While working on something else, I noticed that the kernel would start
accepting interrupts again after crashing in an interrupt handler. Since
the kernel is already in inconsistent state, enabling interrupts is
dangerous and opens up risk of kernel state deteriorating further.
Interrupts do get enabled via what looks like an unintended side effect of
spin_unlock_irq, so switch to the more cautious
spin_lock_irqsave/spin_unlock_irqrestore instead.
Fixes: 76d2a0493a17 ("RISC-V: Init and Halt Code") Signed-off-by: Mattias Nissler <mnissler@rivosinc.com> Reviewed-by: Björn Töpel <bjorn@kernel.org> Link: https://lore.kernel.org/r/20230215144828.3370316-1-mnissler@rivosinc.com Cc: stable@vger.kernel.org Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
I don't think we can actually die() without a regs pointer, but the
compiler was warning about a NULL check after a dereference. It seems
prudent to just avoid the possibly-NULL dereference, given that when
die()ing the system is already toast so who knows how we got there.
Running a rt-kernel base on 6.2.0-rc3-rt1 on an Ampere Altra outputs
the following:
BUG: sleeping function called from invalid context at kernel/locking/spinlock_rt.c:46
in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 9, name: kworker/u320:0
preempt_count: 2, expected: 0
RCU nest depth: 0, expected: 0
3 locks held by kworker/u320:0/9:
#0: ffff3fff8c27d128 ((wq_completion)efi_rts_wq){+.+.}-{0:0}, at: process_one_work (./include/linux/atomic/atomic-long.h:41)
#1: ffff80000861bdd0 ((work_completion)(&efi_rts_work.work)){+.+.}-{0:0}, at: process_one_work (./include/linux/atomic/atomic-long.h:41)
#2: ffffdf7e1ed3e460 (efi_rt_lock){+.+.}-{3:3}, at: efi_call_rts (drivers/firmware/efi/runtime-wrappers.c:101)
Preemption disabled at:
efi_virtmap_load (./arch/arm64/include/asm/mmu_context.h:248)
CPU: 0 PID: 9 Comm: kworker/u320:0 Tainted: G W 6.2.0-rc3-rt1
Hardware name: WIWYNN Mt.Jade Server System B81.03001.0005/Mt.Jade Motherboard, BIOS 1.08.20220218 (SCP: 1.08.20220218) 2022/02/18
Workqueue: efi_rts_wq efi_call_rts
Call trace:
dump_backtrace (arch/arm64/kernel/stacktrace.c:158)
show_stack (arch/arm64/kernel/stacktrace.c:165)
dump_stack_lvl (lib/dump_stack.c:107 (discriminator 4))
dump_stack (lib/dump_stack.c:114)
__might_resched (kernel/sched/core.c:10134)
rt_spin_lock (kernel/locking/rtmutex.c:1769 (discriminator 4))
efi_call_rts (drivers/firmware/efi/runtime-wrappers.c:101)
[...]
This seems to come from commit ff7a167961d1 ("arm64: efi: Execute
runtime services from a dedicated stack") which adds a spinlock. This
spinlock is taken through:
efi_call_rts()
\-efi_call_virt()
\-efi_call_virt_pointer()
\-arch_efi_call_virt_setup()
Make 'efi_rt_lock' a raw_spinlock to avoid being preempted.
[ardb: The EFI runtime services are called with a different set of
translation tables, and are permitted to use the SIMD registers.
The context switch code preserves/restores neither, and so EFI
calls must be made with preemption disabled, rather than only
disabling migration.]
Fixes: ff7a167961d1 ("arm64: efi: Execute runtime services from a dedicated stack") Signed-off-by: Pierre Gondois <pierre.gondois@arm.com> Cc: <stable@vger.kernel.org> # v6.1+ Signed-off-by: Ard Biesheuvel <ardb@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
By default, non-mq drivers do not support nowait. This causes io_uring
to use a slower path as the driver cannot be trust not to block. brd
can safely set the nowait flag, as worst case all it does is a NOIO
allocation.
For io_uring, this makes a substantial difference. Before:
On platforms that do not support IOMMU Extended capability bit 0
Page-walk Coherency, CPU caches are not snooped when IOMMU is accessing
any translation structures. IOMMU access goes only directly to
memory. Intel IOMMU code was missing a flush for the PASID table
directory that resulted in the unrecoverable fault as shown below.
This patch adds clflush calls whenever allocating and updating
a PASID table directory to ensure cache coherency.
On the reverse direction, there's no need to clflush the PASID directory
pointer when we deactivate a context entry in that IOMMU hardware will
not see the old PASID directory pointer after we clear the context entry.
PASID directory entries are also never freed once allocated.
If the storage gives a corrupted node block due to short power failure and
reset, f2fs stops the entire operations by setting the checkpoint failure flag.
Let's give more chances to live by re-issuing IOs for a while in such critical
path.
This patch tries to mitigate lock contention between f2fs_write_checkpoint and
f2fs_get_node_info along with nat_tree_lock.
The idea is, if checkpoint is currently running, other threads that try to grab
nat_tree_lock would be better to wait for checkpoint.
Reviewed-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Stable-dep-of: 3aa51c61cb4a ("f2fs: retry to update the inode page given data corruption") Signed-off-by: Sasha Levin <sashal@kernel.org>
Let's cache nat entry if there's no lock contention only.
