Unbinding amdgpu has no problems, but binding it again leads to an
error of sysfs file already existing. This is because it wasn't
actually cleaned up on unbind. Add the missing cleanup step.
Fixes: 547aad32ed ("drm/amdgpu: add VCN4 ip block support")
Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
amdgpu_acpi_detect() calls some helper functions it calls have large
local structures. When the compiler inlines these helpers, their local
data adds to the amdgpu_acpi_detect() stack frame.
Mark the helpers with noinline_for_stack:
- amdgpu_atif_verify_interface()
- amdgpu_atif_get_notification_params()
- amdgpu_atif_query_backlight_caps()
- amdgpu_atcs_verify_interface()
- amdgpu_acpi_enumerate_xcc()
This keeps the large temporary objects inside the helper’s own stack
frame instead of being inlined into the caller, preventing the caller
from growing beyond the stack limit.
Fixes the below with gcc W=1:
drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c:1403:6: warning: stack frame size (1688) exceeds limit (1024) in 'amdgpu_acpi_detect' [-Wframe-larger-than]
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Christian König <christian.koenig@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Ensure the dma state is initialized when we're not using the contiguous
iova, otherwise the caller may be using a stale state from a previous
request that could use the coalesed iova allocation.
Fixes: 2f6b2565d4 ("block: accumulate memory segment gaps per bio")
Reported-by: Sebastian Ott <sebott@redhat.com>
Tested-by: Sebastian Ott <sebott@redhat.com>
Signed-off-by: Keith Busch <kbusch@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Segment info indexing also used sizeof(struct) instead of the
4K metadata stride, so info_index could point between slots and
subsequent writes would advance incorrectly. Derive info_index
from the pointer returned by the segment meta search using
PCACHE_SEG_INFO_SIZE and advance to the next slot for future
updates.
Signed-off-by: Li Chen <chenl311@chinatelecom.cn>
Signed-off-by: Dongsheng Yang <dongsheng.yang@linux.dev>
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Reviewed-by: Zheng Gu <cengku@gmail.com>
Cc: stable@vger.kernel.org # 6.18
The on-media cache_info index used sizeof(struct) instead of the
4K metadata stride, so gc_percent updates from dmsetup message
were written between slots and lost after reboot. Use
PCACHE_CACHE_INFO_SIZE in get_cache_info_addr() and align
info_index with the slot returned by pcache_meta_find_latest().
Signed-off-by: Li Chen <chenl311@chinatelecom.cn>
Signed-off-by: Dongsheng Yang <dongsheng.yang@linux.dev>
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Reviewed-by: Zheng Gu <cengku@gmail.com>
Cc: stable@vger.kernel.org # 6.18
In dm-pcache, in order to ensure crash-consistency, a dual-copy scheme
is used to alternately update metadata, and there is a slot index that
records the current slot. However, in the write path the current
implementation writes directly to the current slot indexed by slot
index, and then advances the slot — which ends up overwriting the
existing slot, violating the crash-consistency guarantee.
This patch fixes that behavior, preventing metadata from being
overwritten incorrectly.
In addition, this patch add a missing pmem_wmb() after memcpy_flushcache().
Signed-off-by: Dongsheng Yang <dongsheng.yang@linux.dev>
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Reviewed-by: Zheng Gu <cengku@gmail.com>
Cc: stable@vger.kernel.org # 6.18
Also enhance possible takeover/reshape information and do some reformatting.
Signed-off-by: Heinz Mauelshagen <heinzm@redhat.com>
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
The log_writes_kthread() calls try_to_freeze() but lacks set_freezable(),
rendering the freeze attempt ineffective since kernel threads are
non-freezable by default. This prevents proper thread suspension during
system suspend/hibernate.
Add set_freezable() to explicitly mark the thread as freezable.
Fixes: 0e9cebe724 ("dm: add log writes target")
Signed-off-by: Haotian Zhang <vulab@iscas.ac.cn>
Reviewed-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
rs->raid_type is assigned from get_raid_type_by_ll(), which may return
NULL. This NULL value could be dereferenced later in the condition
'if (!(rs_is_raid10(rs) && rt_is_raid0(rs->raid_type)))'.
Add a fail-fast check to return early with an error if raid_type is NULL,
similar to other uses of this function.
Found by Linux Verification Center (linuxtesting.org) with Svace.
