]> www.infradead.org Git - users/jedix/linux-maple.git/log
users/jedix/linux-maple.git
10 months agoshmem_quota: build the object file conditionally to the config option
Carlos Maiolino [Wed, 17 Jul 2024 06:37:27 +0000 (08:37 +0200)]
shmem_quota: build the object file conditionally to the config option

Initially I added shmem-quota to obj-y, move it to the correct place and
remove the unneeded full file #ifdef

Link: https://lkml.kernel.org/r/20240717063737.910840-1-cem@kernel.org
Signed-off-by: Carlos Maiolino <cmaiolino@redhat.com>
Suggested-by: Aristeu Rozanski <aris@redhat.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: fix typo in Kconfig
Valdis Kletnieks [Sat, 13 Jul 2024 06:59:50 +0000 (02:59 -0400)]
mm: fix typo in Kconfig

Fix typo in Kconfig help

Link: https://lkml.kernel.org/r/78656.1720853990@turing-police
Fixes: e93d4166b40a ("mm: memcg: put cgroup v1-specific code under a config option")
Signed-off-by: Valdis Kletnieks <valdis.kletnieks@vt.edu>
Acked-by: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Muchun Song <muchun.song@linux.dev>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: shmem: move shmem_huge_global_enabled() into shmem_allowable_huge_orders()
Baolin Wang [Mon, 22 Jul 2024 05:43:19 +0000 (13:43 +0800)]
mm: shmem: move shmem_huge_global_enabled() into shmem_allowable_huge_orders()

Move shmem_huge_global_enabled() into shmem_allowable_huge_orders(), so
that shmem_allowable_huge_orders() can also help to find the allowable
huge orders for tmpfs.  Moreover the shmem_huge_global_enabled() can
become static.  While we are at it, passing the vma instead of mm for
shmem_huge_global_enabled() makes code cleaner.

No functional changes.

Link: https://lkml.kernel.org/r/8e825146bb29ee1a1c7bd64d2968ff3e19be7815.1721626645.git.baolin.wang@linux.alibaba.com
Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>
Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Barry Song <21cnbao@gmail.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Lance Yang <ioworker0@gmail.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: shmem: rename shmem_is_huge() to shmem_huge_global_enabled()
Baolin Wang [Mon, 22 Jul 2024 05:43:18 +0000 (13:43 +0800)]
mm: shmem: rename shmem_is_huge() to shmem_huge_global_enabled()

shmem_is_huge() is now used to check if the top-level huge page is
enabled, thus rename it to reflect its usage.

Link: https://lkml.kernel.org/r/da53296e0ab6359aa083561d9dc01e4223d60fbe.1721626645.git.baolin.wang@linux.alibaba.com
Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>
Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Barry Song <21cnbao@gmail.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Lance Yang <ioworker0@gmail.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: shmem: simplify the suitable huge orders validation for tmpfs
Baolin Wang [Mon, 22 Jul 2024 05:43:17 +0000 (13:43 +0800)]
mm: shmem: simplify the suitable huge orders validation for tmpfs

Patch series "Some cleanups for shmem", v3.

This series does some cleanups to reuse code, rename functions and
simplify logic to make code more clear.  No functional changes are
expected.

This patch (of 3):

Move the suitable huge orders validation into shmem_suitable_orders() for
tmpfs, which can reuse some code to simplify the logic.

In addition, we don't have special handling for the error code -E2BIG when
checking for conflicts with PMD sized THP in the pagecache for tmpfs,
instead, it will just fallback to order-0 allocations like this patch
does, so this simplification will not add functional changes.

Link: https://lkml.kernel.org/r/cover.1721626645.git.baolin.wang@linux.alibaba.com
Link: https://lkml.kernel.org/r/965985dd6d322929d78a0beee0dafa1c2a1b81e2.1721626645.git.baolin.wang@linux.alibaba.com
Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>
Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Barry Song <21cnbao@gmail.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Lance Yang <ioworker0@gmail.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: kvrealloc: properly document __GFP_ZERO behavior
Danilo Krummrich [Tue, 30 Jul 2024 18:49:44 +0000 (20:49 +0200)]
mm: kvrealloc: properly document __GFP_ZERO behavior

Properly document that if __GFP_ZERO logic is requested, callers must
ensure that, starting with the initial memory allocation, every subsequent
call to this API for the same memory allocation is flagged with
__GFP_ZERO.  Otherwise, it is possible that __GFP_ZERO is not fully
honored by this API.

Link: https://lkml.kernel.org/r/20240730185049.6244-5-dakr@kernel.org
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: kvrealloc: disable KASAN when switching to vmalloc
Danilo Krummrich [Tue, 30 Jul 2024 18:49:41 +0000 (20:49 +0200)]
mm: kvrealloc: disable KASAN when switching to vmalloc

Disable KASAN accessibility checks when switching from a kmalloc buffer
to a vmalloc buffer.

Link: https://lkml.kernel.org/r/20240730185049.6244-2-dakr@kernel.org
Fixes: 923a26b4c679 ("mm: kvmalloc: align kvrealloc() with krealloc()")
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: (k)vrealloc: document concurrency restrictions
Danilo Krummrich [Thu, 25 Jul 2024 12:54:37 +0000 (14:54 +0200)]
mm: (k)vrealloc: document concurrency restrictions

Document that concurrent calls to vrealloc() with itself or vfree() for
the same memory allocation are not permitted, since it leads to undefined
behavior.

Consequently, the same applies for kvrealloc(), hence also document it
for kvrealloc().

Link: https://lkml.kernel.org/r/20240725125442.4957-1-dakr@kernel.org
Suggested-by: Christoph Hellwig <hch@infradead.org>
Suggested-by: Uladzislau Rezki <urezki@gmail.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: kvmalloc: align kvrealloc() with krealloc()
Danilo Krummrich [Mon, 22 Jul 2024 16:29:24 +0000 (18:29 +0200)]
mm: kvmalloc: align kvrealloc() with krealloc()

Besides the obvious (and desired) difference between krealloc() and
kvrealloc(), there is some inconsistency in their function signatures and
behavior:

 - krealloc() frees the memory when the requested size is zero, whereas
   kvrealloc() simply returns a pointer to the existing allocation.

 - krealloc() behaves like kmalloc() if a NULL pointer is passed, whereas
   kvrealloc() does not accept a NULL pointer at all and, if passed,
   would fault instead.

 - krealloc() is self-contained, whereas kvrealloc() relies on the caller
   to provide the size of the previous allocation.

Inconsistent behavior throughout allocation APIs is error prone, hence
make kvrealloc() behave like krealloc(), which seems superior in all
mentioned aspects.

Besides that, implementing kvrealloc() by making use of krealloc() and
vrealloc() provides oppertunities to grow (and shrink) allocations more
efficiently.  For instance, vrealloc() can be optimized to allocate and
map additional pages to grow the allocation or unmap and free unused pages
to shrink the allocation.

Link: https://lkml.kernel.org/r/20240722163111.4766-3-dakr@kernel.org
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Chandan Babu R <chandan.babu@oracle.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Kees Cook <kees@kernel.org>
Cc: Marc Zyngier <maz@kernel.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Miguel Ojeda <ojeda@kernel.org>
Cc: Oliver Upton <oliver.upton@linux.dev>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Wedson Almeida Filho <wedsonaf@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: vrealloc: properly document __GFP_ZERO behavior
Danilo Krummrich [Tue, 30 Jul 2024 18:49:43 +0000 (20:49 +0200)]
mm: vrealloc: properly document __GFP_ZERO behavior

Properly document that if __GFP_ZERO logic is requested, callers must
ensure that, starting with the initial memory allocation, every subsequent
call to this API for the same memory allocation is flagged with
__GFP_ZERO.  Otherwise, it is possible that __GFP_ZERO is not fully
honored by this API.

Link: https://lkml.kernel.org/r/20240730185049.6244-4-dakr@kernel.org
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: vrealloc: consider spare memory for __GFP_ZERO
Danilo Krummrich [Tue, 30 Jul 2024 18:49:42 +0000 (20:49 +0200)]
mm: vrealloc: consider spare memory for __GFP_ZERO

Zero spare memory when shrinking a buffer with __GFP_ZERO.

Link: https://lkml.kernel.org/r/20240730185049.6244-3-dakr@kernel.org
Fixes: 1f39ee9615a8 ("mm: vmalloc: implement vrealloc()")
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: (k)vrealloc: document concurrency restrictions
Danilo Krummrich [Thu, 25 Jul 2024 12:54:37 +0000 (14:54 +0200)]
mm: (k)vrealloc: document concurrency restrictions

Document that concurrent calls to vrealloc() with itself or vfree() for
the same memory allocation are not permitted, since it leads to undefined
behavior.

Consequently, the same applies for kvrealloc(), hence also document it
for kvrealloc().

Link: https://lkml.kernel.org/r/20240725125442.4957-1-dakr@kernel.org
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Suggested-by: Christoph Hellwig <hch@infradead.org>
Suggested-by: Uladzislau Rezki <urezki@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: vrealloc: fix missing nommu implementation
Danilo Krummrich [Thu, 25 Jul 2024 14:11:59 +0000 (16:11 +0200)]
mm: vrealloc: fix missing nommu implementation

Commit 5ec6d992f460 ("mm: vmalloc: implement vrealloc()") introduces
vrealloc(), but lacks the implementation of the nommu version of this
function.

Note that currently nommu isn't actually broken without this.  The only
user, kvrealloc(), never actually calls into vrealloc() with nommu, since
it's guarded by is_vmalloc_addr(), which for nommu always returns false.

However, since this becomes a real issue once vrealloc() is used by other
code, fix this.

Link: https://lkml.kernel.org/r/20240725141227.13954-1-dakr@kernel.org
Fixes: 5ec6d992f460 ("mm: vmalloc: implement vrealloc()")
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: vmalloc: implement vrealloc()
Danilo Krummrich [Mon, 22 Jul 2024 16:29:23 +0000 (18:29 +0200)]
mm: vmalloc: implement vrealloc()

Patch series "Align kvrealloc() with krealloc()", v2.

Besides the obvious (and desired) difference between krealloc() and
kvrealloc(), there is some inconsistency in their function signatures and
behavior:

 - krealloc() frees the memory when the requested size is zero, whereas
   kvrealloc() simply returns a pointer to the existing allocation.

 - krealloc() behaves like kmalloc() if a NULL pointer is passed, whereas
   kvrealloc() does not accept a NULL pointer at all and, if passed, would fault
   instead.

 - krealloc() is self-contained, whereas kvrealloc() relies on the caller to
   provide the size of the previous allocation.

Inconsistent behavior throughout allocation APIs is error prone, hence
make kvrealloc() behave like krealloc(), which seems superior in all
mentioned aspects.

In order to be able to get rid of kvrealloc()'s oldsize parameter,
introduce vrealloc() and make use of it in kvrealloc().

Making use of vrealloc() in kvrealloc() also provides oppertunities to
grow (and shrink) allocations more efficiently.  For instance, vrealloc()
can be optimized to allocate and map additional pages to grow the
allocation or unmap and free unused pages to shrink the allocation.

Besides the above, those functions are required by Rust's allocator abstractons
[1] (rework based on this series in [2]). With `Vec` or `KVec` respectively,
potentially growing (and shrinking) data structures are rather common.

[1] https://lore.kernel.org/lkml/20240704170738.3621-1-dakr@redhat.com/
[2] https://git.kernel.org/pub/scm/linux/kernel/git/dakr/linux.git/log/?h=rust/mm

This patch (of 2):

Implement vrealloc() analogous to krealloc().

Currently, krealloc() requires the caller to pass the size of the previous
memory allocation, which, instead, should be self-contained.

We attempt to fix this in a subsequent patch which, in order to do so,
requires vrealloc().

Besides that, we need realloc() functions for kernel allocators in Rust
too.  With `Vec` or `KVec` respectively, potentially growing (and
shrinking) data structures are rather common.

Link: https://lkml.kernel.org/r/20240722163111.4766-1-dakr@kernel.org
Link: https://lkml.kernel.org/r/20240722163111.4766-2-dakr@kernel.org
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Chandan Babu R <chandan.babu@oracle.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Kees Cook <kees@kernel.org>
Cc: Marc Zyngier <maz@kernel.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Miguel Ojeda <ojeda@kernel.org>
Cc: Oliver Upton <oliver.upton@linux.dev>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Uladzislau Rezki <urezki@gmail.com>
Cc: Wedson Almeida Filho <wedsonaf@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: add node_reclaim successes to VM event counters
Matthew Cassell [Mon, 22 Jul 2024 17:13:16 +0000 (17:13 +0000)]
mm: add node_reclaim successes to VM event counters

/proc/vmstat currently shows the number of node_reclaim() failures when
vm.zone_reclaim_mode is set appropriately.  It would be convenient to have
the number of successes right next to zone_reclaim_failed (similar to
compaction and migration).

While just a trivially addition to the vmstat file.  It was helpful during
benchmarking to not have to probe node_reclaim() to observe the
success/failure ratio.

Link: https://lkml.kernel.org/r/20240722171316.7517-1-mcassell411@gmail.com
Signed-off-by: Matthew Cassell <mcassell411@gmail.com>
Cc: Domenico Cerasuolo <cerasuolodomenico@gmail.com>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Li Zhijian <lizhijian@fujitsu.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agokexec_file: fix elfcorehdr digest exclusion when CONFIG_CRASH_HOTPLUG=y
Petr Tesarik [Mon, 5 Aug 2024 15:07:50 +0000 (17:07 +0200)]
kexec_file: fix elfcorehdr digest exclusion when CONFIG_CRASH_HOTPLUG=y

Fix the condition to exclude the elfcorehdr segment from the SHA digest
calculation.

The j iterator is an index into the output sha_regions[] array, not into
the input image->segment[] array.  Once it reaches
image->elfcorehdr_index, all subsequent segments are excluded.  Besides,
if the purgatory segment precedes the elfcorehdr segment, the elfcorehdr
may be wrongly included in the calculation.

Link: https://lkml.kernel.org/r/20240805150750.170739-1-petr.tesarik@suse.com
Fixes: f7cc804a9fd4 ("kexec: exclude elfcorehdr from the segment digest")
Signed-off-by: Petr Tesarik <ptesarik@suse.com>
Acked-by: Baoquan He <bhe@redhat.com>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: Hari Bathini <hbathini@linux.ibm.com>
Cc: Sourabh Jain <sourabhjain@linux.ibm.com>
Cc: Eric DeVolder <eric_devolder@yahoo.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agoresource: fix region_intersects() for CXL memory
Huang Ying [Fri, 16 Aug 2024 02:07:23 +0000 (10:07 +0800)]
resource: fix region_intersects() for CXL memory

On a system with CXL memory installed, the resource tree (/proc/iomem)
related to CXL memory looks like something as follows.

