After the blamed commit below, the TCP sockets (and the MPTCP subflows)
can build egress packets larger than 64K. That exceeds the maximum DSS
data size, the length being misrepresent on the wire and the stream being
corrupted, as later observed on the receiver:
Objcg vectors attached to slab pages to store slab object ownership
information are allocated using gfp flags for the original slab
allocation. Depending on slab page order and the size of slab objects,
objcg vector can take several pages.
If the original allocation was done with the __GFP_NOFAIL flag, it
triggered a warning in the page allocation code. Indeed, order > 1 pages
should not been allocated with the __GFP_NOFAIL flag.
Fix this by simply dropping the __GFP_NOFAIL flag when allocating the
objcg vector. It effectively allows to skip the accounting of a single
slab object under a heavy memory pressure.
An alternative would be to implement the mechanism to fallback to order-0
allocations for accounting metadata, which is also not perfect because it
will increase performance penalty and memory footprint of the kernel
memory accounting under memory pressure.
Link: https://lkml.kernel.org/r/ZUp8ZFGxwmCx4ZFr@P9FQF9L96D.corp.robot.car Signed-off-by: Roman Gushchin <roman.gushchin@linux.dev> Reported-by: Christoph Lameter <cl@linux.com> Closes: https://lkml.kernel.org/r/6b42243e-f197-600a-5d22-56bd728a5ad8@gentwo.org Acked-by: Shakeel Butt <shakeelb@google.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
While qualifiying the 6.4 release, the following warning was detected in
messages:
vmstat_refresh: nr_file_hugepages -15664
The warning is caused by the incorrect updating of the NR_FILE_THPS
counter in the function split_huge_page_to_list. The if case is checking
for folio_test_swapbacked, but the else case is missing the check for
folio_test_pmd_mappable. The other functions that manipulate the counter
like __filemap_add_folio and filemap_unaccount_folio have the
corresponding check.
I have a test case, which reproduces the problem. It can be found here:
https://github.com/sroeschus/testcase/blob/main/vmstat_refresh/madv.c
The test case reproduces on an XFS filesystem. Running the same test
case on a BTRFS filesystem does not reproduce the problem.
AFAIK version 6.1 until 6.6 are affected by this problem.
[akpm@linux-foundation.org: whitespace fix]
[shr@devkernel.io: test for folio_test_pmd_mappable()] Link: https://lkml.kernel.org/r/20231108171517.2436103-1-shr@devkernel.io Link: https://lkml.kernel.org/r/20231106181918.1091043-1-shr@devkernel.io Signed-off-by: Stefan Roesch <shr@devkernel.io> Co-debugged-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org> Reviewed-by: David Hildenbrand <david@redhat.com> Reviewed-by: Yang Shi <shy828301@gmail.com> Cc: Rik van Riel <riel@surriel.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When GL9750 enters ASPM L1 sub-states, it will stay at L1.1 and will not
enter L1.2. The workaround is to toggle PM state to allow GL9750 to enter
ASPM L1.2.
The pt_level uses CONFIG_PGTABLE_LEVELS to display page table names.
But if page mode is downgraded from kernel cmdline or restricted by
the hardware in 64BIT, it will give a wrong name.
Like, using no4lvl for sv39, ptdump named the 1G-mapping as "PUD"
that should be "PGD":
0xffffffd840000000-0xffffffd900000000 0x00000000c0000000 3G PUD D A G . . W R V
So select "P4D/PUD" or "PGD" via pgtable_l5/4_enabled to correct it.
The interrupt entries are expected to be in the .irqentry.text section.
For example, for kprobes to work properly, exception code cannot be
probed; this is ensured by blacklisting addresses in the .irqentry.text
section.
Actually it is a part of Conor's
commit aae538cd03bc ("riscv: fix detection of toolchain
Zihintpause support").
It is looks like a merge issue. Samuel's
commit 0b1d60d6dd9e ("riscv: Fix build with
CONFIG_CC_OPTIMIZE_FOR_SIZE=y") do not base on Conor's commit and
revert to __riscv_zihintpause. So this patch can fix it.
A recent change to the optimization pipeline in LLVM reveals some
fragility around the inlining of LoongArch's __percpu functions, which
manifests as a BUILD_BUG() failure:
In file included from kernel/sched/build_policy.c:17:
In file included from include/linux/sched/cputime.h:5:
In file included from include/linux/sched/signal.h:5:
In file included from include/linux/rculist.h:11:
In file included from include/linux/rcupdate.h:26:
In file included from include/linux/irqflags.h:18:
arch/loongarch/include/asm/percpu.h:97:3: error: call to '__compiletime_assert_51' declared with 'error' attribute: BUILD_BUG failed
97 | BUILD_BUG();
| ^
include/linux/build_bug.h:59:21: note: expanded from macro 'BUILD_BUG'
59 | #define BUILD_BUG() BUILD_BUG_ON_MSG(1, "BUILD_BUG failed")
| ^
include/linux/build_bug.h:39:37: note: expanded from macro 'BUILD_BUG_ON_MSG'
39 | #define BUILD_BUG_ON_MSG(cond, msg) compiletime_assert(!(cond), msg)
| ^
include/linux/compiler_types.h:425:2: note: expanded from macro 'compiletime_assert'
425 | _compiletime_assert(condition, msg, __compiletime_assert_, __COUNTER__)
| ^
include/linux/compiler_types.h:413:2: note: expanded from macro '_compiletime_assert'
413 | __compiletime_assert(condition, msg, prefix, suffix)
| ^
include/linux/compiler_types.h:406:4: note: expanded from macro '__compiletime_assert'
406 | prefix ## suffix(); \
| ^
<scratch space>:86:1: note: expanded from here
86 | __compiletime_assert_51
| ^
1 error generated.
If these functions are not inlined (which the compiler is free to do
even with functions marked with the standard 'inline' keyword), the
BUILD_BUG() in the default case cannot be eliminated since the compiler
cannot prove it is never used, resulting in a build failure due to the
error attribute.
Mark these functions as __always_inline to guarantee inlining so that
the BUILD_BUG() only triggers when the default case genuinely cannot be
eliminated due to an unexpected size.
When inserting a DRC-cached response into the reply buffer, ensure
that the reply buffer's xdr_stream is updated properly. Otherwise
the server will send a garbage response.
Cc: stable@vger.kernel.org # v6.3+ Reviewed-by: Jeff Layton <jlayton@kernel.org> Tested-by: Jeff Layton <jlayton@kernel.org> Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The commit 5721d4e5a9cd enhanced dm-verity, so that it can verify blocks
from tasklets rather than from workqueues. This reportedly improves
performance significantly.
However, dm-verity was using the flag CRYPTO_TFM_REQ_MAY_SLEEP from
tasklets which resulted in warnings about sleeping function being called
from non-sleeping context.
This commit fixes dm-verity so that it doesn't use the flags
CRYPTO_TFM_REQ_MAY_SLEEP and CRYPTO_TFM_REQ_MAY_BACKLOG from tasklets. The
crypto API would do GFP_ATOMIC allocation instead, it could return -ENOMEM
and we catch -ENOMEM in verity_tasklet and requeue the request to the
workqueue.
dm-bufio has a no-sleep mode. When activated (with the
DM_BUFIO_CLIENT_NO_SLEEP flag), the bufio client is read-only and we
could call dm_bufio_get from tasklets. This is used by dm-verity.
