Inki Dae [Thu, 18 May 2023 23:55:05 +0000 (08:55 +0900)]
drm/exynos: vidi: fix a wrong error return
Fix a wrong error return by dropping an error return.
When vidi driver is remvoed, if ctx->raw_edid isn't same as fake_edid_info
then only what we have to is to free ctx->raw_edid so that driver removing
can work correctly - it's not an error case.
The originator of an SRH places the first hop's IPv6 address in the IPv6
header's IPv6 Destination Address and the second hop's IPv6 address as
the first address in Addresses[1..n].
The CmprI and CmprE fields indicate the number of prefix octets that are
shared with the IPv6 Destination Address. When CmprI or CmprE is not 0,
Addresses[1..n] are compressed as follows:
Segments Left indicates the number of route segments remaining. When the
value is not zero, the SRH is forwarded to the next hop. Its address
is extracted from Addresses[n - Segment Left + 1] and swapped with IPv6
Destination Address.
When Segment Left is greater than or equal to 2, the size of SRH is not
changed because Addresses[1..n-1] are decompressed and recompressed with
CmprI.
OTOH, when Segment Left changes from 1 to 0, the new SRH could have a
different size because Addresses[1..n-1] are decompressed with CmprI and
recompressed with CmprE.
Let's say CmprI is 15 and CmprE is 0. When we receive SRH with Segment
Left >= 2, Addresses[1..n-1] have 1 byte for each, and Addresses[n] has
16 bytes. When Segment Left is 1, Addresses[1..n-1] is decompressed to
16 bytes and not recompressed. Finally, the new SRH will need more room
in the header, and the size is (16 - 1) * (n - 1) bytes.
Here the max value of n is 255 as Segment Left is u8, so in the worst case,
we have to allocate 3825 bytes in the skb headroom. However, now we only
allocate a small fixed buffer that is IPV6_RPL_SRH_WORST_SWAP_SIZE (16 + 7
bytes). If the decompressed size overflows the room, skb_push() hits BUG()
below [0].
Instead of allocating the fixed buffer for every packet, let's allocate
enough headroom only when we receive SRH with Segment Left 1.
Erico Nunes [Tue, 6 Jun 2023 14:32:47 +0000 (16:32 +0200)]
drm/lima: fix sched context destroy
The drm sched entity must be flushed before finishing, to account for
jobs potentially still in flight at that time.
Lima did not do this flush until now, so switch the destroy call to the
drm_sched_entity_destroy() wrapper which will take care of that.
This fixes a regression on lima which started since the rework in
commit 2fdb8a8f07c2 ("drm/scheduler: rework entity flush, kill and fini")
where some specific types of applications may hang indefinitely.
netfilter: ipset: Add schedule point in call_ad().
syzkaller found a repro that causes Hung Task [0] with ipset. The repro
first creates an ipset and then tries to delete a large number of IPs
from the ipset concurrently:
The first deleting thread hogs a CPU with nfnl_lock(NFNL_SUBSYS_IPSET)
held, and other threads wait for it to be released.
Previously, the same issue existed in set->variant->uadt() that could run
so long under ip_set_lock(set). Commit 5e29dc36bd5e ("netfilter: ipset:
Rework long task execution when adding/deleting entries") tried to fix it,
but the issue still exists in the caller with another mutex.
While adding/deleting many IPs, we should release the CPU periodically to
prevent someone from abusing ipset to hang the system.
Note we need to increment the ipset's refcnt to prevent the ipset from
being destroyed while rescheduling.
netfilter: conntrack: fix NULL pointer dereference in nf_confirm_cthelper
An nf_conntrack_helper from nf_conn_help may become NULL after DNAT.
Observed when TCP port 1720 (Q931_PORT), associated with h323 conntrack
helper, is DNAT'ed to another destination port (e.g. 1730), while
nfqueue is being used for final acceptance (e.g. snort).
This happenned after transition from kernel 4.14 to 5.10.161.
Workarounds:
* keep the same port (1720) in DNAT
* disable nfqueue
* disable/unload h323 NAT helper
Jeremy Sowden [Thu, 25 May 2023 14:07:24 +0000 (15:07 +0100)]
netfilter: nft_bitwise: fix register tracking
At the end of `nft_bitwise_reduce`, there is a loop which is intended to
update the bitwise expression associated with each tracked destination
register. However, currently, it just updates the first register
repeatedly. Fix it.
Fixes: 34cc9e52884a ("netfilter: nf_tables: cancel tracking for clobbered destination registers") Signed-off-by: Jeremy Sowden <[email protected]> Signed-off-by: Florian Westphal <[email protected]> Signed-off-by: Pablo Neira Ayuso <[email protected]>
In net/netlink/af_netlink.c, function netlink_getsockopt(), for NETLINK_LIST_MEMBERSHIPS,
nlk->ngroups equals to 36. Before Commit f4e4534850a9, the optlen is calculated as
ALIGN(nlk->ngroups / 8, sizeof(u32)) = 4
After that commit, the optlen is
ALIGN(BITS_TO_BYTES(nlk->ngroups), sizeof(u32)) = 8
Fix the test by setting the expected optlen to be 8.
Chris Mason [Mon, 5 Jun 2023 19:03:15 +0000 (12:03 -0700)]
btrfs: properly enable async discard when switching from RO->RW
The async discard uses the BTRFS_FS_DISCARD_RUNNING bit in the fs_info
to force discards off when the filesystem has aborted or we're generally
not able to run discards. This gets flipped on when we're mounted rw,
and also when we go from ro->rw.
Commit 63a7cb13071842 ("btrfs: auto enable discard=async when possible")
enabled async discard by default, and this meant
"mount -o ro /dev/xxx /yyy" had async discards turned on.
Unfortunately, this meant our check in btrfs_remount_cleanup() would see
that discards are already on:
/* If we toggled discard async */
if (!btrfs_raw_test_opt(old_opts, DISCARD_ASYNC) &&
btrfs_test_opt(fs_info, DISCARD_ASYNC))
btrfs_discard_resume(fs_info);
So, we'd never call btrfs_discard_resume() when remounting the root
filesystem from ro->rw.
drgn shows this really nicely:
import os
import sys
from drgn.helpers.linux.fs import path_lookup
from drgn import NULL, Object, Type, cast
BTRFS_FS_DISCARD_RUNNING = 1 << prog['BTRFS_FS_DISCARD_RUNNING']
if fs_info.flags & BTRFS_FS_DISCARD_RUNNING:
print("discard running flag is on")
else:
print("discard running flag is off")
[root]# mount | grep nvme
/dev/nvme0n1p3 on / type btrfs
(rw,relatime,compress-force=zstd:3,ssd,discard=async,space_cache=v2,subvolid=5,subvol=/)
[root]# ./discard_running.drgn
discard running flag is off
[root]# mount -o remount,discard=sync /
[root]# mount -o remount,discard=async /
[root]# ./discard_running.drgn
discard running flag is on
The fix is to call btrfs_discard_resume() when we're going from ro->rw.
It already checks to make sure the async discard flag is on, so it'll do
the right thing.
Vladimir Oltean [Mon, 29 May 2023 22:34:02 +0000 (01:34 +0300)]
spi: fsl-dspi: avoid SCK glitches with continuous transfers
The DSPI controller has configurable timing for
(a) tCSC: the interval between the assertion of the chip select and the
first clock edge
(b) tASC: the interval between the last clock edge and the deassertion
of the chip select
What is a bit surprising, but is documented in the figure "Example of
continuous transfer (CPHA=1, CONT=1)" in the datasheet, is that when the
chip select stays asserted between multiple TX FIFO writes, the tCSC and
tASC times still apply. With CONT=1, chip select remains asserted, but
SCK takes a break and goes to the idle state for tASC + tCSC ns.
