Linus Torvalds [Fri, 9 Jun 2023 01:52:54 +0000 (18:52 -0700)]
Merge tag 'cgroup-for-6.4-rc5-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup
Pull cgroup fixes from Tejun Heo:
- Fix css_set reference leaks on fork failures
- Fix CPU hotplug locking in cgroup_transfer_tasks() which is used by
cgroup1 cpuset
- Doc update
* tag 'cgroup-for-6.4-rc5-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
cgroup: Documentation: Clarify usage of memory limits
cgroup: always put cset in cgroup_css_set_put_fork
cgroup: fix missing cpus_read_{lock,unlock}() in cgroup_transfer_tasks()
Dave Airlie [Fri, 9 Jun 2023 01:20:22 +0000 (11:20 +1000)]
Merge tag 'drm-msm-fixes-2023-06-08' of https://gitlab.freedesktop.org/drm/msm into drm-fixes
A few more late fixes for v6.4-rc6
+ Fix max segment size to address splat on newer a6xx
+ Disable PSR by default w/ modparam to re-enable, since there
still seems to be a lingering issue
+ Fix HPD issue
+ Fix issue with unitialized GMU mutex
Move GMU mutex initialization earlier to make sure that it is always
initialized. a6xx_destroy can be called from ther failure path before
GMU initialization.
Fixes: 4cd15a3e8b36 ("drm/msm/a6xx: Make GPU destroy a bit safer") Cc: Douglas Anderson <[email protected]> Signed-off-by: Dmitry Baryshkov <[email protected]> Reviewed-by: Douglas Anderson <[email protected]>
Patchwork: https://patchwork.freedesktop.org/patch/531540/ Signed-off-by: Rob Clark <[email protected]>
Kuogee Hsieh [Tue, 23 May 2023 21:52:36 +0000 (14:52 -0700)]
drm/msm/dp: enable HDP plugin/unplugged interrupts at hpd_enable/disable
The internal_hpd flag is set to true by dp_bridge_hpd_enable() and set to
false by dp_bridge_hpd_disable() to handle GPIO pinmuxed into DP controller
case. HDP related interrupts can not be enabled until internal_hpd is set
to true. At current implementation dp_display_config_hpd() will initialize
DP host controller first followed by enabling HDP related interrupts if
internal_hpd was true at that time. Enable HDP related interrupts depends on
internal_hpd status may leave system with DP driver host is in running state
but without HDP related interrupts being enabled. This will prevent external
display from being detected. Eliminated this dependency by moving HDP related
interrupts enable/disable be done at dp_bridge_hpd_enable/disable() directly
regardless of internal_hpd status.
Changes in V3:
-- dp_catalog_ctrl_hpd_enable() and dp_catalog_ctrl_hpd_disable()
-- rewording ocmmit text
Changes in V4:
-- replace dp_display_config_hpd() with dp_display_host_start()
-- move enable_irq() at dp_display_host_start();
Changes in V5:
-- replace dp_display_host_start() with dp_display_host_init()
Changes in V6:
-- squash remove enable_irq() and disable_irq()
Dave Airlie [Fri, 9 Jun 2023 01:10:09 +0000 (11:10 +1000)]
Merge tag 'drm-misc-fixes-2023-06-08' of git://anongit.freedesktop.org/drm/drm-misc into drm-fixes
drm-misc-fixes for v6.4-rc6:
- resume and modeset fixes for ast.
- Fill in fb-helper vars more correctly.
- Assorted ivpu fixes.
- lima context destroy fix.
Linus Torvalds [Thu, 8 Jun 2023 20:54:26 +0000 (13:54 -0700)]
Merge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Pull arm64 fixes from Will Deacon:
"Two tiny arm64 fixes for -rc6.
One fixes a build breakage when MAX_ORDER can be nonsensical if
CONFIG_EXPERT=y and the other fixes the address masking for perf's
page fault software events so that it is consistent amongst them:
- Fix build breakage due to bogus MAX_ORDER definitions on !4k pages
- Avoid masking fault address for perf software events"
* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
arm64: mm: pass original fault address to handle_mm_fault() in PER_VMA_LOCK block
arm64: Remove the ARCH_FORCE_MAX_ORDER config input prompt
Wei Fang [Wed, 7 Jun 2023 09:10:48 +0000 (17:10 +0800)]
net: enetc: correct the indexes of highest and 2nd highest TCs
For ENETC hardware, the TCs are numbered from 0 to N-1, where N
is the number of TCs. Numerically higher TC has higher priority.
It's obvious that the highest priority TC index should be N-1 and
the 2nd highest priority TC index should be N-2.
However, the previous logic uses netdev_get_prio_tc_map() to get
the indexes of highest priority and 2nd highest priority TCs, it
does not make sense and is incorrect to give a "tc" argument to
netdev_get_prio_tc_map(). So the driver may get the wrong indexes
of the two highest priotiry TCs which would lead to failed to set
the CBS for the two highest priotiry TCs.
In this example, the previous logic deems the indexes of the two
highest priotiry TCs should be 3 and 2. Actually, the indexes are
5 and 4, because the number of TCs is 6. So it would be failed to
configure the CBS for the two highest priority TCs.
