rhashtable_walk_start_check() has two restart paths when resuming a walk.
When iter->walker.tbl is valid, it re-validates iter->p against the table
and sets iter->p = NULL if the object is gone. When iter->walker.tbl is
NULL (table was freed during resize), it resets slot and skip but forgets
to clear iter->p.
rhashtable_walk_next() then dereferences the stale iter->p, reading
freed memory. This is a use-after-free.
Any caller that does multi-fragment rhashtable walks across
walk_stop/walk_start boundaries is affected. Concrete cases include
netlink_diag (__netlink_diag_dump in net/netlink/diag.c) and TIPC
(tipc_nl_sk_walk in net/tipc/socket.c).
Crash stack (netlink_diag):
BUG: KASAN: slab-use-after-free in rhashtable_walk_next+0x365/0x3c0
Read of size 8 at addr ffff88801a9d2438 (freed kmalloc-2k, offset 1080)
Call Trace:
rhashtable_walk_next+0x365/0x3c0 (lib/rhashtable.c:1016)
__netlink_diag_dump+0x160/0x760 (net/netlink/diag.c:122)
netlink_diag_dump+0xc2/0x240
netlink_dump+0x5bc/0x1270
netlink_recvmsg+0x7a3/0x980
sock_recvmsg+0x1bc/0x200
__sys_recvfrom+0x1d4/0x2c0
Merge tag 'landlock-7.2-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/mic/linux
Pull landlock fix from Mickaël Salaün:
"This fixes TCP Fast Open support, specific test environments, and doc
warnings"
* tag 'landlock-7.2-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/mic/linux:
selftests/landlock: Skip scoped_signal subtest with MSG_OOB if not available
selftests/landlock: Fix screwed up pointers in the scoped_signal_test
landlock: Update formatting
landlock: Fix kernel-doc for the nested quiet layer flag
selftests/landlock: Add test for TCP fast open
landlock: Fix TCP Fast Open connection bypass
Mikko Perttunen [Wed, 3 Jun 2026 08:37:49 +0000 (17:37 +0900)]
gpu: host1x: Fix use-after-free in host1x_bo_clear_cached_mappings
__host1x_bo_unpin() drops the last reference to the mapping and frees
it, so we can't dereference mapping afterwards. The cache itself
outlives the mapping, so use the cache local variable instead.
Merge tag 'xfs-fixes-7.2-rc4' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux
Pull xfs fixes from Carlos Maiolino:
"This contains mostly a series of bug fixes found by different LLM
models"
* tag 'xfs-fixes-7.2-rc4' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux: (21 commits)
xfs: don't zap bmbt forks if they are MAXLEVELS tall
xfs: clamp timestamp nanoseconds correctly
xfs: fully check the parent handle when it points to the rootdir
xfs: handle non-inode owners for rtrmap record checking
xfs: fix off-by-one error when calling xchk_xref_has_rt_owner
xfs: set xfarray killable sort correctly
xfs: grab rtrmap btree when checking rgsuper
xfs: write the rg superblock when fixing it
xfs: use the rt version of the cow staging checker
xfs: use rtrefcount btree cursor in xchk_xref_is_rt_cow_staging
xfs: don't wrap around quota ids in dqiterate
xfs: move cow_replace_mapping to xfs_bmap_util.c
xfs: make cow repair somewhat flaky when debugging knob enabled
xfs: don't replace the wrong part of the cow fork
xfs: resample the data fork mapping after cycling ILOCK
xfs: fix null pointer dereference in tracepoint
xfs: use xfs_csn_t for xlog_cil_push_now() push_seq parameter
xfs: tie zoned sysfs lifetime to zone info
xfs: fail recovery on a committed log item with no regions
xfs: splice unsorted log items back to the transaction after the loop
...
Merge tag 'erofs-for-7.2-rc4-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs
Pull erofs fixes from Gao Xiang:
- Fix sanity checks for ztailpacking tail pclusters to avoid
false corruption reports
- Use more informative s_id for file-backed mounts
- Hide the meaningless "cache_strategy=" mount option on plain
(uncompressed) filesystems
- Remove the unneeded erofs_is_ishare_inode() helper
* tag 'erofs-for-7.2-rc4-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs:
erofs: hide "cache_strategy=" for plain filesystems
erofs: get rid of erofs_is_ishare_inode() helper
erofs: relax sanity check for tail pclusters due to ztailpacking
erofs: use more informative s_id for file-backed mounts
Merge tag 'pm-7.2-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Pull power management fixes from Rafael Wysocki:
"These fix two cpufreq issues, one in the intel_pstate driver and one
in the core:
- Make cpufreq_update_pressure() use cpuinfo.max_freq as the default
reference frequency when arch_scale_freq_ref() returns 0 to allow
the scheduler to still take CPU frequency caps into account in
those cases (Rafael Wysocki)
- Use the HWP guaranteed performance level as the full capacity
performance in intel_pstate on hybrid systems when turbo
frequencies are not allowed to be used to make scale-invariance
work as expected in those cases (Rafael Wysocki)"
* tag 'pm-7.2-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
cpufreq: Make cpufreq_update_pressure() fall back to cpuinfo.max_freq
cpufreq: intel_pstate: Set non-turbo capacity to HWP_GUARANTEED_PERF()
Merge tag 'pmdomain-v7.2-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/ulfh/linux-pm
Pull pmdomain fixes from Ulf Hansson:
"imx:
- Assign child domains for imx93 to prevent power off when in use
- Fix i.MX8MP power up sequences
mediatek:
- Fix possible nullptr in HWV cleanup/on-check"
* tag 'pmdomain-v7.2-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/ulfh/linux-pm:
pmdomain: mediatek: Fix possible nullptr KP in HWV cleanup/on-check
pmdomain: imx: Fix i.MX8MP VC8000E power up sequence
pmdomain: imx: Fix i.MX8MP power notifier
pmdomain: imx93-blk-ctrl: Extract PHY as shared domain for DSI/CSI
dt-bindings: power: imx93: Add MIPI PHY power domain
Compare the sampled clock values instead of their addresses. Comparing
addresses leaves the samples unsorted, preventing the code from discarding
the minimum and maximum samples.
regulator: ltc3676: Fix incorrect IRQSTAT bit offsets
The LTC3676_IRQSTAT_* bit definitions do not match the IRQSTAT
(Interrupt Request Status) register layout documented in Table 15
of the LTC3676/LTC3676-1 datasheet:
bit 0 - Pushbutton Status Active
bit 1 - Hard Reset Occurred
bit 2 - PGOOD Timeout Occurred
bit 3 - Undervoltage Warning
bit 4 - Undervoltage Standby (Fault) Occurred
bit 5 - Overtemperature Warning
bit 6 - Overtemperature Standby (Fault) Occurred
bit 7 - Reserved
The driver instead defines these starting at bit 3, one bit higher
than the datasheet specifies, which causes ltc3676_regulator_isr()
to check the wrong status bits and misreport (or miss) PGOOD
timeout, undervoltage and thermal warning/fault conditions.
ksmbd: validate compound request size before reading StructureSize2
When ksmbd validates a compound (chained) SMB2 request,
ksmbd_smb2_check_message() reads pdu->StructureSize2 without first
checking that the compound element is large enough to contain it.
StructureSize2 is a 2-byte field at offset 64
(__SMB2_HEADER_STRUCTURE_SIZE) from the start of each element.
The compound-walking logic only guarantees that a full 64-byte SMB2
header is present for the trailing element: when NextCommand is 0, len is
reduced to the number of bytes remaining after next_smb2_rcv_hdr_off. A
remote client can craft a compound request whose last element has exactly
64 bytes, so the 2-byte StructureSize2 read at offset 64 extends one byte
past the receive buffer, producing a slab-out-of-bounds read.
BUG: KASAN: slab-out-of-bounds in ksmbd_smb2_check_message (fs/smb/server/smb2misc.c:402)
Read of size 2 at addr ffff888012ae31ac by task kworker/0:1/14
The buggy address is located 172 bytes inside of allocated 173-byte region
Workqueue: ksmbd-io handle_ksmbd_work
Call Trace:
...
kasan_report (mm/kasan/report.c:595)
ksmbd_smb2_check_message (fs/smb/server/smb2misc.c:402)
handle_ksmbd_work (fs/smb/server/server.c:119)
process_one_work (kernel/workqueue.c:3314)
worker_thread (kernel/workqueue.c:3397)
kthread (kernel/kthread.c:436)
ret_from_fork (arch/x86/kernel/process.c:158)
ret_from_fork_asm (arch/x86/entry/entry_64.S:245)
Reject any compound element that is too small to hold StructureSize2
before dereferencing it.
Fixes: e2f34481b24d ("cifsd: add server-side procedures for SMB3") Reported-by: AutonomousCodeSecurity@microsoft.com Signed-off-by: Xiang Mei (Microsoft) <xmei5@asu.edu> Acked-by: Namjae Jeon <linkinjeon@kernel.org> Signed-off-by: Steve French <stfrench@microsoft.com>
Gil Portnoy [Mon, 13 Jul 2026 00:20:09 +0000 (09:20 +0900)]
ksmbd: lock the binding preauth session in smb3_preauth_hash_rsp
smb3_preauth_hash_rsp() computes the SMB3.1.1 preauth integrity hash on
the response path. For a binding SESSION_SETUP it looks up the
per-connection preauth_session and reads its Preauth_HashValue.
smb2_sess_setup() frees that preauth_session under ksmbd_conn_lock().
Two SMB2 requests on one connection can run concurrently, so an unlocked
lookup and hash can use a preauth_session after another worker frees it.
Take ksmbd_conn_lock() before selecting conn->binding and hold it across
the selected preauth hash lookup and update. This preserves the existing
hash selection while preventing the lookup-to-use lifetime race.
Fixes: 1c5daa2ea924 ("ksmbd: handle channel binding with a different user") Signed-off-by: Gil Portnoy <dddhkts1@gmail.com> Acked-by: Namjae Jeon <linkinjeon@kernel.org> Signed-off-by: Steve French <stfrench@microsoft.com>
Gil Portnoy [Sun, 12 Jul 2026 00:00:00 +0000 (00:00 +0000)]
ksmbd: remove stale channels from all sessions on teardown
ksmbd_sessions_deregister() removes a connection's channels from other
sessions' channel lists only while conn->binding is still set:
if (conn->binding) {
hash_for_each_safe(sessions_table, ...)
ksmbd_chann_del(conn, sess);
}
conn->binding is a transient flag: it is cleared once a binding
SESSION_SETUP completes, and also by a subsequent non-binding
SESSION_SETUP on the same connection (a reauthentication on a bound
channel, or a new SessionId==0 setup). A connection that has bound a
channel into another session's ksmbd_chann_list and then clears
conn->binding leaves that channel behind when it disconnects: the
channel, whose chann->conn points at the now freed struct ksmbd_conn,
stays on the owner session's list.
When the owning connection later tears down, the second loop
dereferences the stale channel:
which is a use-after-free write into the freed ksmbd_conn (the same
stale channel is also walked by show_proc_session() through /proc). The
session is leaked as well, because its channel list never empties.
Remove the conn->binding gate so a connection always removes its
channels from every session on teardown.
Fixes: faf8578c77f3 ("ksmbd: find bound sessions during reauthentication") Signed-off-by: Gil Portnoy <dddhkts1@gmail.com> Acked-by: Namjae Jeon <linkinjeon@kernel.org> Signed-off-by: Steve French <stfrench@microsoft.com>
Gil Portnoy [Sun, 12 Jul 2026 00:00:00 +0000 (00:00 +0000)]
ksmbd: fix stack buffer overflow in multichannel session-key copy
Commit 4b706360ffb7 ("ksmbd: fix multichannel binding and enforce channel
limit") moved the binding-path session key out of the session-wide
sess->sess_key (CIFS_KEY_SIZE = 40) into a new per-channel buffer, and
sized both that buffer and the on-stack copy used during binding with
SMB2_NTLMV2_SESSKEY_SIZE (16):
On a binding SESSION_SETUP, auth_key points at the 16-byte channel_key,
so a client that supplies an NTLM EncryptedRandomSessionKey of up to 40
bytes (with NTLMSSP_NEGOTIATE_KEY_EXCH), or a Kerberos ticket whose
session key is longer than 16 bytes (a normal AES256 key is 32), writes
past the 16-byte stack buffer -- up to a 24-byte kernel stack overflow.
KASAN reports it as a stack-out-of-bounds write in arc4_crypt() called
from ksmbd_decode_ntlmssp_auth_blob().
The destinations must be able to hold the full session key the length
checks already permit. Size the per-channel key buffer and the two
on-stack channel_key buffers with CIFS_KEY_SIZE, matching sess->sess_key.
Fixes: 4b706360ffb7 ("ksmbd: fix multichannel binding and enforce channel limit") Signed-off-by: Gil Portnoy <dddhkts1@gmail.com> Acked-by: Namjae Jeon <linkinjeon@kernel.org> Signed-off-by: Steve French <stfrench@microsoft.com>
Gil Portnoy [Fri, 10 Jul 2026 11:27:10 +0000 (20:27 +0900)]
ksmbd: fix memory leak of xattr_stream_name in smb2_rename()
On an SMB2 SET_INFO(FileRenameInformation) whose target names an alternate
data stream, smb2_rename() obtains a formatted stream-name string from
ksmbd_vfs_xattr_stream_name(), which allocates it with kasprintf() and
returns it through an out-param:
xattr_stream_name is declared inside the alternate-data-stream block, but
the out: label is outside that block and frees only new_name, so it cannot
release xattr_stream_name. ksmbd_vfs_setxattr() takes a const char * and
only reads the name, so it does not take ownership either. Both the
setxattr-failure and the success path therefore leak the kasprintf()'d
string. An authenticated client with a writable share can leak kernel
memory on every stream rename, exhausting kernel memory over time.