Unfortunately, commit 450e8dee51aa ("dm bufio: improve concurrent IO
performance") broke this and the kernel would warn that cache_get()
was calling down_read() from no-sleeping context. The bug can be
reproduced by using "veritysetup open" with the "--use-tasklets"
flag.
This commit fixes dm-bufio, so that the tasklet mode works again, by
expanding use of the 'no_sleep_enabled' static_key to conditionally
use either a rw_semaphore or rwlock_t (which are colocated in the
buffer_tree structure using a union).
Initialise the try sink compose rectangle size to the sink compose
rectangle for binner and scaler sub-devices. This was missed due to the
faulty condition that lead to the compose rectangles to be initialised for
the pixel array sub-device where it is not relevant.
The hfi parser, parses the capabilities received from venus firmware and
copies them to core capabilities. Consider below api, for example,
fill_caps - In this api, caps in core structure gets updated with the
number of capabilities received in firmware data payload. If the same api
is called multiple times, there is a possibility of copying beyond the max
allocated size in core caps.
Similar possibilities in fill_raw_fmts and fill_profile_level functions.
Buffer requirement, for different buffer type, comes from video firmware.
While copying these requirements, there is an OOB possibility when the
payload from firmware is more than expected size. Fix the check to avoid
the OOB possibility.
Supported codec bitmask is populated from the payload from venus firmware.
There is a possible case when all the bits in the codec bitmask is set. In
such case, core cap for decoder is filled and MAX_CODEC_NUM is utilized.
Now while filling the caps for encoder, it can lead to access the caps
array beyong 32 index. Hence leading to OOB write.
The fix counts the supported encoder and decoder. If the count is more than
max, then it skips accessing the caps.
When transmitting, infrared drivers expect an odd number of samples; iow
without a trailing space. No problems have been observed so far, so
this is just belt and braces.
Fixes: 9b6192589be7 ("media: lirc: implement scancode sending") Cc: stable@vger.kernel.org Signed-off-by: Sean Young <sean@mess.org> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Let's allocate the extent_cache tree without dynamic conditions to avoid a
missing condition causing a panic as below.
# create a file w/ a compressed flag
# disable the compression
# panic while updating extent_cache
F2FS-fs (dm-64): Swapfile: last extent is not aligned to section
F2FS-fs (dm-64): Swapfile (3) is not align to section: 1) creat(), 2) ioctl(F2FS_IOC_SET_PIN_FILE), 3) fallocate(2097152 * N)
Adding 124996k swap on ./swap-file. Priority:0 extents:2 across:17179494468k
==================================================================
BUG: KASAN: null-ptr-deref in instrument_atomic_read_write out/common/include/linux/instrumented.h:101 [inline]
BUG: KASAN: null-ptr-deref in atomic_try_cmpxchg_acquire out/common/include/asm-generic/atomic-instrumented.h:705 [inline]
BUG: KASAN: null-ptr-deref in queued_write_lock out/common/include/asm-generic/qrwlock.h:92 [inline]
BUG: KASAN: null-ptr-deref in __raw_write_lock out/common/include/linux/rwlock_api_smp.h:211 [inline]
BUG: KASAN: null-ptr-deref in _raw_write_lock+0x5a/0x110 out/common/kernel/locking/spinlock.c:295
Write of size 4 at addr 0000000000000030 by task syz-executor154/3327
With gcc and W=1 option, there's a warning like this:
fs/f2fs/compress.c: In function ‘f2fs_init_page_array_cache’:
fs/f2fs/compress.c:1984:47: error: ‘%u’ directive writing between
1 and 7 bytes into a region of size between 5 and 8
[-Werror=format-overflow=]
1984 | sprintf(slab_name, "f2fs_page_array_entry-%u:%u", MAJOR(dev),
MINOR(dev));
| ^~
String "f2fs_page_array_entry-%u:%u" can up to 35. The first "%u" can up
to 4 and the second "%u" can up to 7, so total size is "24 + 4 + 7 = 35".
slab_name's size should be 35 rather than 32.
Cc: stable@vger.kernel.org Signed-off-by: Su Hui <suhui@nfschina.com> Reviewed-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently we set SMBHSTCNT_LAST_BYTE only after the host has started
receiving the last byte. If we get e.g. preempted before setting
SMBHSTCNT_LAST_BYTE, the host may be finished with receiving the byte
before SMBHSTCNT_LAST_BYTE is set.
Therefore change the code to set SMBHSTCNT_LAST_BYTE before writing
SMBHSTSTS_BYTE_DONE for the byte before the last byte. Now the code
is also consistent with what we do in i801_isr_byte_done().
Reported-by: Jean Delvare <jdelvare@suse.com> Closes: https://lore.kernel.org/linux-i2c/20230828152747.09444625@endymion.delvare/ Cc: stable@vger.kernel.org Acked-by: Andi Shyti <andi.shyti@kernel.org> Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com> Reviewed-by: Jean Delvare <jdelvare@suse.de> Signed-off-by: Wolfram Sang <wsa@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In gfs2_fill_super(), when mounting a gfs2 filesystem is interrupted,
kthread_create() can return -EINTR. When that happens, we roll back
what has already been done and abort the mount.
Since commit 62dd0f98a0e5 ("gfs2: Flag a withdraw if init_threads()
fails), we are calling gfs2_withdraw_delayed() in gfs2_fill_super();
first via gfs2_make_fs_rw(), then directly. But gfs2_withdraw_delayed()
only marks the filesystem as withdrawing and relies on a caller further
up the stack to do the actual withdraw, which doesn't exist in the
gfs2_fill_super() case. Because the filesystem is marked as withdrawing
/ withdrawn, function gfs2_lm_unmount() doesn't release the dlm
lockspace, so when we try to mount that filesystem again, we get:
Since commit b77b4a4815a9 ("gfs2: Rework freeze / thaw logic"), the
deadlock this gfs2_withdraw_delayed() call was supposed to work around
cannot occur anymore because freeze_go_callback() won't take the
sb->s_umount semaphore unconditionally anymore, so we can get rid of the
gfs2_withdraw_delayed() in gfs2_fill_super() entirely.
Reported-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com> Cc: stable@vger.kernel.org # v6.5+ Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Background: Turris Omnia (Armada 385); eth2 (mvneta) connected to SFP bus;
SFP module is present, but no fiber connected, so definitely no carrier.
After booting, eth2 is down, but netdev LED trigger surprisingly reports
link active. Then, after "ip link set eth2 up", the link indicator goes
away - as I would have expected it from the beginning.
It turns out, that the default carrier state after netdev creation is
"carrier ok". Some ethernet drivers explicitly call netif_carrier_off
during probing, others (like mvneta) don't - which explains the current
behaviour: only when the device is brought up, phylink_start calls
netif_carrier_off.
Fix this for all drivers using phylink, by calling netif_carrier_off in
phylink_create.