In other words, the default values (of 0 and 0 ns) result in SCK
glitches where the SCK transition to the idle state, as well as the SCK
transition from the idle state, will have no delay in between, and it
may appear that a SCK cycle has simply gone missing. The resulting
timing violation might cause data corruption in many peripherals, as
their chip select is asserted.
The driver has device tree bindings for tCSC ("fsl,spi-cs-sck-delay")
and tASC ("fsl,spi-sck-cs-delay"), but these are only specified to apply
when the chip select toggles in the first place, and this timing
characteristic depends on each peripheral. Many peripherals do not have
explicit timing requirements, so many device trees do not have these
properties present at all.
Nonetheless, the lack of SCK glitches is a common sense requirement, and
since the SCK stays in the idle state during transfers for tCSC+tASC ns,
and that in itself should look like half a cycle, then let's ensure that
tCSC and tASC are at least a quarter of a SCK period, such that their
sum is at least half of one.
Chris Chiu [Tue, 6 Jun 2023 14:57:47 +0000 (22:57 +0800)]
ALSA: hda/realtek: Enable 4 amplifiers instead of 2 on a HP platform
In the commit 7bb62340951a ("ALSA: hda/realtek: fix speaker, mute/micmute
LEDs not work on a HP platform"), speakers and LEDs are fixed but only 2
CS35L41 amplifiers on SPI bus connected to Realtek codec are enabled. Need
the ALC245_FIXUP_CS35L41_SPI_4_HP_GPIO_LED to get all amplifiers working.
Kent Gibson [Tue, 6 Jun 2023 12:00:34 +0000 (20:00 +0800)]
gpio: sim: fix memory corruption when adding named lines and unnamed hogs
When constructing the sim, gpio-sim constructs an array of named lines,
sized based on the largest offset of any named line, and then initializes
that array with the names of all lines, including unnamed hogs with higher
offsets. In doing so it writes NULLs beyond the extent of the array.
Add a check that only named lines are used to initialize the array.
Fixes: cb8c474e79be ("gpio: sim: new testing module") Signed-off-by: Kent Gibson<[email protected]> Signed-off-by: Bartosz Golaszewski <[email protected]>
Linus Torvalds [Tue, 6 Jun 2023 13:18:28 +0000 (06:18 -0700)]
Merge tag 'spi-fix-v6.4-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi
Pull spi fixes from Mark Brown:
"A small collection of driver specific fixes, none of them particularly
remarkable or severe"
* tag 'spi-fix-v6.4-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi:
spi: qup: Request DMA before enabling clocks
spi: mt65xx: make sure operations completed before unloading
spi: lpspi: disable lpspi module irq in DMA mode
Takashi Iwai [Tue, 6 Jun 2023 13:09:42 +0000 (15:09 +0200)]
Merge tag 'asoc-fix-v6.4-rc6' of https://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound into for-linus
ASoC: Fixes for v6.4
A lot of routine driver specific fixes here, nothing in the core though
there are a couple of fixes for the generic cards. There's also a few
new quirks for x86 platforms.
Arnd Bergmann [Tue, 6 Jun 2023 13:05:30 +0000 (15:05 +0200)]
Merge tag 'ffa-fix-6.4-2' of git://git.kernel.org/pub/scm/linux/kernel/git/sudeep.holla/linux into arm/fixes
Arm FF-A fix for v6.4
A single fix addressing another MBZ field being non-zero and non-compliant
resulting in the rejection of certain memory interface transmissions by
the receivers(secure partitions). The issue was addressed before but missed
to address this one field which is part of this change.
* tag 'ffa-fix-6.4-2' of git://git.kernel.org/pub/scm/linux/kernel/git/sudeep.holla/linux:
firmware: arm_ffa: Set handle field to zero in memory descriptor
Linus Torvalds [Tue, 6 Jun 2023 12:49:06 +0000 (05:49 -0700)]
Merge tag 'gfs2-v6.4-rc4-fix' of git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2
Pull gfs2 fix from Andreas Gruenbacher:
- Don't get stuck writing page onto itself under direct I/O
* tag 'gfs2-v6.4-rc4-fix' of git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2:
gfs2: Don't get stuck writing page onto itself under direct I/O
accel/ivpu: Reserve all non-command bo's using DMA_RESV_USAGE_BOOKKEEP
Use DMA_RESV_USAGE_BOOKKEEP reservation for buffer objects, except for
command buffers for which we use DMA_RESV_USAGE_WRITE (since VPU can
write to command buffer context save area).
Linus Torvalds [Tue, 6 Jun 2023 12:42:21 +0000 (05:42 -0700)]
Merge tag 'platform-drivers-x86-v6.4-4' of git://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86
Pull x86 platform driver fixes from Hans de Goede:
- various Microsoft Surface support fixes
- one fix for the INT3472 driver
* tag 'platform-drivers-x86-v6.4-4' of git://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86:
platform/x86: int3472: Avoid crash in unregistering regulator gpio
platform/surface: aggregator_tabletsw: Add support for book mode in POS subsystem
platform/surface: aggregator_tabletsw: Add support for book mode in KIP subsystem
platform/surface: aggregator: Allow completion work-items to be executed in parallel
platform/surface: aggregator: Make to_ssam_device_driver() respect constness
Balint Dobszay [Thu, 1 Jun 2023 14:07:49 +0000 (16:07 +0200)]
firmware: arm_ffa: Set handle field to zero in memory descriptor
As described in the commit 111a833dc5cb ("firmware: arm_ffa: Set
reserved/MBZ fields to zero in the memory descriptors") some fields in
the memory descriptor have to be zeroed explicitly. The handle field is
one of these, but it was left out from that change, fix this now.
Takashi Iwai [Tue, 6 Jun 2023 09:38:55 +0000 (11:38 +0200)]
ALSA: hda: Fix kctl->id initialization
HD-audio core code replaces the kctl->id.index of SPDIF-related
controls after assigning via snd_ctl_add(). This doesn't work any
longer with the new Xarray lookup change. The change of the kctl->id
content has to be done via snd_ctl_rename_id() helper, instead.
Takashi Iwai [Tue, 6 Jun 2023 09:38:54 +0000 (11:38 +0200)]
ALSA: gus: Fix kctl->id initialization
GUS driver replaces the kctl->id.index after assigning the kctl via
snd_ctl_add(). This doesn't work any longer with the new Xarray
lookup change. It has to be set before snd_ctl_add() call instead.
Takashi Iwai [Tue, 6 Jun 2023 09:38:53 +0000 (11:38 +0200)]
ALSA: cmipci: Fix kctl->id initialization
cmipci driver replaces the kctl->id.device after assigning the kctl
via snd_ctl_add(). This doesn't work any longer with the new Xarray
lookup change. It has to be set before snd_ctl_add() call instead.
Takashi Iwai [Tue, 6 Jun 2023 09:38:52 +0000 (11:38 +0200)]
ALSA: ymfpci: Fix kctl->id initialization
ymfpci driver replaces the kctl->id.device after assigning the kctl
via snd_ctl_add(). This doesn't work any longer with the new Xarray
lookup change. It has to be set before snd_ctl_add() call instead.
