r/20250502155104.2201469-1-daniele.ceraolospurio@intel.com
(cherry picked from commit 12370bfcc4f0bdf70279ec5b570eb298963422b5)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
5 weeks agodrm/xe/tests/mocs: Hold XE_FORCEWAKE_ALL for LNCF regs
Tejas Upadhyay [Mon, 28 Apr 2025 08:23:57 +0000 (13:53 +0530)]
drm/xe/tests/mocs: Hold XE_FORCEWAKE_ALL for LNCF regs

LNCF registers report wrong values when XE_FORCEWAKE_GT
only is held. Holding XE_FORCEWAKE_ALL ensures correct
operations on LNCF regs.

V2(Himal):
 - Use xe_force_wake_ref_has_domain

Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/1999
Fixes: a6a4ea6d7d37 ("drm/xe: Add mocs kunit")
Reviewed-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250428082357.1730068-1-tejas.upadhyay@intel.com
Signed-off-by: Tejas Upadhyay <tejas.upadhyay@intel.com>
(cherry picked from commit 70a2585e582058e94fe4381a337be42dec800337)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
5 weeks agodrm/xe: Add page queue multiplier
Matthew Brost [Tue, 8 Apr 2025 15:59:15 +0000 (08:59 -0700)]
drm/xe: Add page queue multiplier

For an unknown reason the math to determine the PF queue size does is
not correct - compute UMD applications are overflowing the PF queue
which is fatal. A multippier of 8 fixes the problem.

Fixes: 3338e4f90c14 ("drm/xe: Use topology to determine page fault queue size")
Cc: stable@vger.kernel.org
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Jagmeet Randhawa <jagmeet.randhawa@intel.com>
Link: https://lore.kernel.org/r/20250408155915.78770-1-matthew.brost@intel.com
(cherry picked from commit 29582e0ea75c95668d168b12406e3c56cf5a73c4)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
5 weeks agoMerge tag 'vfio-v6.15-rc6' of https://github.com/awilliam/linux-vfio
Linus Torvalds [Thu, 8 May 2025 19:09:22 +0000 (12:09 -0700)]
Merge tag 'vfio-v6.15-rc6' of https://github.com/awilliam/linux-vfio

Pull vfio fix from Alex Williamson:

 - Fix an issue in vfio-pci huge_fault handling by aligning faults to
   the order, resulting in deterministic use of huge pages.  This
   avoids a race where simultaneous aligned and unaligned faults to
   the same PMD can result in a VM_FAULT_OOM and subsequent VM crash.
   (Alex Williamson)

* tag 'vfio-v6.15-rc6' of https://github.com/awilliam/linux-vfio:
  vfio/pci: Align huge faults to order

5 weeks agoMerge tag 'riscv-fixes-6.15-rc6' of ssh://gitolite.kernel.org/pub/scm/linux/kernel...
Palmer Dabbelt [Thu, 8 May 2025 16:40:21 +0000 (09:40 -0700)]
Merge tag 'riscv-fixes-6.15-rc6' of ssh://gitolite./linux/kernel/git/alexghiti/linux into fixes

riscv fixes for 6.15-rc6

- A fix to handle compressed halfword load/store instructions misaligned accesses
- A fix to allow user memory access while handling a misaligned access
- 2 fixes to return an error if the pointer masking extension is not implemented on the platform but userspace still tries to access it, which caused oops on some early platforms
- A fix to prevent the stripping of .rela.dyn so that a vmlinux loaded by kexec can successfully boot

* tag 'riscv-fixes-6.15-rc6' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/alexghiti/linux:
  riscv: Disallow PR_GET_TAGGED_ADDR_CTRL without Supm
  scripts: Do not strip .rela.dyn section
  riscv: Fix kernel crash due to PR_SET_TAGGED_ADDR_CTRL
  riscv: misaligned: use get_user() instead of __get_user()
  riscv: misaligned: enable IRQs while handling misaligned accesses
  riscv: misaligned: factorize trap handling
  riscv: misaligned: Add handling for ZCB instructions

5 weeks agodrm/amdgpu/hdp7: use memcfg register to post the write for HDP flush
Alex Deucher [Wed, 30 Apr 2025 16:50:02 +0000 (12:50 -0400)]
drm/amdgpu/hdp7: use memcfg register to post the write for HDP flush

Reading back the remapped HDP flush register seems to cause
problems on some platforms. All we need is a read, so read back
the memcfg register.