Fixes: c431047c4efe ("enetc: add support Credit Based Shaper(CBS) for hardware offload") Signed-off-by: Wei Fang <[email protected]> Reviewed-by: Vladimir Oltean <[email protected]> Reviewed-by: Maciej Fijalkowski <[email protected]> Signed-off-by: David S. Miller <[email protected]>
netfilter: nf_tables: incorrect error path handling with NFT_MSG_NEWRULE
In case of error when adding a new rule that refers to an anonymous set,
deactivate expressions via NFT_TRANS_PREPARE state, not NFT_TRANS_RELEASE.
Thus, the lookup expression marks anonymous sets as inactive in the next
generation to ensure it is not reachable in this transaction anymore and
decrement the set refcount as introduced by c1592a89942e ("netfilter:
nf_tables: deactivate anonymous set from preparation phase"). The abort
step takes care of undoing the anonymous set.
This is also consistent with rule deletion, where NFT_TRANS_PREPARE is
used. Note that this error path is exercised in the preparation step of
the commit protocol. This patch replaces nf_tables_rule_release() by the
deactivate and destroy calls, this time with NFT_TRANS_PREPARE.
Due to this incorrect error handling, it is possible to access a
dangling pointer to the anonymous set that remains in the transaction
list.
Mike Christie [Wed, 7 Jun 2023 19:23:38 +0000 (14:23 -0500)]
vhost: Fix worker hangs due to missed wake up calls
We can race where we have added work to the work_list, but
vhost_task_fn has passed that check but not yet set us into
TASK_INTERRUPTIBLE. wake_up_process will see us in TASK_RUNNING and
just return.
This bug was intoduced in commit f9010dbdce91 ("fork, vhost: Use
CLONE_THREAD to fix freezer/ps regression") when I moved the setting
of TASK_INTERRUPTIBLE to simplfy the code and avoid get_signal from
logging warnings about being in the wrong state. This moves the setting
of TASK_INTERRUPTIBLE back to before we test if we need to stop the
task to avoid a possible race there as well. We then have vhost_worker
set TASK_RUNNING if it finds work similar to before.
Mike Christie [Wed, 7 Jun 2023 19:23:37 +0000 (14:23 -0500)]
vhost: Fix crash during early vhost_transport_send_pkt calls
If userspace does VHOST_VSOCK_SET_GUEST_CID before VHOST_SET_OWNER we
can race where:
1. thread0 calls vhost_transport_send_pkt -> vhost_work_queue
2. thread1 does VHOST_SET_OWNER which calls vhost_worker_create.
3. vhost_worker_create will set the dev->worker pointer before setting
the worker->vtsk pointer.
4. thread0's vhost_work_queue will see the dev->worker pointer is
set and try to call vhost_task_wake using not yet set worker->vtsk
pointer.
5. We then crash since vtsk is NULL.
Before commit 6e890c5d5021 ("vhost: use vhost_tasks for worker
threads"), we only had the worker pointer so we could just check it to
see if VHOST_SET_OWNER has been done. After that commit we have the
vhost_worker and vhost_task pointer, so we can now hit the bug above.
This patch embeds the vhost_worker in the vhost_dev and moves the work
list initialization back to vhost_dev_init, so we can just check the
worker.vtsk pointer to check if VHOST_SET_OWNER has been done like
before.
vhost_net: revert upend_idx only on retriable error
Fix possible virtqueue used buffers leak and corresponding stuck
in case of temporary -EIO from sendmsg() which is produced by
tun driver while backend device is not up.
In case of no-retriable error and zcopy do not revert upend_idx
to pass packet data (that is update used_idx in corresponding
vhost_zerocopy_signal_used()) as if packet data has been
transferred successfully.
v2: set vq->heads[ubuf->desc].len equal to VHOST_DMA_DONE_LEN
in case of fake successful transmit.
Shannon Nelson [Mon, 24 Apr 2023 22:50:29 +0000 (15:50 -0700)]
vhost_vdpa: tell vqs about the negotiated
As is done in the net, iscsi, and vsock vhost support, let the vdpa vqs
know about the features that have been negotiated. This allows vhost
to more safely make decisions based on the features, such as when using
PACKED vs split queues.
Dragos Tatulea [Tue, 16 May 2023 09:58:01 +0000 (12:58 +0300)]
vdpa/mlx5: Fix hang when cvq commands are triggered during device unregister
Currently the vdpa device is unregistered after the workqueue that
processes vq commands is disabled. However, the device unregister
process can still send commands to the cvq (a vlan delete for example)
which leads to a hang because the handing workqueue has been disabled
and the command never finishes:
The fix is to disable and destroy the workqueue after the device
unregister. It is expected that vhost will not trigger kicks after
the unregister. But even if it would, the wq is disabled already by
setting the pointer to NULL (done so in the referenced commit).
Sheng Zhao [Tue, 30 May 2023 03:36:26 +0000 (11:36 +0800)]
vduse: avoid empty string for dev name
Syzkaller hits a kernel WARN when the first character of the dev name
provided is NULL. Solution is to add a NULL check before calling
cdev_device_add() in vduse_create_dev().
Kamil Maziarz [Tue, 6 Jun 2023 10:33:58 +0000 (12:33 +0200)]
ice: Fix XDP memory leak when NIC is brought up and down
Fix the buffer leak that occurs while switching
the port up and down with traffic and XDP by
checking for an active XDP program and freeing all empty TX buffers.