Randy Dunlap [Fri, 26 May 2023 04:45:19 +0000 (21:45 -0700)]
accel/ivpu: ivpu_ipc needs GENERIC_ALLOCATOR
Drivers that use the gen_pool*() family of functions should
select GENERIC_ALLOCATOR to prevent build errors like these:
ld: drivers/accel/ivpu/ivpu_ipc.o: in function `gen_pool_free':
include/linux/genalloc.h:172: undefined reference to `gen_pool_free_owner'
ld: drivers/accel/ivpu/ivpu_ipc.o: in function `gen_pool_alloc_algo':
include/linux/genalloc.h:138: undefined reference to `gen_pool_alloc_algo_owner'
ld: drivers/accel/ivpu/ivpu_ipc.o: in function `gen_pool_free':
include/linux/genalloc.h:172: undefined reference to `gen_pool_free_owner'
ld: drivers/accel/ivpu/ivpu_ipc.o: in function `ivpu_ipc_init':
drivers/accel/ivpu/ivpu_ipc.c:441: undefined reference to `devm_gen_pool_create'
ld: drivers/accel/ivpu/ivpu_ipc.o: in function `gen_pool_add_virt':
include/linux/genalloc.h:104: undefined reference to `gen_pool_add_owner'
Manish Chopra [Mon, 5 Jun 2023 11:26:00 +0000 (16:56 +0530)]
qed/qede: Fix scheduling while atomic
Statistics read through bond interface via sysfs causes
below bug and traces as it triggers the bonding module to
collect the slave device statistics while holding the spinlock,
beneath that qede->qed driver statistics flow gets scheduled out
due to usleep_range() used in PTT acquire logic
Fix this by collecting the statistics asynchronously from a periodic
delayed work scheduled at default stats coalescing interval and return
the recent copy of statisitcs from .ndo_get_stats64(), also add ability
to configure/retrieve stats coalescing interval using below commands -
Linus Torvalds [Tue, 6 Jun 2023 11:36:02 +0000 (04:36 -0700)]
Merge tag 'for-linus-2023060501' of git://git.kernel.org/pub/scm/linux/kernel/git/hid/hid
Pull HID fix from Jiri Kosina:
- Final, confirmed fix for regression causing some devices connected
via Logitech HID++ Unifying receiver take too long to initialize
(Benjamin Tissoires)
* tag 'for-linus-2023060501' of git://git.kernel.org/pub/scm/linux/kernel/git/hid/hid:
HID: hidpp: terminate retry loop on success
wifi: iwlwifi: mvm: Fix -Warray-bounds bug in iwl_mvm_wait_d3_notif()
kmemdup() at line 2735 is not duplicating enough memory for
notif->tid_tear_down and notif->station_id. As it only duplicates
612 bytes: up to offsetofend(struct iwl_wowlan_info_notif,
received_beacons), this is the range of [0, 612) bytes.
Fix this by allocating space for the whole notif object and zero out the
remaining space in memory after member station_id.
This also fixes the following -Warray-bounds issues:
CC drivers/net/wireless/intel/iwlwifi/mvm/d3.o
drivers/net/wireless/intel/iwlwifi/mvm/d3.c: In function ‘iwl_mvm_wait_d3_notif’:
drivers/net/wireless/intel/iwlwifi/mvm/d3.c:2743:30: warning: array subscript ‘struct iwl_wowlan_info_notif[0]’ is partly outside array bounds of ‘unsigned char[612]’ [-Warray-bounds=]
2743 | notif->tid_tear_down = notif_v1->tid_tear_down;
|
from drivers/net/wireless/intel/iwlwifi/mvm/d3.c:7:
In function ‘kmemdup’,
inlined from ‘iwl_mvm_wait_d3_notif’ at drivers/net/wireless/intel/iwlwifi/mvm/d3.c:2735:12:
include/linux/fortify-string.h:765:16: note: object of size 612 allocated by ‘__real_kmemdup’
765 | return __real_kmemdup(p, size, gfp);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/net/wireless/intel/iwlwifi/mvm/d3.c: In function ‘iwl_mvm_wait_d3_notif’:
drivers/net/wireless/intel/iwlwifi/mvm/d3.c:2744:30: warning: array subscript ‘struct iwl_wowlan_info_notif[0]’ is partly outside array bounds of ‘unsigned char[612]’ [-Warray-bounds=]
2744 | notif->station_id = notif_v1->station_id;
| ^~
In function ‘kmemdup’,
inlined from ‘iwl_mvm_wait_d3_notif’ at drivers/net/wireless/intel/iwlwifi/mvm/d3.c:2735:12:
include/linux/fortify-string.h:765:16: note: object of size 612 allocated by ‘__real_kmemdup’
765 | return __real_kmemdup(p, size, gfp);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~
Currently, whenever an EMA beacon is formed, due to is_template
argument being false from the caller, the switch count is always
decremented once which is wrong.
Also if switch count is equal to profile periodicity, this makes
the switch count to reach till zero which triggers a WARN_ON_ONCE.
Don't do link address translation for beacons and probe responses,
this leads to reporting multiple scan list entries for the same AP
(one with the MLD address) which just breaks things.
We might need to extend this in the future for some other (action)
frames that aren't MLD addressed.
Johannes Berg [Sun, 4 Jun 2023 09:11:16 +0000 (12:11 +0300)]
wifi: mac80211: mlme: fix non-inheritence element
There were two bugs when creating the non-inheritence
element:
1) 'at_extension' needs to be declared outside the loop,
otherwise the value resets every iteration and we
can never really switch properly
2) 'added' never got set to true, so we always cut off
the extension element again at the end of the function
This shows another issue that we might add a list but no
extension list, but we need to make the extension list a
zero-length one in that case.
Fix all these issues. While at it, add a comment explaining
the trim.
Ilya Dryomov [Mon, 5 Jun 2023 14:33:35 +0000 (16:33 +0200)]
rbd: get snapshot context after exclusive lock is ensured to be held
Move capturing the snapshot context into the image request state
machine, after exclusive lock is ensured to be held for the duration of
dealing with the image request. This is needed to ensure correctness
of fast-diff states (OBJECT_EXISTS vs OBJECT_EXISTS_CLEAN) and object
deltas computed based off of them. Otherwise the object map that is
forked for the snapshot isn't guaranteed to accurately reflect the
contents of the snapshot when the snapshot is taken under I/O. This
breaks differential backup and snapshot-based mirroring use cases with
fast-diff enabled: since some object deltas may be incomplete, the
destination image may get corrupted.
Ilya Dryomov [Mon, 5 Jun 2023 14:33:35 +0000 (16:33 +0200)]
rbd: move RBD_OBJ_FLAG_COPYUP_ENABLED flag setting
Move RBD_OBJ_FLAG_COPYUP_ENABLED flag setting into the object request
state machine to allow for the snapshot context to be captured in the
image request state machine rather than in rbd_queue_workfn().
Sungwoo Kim [Sat, 3 Jun 2023 12:28:09 +0000 (08:28 -0400)]
Bluetooth: L2CAP: Add missing checks for invalid DCID
When receiving a connect response we should make sure that the DCID is
within the valid range and that we don't already have another channel
allocated for the same DCID.
Missing checks may violate the specification (BLUETOOTH CORE SPECIFICATION
Version 5.4 | Vol 3, Part A, Page 1046).