490000000-50fffffff : CXL Window 0
  490000000-50fffffff : region0
    490000000-50fffffff : dax0.0
      490000000-50fffffff : System RAM (kmem)

When the following command line is run to try writing some memory in
CXL memory range,

 $ dd if=data of=/dev/mem bs=1k seek=19136512 count=1
 dd: error writing '/dev/mem': Bad address
 1+0 records in
 0+0 records out
 0 bytes copied, 0.0283507 s, 0.0 kB/s

the command fails as expected.  However, the error code is wrong.  It
should be "Operation not permitted" instead of "Bad address".  And,
the following warning is reported in kernel log.

 ioremap on RAM at 0x0000000490000000 - 0x0000000490000fff
 WARNING: CPU: 2 PID: 416 at arch/x86/mm/ioremap.c:216 __ioremap_caller.constprop.0+0x131/0x35d
 Modules linked in: cxl_pmem libnvdimm cbc encrypted_keys cxl_pmu
 CPU: 2 UID: 0 PID: 416 Comm: dd Not tainted 6.11.0-rc3-kvm #40
 Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.16.3-0-ga6ed6b701f0a-prebuilt.qemu.org 04/01/2014
 RIP: 0010:__ioremap_caller.constprop.0+0x131/0x35d
 Code: 2d 80 3d 24 6a a1 02 00 75 c1 48 8d 54 24 70 48 8d b4 24 90 00 00 00 48 c7 c7 40 3a 05 8a c6 05 07 6a a1 02 01 e8 a3 a0 01 00 <0f> 0b eb 9d 48 8b 84 24 90 00 00 00 48 8d 4c 24 60 89 ea 48 bf 00
 RSP: 0018:ffff888105387bf0 EFLAGS: 00010282
 RAX: 0000000000000000 RBX: 0000000490000fff RCX: 0000000000000000
 RDX: 0000000000000001 RSI: 0000000000000003 RDI: ffffed1020a70f73
 RBP: 0000000000000000 R08: ffffed100d9fce92 R09: 0000000000000001
 R10: ffffffff892348e7 R11: ffffed100d9fce91 R12: 0000000490000000
 R13: 0000000000000001 R14: 0000000000000001 R15: ffff888105387ca0
 FS:  00007f86c438c740(0000) GS:ffff88806ce00000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: 000055ba75b1b818 CR3: 0000000005231000 CR4: 0000000000350eb0
 Call Trace:
  <TASK>
  ? __warn+0xd7/0x1b8
  ? __ioremap_caller.constprop.0+0x131/0x35d
  ? report_bug+0x136/0x19e
  ? __ioremap_caller.constprop.0+0x131/0x35d
  ? handle_bug+0x3c/0x64
  ? exc_invalid_op+0x13/0x38
  ? asm_exc_invalid_op+0x16/0x20
  ? irq_work_claim+0x16/0x38
  ? __ioremap_caller.constprop.0+0x131/0x35d
  ? tracer_hardirqs_off+0x18/0x16d
  ? kmem_cache_debug_flags+0x16/0x23
  ? memremap+0xcb/0x184
  ? iounmap+0xfe/0xfe
  ? lock_sync+0xc7/0xc7
  ? lock_sync+0xc7/0xc7
  ? rcu_is_watching+0x1c/0x38
  ? do_raw_read_unlock+0x37/0x42
  ? _raw_read_unlock+0x1f/0x2f
  memremap+0xcb/0x184
  ? pfn_valid+0x159/0x159
  ? resource_is_exclusive+0xba/0xc5
  xlate_dev_mem_ptr+0x25/0x2f
  write_mem+0x94/0xfb
  vfs_write+0x128/0x26d
  ? kernel_write+0x89/0x89
  ? rcu_is_watching+0x1c/0x38
  ? __might_fault+0x72/0xba
  ? __might_fault+0x72/0xba
  ? rcu_is_watching+0x1c/0x38
  ? lock_release+0xba/0x13e
  ? files_lookup_fd_raw+0x40/0x4b
  ? __fget_light+0x64/0x89
  ksys_write+0xac/0xfe
  ? __ia32_sys_read+0x40/0x40
  ? tracer_hardirqs_off+0x18/0x16d
  ? tracer_hardirqs_on+0x11/0x146
  do_syscall_64+0x9a/0xfd
  entry_SYSCALL_64_after_hwframe+0x4b/0x53
 RIP: 0033:0x7f86c4487140
 Code: 40 00 48 8b 15 c1 9c 0d 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b7 0f 1f 00 80 3d a1 24 0e 00 00 74 17 b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 58 c3 0f 1f 80 00 00 00 00 48 83 ec 28 48 89
 RSP: 002b:00007ffca9f62af8 EFLAGS: 00000202 ORIG_RAX: 0000000000000001
 RAX: ffffffffffffffda RBX: 0000000000000400 RCX: 00007f86c4487140
 RDX: 0000000000000400 RSI: 000055ba75b1a000 RDI: 0000000000000001
 RBP: 000055ba75b1a000 R08: 0000000000000000 R09: 00007f86c457c080
 R10: 00007f86c43a84d0 R11: 0000000000000202 R12: 0000000000000000
 R13: 0000000000000000 R14: 000055ba75b1a000 R15: 0000000000000400
  </TASK>
 irq event stamp: 0
 hardirqs last  enabled at (0): [<0000000000000000>] 0x0
 hardirqs last disabled at (0): [<ffffffff89091e85>] copy_process+0xb60/0x255f
 softirqs last  enabled at (0): [<ffffffff89091e85>] copy_process+0xb60/0x255f
 softirqs last disabled at (0): [<0000000000000000>] 0x0

After investigation, we found the following bug.

In the above resource tree, "System RAM" is a descendant of "CXL Window 0"
instead of a top level resource.  So, region_intersects() will report no
System RAM resources in the CXL memory region incorrectly, because it only
checks the top level resources.  Consequently, devmem_is_allowed() will
return 1 (allow access via /dev/mem) for CXL memory region incorrectly.
Fortunately, ioremap() doesn't allow to map System RAM and reject the
access.

However, region_intersects() needs to be fixed to work correctly with the
resources tree with CXL Window as above.  To fix it, we will search
matched resources in the descendant resources too.  So, we will not miss
any matched resources anymore.

Link: https://lkml.kernel.org/r/20240816020723.771196-1-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Jonathan Cameron <jonathan.cameron@huawei.com>
Cc: Dave Jiang <dave.jiang@intel.com>
Cc: Alison Schofield <alison.schofield@intel.com>
Cc: Vishal Verma <vishal.l.verma@intel.com>
Cc: Ira Weiny <ira.weiny@intel.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: Baoquan He <bhe@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm/slub: add check for s->flags in the alloc_tagging_slab_free_hook
Hao Ge [Fri, 16 Aug 2024 01:33:36 +0000 (09:33 +0800)]
mm/slub: add check for s->flags in the alloc_tagging_slab_free_hook

When enable CONFIG_MEMCG & CONFIG_KFENCE & CONFIG_KMEMLEAK, the following
warning always occurs,This is because the following call stack occurred:
mem_pool_alloc
    kmem_cache_alloc_noprof
        slab_alloc_node
            kfence_alloc

Once the kfence allocation is successful,slab->obj_exts will not be empty,
because it has already been assigned a value in kfence_init_pool.

Since in the prepare_slab_obj_exts_hook function,we perform a check for
s->flags & (SLAB_NO_OBJ_EXT | SLAB_NOLEAKTRACE),the alloc_tag_add function
will not be called as a result.Therefore,ref->ct remains NULL.

However,when we call mem_pool_free,since obj_ext is not empty, it
eventually leads to the alloc_tag_sub scenario being invoked.  This is
where the warning occurs.

So we should add corresponding checks in the alloc_tagging_slab_free_hook.
For __GFP_NO_OBJ_EXT case,I didn't see the specific case where it's using
kfence,so I won't add the corresponding check in
alloc_tagging_slab_free_hook for now.

[    3.734349] ------------[ cut here ]------------
[    3.734807] alloc_tag was not set
[    3.735129] WARNING: CPU: 4 PID: 40 at ./include/linux/alloc_tag.h:130 kmem_cache_free+0x444/0x574
[    3.735866] Modules linked in: autofs4
[    3.736211] CPU: 4 UID: 0 PID: 40 Comm: ksoftirqd/4 Tainted: G        W          6.11.0-rc3-dirty #1
[    3.736969] Tainted: [W]=WARN
[    3.737258] Hardware name: QEMU KVM Virtual Machine, BIOS unknown 2/2/2022
[    3.737875] pstate: 60400005 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[    3.738501] pc : kmem_cache_free+0x444/0x574
[    3.738951] lr : kmem_cache_free+0x444/0x574
[    3.739361] sp : ffff80008357bb60
[    3.739693] x29: ffff80008357bb70 x28: 0000000000000000 x27: 0000000000000000
[    3.740338] x26: ffff80008207f000 x25: ffff000b2eb2fd60 x24: ffff0000c0005700
[    3.740982] x23: ffff8000804229e4 x22: ffff800082080000 x21: ffff800081756000
[    3.741630] x20: fffffd7ff8253360 x19: 00000000000000a8 x18: ffffffffffffffff
[    3.742274] x17: ffff800ab327f000 x16: ffff800083398000 x15: ffff800081756df0
[    3.742919] x14: 0000000000000000 x13: 205d344320202020 x12: 5b5d373038343337
[    3.743560] x11: ffff80008357b650 x10: 000000000000005d x9 : 00000000ffffffd0
[    3.744231] x8 : 7f7f7f7f7f7f7f7f x7 : ffff80008237bad0 x6 : c0000000ffff7fff
[    3.744907] x5 : ffff80008237ba78 x4 : ffff8000820bbad0 x3 : 0000000000000001
[    3.745580] x2 : 68d66547c09f7800 x1 : 68d66547c09f7800 x0 : 0000000000000000
[    3.746255] Call trace:
[    3.746530]  kmem_cache_free+0x444/0x574
[    3.746931]  mem_pool_free+0x44/0xf4
[    3.747306]  free_object_rcu+0xc8/0xdc
[    3.747693]  rcu_do_batch+0x234/0x8a4
[    3.748075]  rcu_core+0x230/0x3e4
[    3.748424]  rcu_core_si+0x14/0x1c
[    3.748780]  handle_softirqs+0x134/0x378
[    3.749189]  run_ksoftirqd+0x70/0x9c
[    3.749560]  smpboot_thread_fn+0x148/0x22c
[    3.749978]  kthread+0x10c/0x118
[    3.750323]  ret_from_fork+0x10/0x20
[    3.750696] ---[ end trace 0000000000000000 ]---

Link: https://lkml.kernel.org/r/20240816013336.17505-1-hao.ge@linux.dev
Fixes: 4b8736964640 ("mm/slab: add allocation accounting into slab allocation and free paths")
Signed-off-by: Hao Ge <gehao@kylinos.cn>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Kees Cook <kees@kernel.org>
Cc: Kent Overstreet <kent.overstreet@linux.dev>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agokunit/overflow: fix UB in overflow_allocation_test
Ivan Orlov [Thu, 15 Aug 2024 00:04:31 +0000 (01:04 +0100)]
kunit/overflow: fix UB in overflow_allocation_test

The 'device_name' array doesn't exist out of the
'overflow_allocation_test' function scope.  However, it is being used
as a driver name when calling 'kunit_driver_create' from
'kunit_device_register'.  It produces the kernel panic with KASAN
enabled due to undefined behaviour.

Since this variable is used in one place only, remove it and pass the
device name into kunit_device_register directly as an ascii string.

Link: https://lkml.kernel.org/r/20240815000431.401869-1-ivan.orlov0322@gmail.com
Fixes: ca90800a91ba ("test_overflow: Add memory allocation overflow tests")
Signed-off-by: Ivan Orlov <ivan.orlov0322@gmail.com>
Tested-by: Erhard Furtner <erhard_f@mailbox.org>
Reviewed-by: David Gow <davidgow@google.com>
Cc: Kees Cook <kees@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agonilfs2: fix state management in error path of log writing function
Ryusuke Konishi [Wed, 14 Aug 2024 10:11:19 +0000 (19:11 +0900)]
nilfs2: fix state management in error path of log writing function

After commit a694291a6211 ("nilfs2: separate wait function from
nilfs_segctor_write") was applied, the log writing function
nilfs_segctor_do_construct() was able to issue I/O requests continuously
even if user data blocks were split into multiple logs across segments,
but two potential flaws were introduced in its error handling.

First, if nilfs_segctor_begin_construction() fails while creating the
second or subsequent logs, the log writing function returns without
calling nilfs_segctor_abort_construction(), so the writeback flag set on
pages/folios will remain uncleared.  This causes page cache operations to
hang waiting for the writeback flag.  For example,
truncate_inode_pages_final(), which is called via nilfs_evict_inode() when
an inode is evicted from memory, will hang.

Second, the NILFS_I_COLLECTED flag set on normal inodes remain uncleared.
As a result, if the next log write involves checkpoint creation, that's
fine, but if a partial log write is performed that does not, inodes with
NILFS_I_COLLECTED set are erroneously removed from the "sc_dirty_files"
list, and their data and b-tree blocks may not be written to the device,
corrupting the block mapping.

Fix these issues by uniformly calling nilfs_segctor_abort_construction()
on failure of each step in the loop in nilfs_segctor_do_construct(),
having it clean up logs and segment usages according to progress, and
correcting the conditions for calling nilfs_redirty_inodes() to ensure
that the NILFS_I_COLLECTED flag is cleared.

Link: https://lkml.kernel.org/r/20240814101119.4070-1-konishi.ryusuke@gmail.com
Fixes: a694291a6211 ("nilfs2: separate wait function from nilfs_segctor_write")
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Tested-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agonilfs2: fix missing cleanup on rollforward recovery error
Ryusuke Konishi [Sat, 10 Aug 2024 06:52:42 +0000 (15:52 +0900)]
nilfs2: fix missing cleanup on rollforward recovery error

In an error injection test of a routine for mount-time recovery, KASAN
found a use-after-free bug.

It turned out that if data recovery was performed using partial logs
created by dsync writes, but an error occurred before starting the log
writer to create a recovered checkpoint, the inodes whose data had been
recovered were left in the ns_dirty_files list of the nilfs object and
were not freed.

Fix this issue by cleaning up inodes that have read the recovery data if
the recovery routine fails midway before the log writer starts.

Link: https://lkml.kernel.org/r/20240810065242.3701-1-konishi.ryusuke@gmail.com
Fixes: 0f3e1c7f23f8 ("nilfs2: recovery functions")
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Tested-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agonilfs2: protect references to superblock parameters exposed in sysfs
Ryusuke Konishi [Sun, 11 Aug 2024 10:03:20 +0000 (19:03 +0900)]
nilfs2: protect references to superblock parameters exposed in sysfs

The superblock buffers of nilfs2 can not only be overwritten at runtime
for modifications/repairs, but they are also regularly swapped, replaced
during resizing, and even abandoned when degrading to one side due to
backing device issues.  So, accessing them requires mutual exclusion using
the reader/writer semaphore "nilfs->ns_sem".

Some sysfs attribute show methods read this superblock buffer without the
necessary mutual exclusion, which can cause problems with pointer
dereferencing and memory access, so fix it.

Link: https://lkml.kernel.org/r/20240811100320.9913-1-konishi.ryusuke@gmail.com
Fixes: da7141fb78db ("nilfs2: add /sys/fs/nilfs2/<device> group")
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agouserfaultfd: don't BUG_ON() if khugepaged yanks our page table
Jann Horn [Tue, 13 Aug 2024 20:25:22 +0000 (22:25 +0200)]
userfaultfd: don't BUG_ON() if khugepaged yanks our page table

Since khugepaged was changed to allow retracting page tables in file
mappings without holding the mmap lock, these BUG_ON()s are wrong - get
rid of them.

We could also remove the preceding "if (unlikely(...))" block, but then we
could reach pte_offset_map_lock() with transhuge pages not just for file
mappings but also for anonymous mappings - which would probably be fine
but I think is not necessarily expected.

Link: https://lkml.kernel.org/r/20240813-uffd-thp-flip-fix-v2-2-5efa61078a41@google.com
Fixes: 1d65b771bc08 ("mm/khugepaged: retract_page_tables() without mmap or vma lock")
Signed-off-by: Jann Horn <jannh@google.com>
Reviewed-by: Qi Zheng <zhengqi.arch@bytedance.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Pavel Emelyanov <xemul@virtuozzo.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agouserfaultfd: fix checks for huge PMDs
Jann Horn [Tue, 13 Aug 2024 20:25:21 +0000 (22:25 +0200)]
userfaultfd: fix checks for huge PMDs

Patch series "userfaultfd: fix races around pmd_trans_huge() check", v2.