Free xattr_stream_name after its use, before the block's goto out. The
two earlier goto out paths never assign the variable, so there is no
double-free.
Signed-off-by: Gil Portnoy <dddhkts1@gmail.com> Acked-by: Namjae Jeon <linkinjeon@kernel.org> Signed-off-by: Steve French <stfrench@microsoft.com>
Gil Portnoy [Fri, 10 Jul 2026 11:23:34 +0000 (20:23 +0900)]
ksmbd: zero the smb2_read alignment tail to avoid an infoleak
Commit 6b9a2e09d4cc ("ksmbd: avoid zeroing the read buffer in smb2_read()")
switched the SMB2 READ payload buffer from kvzalloc() to kvmalloc(), on the
premise that only the nbytes actually read are ever transmitted, so the
ALIGN(length, 8) tail need not be initialized.
That premise does not hold for a compound response. ksmbd_vfs_read() fills
only nbytes, leaving [nbytes, ALIGN(length, 8)) uninitialized. The aux
payload is pinned as the last response iov with iov_len == nbytes, but when
the READ is a member of a compound, init_chained_smb2_rsp() 8-byte-aligns
the previous member by extending that same iov:
so up to 7 uninitialized bytes of the kvmalloc()'d slab tail are sent
to the client. When the read length is small the buffer is served from
a general kmalloc slab, so those bytes can be stale kernel-heap
contents, including pointer values -- an information leak usable to
defeat KASLR.
An authenticated client triggers it with a compound request containing a
READ whose returned nbytes is not 8-aligned (for example [READ, CLOSE] with
a 1-byte read).
Zero only the alignment tail after the read, preserving the bulk
no-zeroing optimization of 6b9a2e09d4cc.
Fixes: 6b9a2e09d4cc ("ksmbd: avoid zeroing the read buffer in smb2_read()") Cc: stable@vger.kernel.org Signed-off-by: Gil Portnoy <dddhkts1@gmail.com> Acked-by: Namjae Jeon <linkinjeon@kernel.org> Signed-off-by: Steve French <stfrench@microsoft.com>
ksmbd: pin conn during async oplock break notification
smb2_oplock_break_noti() and smb2_lease_break_noti() store a ksmbd_conn
pointer in an async ksmbd_work and then queue that work on ksmbd-io. The
work only increments conn->r_count, which prevents teardown from passing
the pending-request wait after the increment, but it does not pin the
struct ksmbd_conn object.
If connection teardown races with an oplock break notification, the last
conn reference can be dropped before the queued worker finishes. The
worker then uses the freed conn in ksmbd_conn_write() and
ksmbd_conn_r_count_dec().
Take a real conn reference when publishing the conn pointer to the async
work item, and drop it after the notification work has decremented
r_count. Apply the same lifetime rule to lease break notification, which
uses the same work->conn pattern.
Fixes: 3aa660c05924 ("ksmbd: prevent connection release during oplock break notification") Signed-off-by: Qihang <q.h.hack.winter@gmail.com> Acked-by: Namjae Jeon <linkinjeon@kernel.org> Signed-off-by: Steve French <stfrench@microsoft.com>
AllocationSize is a fully client-controlled __le64 field; the only
validation performed by the caller (smb2_set_info_file(), case
FILE_ALLOCATION_INFORMATION) is that the fixed buffer is at least
sizeof(struct smb2_file_alloc_info) == 8 bytes. The value itself is
never range-checked before this arithmetic.
When AllocationSize is close to U64_MAX (e.g. 0xffffffffffffffff),
"AllocationSize + 511" wraps around mod 2^64 to a small number
(0xffffffffffffffff + 511 = 510), so alloc_blks becomes 0. Since any
existing regular file has stat.blocks > 0, the function then takes
the "shrink" branch and calls:
silently truncating the file to size 0, even though the client asked
to grow the allocation to (what looks like) the maximum possible
size. The trailing "if (size < alloc_blks * 512) i_size_write(inode,
size);" restore is guarded by a comparison that is never true once
alloc_blks == 0, so the truncation is not undone. This lets an
authenticated SMB client that already holds an open handle with
FILE_WRITE_DATA on a file silently truncate that same file to size 0
via a single crafted SET_INFO(FILE_ALLOCATION_INFORMATION) request
advertising a near-U64_MAX AllocationSize, even though the request
asks to grow the file's allocation rather than shrink it. This is a
functional/data-loss bug, not a privilege-boundary
violation: the same client could already truncate the file via
FILE_END_OF_FILE_INFORMATION or a plain write.
Fix it by validating AllocationSize against MAX_LFS_FILESIZE, the
same upper bound the VFS itself uses to reject unrepresentable file
sizes, before doing the "+511" rounding, and rejecting oversized
values with -EINVAL. Bounding AllocationSize to
MAX_LFS_FILESIZE - 511 guarantees the "+511" addition cannot wrap,
and that the subsequent "alloc_blks * 512" values passed to
vfs_fallocate() and ksmbd_vfs_truncate() stay within a representable
loff_t as well.
No legitimate SMB client asks for an allocation size anywhere near
2^64 bytes, so this only rejects a value that was previously
silently misinterpreted as zero.
Runtime-verified on a v6.19 KASAN test stand: sending SET_INFO
(FILE_ALLOCATION_INFORMATION) with AllocationSize = 0xffffffffffffffff
against ksmbd now returns -EINVAL and leaves the target file's size
unchanged, where the unpatched kernel truncated it from 4096 to 0
bytes.
Fixes: e2f34481b24d ("cifsd: add server-side procedures for SMB3") Cc: stable@vger.kernel.org Signed-off-by: Ibrahim Hashimov <security@auditcode.ai> Assisted-by: AuditCode-AI:2026.07 Acked-by: Namjae Jeon <linkinjeon@kernel.org> Signed-off-by: Steve French <stfrench@microsoft.com>
drm/xe/guc: Hold device ref until queue teardown completes
GuC exec queue destruction can run asynchronously. If the final device
put happens from a destroy worker, drmm cleanup can end up draining
the same workqueue and deadlock.
Hold a drm_device reference for the queue lifetime and drop it after
queue teardown completes. This keeps drmm cleanup from running while
async destroy work is still pending.
Move GuC destroy work to a module-lifetime Xe workqueue and flush it
on PCI remove so hot-unbind/rebind still waits for pending destroy work.
With queue-held device refs, guc_submit_sw_fini() cannot run with live
GuC IDs. Replace the fini wait with an assertion and remove the unused
fini_wq.
v2:
- Rebase
v3:
- Switch to queue-lifetime drm_dev_get()/drm_dev_put() model. (Matt)
- Queue async teardown on system_dfl_wq instead of xe->destroy_wq. (Matt)
- Drop separate deferred drm_dev_put worker.
- Remove stale drain_workqueue(xe->destroy_wq) from guc_submit_sw_fini().
v4:
- Replace the guc_submit_sw_fini() wait with an assertion and remove
the now-unused fini_wq. (sashiko)
v5:
- Move destroy work to a module-lifetime Xe workqueue instead of
system_dfl_wq. (Matt)
- Flush the module-lifetime destroy workqueue during PCI remove to
preserve the old device-remove wait semantics.
v6:
- Keep SVM pagemap destroy work on the per-device destroy_wq to avoid
letting it outlive the xe_device/drm_device. (Sashiko)
- Use WQ_MEM_RECLAIM for xe->destroy_wq because SVM pagemap destroy work
can be queued from the reclaim path.
v7:
- Drop the per-device xe->destroy_wq and use the module-level destroy WQ
for SVM pagemap destroy as well. (Matt)
- Rename xe_exec_queue_destroy_wq_*() helpers to xe_destroy_wq_*()
helpers because the WQ is no longer exec-queue specific. (Matt)
v8:
- Rebase.
v9:
- Keep SVM pagemap destroy work on the per-device WQ_MEM_RECLAIM
destroy_wq because it can be queued from reclaim and embeds
the dev_pagemap used by devres teardown. (Sashiko)
- Keep the module-level destroy WQ GuC-only and drop WQ_MEM_RECLAIM
from it.
- Update the module-WQ kdoc to document the GuC/SVM split.
v10:
- Keep xe->destroy_wq per-cpu while adding WQ_MEM_RECLAIM to fix the
workqueue allocation warning.
v11:
- Drop the SVM pagemap destroy comment as it was revision-specific.
(Thomas)
v12:
- Rebase.
Fixes: 2d2be279f1ca ("drm/xe: fix UAF around queue destruction") Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com> Cc: Tejas Upadhyay <tejas.upadhyay@intel.com> Reviewed-by: Matthew Brost <matthew.brost@intel.com> Signed-off-by: Arvind Yadav <arvind.yadav@intel.com> Link: https://patch.msgid.link/20260716062624.211396-1-arvind.yadav@intel.com Signed-off-by: Tejas Upadhyay <tejas.upadhyay@intel.com>
(cherry picked from commit da1124abac689cc2b1d8995e5f0a816f8a122edb) Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Zongyao Bai [Tue, 14 Jul 2026 23:24:32 +0000 (23:24 +0000)]
drm/xe/pt: Reset current_op in xe_pt_update_ops_init()
xe_pt_update_ops_init() fails to reset current_op to 0. On the
vm_bind path, ops_execute() calls xe_pt_update_ops_prepare() inside
the xe_validation_guard() / drm_exec_until_all_locked() loop. When
that loop retries due to lock contention or OOM eviction
(drm_exec_retry_on_contention() / xe_validation_retry_on_oom()),
xe_pt_update_ops_prepare() runs again on the same vops, and each
call to bind_op_prepare() increments current_op without resetting it.
After N retries current_op exceeds the array size allocated by
xe_vma_ops_alloc(), causing an out-of-bounds write into
SLUB-poisoned memory and a subsequent UAF crash in
xe_migrate_update_pgtables_cpu() when reading the corrupted pt_op->bind.
Also reset needs_svm_lock and needs_invalidation which are derived in
the same prepare pass and would otherwise cause wrong migrate ops
selection and redundant TLB invalidation on retry.
Fix this by resetting current_op, needs_svm_lock and needs_invalidation
in xe_pt_update_ops_init().
v2 (Matt):
- Add details in commit message.
- Add Fixes tag and Cc to stable@vger.kernel.org
Fixes: e8babb280b5e ("drm/xe: Convert multiple bind ops into single job") Suggested-by: Matthew Auld <matthew.auld@intel.com> Cc: stable@vger.kernel.org Assisted-by: GitHub-Copilot:claude-sonnet-4.6 Signed-off-by: Zongyao Bai <zongyao.bai@intel.com> Reviewed-by: Matthew Brost <matthew.brost@intel.com> Signed-off-by: Matthew Brost <matthew.brost@intel.com> Link: https://patch.msgid.link/20260714232433.2737533-1-zongyao.bai@intel.com
(cherry picked from commit 046045543e530605c441063535e7dca0075369a6) Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Starting on LNL the WOPCM size is 8MB instead of 4, so we need to avoid
using the [0, 8MB) range of the GGTT as that can be unaccessible from
the microcontrollers.
Note that the proper long-term fix here is to read the WOPCM size from
the HW, but that is a more serious rework that would be difficult to
backport, so we can do that as a follow-up.
Fixes: 9c57bc08652a ("drm/xe/lnl: Drop force_probe requirement") Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Shuicheng Lin <shuicheng.lin@intel.com> Cc: Matt Roper <matthew.d.roper@intel.com> Reviewed-by: Shuicheng Lin <shuicheng.lin@intel.com> Link: https://patch.msgid.link/20260713221758.3285744-2-daniele.ceraolospurio@intel.com
(cherry picked from commit 3033b0b24ed0e2f5e56bdd4d9c183417c365a45b) Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Matthew Brost [Tue, 14 Jul 2026 06:24:40 +0000 (23:24 -0700)]
drm/xe/vf: Fix VF CCS attach/detach race with in-flight BO moves
xe_bo_move() attaches VF CCS read/write batch buffers (BBs) to a BO
after it transitions NULL/SYSTEM -> TT, and detaches them after it
transitions TT -> SYSTEM. Both operations were done synchronously on
the CPU immediately after building the move's copy/clear fence,
without waiting for that fence to signal. This creates two races with
VF migration:
- Attach happens too late relative to the copy job it is meant to
protect. If the copy job is submitted before the CCS BBs are
attached, a VF migration event that pauses execution mid-copy can
observe partially copied CCS metadata without the attach state
needed to correctly save/restore it.
- Detach happens too early relative to the copy job that moves data
out of TT. The CCS BBs are torn down right after the copy fence is
obtained, while the actual blit may still be in flight. A VF
migration event that pauses execution mid-copy can then race the
save/restore path against the still-running blit, and the CCS BBs
it would need to make sense of the paused state have already been
removed.
Fix both races:
- Move the attach call to before the copy/clear job is submitted, so
the CCS BBs are already registered by the time the copy runs. On
attach failure, unwind and bail out of the move. xe_migrate_ccs_rw_copy()
now takes the destination resource explicitly, since bo->ttm.resource
is not updated to the new resource until after the move commits.
- Detach only after explicitly waiting for the copy fence to signal,
instead of tearing down the CCS BBs immediately after obtaining it.
While here, also fix xe_sriov_vf_ccs_attach_bo() to properly unwind and
propagate errors: the per-context loop previously never broke out on
error, silently discarding earlier failures. Unwind by clearing each
attached context directly via xe_migrate_ccs_rw_copy_clear() instead of
reusing xe_sriov_vf_ccs_detach_bo(), which requires both contexts to be
attached before it will clean up either one.