Fixes: 40624183c202 ("Bluetooth: L2CAP: Add missing checks for invalid LE DCID") Signed-off-by: Sungwoo Kim <[email protected]> Signed-off-by: Luiz Augusto von Dentz <[email protected]>
Pauli Virtanen [Thu, 1 Jun 2023 06:34:45 +0000 (09:34 +0300)]
Bluetooth: ISO: use correct CIS order in Set CIG Parameters event
The order of CIS handle array in Set CIG Parameters response shall match
the order of the CIS_ID array in the command (Core v5.3 Vol 4 Part E Sec
7.8.97). We send CIS_IDs mainly in the order of increasing CIS_ID (but
with "last" CIS first if it has fixed CIG_ID). In handling of the
reply, we currently assume this is also the same as the order of
hci_conn in hdev->conn_hash, but that is not true.
Match the correct hci_conn to the correct handle by matching them based
on the CIG+CIS combination. The CIG+CIS combination shall be unique for
ISO_LINK hci_conn at state >= BT_BOUND, which we maintain in
hci_le_set_cig_params.
Fixes: 26afbd826ee3 ("Bluetooth: Add initial implementation of CIS connections") Signed-off-by: Pauli Virtanen <[email protected]> Signed-off-by: Luiz Augusto von Dentz <[email protected]>
Pauli Virtanen [Thu, 1 Jun 2023 06:34:44 +0000 (09:34 +0300)]
Bluetooth: ISO: don't try to remove CIG if there are bound CIS left
Consider existing BOUND & CONNECT state CIS to block CIG removal.
Otherwise, under suitable timing conditions we may attempt to remove CIG
while Create CIS is pending, which fails.
Fixes: 26afbd826ee3 ("Bluetooth: Add initial implementation of CIS connections") Signed-off-by: Pauli Virtanen <[email protected]> Signed-off-by: Luiz Augusto von Dentz <[email protected]>
Ying Hsu [Wed, 31 May 2023 03:44:56 +0000 (03:44 +0000)]
Bluetooth: Fix l2cap_disconnect_req deadlock
L2CAP assumes that the locks conn->chan_lock and chan->lock are
acquired in the order conn->chan_lock, chan->lock to avoid
potential deadlock.
For example, l2sock_shutdown acquires these locks in the order:
mutex_lock(&conn->chan_lock)
l2cap_chan_lock(chan)
However, l2cap_disconnect_req acquires chan->lock in
l2cap_get_chan_by_scid first and then acquires conn->chan_lock
before calling l2cap_chan_del. This means that these locks are
acquired in unexpected order, which leads to potential deadlock:
l2cap_chan_lock(c)
mutex_lock(&conn->chan_lock)
This patch releases chan->lock before acquiring the conn_chan_lock
to avoid the potential deadlock.
Fixes: a2a9339e1c9d ("Bluetooth: L2CAP: Fix use-after-free in l2cap_disconnect_{req,rsp}") Signed-off-by: Ying Hsu <[email protected]> Signed-off-by: Luiz Augusto von Dentz <[email protected]>
Johan Hovold [Wed, 31 May 2023 08:57:59 +0000 (10:57 +0200)]
Bluetooth: hci_qca: fix debugfs registration
Since commit 3e4be65eb82c ("Bluetooth: hci_qca: Add poweroff support
during hci down for wcn3990"), the setup callback which registers the
debugfs interface can be called multiple times.
This specifically leads to the following error when powering on the
controller:
debugfs: Directory 'ibs' with parent 'hci0' already present!
Add a driver flag to avoid trying to register the debugfs interface more
than once.
Fixes: 3e4be65eb82c ("Bluetooth: hci_qca: Add poweroff support during hci down for wcn3990") Cc: [email protected] # 4.20 Signed-off-by: Johan Hovold <[email protected]> Signed-off-by: Luiz Augusto von Dentz <[email protected]>
Johan Hovold [Wed, 31 May 2023 08:57:58 +0000 (10:57 +0200)]
Bluetooth: fix debugfs registration
Since commit ec6cef9cd98d ("Bluetooth: Fix SMP channel registration for
unconfigured controllers") the debugfs interface for unconfigured
controllers will be created when the controller is configured.
There is however currently nothing preventing a controller from being
configured multiple time (e.g. setting the device address using btmgmt)
which results in failed attempts to register the already registered
debugfs entries:
debugfs: File 'features' in directory 'hci0' already present!
debugfs: File 'manufacturer' in directory 'hci0' already present!
debugfs: File 'hci_version' in directory 'hci0' already present!
...
debugfs: File 'quirk_simultaneous_discovery' in directory 'hci0' already present!
Add a controller flag to avoid trying to register the debugfs interface
more than once.
Fixes: ec6cef9cd98d ("Bluetooth: Fix SMP channel registration for unconfigured controllers") Cc: [email protected] # 4.0 Signed-off-by: Johan Hovold <[email protected]> Signed-off-by: Luiz Augusto von Dentz <[email protected]>
Zhengping Jiang [Thu, 25 May 2023 00:11:58 +0000 (17:11 -0700)]
Bluetooth: hci_sync: add lock to protect HCI_UNREGISTER
When the HCI_UNREGISTER flag is set, no jobs should be scheduled. Fix
potential race when HCI_UNREGISTER is set after the flag is tested in
hci_cmd_sync_queue.
Fixes: 0b94f2651f56 ("Bluetooth: hci_sync: Fix queuing commands when HCI_UNREGISTER is set") Signed-off-by: Zhengping Jiang <[email protected]> Signed-off-by: Luiz Augusto von Dentz <[email protected]>
Pauli Virtanen [Sun, 21 May 2023 15:48:29 +0000 (15:48 +0000)]
Bluetooth: ISO: Fix CIG auto-allocation to select configurable CIG
Make CIG auto-allocation to select the first CIG_ID that is still
configurable. Also use correct CIG_ID range (see Core v5.3 Vol 4 Part E
Sec 7.8.97 p.2553).
Previously, it would always select CIG_ID 0 regardless of anything,
because cis_list with data.cis == 0xff (BT_ISO_QOS_CIS_UNSET) would not
count any CIS. Since we are not adding CIS here, use find_cis instead.
Fixes: 26afbd826ee3 ("Bluetooth: Add initial implementation of CIS connections") Signed-off-by: Pauli Virtanen <[email protected]> Signed-off-by: Luiz Augusto von Dentz <[email protected]>
Pauli Virtanen [Sun, 21 May 2023 15:48:28 +0000 (15:48 +0000)]
Bluetooth: ISO: consider right CIS when removing CIG at cleanup
When looking for CIS blocking CIG removal, consider only the CIS with
the right CIG ID. Don't try to remove CIG with unset CIG ID.
Fixes: 26afbd826ee3 ("Bluetooth: Add initial implementation of CIS connections") Signed-off-by: Pauli Virtanen <[email protected]> Signed-off-by: Luiz Augusto von Dentz <[email protected]>
Dan Schatzberg [Thu, 1 Jun 2023 18:38:19 +0000 (11:38 -0700)]
cgroup: Documentation: Clarify usage of memory limits
The existing documentation refers to memory.high as the "main mechanism
to control memory usage." This seems incorrect to me - memory.high can
result in reclaim pressure which simply leads to stalls unless some
external component observes and actions on it (e.g. systemd-oomd can be
used for this purpose). While this is feasible, users are unaware of
this interaction and are led to believe that memory.high alone is an
effective mechanism for limiting memory.
The documentation should recommend the use of memory.max as the
effective way to enforce memory limits - it triggers reclaim and results
in OOM kills by itself.
And we currently don't have an attach type for NETLINK BPF link.