Reviewed-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Stable-dep-of: 3aa51c61cb4a ("f2fs: retry to update the inode page given data corruption") Signed-off-by: Sasha Levin <sashal@kernel.org>
The condition determining whether the preallocation can be used had
an off-by-one error so we didn't discard preallocation when new
allocation was just following it. This can then confuse code in
inode_getblk().
CC: stable@vger.kernel.org Fixes: 16d055656814 ("udf: Discard preallocation before extending file with a hole") Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Sasha Levin <sashal@kernel.org>
The scand kthread can send dlm messages out, especially dlm remove
messages to free memory for unused rsb on other nodes. To send out dlm
messages, midcomms must be initialized. This patch moves the midcomms
start before scand is started.
Cc: stable@vger.kernel.org Fixes: e7fd41792fc0 ("[DLM] The core of the DLM for GFS2/CLVM") Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
This patch introduces leftovers of init, start, stop and exit
functionality. The dlm application layer should always call the midcomms
layer which getting aware of such event and redirect it to the lowcomms
layer. Some functionality which is currently handled inside the start
functionality of midcomms and lowcomms should be handled in the init
functionality as it only need to be initialized once when dlm is loaded.
Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
Stable-dep-of: aad633dc0cf9 ("fs: dlm: start midcomms before scand") Signed-off-by: Sasha Levin <sashal@kernel.org>
This patch will fix a small issue when printing out that
dlm_midcomms_start() failed to start and it was printing out that the
dlm subcomponent lowcomms was failed but lowcomms is behind the midcomms
layer.
Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
Stable-dep-of: aad633dc0cf9 ("fs: dlm: start midcomms before scand") Signed-off-by: Sasha Levin <sashal@kernel.org>
Emulate ICR writes on AVIC IPI failures due to invalid targets using the
same logic as failures due to invalid types. AVIC acceleration fails if
_any_ of the targets are invalid, and crucially VM-Exits before sending
IPIs to targets that _are_ valid. In logical mode, the destination is a
bitmap, i.e. a single IPI can target multiple logical IDs. Doing nothing
causes KVM to drop IPIs if at least one target is valid and at least one
target is invalid.
Fixes: 18f40c53e10f ("svm: Add VMEXIT handlers for AVIC") Cc: stable@vger.kernel.org Reviewed-by: Paolo Bonzini <pbonzini@redhat.com> Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com> Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230106011306.85230-5-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Don't bother rewriting the ICR value into the vAPIC page on an AVIC IPI
virtualization failure, the access is a trap, i.e. the value has already
been written to the vAPIC page. The one caveat is if hardware left the
BUSY flag set (which appears to happen somewhat arbitrarily), in which
case go through the "nodecode" APIC-write path in order to clear the BUSY
flag.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20220204214205.3306634-6-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Stable-dep-of: 5aede752a839 ("KVM: SVM: Process ICR on AVIC IPI delivery failure due to invalid target") Signed-off-by: Sasha Levin <sashal@kernel.org>
Register /dev/kvm, i.e. expose KVM to userspace, only after all other
setup has completed. Once /dev/kvm is exposed, userspace can start
invoking KVM ioctls, creating VMs, etc... If userspace creates a VM
before KVM is done with its configuration, bad things may happen, e.g.
KVM will fail to properly migrate vCPU state if a VM is created before
KVM has registered preemption notifiers.
Cc: stable@vger.kernel.org Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20221130230934.1014142-2-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Allocating cpumask dynamically in zalloc_cpumask_var() is not ideal.
Allocation is somewhat slow and can (in theory and when CPUMASK_OFFSTACK)
fail. kvm_make_all_cpus_request_except() already disables preemption so
we can use pre-allocated per-cpu cpumasks instead.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com> Reviewed-by: Sean Christopherson <seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20210903075141.403071-8-vkuznets@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Stable-dep-of: 2b0128127373 ("KVM: Register /dev/kvm as the _very_ last thing during initialization") Signed-off-by: Sasha Levin <sashal@kernel.org>
Iterating over set bits in 'vcpu_bitmap' should be faster than going
through all vCPUs, especially when just a few bits are set.
Drop kvm_make_vcpus_request_mask() call from kvm_make_all_cpus_request_except()
to avoid handling the special case when 'vcpu_bitmap' is NULL, move the
code to kvm_make_all_cpus_request_except() itself.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com> Reviewed-by: Sean Christopherson <seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20210903075141.403071-5-vkuznets@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Stable-dep-of: 2b0128127373 ("KVM: Register /dev/kvm as the _very_ last thing during initialization") Signed-off-by: Sasha Levin <sashal@kernel.org>
cb_context should be freed on the error path in nfc_se_io as stated by
commit 25ff6f8a5a3b ("nfc: fix memory leak of se_io context in
nfc_genl_se_io").
Make the error path in nfc_se_io unwind everything in reverse order, i.e.
free the cb_context after unlocking the device.