Fixes: 864690cf4dd6 ("drm/xe/vf: Attach and detach CCS copy commands with BO") Cc: Michal Wajdeczko <michal.wajdeczko@intel.com> Cc: Matthew Auld <matthew.auld@intel.com> Cc: Michał Winiarski <michal.winiarski@intel.com> Cc: Satyanarayana K V P <satyanarayana.k.v.p@intel.com> Assisted-by: GitHub_Copilot:claude-sonnet-5 Signed-off-by: Matthew Brost <matthew.brost@intel.com> Acked-by: Satyanarayana K V P <satyanarayana.k.v.p@intel.com> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Link: https://patch.msgid.link/20260714062440.3421225-1-matthew.brost@intel.com
(cherry picked from commit d45ad0aa7a1eb5d7288b5ed948b05695611dc39e) Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Alexander Usyskin [Tue, 14 Jul 2026 05:54:17 +0000 (08:54 +0300)]
drm/xe/nvm: fix writable override for CRI
The witable override should be set when FDO_MODE bit is enabled.
Fix the comparison to distingush this case from legacy systems
where bit should be disabled to have override.
An imported dma-buf BO is created as a ttm_bo_type_sg BO whose
reservation object is the exporter's dma_buf->resv. The importer,
however, only takes a dma-buf reference after a successful
dma_buf_dynamic_attach(). Until then nothing keeps the exporter alive,
so if the exporter is freed while the BO still references its resv, a
later access to that resv is a use-after-free:
Oops: general protection fault, probably for non-canonical address
0x6b6b6b6b6b6b6b9c
Workqueue: ttm ttm_bo_delayed_delete [ttm]
RIP: 0010:mutex_can_spin_on_owner+0x3f/0xc0
This can be reached on two paths:
- dma_buf_dynamic_attach() fails, or
- ttm_bo_init_reserved() fails during BO creation.
In both cases the BO already has bo->base.resv pointing at the exporter
resv, and sg BOs are always torn down via ttm_bo_delayed_delete(), which
locks bo->base.resv asynchronously - potentially after the exporter has
been freed.
Take the dma-buf reference in xe_bo_init_locked(), before
ttm_bo_init_reserved(), so it also covers a creation failure there, and
release it in xe_ttm_bo_destroy(). The reference is held for the whole
BO lifetime, keeping the shared resv alive on every path.
v2:
- Reworked the fix to avoid creating the imported sg BO before
dma_buf_dynamic_attach() succeeds.
- Attach with importer_priv == NULL and make invalidate_mappings ignore
incomplete imports.
v3:
- Dropped the xe-side reordering approach since importer_priv must be
valid when dma_buf_dynamic_attach() publishes the attachment.
- Per Christian's suggestion on the v1 thread, keyed the check on
import_attach rather than removing the sg guard entirely.
- Fixes both xe and amdgpu in a single TTM patch.
v4:
- Moved import_attach check to after dma_resv_copy_fences() so fences
are copied before returning for successful imports (Thomas).
- Removed exporter-alive claim from commit message (Thomas).
v5:
- Add drm/xe patch to keep imported sg BOs off the LRU before attach
succeeds; the TTM fix alone is not sufficient for xe if the BO is
already LRU-visible. (Thomas)
v4 patch:
https://patchwork.freedesktop.org/patch/736663/?series=169129&rev=2
- Patch 1 (drm/ttm) carries Christian's Reviewed-by from v4.
v6:
- Reworked the fix based on Thomas' suggestion. Instead of the TTM resv
individualization (v1-v5) plus the xe off-LRU/placement handling (v5),
just hold a dma-buf reference for the imported BO lifetime so the
shared resv can never be freed while the BO still references it.
Single xe patch, no TTM change. (Thomas)
- Take the reference in xe_bo_init_locked() before ttm_bo_init_reserved()
so a TTM creation failure is covered too (Thomas).
- Dropped the v5 series (drm/ttm + drm/xe off-LRU); the off-LRU approach
also regressed in CI BAT via ttm_bo_pipeline_gutting() creating a ghost
BO that outlived the exporter.
Link to v5: https://patchwork.freedesktop.org/series/169984/
v7:
- Move changelog above --- so it stays in the commit message.
- Reorder changelog entries oldest-to-newest. (Thomas)
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/work_items/8023 Fixes: dd08ebf6c352 ("drm/xe: Introduce a new DRM driver for Intel GPUs") Cc: stable@vger.kernel.org Cc: Thomas Hellstrom <thomas.hellstrom@linux.intel.com> Cc: Christian Konig <christian.koenig@amd.com> Cc: Matthew Auld <matthew.auld@intel.com> Suggested-by: Thomas Hellstrom <thomas.hellstrom@linux.intel.com> Assisted-by: GitHub_Copilot:claude-sonnet-4.6 Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com> Signed-off-by: Nitin Gote <nitin.r.gote@intel.com> Signed-off-by: Matthew Auld <matthew.auld@intel.com> Link: https://patch.msgid.link/20260710191027.260160-2-nitin.r.gote@intel.com
(cherry picked from commit 3516f3fae6be35642f8f06f8a218da6425c0306a) Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
drm/xe/vm: Fix BO prefetch with CONSULT_MEM_ADVISE_PREF_LOC
When prefetch region is DRM_XE_CONSULT_MEM_ADVISE_PREF_LOC for a BO VMA,
the code used it as an index into region_to_mem_type[], causing an
out-of-bounds access since the value is -1.
Resolve the preferred location for BO VMAs directly: local VRAM on dGFX
(using the BO's tile placement) or system memory on iGPU.
Discovered using AI-assisted static analysis confirmed by Intel Product
Security.
v2:
-Fix null dereference
Reported-by: Martin Hodo <martin.hodo@intel.com> Fixes: c1bb69a2e8e2 ("drm/xe/svm: Consult madvise preferred location in prefetch") Cc: Matthew Brost <matthew.brost@intel.com> Cc: stable@vger.kernel.org Reviewed-by: Matthew Brost <matthew.brost@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20260624174943.2808767-2-himal.prasad.ghimiray@intel.com Signed-off-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
(cherry picked from commit d9a4906ac03be9f6ed3f3b45c56c866b867fd75b) Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Pull kvm fixes from Paolo Bonzini:
"arm64:
- Fix an accounting buglet when reclaiming pages from a protected
guest
- Fix a bunch of architectural compliance issues when injecting a
synthesised exception, most of which were missing the PSTATE.IL bit
indicating a 32bit-wide instruction
- Another set of fixes addressing issues with translation of
VNCR_EL2, including corner cases where the guest point that
register at a RO page...
- Don't warn when trapping accesses to ZCR_EL2 from an L2 guest, as
that's not unexpected at all
- Address a bunch of races with LPI migration vs LPIs being disabled
- Fix a total howler of a bug combining FEAT_MOPS and NV, resulting
in exception returning in the wrong place...
- Move locking for kvm_io_bus_get_dev() into the caller, ensuring
race-free checks that the returned object is of the correct type
- Fix initialisation of the page-table walk level when relaxing
permissions
- Correctly update the XN attribute when relaxing permissions
- Fix the sign extension of loads from emulated MMIO regions
- Assorted collection of fixes for pKVM's FFA proxy, together with a
couple of FFA driver adjustments
- Coerce Fuad Tabba into a reviewer role, and may his Inbox catch
fire!
s390:
- more gmap KVM memory management fixes
- PCI passthru fixes
x86:
- Fix a bug where KVM will trigger a UAF if updating IOMMU IRTEs
fails when registering an IRQ-bypass producer
- Ignore pending PV EOI instead of BUG()ing the host if the feature
was disabled by the guest
- Fix nVMX bugs where KVM would run L1 with an L1-controlled CR3
after a failed "late" consistency check when KVM is NOT using EPT
- Disallow intra-host migration/mirroring of SNP VMs as KVM doesn't
yet support moving/mirroring SNP state
- Fix a TOCTOU bug in KVM's handling of the "trusted" CPUID for TDX
guests
- Fix a NULL pointer deref in trace_kvm_inj_exception() where a
change to the core infrastructure missed KVM's unique (ab)use of
__print_symbolic()
- Put vmcs12 pages if nested VM-Enter fails due to invalid guest
state
- Fix TLB conflicts between two VMs if one of them VM is run on a CPU
before and after it is hotplugged"
* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (48 commits)
KVM: SVM: Bump asid_generation on CPU online to avoid ASID collision after hotplug
KVM: nVMX: Put vmcs12 pages if nested VM-Enter fails due to invalid guest state
KVM: x86: Fix null pointer deref due to dummy array in trace_kvm_inj_exception()
KVM: TDX: Reject concurrent change to CPUID entry count
KVM: selftests: Verify SNP VMs are rejected from migration and mirroring
KVM: SEV: Do not allow intra-host migration/mirroring of SNP VMs
KVM: s390: pci: Fix handling of AIF enable without AISB
KVM: s390: Improve kvm_s390_vm_stop_migration()
KVM: s390: Fix dat_crste_walk_range() early return
KVM: s390: vsie: Avoid potential deadlock with real spaces
KVM: s390: pci: Fix GISC refcount leak on AIF enable failure
KVM: nVMX: Don't use vmcs01.GUEST_CR3 to snapshot L1's CR3 when EPT is disabled
KVM: nVMX: Move vTPR vs. TPR Threshold consistency check into "normal" checks
KVM: x86: Ignore pending PV EOI if the vCPU has since disabled PV EOIs
KVM: x86: Nullify irqfd->producer if updating IRTE for bypass fails
KVM: arm64: Fix propagation of TLBI level in kvm_pgtable_stage2_relax_perms()
firmware: arm_ffa: Fix Endpoint Memory Access Descriptor offset calculation
firmware: arm_ffa: Fix out-of-bound writes in ffa_setup_and_transmit()
KVM: arm64: Zero out the stack initialized data in the FFA handler
KVM: arm64: Ensure FFA ranges are page aligned
...
Christoph Hellwig [Thu, 16 Jul 2026 09:12:46 +0000 (11:12 +0200)]
block: fix aligning of bounced dio read bios
bio_iov_iter_align_down expects the "normal" biovec layout from vector 0,
while bio_iov_iter_bounce_read abuses vector 0 for a bounce buffer
allocation. Pass an explicit bvec to bio_iov_iter_align_down to deal
with this case to avoid a double unpin.
Additionally we need to free the folio if no bio_vec could be added,
and adjust the size of the first bio_vec that contains the bounce buffer
when the I/O size is aligned down.
Fixes: e7b8b3c5b2a6 ("block: align down bounces bios") Reported-by: 0wnerD1ed <l7z@0b1t.tech> Signed-off-by: Christoph Hellwig <hch@lst.de> Tested-by: 0wnerD1ed <l7z@0b1t.tech> Reviewed-by: Keith Busch <kbusch@kernel.org> Link: https://patch.msgid.link/20260716091306.316625-3-hch@lst.de Signed-off-by: Jens Axboe <axboe@kernel.dk>
Christoph Hellwig [Thu, 16 Jul 2026 09:12:45 +0000 (11:12 +0200)]
block: handle huge zero folios in bio_free_folios
When CONFIG_PERSISTENT_HUGE_ZERO_FOLIO is enabled, iomap_dio_zero() can
add a huge zero folio to a zeroing bio, which needs special treatment
in bio_free_folios by also checking is_huge_zero_folio() in addition to
is_zero_folio().
Fixes: 8dd5e7c75d7b ("block: add helpers to bounce buffer an iov_iter into bios") Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hannes Reinecke <hare@kernel.org> Link: https://patch.msgid.link/20260716091306.316625-2-hch@lst.de Signed-off-by: Jens Axboe <axboe@kernel.dk>
Joseph Qi [Thu, 9 Jul 2026 02:01:45 +0000 (10:01 +0800)]
block: try slab allocation in bio_alloc_bioset() before mempool
When the per-CPU bio cache is enabled but empty, bio_alloc_percpu_cache()
returns NULL and bio_alloc_bioset() falls straight through to the mempool
fallback:
if (unlikely(!bio)) {
if (!(saved_gfp & __GFP_DIRECT_RECLAIM))
return NULL;
...
}
For non-sleeping allocations (no __GFP_DIRECT_RECLAIM) this returns NULL
without ever attempting a slab allocation, even when there is plenty of
free memory.
Commit b520c4eef83d ("block: split bio_alloc_bioset more clearly into a
fast and slowpath") introduced this. Before it, a percpu cache miss fell
through to mempool_alloc(), which attempted the underlying slab allocation
first and only failed when that slab allocation failed. The restructuring
dropped the slab attempt that non-sleeping callers of a cache-enabled
bioset (such as the default fs_bio_set used by bio_alloc()) relied on.
Try a slab allocation with optimistic GFP_ flags before falling back to
the mempool whenever the bio is still NULL, so both the cache-empty and
non-cache paths share the same slab attempt. This restores the previous
behavior for non-sleeping allocations.
Fixes: b520c4eef83d ("block: split bio_alloc_bioset more clearly into a fast and slowpath") Suggested-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Joseph Qi <joseph.qi@linux.alibaba.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Link: https://patch.msgid.link/20260709020145.4011533-1-joseph.qi@linux.alibaba.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
Jackie Liu [Wed, 15 Jul 2026 07:33:41 +0000 (15:33 +0800)]
block: show operation in error injection rules
Rules listed through the error_injection debugfs file omit the block
operation they match. As a result, rules that differ only in operation
are indistinguishable even though op is mandatory when adding a rule.
block: serialize elevator changes for the same queue using a writer lock
When elevator_change() is called concurrently for the same queue, the
elevator_change_done() function runs concurrently as well. This function
adds or deletes kobjects for the debugfs entry of the queue. Then the
concurrent calls cause memory corruption of the kobjects and result in a
process hang. The core part of the elevator switch is protected by queue
freeze and q->elevator_lock. However, since the commit 559dc11143eb
("block: move elv_register[unregister]_queue out of elevator_lock"), the
elevator_change_done() is not serialized. Hence the memory corruption
and the hang.