Fixes: 33e53f0685 ("dm raid: introduce extended superblock and new raid types to support takeover/reshaping")
Signed-off-by: Alexey Simakov <bigalex934@gmail.com>
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
There is reported 'scheduling while atomic' bug when using dm-snapshot on
real-time kernels. The reason for the bug is that the hlist_bl code does
preempt_disable() when taking the lock and the kernel attempts to take
other spinlocks while holding the hlist_bl lock.
Fix this by converting a hlist_bl spinlock into a regular spinlock.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Reported-by: Jiping Ma <jiping.ma2@windriver.com>
__blkdev_issue_discard() always returns 0, making all error checking
at call sites dead code.
For dm-thin change issue_discard() return type to void, in
passdown_double_checking_shared_status() remove the r assignment from
return value of the issue_discard(), for end_discard() hardcode value of
r to 0 that matches only value returned from __blkdev_issue_discard().
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Chaitanya Kulkarni <ckulkarnilinux@gmail.com>
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Ben will be working with me as a maintainer, so add him to the
MAINTAINERS file.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Acked-by: Mike Snitzer <snitzer@kernel.org>
There's no point to the MPATHF_RETAIN_ATTACHED_HW_HANDLER flag any more.
The way setup_scsi_dh() worked, if that flag wasn't set, it would
attempt to attach any passed in hardware handler. This would always fail
if a different hardware handler was attached, which caused
setup_scsi_dh() to rerun as if the flag was set. So the code would
already retain any attached handler, because attaching a different one
would always fail.
Also, the code had a bug. If attached_handler_name was NULL but there
was a scsi device handler attached (because either
scsi_dh_attached_handler_name failed() to allocate a name, a handler got
attached after it was called) the code would loop endlessly.
Instead, ignore MPATHF_RETAIN_ATTACHED_HW_HANDLER, and always free the
passed in handler if *attached_handler_name is set. This simplifies the
code, and avoids the endless loop bug, while keeping the functionality
the same.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Fix kerneldoc warnings across the dm-vdo target. Also
remove some unhelpful or inaccurate doc comments, and fix
some format inconsistencies that did not produce warnings.
No functional changes.
Suggested-by: Sunday Adelodun <adelodunolaoluwa@yahoo.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
There may be devices with physical block size larger than 4k.
If dm-bufio sends I/O that is not aligned on physical block size,
performance is degraded.
The 4k minimum alignment limit is there because some SSDs report logical
and physical block size 512 despite having 4k internally - so dm-bufio
shouldn't send I/Os not aligned on 4k boundary, because they perform
badly (the SSD does read-modify-write for them).
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Reported-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Cc: stable@vger.kernel.org
Allow handling of bios with REQ_ATOMIC flag set.
Don't split these bios and fail them if they overrun the hard limit
"BIO_MAX_VECS << PAGE_SHIFT".
In order to simplify the code, this commit joins the logic that avoids
splitting emulated zone append bios with the logic that avoids
splitting atomic write bios.
Signed-off-by: John Garry <john.g.garry@oracle.com>
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Tested-by: John Garry <john.g.garry@oracle.com>
Any bio with REQ_ATOMIC flag set should never be split or partially
completed, so BUG_ON() on this scenario in dm_accept_partial_bio() (whose
intent is to allow partial completions).
Also, we must reject atomic bio to targets that don't support them,
otherwise this BUG could be triggered by stray bios that have the
REQ_ATOMIC set.
Signed-off-by: John Garry <john.g.garry@oracle.com>
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Tested-by: John Garry <john.g.garry@oracle.com>
v->fec->extra_pool has zero reserved entries, so we can remove it and use
the kernel cache directly.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Reviewed-by: Eric Biggers <ebiggers@kernel.org>
There are two problems with the recursive correction:
1. It may cause denial-of-service. In fec_read_bufs, there is a loop that
has 253 iterations. For each iteration, we may call verity_hash_for_block
recursively. There is a limit of 4 nested recursions - that means that
there may be at most 253^4 (4 billion) iterations. Red Hat QE team
actually created an image that pushes dm-verity to this limit - and this
image just makes the udev-worker process get stuck in the 'D' state.