Fixes: 089381b27abe ("leds: initial support for Turris Omnia LEDs") Cc: stable@vger.kernel.org Suggested-by: Andrew Lunn <andrew@lunn.ch> Signed-off-by: Klaus Kudielka <klaus.kudielka@gmail.com> Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When LAN9303 is MDIO-connected two callchains exist into
mdio->bus->write():
1. switch ports 1&2 ("physical" PHYs):
virtual (switch-internal) MDIO bus (lan9303_switch_ops->phy_{read|write})->
lan9303_mdio_phy_{read|write} -> mdiobus_{read|write}_nested
2. LAN9303 virtual PHY:
virtual MDIO bus (lan9303_phy_{read|write}) ->
lan9303_virt_phy_reg_{read|write} -> regmap -> lan9303_mdio_{read|write}
If the latter functions just take
mutex_lock(&sw_dev->device->bus->mdio_lock) it triggers a LOCKDEP
false-positive splat. It's false-positive because the first
mdio_lock in the second callchain above belongs to virtual MDIO bus, the
second mdio_lock belongs to physical MDIO bus.
Consequent annotation in lan9303_mdio_{read|write} as nested lock
(similar to lan9303_mdio_phy_{read|write}, it's the same physical MDIO bus)
prevents the following splat:
WARNING: possible circular locking dependency detected
5.15.71 #1 Not tainted
------------------------------------------------------
kworker/u4:3/609 is trying to acquire lock: ffff000011531c68 (lan9303_mdio:131:(&lan9303_mdio_regmap_config)->lock){+.+.}-{3:3}, at: regmap_lock_mutex
but task is already holding lock: ffff0000114c44d8 (&bus->mdio_lock){+.+.}-{3:3}, at: mdiobus_read
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #1 (&bus->mdio_lock){+.+.}-{3:3}:
lock_acquire
__mutex_lock
mutex_lock_nested
lan9303_mdio_read
_regmap_read
regmap_read
lan9303_probe
lan9303_mdio_probe
mdio_probe
really_probe
__driver_probe_device
driver_probe_device
__device_attach_driver
bus_for_each_drv
__device_attach
device_initial_probe
bus_probe_device
deferred_probe_work_func
process_one_work
worker_thread
kthread
ret_from_fork
-> #0 (lan9303_mdio:131:(&lan9303_mdio_regmap_config)->lock){+.+.}-{3:3}:
__lock_acquire
lock_acquire.part.0
lock_acquire
__mutex_lock
mutex_lock_nested
regmap_lock_mutex
regmap_read
lan9303_phy_read
dsa_slave_phy_read
__mdiobus_read
mdiobus_read
get_phy_device
mdiobus_scan
__mdiobus_register
dsa_register_switch
lan9303_probe
lan9303_mdio_probe
mdio_probe
really_probe
__driver_probe_device
driver_probe_device
__device_attach_driver
bus_for_each_drv
__device_attach
device_initial_probe
bus_probe_device
deferred_probe_work_func
process_one_work
worker_thread
kthread
ret_from_fork
other info that might help us debug this:
Possible unsafe locking scenario:
CPU0 CPU1
---- ----
lock(&bus->mdio_lock);
lock(lan9303_mdio:131:(&lan9303_mdio_regmap_config)->lock);
lock(&bus->mdio_lock);
lock(lan9303_mdio:131:(&lan9303_mdio_regmap_config)->lock);
*** DEADLOCK ***
5 locks held by kworker/u4:3/609:
#0: ffff000002842938 ((wq_completion)events_unbound){+.+.}-{0:0}, at: process_one_work
#1: ffff80000bacbd60 (deferred_probe_work){+.+.}-{0:0}, at: process_one_work
#2: ffff000007645178 (&dev->mutex){....}-{3:3}, at: __device_attach
#3: ffff8000096e6e78 (dsa2_mutex){+.+.}-{3:3}, at: dsa_register_switch
#4: ffff0000114c44d8 (&bus->mdio_lock){+.+.}-{3:3}, at: mdiobus_read
stack backtrace:
CPU: 1 PID: 609 Comm: kworker/u4:3 Not tainted 5.15.71 #1
Workqueue: events_unbound deferred_probe_work_func
Call trace:
dump_backtrace
show_stack
dump_stack_lvl
dump_stack
print_circular_bug
check_noncircular
__lock_acquire
lock_acquire.part.0
lock_acquire
__mutex_lock
mutex_lock_nested
regmap_lock_mutex
regmap_read
lan9303_phy_read
dsa_slave_phy_read
__mdiobus_read
mdiobus_read
get_phy_device
mdiobus_scan
__mdiobus_register
dsa_register_switch
lan9303_probe
lan9303_mdio_probe
...
Cc: stable@vger.kernel.org Fixes: dc7005831523 ("net: dsa: LAN9303: add MDIO managed mode support") Signed-off-by: Alexander Sverdlin <alexander.sverdlin@siemens.com> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Link: https://lore.kernel.org/r/20231027065741.534971-1-alexander.sverdlin@siemens.com Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fix kernel crash in AP bus code caused by very early invocation of the
config change callback function via SCLP.
After a fresh IML of the machine the crypto cards are still offline and
will get switched online only with activation of any LPAR which has the
card in it's configuration. A crypto card coming online is reported
to the LPAR via SCLP and the AP bus offers a callback function to get
this kind of information. However, it may happen that the callback is
invoked before the AP bus init function is complete. As the callback
triggers a synchronous AP bus scan, the scan may already run but some
internal states are not initialized by the AP bus init function resulting
in a crash like this:
This patch improves the ap_bus_force_rescan() function which is
invoked by the config change callback by checking if a first
initial AP bus scan has been done. If not, the force rescan request
is simple ignored. Anyhow it does not make sense to trigger AP bus
re-scans even before the very first bus scan is complete.
During SMBus block data read process, we have seen high interrupt rate
because of TX_EMPTY irq status while waiting for block length byte (the
first data byte after the address phase). The interrupt handler does not
do anything because the internal state is kept as STATUS_WRITE_IN_PROGRESS.
Hence, we should disable TX_EMPTY IRQ until I2C DesignWare receives
first data byte from I2C device, then re-enable it to resume SMBus
transaction.
It takes 0.789 ms for host to receive data length from slave.
Without the patch, i2c_dw_isr() is called 99 times by TX_EMPTY interrupt.
And it is none after applying the patch.
Cc: stable@vger.kernel.org Co-developed-by: Chuong Tran <chuong@os.amperecomputing.com> Signed-off-by: Chuong Tran <chuong@os.amperecomputing.com> Signed-off-by: Tam Nguyen <tamnguyenchi@os.amperecomputing.com> Acked-by: Jarkko Nikula <jarkko.nikula@linux.intel.com> Reviewed-by: Serge Semin <fancer.lancer@gmail.com> Signed-off-by: Wolfram Sang <wsa@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit abd3ac7902fb ("watchdog: sbsa: Support architecture version 1")
introduced new timer math for watchdog revision 1 with the 48 bit offset
register.
The gwdt->clk and timeout are u32, but the argument being calculated is
u64. Without a cast, the compiler performs u32 operations, truncating
intermediate steps, resulting in incorrect values.
A watchdog revision 1 implementation with a gwdt->clk of 1GHz and a
timeout of 600s writes 3647256576 to the one shot watchdog instead of 300000000000, resulting in the watchdog firing in 3.6s instead of 600s.