Thankfully it's not too late to add it. I see that link_create() in
kernel/bpf/syscall.c just bypasses attach_type check. We shouldn't
have done that. Instead we need to add BPF_NETLINK attach type to enum
bpf_attach_type. And wire all that properly throughout the kernel and
libbpf itself.
This adds BPF_NETFILTER and uses it. This breaks uabi but this
wasn't in any non-rc release yet, so it should be fine.
v2: check link_attack prog type in link_create too
[CAUSE]
In btrfs/027 we test RAID6 with missing devices, in this particular
case, we're repairing a metadata at the end of a data stripe.
But at btrfs_repair_io_failure(), we always pass a full PAGE for repair,
and for subpage case this can cross stripe boundary and lead to the
above BUG_ON().
This metadata repair code is always there, since the introduction of
subpage support, but this can trigger BUG_ON() after the bio split
ability at btrfs_map_bio().
[FIX]
Instead of passing the old PAGE_SIZE, we calculate the correct length
based on the eb size and page size for both regular and subpage cases.
Several device tree files get the polarity of the pendown-gpios
wrong: this signal is active low. Fix up all incorrect flags, so
that operating systems can rely on the flag being correctly set.
Arnd Bergmann [Mon, 5 Jun 2023 15:38:51 +0000 (17:38 +0200)]
Merge tag 'at91-fixes-6.4' of https://git.kernel.org/pub/scm/linux/kernel/git/at91/linux into arm/fixes
AT91 fixes for 6.4
It contains:
- fix imbalanced reference counter for ethernet devices; without it
system hangs after consecutive suspend/resume cycles;
- fix debounce delay property for shutdown controller; the initial DT
property is not what the driver expects.
* tag 'at91-fixes-6.4' of https://git.kernel.org/pub/scm/linux/kernel/git/at91/linux:
ARM: dts: at91: sama7g5ek: fix debounce delay property for shdwc
ARM: at91: pm: fix imbalanced reference counter for ethernet devices
Arnd Bergmann [Mon, 5 Jun 2023 15:37:28 +0000 (17:37 +0200)]
Merge tag 'qcom-arm64-fixes-for-6.4' of https://git.kernel.org/pub/scm/linux/kernel/git/qcom/linux into arm/fixes
Qualcomm ARM64 DeviceTree fixes for 6.4
Register scheme for SM8550 LLCC is corrected to avoid using the wrong
register offsets. SDRAM frequency for misidentified SC7180-lite boards
is handled. The datatype for Soundwire interval on SM8550 is corrected.
The resource controller on SC8280XP is added to the CPU cluster
power-domain to get notified to send cached sleep and wake votes before
going entering the lower power states.
SA8155P power-domains that differ from what's inherited from the SM8150
DeviceTree are adjusted to make the platform boot again.
Remoteproc firmware paths are corrected for Sony Xperia 10 IV.
Cache properties are adjusted across a range of platforms, to meet
changes in the binding.
Panel compatibles are corrected for Xiaomi Mi Pad 5 Pro, to match
binding. Invalid dai-cells are dropped from SC7280 devices, to match
binding.
The incorrect removal of "input-enable" from the LPASS pinctrl node of
SC8280XP was reverted, to get dmic pins in the correct state again.
The incorrect input-enable property is dropped from a msm8974, mdm9615
and apq8026 to resolve a range of DT validation warnings, incorrectly
picked up through the ARM64 tree.
* tag 'qcom-arm64-fixes-for-6.4' of https://git.kernel.org/pub/scm/linux/kernel/git/qcom/linux:
arm64: dts: qcom: sm8550: Use the correct LLCC register scheme
arm64: dts: qcom: sc7180-lite: Fix SDRAM freq for misidentified sc7180-lite boards
arm64: dts: qcom: sm8550: use uint16 for Soundwire interval
arm64: dts: qcom: Split out SA8155P and use correct RPMh power domains
arm64: dts: qcom: sm6375-pdx225: Fix remoteproc firmware paths
arm64: dts: qcom: add missing cache properties
arm64: dts: qcom: use decimal for cache level
arm64: dts: qcom: fix indentation
ARM: dts: qcom: msm8974: remove superfluous "input-enable"
ARM: dts: qcom: mdm9615: remove superfluous "input-enable"
ARM: dts: qcom: apq8026: remove superfluous "input-enable"
arm64: dts: qcom: sm8250-xiaomi-elish-csot: fix panel compatible
arm64: dts: qcom: sm8250-xiaomi-elish-boe: fix panel compatible
arm64: dts: qcom: sc7280-qcard: drop incorrect dai-cells from WCD938x SDW
arm64: dts: qcom: sc7280-idp: drop incorrect dai-cells from WCD938x SDW
arm64: dts: qcom: sc8280xp: Flush RSC sleep & wake votes
arm64: dts: qcom: sc8280xp: Revert "arm64: dts: qcom: sc8280xp: remove superfluous "input-enable""
Arnd Bergmann [Mon, 5 Jun 2023 15:35:28 +0000 (17:35 +0200)]
Merge tag 'qcom-driver-fixes-for-6.4' of https://git.kernel.org/pub/scm/linux/kernel/git/qcom/linux into arm/fixes
Qualcomm driver fixes for 6.4
Error paths is corrected across icc-bwmon, rpmh-rsc, ramp_controller and
rmtfs. The ice module is renamed qcom_ice, to avoid clashing with
existing "ice" driver.
SA8155P-specific RPMh power-domains are introduced to avoid the code
trying to access resources that exists on SM8150, but not on SA8155P.
Lastly, changes to the EDAC driver to fix an issue where the driver
performs mmio based on the wrong register map.
* tag 'qcom-driver-fixes-for-6.4' of https://git.kernel.org/pub/scm/linux/kernel/git/qcom/linux:
EDAC/qcom: Get rid of hardcoded register offsets
EDAC/qcom: Remove superfluous return variable assignment in qcom_llcc_core_setup()
dt-bindings: cache: qcom,llcc: Fix SM8550 description
soc: qcom: rpmhpd: Add SA8155P power domains
dt-bindings: power: qcom,rpmpd: Add SA8155P
soc: qcom: Rename ice to qcom_ice to avoid module name conflict
soc: qcom: rmtfs: Fix error code in probe()
soc: qcom: ramp_controller: Fix an error handling path in qcom_ramp_controller_probe()
soc: qcom: rpmh-rsc: drop redundant unsigned >=0 comparision
soc: qcom: icc-bwmon: fix incorrect error code passed to dev_err_probe()
Linus Torvalds [Mon, 5 Jun 2023 15:31:36 +0000 (11:31 -0400)]
Merge tag 'asym-keys-fix-for-linus-v6.4-rc5' of https://github.com/robertosassu/linux
Pull asymmetric keys fix from Roberto Sassu:
"Here is a small fix to make an unconditional copy of the buffer passed
to crypto operations, to take into account the case of the stack not
in the linear mapping area.
It has been tested and verified to fix the bug"
Acked-by: Herbert Xu <[email protected]> Acked-by: David Howells <[email protected]>
* tag 'asym-keys-fix-for-linus-v6.4-rc5' of https://github.com/robertosassu/linux:
KEYS: asymmetric: Copy sig and digest in public_key_verify_signature()
David S. Miller [Mon, 5 Jun 2023 14:15:57 +0000 (15:15 +0100)]
Merge branch 'mptcp-addr-adv-fixes'
Mat Martineau says:
====================
mptcp: Fixes for address advertisement
Patches 1 and 2 allow address advertisements to be removed without
affecting current connected subflows, and updates associated self tests.