2. It doesn't work. In fec_read_bufs we store data into the variable
"fio->bufs", but fio bufs is shared between recursive invocations, if
"verity_hash_for_block" invoked correction recursively, it would
overwrite partially filled fio->bufs.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Reported-by: Guangwu Zhang <guazhang@redhat.com>
Reviewed-by: Sami Tolvanen <samitolvanen@google.com>
Reviewed-by: Eric Biggers <ebiggers@kernel.org>
Commit 9a7c987fb9 ("crypto: arm64/ghash - Use API partial block
handling") made ghash_finup() pass the wrong buffer to
ghash_do_simd_update(). As a result, ghash-neon now produces incorrect
outputs when the message length isn't divisible by 16 bytes. Fix this.
(I didn't notice this earlier because this code is reached only on CPUs
that support NEON but not PMULL. I haven't yet found a way to get
qemu-system-aarch64 to emulate that configuration.)
Fixes: 9a7c987fb9 ("crypto: arm64/ghash - Use API partial block handling")
Cc: stable@vger.kernel.org
Reported-by: Diederik de Haas <diederik@cknow-tech.com>
Closes: https://lore.kernel.org/linux-crypto/DETXT7QI62KE.F3CGH2VWX1SC@cknow-tech.com/
Tested-by: Diederik de Haas <diederik@cknow-tech.com>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Link: https://lore.kernel.org/r/20251209223417.112294-1-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@kernel.org>
For one command type, cu_idx is assigned before calling memset() on the
command structure. This results in cu_idx being overwritten, causing the
firmware to receive an incomplete or invalid command and leading to
unexpected command failures.
Fix this by moving the memset() call before initializing cu_idx so that
all fields are populated in the correct order.
Fixes: 71829d7f2f ("accel/amdxdna: Use MSG_OP_CHAIN_EXEC_NPU when supported")
Reviewed-by: Mario Limonciello (AMD) <superm1@kernel.org>
Signed-off-by: Lizhi Hou <lizhi.hou@amd.com>
Link: https://patch.msgid.link/20251209211639.1636888-1-lizhi.hou@amd.com
When config table entries don't match with the device to be probed,
currently we fall back to SND_INTEL_DSP_DRIVER_ANY, which means to
allow any drivers to bind with it.
This was set so with the assumption (or hope) that all controller
drivers should cover the devices generally, but in practice, this
caused a problem as reported recently. Namely, when a specific
kconfig for SOF isn't set for the modern Intel chips like Alderlake,
a wrong driver (AVS) got probed and failed. This is because we have
entries like:
#if IS_ENABLED(CONFIG_SND_SOC_SOF_ALDERLAKE)
/* Alder Lake / Raptor Lake */
{
.flags = FLAG_SOF | FLAG_SOF_ONLY_IF_DMIC_OR_SOUNDWIRE,
.device = PCI_DEVICE_ID_INTEL_HDA_ADL_S,
},
....
#endif
so this entry is effective only when CONFIG_SND_SOC_SOF_ALDERLAKE is
set. If not set, there is no matching entry, hence it returns
SND_INTEL_DSP_DRIVER_ANY as fallback. OTOH, if the kconfig is set, it
explicitly falls back to SND_INTEL_DSP_DRIVER_LEGACY when no DMIC or
SoundWire is found -- that was the working scenario. That being said,
the current setup may be broken for modern Intel chips that are
supposed to work with either SOF or legacy driver when the
corresponding kconfig were missing.
For addressing the problem above, this patch changes the fallback
driver to the legacy driver, i.e. return SND_INTEL_DSP_DRIVER_LEGACY
type as much as possible. When CONFIG_SND_HDA_INTEL is also disabled,
the fallback is set to SND_INTEL_DSP_DRIVER_ANY type, just to be sure.
Reported-by: Askar Safin <safinaskar@gmail.com>
Closes: https://lore.kernel.org/all/20251014034156.4480-1-safinaskar@gmail.com/
Tested-by: Askar Safin <safinaskar@gmail.com>
Reviewed-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Link: https://patch.msgid.link/20251210131553.184404-1-tiwai@suse.de
On C10 PHY PLLs the SSC is enabled by programming the
XELPDP_PORT_CLOCK_CTL / XELPDP_SSC_ENABLE_PLLB flag and the
PHY_C10_VDR_PLL 4..8 registers:
- If SSC is enabled XELPDP_SSC_ENABLE_PLLB is set and the
PHY_C10_VDR_PLL registers are programmed to non-zero values.