Force u64 math by casting the first argument (gwdt->clk) as a u64. Make
the order of operations explicit with parenthesis.
-EOPNOTSUPP is the return value that implements a "no-op" hook, not 0.
Without this fix having only the BPF LSM enabled (with no programs
attached) can cause uninitialized variable reads in
nfsd4_encode_fattr(), because the BPF hook returns 0 without touching
the 'ctxlen' variable and the corresponding 'contextlen' variable in
nfsd4_encode_fattr() remains uninitialized, yet being treated as valid
based on the 0 return value.
Cc: stable@vger.kernel.org Fixes: 98e828a0650f ("security: Refactor declaration of LSM hooks") Reported-by: Benjamin Coddington <bcodding@redhat.com> Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com> Signed-off-by: Paul Moore <paul@paul-moore.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The intent for the commit was to be able to detect carrier loss/gain
for just the NIC connected to the BMC. The unwanted effect is a
carrier loss for auxiliary paths also causes the BMC to lose
carrier. The BMC never regains carrier despite the secondary NIC
regaining a link.
This change, when merged, needs to be backported to stable kernels.
5.4-stable, 5.10-stable, 5.15-stable, 6.1-stable, 6.5-stable
Fixes: 3780bb29311e ("ncsi: Propagate carrier gain/loss events to the NCSI controller") CC: stable@vger.kernel.org Signed-off-by: Johnathan Mantey <johnathanx.mantey@intel.com> Reviewed-by: Simon Horman <horms@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
HP 255 G10 has a mute LED that can be made to work using quirk
ALC236_FIXUP_HP_MUTE_LED_COEFBIT2.
Enable already existing quirk - at correct line to keep order
Add ALC295 to pin fall back table.
Remove 5 pin quirks for Dell ALC295.
ALC295 was only support MIC2 for external MIC function.
ALC295 assigned model "ALC269_FIXUP_DELL1_MIC_NO_PRESENCE" for pin
fall back table.
It was assigned wrong model. So, let's remove it.
Fixes: fbc571290d9f ("ALSA: hda/realtek - Fixed Headphone Mic can't record on Dell platform") Signed-off-by: Kailang Yang <kailang@realtek.com> Cc: <stable@vger.kernel.org> Link: https://lore.kernel.org/r/7c1998e873834df98d59bd7e0d08c72e@realtek.com Signed-off-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As reported recently, ALSA core info helper may cause a deadlock at
the forced device disconnection during the procfs operation.
The proc_remove() (that is called from the snd_card_disconnect()
helper) has a synchronization of the pending procfs accesses via
wait_for_completion(). Meanwhile, ALSA procfs helper takes the global
mutex_lock(&info_mutex) at both the proc_open callback and
snd_card_info_disconnect() helper. Since the proc_open can't finish
due to the mutex lock, wait_for_completion() never returns, either,
hence it deadlocks.
This patch is a workaround for avoiding the deadlock scenario above.
The basic strategy is to move proc_remove() call outside the mutex
lock. proc_remove() can work gracefully without extra locking, and it
can delete the tree recursively alone. So, we call proc_remove() at
snd_info_card_disconnection() at first, then delete the rest resources
recursively within the info_mutex lock.
After the change, the function snd_info_disconnect() doesn't do
disconnection by itself any longer, but it merely clears the procfs
pointer. So rename the function to snd_info_clear_entries() for
avoiding confusion.
The similar change is applied to snd_info_free_entry(), too. Since
the proc_remove() is called only conditionally with the non-NULL
entry->p, it's skipped after the snd_info_clear_entries() call.
This happens because -EAGAIN error returned from btrfs_reserve_extent()
called from btrfs_new_extent_direct() is spilling over to the userland.
btrfs_reserve_extent() returns -EAGAIN when there is no active zone
available. Then, the caller should wait for some other on-going IO to
finish a zone and retry the allocation.
This logic is already implemented for buffered write in cow_file_range(),
but it is missing for the direct IO counterpart. Implement the same logic
for it.
Reported-by: Shinichiro Kawasaki <shinichiro.kawasaki@wdc.com> Fixes: 2ce543f47843 ("btrfs: zoned: wait until zone is finished when allocation didn't progress") CC: stable@vger.kernel.org # 6.1+ Tested-by: Shinichiro Kawasaki <shinichiro.kawasaki@wdc.com> Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com> Signed-off-by: David Sterba <dsterba@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Because on v3 inodes, di_flushiter doesn't exist. It overlaps with
zero padding in the inode, except when NREXT64=1 configurations are
in use and the zero padding is no longer padding but holds the 64
bit extent counter.
This manifests obviously on big endian platforms (e.g. s390) because
the log dinode is in host order and the overlap is the LSBs of the
extent count field. It is not noticed on little endian machines
because the overlap is at the MSB end of the extent count field and
we need to get more than 2^^48 extents in the inode before it
manifests. i.e. the heat death of the universe will occur before we
see the problem in little endian machines.
This is a zero-day issue for NREXT64=1 configuraitons on big endian
machines. Fix it by only clearing di_flushiter on v2 inodes during
recovery.
Fixes: 9b7d16e34bbe ("xfs: Introduce XFS_DIFLAG2_NREXT64 and associated helpers")
cc: stable@kernel.org # 5.19+ Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: "Darrick J. Wong" <djwong@kernel.org> Signed-off-by: Chandan Babu R <chandanbabu@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Each smb_rqst struct contains two things: an array of kvecs (rq_iov) that
contains the protocol data for an RPC op and an iterator (rq_iter) that
contains the data payload of an RPC op. When an smb_rqst is allocated
rq_iter is it always cleared, but we don't set it up unless we're going to
use it.
The functions that determines the size of the ciphertext buffer that will
be needed to encrypt a request, cifs_get_num_sgs(), assumes that rq_iter is
always initialised - and employs user_backed_iter() to check that the
iterator isn't user-backed. This used to incidentally work, because
->user_backed was set to false because the iterator has never been
initialised, but with commit f1b4cb650b9a0eeba206d8f069fcdc532bfbcd74[1]
which changes user_backed_iter() to determine this based on the iterator
type insted, a warning is now emitted:
The problem is that rq_iter was never set, so the type is 0 (ie. ITER_UBUF)
which causes user_backed_iter() to return true. The code doesn't
malfunction because it checks the size of the iterator - which is 0.
Fix cifs_get_num_sgs() to ignore rq_iter if its count is 0, thereby
bypassing the warnings.
It might be better to explicitly initialise rq_iter to a zero-length
ITER_BVEC, say, as it can always be reinitialised later.
Fixes: d08089f649a0 ("cifs: Change the I/O paths to use an iterator rather than a page list") Reported-by: Damian Tometzki <damian@riscv-rocks.de> Closes: https://lore.kernel.org/r/ZUfQo47uo0p2ZsYg@fedora.fritz.box/ Tested-by: Damian Tometzki <damian@riscv-rocks.de> Cc: stable@vger.kernel.org
cc: Eric Biggers <ebiggers@kernel.org>
cc: linux-cifs@vger.kernel.org
cc: linux-fsdevel@vger.kernel.org Link: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=f1b4cb650b9a0eeba206d8f069fcdc532bfbcd74 Reviewed-by: Paulo Alcantara (SUSE) <pc@manguebit.com> Signed-off-by: David Howells <dhowells@redhat.com> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If the mount command has specified multichannel as a mount option,
but multichannel is found to be unsupported by the server at the time
of mount, we set chan_max to 1. Which means that the user needs to
remount the share if the server starts supporting multichannel.