The failures are observed when udev-worker writes to a sysfs
queue/scheduler attribute file while the blktests test case block/005
writes to the same attribute file. The failure also can be recreated by
running two processes that write to the same queue/scheduler file
concurrently. The failure is observed since another commit 370ac285f23a
("block: avoid cpu_hotplug_lock depedency on freeze_lock"). This commit
changed the behavior of queue freeze and it unveiled the failure.
Fix the failure by changing elv_iosched_store() to acquire
update_nr_hwq_lock as the writer lock instead of the reader lock. This
serializes the whole elevator switch steps, including the
elevator_change_done() call.
Fixes: 559dc11143eb ("block: move elv_register[unregister]_queue out of elevator_lock") Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com> Reviewed-by: Nilay Shroff <nilay@linux.ibm.com> Reviewed-by: Ming Lei <tom.leiming@gmail.com> Link: https://patch.msgid.link/20260716092237.1305030-1-shinichiro.kawasaki@wdc.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
Jackie Liu [Wed, 15 Jul 2026 07:35:18 +0000 (15:35 +0800)]
block: free copied pages when blk_rq_map_kern() fails
bio_copy_kern() allocates pages that are normally freed by the bio
completion callback. If blk_rq_append_bio() rejects the bio, however,
blk_rq_map_kern() only drops the bio reference. Since bio_put() does not
free pages referenced by the bio vectors, those pages leak.
This can happen when the bio exceeds the queue segment constraints or
when a later mapping cannot be merged into a request built by earlier
calls. Track whether the buffer was copied and free those pages before
dropping the rejected bio.
Fixes: 3a5a39276d2a ("block: allow blk_rq_map_kern to append to requests") Assisted-by: Codex:gpt-5.6-sol Signed-off-by: Jackie Liu <liuyun01@kylinos.cn> Reviewed-by: Christoph Hellwig <hch@lst.de> Link: https://patch.msgid.link/20260715073518.96042-1-liu.yun@linux.dev Signed-off-by: Jens Axboe <axboe@kernel.dk>
Oliver Hartkopp [Sun, 12 Jul 2026 17:59:43 +0000 (19:59 +0200)]
can: isotp: serialize TX state transitions under so->rx_lock
The TX state machine (so->tx.state) is driven from three contexts:
sendmsg() claiming and progressing a transfer, the RX path consuming
Flow Control/echo frames, and two hrtimers timing out a stalled
transfer. Mixing a lock-free cmpxchg() claim in sendmsg() with
hrtimer_cancel() calls made under so->rx_lock elsewhere left windows
where a frame or timer callback could act on a state that had already
moved on, corrupting an unrelated transfer.
so->rx_lock now covers the full lifecycle of a TX claim: sendmsg()
takes it to check so->tx.state is ISOTP_IDLE, switch it to
ISOTP_SENDING, bump so->tx_gen and drain the previous transfer's
timers - all as one critical section. isotp_rcv_fc()/isotp_rcv_cf()
already run under this lock via isotp_rcv(), and isotp_rcv_echo() now
takes it itself, so none of them can ever observe a transfer mid-claim.
This also means a transfer can no longer be handed to sendmsg()'s
cleanup paths (signal or send error) while another thread is
concurrently claiming or finishing it, so those paths can cancel
timers and reset the state unconditionally.
isotp_release() claims the socket the same way, so a racing sendmsg()
sees a consistent ISOTP_SHUTDOWN and skips arming its timer or sending.
Only the hrtimer callbacks stay outside so->rx_lock, since they run
under so->rx_lock's cancellation elsewhere and taking it themselves
would deadlock. so->tx_gen lets them recognize whether the transfer
they timed out is still the one currently active, so they don't
report an error against a transfer that has since completed or been
superseded.
Fixes: e057dd3fc20f ("can: add ISO 15765-2:2016 transport protocol") Reported-by: sashiko-bot@kernel.org Closes: https://lore.kernel.org/linux-can/20260710142146.BDAE61F000E9@smtp.kernel.org/ Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net> Link: https://patch.msgid.link/20260712-isotp-fixes-v10-3-793a1b1ce17f@hartkopp.net Cc: stable@kernel.org Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Oliver Hartkopp [Sun, 12 Jul 2026 17:59:42 +0000 (19:59 +0200)]
can: isotp: fix use-after-free race with concurrent NETDEV_UNREGISTER
isotp_release() looked up the bound network device via dev_get_by_index()
using the stored ifindex. During device unregistration the device is
unlisted from the ifindex hash before the NETDEV_UNREGISTER notifier
chain runs, so a concurrent isotp_release() could find no device, skip
can_rx_unregister() entirely, and still proceed to free the socket.
Since isotp_release() had already removed itself from the isotp
notifier list at that point, isotp_notify() would never get a chance to
clean up either, leaving a stale CAN filter that keeps pointing at the
freed socket.
Fix this the same way raw.c already does: hold a tracked reference to
the bound net_device in the socket (so->dev/so->dev_tracker) from
bind() onward instead of re-resolving it from the ifindex, and
serialize bind()/release() with rtnl_lock() so that so->dev is always
consistent with what the NETDEV_UNREGISTER notifier sees. so->dev
stays valid regardless of ifindex-hash unlisting, and is only ever
cleared by whichever of isotp_release()/isotp_notify() gets there
first, so the filter is always removed exactly once.
isotp_bind() now rejects a (re)bind with -EAGAIN while so->[tx|rx].state
isn't ISOTP_IDLE yet, so a timer left running by a prior
NETDEV_UNREGISTER can't act on a newly bound so->ifindex. Both checks
share the same lock_sock() section, so there is no window in which a
concurrent isotp_notify() clearing so->bound could be missed.
Fixes: e057dd3fc20f ("can: add ISO 15765-2:2016 transport protocol") Reported-by: sashiko-bot@kernel.org Closes: https://lore.kernel.org/linux-can/20260707101420.47F261F000E9@smtp.kernel.org/ Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net> Link: https://patch.msgid.link/20260712-isotp-fixes-v10-2-793a1b1ce17f@hartkopp.net Cc: stable@kernel.org Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Oliver Hartkopp [Sun, 12 Jul 2026 17:59:41 +0000 (19:59 +0200)]
can: isotp: use unconditional synchronize_rcu() in isotp_release()
isotp_notify() unregisters the (RCU) CAN filters via can_rx_unregister()
and clears so->bound without waiting for a grace period. isotp_release()
uses so->bound to decide whether it needs to call synchronize_rcu()
before cancelling so->rxtimer, so when NETDEV_UNREGISTER runs first it
skips that synchronize_rcu() and can cancel the timer while an
in-flight isotp_rcv() is still executing and about to re-arm it via
isotp_send_fc(), leading to a use-after-free timer callback on the
freed socket.
sakisho-bot remarked a problem with rtnl_lock held in isotp_notify(),
therefore make isotp_release() always call synchronize_rcu() before
cancelling the timers, regardless of so->bound. This still closes the
original race (isotp_notify() clearing so->bound without waiting for
in-flight isotp_rcv() callers before isotp_release() cancels the RX
timer) without adding any RCU wait to the netdevice notifier path.
Fixes: 14a4696bc311 ("can: isotp: isotp_release(): omit unintended hrtimer restart on socket release") Closes: https://lore.kernel.org/linux-can/20260707085210.6B6C01F000E9@smtp.kernel.org/ Reported-by: Nico Yip <zdi-disclosures@trendmicro.com> Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net> Link: https://patch.msgid.link/20260712-isotp-fixes-v10-1-793a1b1ce17f@hartkopp.net Cc: stable@kernel.org Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Marc Kleine-Budde [Thu, 16 Jul 2026 08:05:00 +0000 (10:05 +0200)]
Merge patch series "can: bcm: collected fixes"
Oliver Hartkopp <socketcan@hartkopp.net> says:
as there were different patches flying arround to fix CAN_BCM issues and AI
assisted stuff pop's up again and again, I've created this collection to be
applied.
Oliver Hartkopp [Tue, 14 Jul 2026 16:55:33 +0000 (18:55 +0200)]
can: bcm: track a single source interface for ANYDEV timeout/throttle ops
An ANYDEV rx op (ifindex == 0) with an active RX timeout and/or
throttle timer has no defined semantics when matching frames arrive
from several interfaces: bcm_rx_handler() can run concurrently for
the same op on different CPUs, racing hrtimer_cancel()/
bcm_rx_starttimer() against bcm_rx_timeout_handler() and causing
spurious RX_TIMEOUT notifications and last_frames corruption. The
same concurrency lets throttled multiplex frames from different
interfaces clobber the single rx_ifindex/rx_stamp fields shared by
the op.
Add op->if_detected to track the first interface that delivers a
matching frame while a timeout/throttle timer is configured, and
reject frames from any other interface for that op. The claim is
decided in bcm_rx_handler() before hrtimer_cancel() touches
op->timer, so a rejected frame can never disturb the claimed
interface's watchdog. RTR-mode ops are excluded via RX_RTR_FRAME,
independent of kt_ival1/kt_ival2, since those may briefly hold a
stale value from an earlier non-RTR configuration.
The claim is released in bcm_notify() on NETDEV_UNREGISTER and in
bcm_rx_setup() when SETTIMER reconfigures the timer values.
A (re-)claim is only possible on CAN devices in NETREG_REGISTERED
dev->reg_state to cover the release in bcm_notify() where reg_state
becomes NETREG_UNREGISTERING until synchronize_net().
Oliver Hartkopp [Tue, 14 Jul 2026 16:55:32 +0000 (18:55 +0200)]
can: bcm: fix data race on rx_stamp/rx_ifindex in bcm_rx_handler()
For an rx op subscribed on all interfaces (ifindex == 0), the same op
is registered once in the shared per-netns wildcard filter list, so
bcm_rx_handler() can run concurrently on different CPUs for frames
arriving on different net devices.
op->rx_stamp and op->rx_ifindex were written before bcm_rx_update_lock was
taken, allowing concurrent writers to race each other - including a torn
store of the 64-bit rx_stamp on 32-bit platforms.
Beyond a torn store bcm_send_to_user() must report the timestamp/ifindex
of the very same frame whose content it is delivering. So the assignment
is placed in the same unbroken bcm_rx_update_lock section as the content
comparison.
As a side effect, the RTR-request frame feature (which never reach
bcm_send_to_user()) no longer updates rx_stamp/rx_ifindex, since only
the notification path needs them.
Oliver Hartkopp [Tue, 14 Jul 2026 16:55:31 +0000 (18:55 +0200)]
can: bcm: fix stale rx/tx ops after device removal
RX: an RX_SETUP update(!) for an existing op skipped can_rx_register()
unconditionally, even when a concurrent NETDEV_UNREGISTER had already
torn down its registration (op->rx_reg_dev == NULL). This silently
did not re-enable frame delivery for that updated filter. bcm_rx_setup()
now re-registers in that case, while leaving rx_ops with ifindex = 0
(all CAN devices) which never carry a tracked rx_reg_dev registered as-is.
TX: bcm_notify() only handled bo->rx_ops on NETDEV_UNREGISTER, leaving
tx_ops with an active cyclic transmission re-arming its hrtimer
indefinitely to execute bcm_tx_timeout_handler(). Cancelling the hrtimer
prevents the runaway timer and any injection into a later reused ifindex,
since nothing else calls bcm_can_tx() for the op until an explicit
TX_SETUP update re-arms it.
Unlike bcm_rx_unreg(), which clears the tracked rx_reg_dev for rx_ops,
the ifindex is intentionally left unchanged for tx_ops. bcm_tx_setup()
always rejects ifindex 0, so clearing it would strand the op: neither a
later TX_SETUP (bcm_find_op()) nor TX_DELETE (bcm_delete_tx_op()) could
ever find it again, since both require an exact ifindex match.
Oliver Hartkopp [Tue, 14 Jul 2026 16:55:30 +0000 (18:55 +0200)]
can: bcm: add missing device refcount for CAN filter removal
sashiko-bot remarked a problem with a concurrent device unregistration
in isotp.c which also is present in the bcm.c code. A former fix for raw.c
commit c275a176e4b6 ("can: raw: add missing refcount for memory leak fix")
introduced a netdevice_tracker which solves the issue for bcm.c too.
bcm_release(), bcm_delete_rx_op() and bcm_notifier() relied on
dev_get_by_index(ifindex) to re-find the device for an rx_op before
unregistering its filter. If a concurrent NETDEV_UNREGISTER has already
unlisted the device from the ifindex table, that lookup fails and
can_rx_unregister() is silently skipped, leaving a stale CAN filter
pointing at the soon-to-be-freed bcm_op/socket.
Hold a netdev_hold()/netdev_put() tracked reference on op->rx_reg_dev
from the moment the rx filter is registered in bcm_rx_setup() until it
is unregistered in bcm_rx_unreg(), and use that reference directly in
bcm_release() and bcm_delete_rx_op() instead of re-looking the device
up by ifindex.