The pmd_trans_huge() code in mfill_atomic() is wrong in three different
ways depending on kernel version:

1. The pmd_trans_huge() check is racy and can lead to a BUG_ON() (if you hit
   the right two race windows) - I've tested this in a kernel build with
   some extra mdelay() calls. See the commit message for a description
   of the race scenario.
   On older kernels (before 6.5), I think the same bug can even
   theoretically lead to accessing transhuge page contents as a page table
   if you hit the right 5 narrow race windows (I haven't tested this case).
2. As pointed out by Qi Zheng, pmd_trans_huge() is not sufficient for
   detecting PMDs that don't point to page tables.
   On older kernels (before 6.5), you'd just have to win a single fairly
   wide race to hit this.
   I've tested this on 6.1 stable by racing migration (with a mdelay()
   patched into try_to_migrate()) against UFFDIO_ZEROPAGE - on my x86
   VM, that causes a kernel oops in ptlock_ptr().
3. On newer kernels (>=6.5), for shmem mappings, khugepaged is allowed
   to yank page tables out from under us (though I haven't tested that),
   so I think the BUG_ON() checks in mfill_atomic() are just wrong.

I decided to write two separate fixes for these (one fix for bugs 1+2, one
fix for bug 3), so that the first fix can be backported to kernels
affected by bugs 1+2.

This patch (of 2):

This fixes two issues.

I discovered that the following race can occur:

  mfill_atomic                other thread
  ============                ============
                              <zap PMD>
  pmdp_get_lockless() [reads none pmd]
  <bail if trans_huge>
  <if none:>
                              <pagefault creates transhuge zeropage>
    __pte_alloc [no-op]
                              <zap PMD>
  <bail if pmd_trans_huge(*dst_pmd)>
  BUG_ON(pmd_none(*dst_pmd))

I have experimentally verified this in a kernel with extra mdelay() calls;
the BUG_ON(pmd_none(*dst_pmd)) triggers.

On kernels newer than commit 0d940a9b270b ("mm/pgtable: allow
pte_offset_map[_lock]() to fail"), this can't lead to anything worse than
a BUG_ON(), since the page table access helpers are actually designed to
deal with page tables concurrently disappearing; but on older kernels
(<=6.4), I think we could probably theoretically race past the two
BUG_ON() checks and end up treating a hugepage as a page table.

The second issue is that, as Qi Zheng pointed out, there are other types
of huge PMDs that pmd_trans_huge() can't catch: devmap PMDs and swap PMDs
(in particular, migration PMDs).

On <=6.4, this is worse than the first issue: If mfill_atomic() runs on a
PMD that contains a migration entry (which just requires winning a single,
fairly wide race), it will pass the PMD to pte_offset_map_lock(), which
assumes that the PMD points to a page table.

Breakage follows: First, the kernel tries to take the PTE lock (which will
crash or maybe worse if there is no "struct page" for the address bits in
the migration entry PMD - I think at least on X86 there usually is no
corresponding "struct page" thanks to the PTE inversion mitigation, amd64
looks different).

If that didn't crash, the kernel would next try to write a PTE into what
it wrongly thinks is a page table.

As part of fixing these issues, get rid of the check for pmd_trans_huge()
before __pte_alloc() - that's redundant, we're going to have to check for
that after the __pte_alloc() anyway.

Backport note: pmdp_get_lockless() is pmd_read_atomic() in older kernels.

Link: https://lkml.kernel.org/r/20240813-uffd-thp-flip-fix-v2-0-5efa61078a41@google.com
Link: https://lkml.kernel.org/r/20240813-uffd-thp-flip-fix-v2-1-5efa61078a41@google.com
Fixes: c1a4de99fada ("userfaultfd: mcopy_atomic|mfill_zeropage: UFFDIO_COPY|UFFDIO_ZEROPAGE preparation")
Signed-off-by: Jann Horn <jannh@google.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jann Horn <jannh@google.com>
Cc: Pavel Emelyanov <xemul@virtuozzo.com>
Cc: Qi Zheng <zhengqi.arch@bytedance.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: vmalloc: ensure vmap_block is initialised before adding to queue
Will Deacon [Mon, 12 Aug 2024 17:16:06 +0000 (18:16 +0100)]
mm: vmalloc: ensure vmap_block is initialised before adding to queue

Commit 8c61291fd850 ("mm: fix incorrect vbq reference in
purge_fragmented_block") extended the 'vmap_block' structure to contain a
'cpu' field which is set at allocation time to the id of the initialising
CPU.

When a new 'vmap_block' is being instantiated by new_vmap_block(), the
partially initialised structure is added to the local 'vmap_block_queue'
xarray before the 'cpu' field has been initialised.  If another CPU is
concurrently walking the xarray (e.g.  via vm_unmap_aliases()), then it
may perform an out-of-bounds access to the remote queue thanks to an
uninitialised index.

This has been observed as UBSAN errors in Android:

 | Internal error: UBSAN: array index out of bounds: 00000000f2005512 [#1] PREEMPT SMP
 |
 | Call trace:
 |  purge_fragmented_block+0x204/0x21c
 |  _vm_unmap_aliases+0x170/0x378
 |  vm_unmap_aliases+0x1c/0x28
 |  change_memory_common+0x1dc/0x26c
 |  set_memory_ro+0x18/0x24
 |  module_enable_ro+0x98/0x238
 |  do_init_module+0x1b0/0x310

Move the initialisation of 'vb->cpu' in new_vmap_block() ahead of the
addition to the xarray.

Link: https://lkml.kernel.org/r/20240812171606.17486-1-will@kernel.org
Fixes: 8c61291fd850 ("mm: fix incorrect vbq reference in purge_fragmented_block")
Signed-off-by: Will Deacon <will@kernel.org>
Reviewed-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Cc: Zhaoyang Huang <zhaoyang.huang@unisoc.com>
Cc: Hailong.Liu <hailong.liu@oppo.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agoselftests: mm: fix build errors on armhf
Muhammad Usama Anjum [Fri, 9 Aug 2024 08:25:11 +0000 (13:25 +0500)]
selftests: mm: fix build errors on armhf

The __NR_mmap isn't found on armhf.  The mmap() is commonly available
system call and its wrapper is present on all architectures.  So it should
be used directly.  It solves problem for armhf and doesn't create problem
for other architectures.

Remove sys_mmap() functions as they aren't doing anything else other than
calling mmap().  There is no need to set errno = 0 manually as glibc
always resets it.

For reference errors are as following:

  CC       seal_elf
seal_elf.c: In function 'sys_mmap':
seal_elf.c:39:33: error: '__NR_mmap' undeclared (first use in this function)
   39 |         sret = (void *) syscall(__NR_mmap, addr, len, prot,
      |                                 ^~~~~~~~~

mseal_test.c: In function 'sys_mmap':
mseal_test.c:90:33: error: '__NR_mmap' undeclared (first use in this function)
   90 |         sret = (void *) syscall(__NR_mmap, addr, len, prot,
      |                                 ^~~~~~~~~

Link: https://lkml.kernel.org/r/20240809082511.497266-1-usama.anjum@collabora.com
Fixes: 4926c7a52de7 ("selftest mm/mseal memory sealing")
Signed-off-by: Muhammad Usama Anjum <usama.anjum@collabora.com>
Cc: Jeff Xu <jeffxu@chromium.org>
Cc: Kees Cook <kees@kernel.org>
Cc: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Shuah Khan <shuah@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm/migrate: fix deadlock in migrate_pages_batch() on large folios
Gao Xiang [Mon, 29 Jul 2024 02:13:06 +0000 (10:13 +0800)]
mm/migrate: fix deadlock in migrate_pages_batch() on large folios

Currently, migrate_pages_batch() can lock multiple locked folios with an
arbitrary order.  Although folio_trylock() is used to avoid deadlock as
commit 2ef7dbb26990 ("migrate_pages: try migrate in batch asynchronously
firstly") mentioned, it seems try_split_folio() is still missing.

It was found by compaction stress test when I explicitly enable EROFS
compressed files to use large folios, which case I cannot reproduce with
the same workload if large folio support is off (current mainline).
Typically, filesystem reads (with locked file-backed folios) could use
another bdev/meta inode to load some other I/Os (e.g.  inode extent
metadata or caching compressed data), so the locking order will be:

  file-backed folios  (A)
     bdev/meta folios (B)

The following calltrace shows the deadlock:
   Thread 1 takes (B) lock and tries to take folio (A) lock
   Thread 2 takes (A) lock and tries to take folio (B) lock

[Thread 1]
INFO: task stress:1824 blocked for more than 30 seconds.
      Tainted: G           OE      6.10.0-rc7+ #6
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
task:stress          state:D stack:0     pid:1824  tgid:1824  ppid:1822   flags:0x0000000c
Call trace:
 __switch_to+0xec/0x138
 __schedule+0x43c/0xcb0
 schedule+0x54/0x198
 io_schedule+0x44/0x70
 folio_wait_bit_common+0x184/0x3f8
<-- folio mapping ffff00036d69cb18 index 996  (**)
 __folio_lock+0x24/0x38
 migrate_pages_batch+0x77c/0xea0 // try_split_folio (mm/migrate.c:1486:2)
// migrate_pages_batch (mm/migrate.c:1734:16)
<--- LIST_HEAD(unmap_folios) has
..
folio mapping 0xffff0000d184f1d8 index 1711;   (*)
folio mapping 0xffff0000d184f1d8 index 1712;
..
 migrate_pages+0xb28/0xe90
 compact_zone+0xa08/0x10f0
 compact_node+0x9c/0x180
 sysctl_compaction_handler+0x8c/0x118
 proc_sys_call_handler+0x1a8/0x280
 proc_sys_write+0x1c/0x30
 vfs_write+0x240/0x380
 ksys_write+0x78/0x118
 __arm64_sys_write+0x24/0x38
 invoke_syscall+0x78/0x108
 el0_svc_common.constprop.0+0x48/0xf0
 do_el0_svc+0x24/0x38
 el0_svc+0x3c/0x148
 el0t_64_sync_handler+0x100/0x130
 el0t_64_sync+0x190/0x198

[Thread 2]
INFO: task stress:1825 blocked for more than 30 seconds.
      Tainted: G           OE      6.10.0-rc7+ #6
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
task:stress          state:D stack:0     pid:1825  tgid:1825  ppid:1822   flags:0x0000000c
Call trace:
 __switch_to+0xec/0x138
 __schedule+0x43c/0xcb0
 schedule+0x54/0x198
 io_schedule+0x44/0x70
 folio_wait_bit_common+0x184/0x3f8
<-- folio = 0xfffffdffc6b503c0 (mapping == 0xffff0000d184f1d8 index == 1711) (*)
 __folio_lock+0x24/0x38
 z_erofs_runqueue+0x384/0x9c0 [erofs]
 z_erofs_readahead+0x21c/0x350 [erofs]       <-- folio mapping 0xffff00036d69cb18 range from [992, 1024] (**)
 read_pages+0x74/0x328
 page_cache_ra_order+0x26c/0x348
 ondemand_readahead+0x1c0/0x3a0
 page_cache_sync_ra+0x9c/0xc0
 filemap_get_pages+0xc4/0x708
 filemap_read+0x104/0x3a8
 generic_file_read_iter+0x4c/0x150
 vfs_read+0x27c/0x330
 ksys_pread64+0x84/0xd0
 __arm64_sys_pread64+0x28/0x40
 invoke_syscall+0x78/0x108
 el0_svc_common.constprop.0+0x48/0xf0
 do_el0_svc+0x24/0x38
 el0_svc+0x3c/0x148
 el0t_64_sync_handler+0x100/0x130
 el0t_64_sync+0x190/0x198

Link: https://lkml.kernel.org/r/20240729021306.398286-1-hsiangkao@linux.alibaba.com
Fixes: 5dfab109d519 ("migrate_pages: batch _unmap and _move")
Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
Reviewed-by: "Huang, Ying" <ying.huang@intel.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agoalloc_tag: mark pages reserved during CMA activation as not tagged
Suren Baghdasaryan [Tue, 13 Aug 2024 15:07:57 +0000 (08:07 -0700)]
alloc_tag: mark pages reserved during CMA activation as not tagged

During CMA activation, pages in CMA area are prepared and then freed
without being allocated.  This triggers warnings when memory allocation
debug config (CONFIG_MEM_ALLOC_PROFILING_DEBUG) is enabled.  Fix this by
marking these pages not tagged before freeing them.

Link: https://lkml.kernel.org/r/20240813150758.855881-2-surenb@google.com
Fixes: d224eb0287fb ("codetag: debug: mark codetags for reserved pages as empty")
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Kent Overstreet <kent.overstreet@linux.dev>
Cc: Pasha Tatashin <pasha.tatashin@soleen.com>
Cc: Sourav Panda <souravpanda@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: <stable@vger.kernel.org> [6.10]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agoalloc_tag: introduce clear_page_tag_ref() helper function
Suren Baghdasaryan [Tue, 13 Aug 2024 15:07:56 +0000 (08:07 -0700)]
alloc_tag: introduce clear_page_tag_ref() helper function

In several cases we are freeing pages which were not allocated using
common page allocators.  For such cases, in order to keep allocation
accounting correct, we should clear the page tag to indicate that the page
being freed is expected to not have a valid allocation tag.  Introduce
clear_page_tag_ref() helper function to be used for this.

Link: https://lkml.kernel.org/r/20240813150758.855881-1-surenb@google.com
Fixes: d224eb0287fb ("codetag: debug: mark codetags for reserved pages as empty")
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Suggested-by: David Hildenbrand <david@redhat.com>
Acked-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Kent Overstreet <kent.overstreet@linux.dev>
Cc: Sourav Panda <souravpanda@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: <stable@vger.kernel.org> [6.10]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agocrash: fix riscv64 crash memory reserve dead loop
Jinjie Ruan [Mon, 12 Aug 2024 06:20:17 +0000 (14:20 +0800)]
crash: fix riscv64 crash memory reserve dead loop

On RISCV64 Qemu machine with 512MB memory, cmdline "crashkernel=500M,high"
will cause system stall as below:

 Zone ranges:
   DMA32    [mem 0x0000000080000000-0x000000009fffffff]
   Normal   empty
 Movable zone start for each node
 Early memory node ranges
   node   0: [mem 0x0000000080000000-0x000000008005ffff]
   node   0: [mem 0x0000000080060000-0x000000009fffffff]
 Initmem setup node 0 [mem 0x0000000080000000-0x000000009fffffff]
(stall here)

commit 5d99cadf1568 ("crash: fix x86_32 crash memory reserve dead loop
bug") fix this on 32-bit architecture.  However, the problem is not
completely solved.  If `CRASH_ADDR_LOW_MAX = CRASH_ADDR_HIGH_MAX` on
64-bit architecture, for example, when system memory is equal to
CRASH_ADDR_LOW_MAX on RISCV64, the following infinite loop will also
occur:

-> reserve_crashkernel_generic() and high is true
   -> alloc at [CRASH_ADDR_LOW_MAX, CRASH_ADDR_HIGH_MAX] fail
      -> alloc at [0, CRASH_ADDR_LOW_MAX] fail and repeatedly
         (because CRASH_ADDR_LOW_MAX = CRASH_ADDR_HIGH_MAX).

As Catalin suggested, do not remove the ",high" reservation fallback to
",low" logic which will change arm64's kdump behavior, but fix it by
skipping the above situation similar to commit d2f32f23190b ("crash: fix
x86_32 crash memory reserve dead loop").

After this patch, it print:
cannot allocate crashkernel (size:0x1f400000)

Link: https://lkml.kernel.org/r/20240812062017.2674441-1-ruanjinjie@huawei.com
Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Suggested-by: Catalin Marinas <catalin.marinas@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: Baoquan He <bhe@redhat.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Dave Young <dyoung@redhat.com>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <paul.walmsley@sifive.com>
Cc: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agoselftests: memfd_secret: don't build memfd_secret test on unsupported arches
Muhammad Usama Anjum [Fri, 9 Aug 2024 07:56:42 +0000 (12:56 +0500)]
selftests: memfd_secret: don't build memfd_secret test on unsupported arches

[1] mentions that memfd_secret is only supported on arm64, riscv, x86 and
x86_64 for now.  It doesn't support other architectures.  I found the
build error on arm and decided to send the fix as it was creating noise on
KernelCI:

memfd_secret.c: In function 'memfd_secret':
memfd_secret.c:42:24: error: '__NR_memfd_secret' undeclared (first use in this function);
did you mean 'memfd_secret'?
   42 |         return syscall(__NR_memfd_secret, flags);
      |                        ^~~~~~~~~~~~~~~~~
      |                        memfd_secret

Hence I'm adding condition that memfd_secret should only be compiled on
supported architectures.