Fixes: 689275140cb8 ("drm/amdgpu/hdp7.0: do a posting read when flushing HDP")
Reported-by: Alexey Klimov <alexey.klimov@linaro.org>
Link: https://lists.freedesktop.org/archives/amd-gfx/2025-April/123150.html
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4119
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3908
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit dbc064adfcf9095e7d895bea87b2f75c1ab23236)
Cc: stable@vger.kernel.org
5 weeks agodrm/amdgpu/hdp6: use memcfg register to post the write for HDP flush
Alex Deucher [Wed, 30 Apr 2025 16:48:51 +0000 (12:48 -0400)]
drm/amdgpu/hdp6: use memcfg register to post the write for HDP flush

Reading back the remapped HDP flush register seems to cause
problems on some platforms. All we need is a read, so read back
the memcfg register.

Fixes: abe1cbaec6cf ("drm/amdgpu/hdp6.0: do a posting read when flushing HDP")
Reported-by: Alexey Klimov <alexey.klimov@linaro.org>
Link: https://lists.freedesktop.org/archives/amd-gfx/2025-April/123150.html
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4119
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3908
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 84141ff615951359c9a99696fd79a36c465ed847)
Cc: stable@vger.kernel.org
5 weeks agodrm/amdgpu/hdp5.2: use memcfg register to post the write for HDP flush
Alex Deucher [Wed, 30 Apr 2025 16:47:37 +0000 (12:47 -0400)]
drm/amdgpu/hdp5.2: use memcfg register to post the write for HDP flush

Reading back the remapped HDP flush register seems to cause
problems on some platforms. All we need is a read, so read back
the memcfg register.

Fixes: f756dbac1ce1 ("drm/amdgpu/hdp5.2: do a posting read when flushing HDP")
Reported-by: Alexey Klimov <alexey.klimov@linaro.org>
Link: https://lists.freedesktop.org/archives/amd-gfx/2025-April/123150.html
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4119
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3908
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 4a89b7698e771914b4d5b571600c76e2fdcbe2a9)
Cc: stable@vger.kernel.org
5 weeks agodrm/amdgpu/hdp5: use memcfg register to post the write for HDP flush
Alex Deucher [Wed, 30 Apr 2025 16:46:56 +0000 (12:46 -0400)]
drm/amdgpu/hdp5: use memcfg register to post the write for HDP flush

Reading back the remapped HDP flush register seems to cause
problems on some platforms. All we need is a read, so read back
the memcfg register.

Fixes: cf424020e040 ("drm/amdgpu/hdp5.0: do a posting read when flushing HDP")
Reported-by: Alexey Klimov <alexey.klimov@linaro.org>
Link: https://lists.freedesktop.org/archives/amd-gfx/2025-April/123150.html
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4119
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3908
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit a5cb344033c7598762e89255e8ff52827abb57a4)
Cc: stable@vger.kernel.org
5 weeks agoMerge tag 'net-6.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Linus Torvalds [Thu, 8 May 2025 15:33:56 +0000 (08:33 -0700)]
Merge tag 'net-6.15-rc6' of git://git./linux/kernel/git/netdev/net

Pull networking fixes from Paolo Abeni:
 "Including fixes from CAN, WiFi and netfilter.

  We have still a comple of regressions open due to the recent
  drivers locking refactor. The patches are in-flight, but not
  ready yet.

  Current release - regressions:

   - core: lock netdevices during dev_shutdown

   - sch_htb: make htb_deactivate() idempotent

   - eth: virtio-net: don't re-enable refill work too early

  Current release - new code bugs:

   - eth: icssg-prueth: fix kernel panic during concurrent Tx queue
     access

  Previous releases - regressions:

   - gre: fix again IPv6 link-local address generation.