Fixes: efc2214b6047 ("ice: Add support for XDP") Signed-off-by: Kamil Maziarz <[email protected]> Tested-by: Chandan Kumar Rout <[email protected]> (A Contingent Worker at Intel) Acked-by: Maciej Fijalkowski <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
As riscv do not select ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE,
the +0($arg2) addr is processed as a kernel address though it is a
userspace address, cause the above filename=(fault) print. So select
ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE to avoid the issue, after that the
kprobe trace is ok as below:
Linus Torvalds [Thu, 8 Jun 2023 15:46:58 +0000 (08:46 -0700)]
Merge tag 'xfs-6.4-rc5-fixes' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux
Pull xfs fixes from Dave Chinner:
"These are a set of regression fixes discovered on recent kernels. I
was hoping to send this to you a week and half ago, but events out of
my control delayed finalising the changes until early this week.
Whilst the diffstat looks large for this stage of the merge window, a
large chunk of it comes from moving the guts of one function from one
file to another i.e. it's the same code, it is just run in a different
context where it is safe to hold a specific lock. Otherwise the
individual changes are relatively small and straigtht forward.
Summary:
- Propagate unlinked inode list corruption back up to log recovery
(regression fix)
- improve corruption detection for AGFL entries, AGFL indexes and
XEFI extents (syzkaller fuzzer oops report)
- Improve extent merging detection in scrub (regression fix)
- Fix a new undefined high bit shift (regression fix)
- Fix for AGF vs inode cluster buffer deadlock (regression fix)"
* tag 'xfs-6.4-rc5-fixes' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
xfs: collect errors from inodegc for unlinked inode recovery
xfs: validate block number being freed before adding to xefi
xfs: validity check agbnos on the AGFL
xfs: fix agf/agfl verification on v4 filesystems
xfs: fix double xfs_perag_rele() in xfs_filestream_pick_ag()
xfs: fix broken logic when detecting mergeable bmap records
xfs: Fix undefined behavior of shift into sign bit
xfs: fix AGF vs inode cluster buffer deadlock
xfs: defered work could create precommits
xfs: restore allocation trylock iteration
xfs: buffer pins need to hold a buffer reference
Theodore Ts'o [Thu, 8 Jun 2023 14:06:40 +0000 (10:06 -0400)]
ext4: only check dquot_initialize_needed() when debugging
ext4_xattr_block_set() relies on its caller to call dquot_initialize()
on the inode. To assure that this has happened there are WARN_ON
checks. Unfortunately, this is subject to false positives if there is
an antagonist thread which is flipping the file system at high rates
between r/o and rw. So only do the check if EXT4_XATTR_DEBUG is
enabled.
Qu Wenruo [Tue, 6 Jun 2023 07:08:28 +0000 (15:08 +0800)]
btrfs: scrub: also report errors hit during the initial read
[BUG]
After the recent scrub rework introduced in commit e02ee89baa66 ("btrfs:
scrub: switch scrub_simple_mirror() to scrub_stripe infrastructure"),
btrfs scrub no longer reports repaired errors any more:
[CAUSE]
In the reworked scrub, we update the scrub progress inside
scrub_stripe_report_errors(), using various bitmaps to update the
result.
For example for csum_errors, we use bitmap_weight() of
stripe->csum_error_bitmap.
Unfortunately at that stage, all error bitmaps (except
init_error_bitmap) are the result of the latest repair attempt, thus if
the stripe is fully repaired, those error bitmaps will all be empty,
resulting the above output mismatch.
To fix this, record the number of errors into stripe->init_nr_*_errors.
Since we don't really care about where those errors are, we only need to
record the number of errors.
Then in scrub_stripe_report_errors(), use those initial numbers to
update the progress other than using the latest error bitmaps.
Fixes: e02ee89baa66 ("btrfs: scrub: switch scrub_simple_mirror() to scrub_stripe infrastructure") Signed-off-by: Qu Wenruo <[email protected]> Reviewed-by: David Sterba <[email protected]> Signed-off-by: David Sterba <[email protected]>
Scrub device $dev2 (id 2) done
Scrub started: Tue Jun 6 09:59:14 2023
Status: finished
Duration: 0:00:00
[...]
corrected_errors: 16 <<< Still has corrupted sectors
last_physical: 1351614464
# btrfs scrub start -BrRd $mnt
Scrub device $dev1 (id 1) done
Scrub started: Tue Jun 6 10:00:17 2023
Status: finished
Duration: 0:00:00
[...]
corrected_errors: 0 <<< No more errors
last_physical: 1372585984
Scrub device $dev2 (id 2) done
[...]
corrected_errors: 0 <<< No more errors
last_physical: 1372585984
[CAUSE]
In the newly reworked scrub code, repair is always submitted no matter
if we're doing a read-only scrub.
[FIX]
Fix it by skipping the write submission if the scrub is a read-only one.
Unfortunately for the report part, even for a read-only scrub we will
still report it as corrected errors, as we know it's repairable, even we
won't really submit the write.
Fixes: e02ee89baa66 ("btrfs: scrub: switch scrub_simple_mirror() to scrub_stripe infrastructure") Signed-off-by: Qu Wenruo <[email protected]> Reviewed-by: David Sterba <[email protected]> Signed-off-by: David Sterba <[email protected]>
Charles Keepax [Fri, 2 Jun 2023 10:11:36 +0000 (11:11 +0100)]
soundwire: stream: Add missing clear of alloc_slave_rt
The current path that skips allocating the slave runtime does not clear
the alloc_slave_rt flag, this is clearly incorrect. Add the missing
clear, so the runtime won't be erroneously cleaned up.