- If SSC is disabled XELPDP_SSC_ENABLE_PLLB is cleared and the
PHY_C10_VDR_PLL registers are programmed to zeroed-out values.
The driver's state checker verifies if the above settings are consistent,
i.e. if XELPDP_SSC_ENABLE_PLLB being set corresponds to the
PHY_C10_VDR_PLL registers being zeroed-out or not.
On WCL the BIOS programs non-zero values to the PHY_C10_VDR_PLL 4..8
registers, but does not set the XELPDP_SSC_ENABLE_PLLB flag. This will
trigger the following PLL state check warning during driver loading:
<4>[ 44.457809] xe 0000:00:02.0: [drm] PHY B: SSC enabled state (no), doesn't match PLL configuration (SSC-enabled)
<4>[ 44.457833] WARNING: CPU: 4 PID: 298 at drivers/gpu/drm/i915/display/intel_cx0_phy.c:2281 intel_cx0pll_readout_hw_state+0x221/0x620 [xe]
It's not clear whether the HW uses the PHY_C10_VDR_PLL 4..8 register
values if the XELPDP_SSC_ENABLE_PLLB flag is cleared, or just ignores
them in this case. Since the driver always programs the register values
according to the above, it still makes sense to verify that the
programming happened correctly.
To avoid the state check WARN during driver loading due to the way BIOS
programs the registers, convert the WARN to a debug message.
While at it clarify the debug message.
v2: Clarify the debug message. (Jani)
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Mika Kahola <mika.kahola@intel.com>
Reviewed-by: Mika Kahola <mika.kahola@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patch.msgid.link/20251209153407.1791839-1-imre.deak@intel.com
drm_gem_get_unmapped_area() relies on mm_get_unmapped_area() which is
only available if CONFIG_MMU=y.
Fixes: 99bda20d6d ("drm/gem: Introduce drm_gem_get_unmapped_area() fop")
Cc: Loïc Molinari <loic.molinari@collabora.com>
Reviewed-by: Loïc Molinari <loic.molinari@collabora.com>
Link: https://patch.msgid.link/20251209171151.2449120-1-boris.brezillon@collabora.com
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Now as Selective Fetch is performing full frame update on async flip and
vblank evasion is done as needed we can allow async flip even when
Selective Fetch is enabled.
Signed-off-by: Jouni Högander <jouni.hogander@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20251204070718.1090778-4-jouni.hogander@intel.com
According to bspec selective fetch is not supported with async flips and
instructing full frame update on async flip.
v4:
- check crtc_state->async_flip_planes in
psr2_sel_fetch_pipe_state_supported
v3:
- rebase
- fix old_crtc_state->pipe_srcsz_early_tpt
- fix using intel_atomic_get_new_crtc_state
v2:
- check also crtc_state->async_flip_planes in
psr2_sel_fetch_plane_state_supported
Bspec: 55229
Signed-off-by: Jouni Högander <jouni.hogander@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20251204070718.1090778-3-jouni.hogander@intel.com
Currently plane id bit is set in crtc_state->async_flip_planes only when
async flip toggle workaround is needed. We want to utilize
crtc_state->async_flip_planes further in Selective Fetch calculation.
v2:
- rework if-else if to if-if
- added comment updated
Signed-off-by: Jouni Högander <jouni.hogander@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20251204070718.1090778-2-jouni.hogander@intel.com
When CONFIG_SLUB_TINY is enabled, kfree_nolock() calls kasan_slab_free()
before defer_free(). On ARM64 with MTE (Memory Tagging Extension),
kasan_slab_free() poisons the memory and changes the tag from the
original (e.g., 0xf3) to a poison tag (0xfe).
When defer_free() then tries to write to the freed object to build the
deferred free list via llist_add(), the pointer still has the old tag,
causing a tag mismatch and triggering a KASAN use-after-free report:
BUG: KASAN: slab-use-after-free in defer_free+0x3c/0xbc mm/slub.c:6537
Write at addr f3f000000854f020 by task kworker/u8:6/983
Pointer tag: [f3], memory tag: [fe]
Fix this by calling kasan_reset_tag() before accessing the freed memory.
This is safe because defer_free() is part of the allocator itself and is
expected to manipulate freed memory for bookkeeping purposes.
Fixes: af92793e52 ("slab: Introduce kmalloc_nolock() and kfree_nolock().")