   - eth: b53: fix learning on VLAN unaware bridges

  Previous releases - always broken:

   - wifi: fix out-of-bounds access during multi-link element
     defragmentation

   - can:
       - initialize spin lock on device probe
       - fix order of unregistration calls

   - openvswitch: fix unsafe attribute parsing in output_userspace()

   - eth:
       - virtio-net: fix total qstat values
       - mtk_eth_soc: reset all TX queues on DMA free
       - fbnic: firmware IPC mailbox fixes"

* tag 'net-6.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (55 commits)
  virtio-net: fix total qstat values
  net: export a helper for adding up queue stats
  fbnic: Do not allow mailbox to toggle to ready outside fbnic_mbx_poll_tx_ready
  fbnic: Pull fbnic_fw_xmit_cap_msg use out of interrupt context
  fbnic: Improve responsiveness of fbnic_mbx_poll_tx_ready
  fbnic: Cleanup handling of completions
  fbnic: Actually flush_tx instead of stalling out
  fbnic: Add additional handling of IRQs
  fbnic: Gate AXI read/write enabling on FW mailbox
  fbnic: Fix initialization of mailbox descriptor rings
  net: dsa: b53: do not set learning and unicast/multicast on up
  net: dsa: b53: fix learning on VLAN unaware bridges
  net: dsa: b53: fix toggling vlan_filtering
  net: dsa: b53: do not program vlans when vlan filtering is off
  net: dsa: b53: do not allow to configure VLAN 0
  net: dsa: b53: always rejoin default untagged VLAN on bridge leave
  net: dsa: b53: fix VLAN ID for untagged vlan on bridge leave
  net: dsa: b53: fix flushing old pvid VLAN on pvid change
  net: dsa: b53: fix clearing PVID of a port
  net: dsa: b53: keep CPU port always tagged again
  ...

5 weeks agoMerge tag 's390-6.15-4' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux
Linus Torvalds [Thu, 8 May 2025 15:29:13 +0000 (08:29 -0700)]
Merge tag 's390-6.15-4' of git://git./linux/kernel/git/s390/linux

Pull s390 fixes from Heiko Carstens:

 - Fix potential use-after-free bug and missing error handling in PCI
   code

 - Fix dcssblk build error

 - Fix last breaking event handling in case of stack corruption to allow
   for better error reporting

 - Update defconfigs

* tag 's390-6.15-4' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
  s390/pci: Fix duplicate pci_dev_put() in disable_slot() when PF has child VFs
  s390/pci: Fix missing check for zpci_create_device() error return
  s390: Update defconfigs
  s390/dcssblk: Fix build error with CONFIG_DAX=m and CONFIG_DCSSBLK=y
  s390/entry: Fix last breaking event handling in case of stack corruption
  s390/configs: Enable options required for TC flow offload
  s390/configs: Enable VDPA on Nvidia ConnectX-6 network card

5 weeks agoMerge tag 'v6.15-rc5-ksmbd-server-fixes' of git://git.samba.org/ksmbd
Linus Torvalds [Thu, 8 May 2025 15:22:35 +0000 (08:22 -0700)]
Merge tag 'v6.15-rc5-ksmbd-server-fixes' of git://git.samba.org/ksmbd

Pull smb server fixes from Steve French:

 - Fix UAF closing file table (e.g. in tree disconnect)

 - Fix potential out of bounds write

 - Fix potential memory leak parsing lease state in open

 - Fix oops in rename with empty target

* tag 'v6.15-rc5-ksmbd-server-fixes' of git://git.samba.org/ksmbd:
  ksmbd: Fix UAF in __close_file_table_ids
  ksmbd: prevent out-of-bounds stream writes by validating *pos
  ksmbd: fix memory leak in parse_lease_state()
  ksmbd: prevent rename with empty string

5 weeks agoMerge tag 'nvme-6.15-2025-05-08' of git://git.infradead.org/nvme into block-6.15 block-6.15-20250509
Jens Axboe [Thu, 8 May 2025 15:08:23 +0000 (09:08 -0600)]
Merge tag 'nvme-6.15-2025-05-08' of git://git.infradead.org/nvme into block-6.15

Pull NVMe fix from Christoph:

"nvme fixes for linux 6.15

 - unblock ctrl state transition for firmware update (Daniel Wagner)"