As per the new udp tunnel framework, drivers which need to know the
details of a port entry (i.e. port type) when it gets deleted should
use the .set_port / .unset_port callbacks.
Implementing the current .udp_tunnel_sync callback would mean that the
deleted tunnel port entry would be all zeros. This used to work on
older firmware because it would not check the input when deleting a
tunnel port. With newer firmware, the delete will now fail and
subsequent tunnel port allocation will fail as a result.
Pavan Chebbi [Wed, 7 Jun 2023 07:54:08 +0000 (00:54 -0700)]
bnxt_en: Prevent kernel panic when receiving unexpected PHC_UPDATE event
The firmware can send PHC_RTC_UPDATE async event on a PF that may not
have PTP registered. In such a case, there will be a null pointer
deference for bp->ptp_cfg when we try to handle the event.
Fix it by not registering for this event with the firmware if !bp->ptp_cfg.
Also, check that bp->ptp_cfg is valid before proceeding when we receive
the event.
Fixes: 8bcf6f04d4a5 ("bnxt_en: Handle async event when the PHC is updated in RTC mode") Signed-off-by: Pavan Chebbi <[email protected]> Signed-off-by: Michael Chan <[email protected]> Signed-off-by: Paolo Abeni <[email protected]>
Vikas Gupta [Wed, 7 Jun 2023 07:54:07 +0000 (00:54 -0700)]
bnxt_en: Skip firmware fatal error recovery if chip is not accessible
Driver starts firmware fatal error recovery by detecting
heartbeat failure or fw reset count register changing. But
these checks are not reliable if the device is not accessible.
This can happen while DPC (Downstream Port containment) is in
progress. Skip firmware fatal recovery if pci_device_is_present()
returns false.
Somnath Kotur [Wed, 7 Jun 2023 07:54:06 +0000 (00:54 -0700)]
bnxt_en: Query default VLAN before VNIC setup on a VF
We need to call bnxt_hwrm_func_qcfg() on a VF to query the default
VLAN that may be setup by the PF. If a default VLAN is enabled,
the VF cannot support VLAN acceleration on the receive side and
the VNIC must be setup to strip out the default VLAN tag. If a
default VLAN is not enabled, the VF can support VLAN acceleration
on the receive side. The VNIC should be set up to strip or not
strip the VLAN based on the RX VLAN acceleration setting.
Without this call to determine the default VLAN before calling
bnxt_setup_vnic(), the VNIC may not be set up correctly. For
example, bnxt_setup_vnic() may set up to strip the VLAN tag based
on stale default VLAN information. If RX VLAN acceleration is
not enabled, the VLAN tag will be incorrectly stripped and the
RX data path will not work correctly.
Fixes: cf6645f8ebc6 ("bnxt_en: Add function for VF driver to query default VLAN.") Reviewed-by: Pavan Chebbi <[email protected]> Signed-off-by: Somnath Kotur <[email protected]> Signed-off-by: Michael Chan <[email protected]> Signed-off-by: Paolo Abeni <[email protected]>
Sreekanth Reddy [Wed, 7 Jun 2023 07:54:05 +0000 (00:54 -0700)]
bnxt_en: Don't issue AP reset during ethtool's reset operation
Only older NIC controller's firmware uses the PROC AP reset type.
Firmware on 5731X/5741X and newer chips does not support this reset
type. When bnxt_reset() issues a series of resets, this PROC AP
reset may actually fail on these newer chips because the firmware
is not ready to accept this unsupported command yet. Avoid this
unnecessary error by skipping this reset type on chips that don't
support it.
selftests/bpf: Add test cases to assert proper ID tracking on spill
The previous commit fixed a verifier bypass by ensuring that ID is not
preserved on narrowing spills. Add the test cases to check the
problematic patterns.
The following scenario describes a bug in the verifier where it
incorrectly concludes about equivalent scalar IDs which could lead to
verifier bypass in privileged mode:
1. Prepare a 32-bit rogue number.
2. Put the rogue number into the upper half of a 64-bit register, and
roll a random (unknown to the verifier) bit in the lower half. The
rest of the bits should be zero (although variations are possible).
3. Assign an ID to the register by MOVing it to another arbitrary
register.
4. Perform a 32-bit spill of the register, then perform a 32-bit fill to
another register. Due to a bug in the verifier, the ID will be
preserved, although the new register will contain only the lower 32
bits, i.e. all zeros except one random bit.
At this point there are two registers with different values but the same
ID, which means the integrity of the verifier state has been corrupted.
5. Compare the new 32-bit register with 0. In the branch where it's
equal to 0, the verifier will believe that the original 64-bit
register is also 0, because it has the same ID, but its actual value
still contains the rogue number in the upper half.
Some optimizations of the verifier prevent the actual bypass, so
extra care is needed: the comparison must be between two registers,
and both branches must be reachable (this is why one random bit is
needed). Both branches are still suitable for the bypass.
6. Right shift the original register by 32 bits to pop the rogue number.
7. Use the rogue number as an offset with any pointer. The verifier will
believe that the offset is 0, while in reality it's the given number.