Cc: stable@vger.kernel.org
Reported-by: syzbot+7a25305a76d872abcfa1@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=7a25305a76d872abcfa1
Tested-by: syzbot+7a25305a76d872abcfa1@syzkaller.appspotmail.com
Signed-off-by: Deepanshu Kartikey <kartikey406@gmail.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Link: https://patch.msgid.link/20251210022024.3255826-1-kartikey406@gmail.com
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
__do_user_fault() may be called with indeterminent interrupt enable
state, which means we may be preemptive at this point. This causes
problems when calling harden_branch_predictor(). For example, when
called from a data abort, do_alignment_fault()->do_bad_area().
Move harden_branch_predictor() out of __do_user_fault() and into the
calling contexts.
Moving it into do_kernel_address_page_fault(), we can be sure that
interrupts will be disabled here.
Converting do_translation_fault() to use do_kernel_address_page_fault()
rather than do_bad_area() means that we keep branch predictor handling
for translation faults. Interrupts will also be disabled at this call
site.
do_sect_fault() needs special handling, so detect user mode accesses
to kernel-addresses, and add an explicit call to branch predictor
hardening.
Finally, add branch predictor hardening to do_alignment() for the
faulting case (user mode accessing kernel addresses) before interrupts
are enabled.
This should cover all cases where harden_branch_predictor() is called,
ensuring that it is always has interrupts disabled, also ensuring that
it is called early in each call path.
Reviewed-by: Xie Yuanbin <xieyuanbin1@huawei.com>
Tested-by: Xie Yuanbin <xieyuanbin1@huawei.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Zizhi Wo reports:
"During the execution of hash_name()->load_unaligned_zeropad(), a
potential memory access beyond the PAGE boundary may occur. For
example, when the filename length is near the PAGE_SIZE boundary.
This triggers a page fault, which leads to a call to
do_page_fault()->mmap_read_trylock(). If we can't acquire the lock,
we have to fall back to the mmap_read_lock() path, which calls
might_sleep(). This breaks RCU semantics because path lookup occurs
under an RCU read-side critical section."
This is seen with CONFIG_DEBUG_ATOMIC_SLEEP=y and CONFIG_KFENCE=y.
Kernel addresses (with the exception of the vectors/kuser helper
page) do not have VMAs associated with them. If the vectors/kuser
helper page faults, then there are two possibilities:
1. if the fault happened while in kernel mode, then we're basically
dead, because the CPU won't be able to vector through this page
to handle the fault.
2. if the fault happened while in user mode, that means the page was
protected from user access, and we want to fault anyway.
Thus, we can handle kernel addresses from any context entirely
separately without going anywhere near the mmap lock. This gives us
an entirely non-sleeping path for all kernel mode kernel address
faults.
As we handle the kernel address faults before interrupts are enabled,
this change has the side effect of improving the branch predictor
hardening, but does not completely solve the issue.
Reported-by: Zizhi Wo <wozizhi@huaweicloud.com>
Reported-by: Xie Yuanbin <xieyuanbin1@huawei.com>
Link: https://lore.kernel.org/r/20251126090505.3057219-1-wozizhi@huaweicloud.com
Reviewed-by: Xie Yuanbin <xieyuanbin1@huawei.com>
Tested-by: Xie Yuanbin <xieyuanbin1@huawei.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Allow __do_kernel_fault() to detect the execution of memory, so we can
provide the same fault message as do_page_fault() would do. This is
required when we split the kernel address fault handling from the
main do_page_fault() code path.
Reviewed-by: Xie Yuanbin <xieyuanbin1@huawei.com>
Tested-by: Xie Yuanbin <xieyuanbin1@huawei.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Jakub says: "We try to reserve SKIP for tests skipped because tool is
missing in env, something isn't built into the kernel etc."
use xfail, we can't force the race condition to appear at will
so its expected that the test 'fails' occasionally.
Fixes: 78a5883635 ("selftests: netfilter: add conntrack clash resolution test case")
Reported-by: Jakub Kicinski <kuba@kernel.org>
Closes: https://lore.kernel.org/netdev/20251206175647.5c32f419@kernel.org/
Signed-off-by: Florian Westphal <fw@strlen.de>
Always set nf_flow_route tuple out ifindex even if the indev is not one
of the flowtable configured devices since otherwise the outdev lookup in
nf_flow_offload_ip_hook() or nf_flow_offload_ipv6_hook() for
FLOW_OFFLOAD_XMIT_NEIGH flowtable entries will fail.