Also check in run_vmtests script if memfd_secret binary is present before
executing it.

Link: https://lkml.kernel.org/r/20240812061522.1933054-1-usama.anjum@collabora.com
Link: https://lore.kernel.org/all/20210518072034.31572-7-rppt@kernel.org/
Link: https://lkml.kernel.org/r/20240809075642.403247-1-usama.anjum@collabora.com
Fixes: 76fe17ef588a ("secretmem: test: add basic selftest for memfd_secret(2)")
Signed-off-by: Muhammad Usama Anjum <usama.anjum@collabora.com>
Reviewed-by: Shuah Khan <skhan@linuxfoundation.org>
Acked-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Cc: Mike Rapoport (Microsoft) <rppt@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <paul.walmsley@sifive.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: fix endless reclaim on machines with unaccepted memory
Kirill A. Shutemov [Fri, 9 Aug 2024 11:48:47 +0000 (14:48 +0300)]
mm: fix endless reclaim on machines with unaccepted memory

Unaccepted memory is considered unusable free memory, which is not counted
as free on the zone watermark check.  This causes get_page_from_freelist()
to accept more memory to hit the high watermark, but it creates problems
in the reclaim path.

The reclaim path encounters a failed zone watermark check and attempts to
reclaim memory.  This is usually successful, but if there is little or no
reclaimable memory, it can result in endless reclaim with little to no
progress.  This can occur early in the boot process, just after start of
the init process when the only reclaimable memory is the page cache of the
init executable and its libraries.

Make unaccepted memory free from watermark check point of view.  This way
unaccepted memory will never be the trigger of memory reclaim.  Accept
more memory in the get_page_from_freelist() if needed.

Link: https://lkml.kernel.org/r/20240809114854.3745464-2-kirill.shutemov@linux.intel.com
Fixes: dcdfdd40fa82 ("mm: Add support for unaccepted memory")
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Reported-by: Jianxiong Gao <jxgao@google.com>
Acked-by: David Hildenbrand <david@redhat.com>
Tested-by: Jianxiong Gao <jxgao@google.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Mike Rapoport (Microsoft) <rppt@kernel.org>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: <stable@vger.kernel.org> [6.5+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agoselftests/mm: compaction_test: fix off by one in check_compaction()
Dan Carpenter [Fri, 9 Aug 2024 12:32:30 +0000 (15:32 +0300)]
selftests/mm: compaction_test: fix off by one in check_compaction()

The "initial_nr_hugepages" variable is unsigned long so it takes up to 20
characters to print, plus 1 more character for the NUL terminator.
Unfortunately, this buffer is not quite large enough for the terminator to
fit.  Also use snprintf() for a belt and suspenders approach.

Link: https://lkml.kernel.org/r/87470c06-b45a-4e83-92ff-aac2e7b9c6ba@stanley.mountain
Fixes: fb9293b6b015 ("selftests/mm: compaction_test: fix bogus test success and reduce probability of OOM-killer invocation")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Cc: Shuah Khan <shuah@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm/numa: no task_numa_fault() call if PMD is changed
Zi Yan [Fri, 9 Aug 2024 14:59:05 +0000 (10:59 -0400)]
mm/numa: no task_numa_fault() call if PMD is changed

When handling a numa page fault, task_numa_fault() should be called by a
process that restores the page table of the faulted folio to avoid
duplicated stats counting.  Commit c5b5a3dd2c1f ("mm: thp: refactor NUMA
fault handling") restructured do_huge_pmd_numa_page() and did not avoid
task_numa_fault() call in the second page table check after a numa
migration failure.  Fix it by making all !pmd_same() return immediately.

This issue can cause task_numa_fault() being called more than necessary
and lead to unexpected numa balancing results (It is hard to tell whether
the issue will cause positive or negative performance impact due to
duplicated numa fault counting).

Link: https://lkml.kernel.org/r/20240809145906.1513458-3-ziy@nvidia.com
Fixes: c5b5a3dd2c1f ("mm: thp: refactor NUMA fault handling")
Reported-by: "Huang, Ying" <ying.huang@intel.com>
Closes: https://lore.kernel.org/linux-mm/87zfqfw0yw.fsf@yhuang6-desk2.ccr.corp.intel.com/
Signed-off-by: Zi Yan <ziy@nvidia.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Yang Shi <shy828301@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm/numa: no task_numa_fault() call if PTE is changed
Zi Yan [Fri, 9 Aug 2024 14:59:04 +0000 (10:59 -0400)]
mm/numa: no task_numa_fault() call if PTE is changed

When handling a numa page fault, task_numa_fault() should be called by a
process that restores the page table of the faulted folio to avoid
duplicated stats counting.  Commit b99a342d4f11 ("NUMA balancing: reduce
TLB flush via delaying mapping on hint page fault") restructured
do_numa_page() and did not avoid task_numa_fault() call in the second page
table check after a numa migration failure.  Fix it by making all
!pte_same() return immediately.

This issue can cause task_numa_fault() being called more than necessary
and lead to unexpected numa balancing results (It is hard to tell whether
the issue will cause positive or negative performance impact due to
duplicated numa fault counting).

Link: https://lkml.kernel.org/r/20240809145906.1513458-2-ziy@nvidia.com
Fixes: b99a342d4f11 ("NUMA balancing: reduce TLB flush via delaying mapping on hint page fault")
Signed-off-by: Zi Yan <ziy@nvidia.com>
Reported-by: "Huang, Ying" <ying.huang@intel.com>
Closes: https://lore.kernel.org/linux-mm/87zfqfw0yw.fsf@yhuang6-desk2.ccr.corp.intel.com/
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Yang Shi <shy828301@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm/vmalloc: fix page mapping if vm_area_alloc_pages() with high order fallback to...
Hailong Liu [Thu, 8 Aug 2024 12:19:56 +0000 (20:19 +0800)]
mm/vmalloc: fix page mapping if vm_area_alloc_pages() with high order fallback to order 0

The __vmap_pages_range_noflush() assumes its argument pages** contains
pages with the same page shift.  However, since commit e9c3cda4d86e ("mm,
vmalloc: fix high order __GFP_NOFAIL allocations"), if gfp_flags includes
__GFP_NOFAIL with high order in vm_area_alloc_pages() and page allocation
failed for high order, the pages** may contain two different page shifts
(high order and order-0).  This could lead __vmap_pages_range_noflush() to
perform incorrect mappings, potentially resulting in memory corruption.

Users might encounter this as follows (vmap_allow_huge = true, 2M is for
PMD_SIZE):

kvmalloc(2M, __GFP_NOFAIL|GFP_X)
    __vmalloc_node_range_noprof(vm_flags=VM_ALLOW_HUGE_VMAP)
        vm_area_alloc_pages(order=9) ---> order-9 allocation failed and fallback to order-0
            vmap_pages_range()
                vmap_pages_range_noflush()
                    __vmap_pages_range_noflush(page_shift = 21) ----> wrong mapping happens

We can remove the fallback code because if a high-order allocation fails,
__vmalloc_node_range_noprof() will retry with order-0.  Therefore, it is
unnecessary to fallback to order-0 here.  Therefore, fix this by removing
the fallback code.

Link: https://lkml.kernel.org/r/20240808122019.3361-1-hailong.liu@oppo.com
Fixes: e9c3cda4d86e ("mm, vmalloc: fix high order __GFP_NOFAIL allocations")
Signed-off-by: Hailong Liu <hailong.liu@oppo.com>
Reported-by: Tangquan Zheng <zhengtangquan@oppo.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Acked-by: Barry Song <baohua@kernel.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm/memory-failure: use raw_spinlock_t in struct memory_failure_cpu
Waiman Long [Tue, 6 Aug 2024 16:41:07 +0000 (12:41 -0400)]
mm/memory-failure: use raw_spinlock_t in struct memory_failure_cpu

The memory_failure_cpu structure is a per-cpu structure.  Access to its
content requires the use of get_cpu_var() to lock in the current CPU and
disable preemption.  The use of a regular spinlock_t for locking purpose
is fine for a non-RT kernel.

Since the integration of RT spinlock support into the v5.15 kernel, a
spinlock_t in a RT kernel becomes a sleeping lock and taking a sleeping
lock in a preemption disabled context is illegal resulting in the
following kind of warning.

  [12135.732244] BUG: sleeping function called from invalid context at kernel/locking/spinlock_rt.c:48
  [12135.732248] in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 270076, name: kworker/0:0
  [12135.732252] preempt_count: 1, expected: 0
  [12135.732255] RCU nest depth: 2, expected: 2
    :
  [12135.732420] Hardware name: Dell Inc. PowerEdge R640/0HG0J8, BIOS 2.10.2 02/24/2021
  [12135.732423] Workqueue: kacpi_notify acpi_os_execute_deferred
  [12135.732433] Call Trace:
  [12135.732436]  <TASK>
  [12135.732450]  dump_stack_lvl+0x57/0x81
  [12135.732461]  __might_resched.cold+0xf4/0x12f
  [12135.732479]  rt_spin_lock+0x4c/0x100
  [12135.732491]  memory_failure_queue+0x40/0xe0
  [12135.732503]  ghes_do_memory_failure+0x53/0x390
  [12135.732516]  ghes_do_proc.constprop.0+0x229/0x3e0
  [12135.732575]  ghes_proc+0xf9/0x1a0
  [12135.732591]  ghes_notify_hed+0x6a/0x150
  [12135.732602]  notifier_call_chain+0x43/0xb0
  [12135.732626]  blocking_notifier_call_chain+0x43/0x60
  [12135.732637]  acpi_ev_notify_dispatch+0x47/0x70
  [12135.732648]  acpi_os_execute_deferred+0x13/0x20
  [12135.732654]  process_one_work+0x41f/0x500
  [12135.732695]  worker_thread+0x192/0x360
  [12135.732715]  kthread+0x111/0x140
  [12135.732733]  ret_from_fork+0x29/0x50
  [12135.732779]  </TASK>

Fix it by using a raw_spinlock_t for locking instead.

Also move the pr_err() out of the lock critical section and after
put_cpu_ptr() to avoid indeterminate latency and the possibility of sleep
with this call.

[longman@redhat.com: don't hold percpu ref across pr_err(), per Miaohe]
Link: https://lkml.kernel.org/r/20240807181130.1122660-1-longman@redhat.com
Link: https://lkml.kernel.org/r/20240806164107.1044956-1-longman@redhat.com
Fixes: 0f383b6dc96e ("locking/spinlock: Provide RT variant")
Signed-off-by: Waiman Long <longman@redhat.com>
Acked-by: Miaohe Lin <linmiaohe@huawei.com>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: Len Brown <len.brown@intel.com>
Cc: Naoya Horiguchi <nao.horiguchi@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: don't account memmap per-node
Pasha Tatashin [Thu, 8 Aug 2024 21:34:36 +0000 (21:34 +0000)]
mm: don't account memmap per-node

Fix invalid access to pgdat during hot-remove operation:
ndctl users reported a GPF when trying to destroy a namespace:
$ ndctl destroy-namespace all -r all -f
 Segmentation fault
 dmesg:
 Oops: general protection fault, probably for
 non-canonical address 0xdffffc0000005650: 0000 [#1] PREEMPT SMP KASAN
 PTI
 KASAN: probably user-memory-access in range
 [0x000000000002b280-0x000000000002b287]
 CPU: 26 UID: 0 PID: 1868 Comm: ndctl Not tainted 6.11.0-rc1 #1
 Hardware name: Dell Inc. PowerEdge R640/08HT8T, BIOS
 2.20.1 09/13/2023
 RIP: 0010:mod_node_page_state+0x2a/0x110

cxl-test users report a GPF when trying to unload the test module:
$ modrpobe -r cxl-test
 dmesg
 BUG: unable to handle page fault for address: 0000000000004200
 #PF: supervisor read access in kernel mode
 #PF: error_code(0x0000) - not-present page
 PGD 0 P4D 0
 Oops: Oops: 0000 [#1] PREEMPT SMP PTI
 CPU: 0 UID: 0 PID: 1076 Comm: modprobe Tainted: G O N 6.11.0-rc1 #197
 Tainted: [O]=OOT_MODULE, [N]=TEST
 Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/15
 RIP: 0010:mod_node_page_state+0x6/0x90

Currently, when memory is hot-plugged or hot-removed the accounting is
done based on the assumption that memmap is allocated from the same node
as the hot-plugged/hot-removed memory, which is not always the case.

In addition, there are challenges with keeping the node id of the memory
that is being remove to the time when memmap accounting is actually
performed: since this is done after remove_pfn_range_from_zone(), and
also after remove_memory_block_devices(). Meaning that we cannot use
pgdat nor walking though memblocks to get the nid.

Given all of that, account the memmap overhead system wide instead.

For this we are going to be using global atomic counters, but given that
memmap size is rarely modified, and normally is only modified either
during early boot when there is only one CPU, or under a hotplug global
mutex lock, therefore there is no need for per-cpu optimizations.

Also, while we are here rename nr_memmap to nr_memmap_pages, and
nr_memmap_boot to nr_memmap_boot_pages to be self explanatory that the
units are in page count.

[pasha.tatashin@soleen.com: address a few nits from David Hildenbrand]
Link: https://lkml.kernel.org/r/20240809191020.1142142-4-pasha.tatashin@soleen.com
Link: https://lkml.kernel.org/r/20240809191020.1142142-4-pasha.tatashin@soleen.com
Link: https://lkml.kernel.org/r/20240808213437.682006-4-pasha.tatashin@soleen.com
Fixes: 15995a352474 ("mm: report per-page metadata information")
Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Reported-by: Yi Zhang <yi.zhang@redhat.com>
Closes: https://lore.kernel.org/linux-cxl/CAHj4cs9Ax1=CoJkgBGP_+sNu6-6=6v=_L-ZBZY0bVLD3wUWZQg@mail.gmail.com
Reported-by: Alison Schofield <alison.schofield@intel.com>
Closes: https://lore.kernel.org/linux-mm/Zq0tPd2h6alFz8XF@aschofie-mobl2/#t
Tested-by: Dan Williams <dan.j.williams@intel.com>
Tested-by: Alison Schofield <alison.schofield@intel.com>
Acked-by: David Hildenbrand <david@redhat.com>
Acked-by: David Rientjes <rientjes@google.com>
Tested-by: Yi Zhang <yi.zhang@redhat.com>
Cc: Domenico Cerasuolo <cerasuolodomenico@gmail.com>
Cc: Fan Ni <fan.ni@samsung.com>
Cc: Joel Granados <j.granados@samsung.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Li Zhijian <lizhijian@fujitsu.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Sourav Panda <souravpanda@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: add system wide stats items category
Pasha Tatashin [Thu, 8 Aug 2024 21:34:35 +0000 (21:34 +0000)]
mm: add system wide stats items category

/proc/vmstat contains events and stats, events can only grow, but stats
can grow and shrink.

vmstat has the following:
-------------------------
NR_VM_ZONE_STAT_ITEMS: per-zone stats
NR_VM_NUMA_EVENT_ITEMS: per-numa events
NR_VM_NODE_STAT_ITEMS: per-numa stats
NR_VM_WRITEBACK_STAT_ITEMS: system-wide background-writeback and
dirty-throttling tresholds.
NR_VM_EVENT_ITEMS: system-wide events
-------------------------

Rename NR_VM_WRITEBACK_STAT_ITEMS to NR_VM_STAT_ITEMS, to track the
system-wide stats, we are going to add per-page metadata stats to this
category in the next patch.

Also delete unused writeback_stat_name().