The fix is similar to the 32-bit BPF_MOV handling in check_alu_op for
SCALAR_VALUE. If the spill is narrowing the actual register value, don't
keep the ID, make sure it's reset to 0.
Randy Dunlap [Thu, 8 Jun 2023 02:54:24 +0000 (19:54 -0700)]
eeprom: at24: also select REGMAP
Selecting only REGMAP_I2C can leave REGMAP unset, causing build errors,
so also select REGMAP to prevent the build errors.
../drivers/misc/eeprom/at24.c:540:42: warning: 'struct regmap_config' declared inside parameter list will not be visible outside of this definition or declaration
540 | struct regmap_config *regmap_config)
../drivers/misc/eeprom/at24.c: In function 'at24_make_dummy_client':
../drivers/misc/eeprom/at24.c:552:18: error: implicit declaration of function 'devm_regmap_init_i2c' [-Werror=implicit-function-declaration]
552 | regmap = devm_regmap_init_i2c(dummy_client, regmap_config);
../drivers/misc/eeprom/at24.c:552:16: warning: assignment to 'struct regmap *' from 'int' makes pointer from integer without a cast [-Wint-conversion]
552 | regmap = devm_regmap_init_i2c(dummy_client, regmap_config);
../drivers/misc/eeprom/at24.c: In function 'at24_probe':
../drivers/misc/eeprom/at24.c:586:16: error: variable 'regmap_config' has initializer but incomplete type
586 | struct regmap_config regmap_config = { };
../drivers/misc/eeprom/at24.c:586:30: error: storage size of 'regmap_config' isn't known
586 | struct regmap_config regmap_config = { };
../drivers/misc/eeprom/at24.c:586:30: warning: unused variable 'regmap_config' [-Wunused-variable]
Xiubo Li [Thu, 1 Jun 2023 00:59:31 +0000 (08:59 +0800)]
ceph: fix use-after-free bug for inodes when flushing capsnaps
There is a race between capsnaps flush and removing the inode from
'mdsc->snap_flush_list' list:
== Thread A == == Thread B ==
ceph_queue_cap_snap()
-> allocate 'capsnapA'
->ihold('&ci->vfs_inode')
->add 'capsnapA' to 'ci->i_cap_snaps'
->add 'ci' to 'mdsc->snap_flush_list'
...
== Thread C ==
ceph_flush_snaps()
->__ceph_flush_snaps()
->__send_flush_snap()
handle_cap_flushsnap_ack()
->iput('&ci->vfs_inode')
this also will release 'ci'
...
== Thread D ==
ceph_handle_snap()
->flush_snaps()
->iterate 'mdsc->snap_flush_list'
->get the stale 'ci'
->remove 'ci' from ->ihold(&ci->vfs_inode) this
'mdsc->snap_flush_list' will WARNING
To fix this we will increase the inode's i_count ref when adding 'ci'
to the 'mdsc->snap_flush_list' list.
accel/ivpu: Do not trigger extra VPU reset if the VPU is idle
Turning off the PLL and entering D0i3 will reset the VPU so
an explicit IP reset is redundant.
But if the VPU is active, it may interfere with PLL disabling
and to avoid that, we have to issue an additional IP reset
to silence the VPU before turning off the PLL.
Jakub Kicinski [Thu, 8 Jun 2023 04:56:01 +0000 (21:56 -0700)]
Merge tag 'batadv-net-pullrequest-20230607' of git://git.open-mesh.org/linux-merge
Simon Wunderlich says:
====================
Here is a batman-adv bugfix:
- fix a broken sync while rescheduling delayed work,
by Vladislav Efanov
* tag 'batadv-net-pullrequest-20230607' of git://git.open-mesh.org/linux-merge:
batman-adv: Broken sync while rescheduling delayed work
====================
Jakub Kicinski [Thu, 8 Jun 2023 04:47:11 +0000 (21:47 -0700)]
Merge tag 'for-netdev' of https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf
Daniel Borkmann says:
====================
pull-request: bpf 2023-06-07
We've added 7 non-merge commits during the last 7 day(s) which contain
a total of 12 files changed, 112 insertions(+), 7 deletions(-).
The main changes are:
1) Fix a use-after-free in BPF's task local storage, from KP Singh.
2) Make struct path handling more robust in bpf_d_path, from Jiri Olsa.
3) Fix a syzbot NULL-pointer dereference in sockmap, from Eric Dumazet.
4) UAPI fix for BPF_NETFILTER before final kernel ships,
from Florian Westphal.
5) Fix map-in-map array_map_gen_lookup code generation where elem_size was
not being set for inner maps, from Rhys Rustad-Elliott.
6) Fix sockopt_sk selftest's NETLINK_LIST_MEMBERSHIPS assertion,
from Yonghong Song.
* tag 'for-netdev' of https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf:
bpf: Add extra path pointer check to d_path helper
selftests/bpf: Fix sockopt_sk selftest
bpf: netfilter: Add BPF_NETFILTER bpf_attach_type
selftests/bpf: Add access_inner_map selftest
bpf: Fix elem_size not being set for inner maps
bpf: Fix UAF in task local storage
bpf, sockmap: Avoid potential NULL dereference in sk_psock_verdict_data_ready()
====================
Ben Hutchings [Fri, 2 Jun 2023 18:28:15 +0000 (20:28 +0200)]
lib: cpu_rmap: Fix potential use-after-free in irq_cpu_rmap_release()
irq_cpu_rmap_release() calls cpu_rmap_put(), which may free the rmap.