The above issue occurs in the following configuration since IP6IP6
tunnel does not support flowtable acceleration yet:
$ip addr show
5: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
link/ether 00:11:22:33:22:55 brd ff:ff:ff:ff:ff:ff link-netns ns1
inet6 2001:db8:1::2/64 scope global nodad
valid_lft forever preferred_lft forever
inet6 fe80::211:22ff:fe33:2255/64 scope link tentative proto kernel_ll
valid_lft forever preferred_lft forever
6: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
link/ether 00:22:22:33:22:55 brd ff:ff:ff:ff:ff:ff link-netns ns3
inet6 2001:db8:2::1/64 scope global nodad
valid_lft forever preferred_lft forever
inet6 fe80::222:22ff:fe33:2255/64 scope link tentative proto kernel_ll
valid_lft forever preferred_lft forever
7: tun0@NONE: <POINTOPOINT,NOARP,UP,LOWER_UP> mtu 1452 qdisc noqueue state UNKNOWN group default qlen 1000
link/tunnel6 2001:db8:2::1 peer 2001:db8:2::2 permaddr a85:e732:2c37::
inet6 2002:db8:1::1/64 scope global nodad
valid_lft forever preferred_lft forever
inet6 fe80::885:e7ff:fe32:2c37/64 scope link proto kernel_ll
valid_lft forever preferred_lft forever
$ip -6 route show
2001:db8:1::/64 dev eth0 proto kernel metric 256 pref medium
2001:db8:2::/64 dev eth1 proto kernel metric 256 pref medium
2002:db8:1::/64 dev tun0 proto kernel metric 256 pref medium
default via 2002:db8:1::2 dev tun0 metric 1024 pref medium
$nft list ruleset
table inet filter {
flowtable ft {
hook ingress priority filter
devices = { eth0, eth1 }
}
chain forward {
type filter hook forward priority filter; policy accept;
meta l4proto { tcp, udp } flow add @ft
}
}
Fixes: b5964aac51 ("netfilter: flowtable: consolidate xmit path")
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
The IPv4 code path in __ip_vs_get_out_rt() calls dst_link_failure()
without ensuring skb->dev is set, leading to a NULL pointer dereference
in fib_compute_spec_dst() when ipv4_link_failure() attempts to send
ICMP destination unreachable messages.
The issue emerged after commit ed0de45a10 ("ipv4: recompile ip options
in ipv4_link_failure") started calling __ip_options_compile() from
ipv4_link_failure(). This code path eventually calls fib_compute_spec_dst()
which dereferences skb->dev. An attempt was made to fix the NULL skb->dev
dereference in commit 0113d9c9d1 ("ipv4: fix null-deref in
ipv4_link_failure"), but it only addressed the immediate dev_net(skb->dev)
dereference by using a fallback device. The fix was incomplete because
fib_compute_spec_dst() later in the call chain still accesses skb->dev
directly, which remains NULL when IPVS calls dst_link_failure().
The crash occurs when:
1. IPVS processes a packet in NAT mode with a misconfigured destination
2. Route lookup fails in __ip_vs_get_out_rt() before establishing a route
3. The error path calls dst_link_failure(skb) with skb->dev == NULL
4. ipv4_link_failure() → ipv4_send_dest_unreach() →
__ip_options_compile() → fib_compute_spec_dst()
5. fib_compute_spec_dst() dereferences NULL skb->dev
Apply the same fix used for IPv6 in commit 326bf17ea5 ("ipvs: fix
ipv6 route unreach panic"): set skb->dev from skb_dst(skb)->dev before
calling dst_link_failure().