Patches 3 and 4 correctly track (and allow removal of) addresses that
were implicitly announced as part of subflow creation. Also updates
associated self tests.
Patch 5 makes subflow and address announcement counters work consistently
between the userspace and in-kernel path managers.
====================
Geliang Tang [Mon, 5 Jun 2023 03:25:21 +0000 (20:25 -0700)]
mptcp: update userspace pm infos
Increase pm subflows counter on both server side and client side when
userspace pm creates a new subflow, and decrease the counter when it
closes a subflow.
Increase add_addr_signaled counter in mptcp_nl_cmd_announce() when the
address is announced by userspace PM.
This modification is similar to how the in-kernel PM is updating the
counter: when additional subflows are created/removed.
Fixes: 9ab4807c84a4 ("mptcp: netlink: Add MPTCP_PM_CMD_ANNOUNCE") Fixes: 702c2f646d42 ("mptcp: netlink: allow userspace-driven subflow establishment") Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/329 Cc: [email protected] Reviewed-by: Matthieu Baerts <[email protected]> Signed-off-by: Geliang Tang <[email protected]> Signed-off-by: Mat Martineau <[email protected]> Signed-off-by: David S. Miller <[email protected]>
To align with what is done by the in-kernel PM, update userspace pm
subflow selftests, by sending the a remove_addrs command together
before the remove_subflows command. This will get a RM_ADDR in
chk_rm_nr().
Geliang Tang [Mon, 5 Jun 2023 03:25:19 +0000 (20:25 -0700)]
mptcp: add address into userspace pm list
Add the address into userspace_pm_local_addr_list when the subflow is
created. Make sure it can be found in mptcp_nl_cmd_remove(). And delete
it in the new helper mptcp_userspace_pm_delete_local_addr().
By doing this, the "REMOVE" command also works with subflows that have
been created via the "SUB_CREATE" command instead of restricting to
the addresses that have been announced via the "ANNOUNCE" command.
Geliang Tang [Mon, 5 Jun 2023 03:25:18 +0000 (20:25 -0700)]
selftests: mptcp: update userspace pm addr tests
This patch is linked to the previous commit ("mptcp: only send RM_ADDR in
nl_cmd_remove").
To align with what is done by the in-kernel PM, update userspace pm addr
selftests, by sending a remove_subflows command together after the
remove_addrs command.
Fixes: d9a4594edabf ("mptcp: netlink: Add MPTCP_PM_CMD_REMOVE") Fixes: 97040cf9806e ("selftests: mptcp: userspace pm address tests") Cc: [email protected] Reviewed-by: Matthieu Baerts <[email protected]> Signed-off-by: Geliang Tang <[email protected]> Signed-off-by: Mat Martineau <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Jouni Högander [Tue, 30 May 2023 10:16:49 +0000 (13:16 +0300)]
drm/i915: Use 18 fast wake AUX sync len
HW default for wake sync pulses is 18. 10 precharge and 8 preamble. There
is no reason to change this especially as it is causing problems with
certain eDP panels.
net: stmmac: dwmac-qcom-ethqos: fix a regression on EMAC < 3
We must not assign plat_dat->dwmac4_addrs unconditionally as for
structures which don't set them, this will result in the core driver
using zeroes everywhere and breaking the driver for older HW. On EMAC < 2
the address should remain NULL.
Marek Behún [Sun, 21 May 2023 12:19:40 +0000 (14:19 +0200)]
i2c: mv64xxx: Fix reading invalid status value in atomic mode
There seems to be a bug within the mv64xxx I2C controller, wherein the
status register may not necessarily contain valid value immediately
after the IFLG flag is set in the control register.
My theory is that the controller:
- first sets the IFLG in control register
- then updates the status register
- then raises an interrupt
This may sometime cause weird bugs when in atomic mode, since in this
mode we do not wait for an interrupt, but instead we poll the control
register for IFLG and read status register immediately after.
I encountered -ENXIO from mv64xxx_i2c_fsm() due to this issue when using
this driver in atomic mode.
Note that I've only seen this issue on Armada 385, I don't know whether
other SOCs with this controller are also affected. Also note that this
fix has been in U-Boot for over 4 years [1] without anybody complaining,
so it should not cause regressions.
David S. Miller [Mon, 5 Jun 2023 10:21:10 +0000 (11:21 +0100)]
Merge tag 'linux-can-fixes-for-6.4-20230605' of git://git.kernel.org/pub/scm/linux/kernel/git/mkl/linux-can
Marc Kleine-Budde says:
====================
this is a pull request of 3 patches for net/master.
All 3 patches target the j1939 stack.
The 1st patch is by Oleksij Rempel and fixes the error queue handling
for (E)TP sessions that run into timeouts.
The last 2 patches are by Fedor Pchelkin and fix a potential
use-after-free in j1939_netdev_start() if j1939_can_rx_register()
fails.
====================
David Zheng [Wed, 24 May 2023 18:14:59 +0000 (11:14 -0700)]
i2c: designware: fix idx_write_cnt in read loop
With IC_INTR_RX_FULL slave interrupt handler reads data in a loop until
RX FIFO is empty. When testing with the slave-eeprom, each transaction
has 2 bytes for address/index and 1 byte for value, the address byte
can be written as data byte due to dropping STOP condition.
In the test below, the master continuously writes to the slave, first 2
bytes are index, 3rd byte is value and follow by a STOP condition.
Upon receiving STOP condition slave eeprom would reset `idx_write_cnt` so
next 2 bytes can be treated as buffer index for upcoming transaction.
Supposedly the slave eeprom buffer would be written as
When CPU load is high the slave irq handler may not read fast enough,
the interrupt status can be seen as 0x204 with both DW_IC_INTR_STOP_DET
(0x200) and DW_IC_INTR_RX_FULL (0x4) bits. The slave device may see
the transactions below.
0x1 STATUS SLAVE_ACTIVITY=0x1 : RAW_INTR_STAT=0x1594 : INTR_STAT=0x4
0x1 STATUS SLAVE_ACTIVITY=0x1 : RAW_INTR_STAT=0x1594 : INTR_STAT=0x4
0x1 STATUS SLAVE_ACTIVITY=0x1 : RAW_INTR_STAT=0x1594 : INTR_STAT=0x4
0x1 STATUS SLAVE_ACTIVITY=0x1 : RAW_INTR_STAT=0x1794 : INTR_STAT=0x204
0x1 STATUS SLAVE_ACTIVITY=0x0 : RAW_INTR_STAT=0x1790 : INTR_STAT=0x200
0x1 STATUS SLAVE_ACTIVITY=0x1 : RAW_INTR_STAT=0x1594 : INTR_STAT=0x4
0x1 STATUS SLAVE_ACTIVITY=0x1 : RAW_INTR_STAT=0x1594 : INTR_STAT=0x4
0x1 STATUS SLAVE_ACTIVITY=0x1 : RAW_INTR_STAT=0x1594 : INTR_STAT=0x4
After `D1` is received, read loop continues to read `00` which is the
first bype of next index. Since STOP condition is ignored by the loop,
eeprom buffer index increased to `D2` and `00` is written as value.
The fix is to use `FIRST_DATA_BYTE` (bit 11) in `IC_DATA_CMD` to split
the transactions. The first index byte in this case would have bit 11
set. Check this indication to inject I2C_SLAVE_WRITE_REQUESTED event
which will reset `idx_write_cnt` in slave eeprom.
We got multiple syzbot reports, all duplicates of the following [1]
syzbot managed to install fq_pie with a zero TCA_FQ_PIE_QUANTUM,
thus triggering infinite loops.