This change removes this reset. What it means is that if the user
specified multichannel or max_channels during mount, and at this
time, multichannel is not supported, but the server starts supporting
it at a later point, the client will be capable of scaling out the
number of channels.
Cc: stable@vger.kernel.org Signed-off-by: Shyam Prasad N <sprasad@microsoft.com> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
During a session reconnect, it is possible that the
server moved to another physical server (happens in case
of Azure files). So at this time, force a query of server
interfaces again (in case of multichannel session), such
that the secondary channels connect to the right
IP addresses (possibly updated now).
Cc: stable@vger.kernel.org Signed-off-by: Shyam Prasad N <sprasad@microsoft.com> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We introduced a helper function to be used by non-cifsd threads to
mark the connection for reconnect. For multichannel, when only
a particular channel needs to be reconnected, this had a bug.
This change fixes that by marking that particular channel
for reconnect.
Fixes: dca65818c80c ("cifs: use a different reconnect helper for non-cifsd threads") Cc: stable@vger.kernel.org Signed-off-by: Shyam Prasad N <sprasad@microsoft.com> Reviewed-by: Paulo Alcantara (SUSE) <pc@manguebit.com> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
All release_mid() callers seem to hold a reference of @mid so there is
no need to call kref_put(&mid->refcount, __release_mid) under
@server->mid_lock spinlock. If they don't, then an use-after-free bug
would have occurred anyways.
By getting rid of such spinlock also fixes a potential deadlock as
shown below
CPU 0 CPU 1
------------------------------------------------------------------
cifs_demultiplex_thread() cifs_debug_data_proc_show()
release_mid()
spin_lock(&server->mid_lock);
spin_lock(&cifs_tcp_ses_lock)
spin_lock(&server->mid_lock)
__release_mid()
smb2_find_smb_tcon()
spin_lock(&cifs_tcp_ses_lock) *deadlock*
Cc: stable@vger.kernel.org Signed-off-by: Paulo Alcantara (SUSE) <pc@manguebit.com> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The following UAF was triggered when running fstests generic/072 with
KASAN enabled against Windows Server 2022 and mount options
'multichannel,max_channels=2,vers=3.1.1,mfsymlinks,noperm'
BUG: KASAN: slab-use-after-free in smb2_query_info_compound+0x423/0x6d0 [cifs]
Read of size 8 at addr ffff888014941048 by task xfs_io/27534
This is a race between open_cached_dir() and cached_dir_lease_break()
where the cache entry for the open directory handle receives a lease
break while creating it. And before returning from open_cached_dir(),
we put the last reference of the new @cfid because of
!@cfid->has_lease.
Besides the UAF, while running xfstests a lot of missed lease breaks
have been noticed in tests that run several concurrent statfs(2) calls
on those cached fids
CIFS: VFS: \\w22-root1.gandalf.test No task to wake, unknown frame...
CIFS: VFS: \\w22-root1.gandalf.test Cmd: 18 Err: 0x0 Flags: 0x1...
CIFS: VFS: \\w22-root1.gandalf.test smb buf 00000000715bfe83 len 108
CIFS: VFS: Dump pending requests:
CIFS: VFS: \\w22-root1.gandalf.test No task to wake, unknown frame...
CIFS: VFS: \\w22-root1.gandalf.test Cmd: 18 Err: 0x0 Flags: 0x1...
CIFS: VFS: \\w22-root1.gandalf.test smb buf 000000005aa7316e len 108
...
To fix both, in open_cached_dir() ensure that @cfid->has_lease is set
right before sending out compounded request so that any potential
lease break will be get processed by demultiplex thread while we're
still caching @cfid. And, if open failed for some reason, re-check
@cfid->has_lease to decide whether or not put lease reference.
Cc: stable@vger.kernel.org Signed-off-by: Paulo Alcantara (SUSE) <pc@manguebit.com> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: stable@vger.kernel.org Signed-off-by: Paulo Alcantara (SUSE) <pc@manguebit.com> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When multiple mounts are to the same share from the same client it was not
possible to determine which section of /proc/fs/cifs/Stats (and DebugData)
correspond to that mount. In some recent examples this turned out to be
a significant problem when trying to analyze performance data - since
there are many cases where unless we know the tree id and session id we
can't figure out which stats (e.g. number of SMB3.1.1 requests by type,
the total time they take, which is slowest, how many fail etc.) apply to
which mount. The only existing loosely related ioctl CIFS_IOC_GET_MNT_INFO
does not return the information needed to uniquely identify which tcon
is which mount although it does return various flags and device info.
Add a cifs.ko ioctl CIFS_IOC_GET_TCON_INFO (0x800ccf0c) to return tid,
session id, tree connect count.
Cc: stable@vger.kernel.org Reviewed-by: Shyam Prasad N <sprasad@microsoft.com> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
For example:
touch -h -t 02011200 testfile
where testfile is a symlink would not change the timestamp, but
touch -t 02011200 testfile
does work to change the timestamp of the target
Suggested-by: David Howells <dhowells@redhat.com> Reported-by: Micah Veilleux <micah.veilleux@iba-group.com> Closes: https://bugzilla.samba.org/show_bug.cgi?id=14476 Cc: stable@vger.kernel.org Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fixes some xfstests including generic/564 and generic/157
The "sfu" mount option can be useful for creating special files (character
and block devices in particular) but could not create FIFOs. It did
recognize existing empty files with the "system" attribute flag as FIFOs
but this is too general, so to support creating FIFOs more safely use a new
tag (but the same length as those for char and block devices ie "IntxLNK"
and "IntxBLK") "LnxFIFO" to indicate that the file should be treated as a
FIFO (when mounted with the "sfu"). For some additional context note that
"sfu" followed the way that "Services for Unix" on Windows handled these
special files (at least for character and block devices and symlinks),
which is different than newer Windows which can handle special files
as reparse points (which isn't an option to many servers).
Cc: stable@vger.kernel.org Reviewed-by: Paulo Alcantara (SUSE) <pc@manguebit.com> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
struct timespec64 has unused bits in the tv_nsec field that can be used
for other purposes. In future patches, we're going to change how the
inode->i_ctime is accessed in certain inodes in order to make use of
them. In order to do that safely though, we'll need to eradicate raw
accesses of the inode->i_ctime field from the kernel.
Add new accessor functions for the ctime that we use to replace them.
Reviewed-by: Jan Kara <jack@suse.cz> Reviewed-by: Luis Chamberlain <mcgrof@kernel.org> Signed-off-by: Jeff Layton <jlayton@kernel.org> Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Message-Id: <20230705185812.579118-2-jlayton@kernel.org> Signed-off-by: Christian Brauner <brauner@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Use the low-power states of the underlying platform to enable runtime PM.
If the platform doesn't support runtime D3, then enabling default RPM will
result in the controller malfunctioning, as in the case of hotplug devices
not being detected because of a failed interrupt generation.