So we need to clear the pointer to our glue structure in rmap before
doing that, not after.
netfilter: nf_tables: integrate pipapo into commit protocol
The pipapo set backend follows copy-on-update approach, maintaining one
clone of the existing datastructure that is being updated. The clone
and current datastructures are swapped via rcu from the commit step.
The existing integration with the commit protocol is flawed because
there is no operation to clean up the clone if the transaction is
aborted. Moreover, the datastructure swap happens on set element
activation.
This patch adds two new operations for sets: commit and abort, these new
operations are invoked from the commit and abort steps, after the
transactions have been digested, and it updates the pipapo set backend
to use it.
This patch adds a new ->pending_update field to sets to maintain a list
of sets that require this new commit and abort operations.
Fixes: 3c4287f62044 ("nf_tables: Add set type for arbitrary concatenation of ranges") Signed-off-by: Pablo Neira Ayuso <[email protected]>
Chia-I Wu [Thu, 1 Jun 2023 21:48:08 +0000 (14:48 -0700)]
drm/amdgpu: fix xclk freq on CHIP_STONEY
According to Alex, most APUs from that time seem to have the same issue
(vbios says 48Mhz, actual is 100Mhz). I only have a CHIP_STONEY so I
limit the fixup to CHIP_STONEY
Min Li [Sat, 3 Jun 2023 07:43:45 +0000 (15:43 +0800)]
drm/radeon: fix race condition UAF in radeon_gem_set_domain_ioctl
Userspace can race to free the gobj(robj converted from), robj should not
be accessed again after drm_gem_object_put, otherwith it will result in
use-after-free.
Samson Tam [Tue, 9 May 2023 20:40:19 +0000 (16:40 -0400)]
drm/amd/display: add ODM case when looking for first split pipe
[Why]
When going from ODM 2:1 single display case to max displays, second
odm pipe needs to be repurposed for one of the new single displays.
However, acquire_first_split_pipe() only handles MPC case and not
ODM case
[How]
Add ODM conditions in acquire_first_split_pipe()
Add commit_minimal_transition_state() in commit_streams() to handle
odm 2:1 exit first, and then process new streams
Handle ODM condition in commit_minimal_transition_state()
drm/amd: Make lack of `ACPI_FADT_LOW_POWER_S0` or `CONFIG_AMD_PMC` louder during suspend path
Users have reported that s2idle wasn't working on OEM Phoenix systems,
but it was root caused to be because `CONFIG_AMD_PMC` wasn't set in
the distribution kernel config.
To make this more apparent, raise the messaging to err instead of warn.
Horatio Zhang [Mon, 29 May 2023 18:23:37 +0000 (14:23 -0400)]
drm/amdgpu: fix Null pointer dereference error in amdgpu_device_recover_vram
Use the function of amdgpu_bo_vm_destroy to handle the resource release
of shadow bo. During the amdgpu_mes_self_test, shadow bo released, but
vmbo->shadow_list was not, which caused a null pointer reference error
in amdgpu_device_recover_vram when GPU reset.
Fixes: 6c032c37ac3e ("drm/amdgpu: Fix vram recover doesn't work after whole GPU reset (v2)") Signed-off-by: xinhui pan <[email protected]> Signed-off-by: Horatio Zhang <[email protected]> Acked-by: Feifei Xu <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
commit cf488dcd0ab7 ("drm/amd: Allow s0ix without BIOS support") showed
improvements to power consumption over suspend when s0ix wasn't enabled in
BIOS and the system didn't support S3.
This patch however was misguided because the reason the system didn't
support S3 was because SMT was disabled in OEM BIOS setup.
This prevented the BIOS from allowing S3.
Also allowing GPUs to use the s2idle path actually causes problems if
they're invoked on systems that may not support s2idle in the platform
firmware. `systemd` has a tendency to try to use `s2idle` if `deep` fails
for any reason, which could lead to unexpected flows.
The original commit also fixed a problem during resume from suspend to idle
without hardware support, but this is no longer necessary with commit ca4751866397 ("drm/amd: Don't allow s0ix on APUs older than Raven")
Revert commit cf488dcd0ab7 ("drm/amd: Allow s0ix without BIOS support")
to make it match the expected behavior again.
Chen-Yu Tsai [Fri, 26 May 2023 07:35:37 +0000 (15:35 +0800)]
notifier: Initialize new struct srcu_usage field
In commit 95433f726301 ("srcu: Begin offloading srcu_struct fields to
srcu_update"), a new struct srcu_usage field was added, but was not
properly initialized. This led to a "spinlock bad magic" BUG when the
SRCU notifier was ever used. This was observed in the MediaTek CCI
devfreq driver on next-20230525. The trimmed stack trace is as follows:
Thomas Gleixner [Tue, 6 Jun 2023 23:14:05 +0000 (01:14 +0200)]
MAINTAINERS: Add entry for debug objects
This is overdue and an oversight.
Add myself to this file deespite the fact that I'm trying to reduce the
number of entries in this file which have my name attached, but in the
hope that patches wont get picked up elsewhere completely unreviewed and
unnoticed.