Link: https://lkml.kernel.org/r/20240809191020.1142142-2-pasha.tatashin@soleen.com
Link: https://lkml.kernel.org/r/20240808213437.682006-3-pasha.tatashin@soleen.com
Fixes: 15995a352474 ("mm: report per-page metadata information")
Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Suggested-by: Yosry Ahmed <yosryahmed@google.com>
Tested-by: Alison Schofield <alison.schofield@intel.com>
Acked-by: David Hildenbrand <david@redhat.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Domenico Cerasuolo <cerasuolodomenico@gmail.com>
Cc: Joel Granados <j.granados@samsung.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Li Zhijian <lizhijian@fujitsu.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Sourav Panda <souravpanda@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yi Zhang <yi.zhang@redhat.com>
Cc: Fan Ni <fan.ni@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm: don't account memmap on failure
Pasha Tatashin [Thu, 8 Aug 2024 21:34:34 +0000 (21:34 +0000)]
mm: don't account memmap on failure

Patch series "Fixes for memmap accounting", v4.

Memmap accounting provides us with observability of how much memory is
used for per-page metadata: i.e. "struct page"'s and "struct page_ext".
It also provides with information of how much was allocated using
boot allocator (i.e. not part of MemTotal), and how much was allocated
using buddy allocated (i.e. part of MemTotal).

This small series fixes a few problems that were discovered with the
original patch.

This patch (of 3):

When we fail to allocate the mmemmap in alloc_vmemmap_page_list(), do not
account any already-allocated pages: we're going to free all them before
we return from the function.

Link: https://lkml.kernel.org/r/20240809191020.1142142-1-pasha.tatashin@soleen.com
Link: https://lkml.kernel.org/r/20240808213437.682006-1-pasha.tatashin@soleen.com
Link: https://lkml.kernel.org/r/20240808213437.682006-2-pasha.tatashin@soleen.com
Fixes: 15995a352474 ("mm: report per-page metadata information")
Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Reviewed-by: Fan Ni <fan.ni@samsung.com>
Reviewed-by: Yosry Ahmed <yosryahmed@google.com>
Acked-by: David Hildenbrand <david@redhat.com>
Tested-by: Alison Schofield <alison.schofield@intel.com>
Reviewed-by: Muchun Song <muchun.song@linux.dev>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Domenico Cerasuolo <cerasuolodomenico@gmail.com>
Cc: Joel Granados <j.granados@samsung.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Li Zhijian <lizhijian@fujitsu.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Sourav Panda <souravpanda@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yi Zhang <yi.zhang@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomm/hugetlb: fix hugetlb vs. core-mm PT locking
David Hildenbrand [Thu, 1 Aug 2024 20:47:48 +0000 (22:47 +0200)]
mm/hugetlb: fix hugetlb vs. core-mm PT locking

We recently made GUP's common page table walking code to also walk hugetlb
VMAs without most hugetlb special-casing, preparing for the future of
having less hugetlb-specific page table walking code in the codebase.
Turns out that we missed one page table locking detail: page table locking
for hugetlb folios that are not mapped using a single PMD/PUD.

Assume we have hugetlb folio that spans multiple PTEs (e.g., 64 KiB
hugetlb folios on arm64 with 4 KiB base page size).  GUP, as it walks the
page tables, will perform a pte_offset_map_lock() to grab the PTE table
lock.

However, hugetlb that concurrently modifies these page tables would
actually grab the mm->page_table_lock: with USE_SPLIT_PTE_PTLOCKS, the
locks would differ.  Something similar can happen right now with hugetlb
folios that span multiple PMDs when USE_SPLIT_PMD_PTLOCKS.

This issue can be reproduced [1], for example triggering:

[ 3105.936100] ------------[ cut here ]------------
[ 3105.939323] WARNING: CPU: 31 PID: 2732 at mm/gup.c:142 try_grab_folio+0x11c/0x188
[ 3105.944634] Modules linked in: [...]
[ 3105.974841] CPU: 31 PID: 2732 Comm: reproducer Not tainted 6.10.0-64.eln141.aarch64 #1
[ 3105.980406] Hardware name: QEMU KVM Virtual Machine, BIOS edk2-20240524-4.fc40 05/24/2024
[ 3105.986185] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[ 3105.991108] pc : try_grab_folio+0x11c/0x188
[ 3105.994013] lr : follow_page_pte+0xd8/0x430
[ 3105.996986] sp : ffff80008eafb8f0
[ 3105.999346] x29: ffff80008eafb900 x28: ffffffe8d481f380 x27: 00f80001207cff43
[ 3106.004414] x26: 0000000000000001 x25: 0000000000000000 x24: ffff80008eafba48
[ 3106.009520] x23: 0000ffff9372f000 x22: ffff7a54459e2000 x21: ffff7a546c1aa978
[ 3106.014529] x20: ffffffe8d481f3c0 x19: 0000000000610041 x18: 0000000000000001
[ 3106.019506] x17: 0000000000000001 x16: ffffffffffffffff x15: 0000000000000000
[ 3106.024494] x14: ffffb85477fdfe08 x13: 0000ffff9372ffff x12: 0000000000000000
[ 3106.029469] x11: 1fffef4a88a96be1 x10: ffff7a54454b5f0c x9 : ffffb854771b12f0
[ 3106.034324] x8 : 0008000000000000 x7 : ffff7a546c1aa980 x6 : 0008000000000080
[ 3106.038902] x5 : 00000000001207cf x4 : 0000ffff9372f000 x3 : ffffffe8d481f000
[ 3106.043420] x2 : 0000000000610041 x1 : 0000000000000001 x0 : 0000000000000000
[ 3106.047957] Call trace:
[ 3106.049522]  try_grab_folio+0x11c/0x188
[ 3106.051996]  follow_pmd_mask.constprop.0.isra.0+0x150/0x2e0
[ 3106.055527]  follow_page_mask+0x1a0/0x2b8
[ 3106.058118]  __get_user_pages+0xf0/0x348
[ 3106.060647]  faultin_page_range+0xb0/0x360
[ 3106.063651]  do_madvise+0x340/0x598

Let's make huge_pte_lockptr() effectively use the same PT locks as any
core-mm page table walker would.  Add ptep_lockptr() to obtain the PTE
page table lock using a pte pointer -- unfortunately we cannot convert
pte_lockptr() because virt_to_page() doesn't work with kmap'ed page tables
we can have with CONFIG_HIGHPTE.

Handle CONFIG_PGTABLE_LEVELS correctly by checking in reverse order, such
that when e.g., CONFIG_PGTABLE_LEVELS==2 with
PGDIR_SIZE==P4D_SIZE==PUD_SIZE==PMD_SIZE will work as expected.  Document
why that works.

There is one ugly case: powerpc 8xx, whereby we have an 8 MiB hugetlb
folio being mapped using two PTE page tables.  While hugetlb wants to take
the PMD table lock, core-mm would grab the PTE table lock of one of both
PTE page tables.  In such corner cases, we have to make sure that both
locks match, which is (fortunately!) currently guaranteed for 8xx as it
does not support SMP and consequently doesn't use split PT locks.

[1] https://lore.kernel.org/all/1bbfcc7f-f222-45a5-ac44-c5a1381c596d@redhat.com/

Link: https://lkml.kernel.org/r/20240801204748.99107-1-david@redhat.com
Fixes: 9cb28da54643 ("mm/gup: handle hugetlb in the generic follow_page_mask code")
Signed-off-by: David Hildenbrand <david@redhat.com>
Acked-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Baolin Wang <baolin.wang@linux.alibaba.com>
Tested-by: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
10 months agomseal: fix is_madv_discard()
Pedro Falcato [Wed, 7 Aug 2024 17:33:35 +0000 (18:33 +0100)]
mseal: fix is_madv_discard()

is_madv_discard did its check wrong. MADV_ flags are not bitwise,
they're normal sequential numbers. So, for instance:
behavior & (/* ... */ | MADV_REMOVE)

tagged both MADV_REMOVE and MADV_RANDOM (bit 0 set) as discard
operations.

As a result the kernel could erroneously block certain madvises (e.g
MADV_RANDOM or MADV_HUGEPAGE) on sealed VMAs due to them sharing bits
with blocked MADV operations (e.g REMOVE or WIPEONFORK).

This is obviously incorrect, so use a switch statement instead.

Link: https://lkml.kernel.org/r/20240807173336.2523757-1-pedro.falcato@gmail.com
Link: https://lkml.kernel.org/r/20240807173336.2523757-2-pedro.falcato@gmail.com
Fixes: 8be7258aad44 ("mseal: add mseal syscall")
Signed-off-by: Pedro Falcato <pedro.falcato@gmail.com>
Tested-by: Jeff Xu <jeffxu@chromium.org>
Reviewed-by: Jeff Xu <jeffxu@chromium.org>
Cc: Kees Cook <kees@kernel.org>
Cc: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Shuah Khan <shuah@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 months agoLinux 6.11-rc3
Linus Torvalds [Sun, 11 Aug 2024 21:27:14 +0000 (14:27 -0700)]
Linux 6.11-rc3

11 months agoMerge tag 'x86-urgent-2024-08-11' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 11 Aug 2024 17:20:29 +0000 (10:20 -0700)]
Merge tag 'x86-urgent-2024-08-11' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull x86 fixes from Thomas Gleixner:

 - Fix 32-bit PTI for real.

   pti_clone_entry_text() is called twice, once before initcalls so that
   initcalls can use the user-mode helper and then again after text is
   set read only. Setting read only on 32-bit might break up the PMD
   mapping, which makes the second invocation of pti_clone_entry_text()
   find the mappings out of sync and failing.

   Allow the second call to split the existing PMDs in the user mapping
   and synchronize with the kernel mapping.

 - Don't make acpi_mp_wake_mailbox read-only after init as the mail box
   must be writable in the case that CPU hotplug operations happen after
   boot. Otherwise the attempt to start a CPU crashes with a write to
   read only memory.

 - Add a missing sanity check in mtrr_save_state() to ensure that the
   fixed MTRR MSRs are supported.

   Otherwise mtrr_save_state() ends up in a #GP, which is fixed up, but
   the WARN_ON() can bring systems down when panic on warn is set.

* tag 'x86-urgent-2024-08-11' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/mtrr: Check if fixed MTRRs exist before saving them
  x86/paravirt: Fix incorrect virt spinlock setting on bare metal
  x86/acpi: Remove __ro_after_init from acpi_mp_wake_mailbox
  x86/mm: Fix PTI for i386 some more

11 months agoMerge tag 'timers-urgent-2024-08-11' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 11 Aug 2024 17:15:34 +0000 (10:15 -0700)]
Merge tag 'timers-urgent-2024-08-11' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull time keeping fixes from Thomas Gleixner:

 - Fix a couple of issues in the NTP code where user supplied values are
   neither sanity checked nor clamped to the operating range. This
   results in integer overflows and eventualy NTP getting out of sync.

   According to the history the sanity checks had been removed in favor
   of clamping the values, but the clamping never worked correctly under
   all circumstances. The NTP people asked to not bring the sanity
   checks back as it might break existing applications.

   Make the clamping work correctly and add it where it's missing

 - If adjtimex() sets the clock it has to trigger the hrtimer subsystem
   so it can adjust and if the clock was set into the future expire
   timers if needed. The caller should provide a bitmask to tell
   hrtimers which clocks have been adjusted.

   adjtimex() uses not the proper constant and uses CLOCK_REALTIME
   instead, which is 0. So hrtimers adjusts only the clocks, but does
   not check for expired timers, which might make them expire really
   late. Use the proper bitmask constant instead.

* tag 'timers-urgent-2024-08-11' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  timekeeping: Fix bogus clock_was_set() invocation in do_adjtimex()
  ntp: Safeguard against time_constant overflow
  ntp: Clamp maxerror and esterror to operating range

11 months agoMerge tag 'irq-urgent-2024-08-11' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 11 Aug 2024 17:07:52 +0000 (10:07 -0700)]
Merge tag 'irq-urgent-2024-08-11' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull irq fixes from Thomas Gleixner:
 "Three small fixes for interrupt core and drivers:

   - The interrupt core fails to honor caller supplied affinity hints
     for non-managed interrupts and uses the system default affinity on
     startup instead. Set the missing flag in the descriptor to tell the
     core to use the provided affinity.

   - Fix a shift out of bounds error in the Xilinx driver

   - Handle switching to level trigger correctly in the RISCV APLIC
     driver. It failed to retrigger the interrupt which causes it to
     become stale"

* tag 'irq-urgent-2024-08-11' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  irqchip/riscv-aplic: Retrigger MSI interrupt on source configuration
  irqchip/xilinx: Fix shift out of bounds
  genirq/irqdesc: Honor caller provided affinity in alloc_desc()

11 months agoMerge tag 'usb-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Linus Torvalds [Sun, 11 Aug 2024 16:55:32 +0000 (09:55 -0700)]
Merge tag 'usb-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb

Pull USB fixes from Greg KH:
 "Here are a number of small USB driver fixes for reported issues for
  6.11-rc3. Included in here are:

   - usb serial driver MODULE_DESCRIPTION() updates

   - usb serial driver fixes

   - typec driver fixes

   - usb-ip driver fix

   - gadget driver fixes

   - dt binding update

  All of these have been in linux-next with no reported issues"

* tag 'usb-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
  usb: typec: ucsi: Fix a deadlock in ucsi_send_command_common()
  usb: typec: tcpm: avoid sink goto SNK_UNATTACHED state if not received source capability message
  usb: gadget: f_fs: pull out f->disable() from ffs_func_set_alt()
  usb: gadget: f_fs: restore ffs_func_disable() functionality
  USB: serial: debug: do not echo input by default
  usb: typec: tipd: Delete extra semi-colon
  usb: typec: tipd: Fix dereferencing freeing memory in tps6598x_apply_patch()
  usb: gadget: u_serial: Set start_delayed during suspend
  usb: typec: tcpci: Fix error code in tcpci_check_std_output_cap()
  usb: typec: fsa4480: Check if the chip is really there
  usb: gadget: core: Check for unset descriptor
  usb: vhci-hcd: Do not drop references before new references are gained
  usb: gadget: u_audio: Check return codes from usb_ep_enable and config_ep_by_speed.
  usb: gadget: midi2: Fix the response for FB info with block 0xff
  dt-bindings: usb: microchip,usb2514: Add USB2517 compatible
  USB: serial: garmin_gps: use struct_size() to allocate pkt
  USB: serial: garmin_gps: annotate struct garmin_packet with __counted_by
  USB: serial: add missing MODULE_DESCRIPTION() macros
  USB: serial: spcp8x5: remove unused struct 'spcp8x5_usb_ctrl_arg'

11 months agoMerge tag 'tty-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty
Linus Torvalds [Sun, 11 Aug 2024 16:51:29 +0000 (09:51 -0700)]
Merge tag 'tty-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty

Pull tty / serial driver fixes from Greg KH:
 "Here are some small tty and serial driver fixes for reported problems
  for 6.11-rc3. Included in here are:

   - sc16is7xx serial driver fixes

   - uartclk bugfix for a divide by zero issue

   - conmakehash userspace build issue fix

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

* tag 'tty-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty:
  tty: vt: conmakehash: cope with abs_srctree no longer in env
  serial: sc16is7xx: fix invalid FIFO access with special register set
  serial: sc16is7xx: fix TX fifo corruption
  serial: core: check uartclk for zero to avoid divide by zero

11 months agoMerge tag 'driver-core-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 11 Aug 2024 16:38:38 +0000 (09:38 -0700)]
Merge tag 'driver-core-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core

Pull driver core / documentation fixes from Greg KH:
 "Here are some small fixes, and some documentation updates for
  6.11-rc3. Included in here are:

   - embargoed hardware documenation updates based on a lot of review by
     legal-types in lots of companies to try to make the process a _bit_
     easier for us to manage over time.

   - rust firmware documentation fix

   - driver detach race fix for the fix that went into 6.11-rc1

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

* tag 'driver-core-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core:
  driver core: Fix uevent_show() vs driver detach race
  Documentation: embargoed-hardware-issues.rst: add a section documenting the "early access" process
  Documentation: embargoed-hardware-issues.rst: minor cleanups and fixes
  rust: firmware: fix invalid rustdoc link

11 months agoMerge tag 'char-misc-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregk...
Linus Torvalds [Sun, 11 Aug 2024 16:32:54 +0000 (09:32 -0700)]
Merge tag 'char-misc-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc

Pull char/misc fixes from Greg KH:
 "Here are some small char/misc/other driver fixes for 6.11-rc3 for
  reported issues. Included in here are:

   - binder driver fixes

   - fsi MODULE_DESCRIPTION() additions (people seem to love them...)