KASAN: null-ptr-deref in range [0x0000000000000328-0x000000000000032f]
CPU: 1 PID: 12732 Comm: syz.1.3469 Not tainted 6.6.114 #2
RIP: 0010:__in_dev_get_rcu include/linux/inetdevice.h:233
RIP: 0010:fib_compute_spec_dst+0x17a/0x9f0 net/ipv4/fib_frontend.c:285
Call Trace:
<TASK>
spec_dst_fill net/ipv4/ip_options.c:232
spec_dst_fill net/ipv4/ip_options.c:229
__ip_options_compile+0x13a1/0x17d0 net/ipv4/ip_options.c:330
ipv4_send_dest_unreach net/ipv4/route.c:1252
ipv4_link_failure+0x702/0xb80 net/ipv4/route.c:1265
dst_link_failure include/net/dst.h:437
__ip_vs_get_out_rt+0x15fd/0x19e0 net/netfilter/ipvs/ip_vs_xmit.c:412
ip_vs_nat_xmit+0x1d8/0xc80 net/netfilter/ipvs/ip_vs_xmit.c:764
Fixes: ed0de45a10 ("ipv4: recompile ip options in ipv4_link_failure")
Signed-off-by: Slavin Liu <slavin452@gmail.com>
Acked-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Florian Westphal <fw@strlen.de>
There are some situations where ct might be leaked as error paths are
skipping the refcounted check and return immediately. In order to solve
it make sure that the check is always called.
Fixes: be102eb6a0 ("netfilter: nf_conncount: rework API to use sk_buff directly")
Signed-off-by: Fernando Fernandez Mancera <fmancera@suse.de>
Signed-off-by: Florian Westphal <fw@strlen.de>
None of the RBD code directly requires CRC32, CRYPTO, or CRYPTO_AES.
These options are needed by CEPH_LIB code and they are selected there
directly.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Dongsheng Yang <dongsheng.yang@linux.dev>
None of the CEPH_FS code directly requires CRC32, CRYPTO, or CRYPTO_AES.
These options do get selected indirectly anyway via CEPH_LIB, which does
need them, but there is no need for CEPH_FS to select them too.
Signed-off-by: Eric Biggers <ebiggers@kernel.org>
Reviewed-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
If the osdmap is (maliciously) corrupted such that the encoded length
of ceph_pg_pool envelope is less than what is expected for a particular
encoding version, out-of-bounds reads may ensue because the only bounds
check that is there is based on that length value.
This patch adds explicit bounds checks for each field that is decoded
or skipped.
Cc: stable@vger.kernel.org
Reported-by: ziming zhang <ezrakiez@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Xiubo Li <xiubli@redhat.com>
Tested-by: ziming zhang <ezrakiez@gmail.com>
In a few cases the code compares 32-bit value to a SIZE_MAX derived
constant which is much higher than that value on 64-bit platforms,
Clang, in particular, is not happy about this
net/ceph/osdmap.c:1441:10: error: result of comparison of constant 4611686018427387891 with expression of type 'u32' (aka 'unsigned int') is always false [-Werror,-Wtautological-constant-out-of-range-compare]
1441 | if (len > (SIZE_MAX - sizeof(*pg)) / sizeof(u32))
| ~~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
net/ceph/osdmap.c:1624:10: error: result of comparison of constant 2305843009213693945 with expression of type 'u32' (aka 'unsigned int') is always false [-Werror,-Wtautological-constant-out-of-range-compare]
1624 | if (len > (SIZE_MAX - sizeof(*pg)) / (2 * sizeof(u32)))
| ~~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Fix this by casting to size_t. Note, that possible replacement of SIZE_MAX
by U32_MAX may lead to the behaviour changes on the corner cases.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Viacheslav Dubeyko <Slava.Dubeyko@ibm.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
In a few cases the code compares 32-bit value to a SIZE_MAX derived
constant which is much higher than that value on 64-bit platforms,
Clang, in particular, is not happy about this
fs/ceph/snap.c:377:10: error: result of comparison of constant 2305843009213693948 with expression of type 'u32' (aka 'unsigned int') is always false [-Werror,-Wtautological-constant-out-of-range-compare]
377 | if (num > (SIZE_MAX - sizeof(*snapc)) / sizeof(u64))
| ~~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Fix this by casting to size_t. Note, that possible replacement of SIZE_MAX
by U32_MAX may lead to the behaviour changes on the corner cases.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Viacheslav Dubeyko <Slava.Dubeyko@ibm.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
This patch adds trace points to the Ceph filesystem MDS client:
- request submission (CEPH_MSG_CLIENT_REQUEST) and completion
(CEPH_MSG_CLIENT_REPLY)
- capabilities (CEPH_MSG_CLIENT_CAPS)
These are the central pieces that are useful for analyzing MDS
latency/performance problems from the client's perspective.