When calculating the pfn for the iitlbt/idtlbt instruction, do not
drop the upper 5 address bits. This doesn't seem to have an effect
on physical hardware which uses less physical address bits, but in
qemu the missing bits are visible.
Bail out early with error message when trying to boot a 64-bit kernel on
32-bit machines. This fixes the previous commit to include the check for
true 64-bit kernels as well.
When dealing with hugetlb pages, manipulating struct page pointers
directly can get to wrong struct page, since struct page is not guaranteed
to be contiguous on SPARSEMEM without VMEMMAP. Use nth_page() to handle
it properly.
A wrong or non-existing page might be tried to be grabbed, either
leading to a non freeable page or kernel memory access errors. No bug
is reported. It comes from code inspection.
Link: https://lkml.kernel.org/r/20230913201248.452081-3-zi.yan@sent.com Fixes: 57a196a58421 ("hugetlb: simplify hugetlb handling in follow_page_mask") Signed-off-by: Zi Yan <ziy@nvidia.com> Reviewed-by: Muchun Song <songmuchun@bytedance.com> Cc: David Hildenbrand <david@redhat.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
follow_page() doesn't use FOLL_PIN, meanwhile hugetlb seems to not be the
target of FOLL_WRITE either. However add the checks.
Namely, either the need to CoW due to missing write bit, or proper
unsharing on !AnonExclusive pages over R/O pins to reject the follow page.
That brings this function closer to follow_hugetlb_page().
So we don't care before, and also for now. But we'll care if we switch
over slow-gup to use hugetlb_follow_page_mask(). We'll also care when to
return -EMLINK properly, as that's the gup internal api to mean "we should
unshare". Not really needed for follow page path, though.
When at it, switching the try_grab_page() to use WARN_ON_ONCE(), to be
clear that it just should never fail. When error happens, instead of
setting page==NULL, capture the errno instead.
Link: https://lkml.kernel.org/r/20230628215310.73782-3-peterx@redhat.com Signed-off-by: Peter Xu <peterx@redhat.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Reviewed-by: David Hildenbrand <david@redhat.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Hugh Dickins <hughd@google.com> Cc: James Houghton <jthoughton@google.com> Cc: Jason Gunthorpe <jgg@nvidia.com> Cc: John Hubbard <jhubbard@nvidia.com> Cc: Kirill A . Shutemov <kirill@shutemov.name> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Yang Shi <shy828301@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Stable-dep-of: 426056efe835 ("mm/hugetlb: use nth_page() in place of direct struct page manipulation") Signed-off-by: Sasha Levin <sashal@kernel.org>
The stuttering code isn't functioning as expected. Ideally, it should
pause the torture threads for a designated period before resuming. Yet,
it fails to halt the test for the correct duration. Additionally, a race
condition exists, potentially causing the stuttering code to pause for
an extended period if the 'spt' variable is non-zero due to the stutter
orchestration thread's inadequate CPU time.
Moreover, over-stuttering can hinder RCU's progress on TREE07 kernels.
This happens as the stuttering code may run within a softirq due to RCU
callbacks. Consequently, ksoftirqd keeps a CPU busy for several seconds,
thus obstructing RCU's progress. This situation triggers a warning
message in the logs:
This warning suggests that an RCU torture object, although invisible to
RCU readers, couldn't make it past the pipe array and be freed -- a
strong indication that there weren't enough grace periods during the
stutter interval.
To address these issues, this patch sets the "stutter end" time to an
absolute point in the future set by the main stutter thread. This is
then used for waiting in stutter_wait(). While the stutter thread still
defines this absolute time, the waiters' waiting logic doesn't rely on
the stutter thread receiving sufficient CPU time to halt the stuttering
as the halting is now self-controlled.
Cc: stable@vger.kernel.org Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org> Signed-off-by: Paul E. McKenney <paulmck@kernel.org> Signed-off-by: Frederic Weisbecker <frederic@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The current torture-test sleeps are waiting for a duration, but there
are situations where it is better to wait for an absolute time, for
example, when ending a stutter interval. This commit therefore adds
an hrtimer mode parameter to torture_hrtimeout_ns(). Why not also the
other torture_hrtimeout_*() functions? The theory is that most absolute
times will be in nanoseconds, especially not (say) jiffies.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org> Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Stable-dep-of: cca42bd8eb1b ("rcutorture: Fix stuttering races and other issues") Signed-off-by: Sasha Levin <sashal@kernel.org>
In order to gain better race coverage, move the test start/stop
waits in stutter_wait() to torture_hrtimeout_jiffies().
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Stable-dep-of: cca42bd8eb1b ("rcutorture: Fix stuttering races and other issues") Signed-off-by: Sasha Levin <sashal@kernel.org>
Given that it is expected that more code will use torture_hrtimeout_*(),
including for longer timeouts, make it use TASK_IDLE instead of
TASK_UNINTERRUPTIBLE.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Stable-dep-of: cca42bd8eb1b ("rcutorture: Fix stuttering races and other issues") Signed-off-by: Sasha Levin <sashal@kernel.org>
This commit adds a kthread-creation callback to the
_torture_create_kthread() function, which allows callers of a new
torture_create_kthread_cb() macro to specify a function to be invoked
after the kthread is created but before it is awakened for the first time.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org> Cc: Dietmar Eggemann <dietmar.eggemann@arm.com> Cc: Josh Triplett <josh@joshtriplett.org> Cc: Juri Lelli <juri.lelli@redhat.com> Cc: Valentin Schneider <vschneid@redhat.com> Cc: Dietmar Eggemann <dietmar.eggemann@arm.com> Cc: kernel-team@android.com Reviewed-by: Joel Fernandes (Google) <joel@joelfernandes.org> Acked-by: John Stultz <jstultz@google.com>
Stable-dep-of: cca42bd8eb1b ("rcutorture: Fix stuttering races and other issues") Signed-off-by: Sasha Levin <sashal@kernel.org>
Commit ac91e6980563 ("PCI: Unify delay handling for reset and resume")
shortened an unconditional 1 sec delay after a Secondary Bus Reset to 100
msec for PCIe (per PCIe r6.1 sec 6.6.1). The 1 sec delay is only required
for Conventional PCI.
But it turns out that there are PCIe devices which require a longer delay
than prescribed before first config space access after reset recovery or
resume from D3cold:
Chad reports that a "VideoPropulsion Torrent QN16e" MPEG QAM Modulator
"raises a PCI system error (PERR), as reported by the IPMI event log, and
the hardware itself would suffer a catastrophic event, cycling the server"
unless the longer delay is observed.
The card is specified to conform to PCIe r1.0 and indeed only supports Gen1
speed (2.5 GT/s) according to lspci. PCIe r1.0 sec 7.6 prescribes the same
100 msec delay as PCIe r6.1 sec 6.6.1:
To allow components to perform internal initialization, system software
must wait for at least 100 ms from the end of a reset (cold/warm/hot)
before it is permitted to issue Configuration Requests
The behavior of the Torrent QN16e card thus appears to be a quirk. Treat
it as such and lengthen the reset delay for this specific device.