Oliver Hartkopp [Tue, 14 Jul 2026 16:55:29 +0000 (18:55 +0200)]
can: bcm: validate frame length in bcm_rx_setup() for RTR replies
bcm_tx_setup() validates cf->len against the CAN/CAN FD DLC limits
before installing frames for TX_SETUP, but bcm_rx_setup() never did
the same for the RTR-reply frame configured via RX_SETUP with
RX_RTR_FRAME.
Oliver Hartkopp [Tue, 14 Jul 2026 16:55:28 +0000 (18:55 +0200)]
can: bcm: extend bcm_tx_lock usage for data and timer updates
Stage new CAN frame content for an existing tx op into a kmalloc()'d
buffer and validate it there, mirroring the approach already used in
bcm_rx_setup(). Only copy the validated data into op->frames while
holding op->bcm_tx_lock, so bcm_can_tx() and bcm_tx_timeout_handler()
can no longer observe a partially updated or unvalidated frame.
Add a missing error path for memcpy_from_msg() when copying CAN frame
data from userspace.
Also move the kt_ival1/kt_ival2/ival1/ival2 updates in bcm_tx_setup()
under op->bcm_tx_lock, and read kt_ival1/kt_ival2/count under the same
lock in bcm_tx_set_expiry() and bcm_tx_timeout_handler(), closing the
torn 64-bit ktime_t read on 32-bit platforms.
Oliver Hartkopp [Tue, 14 Jul 2026 16:55:27 +0000 (18:55 +0200)]
can: bcm: add missing rcu list annotations and operations
sashiko-bot remarked the missing use of list_add_rcu() in
bcm_[rx|tx]_setup() to have a proper initialized bcm_op structure
when bcm_proc_show() traverses the bcm_op's under rcu_read_lock().
To cover all initial settings of the bcm_op's the list_add_rcu() calls
are moved to the end of the setup code.
While at it, also fix the mirroring removal side: bcm_release() called
bcm_remove_op() - which frees the op via call_rcu() - on ops that were
still linked in bo->tx_ops/bo->rx_ops, without list_del_rcu() first.
Unlink each op with list_del_rcu() before handing it to bcm_remove_op(),
matching the existing pattern in bcm_delete_tx_op()/bcm_delete_rx_op().
Reported-by: sashiko-reviews@lists.linux.dev Closes: https://lore.kernel.org/linux-can/20260610094654.A1FFE1F00893@smtp.kernel.org/ Fixes: dac5e6249159 ("can: bcm: add missing rcu read protection for procfs content") Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net> Link: https://patch.msgid.link/20260714-bcm_fixes-v15-5-562f7e3e42da@hartkopp.net Cc: stable@kernel.org Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Oliver Hartkopp [Tue, 14 Jul 2026 16:55:26 +0000 (18:55 +0200)]
can: bcm: fix CAN frame rx/tx statistics
KCSAN detected a data race within the bcm_rx_handler() when two CAN frames
have been simultaneously received and processed in a single rx op by two
different CPUs.
Use atomic operations with (signed) long data types to access the
statistics in the hot path to fix the KCSAN complaint.
Additionally simplify the update and check of statistics overflow by
using the atomic operations in separate bcm_update_[rx|tx]_stats()
functions. The rx variant runs under bcm_rx_update_lock to prevent
races when resetting the two rx counters; the tx variant runs under
bcm_tx_lock and only needs to guard its own counter's overflow.
As the rx path resets its values already at LONG_MAX / 100, there is
no conflict between the two locking domains (bcm_rx_update_lock vs.
bcm_tx_lock) even for ops that use both paths.
The rx statistics update and the frames_filtered update in
bcm_rx_changed() were previously performed in two separate
bcm_rx_update_lock sections. For an rx op subscribed on all interfaces
(ifindex == 0), bcm_rx_handler() can run concurrently on different
CPUs, so a counter reset by one CPU between these two sections could
leave frames_filtered larger than frames_abs on another CPU, producing
a bogus (even negative) reduction percentage in procfs. Update the
statistics in the same critical section as bcm_rx_changed() to close
this gap, which also removes the now unneeded extra lock/unlock pair
around the traffic_flags calculation.
Oliver Hartkopp [Tue, 14 Jul 2026 16:55:25 +0000 (18:55 +0200)]
can: bcm: add locking when updating filter and timer values
KCSAN detected a simultaneous access to timer values that can be
overwritten in bcm_rx_setup() when updating timer and filter content
while bcm_rx_handler(), bcm_rx_timeout_handler() or bcm_rx_thr_handler()
run concurrently on incoming CAN traffic.
Protect the timer (ival1/ival2/kt_ival1/kt_ival2/kt_lastmsg) and filter
(nframes/flags/frames/last_frames) updates in bcm_rx_setup() with a new
per-op bcm_rx_update_lock, taken with the matching scope in the RX
handlers. memcpy_from_msg() is staged into a temporary buffer before the
lock is taken, since it can sleep and must not run under a spinlock.
hrtimer_cancel() is always called without bcm_rx_update_lock held, since
bcm_rx_timeout_handler()/bcm_rx_thr_handler() take the same lock and a
running callback would otherwise deadlock against the canceller.
Also close a related race: bcm_rx_setup() cleared the RTR flag in the
stored reply frame's can_id as a separate, unprotected step after the
frame content was already installed, so a concurrent bcm_rx_handler()
could transmit a stale reply with CAN_RTR_FLAG still set. Fold that
normalization into the initial frame preparation instead (on the staged
buffer for updates, directly on op->frames pre-registration for new
ops), so the installed frame is always atomically self-consistent.
bcm_rx_handler()'s RX_RTR_FRAME check now takes a lock-protected
snapshot of op->flags before deciding whether to call bcm_can_tx(),
but does not hold the lock across that call.
Also take a lock-protected snapshot of the currframe in bcm_can_tx()
to avoid partly overwrites by content updates in bcm_tx_setup().
Finally check if a TX_RESET_MULTI_IDX/SETTIMER might have reset
op->currframe between the two locked sections in bcm_can_tx().
Omit calling hrtimer_forward() with zero interval in bcm_rx_thr_handler().
kt_ival2 may have been concurrently cleared by bcm_rx_setup() before it
cancels this timer, so check kt_ival2 inside the bcm_rx_update_lock.
Fixes: c2aba69d0c36 ("can: bcm: add locking for bcm_op runtime updates") Reported-by: syzbot+75e5e4ae00c3b4bb544e@syzkaller.appspotmail.com Closes: https://lore.kernel.org/linux-can/6975d5cf.a00a0220.33ccc7.0022.GAE@google.com/ Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net> Link: https://patch.msgid.link/20260714-bcm_fixes-v15-3-562f7e3e42da@hartkopp.net Cc: stable@kernel.org Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Oliver Hartkopp [Tue, 14 Jul 2026 16:55:24 +0000 (18:55 +0200)]
can: bcm: fix lockless bound/ifindex race and silent RX_SETUP failure
bcm_sendmsg() reads bo->ifindex and checks bo->bound before taking
lock_sock(), while bcm_notify(), bcm_connect() and bcm_release() all
mutate both fields under that same lock. Because the lockless reads
and the locked writes are unordered with respect to each other, a
racing bcm_notify() (device unregister) or bcm_connect() (concurrent
bind on another thread sharing the socket) can make bcm_sendmsg()
observe an inconsistent combination, e.g. a stale bound=1 together
with the now-cleared ifindex=0, silently turning a socket bound to a
specific CAN interface into one that also matches "any" interface.
Keep the lockless bo->bound check purely as a fast-path reject, and
move the ifindex read (and a bo->bound re-check) into the locked
section, where every writer already serializes. This removes the
possibility of observing the two fields torn against each other,
rather than trying to fix it with more READ_ONCE()/WRITE_ONCE() pairs
on two independently updated fields. Annotate the now-purely-lockless
bo->bound accesses consistently across all its write sites.
Also fix bcm_rx_setup() silently returning success when the target
device disappears concurrently instead of reporting -ENODEV, so a
broken RX op is no longer left registered as if it had succeeded.
Lee Jones [Tue, 14 Jul 2026 16:55:23 +0000 (18:55 +0200)]
can: bcm: defer rx_op deallocation to workqueue to fix thrtimer UAF
Commit f1b4e32aca08 ("can: bcm: use call_rcu() instead of costly
synchronize_rcu()") replaced synchronize_rcu() in bcm_delete_rx_op()
with call_rcu() and introduced the RX_NO_AUTOTIMER flag.
However, this flag check was omitted for thrtimer in the packet rx
fast-path. During BCM RX operation teardown, a concurrent RCU reader
(bcm_rx_handler) can race and re-arm thrtimer via
bcm_rx_update_and_send() after call_rcu() has been scheduled. Once
the RCU grace period elapses, bcm_op is freed. The subsequently
firing thrtimer then dereferences the deallocated op, causing a UAF.
Adding flag checks to the rx fast-path (bcm_rx_update_and_send) does not
fully close the TOCTOU race and introduces latency for every CAN frame.
Conversely, calling hrtimer_cancel() directly inside the RCU callback
(softirq context) is fatal as hrtimer_cancel() can sleep, triggering
a "scheduling while atomic" panic.
Resolve this by deferring the timer cancellation and memory free to a
dedicated unbound workqueue (bcm_wq). The RCU callback now queues a
work item to bcm_wq, which safely cancels both timers and deallocates
memory in sleepable process context. A dedicated workqueue is used to
prevent system-wide WQ saturation and is cleanly flushed/destroyed
on module unload to avoid rmmod page faults.
Since the deferred work can now outlive the calling context by an
unbounded amount, also take a reference on op->sk when it is assigned
and drop it only once the deferred work has cancelled both timers, so a
socket can no longer be freed out from under a still-armed timer whose
callback (bcm_send_to_user()) dereferences op->sk.
Fixes: f1b4e32aca08 ("can: bcm: use call_rcu() instead of costly synchronize_rcu()") Tested-by: Feng Xue <feng.xue@outlook.com> Tested-by: Oliver Hartkopp <socketcan@hartkopp.net> Signed-off-by: Lee Jones <lee@kernel.org> Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net> Link: https://patch.msgid.link/20260714-bcm_fixes-v15-1-562f7e3e42da@hartkopp.net Cc: stable@kernel.org Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Stéphane Grosjean [Thu, 7 May 2026 08:22:27 +0000 (10:22 +0200)]
can: peak: Modification of references to email accounts being deleted
Following the sale of PEAK-System France by HMS-Networks, this update is
intended to change all my @hms-networks.com email addresses to my new
@peak-system.fr address.
Shuhao Fu [Thu, 7 May 2026 08:22:26 +0000 (10:22 +0200)]
can: j1939: fix lockless local-destination check
j1939_priv.ents[].nusers is documented as protected by priv->lock, and
its updates already happen under that lock. j1939_can_recv() also reads
it under read_lock_bh(). However, j1939_session_skb_queue() and
j1939_tp_send() still read priv->ents[da].nusers without taking the
lock.
Those transport-side checks decide whether to set J1939_ECU_LOCAL_DST, so
they can race with j1939_local_ecu_get() and j1939_local_ecu_put() while
userspace is binding or releasing sockets concurrently with TP traffic.
This can misclassify TP/ETP sessions as local or remote and take the wrong
transport path.
Fix both transport paths by routing the destination-locality check through
a helper that reads ents[].nusers under read_lock_bh(&priv->lock).
Fixes: 9d71dd0c7009 ("can: add support of SAE J1939 protocol") Signed-off-by: Shuhao Fu <sfual@cse.ust.hk> Tested-by: Oleksij Rempel <o.rempel@pengutronix.de> Acked-by: Oleksij Rempel <o.rempel@pengutronix.de> Link: https://patch.msgid.link/20260419140614.GA4041240@chcpu16 Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
KVM: SVM: Bump asid_generation on CPU online to avoid ASID collision after hotplug
If a vCPU stays scheduled out (or blocked) while the last pCPU it ran
on goes through a hotplug cycle (online->offline->online), and the vCPU
then resumes execution on the same pCPU, then it is possible for it to
run with an ASID that has now been assigned to a different vCPU,
resulting in stale TLB translations being used.
svm_enable_virtualization_cpu() resets asid_generation to 1 and sets
next_asid to max_asid + 1 on every CPU online event, including hotplug
cycles. Because next_asid starts beyond the pool boundary, the first
call to new_asid() after an online event always wraps the pool,
incrementing asid_generation to 2 and assigning ASIDs starting from
min_asid.
Consider two vCPUs from different VMs, vCPU-A pinned to CPU-X holding
asid_generation=2 and ASID=N from before the hotplug event:
1. CPU-X goes offline and back online: asid_generation resets to 1,
next_asid = max_asid + 1.
2. One or more vCPUs migrate to CPU-X and call new_asid(), wrapping
the pool and consuming ASIDs starting from min_asid. Eventually
vCPU-B from a different VM is assigned asid_generation=2, ASID=N
— the same ASID that vCPU-A held before the hotplug.
3. vCPU-A enters pre_svm_run() on CPU-X: current_vmcb->cpu is
unchanged so the migration branch is skipped. Its saved
asid_generation=2 matches sd->asid_generation=2, so the generation
check silently passes and vCPU-A continues running with ASID=N —
the same ASID just freshly assigned to vCPU-B.
Both vCPUs from different VMs now run on CPU-X with the same ASID,
causing them to share NPT TLB entries and producing stale translations.
The collision manifests as a KVM internal error (Suberror: 1, emulation
failure). The NPT page fault reports a faulting GPA far outside the
VM's physical memory range — a sign of stale TLB translations being
used. KVM falls back to instruction emulation, which fails on
FPU/XSave instructions (XRSTOR, STMXCSR) that the emulator does not
implement.