* tag 'nvme-6.15-2025-05-08' of git://git.infradead.org/nvme:
  nvme: unblock ctrl state transition for firmware update

5 weeks agoblock: remove test of incorrect io priority level
Aaron Lu [Thu, 8 May 2025 08:30:36 +0000 (16:30 +0800)]
block: remove test of incorrect io priority level

Ever since commit eca2040972b4("scsi: block: ioprio: Clean up interface
definition"), the macro IOPRIO_PRIO_LEVEL() will mask the level value to
something between 0 and 7 so necessarily, level will always be lower than
IOPRIO_NR_LEVELS(8).

Remove this obsolete check.

Reported-by: Kexin Wei <ys.weikexin@h3c.com>
Cc: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Aaron Lu <ziqianlu@bytedance.com>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Link: https://lore.kernel.org/r/20250508083018.GA769554@bytedance
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 weeks agoKVM: SVM: Set/clear SRSO's BP_SPEC_REDUCE on 0 <=> 1 VM count transitions
Sean Christopherson [Mon, 5 May 2025 18:03:00 +0000 (11:03 -0700)]
KVM: SVM: Set/clear SRSO's BP_SPEC_REDUCE on 0 <=> 1 VM count transitions

Set the magic BP_SPEC_REDUCE bit to mitigate SRSO when running VMs if and
only if KVM has at least one active VM.  Leaving the bit set at all times
unfortunately degrades performance by a wee bit more than expected.

Use a dedicated spinlock and counter instead of hooking virtualization
enablement, as changing the behavior of kvm.enable_virt_at_load based on
SRSO_BP_SPEC_REDUCE is painful, and has its own drawbacks, e.g. could
result in performance issues for flows that are sensitive to VM creation
latency.

Defer setting BP_SPEC_REDUCE until VMRUN is imminent to avoid impacting
performance on CPUs that aren't running VMs, e.g. if a setup is using
housekeeping CPUs.  Setting BP_SPEC_REDUCE in task context, i.e. without
blasting IPIs to all CPUs, also helps avoid serializing 1<=>N transitions
without incurring a gross amount of complexity (see the Link for details
on how ugly coordinating via IPIs gets).

Link: https://lore.kernel.org/all/aBOnzNCngyS_pQIW@google.com
Fixes: 8442df2b49ed ("x86/bugs: KVM: Add support for SRSO_MSR_FIX")
Reported-by: Michael Larabel <Michael@michaellarabel.com>
Closes: https://www.phoronix.com/review/linux-615-amd-regression
Cc: Borislav Petkov <bp@alien8.de>
Tested-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20250505180300.973137-1-seanjc@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
5 weeks agoriscv: Disallow PR_GET_TAGGED_ADDR_CTRL without Supm
Samuel Holland [Wed, 7 May 2025 14:52:18 +0000 (07:52 -0700)]
riscv: Disallow PR_GET_TAGGED_ADDR_CTRL without Supm

When the prctl() interface for pointer masking was added, it did not
check that the pointer masking ISA extension was supported, only the
individual submodes. Userspace could still attempt to disable pointer
masking and query the pointer masking state. commit 81de1afb2dd1
("riscv: Fix kernel crash due to PR_SET_TAGGED_ADDR_CTRL") disallowed
the former, as the senvcfg write could crash on older systems.
PR_GET_TAGGED_ADDR_CTRL state does not crash, because it reads only
kernel-internal state and not senvcfg, but it should still be disallowed
for consistency.

Fixes: 09d6775f503b ("riscv: Add support for userspace pointer masking")
Signed-off-by: Samuel Holland <samuel.holland@sifive.com>
Reviewed-by: Nam Cao <namcao@linutronix.de>
Link: https://lore.kernel.org/r/20250507145230.2272871-1-samuel.holland@sifive.com
Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
5 weeks agoscripts: Do not strip .rela.dyn section
Alexandre Ghiti [Tue, 8 Apr 2025 07:28:51 +0000 (09:28 +0200)]
scripts: Do not strip .rela.dyn section

The .rela.dyn section contains runtime relocations and is only emitted
for a relocatable kernel.

riscv uses this section to relocate the kernel at runtime but that section
is stripped from vmlinux. That prevents kexec to successfully load vmlinux
since it does not contain the relocations info needed.