Suggested-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> Signed-off-by: Fedor Pchelkin <pchelkin@ispras.ru> Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> Link: https://lore.kernel.org/r/20230306212650.230322-1-pchelkin@ispras.ru Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If the boot loader inode has never been used before, the
EXT4_IOC_SWAP_BOOT inode will initialize it, including setting the
i_size to 0. However, if the "never before used" boot loader has a
non-zero i_size, then i_disksize will be non-zero, and the
inconsistency between i_size and i_disksize can trigger a kernel
warning:
Above issue happens as follows:
ext4_iget
ext4_find_inline_data_nolock ->i_inline_off=164 i_inline_size=60
ext4_try_add_inline_entry
__ext4_mark_inode_dirty
ext4_expand_extra_isize_ea ->i_extra_isize=32 s_want_extra_isize=44
ext4_xattr_shift_entries
->after shift i_inline_off is incorrect, actually is change to 176
ext4_try_add_inline_entry
ext4_update_inline_dir
get_max_inline_xattr_value_size
if (EXT4_I(inode)->i_inline_off)
entry = (struct ext4_xattr_entry *)((void *)raw_inode +
EXT4_I(inode)->i_inline_off);
free += EXT4_XATTR_SIZE(le32_to_cpu(entry->e_value_size));
->As entry is incorrect, then 'free' may be negative
ext4_update_inline_data
value = kzalloc(len, GFP_NOFS);
-> len is unsigned int, maybe very large, then trigger warning when
'kzalloc()'
To resolve the above issue we need to update 'i_inline_off' after
'ext4_xattr_shift_entries()'. We do not need to set
EXT4_STATE_MAY_INLINE_DATA flag here, since ext4_mark_inode_dirty()
already sets this flag if needed. Setting EXT4_STATE_MAY_INLINE_DATA
when it is needed may trigger a BUG_ON in ext4_writepages().
Reported-by: syzbot+d30838395804afc2fa6f@syzkaller.appspotmail.com Cc: stable@kernel.org Signed-off-by: Ye Bin <yebin10@huawei.com> Reviewed-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20230307015253.2232062-3-yebin@huaweicloud.com Signed-off-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The only caller of ext4_find_inline_data_nolock() that needs setting of
EXT4_STATE_MAY_INLINE_DATA flag is ext4_iget_extra_inode(). In
ext4_write_inline_data_end() we just need to update inode->i_inline_off.
Since we are going to add one more caller that does not need to set
EXT4_STATE_MAY_INLINE_DATA, just move setting of EXT4_STATE_MAY_INLINE_DATA
out to ext4_iget_extra_inode().
Signed-off-by: Ye Bin <yebin10@huawei.com> Cc: stable@kernel.org Reviewed-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20230307015253.2232062-2-yebin@huaweicloud.com Signed-off-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
For GETFSMAP calls, the caller selects a physical block device by
writing its block number into fsmap_head.fmh_keys[01].fmr_device.
To query mappings for a subrange of the device, the starting byte of the
range is written to fsmap_head.fmh_keys[0].fmr_physical and the last
byte of the range goes in fsmap_head.fmh_keys[1].fmr_physical.
IOWs, to query what mappings overlap with bytes 3-14 of /dev/sda, you'd
set the inputs as follows:
Which would return you whatever is mapped in the 12 bytes starting at
physical offset 3.
The crash is due to insufficient range validation of keys[1] in
ext4_getfsmap_datadev. On 1k-block filesystems, block 0 is not part of
the filesystem, which means that s_first_data_block is nonzero.
ext4_get_group_no_and_offset subtracts this quantity from the blocknr
argument before cracking it into a group number and a block number
within a group. IOWs, block group 0 spans blocks 1-8192 (1-based)
instead of 0-8191 (0-based) like what happens with larger blocksizes.
The net result of this encoding is that blocknr < s_first_data_block is
not a valid input to this function. The end_fsb variable is set from
the keys that are copied from userspace, which means that in the above
example, its value is zero. That leads to an underflow here:
Leaving an impossibly large group number (2^32-1) in blocknr.
ext4_getfsmap_check_keys checked that keys[0].fmr_physical and
keys[1].fmr_physical are in increasing order, but
ext4_getfsmap_datadev adjusts keys[0].fmr_physical to be at least
s_first_data_block. This implies that we have to check it again after
the adjustment, which is the piece that I forgot.
A significant number of xfstests can cause ext4 to log one or more
warning messages when they are run on a test file system where the
inline_data feature has been enabled. An example:
"EXT4-fs warning (device vdc): ext4_dirblock_csum_set:425: inode
#16385: comm fsstress: No space for directory leaf checksum. Please
run e2fsck -D."
The xfstests include: ext4/057, 058, and 307; generic/013, 051, 068,
070, 076, 078, 083, 232, 269, 270, 390, 461, 475, 476, 482, 579, 585,
589, 626, 631, and 650.
In this situation, the warning message indicates a bug in the code that
performs the RENAME_WHITEOUT operation on a directory entry that has
been stored inline. It doesn't detect that the directory is stored
inline, and incorrectly attempts to compute a dirent block checksum on
the whiteout inode when creating it. This attempt fails as a result
of the integrity checking in get_dirent_tail (usually due to a failure
to match the EXT4_FT_DIR_CSUM magic cookie), and the warning message
is then emitted.
Fix this by simply collecting the inlined data state at the time the
search for the source directory entry is performed. Existing code
handles the rest, and this is sufficient to eliminate all spurious
warning messages produced by the tests above. Go one step further
and do the same in the code that resets the source directory entry in
the event of failure. The inlined state should be present in the
"old" struct, but given the possibility of a race there's no harm
in taking a conservative approach and getting that information again
since the directory entry is being reread anyway.