Fix this by incrementing asid_generation instead of resetting it to 1
in svm_enable_virtualization_cpu(). On module load, asid_generation
starts at 0 (memset) and the increment produces 1, identical to the
old behaviour. On subsequent hotplug cycles the generation advances
beyond any value a vCPU previously observed on this CPU, so the
generation check in pre_svm_run() reliably forces new_asid() on every
vCPU after every hotplug cycle.
Fixes: 774c47f1d78e ("[PATCH] KVM: cpu hotplug support") Reported-by: Chandrakanth Silveru <Chandrakanth.Silveru@amd.com> Tested-by: Srikanth Aithal <Srikanth.Aithal@amd.com> Reviewed-by: K Prateek Nayak <kprateek.nayak@amd.com> Reviewed-by: Tom Lendacky <thomas.lendacky@amd.com> Signed-off-by: Nikunj A Dadhania <nikunj@amd.com>
Message-ID: <20260715063506.672432-1-nikunj@amd.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Merge tag 'linux_kselftest-kunit-fixes-7.2-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest
Pull kunit fix from Shuah Khan:
"Fix warning suppressions with kunit built as module:
CONFIG_KUNIT is a tristate symbol but the warning suppression code in
lib/bug.c is only built if it's built-in due to it using a plain
#ifdef, rendering warning suppressions broken for kunit build as
loadable module.
kunit_is_suppressed_warning() already has a stub for when kunit is
disabled so drop that guard entirely"
* tag 'linux_kselftest-kunit-fixes-7.2-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest:
bug: fix warning suppressions with kunit built as module
Merge tag 'linux_kselftest-fixes-7.2-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest
Pull kselftest fixes from Shuah Khan:
- Fix ftrace reading enabled_func test in add_remove_fprobe_module test
- Fix tracing trigger-hist-poll.tc to use sched_process_exit
* tag 'linux_kselftest-fixes-7.2-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest:
selftests/tracing: Have trigger-hist-poll.tc use sched_process_exit
selftests/ftrace: Fix reading enabled_functions in add_remove_fprobe_module test
block: do not warn when doing greedy allocation in folio_alloc_greedy()
During one of my local btrfs fstests runs, folio_alloc() inside
folio_alloc_greedy() triggered an allocation failure report when trying
to allocate an order-4 folio.
The kernel is from the latest development branch, which is utilizing
the IOMAP_DIO_BOUNCE flag for direct writes when the inode requires
checksum.
Unfortunately I didn't save the full log, only the function and the
order.
When the IOMAP_DIO_BOUNCE flag is utilized, we will hit the following call
chain:
Michael Bommarito [Tue, 14 Jul 2026 11:48:06 +0000 (07:48 -0400)]
partitions: aix: bound the lvd scan to one sector
aix_partition() reads the logical-volume descriptor array as a single
sector and then scans it:
if (numlvs && (d = read_part_sector(state, vgda_sector + 1, §))) {
struct lvd *p = (struct lvd *)d;
...
for (i = 0; foundlvs < numlvs && i < state->limit; i++) {
lvip[i].pps_per_lv = be16_to_cpu(p[i].num_lps);
p points at a single 512-byte sector, which holds SECTOR_SIZE /
sizeof(struct lvd) = 16 entries, but the loop runs until foundlvs reaches
the on-disk numlvs or i reaches state->limit (DISK_MAX_PARTS, 256).
numlvs is an on-disk __be16 read straight from the volume group
descriptor and is not validated, so a crafted AIX image with numlvs
larger than 16 and lvd entries whose num_lps fields are zero (so foundlvs
never advances) drives the loop to read p[i] well past the end of the
read sector buffer.
Commit d97a86c170b4 ("partitions: aix.c: off by one bug") hardened the
matching write of lvip[lv_ix] in 2014 but left this read loop unbounded.
Bound the scan to the number of struct lvd entries that fit in the
sector that was actually read.
Tao Cui [Wed, 15 Jul 2026 13:24:07 +0000 (21:24 +0800)]
blk-cgroup: fix leaks and online flag on radix_tree_insert failure
When radix_tree_insert() fails in blkg_create(), the error path has two
issues:
1. blkg->online is set to true unconditionally, even when the blkg was
never fully inserted. Move the assignment inside the success block.
2. The error path calls blkg_put() without first calling
percpu_ref_kill(). Because the refcount is still in percpu mode,
percpu_ref_put() only does this_cpu_sub() without checking for zero,
so blkg_release() is never triggered. This permanently leaks the
blkg memory, its percpu iostat, policy data, the parent blkg
reference, and the cgroup css reference — the latter preventing the
cgroup from ever being destroyed.
Fix by replacing blkg_put() with percpu_ref_kill(), matching the pattern
used in blkg_destroy().
Commit 267ec4d7223a ("loop: fix partition scan race between udev and
loop_reread_partitions()") stopped disk_force_media_change() from
setting GD_NEED_PART_SCAN because loop devices with LO_FLAGS_PARTSCAN
rescan partitions explicitly. However, partitions can also be added
manually with BLKPG while LO_FLAGS_PARTSCAN is clear.
When such a loop device is detached, __loop_clr_fd() skips
bdev_disk_changed(). Without GD_NEED_PART_SCAN, reopening the unbound
device no longer performs the previous lazy cleanup, leaving dead
partition devices behind. A subsequent LOOP_CONFIGURE can then fail its
partition scan with -EBUSY, as seen in blktests loop/009 after loop/008.
Call bdev_disk_changed() unconditionally during __loop_clr_fd(). The
disk capacity is already zero and the release path holds open_mutex, so
this drops all partitions without rescanning the detached backing file.
The new blktests loop/013 case covers this sequence by adding a partition
with BLKPG without LO_FLAGS_PARTSCAN, detaching the loop device, and
checking that the partition is gone when the device is reopened.
Fixes: 267ec4d7223a ("loop: fix partition scan race between udev and loop_reread_partitions()") Reported-by: kernel test robot <oliver.sang@intel.com> Closes: https://lore.kernel.org/oe-lkp/202607150754.b660f5b9-lkp@intel.com Signed-off-by: Daan De Meyer <daan@amutable.com> Link: https://patch.msgid.link/20260715-b4-loop-partition-cleanup-v1-1-b9f59910cd1e@amutable.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
This is problematic when the compiler emits the final
"return plug->cur_ktime" as a reload from memory, after PF_BLOCK_TS has
already been set.
Since the cached timestamp is now invalidated from finish_task_switch()
(fad156c2af22 "block: invalidate cached plug timestamp after task
switch"), a task preempted between setting PF_BLOCK_TS and that reload
has plug->cur_ktime zeroed by blk_plug_invalidate_ts() when it is
scheduled back in. The reload then returns 0.
A 0 handed back here is stored as a start timestamp -- e.g.
blk_account_io_start() writes it to rq->start_time_ns -- and later
subtracted from "now". blk_account_io_done() then adds (now - 0), i.e.
roughly the system uptime, to the per-group nsecs[] counters. On an
otherwise idle, healthy device this appears as sudden ~uptime-sized jumps
in the diskstats time fields (write_ticks/discard_ticks/time_in_queue).
The solution is to be explicit in our reads and writes to this field
that is preemption volatile. We also add a barrier() to ensure that any
setting of PF_BLOCK_TS is ordered to happen after the cur_ktime update.
This issue was discovered using AI-assisted kprobes looking for paths
that were leaking zeroed timestamps in a live system, based on the
observation that we were sometimes seeing uptime-sized jumps in kernel
exported counters. This was flagged by NodeDiskIOSaturation
prometheus alerts that started firing on all hosts post 7.1.3 kernel
upgrade, due to node-exporter now exporting a nonsensical
node_disk_io_time_weighted_seconds_total.
Merge tag 'scmi-ffa-fixes-7.2' of https://git.kernel.org/pub/scm/linux/kernel/git/sudeep.holla/linux into arm/fixes
Arm SCMI/FF-A fixes for v7.2
Fix two runtime issues in the SCMI framework. Use full 64-bit division
when rounding range-based clock rates, avoiding divisor truncation and
a possible divide-by-zero on 32-bit systems. Rate-limit notification
queue-full warnings emitted from interrupt context to prevent printk
floods and prolonged system stalls during notification bursts. Also
correct a grammar error in the ARM_SCMI_POWER_CONTROL Kconfig help
text.
Fix the FF-A driver RX/TX buffer sizing logic to respect the maximum
buffer size advertised by firmware, while retaining compatibility with
older implementations that may reject PAGE_SIZE-rounded buffers.
Also fix a NULL pointer dereference in ffa_partition_info_get() by
rejecting NULL UUID strings before passing them to uuid_parse().
* tag 'scmi-ffa-fixes-7.2' of https://git.kernel.org/pub/scm/linux/kernel/git/sudeep.holla/linux:
firmware: arm_scmi: Rate-limit queue-full warnings in IRQ context
firmware: arm_scmi: Use 64-bit division for clock rate rounding
firmware: arm_scmi: Grammar s/may needed/may be needed/
firmware: arm_ffa: Fix NULL dereference in ffa_partition_info_get()
firmware: arm_ffa: Respect firmware advertised RX/TX buffer size limits
Mark Harris [Tue, 14 Jul 2026 00:30:56 +0000 (17:30 -0700)]
riscv: hwprobe: Avoid uninitialized read in hwprobe_get_cpus()
When cpusetsize < cpumask_size(), hwprobe_get_cpus() did not fully
initialize its copy of the cpu mask, which could cause non-deterministic
results from the riscv_hwprobe syscall on a system with more than 8 CPUs
when the supplied cpu mask is empty. Address this by fully initializing
the cpu mask.
Fixes: e178bf146e4b ("RISC-V: hwprobe: Introduce which-cpus flag") Signed-off-by: Mark Harris <mark.hsj@gmail.com> Reviewed-by: Nam Cao <namcao@linutronix.de> Reviewed-by: Michael Ellerman <mpe@kernel.org> Link: https://patch.msgid.link/20260714003056.73707-1-mark.hsj@gmail.com Signed-off-by: Paul Walmsley <pjw@kernel.org>
s390/checksum: Fix csum_partial() without vector facility
Currently csum_partial() calls csum_copy() with copy=false and dst=NULL.
On machines without the vector facility, csum_copy() falls back to
cksm(dst, ...), causing the checksum to be calculated from address zero
instead of the source buffer.
The VX implementation already checksums data loaded from src. Make the
fallback do the same by passing src to cksm().
accel/ivpu: Reject firmware log with size smaller than header
fw_log_from_bo() validates the tracing buffer header_size and that the
log fits within the BO, but never checks that log->size is at least
log->header_size. fw_log_print_buffer() then computes:
u32 data_size = log->size - log->header_size;
which underflows to a near-U32_MAX value when firmware reports a log whose
size is smaller than its header. That huge data_size defeats the
log_start/log_end bounds clamps added by commit dd1311bcf0e6 ("accel/ivpu:
Add bounds checks for firmware log indices"), so fw_log_print_lines() reads
far past the small real data region of the BO. A size of 0 also makes
fw_log_from_bo() advance the offset by 0, causing the callers to loop
forever on the same header.
Reject logs whose size is smaller than the header (which also rejects
size == 0).
KVM: riscv: Fix Spectre-v1 in vector register access
User-controlled register indices from the ONE_REG ioctl are used to
index into the vector register buffer (v0..v31). Sanitize the calculated
offset with array_index_nospec() to prevent speculative out-of-bounds
access.
drmm_mutex_init() can fail while registering the managed cleanup action.
When that happens, drmm_add_action_or_reset() destroys the mutex before
returning the error. Continuing initialization would therefore leave the
debugfs GEM object list with an unusable lock.
Propagate the error as is already done for the other managed mutexes in
panthor_device_init().
Fixes: a3707f53eb3f ("drm/panthor: show device-wide list of DRM GEM objects over DebugFS") Signed-off-by: Linmao Li <lilinmao@kylinos.cn> Reviewed-by: Liviu Dudau <liviu.dudau@arm.com> Link: https://patch.msgid.link/20260713082912.321021-1-lilinmao@kylinos.cn Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>
panthor_fw_load() detects truncated firmware images, but jumps to the
common cleanup path without setting ret. If no previous error was recorded,
the function can return 0 and treat the invalid firmware as successfully
loaded.
Set ret to -EINVAL before leaving the truncated-image path.
Sean Christopherson [Tue, 14 Jul 2026 16:03:06 +0000 (09:03 -0700)]
KVM: nVMX: Put vmcs12 pages if nested VM-Enter fails due to invalid guest state
Put all vmcs12 pages if KVM synthesizes a nested VM-Exit due to invalid
guest while emulating VMLAUNCH or VMRESUME. The invalid guest state path
doesn't use nested_vmx_vmexit() as that API is intended to be used if and
only if L2 is active, and the open coded equivalent neglects to put the
vmcs12 pages. Failure to put the vmcs12 pages leaks any pinned pages
(and/or mappings) if L1 retries VMLAUNCH/VMRESUME.
Note, the !from_vmenter scenario doesn't suffer the same problem, as
vmx_get_nested_state_pages() only gets/pins/maps the vmcs12 pages if L2 is
active, i.e. if a "full" VM-Exit is guaranteed before KVM will retry
getting vmcs12 pages.