Fixes: ac91e6980563 ("PCI: Unify delay handling for reset and resume") Link: https://lore.kernel.org/r/47727e792c7f0282dc144e3ec8ce8eb6e713394e.1695304512.git.lukas@wunner.de Reported-by: Chad Schroeder <CSchroeder@sonifi.com> Closes: https://lore.kernel.org/linux-pci/DM6PR16MB2844903E34CAB910082DF019B1FAA@DM6PR16MB2844.namprd16.prod.outlook.com/ Tested-by: Chad Schroeder <CSchroeder@sonifi.com> Signed-off-by: Lukas Wunner <lukas@wunner.de> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Cc: stable@vger.kernel.org # v5.4+ Signed-off-by: Sasha Levin <sashal@kernel.org>
Currently, The imx pgc power domain doesn't set the fwnode
pointer, which results in supply regulator device can't get
consumer imx pgc power domain device from fwnode when creating
a link.
This causes the driver core to instead try to create a link
between the parent gpc device of imx pgc power domain device and
supply regulator device. However, at this point, the gpc device
has already been bound, and the link creation will fail. So adding
the fwnode pointer to the imx pgc power domain device will fix
this issue.
Signed-off-by: Pengfei Li <pengfei.li_1@nxp.com> Tested-by: Emil Kronborg <emil.kronborg@protonmail.com> Fixes: 3fb16866b51d ("driver core: fw_devlink: Make cycle detection more robust") Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20231020185949.537083-1-pengfei.li_1@nxp.com Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The commit c494a447c14e ("soc: bcm: bcm2835-power: Refactor ASB control")
refactored the ASB control by using a general function to handle both
the enable and disable. But this patch introduced a subtle regression:
we need to check if !!(readl(base + reg) & ASB_ACK) == enable, not just
check if (readl(base + reg) & ASB_ACK) == true.
Currently, this is causing an invalid register state in V3D when
unloading and loading the driver, because `bcm2835_asb_disable()` will
return -ETIMEDOUT and `bcm2835_asb_power_off()` will fail to disable the
ASB slave for V3D.
The CXL subsystem, at cxl_mem ->probe() time, establishes a lineage of
ports (struct cxl_port objects) between an endpoint and the root of a
CXL topology. Each port including the endpoint port is attached to the
cxl_port driver.
Given that setup, it follows that when either any port in that lineage
goes through a cxl_port ->remove() event, or the memdev goes through a
cxl_mem ->remove() event. The hierarchy below the removed port, or the
entire hierarchy if the memdev is removed needs to come down.
The delete_endpoint() callback is careful to check whether it is being
called to tear down the hierarchy, or if it is only being called to
teardown the memdev because an ancestor port is going through
->remove().
That care needs to take the device_lock() of the endpoint's parent.
Which requires 2 bugs to be fixed:
1/ A reference on the parent is needed to prevent use-after-free
scenarios like this signature:
BUG: spinlock bad magic on CPU#0, kworker/u56:0/11
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS edk2-20230524-3.fc38 05/24/2023
Workqueue: cxl_port detach_memdev [cxl_core]
RIP: 0010:spin_bug+0x65/0xa0
Call Trace:
do_raw_spin_lock+0x69/0xa0
__mutex_lock+0x695/0xb80
delete_endpoint+0xad/0x150 [cxl_core]
devres_release_all+0xb8/0x110
device_unbind_cleanup+0xe/0x70
device_release_driver_internal+0x1d2/0x210
detach_memdev+0x15/0x20 [cxl_core]
process_one_work+0x1e3/0x4c0
worker_thread+0x1dd/0x3d0
2/ In the case of RCH topologies, the parent device that needs to be
locked is not always @port->dev as returned by cxl_mem_find_port(), use
endpoint->dev.parent instead.
Fixes: 8dd2bc0f8e02 ("cxl/mem: Add the cxl_mem driver") Cc: <stable@vger.kernel.org> Reported-by: Robert Richter <rrichter@amd.com> Closes: http://lore.kernel.org/r/20231018171713.1883517-2-rrichter@amd.com Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Root decoder granularity must match value from CFWMS, which may not
be the region's granularity for non-interleaved root decoders.
So when calculating granularities for host bridge decoders, use the
region's granularity instead of the root decoder's granularity to ensure
the correct granularities are set for the host bridge decoders and any
downstream switch decoders.
Test configuration is 1 host bridge * 2 switches * 2 endpoints per switch.
Region created with 2048 granularity using following command line:
Use "cxl list -PDE | grep granularity" to get a view of the granularity
set at each level of the topology.
Before this patch:
"interleave_granularity":2048,
"interleave_granularity":2048,
"interleave_granularity":512,
"interleave_granularity":2048,
"interleave_granularity":2048,
"interleave_granularity":512,
"interleave_granularity":256,
BIT 20: TIMEOUT error
The module has stalled too long in a frame. This happens when:
- The TX FIFO or RX FIFO is not handled and the bus is stuck in the
middle of a message,
- No STOP was issued and between messages,
- IBI manual is used and no decision was made.
The maximum stall period is 100 μs.
This can be considered as being just a warning as the system IRQ latency
can easily be greater than 100us.
Upon IBIWON timeout, the SDA line will always be kept low if we don't emit
a stop. Calling svc_i3c_master_emit_stop() there will let the bus return to
idle state.
MSTATUS[RXPEND] is only updated after the data transfer cycle started. This
creates an issue when the I3C clock is slow, and the CPU is running fast
enough that MSTATUS[RXPEND] may not be updated when the code reaches
checking point. As a result, mandatory data can be missed.
Add a wait for MSTATUS[COMPLETE] to ensure that all mandatory data is
already in FIFO. It also works without mandatory data.
If an In-Band Interrupt (IBI) occurs and IBI work thread is not immediately
scheduled, when svc_i3c_master_priv_xfers() initiates the I3C transfer and
attempts to send address 0x7e, the target interprets it as an
IBI handler and returns the target address 0x0a.
However, svc_i3c_master_priv_xfers() does not handle this case and proceeds
with other transfers, resulting in incorrect data being returned.
Add IBIWON check in svc_i3c_master_xfer(). In case this situation occurs,
return a failure to the driver.
The ibi work thread operates asynchronously with other transfers, such as
svc_i3c_master_priv_xfers(). Introduce mutex protection to ensure the
completion of the entire i3c/i2c transaction.
Commit 5e42bcbc3fef ("cxl/region: decrement ->nr_targets on error in
cxl_region_attach()") tried to avoid 'eiw' initialization errors when
->nr_targets exceeded 16, by just decrementing ->nr_targets when
cxl_region_setup_targets() failed.
Commit 86987c766276 ("cxl/region: Cleanup target list on attach error")
extended that cleanup to also clear cxled->pos and p->targets[pos]. The
initialization error was incidentally fixed separately by:
Commit 8d4285425714 ("cxl/region: Fix port setup uninitialized variable
warnings") which was merged a few days after 5e42bcbc3fef.
But now the original cleanup when cxl_region_setup_targets() fails
prevents endpoint and switch decoder resources from being reused:
1) the cleanup does not set the decoder's region to NULL, which results
in future dpa_size_store() calls returning -EBUSY
2) the decoder is not properly freed, which results in future commit
errors associated with the upstream switch
Now that the initialization errors were fixed separately, the proper
cleanup for this case is to just return immediately. Then the resources
associated with this target get cleanup up as normal when the failed
region is deleted.