Fixes: b7ff91fd030d ("ext4: find old entry again if failed to rename whiteout") Cc: stable@kernel.org Signed-off-by: Eric Whitney <enwlinux@gmail.com> Reviewed-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20230210173244.679890-1-enwlinux@gmail.com Signed-off-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When writing a page from an encrypted file that is using
filesystem-layer encryption (not inline encryption), ext4 encrypts the
pagecache page into a bounce page, then writes the bounce page.
It also passes the bounce page to wbc_account_cgroup_owner(). That's
incorrect, because the bounce page is a newly allocated temporary page
that doesn't have the memory cgroup of the original pagecache page.
This makes wbc_account_cgroup_owner() not account the I/O to the owner
of the pagecache page as it should.
Fix this by always passing the pagecache page to
wbc_account_cgroup_owner().
To last 2 parameters to cfg80211_get_bss() should be of
the enum ieee80211_bss_type resp. enum ieee80211_privacy types,
which WLAN_CAPABILITY_ESS very much is not.
Fix both cfg80211_get_bss() calls in ioctl_cfg80211.c to pass
the right parameters.
Note that the second call was already somewhat fixed by commenting
out WLAN_CAPABILITY_ESS and passing in 0 instead. This was still
not entirely correct though since that would limit returned
BSS-es to ESS type BSS-es with privacy on.
Currently, calling clone3() with CLONE_NEWTIME in clone_args->flags
fails with -EINVAL. This is because CLONE_NEWTIME intersects with
CSIGNAL. However, CSIGNAL was deprecated when clone3 was introduced in
commit 7f192e3cd316 ("fork: add clone3"), allowing re-use of that part
of clone flags.
Fix this by explicitly allowing CLONE_NEWTIME in clone3_args_valid. This
is also in line with the respective check in check_unshare_flags which
allow CLONE_NEWTIME for unshare().
Fixes: 769071ac9f20 ("ns: Introduce Time Namespace") Cc: Andrey Vagin <avagin@openvz.org> Cc: Christian Brauner <brauner@kernel.org> Cc: stable@vger.kernel.org Signed-off-by: Tobias Klauser <tklauser@distanz.ch> Reviewed-by: Christian Brauner <brauner@kernel.org> Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When MMAP2 has the PERF_RECORD_MISC_MMAP_BUILD_ID flag, it means the
record already has the build-id info. So it marks the DSO as hit, to
skip if the same DSO is not processed if it happens to miss the build-id
later.
But it missed to copy the MMAP2 record itself so it'd fail to symbolize
samples for those regions.
For example, the following generates 249 MMAP2 events.
Fixes: f7fc0d1c915a74ff ("perf inject: Do not inject BUILD_ID record if MMAP2 has it") Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20230223070155.54251-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Remove call_usermodehelper starting /etc/acpi/events/RadioPower.sh that
is not available. This script is not part of the kernel and it is not
officially available on the www. The result is that this lines are just
dead code.
Remove function _rtl92e_dm_check_ac_dc_power calling a script
/etc/acpi/wireless-rtl-ac-dc-power.sh that is not available. This script
is not part of the kernel and it is not available on the www. The result
is that this function is just dead code.
This is a re-do of commit e0e0747de0ea ("scsi: mpt3sas: Fix return value
check of dma_get_required_mask()"), which I ended up undoing in a
mis-merge in commit 62e6e5940c0c ("Merge tag 'scsi-misc' of
git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi").
The original commit message was
scsi: mpt3sas: Fix return value check of dma_get_required_mask()
Fix the incorrect return value check of dma_get_required_mask(). Due to
this incorrect check, the driver was always setting the DMA mask to 63 bit.
Link: https://lore.kernel.org/r/20220913120538.18759-2-sreekanth.reddy@broadcom.com Fixes: ba27c5cf286d ("scsi: mpt3sas: Don't change the DMA coherent mask after allocations") Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
and this fix was lost when I mis-merged the conflict with commit 9df650963bf6 ("scsi: mpt3sas: Don't change DMA mask while reallocating
pools").
When a pool crosses the 4GB boundary region then before reallocating pools
change the coherent DMA mask to 32 bits and keep the normal DMA mask set to
63/64 bits.
Link: https://lore.kernel.org/r/20220825075457.16422-2-sreekanth.reddy@broadcom.com Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Salvatore Bonaccorso <carnil@debian.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As noted in 1a2dcbdde82e ("scsi: mpt3sas: re-do lost mpt3sas DMA mask
fix") in mainline there was a mis-merge in commit 62e6e5940c0c ("Merge
tag 'scsi-misc' of
git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi"). causing that
the fix needed to be redone later on again. To make series of patches
apply cleanly to the stable series where e0e0747de0ea ("scsi: mpt3sas:
Fix return value check of dma_get_required_mask()") was backported,
revert the aforementioned commit.
usb_kill_urb warranties that all the handlers are finished when it
returns, but does not protect against threads that might be handling
asynchronously the urb.
For UVC, the function uvc_ctrl_status_event_async() takes care of
control changes asynchronously.
If the code is executed in the following order:
CPU 0 CPU 1
===== =====
uvc_status_complete()
uvc_status_stop()
uvc_ctrl_status_event_work()
uvc_status_start() -> FAIL
Then uvc_status_start will keep failing and this error will be shown:
<4>[ 5.540139] URB 0000000000000000 submitted while active
drivers/usb/core/urb.c:378 usb_submit_urb+0x4c3/0x528
Let's improve the current situation, by not re-submiting the urb if
we are stopping the status event. Also process the queued work
(if any) during stop.