   - eeprom driver fix

   - Kconfig dependency fix to resolve build issues

   - spmi driver fixes

  All of these have been in linux-next for a while with no reported
  problems"

* tag 'char-misc-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  spmi: pmic-arb: add missing newline in dev_err format strings
  spmi: pmic-arb: Pass the correct of_node to irq_domain_add_tree
  binder_alloc: Fix sleeping function called from invalid context
  binder: fix descriptor lookup for context manager
  char: add missing NetWinder MODULE_DESCRIPTION() macros
  misc: mrvl-cn10k-dpi: add PCI_IOV dependency
  eeprom: ee1004: Fix locking issues in ee1004_probe()
  fsi: add missing MODULE_DESCRIPTION() macros

11 months agoMerge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Linus Torvalds [Sun, 11 Aug 2024 16:28:04 +0000 (09:28 -0700)]
Merge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi

Pull SCSI fixes from James Bottomley:
 "Two core fixes: one to prevent discard type changes (seen on iSCSI)
  during intermittent errors and the other is fixing a lockdep problem
  caused by the queue limits change.

  And one driver fix in ufs"

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  scsi: sd: Keep the discard mode stable
  scsi: sd: Move sd_read_cpr() out of the q->limits_lock region
  scsi: ufs: core: Fix hba->last_dme_cmd_tstamp timestamp updating logic

11 months agoMerge tag 'nfsd-6.11-1' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux
Linus Torvalds [Sat, 10 Aug 2024 17:44:21 +0000 (10:44 -0700)]
Merge tag 'nfsd-6.11-1' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux

Pull nfsd fixes from Chuck Lever:

 - Two minor fixes for recent changes

* tag 'nfsd-6.11-1' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux:
  nfsd: don't set SVC_SOCK_ANONYMOUS when creating nfsd sockets
  sunrpc: avoid -Wformat-security warning

11 months agoMerge tag 'i2c-for-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa...
Linus Torvalds [Sat, 10 Aug 2024 17:28:52 +0000 (10:28 -0700)]
Merge tag 'i2c-for-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux

Pull i2c fixes from Wolfram Sang:

 - Two fixes for SMBusAlert handling in the I2C core: one to avoid an
   endless loop when scanning for handlers and one to make sure handlers
   are always called even if HW has broken behaviour

 - I2C header build fix for when ACPI is enabled but I2C isn't

 - The testunit gets a rename in the code to match the documentation

 - Two fixes for the Qualcomm GENI I2C controller are cleaning up the
   error exit patch in the runtime_resume() function. The first is
   disabling the clock, the second disables the icc on the way out

* tag 'i2c-for-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  i2c: testunit: match HostNotify test name with docs
  i2c: qcom-geni: Add missing geni_icc_disable in geni_i2c_runtime_resume
  i2c: qcom-geni: Add missing clk_disable_unprepare in geni_i2c_runtime_resume
  i2c: Fix conditional for substituting empty ACPI functions
  i2c: smbus: Send alert notifications to all devices if source not found
  i2c: smbus: Improve handling of stuck alerts

11 months agoMerge tag 'dma-mapping-6.11-2024-08-10' of git://git.infradead.org/users/hch/dma...
Linus Torvalds [Sat, 10 Aug 2024 17:19:05 +0000 (10:19 -0700)]
Merge tag 'dma-mapping-6.11-2024-08-10' of git://git.infradead.org/users/hch/dma-mapping

Pull dma-mapping fix from Christoph Hellwig:

 - avoid a deadlock with dma-debug and netconsole (Rik van Riel)

* tag 'dma-mapping-6.11-2024-08-10' of git://git.infradead.org/users/hch/dma-mapping:
  dma-debug: avoid deadlock between dma debug vs printk and netconsole

11 months agoMerge tag 'bcachefs-2024-08-10' of git://evilpiepirate.org/bcachefs
Linus Torvalds [Sat, 10 Aug 2024 17:06:26 +0000 (10:06 -0700)]
Merge tag 'bcachefs-2024-08-10' of git://evilpiepirate.org/bcachefs

Pull more bcachefs fixes from Kent Overstreet:
 "A couple last minute fixes for the new disk accounting

   - fix a bug that was causing ACLs to seemingly "disappear"

   - new on disk format version, bcachefs_metadata_version_disk_accounting_v3

     bcachefs_metadata_version_disk_accounting_v2 accidentally included
     padding in disk_accounting_key; fortunately, 6.11 isn't out yet so
     we can fix this with another version bump"

* tag 'bcachefs-2024-08-10' of git://evilpiepirate.org/bcachefs:
  bcachefs: bcachefs_metadata_version_disk_accounting_v3
  bcachefs: improve bch2_dev_usage_to_text()
  bcachefs: bch2_accounting_invalid()
  bcachefs: Switch to .get_inode_acl()

11 months agoirqchip/riscv-aplic: Retrigger MSI interrupt on source configuration
Yong-Xuan Wang [Fri, 9 Aug 2024 07:10:47 +0000 (15:10 +0800)]
irqchip/riscv-aplic: Retrigger MSI interrupt on source configuration

The section 4.5.2 of the RISC-V AIA specification says that "any write
to a sourcecfg register of an APLIC might (or might not) cause the
corresponding interrupt-pending bit to be set to one if the rectified
input value is high (= 1) under the new source mode."

When the interrupt type is changed in the sourcecfg register, the APLIC
device might not set the corresponding pending bit, so the interrupt might
never become pending.

To handle sourcecfg register changes for level-triggered interrupts in MSI
mode, manually set the pending bit for retriggering interrupt so it gets
retriggered if it was already asserted.

Fixes: ca8df97fe679 ("irqchip/riscv-aplic: Add support for MSI-mode")
Signed-off-by: Yong-Xuan Wang <yongxuan.wang@sifive.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Vincent Chen <vincent.chen@sifive.com>
Reviewed-by: Anup Patel <anup@brainfault.org>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/all/20240809071049.2454-1-yongxuan.wang@sifive.com
11 months agoirqchip/xilinx: Fix shift out of bounds
Radhey Shyam Pandey [Fri, 9 Aug 2024 07:02:24 +0000 (12:32 +0530)]
irqchip/xilinx: Fix shift out of bounds

The device tree property 'xlnx,kind-of-intr' is sanity checked that the
bitmask contains only set bits which are in the range of the number of
interrupts supported by the controller.

The check is done by shifting the mask right by the number of supported
interrupts and checking the result for zero.

The data type of the mask is u32 and the number of supported interrupts is
up to 32. In case of 32 interrupts the shift is out of bounds, resulting in
a mismatch warning. The out of bounds condition is also reported by UBSAN:

  UBSAN: shift-out-of-bounds in irq-xilinx-intc.c:332:22
  shift exponent 32 is too large for 32-bit type 'unsigned int'

Fix it by promoting the mask to u64 for the test.

Fixes: d50466c90724 ("microblaze: intc: Refactor DT sanity check")
Signed-off-by: Radhey Shyam Pandey <radhey.shyam.pandey@amd.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/all/1723186944-3571957-1-git-send-email-radhey.shyam.pandey@amd.com
11 months agoMerge tag '6.11-rc2-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Sat, 10 Aug 2024 04:33:25 +0000 (21:33 -0700)]
Merge tag '6.11-rc2-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6

Pull smb client fixes from Steve French:

 - DFS fix

 - fix for security flags for requiring encryption

 - minor cleanup

* tag '6.11-rc2-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6:
  cifs: cifs_inval_name_dfs_link_error: correct the check for fullpath
  Fix spelling errors in Server Message Block
  smb3: fix setting SecurityFlags when encryption is required

11 months agoMerge tag 'spi-fix-v6.11-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/brooni...
Linus Torvalds [Sat, 10 Aug 2024 04:26:50 +0000 (21:26 -0700)]
Merge tag 'spi-fix-v6.11-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi

Pull spi fixes from Mark Brown:
 "A few SPI fixes: clock rate calculation fixes for the Kunpeng and lpsi
  drivers and a missing registration of a device ID for spidev (which
  had only been updated for DT cases, causing warnings)"

* tag 'spi-fix-v6.11-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi:
  spi: spi-fsl-lpspi: Fix scldiv calculation
  spi: spidev: Add missing spi_device_id for bh2228fv
  spi: hisi-kunpeng: Add verification for the max_frequency provided by the firmware
  spi: hisi-kunpeng: Add validation for the minimum value of speed_hz

11 months agobcachefs: bcachefs_metadata_version_disk_accounting_v3
Kent Overstreet [Fri, 9 Aug 2024 04:25:25 +0000 (00:25 -0400)]
bcachefs: bcachefs_metadata_version_disk_accounting_v3

bcachefs_metadata_version_disk_accounting_v2 erroneously had padding
bytes in disk_accounting_key, which is a problem because we have to
guarantee that all unused bytes in disk_accounting_key are zeroed.

Fortunately 6.11 isn't out yet, so it's cheap to fix this by spinning a
new version.

Reported-by: Gabriel de Perthuis <g2p.code@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
11 months agoMerge tag 'drm-fixes-2024-08-10' of https://gitlab.freedesktop.org/drm/kernel
Linus Torvalds [Fri, 9 Aug 2024 21:00:22 +0000 (14:00 -0700)]
Merge tag 'drm-fixes-2024-08-10' of https://gitlab.freedesktop.org/drm/kernel

Pull drm fixes from Dave Airlie:
 "Weekly regular fixes, mostly amdgpu with i915/xe having a few each,
  and then some misc bits across the board, seems about right for rc3
  time.

  client:
   - fix null ptr deref

  bridge:
   - connector: fix double free

  atomic:
   - fix async flip update

  panel:
   - document panel

  omap:
   - add config dependency

  tests:
   - fix gem shmem test

  drm buddy:
   - Add start address to trim function

  amdgpu:
   - DMCUB fix
   - Fix DET programming on some DCNs
   - DCC fixes
   - DCN 4.0.1 fixes
   - SMU 14.0.x update
   - MMHUB fix
   - DCN 3.1.4 fix
   - GC 12.0 fixes
   - Fix soft recovery error propogation
   - SDMA 7.0 fixes
   - DSC fix

  xe:
   - Fix off-by-one when processing RTP rules
   - Use dma_fence_chain_free in chain fence unused as a sync
   - Fix PL1 disable flow in xe_hwmon_power_max_write
   - Take ref to VM in delayed dump snapshot

  i915:
   - correct dual pps handling for MTL_PCH+ [display]
   - Adjust vma offset for framebuffer mmap offset [gem]
   - Fix Virtual Memory mapping boundaries calculation [gem]
   - Allow evicting to use the requested placement
   - Attempt to get pages without eviction first"

* tag 'drm-fixes-2024-08-10' of https://gitlab.freedesktop.org/drm/kernel: (31 commits)
  drm/xe: Take ref to VM in delayed snapshot
  drm/xe/hwmon: Fix PL1 disable flow in xe_hwmon_power_max_write
  drm/xe: Use dma_fence_chain_free in chain fence unused as a sync
  drm/xe/rtp: Fix off-by-one when processing rules
  drm/amdgpu: Add DCC GFX12 flag to enable address alignment
  drm/amdgpu: correct sdma7 max dw
  drm/amdgpu: Add address alignment support to DCC buffers
  drm/amd/display: Skip Recompute DSC Params if no Stream on Link
  drm/amdgpu: change non-dcc buffer copy configuration
  drm/amdgpu: Forward soft recovery errors to userspace
  drm/amdgpu: add golden setting for gc v12
  drm/buddy: Add start address support to trim function
  drm/amd/display: Add missing program DET segment call to pipe init
  drm/amd/display: Add missing DCN314 to the DML Makefile
  drm/amdgpu: force to use legacy inv in mmhub
  drm/amd/pm: update powerplay structure on smu v14.0.2/3
  drm/amd/display: Add missing mcache registers
  drm/amd/display: Add dcc propagation value
  drm/amd/display: Add missing DET segments programming
  drm/amd/display: Replace dm_execute_dmub_cmd with dc_wake_and_execute_dmub_cmd
  ...

11 months agobcachefs: improve bch2_dev_usage_to_text()
Kent Overstreet [Fri, 9 Aug 2024 03:44:00 +0000 (23:44 -0400)]
bcachefs: improve bch2_dev_usage_to_text()

Add a line for capacity

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
11 months agobcachefs: bch2_accounting_invalid()
Kent Overstreet [Fri, 9 Aug 2024 03:19:59 +0000 (23:19 -0400)]
bcachefs: bch2_accounting_invalid()

Implement bch2_accounting_invalid(); check for junk at the end, and
replicas accounting entries in particular need to be checked or we'll
pop asserts later.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
11 months agoMerge tag 'bitmap-6.11-rc' of https://github.com/norov/linux
Linus Torvalds [Fri, 9 Aug 2024 18:18:09 +0000 (11:18 -0700)]
Merge tag 'bitmap-6.11-rc' of https://github.com/norov/linux

Pull cpumask fix from Yury Norov:
 "Fix for cpumask merge"

[ Mea culpa, this was my mismerge due to too much cut-and-paste - Linus ]

* tag 'bitmap-6.11-rc' of https://github.com/norov/linux:
  cpumask: Fix crash on updating CPU enabled mask

11 months agoMerge tag 'pm-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Linus Torvalds [Fri, 9 Aug 2024 17:44:35 +0000 (10:44 -0700)]
Merge tag 'pm-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull power management fix from Rafael Wysocki:
 "Change the default EPP (energy-performence preference) value for the
  Emerald Rapids processor in the intel_pstate driver.

  Thisshould improve both the performance and energy efficiency (Pedro
  Henrique Kopper)"

* tag 'pm-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  cpufreq: intel_pstate: Update Balance performance EPP for Emerald Rapids

11 months agoMerge tag 'asm-generic-fixes-6.11-2' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 9 Aug 2024 17:23:18 +0000 (10:23 -0700)]
Merge tag 'asm-generic-fixes-6.11-2' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/asm-generic

Pull asm-generic fixes from Arnd Bergmann:
 "There are two more changes to the syscall.tbl conversion: the
  '__NR_newfstat' in the previous bugfix was a mistake and gets reverted
  now, after triple-checking that the contents are now back to what they
  were on all architectures. The __NR_nfsservctl definition is not
  really needed but came up in the same discussion as it had previously
  been defined in uapi/asm-generic/unistd.h and tested for in user
  space.