Use limits similar to sch_fq, with commits 3725a269815b ("pkt_sched: fq: avoid hang when quantum 0") and d9e15a273306 ("pkt_sched: fq: do not accept silly TCA_FQ_QUANTUM")
selftests: alsa: pcm-test: Fix compiler warnings about the format
GCC 11.3.0 issues warnings in this module about wrong sizes of format
specifiers:
pcm-test.c: In function ‘test_pcm_time’:
pcm-test.c:384:68: warning: format ‘%ld’ expects argument of type ‘long int’, but argument 5 \
has type ‘unsigned int’ [-Wformat=]
384 | snprintf(msg, sizeof(msg), "rate mismatch %ld != %ld", rate, rrate);
pcm-test.c:455:53: warning: format ‘%d’ expects argument of type ‘int’, but argument 4 has \
type ‘long int’ [-Wformat=]
455 | "expected %d, wrote %li", rate, frames);
pcm-test.c:462:53: warning: format ‘%d’ expects argument of type ‘int’, but argument 4 has \
type ‘long int’ [-Wformat=]
462 | "expected %d, wrote %li", rate, frames);
pcm-test.c:467:53: warning: format ‘%d’ expects argument of type ‘int’, but argument 4 has \
type ‘long int’ [-Wformat=]
467 | "expected %d, wrote %li", rate, frames);
Simple fix according to compiler's suggestion removed the warnings.
The patch series fixes a possible racy use-after-free scenario
described in 2/2: if j1939_can_rx_register() fails then the concurrent
thread may have already read the invalid priv structure.
The 1/2 makes j1939_netdev_lock a mutex so that access to
j1939_can_rx_register() can be serialized without changing GFP_KERNEL
to GFP_ATOMIC inside can_rx_register(). This seems to be safe.
Note that the patch series has been tested only via Syzkaller and not
with a real device.
Fedor Pchelkin [Fri, 26 May 2023 17:19:10 +0000 (20:19 +0300)]
can: j1939: avoid possible use-after-free when j1939_can_rx_register fails
Syzkaller reports the following failure:
BUG: KASAN: use-after-free in kref_put include/linux/kref.h:64 [inline]
BUG: KASAN: use-after-free in j1939_priv_put+0x25/0xa0 net/can/j1939/main.c:172
Write of size 4 at addr ffff888141c15058 by task swapper/3/0
To avoid this, call j1939_can_rx_register() under j1939_netdev_lock so
that a concurrent thread cannot process j1939_priv before
j1939_can_rx_register() returns.
Found by Linux Verification Center (linuxtesting.org) with Syzkaller.
Fedor Pchelkin [Fri, 26 May 2023 17:19:09 +0000 (20:19 +0300)]
can: j1939: change j1939_netdev_lock type to mutex
It turns out access to j1939_can_rx_register() needs to be serialized,
otherwise j1939_priv can be corrupted when parallel threads call
j1939_netdev_start() and j1939_can_rx_register() fails. This issue is
thoroughly covered in other commit which serializes access to
j1939_can_rx_register().
Change j1939_netdev_lock type to mutex so that we do not need to remove
GFP_KERNEL from can_rx_register().
j1939_netdev_lock seems to be used in normal contexts where mutex usage
is not prohibited.
Found by Linux Verification Center (linuxtesting.org) with Syzkaller.
Oleksij Rempel [Fri, 26 May 2023 08:19:46 +0000 (10:19 +0200)]
can: j1939: j1939_sk_send_loop_abort(): improved error queue handling in J1939 Socket
This patch addresses an issue within the j1939_sk_send_loop_abort()
function in the j1939/socket.c file, specifically in the context of
Transport Protocol (TP) sessions.
Without this patch, when a TP session is initiated and a Clear To Send
(CTS) frame is received from the remote side requesting one data packet,
the kernel dispatches the first Data Transport (DT) frame and then waits
for the next CTS. If the remote side doesn't respond with another CTS,
the kernel aborts due to a timeout. This leads to the user-space
receiving an EPOLLERR on the socket, and the socket becomes active.
However, when trying to read the error queue from the socket with
sock.recvmsg(, , socket.MSG_ERRQUEUE), it returns -EAGAIN,
given that the socket is non-blocking. This situation results in an
infinite loop: the user-space repeatedly calls epoll(), epoll() returns
the socket file descriptor with EPOLLERR, but the socket then blocks on
the recv() of ERRQUEUE.
This patch introduces an additional check for the J1939_SOCK_ERRQUEUE
flag within the j1939_sk_send_loop_abort() function. If the flag is set,
it indicates that the application has subscribed to receive error queue
messages. In such cases, the kernel can communicate the current transfer
state via the error queue. This allows for the function to return early,
preventing the unnecessary setting of the socket into an error state,
and breaking the infinite loop. It is crucial to note that a socket
error is only needed if the application isn't using the error queue, as,
without it, the application wouldn't be aware of transfer issues.
Dave Chinner [Mon, 5 Jun 2023 04:48:15 +0000 (14:48 +1000)]
xfs: collect errors from inodegc for unlinked inode recovery
Unlinked list recovery requires errors removing the inode the from
the unlinked list get fed back to the main recovery loop. Now that
we offload the unlinking to the inodegc work, we don't get errors
being fed back when we trip over a corruption that prevents the
inode from being removed from the unlinked list.
This means we never clear the corrupt unlinked list bucket,
resulting in runtime operations eventually tripping over it and
shutting down.
Fix this by collecting inodegc worker errors and feed them
back to the flush caller. This is largely best effort - the only
context that really cares is log recovery, and it only flushes a
single inode at a time so we don't need complex synchronised
handling. Essentially the inodegc workers will capture the first
error that occurs and the next flush will gather them and clear
them. The flush itself will only report the first gathered error.
In the cases where callers can return errors, propagate the
collected inodegc flush error up the error handling chain.
In the case of inode unlinked list recovery, there are several
superfluous calls to flush queued unlinked inodes -
xlog_recover_iunlink_bucket() guarantees that it has flushed the
inodegc and collected errors before it returns. Hence nothing in the
calling path needs to run a flush, even when an error is returned.
Dave Chinner [Mon, 5 Jun 2023 04:48:15 +0000 (14:48 +1000)]
xfs: validate block number being freed before adding to xefi
Bad things happen in defered extent freeing operations if it is
passed a bad block number in the xefi. This can come from a bogus
agno/agbno pair from deferred agfl freeing, or just a bad fsbno
being passed to __xfs_free_extent_later(). Either way, it's very
difficult to diagnose where a null perag oops in EFI creation
is coming from when the operation that queued the xefi has already
been completed and there's no longer any trace of it around....
Dave Chinner [Mon, 5 Jun 2023 04:48:15 +0000 (14:48 +1000)]
xfs: validity check agbnos on the AGFL
If the agfl or the indexing in the AGF has been corrupted, getting a
block form the AGFL could return an invalid block number. If this
happens, bad things happen. Check the agbno we pull off the AGFL
and return -EFSCORRUPTED if we find somethign bad.
Dave Chinner [Mon, 5 Jun 2023 04:48:15 +0000 (14:48 +1000)]
xfs: fix agf/agfl verification on v4 filesystems
When a v4 filesystem has fl_last - fl_first != fl_count, we do not
not detect the corruption and allow the AGF to be used as it if was
fully valid. On V5 filesystems, we reset the AGFL to empty in these
cases and avoid the corruption at a small cost of leaked blocks.