CPU 0 CPU 1
===== =====
uvc_status_complete()
uvc_status_stop()
uvc_status_start()
uvc_ctrl_status_event_work() -> FAIL
Hopefully, with the usb layer protection this should be enough to cover
all the cases.
Cc: stable@vger.kernel.org Fixes: e5225c820c05 ("media: uvcvideo: Send a control event when a Control Change interrupt arrives") Reviewed-by: Yunke Cao <yunkec@chromium.org> Signed-off-by: Ricardo Ribalda <ribalda@chromium.org> Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The receive path may take the socket right before hci_sock_release(),
but it may enqueue the packets to the socket queues after the call to
skb_queue_purge(), therefore the socket can be destroyed without clear
its queues completely.
Moving these skb_queue_purge() to the hci_sock_destruct() will fix this
issue, because nothing is referencing the socket at this point.
After an error during receiving a packet for a multi-packet DP MST
sideband message, the state tracking which packets have been received
already is not reset. This prevents the reception of subsequent down
messages (due to the pending message not yet completed with an
end-of-message-transfer packet).
Fix the above by resetting the reception state after a packet error.
If the sink gets disconnected during receiving a multi-packet DP MST AUX
down-reply/up-request sideband message, the state keeping track of which
packets have been received already is not reset. This results in a failed
sanity check for the subsequent message packet received after a sink is
reconnected (due to the pending message not yet completed with an
end-of-message-transfer packet), indicated by the
"sideband msg set header failed"
error.
Fix the above by resetting the up/down message reception state after a
disconnect event.
The implementation of 'current' on x86 is very intentionally special: it
is a very common thing to look up, and it uses 'this_cpu_read_stable()'
to get the current thread pointer efficiently from per-cpu storage.
And the keyword in there is 'stable': the current thread pointer never
changes as far as a single thread is concerned. Even if when a thread
is preempted, or moved to another CPU, or even across an explicit call
'schedule()' that thread will still have the same value for 'current'.
It is, after all, the kernel base pointer to thread-local storage.
That's why it's stable to begin with, but it's also why it's important
enough that we have that special 'this_cpu_read_stable()' access for it.
So this is all done very intentionally to allow the compiler to treat
'current' as a value that never visibly changes, so that the compiler
can do CSE and combine multiple different 'current' accesses into one.
However, there is obviously one very special situation when the
currently running thread does actually change: inside the scheduler
itself.
So the scheduler code paths are special, and do not have a 'current'
thread at all. Instead there are _two_ threads: the previous and the
next thread - typically called 'prev' and 'next' (or prev_p/next_p)
internally.
So this is all actually quite straightforward and simple, and not all
that complicated.
Except for when you then have special code that is run in scheduler
context, that code then has to be aware that 'current' isn't really a
valid thing. Did you mean 'prev'? Did you mean 'next'?
In fact, even if then look at the code, and you use 'current' after the
new value has been assigned to the percpu variable, we have explicitly
told the compiler that 'current' is magical and always stable. So the
compiler is quite free to use an older (or newer) value of 'current',
and the actual assignment to the percpu storage is not relevant even if
it might look that way.
Which is exactly what happened in the resctl code, that blithely used
'current' in '__resctrl_sched_in()' when it really wanted the new
process state (as implied by the name: we're scheduling 'into' that new
resctl state). And clang would end up just using the old thread pointer
value at least in some configurations.
This could have happened with gcc too, and purely depends on random
compiler details. Clang just seems to have been more aggressive about
moving the read of the per-cpu current_task pointer around.
The fix is trivial: just make the resctl code adhere to the scheduler
rules of using the prev/next thread pointer explicitly, instead of using
'current' in a situation where it just wasn't valid.
That same code is then also used outside of the scheduler context (when
a thread resctl state is explicitly changed), and then we will just pass
in 'current' as that pointer, of course. There is no ambiguity in that
case.
The fix may be trivial, but noticing and figuring out what went wrong
was not. The credit for that goes to Stephane Eranian.
syzbot sent a hung task report and Eric explains that adversarial
receiver may keep RWIN at 0 for a long time, so we are not guaranteed
to make forward progress. Thread which took tx_lock and went to sleep
may not release tx_lock for hours. Use interruptible sleep where
possible and reschedule the work if it can't take the lock.
If wait_for_completion_timeout() times-out in _cdns_xfer_msg() it
is possible that something could have been written to the RX FIFO.
In this case, we should drain the RX FIFO so that anything in it
doesn't carry over and mess up the next transfer.
Obviously, if we got to this state something went wrong, and we
don't really know the state of everything. The cleanup in this
situation cannot be bullet-proof but we should attempt to avoid
breaking future transaction, if only to reduce the amount of
error noise when debugging the failure from a kernel log.
Note that this patch only implements the draining for blocking
(non-deferred) transfers. The deferred API doesn't have any proper
handling of error conditions and would need some re-design before
implementing cleanup. That is a task for a separate patch...
The response_buf was declared much larger (128 entries) than the number
of responses that could ever be written into it. The Cadence IP is
configurable up to a maximum of 32 entries, and the datasheet says
that RX_FIFO_AVAIL can be 2 larger than this. So allow up to 34
responses.
Also add checking in cdns_read_response() to prevent overflowing
reponse_buf if RX_FIFO_AVAIL contains an unexpectedly large number.