  There are a few more symbols that used to be defined in the old
  unistd.h file, but that are never defined on any other architecture
  using syscall.tbl format. These used to be needed inside of the
  kernel:

     __NR_syscalls
     __NR_arch_specific_syscall
     __NR3264_*

  Searching for these on https://codesearch.debian.net/ shows a few
  packages (rustc, golang, clamav, libseccomp, librsvg, strace) that
  duplicate all the macros from asm/unistd.h, but nothing that actually
  uses the macros, so I concluded that they are fine to omit after all"

* tag 'asm-generic-fixes-6.11-2' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/asm-generic:
  syscalls: add back legacy __NR_nfsservctl macro
  syscalls: fix fstat() entry again

11 months agoMerge tag 'arm-fixes-6.11-1' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc
Linus Torvalds [Fri, 9 Aug 2024 17:06:43 +0000 (10:06 -0700)]
Merge tag 'arm-fixes-6.11-1' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc

Pull ARM SoC fixes from Arnd Bergmann:
 "There are three sets of patches for the soc tree:

   - Marek Behún addresses multiple build time regressions caused by
     changes to the cznic turris-omnia support

   - Dmitry Torokhov fixes a regression in the legacy "gumstix" board
     code he cleaned up earlier

   - The TI K3 maintainers found multiple bugs in the in gpio, audio and
     pcie devicetree nodes"

* tag 'arm-fixes-6.11-1' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc:
  ARM: pxa/gumstix: fix attaching properties to vbus gpio device
  doc: platform: cznic: turris-omnia-mcu: Use double backticks for attribute value
  doc: platform: cznic: turris-omnia-mcu: Fix sphinx-build warning
  platform: cznic: turris-omnia-mcu: Make GPIO code optional
  platform: cznic: turris-omnia-mcu: Make poweroff and wakeup code optional
  platform: cznic: turris-omnia-mcu: Make TRNG code optional
  platform: cznic: turris-omnia-mcu: Make watchdog code optional
  arm64: dts: ti: k3-j784s4-main: Correct McASP DMAs
  arm64: dts: ti: k3-j722s: Fix gpio-range for main_pmx0
  arm64: dts: ti: k3-am62p: Fix gpio-range for main_pmx0
  arm64: dts: ti: k3-am62p: Add gpio-ranges for mcu_gpio0
  arm64: dts: ti: k3-am62-verdin-dahlia: Keep CTRL_SLEEP_MOCI# regulator on
  arm64: dts: ti: k3-j784s4-evm: Consolidate serdes0 references
  arm64: dts: ti: k3-j784s4-evm: Assign only lanes 0 and 1 to PCIe1

11 months agoMerge tag 'probes-fixes-v6.11-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 9 Aug 2024 16:43:46 +0000 (09:43 -0700)]
Merge tag 'probes-fixes-v6.11-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace

Pull kprobe fixes from Masami Hiramatsu:

 - Fix misusing str_has_prefix() parameter order to check symbol prefix
   correctly

 - bpf: remove unused declaring of bpf_kprobe_override

* tag 'probes-fixes-v6.11-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  kprobes: Fix to check symbol prefixes correctly
  bpf: kprobe: remove unused declaring of bpf_kprobe_override

11 months agoMerge tag 'block-6.11-20240809' of git://git.kernel.dk/linux
Linus Torvalds [Fri, 9 Aug 2024 16:35:58 +0000 (09:35 -0700)]
Merge tag 'block-6.11-20240809' of git://git.kernel.dk/linux

Pull block fixes from Jens Axboe:
 "Just a set of cleanups for blk-throttle and nvme structures"

* tag 'block-6.11-20240809' of git://git.kernel.dk/linux:
  nvme: reorganize nvme_ns_head fields
  nvme: change data type of lba_shift
  nvme: remove a field from nvme_ns_head
  nvme: remove unused parameter
  blk-throttle: remove more latency dead-code

11 months agoMerge tag 'io_uring-6.11-20240809' of git://git.kernel.dk/linux
Linus Torvalds [Fri, 9 Aug 2024 16:32:10 +0000 (09:32 -0700)]
Merge tag 'io_uring-6.11-20240809' of git://git.kernel.dk/linux

Pull io_uring fixes from Jens Axboe:
 "Nothing major in here, just two fixes for ensuring that bundle
  recv/send requests always get marked for cleanups, and a single fix to
  ensure that sends with provided buffers only pick a single buffer
  unless the bundle option has been enabled"

* tag 'io_uring-6.11-20240809' of git://git.kernel.dk/linux:
  io_uring/net: don't pick multiple buffers for non-bundle send
  io_uring/net: ensure expanded bundle send gets marked for cleanup
  io_uring/net: ensure expanded bundle recv gets marked for cleanup

11 months agoMerge tag 'sound-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai...
Linus Torvalds [Fri, 9 Aug 2024 16:25:30 +0000 (09:25 -0700)]
Merge tag 'sound-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound

Pull sound fixes from Takashi Iwai:
 "A collection of lots of small changes, almost all device-specific:

   - A series of fixes for ASoC Qualcomm stuff

   - Various fixes for Cirrus ASoC and HD-audio codecs

   - A few AMD ASoC quirks and usual HD-audio quirks

   - Other misc fixes, including a long-time regression in USB-audio"

* tag 'sound-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound: (39 commits)
  ASoC: cs35l56: Patch CS35L56_IRQ1_MASK_18 to the default value
  ASoC: meson: axg-fifo: fix irq scheduling issue with PREEMPT_RT
  MAINTAINERS: Update Cirrus Logic parts to linux-sound mailing list
  ASoC: dt-bindings: qcom,wcd939x: Correct reset GPIO polarity in example
  ASoC: dt-bindings: qcom,wcd938x: Correct reset GPIO polarity in example
  ASoC: dt-bindings: qcom,wcd934x: Correct reset GPIO polarity in example
  ASoC: dt-bindings: qcom,wcd937x: Correct reset GPIO polarity in example
  ASoC: amd: yc: Add quirk entry for OMEN by HP Gaming Laptop 16-n0xxx
  ASoC: codecs: ES8326: button detect issue
  ASoC: amd: yc: Support mic on Lenovo Thinkpad E14 Gen 6
  ALSA: usb-audio: Re-add ScratchAmp quirk entries
  ALSA: hda/realtek: Add Framework Laptop 13 (Intel Core Ultra) to quirks
  ALSA: hda/hdmi: Yet more pin fix for HP EliteDesk 800 G4
  ALSA: hda: Add HP MP9 G4 Retail System AMS to force connect list
  ASoC: cs35l56: Handle OTP read latency over SoundWire
  ASoC: codecs: lpass-macro: fix missing codec version
  ALSA: line6: Fix racy access to midibuf
  ASoC: cs-amp-lib: Fix NULL pointer crash if efi.get_variable is NULL
  ASoC: cs35l56: Stop creating ALSA controls for firmware coefficients
  ASoC: wm_adsp: Add control_add callback and export wm_adsp_control_add()
  ...

11 months agomodule: make waiting for a concurrent module loader interruptible
Linus Torvalds [Fri, 9 Aug 2024 15:33:28 +0000 (08:33 -0700)]
module: make waiting for a concurrent module loader interruptible

The recursive aes-arm-bs module load situation reported by Russell King
is getting fixed in the crypto layer, but this in the meantime fixes the
"recursive load hangs forever" by just making the waiting for the first
module load be interruptible.

This should now match the old behavior before commit 9b9879fc0327
("modules: catch concurrent module loads, treat them as idempotent"),
which used the different "wait for module to be ready" code in
module_patient_check_exists().

End result: a recursive module load will still block, but now a signal
will interrupt it and fail the second module load, at which point the
first module will successfully complete loading.

Fixes: 9b9879fc0327 ("modules: catch concurrent module loads, treat them as idempotent")
Cc: Russell King <linux@armlinux.org.uk>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 months agoMerge tag 'i2c-host-fixes-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel...
Wolfram Sang [Fri, 9 Aug 2024 13:28:08 +0000 (15:28 +0200)]
Merge tag 'i2c-host-fixes-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/andi.shyti/linux into i2c/for-current

Two fixes on the Qualcomm GENI I2C controller are cleaning up the
error exit patch in the runtime_resume() function. The first is
disabling the clock, the second disables the icc on the way out.

11 months agoMerge tag 'asoc-fix-v6.11-rc2' of https://git.kernel.org/pub/scm/linux/kernel/git...
Takashi Iwai [Fri, 9 Aug 2024 07:58:07 +0000 (09:58 +0200)]
Merge tag 'asoc-fix-v6.11-rc2' of https://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound into for-linus

ASoC: Fixes for v6.11

Quite a lot of fixes have come in since the merge window, there's some
repetitive fixes over the Qualcomm drivers increasing the patch count,
along with a large batch of fixes from Cirrus.  We also have some quirks
and some individual fixes.

11 months agoMerge tag 'drm-intel-fixes-2024-08-08' of https://gitlab.freedesktop.org/drm/i915...
Dave Airlie [Fri, 9 Aug 2024 07:16:28 +0000 (17:16 +1000)]
Merge tag 'drm-intel-fixes-2024-08-08' of https://gitlab.freedesktop.org/drm/i915/kernel into drm-fixes

- correct dual pps handling for MTL_PCH+ [display] (Dnyaneshwar Bhadane)
- Adjust vma offset for framebuffer mmap offset [gem] (Andi Shyti)
- Fix Virtual Memory mapping boundaries calculation [gem] (Andi Shyti)
- Allow evicting to use the requested placement (David Gow)
- Attempt to get pages without eviction first (David Gow)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Tvrtko Ursulin <tursulin@igalia.com>
Link: https://patchwork.freedesktop.org/patch/msgid/ZrSFpj20b1LbBhCJ@linux
11 months agoMerge tag 'drm-xe-fixes-2024-08-08' of https://gitlab.freedesktop.org/drm/xe/kernel...
Dave Airlie [Fri, 9 Aug 2024 07:08:55 +0000 (17:08 +1000)]
Merge tag 'drm-xe-fixes-2024-08-08' of https://gitlab.freedesktop.org/drm/xe/kernel into drm-fixes

- Fix off-by-one when processing RTP rules (Lucas)
- Use dma_fence_chain_free in chain fence unused as a sync (Brost)
- Fix PL1 disable flow in xe_hwmon_power_max_write (Karthik)
- Take ref to VM in delayed dump snapshot (Brost)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/ZrUGgVrgTJ_vF2PS@intel.com
11 months agoMerge tag 'amd-drm-fixes-6.11-2024-08-08' of https://gitlab.freedesktop.org/agd5f...
Dave Airlie [Fri, 9 Aug 2024 03:45:52 +0000 (13:45 +1000)]
Merge tag 'amd-drm-fixes-6.11-2024-08-08' of https://gitlab.freedesktop.org/agd5f/linux into drm-fixes

amd-drm-fixes-6.11-2024-08-08:

amdgpu:
- DMCUB fix
- Fix DET programming on some DCNs
- DCC fixes
- DCN 4.0.1 fixes
- SMU 14.0.x update
- MMHUB fix
- DCN 3.1.4 fix
- GC 12.0 fixes
- Fix soft recovery error propogation
- SDMA 7.0 fixes
- DSC fix

drm buddy:
- Add start address to trim function

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240808161134.1227671-1-alexander.deucher@amd.com
11 months agoMerge tag 'drm-misc-fixes-2024-08-08' of https://gitlab.freedesktop.org/drm/misc...
Dave Airlie [Fri, 9 Aug 2024 03:00:59 +0000 (13:00 +1000)]
Merge tag 'drm-misc-fixes-2024-08-08' of https://gitlab.freedesktop.org/drm/misc/kernel into drm-fixes

A fix for drm/client to prevent a null pointer dereference, a fix for a
double-free in drm/bridge-connector, a fix for a gem shmem test, and a
fix for async flips updates.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Maxime Ripard <mripard@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240808-debonair-smoky-mustang-8b6b3f@houat
11 months agocifs: cifs_inval_name_dfs_link_error: correct the check for fullpath
Gleb Korobeynikov [Thu, 8 Aug 2024 15:47:48 +0000 (18:47 +0300)]
cifs: cifs_inval_name_dfs_link_error: correct the check for fullpath

Replace the always-true check tcon->origin_fullpath with
check of server->leaf_fullpath

See https://bugzilla.kernel.org/show_bug.cgi?id=219083

The check of the new @tcon will always be true during mounting,
since @tcon->origin_fullpath will only be set after the tree is
connected to the latest common resource, as well as checking if
the prefix paths from it are fully accessible.

Fixes: 3ae872de4107 ("smb: client: fix shared DFS root mounts with different prefixes")
Reviewed-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Gleb Korobeynikov <gkorobeynikov@astralinux.ru>
Signed-off-by: Steve French <stfrench@microsoft.com>
11 months agoMerge tag 'net-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Linus Torvalds [Thu, 8 Aug 2024 20:51:44 +0000 (13:51 -0700)]
Merge tag 'net-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net

Pull networking fixes from Jakub Kicinski:
 "Including fixes from bluetooth.

  Current release - regressions:

   - eth: bnxt_en: fix memory out-of-bounds in bnxt_fill_hw_rss_tbl() on
     older chips

  Current release - new code bugs:

   - ethtool: fix off-by-one error / kdoc contradicting the code for max
     RSS context IDs

   - Bluetooth: hci_qca:
      - QCA6390: fix support on non-DT platforms
      - QCA6390: don't call pwrseq_power_off() twice
      - fix a NULL-pointer derefence at shutdown

   - eth: ice: fix incorrect assigns of FEC counters

  Previous releases - regressions:

   - mptcp: fix handling endpoints with both 'signal' and 'subflow'
     flags set

   - virtio-net: fix changing ring count when vq IRQ coalescing not
     supported

   - eth: gve: fix use of netif_carrier_ok() during reconfig / reset

  Previous releases - always broken:

   - eth: idpf: fix bugs in queue re-allocation on reconfig / reset

   - ethtool: fix context creation with no parameters

  Misc:

   - linkwatch: use system_unbound_wq to ease RTNL contention"

* tag 'net-6.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (41 commits)
  net: dsa: microchip: disable EEE for KSZ8567/KSZ9567/KSZ9896/KSZ9897.
  ethtool: Fix context creation with no parameters
  net: ethtool: fix off-by-one error in max RSS context IDs
  net: pse-pd: tps23881: include missing bitfield.h header
  net: fec: Stop PPS on driver remove
  net: bcmgenet: Properly overlay PHY and MAC Wake-on-LAN capabilities
  l2tp: fix lockdep splat
  net: stmmac: dwmac4: fix PCS duplex mode decode
  idpf: fix UAFs when destroying the queues
  idpf: fix memleak in vport interrupt configuration
  idpf: fix memory leaks and crashes while performing a soft reset
  bnxt_en : Fix memory out-of-bounds in bnxt_fill_hw_rss_tbl()
  net: dsa: bcm_sf2: Fix a possible memory leak in bcm_sf2_mdio_register()
  net/smc: add the max value of fallback reason count
  Bluetooth: hci_sync: avoid dup filtering when passive scanning with adv monitor
  Bluetooth: l2cap: always unlock channel in l2cap_conless_channel()
  Bluetooth: hci_qca: fix a NULL-pointer derefence at shutdown
  Bluetooth: hci_qca: fix QCA6390 support on non-DT platforms
  Bluetooth: hci_qca: don't call pwrseq_power_off() twice for QCA6390
  ice: Fix incorrect assigns of FEC counts
  ...

11 months agoMerge tag 'trace-v6.11-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace...
Linus Torvalds [Thu, 8 Aug 2024 20:32:59 +0000 (13:32 -0700)]
Merge tag 'trace-v6.11-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace

Pull tracing fixes from Steven Rostedt:

 - Have reading of event format files test if the metadata still exists.

   When a event is freed, a flag (EVENT_FILE_FL_FREED) in the metadata
   is set to state that it is to prevent any new references to it from
   happening while waiting for existing references to close. When the
   last reference closes, the metadata is freed. But the "format" was
   missing a check to this flag (along with some other files) that
   allowed new references to happen, and a use-after-free bug to occur.

 - Have the trace event meta data use the refcount infrastructure
   instead of relying on its own atomic counters.

 - Have tracefs inodes use alloc_inode_sb() for allocation instead of
   using kmem_cache_alloc() directly.

 - Have eventfs_create_dir() return an ERR_PTR instead of NULL as the
   callers expect a real object or an ERR_PTR.

 - Have release_ei() use call_srcu() and not call_rcu() as all the
   protection is on SRCU and not RCU.

 - Fix ftrace_graph_ret_addr() to use the task passed in and not
   current.

 - Fix overflow bug in get_free_elt() where the counter can overflow the
   integer and cause an infinite loop.

 - Remove unused function ring_buffer_nr_pages()

 - Have tracefs freeing use the inode RCU infrastructure instead of
   creating its own.

   When the kernel had randomize structure fields enabled, the rcu field
   of the tracefs_inode was overlapping the rcu field of the inode
   structure, and corrupting it. Instead, use the destroy_inode()
   callback to do the initial cleanup of the code, and then have
   free_inode() free it.