Update the fallback string for the RTL8821CS from realtek,rtl8822cs-bt
to realtek,rtl8723bs-bt. The difference between these two strings is
that the 8822cs enables power saving features that the 8723bs does not,
and in testing the 8821cs seems to have issues with these power saving
modes enabled.
David Howells [Wed, 7 Jun 2023 08:47:13 +0000 (09:47 +0100)]
afs: Fix setting of mtime when creating a file/dir/symlink
kafs incorrectly passes a zero mtime (ie. 1st Jan 1970) to the server when
creating a file, dir or symlink because the mtime recorded in the
afs_operation struct gets passed to the server by the marshalling routines,
but the afs_mkdir(), afs_create() and afs_symlink() functions don't set it.
This gets masked if a file or directory is subsequently modified.
Fix this by filling in op->mtime before calling the create op.
Oliver Upton [Tue, 6 Jun 2023 18:48:14 +0000 (18:48 +0000)]
KVM: arm64: Use raw_smp_processor_id() in kvm_pmu_probe_armpmu()
Sebastian reports that commit 1c913a1c35aa ("KVM: arm64: Iterate
arm_pmus list to probe for default PMU") introduced the following splat
with CONFIG_DEBUG_PREEMPT enabled:
Fix the issue by using the raw variant that bypasses the debug
assertion. While at it, stick all of the nuance and UAPI baggage into a
comment for posterity.
Marc Zyngier [Wed, 7 Jun 2023 14:38:44 +0000 (15:38 +0100)]
KVM: arm64: Restore GICv2-on-GICv3 functionality
When reworking the vgic locking, the vgic distributor registration
got simplified, which was a very good cleanup. But just a tad too
radical, as we now register the *native* vgic only, ignoring the
GICv2-on-GICv3 that allows pre-historic VMs (or so I thought)
to run.
As it turns out, QEMU still defaults to GICv2 in some cases, and
this breaks Nathan's setup!
Fix it by propagating the *requested* vgic type rather than the
host's version.
Alexandre Ghiti [Wed, 7 Jun 2023 12:58:51 +0000 (14:58 +0200)]
riscv: Check the virtual alignment before choosing a map size
We used to only check the alignment of the physical address to decide
which mapping would fit for a certain region of the linear mapping, but
it is not enough since the virtual address must also be aligned, so check
that too.
Alexandre Ghiti [Tue, 6 Jun 2023 13:04:44 +0000 (15:04 +0200)]
riscv: Fix kfence now that the linear mapping can be backed by PUD/P4D/PGD
RISC-V Kfence implementation used to rely on the fact the linear mapping
was backed by at most PMD hugepages, which is not true anymore since
commit 3335068f8721 ("riscv: Use PUD/P4D/PGD pages for the linear
mapping").
Instead of splitting PUD/P4D/PGD mappings afterwards, directly map the
kfence pool region using PTE mappings by allocating this region before
setup_vm_final().
riscv: mm: Ensure prot of VM_WRITE and VM_EXEC must be readable
Commit 8aeb7b17f04e ("RISC-V: Make mmap() with PROT_WRITE imply PROT_READ")
allows riscv to use mmap with PROT_WRITE only, and meanwhile mmap with w+x
is also permitted. However, when userspace tries to access this page with
PROT_WRITE|PROT_EXEC, which causes infinite loop at load page fault as
well as it triggers soft lockup. According to riscv privileged spec,
"Writable pages must also be marked readable". The fix to drop the
`PAGE_COPY_READ_EXEC` and then `PAGE_COPY_EXEC` would be just used instead.
This aligns the other arches (i.e arm64) for protection_map.
It's problem for all stable trees with bpf_d_path helper, which was
added in 5.9.
This issue is fixed in current bpf code, where we identify and mark
trusted pointers, so the above code would fail even to load.
For the sake of the stable trees and to workaround potentially broken
verifier in the future, adding the code that reads the path object from
the passed pointer and verifies it's valid in kernel space.
Kent Gibson [Wed, 7 Jun 2023 06:50:04 +0000 (14:50 +0800)]
gpio: sim: quietly ignore configured lines outside the bank
The user-space policy of the gpio-sim is that configuration for lines
with offsets outside the bounds of the corresponding bank is ignored,
but gpio-sim is still using that configuration when constructing the
sim. In the case of named lines this results in temporarily allocating
space for names that are not used, and for hogs results in errors being
logged when the gpio-sim attempts to register the out of range hog with
gpiolib:
gpiochip_machine_hog: unable to get GPIO desc: -22
Add checks to filter out any line configuration outside the bounds
of the bank when constructing the sim.
Hangyu Hua [Wed, 7 Jun 2023 02:23:01 +0000 (10:23 +0800)]
net: sched: fix possible refcount leak in tc_chain_tmplt_add()
try_module_get will be called in tcf_proto_lookup_ops. So module_put needs
to be called to drop the refcount if ops don't implement the required
function.