The dp and ufp are defined as bool type, the return value type of
function extcon_get_state should be int, so the type of dp and ufp
are modified to int.
./drivers/phy/rockchip/phy-rockchip-typec.c:827:12-14: WARNING: Unsigned expression compared with zero: dp > 0.
Wangxun has verified there is no peer-to-peer between functions for the
below selection of SFxxx, RP1000 and RP2000 NICS. They may be
multi-function devices, but the hardware does not advertise ACS capability.
Add an ACS quirk for these devices so the functions can be in independent
IOMMU groups.
When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time. To make things simpler, just
call debugfs_lookup_and_remove() instead which handles all of the logic
at once.
When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time. To make things simpler, just
call debugfs_lookup_and_remove() instead which handles all of the logic
at once.
When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time. To make things simpler, just
call debugfs_lookup_and_remove() instead which handles all of the logic
at once.
When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time. To make things simpler, just
call debugfs_lookup_and_remove() instead which handles all of the logic at
once.
When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time. To make things simpler, just
call debugfs_lookup_and_remove() instead which handles all of the logic
at once.
A PCI bridge may reside on a bus with other devices as well. The resource
distribution code does not take this into account and therefore it expands
the bridge resource windows too much, not leaving space for the other
devices (or functions of a multifunction device). This leads to an issue
that Jonathan reported when running QEMU with the following topology (QEMU
parameters):
The first e1000 NIC here is another function in the switch upstream port.
This leads to following errors:
pci 0000:00:04.0: bridge window [mem 0x10200000-0x103fffff] to [bus 02-04]
pci 0000:02:00.0: bridge window [mem 0x10200000-0x103fffff] to [bus 03-04]
pci 0000:02:00.1: BAR 0: failed to assign [mem size 0x00020000]
e1000 0000:02:00.1: can't ioremap BAR 0: [??? 0x00000000 flags 0x0]
Fix this by taking into account bridge windows, device BARs and SR-IOV PF
BARs on the bus (PF BARs include space for VF BARS so only account PF
BARs), including the ones belonging to bridges themselves if it has any.
After division the extra resource space per hotplug bridge may not be
aligned according to the window alignment, so align it before passing it
down for further distribution.
At the moment, the UVC function graph is hardcoded IT -> PU -> OT.
To add XU support we need the ability to insert the XU descriptors
into the chain. To facilitate that, make the output terminal's
bSourceID attribute writeable so that we can configure its source.
The color matching descriptors defined in the UVC Specification
contain 3 fields with discrete numeric values representing particular
settings. Enumerate those values so that later code setting them can
be more readable.
The allocation of PageBuffer is 512 bytes in size, but the dereferencing
of struct ms_bootblock_idi (also size 512) happens at a calculated offset
within the allocation, which means the object could potentially extend
beyond the end of the allocation. Avoid this case by just allocating
enough space to catch any accesses beyond the end. Seen with GCC 13:
../drivers/usb/storage/ene_ub6250.c: In function 'ms_lib_process_bootblock':
../drivers/usb/storage/ene_ub6250.c:1050:44: warning: array subscript 'struct ms_bootblock_idi[0]' is partly outside array bounds of 'unsigned char[512]' [-Warray-bounds=]
1050 | if (le16_to_cpu(idi->wIDIgeneralConfiguration) != MS_IDI_GENERAL_CONF)
| ^~
../include/uapi/linux/byteorder/little_endian.h:37:51: note: in definition of macro '__le16_to_cpu'
37 | #define __le16_to_cpu(x) ((__force __u16)(__le16)(x))
| ^
../drivers/usb/storage/ene_ub6250.c:1050:29: note: in expansion of macro 'le16_to_cpu'
1050 | if (le16_to_cpu(idi->wIDIgeneralConfiguration) != MS_IDI_GENERAL_CONF)
| ^~~~~~~~~~~
In file included from ../drivers/usb/storage/ene_ub6250.c:5:
In function 'kmalloc',
inlined from 'ms_lib_process_bootblock' at ../drivers/usb/storage/ene_ub6250.c:942:15:
../include/linux/slab.h:580:24: note: at offset [256, 512] into object of size 512 allocated by 'kmalloc_trace'
580 | return kmalloc_trace(
| ^~~~~~~~~~~~~~
581 | kmalloc_caches[kmalloc_type(flags)][index],
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
582 | flags, size);
| ~~~~~~~~~~~~
Walking the dram->cs array was seen as accesses beyond the first array
item by the compiler. Instead, use the array index directly. This allows
for run-time bounds checking under CONFIG_UBSAN_BOUNDS as well. Seen
with GCC 13 with -fstrict-flex-arrays:
In function 'xhci_mvebu_mbus_config',
inlined from 'xhci_mvebu_mbus_init_quirk' at ../drivers/usb/host/xhci-mvebu.c:66:2:
../drivers/usb/host/xhci-mvebu.c:37:28: warning: array subscript 0 is outside array bounds of 'const struct mbus_dram_window[0]' [-Warray-bounds=]
37 | writel(((cs->size - 1) & 0xffff0000) | (cs->mbus_attr << 8) |
| ~~^~~~~~
When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time. To make things simpler, just
call debugfs_lookup_and_remove() instead which handles all of the logic
at once.
When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time. To make things simpler, just
call debugfs_lookup_and_remove() instead which handles all of the logic
at once.