* tag 'trace-v6.11-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  tracefs: Use generic inode RCU for synchronizing freeing
  ring-buffer: Remove unused function ring_buffer_nr_pages()
  tracing: Fix overflow in get_free_elt()
  function_graph: Fix the ret_stack used by ftrace_graph_ret_addr()
  eventfs: Use SRCU for freeing eventfs_inodes
  eventfs: Don't return NULL in eventfs_create_dir()
  tracefs: Fix inode allocation
  tracing: Use refcount for trace_event_file reference counter
  tracing: Have format file honor EVENT_FILE_FL_FREED

11 months agoMerge tag 'bcachefs-2024-08-08' of git://evilpiepirate.org/bcachefs
Linus Torvalds [Thu, 8 Aug 2024 20:27:31 +0000 (13:27 -0700)]
Merge tag 'bcachefs-2024-08-08' of git://evilpiepirate.org/bcachefs

Pull bcachefs fixes from Kent Overstreet:
 "Assorted little stuff:

   - lockdep fixup for lockdep_set_notrack_class()

   - we can now remove a device when using erasure coding without
     deadlocking, though we still hit other issues

   - the 'allocator stuck' timeout is now configurable, and messages are
     ratelimited. The default timeout has been increased from 10 seconds
     to 30"

* tag 'bcachefs-2024-08-08' of git://evilpiepirate.org/bcachefs:
  bcachefs: Use bch2_wait_on_allocator() in btree node alloc path
  bcachefs: Make allocator stuck timeout configurable, ratelimit messages
  bcachefs: Add missing path_traverse() to btree_iter_next_node()
  bcachefs: ec should not allocate from ro devs
  bcachefs: Improved allocator debugging for ec
  bcachefs: Add missing bch2_trans_begin() call
  bcachefs: Add a comment for bucket helper types
  bcachefs: Don't rely on implicit unsigned -> signed integer conversion
  lockdep: Fix lockdep_set_notrack_class() for CONFIG_LOCK_STAT
  bcachefs: Fix double free of ca->buckets_nouse

11 months agoASoC: cs35l56: Patch CS35L56_IRQ1_MASK_18 to the default value
Simon Trimmer [Wed, 7 Aug 2024 14:26:48 +0000 (14:26 +0000)]
ASoC: cs35l56: Patch CS35L56_IRQ1_MASK_18 to the default value

Device tuning files made with early revision tooling may contain
configuration that can unmask IRQ signals that are owned by the host.

Adding a safe default to the regmap patch ensures that the hardware
matches the driver expectations.

Signed-off-by: Simon Trimmer <simont@opensource.cirrus.com>
Link: https://patch.msgid.link/20240807142648.46932-1-simont@opensource.cirrus.com
Signed-off-by: Mark Brown <broonie@kernel.org>
11 months agoASoC: meson: axg-fifo: fix irq scheduling issue with PREEMPT_RT
Jerome Brunet [Wed, 7 Aug 2024 16:27:03 +0000 (18:27 +0200)]
ASoC: meson: axg-fifo: fix irq scheduling issue with PREEMPT_RT

With PREEMPT_RT enabled a spinlock_t becomes a sleeping lock.

This is usually not a problem with spinlocks used in IRQ context since
IRQ handlers get threaded. However, if IRQF_ONESHOT is set, the primary
handler won't be force-threaded and runs always in hardirq context. This is
a problem because spinlock_t requires a preemptible context on PREEMPT_RT.

In this particular instance, regmap mmio uses spinlock_t to protect the
register access and IRQF_ONESHOT is set on the IRQ. In this case, it is
actually better to do everything in threaded handler and it solves the
problem with PREEMPT_RT.

Reported-by: Arseniy Krasnov <avkrasnov@salutedevices.com>
Closes: https://lore.kernel.org/linux-amlogic/20240729131652.3012327-1-avkrasnov@salutedevices.com
Suggested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Fixes: b11d26660dff ("ASoC: meson: axg-fifo: use threaded irq to check periods")
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Link: https://patch.msgid.link/20240807162705.4024136-1-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
11 months agoMAINTAINERS: Update Cirrus Logic parts to linux-sound mailing list
Charles Keepax [Wed, 7 Aug 2024 14:01:40 +0000 (15:01 +0100)]
MAINTAINERS: Update Cirrus Logic parts to linux-sound mailing list

Now that most kernel work on sound has moved over to the linux-sound
mailing list so should the Cirrus Logic audio parts.

Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Link: https://patch.msgid.link/20240807140140.421359-1-ckeepax@opensource.cirrus.com
Signed-off-by: Mark Brown <broonie@kernel.org>
11 months agoASoC: dt-bindings: qcom,wcd939x: Correct reset GPIO polarity in example
Krzysztof Kozlowski [Tue, 6 Aug 2024 11:49:31 +0000 (13:49 +0200)]
ASoC: dt-bindings: qcom,wcd939x: Correct reset GPIO polarity in example

The reset GPIO of WCD9390/WCD9395 is active low and that's how it is
routed on typical boards, so correct the example DTS to use expected
polarity, instead of IRQ flag (which is a logical mistake on its own).

Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://patch.msgid.link/20240806114931.40090-4-krzysztof.kozlowski@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
11 months agoASoC: dt-bindings: qcom,wcd938x: Correct reset GPIO polarity in example
Krzysztof Kozlowski [Tue, 6 Aug 2024 11:49:30 +0000 (13:49 +0200)]
ASoC: dt-bindings: qcom,wcd938x: Correct reset GPIO polarity in example

The reset GPIO of WCD9380/WCD9385 is active low and that's how it is
routed on typical boards, so correct the example DTS to use expected
polarity.

Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://patch.msgid.link/20240806114931.40090-3-krzysztof.kozlowski@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
11 months agoASoC: dt-bindings: qcom,wcd934x: Correct reset GPIO polarity in example
Krzysztof Kozlowski [Tue, 6 Aug 2024 11:49:29 +0000 (13:49 +0200)]
ASoC: dt-bindings: qcom,wcd934x: Correct reset GPIO polarity in example

The reset GPIO of WCD9340/WCD9341 is active low and that's how it is
routed on typical boards, so correct the example DTS to use expected
polarity.

Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://patch.msgid.link/20240806114931.40090-2-krzysztof.kozlowski@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
11 months agoASoC: dt-bindings: qcom,wcd937x: Correct reset GPIO polarity in example
Krzysztof Kozlowski [Tue, 6 Aug 2024 11:49:28 +0000 (13:49 +0200)]
ASoC: dt-bindings: qcom,wcd937x: Correct reset GPIO polarity in example

The reset GPIO of WCD9370/WCD9375 is active low and that's how it is
routed on typical boards, so correct the example DTS to use expected
polarity.

Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://patch.msgid.link/20240806114931.40090-1-krzysztof.kozlowski@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
11 months agoASoC: amd: yc: Add quirk entry for OMEN by HP Gaming Laptop 16-n0xxx
Takashi Iwai [Wed, 7 Aug 2024 17:02:27 +0000 (19:02 +0200)]
ASoC: amd: yc: Add quirk entry for OMEN by HP Gaming Laptop 16-n0xxx

Fix the missing mic on OMEN by HP Gaming Laptop 16-n0xxx by adding the
quirk entry with the board ID 8A44.

Cc: stable@vger.kernel.org
Link: https://bugzilla.suse.com/show_bug.cgi?id=1227182
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Link: https://patch.msgid.link/20240807170249.16490-1-tiwai@suse.de
Signed-off-by: Mark Brown <broonie@kernel.org>
11 months agoASoC: codecs: ES8326: button detect issue
Zhang Yi [Wed, 7 Aug 2024 02:53:55 +0000 (10:53 +0800)]
ASoC: codecs: ES8326: button detect issue

We find that we need to set snd_jack_types to 0. If not,
there will be a probability of button detection errors

Signed-off-by: Zhang Yi <zhangyi@everest-semi.com>
Link: https://patch.msgid.link/20240807025356.24904-2-zhangyi@everest-semi.com
Signed-off-by: Mark Brown <broonie@kernel.org>
11 months agoASoC: amd: yc: Support mic on Lenovo Thinkpad E14 Gen 6
Krzysztof Stępniak [Wed, 7 Aug 2024 00:12:19 +0000 (02:12 +0200)]
ASoC: amd: yc: Support mic on Lenovo Thinkpad E14 Gen 6

Lenovo Thinkpad E14 Gen 6 (model type 21M3)
needs a quirk entry for internal mic to work.

Signed-off-by: Krzysztof Stępniak <kfs.szk@gmail.com>
Link: https://patch.msgid.link/20240807001219.1147-1-kfs.szk@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
11 months agomodule: warn about excessively long module waits
Linus Torvalds [Thu, 8 Aug 2024 19:29:40 +0000 (12:29 -0700)]
module: warn about excessively long module waits

Russell King reported that the arm cbc(aes) crypto module hangs when
loaded, and Herbert Xu bisected it to commit 9b9879fc0327 ("modules:
catch concurrent module loads, treat them as idempotent"), and noted:

 "So what's happening here is that the first modprobe tries to load a
  fallback CBC implementation, in doing so it triggers a load of the
  exact same module due to module aliases.

  IOW we're loading aes-arm-bs which provides cbc(aes). However, this
  needs a fallback of cbc(aes) to operate, which is made out of the
  generic cbc module + any implementation of aes, or ecb(aes). The
  latter happens to also be provided by aes-arm-cb so that's why it
  tries to load the same module again"

So loading the aes-arm-bs module ends up wanting to recursively load
itself, and the recursive load then ends up waiting for the original
module load to complete.

This is a regression, in that it used to be that we just tried to load
the module multiple times, and then as we went on to install it the
second time we would instead just error out because the module name
already existed.

That is actually also exactly what the original "catch concurrent loads"
patch did in commit 9828ed3f695a ("module: error out early on concurrent
load of the same module file"), but it turns out that it ends up being
racy, in that erroring out before the module has been fully initialized
will cause failures in dependent module loading.

See commit ac2263b588df (which was the revert of that "error out early")
commit for details about why erroring out before the module has been
initialized is actually fundamentally racy.

Now, for the actual recursive module load (as opposed to just
concurrently loading the same module twice), the race is not an issue.

At the same time it's hard for the kernel to see that this is recursion,
because the module load is always done from a usermode helper, so the
recursion is not some simple callchain within the kernel.

End result: this is not the real fix, but this at least adds a warning
for the situation (admittedly much too late for all the debugging pain
that Russell and Herbert went through) and if we can come to a
resolution on how to detect the recursion properly, this re-organizes
the code to make that easier.

Link: https://lore.kernel.org/all/ZrFHLqvFqhzykuYw@shell.armlinux.org.uk/
Reported-by: Russell King <linux@armlinux.org.uk>
Debugged-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 months agobcachefs: Switch to .get_inode_acl()
Kent Overstreet [Wed, 7 Aug 2024 19:42:23 +0000 (15:42 -0400)]
bcachefs: Switch to .get_inode_acl()

.set_acl() requires a dentry, and if one isn't passed it marks the VFS
inode as not having an ACL.

This has been causing inodes with ACLs to have them "disappear" on
bcachefs filesystem, depending on which path those inodes get pulled
into the cache from.

Switching to .get_inode_acl(), like other local filesystems, fixes this.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
11 months agoMerge tag 'nvme-6.11-2024-08-08' of git://git.infradead.org/nvme into block-6.11
Jens Axboe [Thu, 8 Aug 2024 18:27:40 +0000 (12:27 -0600)]
Merge tag 'nvme-6.11-2024-08-08' of git://git.infradead.org/nvme into block-6.11

Pull NVMe fixes from Keith:

"nvme fixes for Linux 6.11

 - Cleanups and improved struct packing (Kanchan)"

* tag 'nvme-6.11-2024-08-08' of git://git.infradead.org/nvme:
  nvme: reorganize nvme_ns_head fields
  nvme: change data type of lba_shift
  nvme: remove a field from nvme_ns_head
  nvme: remove unused parameter

11 months agoMerge tag 'loongarch-fixes-6.11-1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Thu, 8 Aug 2024 18:22:04 +0000 (11:22 -0700)]
Merge tag 'loongarch-fixes-6.11-1' of git://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai/linux-loongson

Pull LoongArch fixes from Huacai Chen:
 "Enable general EFI poweroff method to make poweroff usable on
  hardwares which lack ACPI S5, use accessors to page table entries
  instead of direct dereference to avoid potential problems, and two
  trivial kvm cleanups"

* tag 'loongarch-fixes-6.11-1' of git://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai/linux-loongson:
  LoongArch: KVM: Remove undefined a6 argument comment for kvm_hypercall()
  LoongArch: KVM: Remove unnecessary definition of KVM_PRIVATE_MEM_SLOTS
  LoongArch: Use accessors to page table entries instead of direct dereference
  LoongArch: Enable general EFI poweroff method

11 months agodrm/xe: Take ref to VM in delayed snapshot
Matthew Brost [Thu, 1 Aug 2024 15:41:16 +0000 (08:41 -0700)]
drm/xe: Take ref to VM in delayed snapshot

Kernel BO's don't take a ref to the VM, we need the VM for the
delayed snapshot, so take a ref to the VM in delayed snapshot.

v2:
 - Check for lrc_bo before taking a VM ref (CI)
 - Check lrc_bo->vm before taking / dropping a VM ref (CI)
 - Drop VM in xe_lrc_snapshot_free
v5:
 - Fix commit message wording (Johnathan)

Fixes: 47058633d9c5 ("drm/xe: Move lrc snapshot capturing to xe_lrc.c")
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240801154118.2547543-2-matthew.brost@intel.com
(cherry picked from commit c3bc97d2f102ddd5a8341eeb2dbae2a3e98bb46a)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
11 months agodrm/xe/hwmon: Fix PL1 disable flow in xe_hwmon_power_max_write
Karthik Poosa [Thu, 1 Aug 2024 11:24:24 +0000 (16:54 +0530)]
drm/xe/hwmon: Fix PL1 disable flow in xe_hwmon_power_max_write

In xe_hwmon_power_max_write, for PL1 disable supported case, instead of
returning after PL1 disable, PL1 enable path was also being run.
Fixed it by returning after disable.

v2: Correct typo and grammar in commit message. (Jonathan)

Signed-off-by: Karthik Poosa <karthik.poosa@intel.com>
Fixes: fef6dd12b45a ("drm/xe/hwmon: Protect hwmon rw attributes with hwmon_lock")
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240801112424.1841766-1-karthik.poosa@intel.com
(cherry picked from commit 146458645e505f5eac498759bcd865cf7c0dfd9a)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
11 months agodrm/xe: Use dma_fence_chain_free in chain fence unused as a sync
Matthew Brost [Sat, 27 Jul 2024 01:22:16 +0000 (18:22 -0700)]
drm/xe: Use dma_fence_chain_free in chain fence unused as a sync

A chain fence is uninitialized if not installed in a drm sync obj. Thus
if xe_sync_entry_cleanup is called and sync->chain_fence is non-NULL the
proper cleanup is dma_fence_chain_free rather than a dma-fence put.

Reported-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2411
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2261
Fixes: dd08ebf6c352 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240727012216.2118276-1-matthew.brost@intel.com
(cherry picked from commit 7f7a2da3bf8bc0e0f6c239af495b7050056e889c)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
11 months agodrm/xe/rtp: Fix off-by-one when processing rules
Lucas De Marchi [Fri, 26 Jul 2024 06:43:35 +0000 (23:43 -0700)]
drm/xe/rtp: Fix off-by-one when processing rules

Gustavo noticed an odd "+ 2" in rtp_mark_active() while processing
rtp rules and pointed that it should be "+ 1". In fact, while processing
entries without actions (OOB workarounds), if the WA is activated and
has OR rules, it will also inadvertently activate the very next
workaround.

Test in a LNL B0 platform by moving 18024947630 on top of 16020292621,
makes the latter become active:

$ cat /sys/kernel/debug/dri/0/gt0/workarounds
...
OOB Workarounds
18024947630
16020292621
14018094691
16022287689
13011645652
22019338487_display

In future a kunit test will be added to cover the rtp checks for entries
without actions.

Fixes: fe19328b900c ("drm/xe/rtp: Add support for entries with no action")
Cc: Gustavo Sousa <gustavo.sousa@intel.com>
Reviewed-by: Gustavo Sousa <gustavo.sousa@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240726064337.797576-6-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
(cherry picked from commit fd6797ec50c561f085bc94e3ee26f484a52af79e)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>