Fixes: 9f407f1768d3 ("net: sched: introduce chain templates") Signed-off-by: Hangyu Hua <[email protected]> Reviewed-by: Larysa Zaremba <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Eric Dumazet [Tue, 6 Jun 2023 13:13:04 +0000 (13:13 +0000)]
net: sched: act_police: fix sparse errors in tcf_police_dump()
Fixes following sparse errors:
net/sched/act_police.c:360:28: warning: dereference of noderef expression
net/sched/act_police.c:362:45: warning: dereference of noderef expression
net/sched/act_police.c:362:45: warning: dereference of noderef expression
net/sched/act_police.c:368:28: warning: dereference of noderef expression
net/sched/act_police.c:370:45: warning: dereference of noderef expression
net/sched/act_police.c:370:45: warning: dereference of noderef expression
net/sched/act_police.c:376:45: warning: dereference of noderef expression
net/sched/act_police.c:376:45: warning: dereference of noderef expression
Fixes: d1967e495a8d ("net_sched: act_police: add 2 new attributes to support police 64bit rate and peakrate") Signed-off-by: Eric Dumazet <[email protected]> Reviewed-by: Simon Horman <[email protected]> Acked-by: Jamal Hadi Salim <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Eelco Chaudron [Tue, 6 Jun 2023 11:56:35 +0000 (13:56 +0200)]
net: openvswitch: fix upcall counter access before allocation
Currently, the per cpu upcall counters are allocated after the vport is
created and inserted into the system. This could lead to the datapath
accessing the counters before they are allocated resulting in a kernel
Oops.
Eric Dumazet [Tue, 6 Jun 2023 11:42:33 +0000 (11:42 +0000)]
net: sched: move rtm_tca_policy declaration to include file
rtm_tca_policy is used from net/sched/sch_api.c and net/sched/cls_api.c,
thus should be declared in an include file.
This fixes the following sparse warning:
net/sched/sch_api.c:1434:25: warning: symbol 'rtm_tca_policy' was not declared. Should it be static?
Fixes: e331473fee3d ("net/sched: cls_api: add missing validation of netlink attributes") Signed-off-by: Eric Dumazet <[email protected]> Acked-by: Jamal Hadi Salim <[email protected]> Signed-off-by: David S. Miller <[email protected]>
i2c: sprd: Delete i2c adapter in .remove's error path
If pm runtime resume fails the .remove callback used to exit early. This
resulted in an error message by the driver core but the device gets
removed anyhow. This lets the registered i2c adapter stay around with an
unbound parent device.
So only skip clk disabling if resume failed, but do delete the adapter.
Michal Schmidt [Tue, 6 Jun 2023 17:12:53 +0000 (10:12 -0700)]
ice: make writes to /dev/gnssX synchronous
The current ice driver's GNSS write implementation buffers writes and
works through them asynchronously in a kthread. That's bad because:
- The GNSS write_raw operation is supposed to be synchronous[1][2].
- There is no upper bound on the number of pending writes.
Userspace can submit writes much faster than the driver can process,
consuming unlimited amounts of kernel memory.
A patch that's currently on review[3] ("[v3,net] ice: Write all GNSS
buffers instead of first one") would add one more problem:
- The possibility of waiting for a very long time to flush the write
work when doing rmmod, softlockups.
To fix these issues, simplify the implementation: Drop the buffering,
the write_work, and make the writes synchronous.
I tested this with gpsd and ubxtool.
[1] https://events19.linuxfoundation.org/wp-content/uploads/2017/12/The-GNSS-Subsystem-Johan-Hovold-Hovold-Consulting-AB.pdf
"User interface" slide.
[2] A comment in drivers/gnss/core.c:gnss_write():
/* Ignoring O_NONBLOCK, write_raw() is synchronous. */
[3] https://patchwork.ozlabs.org/project/intel-wired-lan/patch/20230217120541[email protected]/
Tvrtko Ursulin [Mon, 5 Jun 2023 13:11:35 +0000 (14:11 +0100)]
drm/i915/selftests: Add some missing error propagation
Add some missing error propagation in live_parallel_switch.
To avoid needlessly burdening the various backport processes, note I am
not marking it as a fix against any patches and not copying stable since
it is debug/selftests only code.
Reported by Kernel Concurrency Sanitizer on:
CPU: 0 PID: 6912 Comm: syz-executor.5 Not tainted 6.4.0-rc3-syzkaller-00190-g0d85b27b0cc6 #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 05/16/2023
Jakub Kicinski [Wed, 7 Jun 2023 04:36:57 +0000 (21:36 -0700)]
Merge tag 'for-net-2023-06-05' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetooth/bluetooth
Luiz Augusto von Dentz says:
====================
bluetooth pull request for net:
- Fixes to debugfs registration
- Fix use-after-free in hci_remove_ltk/hci_remove_irk
- Fixes to ISO channel support
- Fix missing checks for invalid L2CAP DCID
- Fix l2cap_disconnect_req deadlock
- Add lock to protect HCI_UNREGISTER
* tag 'for-net-2023-06-05' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetooth/bluetooth:
Bluetooth: L2CAP: Add missing checks for invalid DCID
Bluetooth: ISO: use correct CIS order in Set CIG Parameters event
Bluetooth: ISO: don't try to remove CIG if there are bound CIS left
Bluetooth: Fix l2cap_disconnect_req deadlock
Bluetooth: hci_qca: fix debugfs registration
Bluetooth: fix debugfs registration
Bluetooth: hci_sync: add lock to protect HCI_UNREGISTER
Bluetooth: Fix use-after-free in hci_remove_ltk/hci_remove_irk
Bluetooth: ISO: Fix CIG auto-allocation to select configurable CIG
Bluetooth: ISO: consider right CIS when removing CIG at cleanup
====================
projects / linux.git / log