Fixes: 96c66e87deee ("KVM/nVMX: Use kvm_vcpu_map when mapping the virtual APIC page") Fixes: 3278e0492554 ("KVM/nVMX: Use kvm_vcpu_map when mapping the posted interrupt descriptor table") Fixes: fe1911aa443e ("KVM: nVMX: Use kvm_vcpu_map() to get/pin vmcs12's APIC-access page") Reported-by: Minh Nguyen <minhnguyen.080505@gmail.com> Cc: stable@vger.kernel.org Signed-off-by: Sean Christopherson <seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Paolo Bonzini [Wed, 15 Jul 2026 10:12:37 +0000 (12:12 +0200)]
Merge tag 'kvmarm-fixes-7.2-1' of git://git.kernel.org/pub/scm/linux/kernel/git/kvmarm/kvmarm into HEAD
KVM/arm64 fixes for 7.2, take #1
- Fix an accounting buglet when reclaiming pages from a protected
guest
- Fix a bunch of architectural compliance issues when injecting a
synthesised exception, most of which were missing the PSTATE.IL bit
indicating a 32bit-wide instruction
- Another set of fixes addressing issues with translation of VNCR_EL2,
including corner cases where the guest point that register at a RO
page...
- Don't warn when trapping accesses to ZCR_EL2 from an L2 guest, as
that's not unexpected at all
- Address a bunch of races with LPI migration vs LPIs being disabled
- Fix a total howler of a bug combining FEAT_MOPS and NV, resulting in
exception returning in the wrong place...
- Coerce Fuad Tabba into a reviewer role, and may his Inbox catch
fire!
====================
bpf: Reject negative const offsets for buffer pointers
Reject negative effective offsets for PTR_TO_TP_BUFFER and PTR_TO_BUF
accesses. Calculate the effective access start using signed arithmetic
to prevent unsigned access-end accounting from wrapping, and cover both
load-time rejection and the raw tracepoint writable attach-time path.
---
Changes in v5:
- Simplify __check_buffer_access() to reject a negative effective start
after confirming that var_off is constant. Validate the combined
offset instead of rejecting negative instruction offsets separately.
Drop the duplicate BPF_MAX_VAR_OFF check because pointer arithmetic
already bounds constant offsets, and remove the redundant size < 0
check.
- Switch the raw tracepoint writable attach tests from nbd_send_request
to bpf_testmod_test_writable_bare_tp, avoiding the NBD configuration
dependency and its false-pass condition.
- Split the attach coverage into named subtests and require
bpf_raw_tracepoint_open() to return -EINVAL.
- Add verifier coverage for a negative constant PTR_TO_BUF offset.
Changes in v4:
- Correct the Fixes tag to point to 022ac0750883, where pointer offsets
were folded into reg->var_off.
- Drop the end > U32_MAX check, which is unreachable after bounding const
var_off with BPF_MAX_VAR_OFF while keeping instruction offsets and
access sizes bounded.
Changes in v3:
- Check constant var_off against +/-BPF_MAX_VAR_OFF before computing
the effective access range, matching the existing verifier pointer
offset convention.
- Keep explicit rejection of negative instruction offsets and keep
bounded negative constant var_off valid when the effective offset is
non-negative.
Changes in v2:
- Split the kernel fix and selftests into separate patches.
- Add an attach-time raw tracepoint writable test that exercises
max_tp_access against nbd_send_request's writable size.
- Adjust selftest formatting to use the 100 character line width.
Add verifier coverage for constant negative offsets on PTR_TO_TP_BUFFER
and PTR_TO_BUF pointers. Both programs adjust the buffer pointer by -8
and access it at offset zero, so the negative effective start must be
rejected at load time.
Switch the raw tracepoint writable attach checks from nbd_send_request
to bpf_testmod_test_writable_bare_tp, avoiding a dependency on the NBD
tracepoint. Keep the existing past-end case and add a case with a
negative var_off compensated by a positive instruction offset. The
effective start remains non-negative, so the program loads, but its
access end exceeds the writable context size and
bpf_raw_tracepoint_open() must return -EINVAL.
Sun Jian [Tue, 14 Jul 2026 09:38:45 +0000 (02:38 -0700)]
bpf: Reject negative const offsets for buffer pointers
The verifier rejects variable offsets for PTR_TO_TP_BUFFER and PTR_TO_BUF
accesses, but it currently accepts a constant negative offset produced by
pointer arithmetic.
Commit 022ac0750883 ("bpf: use reg->var_off instead of reg->off for
pointers") moved constant pointer offsets from reg->off to reg->var_off.
However, __check_buffer_access() continued to check only the instruction
offset. An access with reg->var_off equal to -8 and an instruction offset
of zero therefore passes verification.
For writable raw tracepoints, the access end is also calculated from the
unsigned reg->var_off.value. An eight-byte access starting at -8 wraps
the calculated end to zero, allowing the program to load and attach
without increasing max_tp_access.
After ensuring that reg->var_off is constant, calculate the effective
access start using signed arithmetic and reject it when it is negative.
Use the validated start to calculate the access end for both
PTR_TO_TP_BUFFER and PTR_TO_BUF.
Fixes: 022ac0750883 ("bpf: use reg->var_off instead of reg->off for pointers") Signed-off-by: Sun Jian <sun.jian.kdev@gmail.com> Acked-by: Shung-Hsi Yu <shung-hsi.yu@suse.com> Cc: stable@vger.kernel.org # 5.2.0 Link: https://patch.msgid.link/20260714093846.18159-2-sun.jian.kdev@gmail.com Signed-off-by: Eduard Zingerman <eddyz87@gmail.com>
====================
bpf, sockmap: Fix FIONREAD for sockets without a verdict program
Sockets added to a sockmap/sockhash with no stream/skb verdict program
attached answer FIONREAD with 0 even when unread data is pending in
sk_receive_queue. Fix tcp_bpf_ioctl() to account for the receive queue
in that case, and add a selftest.
Changes in v3:
- Remove unused sk_psock_msg_inq()
- Link to v2: https://patch.msgid.link/20260708-fionread-no-verdict-v2-0-29dd293621c7@coralogix.com
Changes in v2:
- Split the fix and the selftest into separate patches
- Use READ_ONCE() to read the verdict program pointers
- Link to v1: https://patch.msgid.link/20260707-fionread-no-verdict-v1-1-ce94a72357ec@coralogix.com
bpf, sockmap: Account for receive queue in FIONREAD without a verdict program
tcp_bpf_ioctl() answers SIOCINQ from psock->msg_tot_len, which only
counts bytes in ingress_msg. Without a stream/skb verdict program
nothing is diverted there: data stays in sk_receive_queue, so FIONREAD
returns 0 even though read() returns data.
Add tcp_inq() to the reported value when the psock has no verdict
program. The two queues are disjoint, so bytes redirected into
ingress_msg from other sockets stay correctly accounted through
msg_tot_len.
Luke Wang [Wed, 15 Jul 2026 07:18:18 +0000 (15:18 +0800)]
mmc: sdhci-esdhc-imx: fix resume error handling
Check pm_runtime_force_resume() return value in resume. If it fails
(clock enable failure), return immediately since accessing hardware
registers on an unclocked device would cause a kernel panic.
The early return intentionally skips enable_irq() and
sdhci_disable_irq_wakeups() because the IRQ handler reads
SDHCI_INT_STATUS, which would also fault without clocks. The PM runtime
usage counter leak only affects this already-broken device instance and
is an acceptable tradeoff to preserve system stability.
Remove the return value check for mmc_gpio_set_cd_wake(host->mmc, false)
since disable_irq_wake() called internally always returns 0.
Also return 0 explicitly on the success path instead of propagating
stale return values.
Fixes: 676a83855614 ("mmc: host: sdhci-esdhc-imx: refactor the system PM logic") Acked-by: Adrian Hunter <adrian.hunter@intel.com> Reviewed-by: Frank Li <Frank.Li@nxp.com> Signed-off-by: Luke Wang <ziniu.wang_1@nxp.com> Cc: stable@vger.kernel.org Signed-off-by: Ulf Hansson <ulfh@kernel.org>
Luke Wang [Wed, 15 Jul 2026 07:18:17 +0000 (15:18 +0800)]
mmc: sdhci-esdhc-imx: make non-fatal errors non-blocking in suspend
Make pinctrl_pm_select_sleep_state() and mmc_gpio_set_cd_wake() failures
non-fatal in the suspend path. These failures only mean slightly higher
power consumption or missing CD wakeup capability, but should not block
system suspend.
Also change the function to always return 0 on the success path instead
of propagating non-fatal warning return values.
Fixes: 676a83855614 ("mmc: host: sdhci-esdhc-imx: refactor the system PM logic") Acked-by: Adrian Hunter <adrian.hunter@intel.com> Signed-off-by: Luke Wang <ziniu.wang_1@nxp.com> Cc: stable@vger.kernel.org Signed-off-by: Ulf Hansson <ulfh@kernel.org>
Luke Wang [Wed, 15 Jul 2026 07:18:16 +0000 (15:18 +0800)]
mmc: sdhci-esdhc-imx: use pm_runtime_resume_and_get() in suspend
Replace pm_runtime_get_sync() with pm_runtime_resume_and_get() to
simplify error handling. pm_runtime_resume_and_get() automatically
drops the usage counter on failure, avoiding the need for a separate
pm_runtime_put_noidle() call. If it fails, the device is unclocked and
accessing hardware registers would cause a kernel panic, so return the
error immediately.
Fixes: 676a83855614 ("mmc: host: sdhci-esdhc-imx: refactor the system PM logic") Acked-by: Adrian Hunter <adrian.hunter@intel.com> Signed-off-by: Luke Wang <ziniu.wang_1@nxp.com> Cc: stable@vger.kernel.org Signed-off-by: Ulf Hansson <ulfh@kernel.org>
Luke Wang [Wed, 15 Jul 2026 07:18:15 +0000 (15:18 +0800)]
mmc: sdhci-esdhc-imx: disable irq during suspend to fix unhandled interrupt
When using WIFI out-of-band wakeup, an "irq xxx: nobody cared" warning
occurs. This happens because the usdhc interrupt is not disabled during
system suspend when device_may_wakeup() returns false.
The sequence of events leading to this issue:
1. System enters suspend without disabling usdhc interrupt
(because device_may_wakeup() returns false for usdhc device)
2. WIFI out-of-band wakeup triggers system resume via GPIO interrupt
3. WIFI sends a Card interrupt before usdhc has fully resumed
4. usdhc is still in runtime suspend state and cannot handle the
interrupt properly
5. The unhandled interrupt triggers "nobody cared" warning
Fix this by unconditionally disabling the usdhc interrupt during suspend
and re-enabling it during resume, regardless of the wakeup capability.
This ensures no interrupts are processed during the suspend/resume
transition.
Fixes: 676a83855614 ("mmc: host: sdhci-esdhc-imx: refactor the system PM logic") Acked-by: Adrian Hunter <adrian.hunter@intel.com> Reviewed-by: Haibo Chen <haibo.chen@nxp.com> Signed-off-by: Luke Wang <ziniu.wang_1@nxp.com> Cc: stable@vger.kernel.org Signed-off-by: Ulf Hansson <ulfh@kernel.org>
Luke Wang [Wed, 15 Jul 2026 07:18:14 +0000 (15:18 +0800)]
mmc: sdhci-esdhc-imx: restore pinctrl before restoring ios timing on resume
SDIO devices such as WiFi may keep power during suspend, so the MMC
core skips full card re-initialization on resume and directly restores
the host controller's ios timing to match the card. For DDR mode,
pm_runtime_force_resume() sets DDR_EN before the pin configuration is
restored from sleep state.
This is related to the SoC IP integration: switching pinctrl setting
(changing alt from GPIO to USDHC) impacts the internal loopback path.
If pinctrl configures the pad to GPIO function, once DDR_EN is set, the
DLL delay will be fixed based on the GPIO function loopback path. When
the pinctrl is later changed to USDHC function, the internal loopback
path changes, making the original fixed sample point no longer suitable
for the current loopback path. This causes persistent read CRC errors on
subsequent data transfers.
SD/eMMC running in DDR mode are unaffected as they are fully
re-initialized from legacy timing after resume.
Fix this by restoring the pinctrl state based on current timing mode
using esdhc_change_pinstate() before pm_runtime_force_resume(). This
ensures the correct pin configuration (e.g., 100/200MHz for UHS modes)
is applied before DDR_EN is set. Only restore for non-wakeup devices
since wakeup devices kept their active pin state during suspend.
Fixes: 676a83855614 ("mmc: host: sdhci-esdhc-imx: refactor the system PM logic") Acked-by: Adrian Hunter <adrian.hunter@intel.com> Reviewed-by: Haibo Chen <haibo.chen@nxp.com> Signed-off-by: Luke Wang <ziniu.wang_1@nxp.com> Cc: stable@vger.kernel.org Signed-off-by: Ulf Hansson <ulfh@kernel.org>
Luke Wang [Wed, 15 Jul 2026 07:18:13 +0000 (15:18 +0800)]
mmc: sdhci-esdhc-imx: fix esdhc_change_pinstate() to allow default state restore
esdhc_change_pinstate() checks for pins_100mhz and pins_200mhz at the
top of the function and returns -EINVAL if either is not defined. This
prevents the default case from ever being reached, which means devices
with a sleep pinctrl state but without high-speed pin states (100mhz/
200mhz) can never restore their default pin configuration.
Move the IS_ERR checks for pins_100mhz and pins_200mhz into their
respective switch cases.
Fixes: 676a83855614 ("mmc: host: sdhci-esdhc-imx: refactor the system PM logic") Acked-by: Adrian Hunter <adrian.hunter@intel.com> Reviewed-by: Frank Li <Frank.Li@nxp.com> Signed-off-by: Luke Wang <ziniu.wang_1@nxp.com> Cc: stable@vger.kernel.org Signed-off-by: Ulf Hansson <ulfh@kernel.org>
Luke Wang [Wed, 15 Jul 2026 07:18:12 +0000 (15:18 +0800)]
mmc: sdhci-esdhc-imx: restore DLL override for DDR modes on resume
sdhci_esdhc_imx_hwinit() unconditionally clears ESDHC_DLL_CTRL by
writing zero. For SDIO devices that keep power during system suspend
and operate in DDR mode, the card remains in DDR timing while the host
DLL override configuration is lost.