When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time. To make things simpler, just
call debugfs_lookup_and_remove() instead which handles all of the logic
at once.
Cc: Jakob Koschel <jakobkoschel@gmail.com> Cc: Miaoqian Lin <linmq006@gmail.com> Acked-by: Vladimir Zapolskiy <vz@mleia.com> Link: https://lore.kernel.org/r/20230202153235.2412790-10-gregkh@linuxfoundation.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time. To make things simpler, just
call debugfs_lookup_and_remove() instead which handles all of the logic
at once.
When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time. To make things simpler, just
call debugfs_lookup_and_remove() instead which handles all of the logic
at once.
When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time. To make things simpler, just
call debugfs_lookup_and_remove() instead which handles all of the logic
at once.
When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time. To make things simpler, just
call debugfs_lookup_and_remove() instead which handles all of the logic
at once.
When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time. To make things simpler, just
call debugfs_lookup_and_remove() instead which handles all of the logic
at once.
When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time. To make things simpler, just
call debugfs_lookup_and_remove() instead which handles all of the logic
at once.
When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time. To make things simpler, just
call debugfs_lookup_and_remove() instead which handles all of the logic
at once.
When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time. To make things simpler, just
call debugfs_lookup_and_remove() instead which handles all of the logic
at once.
When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time. To make things simpler, just
call debugfs_lookup_and_remove() instead which handles all of the logic
at once.
Note, the root dentry for the debugfs directory for the device needs to
be saved so we don't have to keep looking it up, which required a bit
more refactoring to properly create and remove it when needed.
Except for isochronous-configured devices, software may set
Max_Read_Request_Size (MRRS) to any value up to 4096. If a device issues a
read request with size greater than the completer's Max_Payload_Size (MPS),
the completer is required to break the response into multiple completions.
Instead of correctly responding with multiple completions to a large read
request, some LS7A Root Ports respond with a Completer Abort. To prevent
this, the MRRS must be limited to an implementation-specific value.
The OS cannot detect that value, so rely on BIOS to configure MRRS before
booting, and quirk the Root Ports so we never set an MRRS larger than that
BIOS value for any downstream device.
N.B. Hot-added devices are not configured by BIOS, and they power up with
MRRS = 512 bytes, so these devices will be limited to 512 bytes. If the
LS7A limit is smaller, those hot-added devices may not work correctly, but
per [1], hotplug is not supported with this chipset revision.
Don't hold sdw_dev_lock while calling the peripheral driver
probe() and remove() callbacks.
Holding sdw_dev_lock around the probe() and remove() calls causes
a theoretical mutex inversion which lockdep will assert on.
During probe() the sdw_dev_lock mutex is taken first and then
ASoC/ALSA locks are taken by the probe() implementation.
During normal operation ASoC can take its locks and then trigger
a runtime resume of the component. The SoundWire resume will then
take sdw_dev_lock. This is the reverse order compared to probe().
It's not necessary to hold sdw_dev_lock when calling the probe()
and remove(), it is only used to prevent the bus core calling the
driver callbacks if there isn't a driver or the driver is removing.
All calls to the driver callbacks are guarded by the 'probed' flag.
So if sdw_dev_lock is held while setting and clearing the 'probed'
flag this is sufficient to guarantee the safety of callback
functions.
Removing the mutex from around the call to probe() means that it
is now possible for a bus event (PING response) to be handled in
parallel with the probe(). But sdw_bus_probe() already has
handling for this by calling the device update_status() after
the probe() has completed.
The GPIO controller component of the sc16is7xx driver is setup too
early, which can result in a race condition where another device tries
to utilise the GPIO lines before the sc16is7xx device has finished
initialising.
This issue manifests itself as an Oops when the GPIO lines are configured:
Unable to handle kernel read from unreadable memory at virtual address
...
pc : sc16is7xx_gpio_direction_output+0x68/0x108 [sc16is7xx]
lr : sc16is7xx_gpio_direction_output+0x4c/0x108 [sc16is7xx]
...
Call trace:
sc16is7xx_gpio_direction_output+0x68/0x108 [sc16is7xx]
gpiod_direction_output_raw_commit+0x64/0x318
gpiod_direction_output+0xb0/0x170
create_gpio_led+0xec/0x198
gpio_led_probe+0x16c/0x4f0
platform_drv_probe+0x5c/0xb0
really_probe+0xe8/0x448
driver_probe_device+0xe8/0x138
__device_attach_driver+0x94/0x118
bus_for_each_drv+0x8c/0xe0
__device_attach+0x100/0x1b8
device_initial_probe+0x28/0x38
bus_probe_device+0xa4/0xb0
deferred_probe_work_func+0x90/0xe0
process_one_work+0x1c4/0x480
worker_thread+0x54/0x430
kthread+0x138/0x150
ret_from_fork+0x10/0x1c
This patch moves the setup of the GPIO controller functions to later in the
probe function, ensuring the sc16is7xx device has already finished
initialising by the time other devices try to make use of the GPIO lines.
The error handling has also been reordered to reflect the new
initialisation order.
LPUART IP has a bug that it treats the CTS as higher priority than the
break signal, which cause the break signal sending through UARTCTRL_SBK
may impacted by the CTS input if the HW flow control is enabled.
Add this workaround patch to fix the IP bug, we can disable CTS before
asserting SBK to avoid any interference from CTS, and re-enable it when
break off.