In the long run, all doutc() calls should be replaced with
tracepoints. This way, the Ceph filesystem can be traced at any time
(without spamming the kernel log). Additionally, trace points can be
used in BPF programs (which can even deference the pointer parameters
and extract more values).
Signed-off-by: Max Kellermann <max.kellermann@ionos.com>
Reviewed-by: Viacheslav Dubeyko <Slava.Dubeyko@ibm.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
OSD client logging has a problem in get_osd() and put_osd().
For one logging output refcount_read() is called twice. If recount
value changes between both calls logging output is not consistent.
This patch prints out only the resulting value.
[ idryomov: don't make the log messages more verbose ]
Signed-off-by: Simon Buttgereit <simon.buttgereit@tu-ilmenau.de>
Reviewed-by: Viacheslav Dubeyko <Slava.Dubeyko@ibm.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
blk_hctx_poll() always checks if the task is running or not, and returns
1 if the task is running. This is a leftover from when polled IO was
purely for synchronous IO, and doesn't make sense anymore when polled IO
is purely asynchronous. Similarly, marking the task as TASK_RUNNING is
also superflous, as the very much has to be running to enter the
function in the first place.
It looks like there has been this judgment for historical reasons, and
in very early versions of this function the user would set the process
state to TASK_UNINTERRUPTIBLE.
Signed-off-by: Diangang Li <lidiangang@bytedance.com>
Signed-off-by: Fengnan Chang <changfengnan@bytedance.com>
[axboe: kill all remnants of task running, pointless now. massage message]
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Shuran Liu says:
====================
bpf: fix bpf_d_path() helper prototype
Hi,
This series fixes a verifier issue with bpf_d_path() and adds a
regression test to cover its use within a hook function.
Patch 1 updates the bpf_d_path() helper prototype so that the second
argument is marked as MEM_WRITE. This makes it explicit to the verifier
that the helper writes into the provided buffer.
Patch 2 extends the existing d_path selftest to cover incorrect verifier
assumptions caused by an incorrect function prototype. The test program calls
bpf_d_path() and checks if the first character of the path can be read.
It ensures the verifier does not assume the buffer remains unwritten.
Changelog
=========
v5:
- Moved the temporary file for the fallocate test from /tmp to /dev/shm
Since bpf CI's 9P filesystem under /tmp does not support fallocate.
v4:
- Use the fallocate hook instead of an LSM hook to simplify the selftest,
as suggested by Matt and Alexei.
- Add a utility function in test_d_path.c to load the BPF program,
improving code reuse.
v3:
- Switch the pathname prefix loop to use bpf_for() instead of
#pragma unroll, as suggested by Matt.
- Remove /tmp/bpf_d_path_test in the test cleanup path.
- Add the missing Reviewed-by tags.
v2:
- Merge the new test into the existing d_path selftest rather than
creating new files.
- Add PID filtering in the LSM program to avoid nondeterministic failures
due to unrelated processes triggering bprm_check_security.
- Synchronize child execution using a pipe to ensure deterministic
updates to the PID.
Thanks for your time and reviews.
====================
Link: https://patch.msgid.link/20251206141210.3148-1-electronlsr@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Add a regression test for bpf_d_path() to cover incorrect verifier
assumptions caused by an incorrect function prototype. The test
attaches to the fallocate hook, calls bpf_d_path() and verifies that
a simple prefix comparison on the returned pathname behaves correctly
after the fix in patch 1. It ensures the verifier does not assume
the buffer remains unwritten.
Co-developed-by: Zesen Liu <ftyg@live.com>
Signed-off-by: Zesen Liu <ftyg@live.com>
Co-developed-by: Peili Gao <gplhust955@gmail.com>
Signed-off-by: Peili Gao <gplhust955@gmail.com>
Co-developed-by: Haoran Ni <haoran.ni.cs@gmail.com>
Signed-off-by: Haoran Ni <haoran.ni.cs@gmail.com>
Signed-off-by: Shuran Liu <electronlsr@gmail.com>
Link: https://lore.kernel.org/r/20251206141210.3148-3-electronlsr@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Since the stolen memory hooks are function pointers, make some of them
optional instead of having to define them for xe.
insert_node, area_address, and area_size are only needed on platforms
not supported by xe.
Reviewed-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Link: https://patch.msgid.link/0dbb460e8bd1df29df98862d08fcdfda03912673.1764930576.git.jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