Extract the DLL override setup from esdhc_set_uhs_signaling() into
a helper esdhc_set_dll_override(), and call it on the resume path
when the card kept power and is using a DDR timing mode.
Fixes: 676a83855614 ("mmc: host: sdhci-esdhc-imx: refactor the system PM logic") Acked-by: Adrian Hunter <adrian.hunter@intel.com> Reviewed-by: Frank Li <Frank.Li@nxp.com> Reviewed-by: Haibo Chen <haibo.chen@nxp.com> Signed-off-by: Luke Wang <ziniu.wang_1@nxp.com> Cc: stable@vger.kernel.org Signed-off-by: Ulf Hansson <ulfh@kernel.org>
Luke Wang [Wed, 15 Jul 2026 07:18:11 +0000 (15:18 +0800)]
mmc: sdhci-esdhc-imx: remove unnecessary mmc_card_wake_sdio_irq check for tuning save/restore
The tuning save/restore during system PM is conditioned on
mmc_card_wake_sdio_irq(), but this check is unrelated to whether
tuning values need to be preserved. The actual requirement is that
the card keeps power during suspend and the controller is a uSDHC.
SDIO devices using out-of-band GPIO wakeup maintain power during
suspend but do not set the SDIO IRQ wake flag. In this case the
tuning delay values are not saved/restored.
Remove the unnecessary mmc_card_wake_sdio_irq() condition from both
the suspend save and resume restore paths.
Fixes: c63d25cdc59a ("mmc: sdhci-esdhc-imx: Save tuning value when card stays powered in suspend") Acked-by: Adrian Hunter <adrian.hunter@intel.com> Reviewed-by: Frank Li <Frank.Li@nxp.com> Reviewed-by: Haibo Chen <haibo.chen@nxp.com> Signed-off-by: Luke Wang <ziniu.wang_1@nxp.com> Cc: stable@vger.kernel.org Signed-off-by: Ulf Hansson <ulfh@kernel.org>
Fix for UDP sockets getting leaked during sockmap lookup/release.
Accompanied by selftests updates.
Two Sashiko's concerns to be addressed separately:
https://lore.kernel.org/bpf/20260626205814.BAC3C1F000E9@smtp.kernel.org/
Signed-off-by: Michal Luczaj <mhal@rbox.co>
---
Changes in v4:
- selftest: drop redundant `if (err)` [Sashiko]
- Link to v3: https://patch.msgid.link/20260702-sockmap-lookup-udp-leak-v3-0-ff8de8782468@rbox.co
Changes in v3:
- selftest: better error handling, ASSERT_*() macros [Sashiko]
- selftest: fix grammar, reorder patches [Kuniyuki]
- Link to v2: https://patch.msgid.link/20260626-sockmap-lookup-udp-leak-v2-0-7e7e201c951a@rbox.co
Changes in v2:
- selftest: drop the original, adapt old tests
- fix: change approach to rejecting unbound UDP [Kuniyuki]
- Link to v1: https://patch.msgid.link/20260623-sockmap-lookup-udp-leak-v1-0-05804f9308e4@rbox.co
To: Alexei Starovoitov <ast@kernel.org>
To: Daniel Borkmann <daniel@iogearbox.net>
To: Andrii Nakryiko <andrii@kernel.org>
To: Eduard Zingerman <eddyz87@gmail.com>
To: Kumar Kartikeya Dwivedi <memxor@gmail.com>
To: Martin KaFai Lau <martin.lau@linux.dev>
To: Song Liu <song@kernel.org>
To: Yonghong Song <yonghong.song@linux.dev>
To: Jiri Olsa <jolsa@kernel.org>
To: Emil Tsalapatis <emil@etsalapatis.com>
To: Shuah Khan <shuah@kernel.org>
To: John Fastabend <john.fastabend@gmail.com>
To: Jakub Sitnicki <jakub@cloudflare.com>
To: Jiayuan Chen <jiayuan.chen@linux.dev>
To: Eric Dumazet <edumazet@google.com>
To: Kuniyuki Iwashima <kuniyu@google.com>
To: Paolo Abeni <pabeni@redhat.com>
To: Willem de Bruijn <willemb@google.com>
To: "David S. Miller" <davem@davemloft.net>
To: Jakub Kicinski <kuba@kernel.org>
To: Simon Horman <horms@kernel.org>
To: Cong Wang <cong.wang@bytedance.com> Cc: bpf@vger.kernel.org Cc: linux-kselftest@vger.kernel.org Cc: linux-kernel@vger.kernel.org Cc: netdev@vger.kernel.org
---
====================
Michal Luczaj [Tue, 7 Jul 2026 04:23:57 +0000 (06:23 +0200)]
bpf, sockmap: Reject unhashed UDP sockets on sockmap update
UDP sockets get SOCK_RCU_FREE set when (auto-)bound. This means
sk_is_refcounted(unbound) = true, while sk_is_refcounted(bound) = false.
Because sockmap accepts unbound UDP sockets, a BPF program can increment a
socket's refcount via lookup. If the socket is subsequently bound, the
transition from unbound to bound causes bpf_sk_release() to skip the
decrement of the refcount, causing a memory leak.
Instead of special-casing for refcounted sockets, reject unhashed UDP
sockets during sockmap updates, as there is no benefit to supporting those.
This effectively reverts the commit under Fixes, with two exceptions:
1. sock_map_sk_state_allowed() maintains a fall-through `return true`.
2. In the spirit of commit b8b8315e39ff ("bpf, sockmap: Remove unhash
handler for BPF sockmap usage"), the proto::unhash BPF handler is not
reintroduced.
Historical note: this issue is related to commit 67312adc96b5 ("bpf: reject
unhashed sockets in bpf_sk_assign").
Fixes: 0c48eefae712 ("sock_map: Lift socket state restriction for datagram sockets") Suggested-by: Kuniyuki Iwashima <kuniyu@google.com> Signed-off-by: Michal Luczaj <mhal@rbox.co> Reviewed-by: Kuniyuki Iwashima <kuniyu@google.com> Reviewed-by: Jakub Sitnicki <jakub@cloudflare.com> Reviewed-by: John Fastabend <john.fastabend@gmail.com> Link: https://lore.kernel.org/bpf/20260707-sockmap-lookup-udp-leak-v4-2-f878346f27ab@rbox.co Signed-off-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
Xie Bo [Wed, 15 Jul 2026 02:03:59 +0000 (10:03 +0800)]
RISC-V: KVM: Serialize virtual interrupt pending state updates
KVM RISC-V tracks guest local interrupt state with two bitmaps:
- irqs_pending: interrupts that should be visible to the guest
- irqs_pending_mask: interrupts whose pending state changed
The current code updates those bitmaps with independent atomic bitops
and assumes a multiple-producer, single-consumer protocol. That model
does not actually hold.
kvm_riscv_vcpu_sync_interrupts() is not a pure consumer. When the guest
changes guest-visible HVIP state, sync_interrupts() writes both
irqs_pending and irqs_pending_mask to reflect the new guest state back
into KVM state. As a result, irqs_pending and irqs_pending_mask form a
single logical state transition, but they are not updated atomically as
a pair.
This allows a race where a newly injected interrupt is lost. For
example:
After that interleaving, a later flush can update HVIP without VSSIP
even though a new virtual interrupt was injected. In practice, the
guest can remain blocked in WFI with work pending.
The same pending/mask protocol is shared by VS soft interrupts, PMU
overflow delivery, and AIA high interrupt synchronization, so the race
is not limited to one interrupt source.
Fix this by serializing all updates to irqs_pending and irqs_pending_mask
with a per-vCPU raw spinlock. This keeps the pending bit and the dirty
mask as one state transition across:
- set/unset interrupt
- guest HVIP sync
- interrupt flush to guest CSR state
- vCPU reset
- AIA CSR writes that clear dirty state
Use non-atomic bitmap operations while holding the lock. Hold the lock
across the AIA sync, flush, and pending checks as well, so both bitmap
words share the same serialization domain.
This intentionally replaces the existing lockless protocol instead of
trying to repair it with additional barriers. The problem is not memory
ordering on a single field; it is that two separate bitmaps encode one
shared state machine while both producers and sync paths can modify
them. A per-vCPU raw spinlock keeps the fix small, local, and suitable
for backporting.
Matthew Brost [Thu, 2 Jul 2026 21:48:15 +0000 (14:48 -0700)]
drm/ttm: Account for NULL and handle pages in ttm_pool_backup
Pages in ttm_pool_backup can be NULL or backup handles
(ttm_backup_page_ptr_is_handle()), neither of which can be passed to
set_pages_array_wb() or freed. Add a dedicated WB pass before the
dma/purge loop that walks allocations using the same i += num_pages
stride, skipping NULL and handle entries, and calls set_pages_array_wb()
once per contiguous run of real pages. Apply the same NULL/handle guard
to the dma/purge loop.
Instead, the entire file is reported as data. SEEK_HOLE(0) returns EOF,
and SEEK_DATA(1M) returns 1M.
This happens because a dirty folio spanning the punched range can be
written back after the punch and refill the hole.
Fix this by flushing and waiting for dirty data in the punched range
before invalidating the page cache and issuing FSCTL_SET_ZERO_DATA.
The xfstests generic/539 pass against Samba/ksmbd with this change.
Signed-off-by: Huiwen He <hehuiwen@kylinos.cn> Reviewed-by: ChenXiaoSong <chenxiaosong@kylinos.cn> Signed-off-by: Steve French <stfrench@microsoft.com>
Andy Shevchenko [Tue, 14 Jul 2026 13:53:33 +0000 (15:53 +0200)]
smb/client: Use EXPORT_SYMBOL_IF_KUNIT() to export symbols in SMB2
Replace EXPORT_SYMBOL_FOR_MODULES() with EXPORT_SYMBOL_IF_KUNIT()
to mark the symbols as visible only if CONFIG_KUNIT is enabled.
Kunit test should import the namespace EXPORTED_FOR_KUNIT_TESTING to
use these marked symbols. This is the standard way for all KUnit
tests.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Reviewed-by: ChenXiaoSong <chenxiaosong@kylinos.cn> Signed-off-by: Steve French <stfrench@microsoft.com>
Andy Shevchenko [Tue, 14 Jul 2026 12:21:20 +0000 (14:21 +0200)]
smb/client: Use EXPORT_SYMBOL_IF_KUNIT() to export symbols
Replace EXPORT_SYMBOL_FOR_MODULES() with EXPORT_SYMBOL_IF_KUNIT()
to mark the symbols as visible only if CONFIG_KUNIT is enabled.
Kunit test should import the namespace EXPORTED_FOR_KUNIT_TESTING to
use these marked symbols. This is the standard way for all KUnit
tests.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Reviewed-by: ChenXiaoSong <chenxiaosong@kylinos.cn> Signed-off-by: Steve French <stfrench@microsoft.com>
Since commit 7dadeaa6e851 ("sched: Further restrict the preemption
modes"), powerpc always has CONFIG_PREEMPTION because only
CONFIG_PREEMPT and CONFIG_PREEMPT_LAZY are possible, even in
dynamic preemption mode (see sched_dynamic_mode).
As a consequence, need_irq_preemption() is always true and can be
removed.
And because commit bee25f97ad24 ("powerpc: Enable GENERIC_ENTRY
feature") includes linux/irq-entry-common.h which already declares
sk_dynamic_irqentry_exit_cond_resched static key, asm/preempt.h
becauses useless and can be removed.
Amit Machhiwal [Sun, 14 Jun 2026 17:34:37 +0000 (23:04 +0530)]
powerpc/dt_cpu_ftrs: Set CPU_FTR_P11_PVR for Power11 and later processors
When using device tree CPU features (dt-cpu-ftrs), the kernel bypasses
the traditional cputable-based CPU identification and instead derives
CPU features from the device tree's "ibm,powerpc-cpu-features" node
provided by firmware.
However, CPU_FTR_P11_PVR is a kernel-internal feature flag used to
identify Power11 and later processors, and is not represented in the
device tree's ISA feature set. While ISA v3.1 support (indicated by
CPU_FTR_ARCH_31) is present on both Power10 and Power11, the
CPU_FTR_P11_PVR flag is specifically needed by code that must
distinguish between Power10 and Power11 processors.
Without this flag set, code that checks for Power11 using
cpu_has_feature(CPU_FTR_P11_PVR) will incorrectly return false on
Power11+ systems using dt-cpu-ftrs, leading to incorrect behavior.
This issue manifests specifically in powernv environments (bare-metal
or QEMU TCG with powernv machine type), where skiboot/OPAL firmware
provides the "ibm,powerpc-cpu-features" node, causing the kernel to
use dt-cpu-ftrs. The issue does not affect pseries guests, where SLOF
firmware does not provide this node, causing the kernel to fall back
to the traditional cputable path (identify_cpu) which correctly sets
CPU_FTR_P11_PVR during PVR-based CPU identification.
In powernv TCG guests, the missing flag causes KVM code to trigger
warnings when attempting to create KVM guests, as cpu_features shows
0x000c00eb8f4fb187 (missing bit 53) instead of the correct
0x002c00eb8f4fb187 (with bit 53 set).
Fix this by setting CPU_FTR_P11_PVR for all processors with
PVR >= PVR_POWER11 when ISA v3.1 support is detected in
cpufeatures_setup_start(). This approach ensures forward
compatibility with future processor generations.
projects / linux.git / log