Fixes: 559d1e45a16d ("riscv: Use --emit-relocs in order to move .rela.dyn in init")
Tested-by: Björn Töpel <bjorn@rivosinc.com>
Reviewed-by: Björn Töpel <bjorn@rivosinc.com>
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Link: https://lore.kernel.org/r/20250408072851.90275-1-alexghiti@rivosinc.com
Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
5 weeks agoriscv: Fix kernel crash due to PR_SET_TAGGED_ADDR_CTRL
Nam Cao [Sun, 4 May 2025 10:19:20 +0000 (12:19 +0200)]
riscv: Fix kernel crash due to PR_SET_TAGGED_ADDR_CTRL

When userspace does PR_SET_TAGGED_ADDR_CTRL, but Supm extension is not
available, the kernel crashes:

Oops - illegal instruction [#1]
    [snip]
epc : set_tagged_addr_ctrl+0x112/0x15a
 ra : set_tagged_addr_ctrl+0x74/0x15a
epc : ffffffff80011ace ra : ffffffff80011a30 sp : ffffffc60039be10
    [snip]
status: 0000000200000120 badaddr: 0000000010a79073 cause: 0000000000000002
    set_tagged_addr_ctrl+0x112/0x15a
    __riscv_sys_prctl+0x352/0x73c
    do_trap_ecall_u+0x17c/0x20c
    andle_exception+0x150/0x15c

Fix it by checking if Supm is available.

Fixes: 09d6775f503b ("riscv: Add support for userspace pointer masking")
Signed-off-by: Nam Cao <namcao@linutronix.de>
Cc: stable@vger.kernel.org
Reviewed-by: Samuel Holland <samuel.holland@sifive.com>
Link: https://lore.kernel.org/r/20250504101920.3393053-1-namcao@linutronix.de
Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
5 weeks agoriscv: misaligned: use get_user() instead of __get_user()
Clément Léger [Tue, 22 Apr 2025 16:23:10 +0000 (18:23 +0200)]
riscv: misaligned: use get_user() instead of __get_user()

Now that we can safely handle user memory accesses while in the
misaligned access handlers, use get_user() instead of __get_user() to
have user memory access checks.

Signed-off-by: Clément Léger <cleger@rivosinc.com>
Reviewed-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Link: https://lore.kernel.org/r/20250422162324.956065-4-cleger@rivosinc.com
Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
5 weeks agoriscv: misaligned: enable IRQs while handling misaligned accesses
Clément Léger [Tue, 22 Apr 2025 16:23:09 +0000 (18:23 +0200)]
riscv: misaligned: enable IRQs while handling misaligned accesses

We can safely reenable IRQs if coming from userspace. This allows to
access user memory that could potentially trigger a page fault.

Fixes: b686ecdeacf6 ("riscv: misaligned: Restrict user access to kernel memory")
Signed-off-by: Clément Léger <cleger@rivosinc.com>
Reviewed-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Link: https://lore.kernel.org/r/20250422162324.956065-3-cleger@rivosinc.com
Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
5 weeks agoriscv: misaligned: factorize trap handling
Clément Léger [Tue, 22 Apr 2025 16:23:08 +0000 (18:23 +0200)]
riscv: misaligned: factorize trap handling

Since both load/store and user/kernel should use almost the same path and
that we are going to add some code around that, factorize it.

Signed-off-by: Clément Léger <cleger@rivosinc.com>
Reviewed-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Link: https://lore.kernel.org/r/20250422162324.956065-2-cleger@rivosinc.com
Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
5 weeks agoMAINTAINERS: Remove entry for Seth Heasley
Andi Shyti [Mon, 5 May 2025 23:15:11 +0000 (01:15 +0200)]
MAINTAINERS: Remove entry for Seth Heasley

Seth's mails bounce back, remove his maintainership.

Signed-off-by: Andi Shyti <andi.shyti@kernel.org>
Link: https://lore.kernel.org/r/20250505231511.3175151-1-andi.shyti@kernel.org