If we don't catch the corruption on V4 filesystems, bad things
happen later when an allocation attempts to trim the free list
and either double-frees stale entries in the AGFl or tries to free
NULLAGBNO entries.
Either way, this is bad. Prevent this from happening by using the
AGFL_NEED_RESET logic for v4 filesysetms, too.
Dave Chinner [Mon, 5 Jun 2023 04:48:15 +0000 (14:48 +1000)]
xfs: fix double xfs_perag_rele() in xfs_filestream_pick_ag()
xfs_bmap_longest_free_extent() can return an error when accessing
the AGF fails. In this case, the behaviour of
xfs_filestream_pick_ag() is conditional on the error. We may
continue the loop, or break out of it. The error handling after the
loop cleans up the perag reference held when the break occurs. If we
continue, the next loop iteration handles cleaning up the perag
reference.
EIther way, we don't need to release the active perag reference when
xfs_bmap_longest_free_extent() fails. Doing so means we do a double
decrement on the active reference count, and this causes tha active
reference count to fall to zero. At this point, new active
references will fail.
This leads to unmount hanging because it tries to grab active
references to that perag, only for it to fail. This happens inside a
loop that retries until a inode tree radix tree tag is cleared,
which cannot happen because we can't get an active reference to the
perag.
Darrick J. Wong [Mon, 5 Jun 2023 04:48:12 +0000 (14:48 +1000)]
xfs: fix broken logic when detecting mergeable bmap records
Commit 6bc6c99a944c was a well-intentioned effort to initiate
consolidation of adjacent bmbt mapping records by setting the PREEN
flag. Consolidation can only happen if the length of the combined
record doesn't overflow the 21-bit blockcount field of the bmbt
recordset. Unfortunately, the length test is inverted, leading to it
triggering on data forks like these:
Note that record 0 has a length of 16777208 512b blocks. This
corresponds to 2097151 4k fsblocks, which is the maximum. Hence the two
records cannot be merged.
However, the logic is still wrong even if we change the in-loop
comparison, because the scope of our examination isn't broad enough
inside the loop to detect mappings like this:
These three records could be merged into two, but one cannot determine
this purely from looking at records 0-1 or 1-2 in isolation.
Hoist the mergability detection outside the loop, and base its decision
making on whether or not a merged mapping could be expressed in fewer
bmbt records. While we're at it, fix the incorrect return type of the
iter function.
Fixes: 336642f79283 ("xfs: alert the user about data/attr fork mappings that could be merged") Signed-off-by: Darrick J. Wong <[email protected]> Reviewed-by: Dave Chinner <[email protected]> Signed-off-by: Dave Chinner <[email protected]>
xfs: Fix undefined behavior of shift into sign bit
With gcc-5:
In file included from ./include/trace/define_trace.h:102:0,
from ./fs/xfs/scrub/trace.h:988,
from fs/xfs/scrub/trace.c:40:
./fs/xfs/./scrub/trace.h: In function ‘trace_raw_output_xchk_fsgate_class’:
./fs/xfs/scrub/scrub.h:111:28: error: initializer element is not constant
#define XREP_ALREADY_FIXED (1 << 31) /* checking our repair work */
^
Shifting the (signed) value 1 into the sign bit is undefined behavior.
Fix this for all definitions in the file by shifting "1U" instead of
"1".
This was exposed by the first user added in commit 466c525d6d35e691
("xfs: minimize overhead of drain wakeups by using jump labels").
Fixes: 160b5a784525e8a4 ("xfs: hoist the already_fixed variable to the scrub context") Signed-off-by: Geert Uytterhoeven <[email protected]> Reviewed-by: Darrick J. Wong <[email protected]> Signed-off-by: Dave Chinner <[email protected]>
Dave Chinner [Sun, 4 Jun 2023 18:08:27 +0000 (04:08 +1000)]
xfs: fix AGF vs inode cluster buffer deadlock
Lock order in XFS is AGI -> AGF, hence for operations involving
inode unlinked list operations we always lock the AGI first. Inode
unlinked list operations operate on the inode cluster buffer,
so the lock order there is AGI -> inode cluster buffer.
For O_TMPFILE operations, this now means the lock order set down in
xfs_rename and xfs_link is AGI -> inode cluster buffer -> AGF as the
unlinked ops are done before the directory modifications that may
allocate space and lock the AGF.
Unfortunately, we also now lock the inode cluster buffer when
logging an inode so that we can attach the inode to the cluster
buffer and pin it in memory. This creates a lock order of AGF ->
inode cluster buffer in directory operations as we have to log the
inode after we've allocated new space for it.
This creates a lock inversion between the AGF and the inode cluster
buffer. Because the inode cluster buffer is shared across multiple
inodes, the inversion is not specific to individual inodes but can
occur when inodes in the same cluster buffer are accessed in
different orders.
To fix this we need move all the inode log item cluster buffer
interactions to the end of the current transaction. Unfortunately,
xfs_trans_log_inode() calls are littered throughout the transactions
with no thought to ordering against other items or locking. This
makes it difficult to do anything that involves changing the call
sites of xfs_trans_log_inode() to change locking orders.
However, we do now have a mechanism that allows is to postpone dirty
item processing to just before we commit the transaction: the
->iop_precommit method. This will be called after all the
modifications are done and high level objects like AGI and AGF
buffers have been locked and modified, thereby providing a mechanism
that guarantees we don't lock the inode cluster buffer before those
high level objects are locked.
This change is largely moving the guts of xfs_trans_log_inode() to
xfs_inode_item_precommit() and providing an extra flag context in
the inode log item to track the dirty state of the inode in the
current transaction. This also means we do a lot less repeated work
in xfs_trans_log_inode() by only doing it once per transaction when
all the work is done.
Dave Chinner [Sun, 4 Jun 2023 18:07:27 +0000 (04:07 +1000)]
xfs: defered work could create precommits
To fix a AGI-AGF-inode cluster buffer deadlock, we need to move
inode cluster buffer operations to the ->iop_precommit() method.
However, this means that deferred operations can require precommits
to be run on the final transaction that the deferred ops pass back
to xfs_trans_commit() context. This will be exposed by attribute
handling, in that the last changes to the inode in the attr set
state machine "disappear" because the precommit operation is not run.
Dave Chinner [Sun, 4 Jun 2023 18:06:27 +0000 (04:06 +1000)]
xfs: restore allocation trylock iteration
It was accidentally dropped when refactoring the allocation code,
resulting in the AG iteration always doing blocking AG iteration.
This results in a small performance regression for a specific fsmark
test that runs more user data writer threads than there are AGs.
Dave Chinner [Sun, 4 Jun 2023 18:05:27 +0000 (04:05 +1000)]
xfs: buffer pins need to hold a buffer reference
When a buffer is unpinned by xfs_buf_item_unpin(), we need to access
the buffer after we've dropped the buffer log item reference count.
This opens a window where we can have two racing unpins for the
buffer item (e.g. shutdown checkpoint context callback processing
racing with journal IO iclog completion processing) and both attempt
to access the buffer after dropping the BLI reference count. If we
are unlucky, the "BLI freed" context wins the race and frees the
buffer before the "BLI still active" case checks the buffer pin
count.
This results in a use after free that can only be triggered
in active filesystem shutdown situations.
To fix this, we need to ensure that buffer existence extends beyond
the BLI reference count checks and until the unpin processing is
complete. This implies that a buffer pin operation must also take a
buffer reference to ensure that the buffer cannot be freed until the
buffer unpin processing is complete.
projects / linux.git / log