Such as for the bluetooth chip power save feature, host can let the BT
chip get into sleep state by sending a UART break signal, and wake it up
by turning off the UART break. If the BT chip enters the sleep mode
successfully, it will pull up the CTS line, if the BT chip is woken up,
it will pull down the CTS line. If without this workaround patch, the
UART TX pin cannot send the break signal successfully as it affected by
the BT CTS pin. After adding this patch, the BT power save feature can
work well.
When specifying an invalid console= device like console=tty3270,
tty_driver_lookup_tty() returns the tty struct without checking
whether index is a valid number.
When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time. To make things simpler, just
call debugfs_lookup_and_remove() instead which handles all of the logic at
once.
The memcpy() in uvc_video_decode_meta() intentionally copies across the
length and flags members and into the trailing buf flexible array.
Split the copy so that the compiler can better reason about (the lack
of) buffer overflows here. Avoid the run-time false positive warning:
memcpy: detected field-spanning write (size 12) of single field "&meta->length" at drivers/media/usb/uvc/uvc_video.c:1355 (size 1)
Additionally fix a typo in the documentation for struct uvc_meta_buf.
Logitech B910 and C910 firmware are unable to recover from a USB
autosuspend. When it resumes, the device is in a state where it only
produces invalid frames. Eg:
$ echo 0xFFFF > /sys/module/uvcvideo/parameters/trace # enable verbose log
$ yavta -c1 -n1 --file='frame#.jpg' --format MJPEG --size=1920x1080 /dev/video1
[350438.435219] uvcvideo: uvc_v4l2_open
[350438.529794] uvcvideo: Resuming interface 2
[350438.529801] uvcvideo: Resuming interface 3
[350438.529991] uvcvideo: Trying format 0x47504a4d (MJPG): 1920x1080.
[350438.529996] uvcvideo: Using default frame interval 33333.3 us (30.0 fps).
[350438.551496] uvcvideo: uvc_v4l2_mmap
[350438.555890] uvcvideo: Device requested 3060 B/frame bandwidth.
[350438.555896] uvcvideo: Selecting alternate setting 11 (3060 B/frame bandwidth).
[350438.556362] uvcvideo: Allocated 5 URB buffers of 32x3060 bytes each.
[350439.316468] uvcvideo: Marking buffer as bad (error bit set).
[350439.316475] uvcvideo: Frame complete (EOF found).
[350439.316477] uvcvideo: EOF in empty payload.
[350439.316484] uvcvideo: frame 1 stats: 149/261/417 packets, 1/149/417 pts (early initial), 416/417 scr, last pts/stc/sof 2976325734/2978107243/249
[350439.384510] uvcvideo: Marking buffer as bad (error bit set).
[350439.384516] uvcvideo: Frame complete (EOF found).
[350439.384518] uvcvideo: EOF in empty payload.
[350439.384525] uvcvideo: frame 2 stats: 265/379/533 packets, 1/265/533 pts (early initial), 532/533 scr, last pts/stc/sof 2979524454/2981305193/316
[350439.448472] uvcvideo: Marking buffer as bad (error bit set).
[350439.448478] uvcvideo: Frame complete (EOF found).
[350439.448480] uvcvideo: EOF in empty payload.
[350439.448487] uvcvideo: frame 3 stats: 265/377/533 packets, 1/265/533 pts (early initial), 532/533 scr, last pts/stc/sof 2982723174/2984503144/382
...(loop)...
The devices can leave this invalid state if the alternate setting of
the streaming interface is toggled.
This patch adds a quirk for this device so it can be autosuspended
properly.
lsusb -v:
Bus 001 Device 049: ID 046d:0821 Logitech, Inc. HD Webcam C910
Device Descriptor:
bLength 18
bDescriptorType 1
bcdUSB 2.00
bDeviceClass 239 Miscellaneous Device
bDeviceSubClass 2
bDeviceProtocol 1 Interface Association
bMaxPacketSize0 64
idVendor 0x046d Logitech, Inc.
idProduct 0x0821 HD Webcam C910
bcdDevice 0.10
iManufacturer 0
iProduct 0
iSerial 1 390022B0
bNumConfigurations 1
On a Webcam from Quanta, we see the following error.
usb 3-5: New USB device found, idVendor=0408, idProduct=30d2, bcdDevice= 0.03
usb 3-5: New USB device strings: Mfr=3, Product=1, SerialNumber=2
usb 3-5: Product: USB2.0 HD UVC WebCam
usb 3-5: Manufacturer: Quanta
usb 3-5: SerialNumber: 0x0001
...
uvcvideo: Found UVC 1.10 device USB2.0 HD UVC WebCam (0408:30d2)
uvcvideo: Failed to initialize entity for entity 5
uvcvideo: Failed to register entities (-22).
The Webcam reports an entity of type UVC_VC_EXTENSION_UNIT. It reports a
string index of '7' associated with that entity. The attempt to read that
string from the camera fails with error -32 (-EPIPE). usb_string() returns
that error, but it is ignored. As result, the entity name is empty. This
later causes v4l2_device_register_subdev() to return -EINVAL, and no
entities are registered as result.
While this appears to be a firmware problem with the camera, the kernel
should still handle the situation gracefully. To do that, check the return
value from usb_string(). If it reports an error, assign the entity's
default name.
projects / thirdparty / kernel / stable.git / log