The ->nr_targets decrement in the error case also helped prevent
a p->targets[] array overflow, so add a new check to prevent against
that overflow.
Tested by trying to create an invalid region for a 2 switch * 2 endpoint
topology, and then following up with creating a valid region.
Fixes: 5e42bcbc3fef ("cxl/region: decrement ->nr_targets on error in cxl_region_attach()") Cc: <stable@vger.kernel.org> Signed-off-by: Jim Harris <jim.harris@samsung.com> Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com> Acked-by: Dan Carpenter <dan.carpenter@linaro.org> Reviewed-by: Dave Jiang <dave.jiang@intel.com> Link: https://lore.kernel.org/r/169703589120.1202031.14696100866518083806.stgit@bgt-140510-bm03.eng.stellus.in Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently the offset into the device when looking for OTP
bits can go outside of the address of the MTD NOR devices,
and if that memory isn't readable, bad things happen
on the IXP4xx (added prints that illustrate the problem before
the crash):
cfi_intelext_otp_walk walk OTP on chip 0 start at reg_prot_offset 0x00000100
ixp4xx_copy_from copy from 0x00000100 to 0xc880dd78
cfi_intelext_otp_walk walk OTP on chip 0 start at reg_prot_offset 0x12000000
ixp4xx_copy_from copy from 0x12000000 to 0xc880dd78
8<--- cut here ---
Unable to handle kernel paging request at virtual address db000000
[db000000] *pgd=00000000
(...)
This happens in this case because the IXP4xx is big endian and
the 32- and 16-bit fields in the struct cfi_intelext_otpinfo are not
properly byteswapped. Compare to how the code in read_pri_intelext()
byteswaps the fields in struct cfi_pri_intelext.
Adding a small byte swapping loop for the OTP in read_pri_intelext()
and the crash goes away.
The problem went unnoticed for many years until I enabled
CONFIG_MTD_OTP on the IXP4xx as well, triggering the bug.
Defining a prctl flag as an int is a footgun because on a 64 bit machine
and with a variadic implementation of prctl (like in musl and glibc), when
used directly as a prctl argument, it can get casted to long with garbage
upper bits which would result in unexpected behaviors.
This patch changes the constant to an unsigned long to eliminate that
possibilities. This does not break UAPI.
I think that a stable backport would be "nice to have": to reduce the
chances that users build binaries that could end up with garbage bits in
their MDWE prctl arguments. We are not aware of anyone having yet
encountered this corner case with MDWE prctls but a backport would reduce
the likelihood it happens, since this sort of issues has happened with
other prctls. But If this is perceived as a backporting burden, I suppose
we could also live without a stable backport.
Link: https://lkml.kernel.org/r/20230828150858.393570-5-revest@chromium.org Fixes: b507808ebce2 ("mm: implement memory-deny-write-execute as a prctl") Signed-off-by: Florent Revest <revest@chromium.org> Suggested-by: Alexey Izbyshev <izbyshev@ispras.ru> Reviewed-by: David Hildenbrand <david@redhat.com> Reviewed-by: Kees Cook <keescook@chromium.org> Acked-by: Catalin Marinas <catalin.marinas@arm.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Ayush Jain <ayush.jain3@amd.com> Cc: Greg Thelen <gthelen@google.com> Cc: Joey Gouly <joey.gouly@arm.com> Cc: KP Singh <kpsingh@kernel.org> Cc: Mark Brown <broonie@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Peter Xu <peterx@redhat.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: Szabolcs Nagy <Szabolcs.Nagy@arm.com> Cc: Topi Miettinen <toiwoton@gmail.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When dealing with hugetlb pages, manipulating struct page pointers
directly can get to wrong struct page, since struct page is not guaranteed
to be contiguous on SPARSEMEM without VMEMMAP. Use pfn calculation to
handle it properly.
Without the fix, a wrong number of page might be skipped. Since skip cannot be
negative, scan_movable_page() will end early and might miss a movable page with
-ENOENT. This might fail offline_pages(). No bug is reported. The fix comes
from code inspection.
Link: https://lkml.kernel.org/r/20230913201248.452081-4-zi.yan@sent.com Fixes: eeb0efd071d8 ("mm,memory_hotplug: fix scan_movable_pages() for gigantic hugepages") Signed-off-by: Zi Yan <ziy@nvidia.com> Reviewed-by: Muchun Song <songmuchun@bytedance.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Patch series "Use nth_page() in place of direct struct page manipulation",
v3.
On SPARSEMEM without VMEMMAP, struct page is not guaranteed to be
contiguous, since each memory section's memmap might be allocated
independently. hugetlb pages can go beyond a memory section size, thus
direct struct page manipulation on hugetlb pages/subpages might give wrong
struct page. Kernel provides nth_page() to do the manipulation properly.
Use that whenever code can see hugetlb pages.
This patch (of 5):
When dealing with hugetlb pages, manipulating struct page pointers
directly can get to wrong struct page, since struct page is not guaranteed
to be contiguous on SPARSEMEM without VMEMMAP. Use nth_page() to handle
it properly.
Without the fix, page_kasan_tag_reset() could reset wrong page tags,
causing a wrong kasan result. No related bug is reported. The fix
comes from code inspection.
Link: https://lkml.kernel.org/r/20230913201248.452081-1-zi.yan@sent.com Link: https://lkml.kernel.org/r/20230913201248.452081-2-zi.yan@sent.com Fixes: 2813b9c02962 ("kasan, mm, arm64: tag non slab memory allocated via pagealloc") Signed-off-by: Zi Yan <ziy@nvidia.com> Reviewed-by: Muchun Song <songmuchun@bytedance.com> Cc: David Hildenbrand <david@redhat.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If the cmma no-dat feature is available the kernel page tables are walked
to identify and mark all pages which are used for address translation (all
region, segment, and page tables). In a subsequent loop all other pages are
marked as "no-dat" pages with the ESSA instruction.
This information is visible to the hypervisor, so that the hypervisor can
optimize purging of guest TLB entries. The initial loop however is
incorrect: only the first three of the four pages which belong to segment
and region tables will be marked as being used for DAT. The last page is
incorrectly marked as no-dat.
If the cmma no-dat feature is available all pages that are not used for
dynamic address translation are marked as "no-dat" with the ESSA
instruction. This information is visible to the hypervisor, so that the
hypervisor can optimize purging of guest TLB entries. This also means that
pages which are used for dynamic address translation must not be marked as
"no-dat", since the hypervisor may then incorrectly not purge guest TLB
entries.
Region and segment tables allocated via vmem_crst_alloc() are incorrectly
marked as "no-dat", as soon as slab_is_available() returns true.
Such tables are allocated e.g. when kernel page tables are split, memory is
hotplugged, or a DCSS segment is loaded.
Fix this by adding the missing arch_set_page_dat() call.
In case of the prep descriptor while the channel is already running, the
CCR register value stored into the channel could already have its EN bit
set. This would lead to a bad transfer since, at start transfer time,
enabling the channel while other registers aren't yet properly set.
To avoid this, ensure to mask the CCR_EN bit when storing the ccr value
into the mdma channel structure.
chameleon_parse_gdd() may fail for different reasons and end up
in the err tag. Make sure we at least always free the mcb_device
allocated with mcb_alloc_dev().
If mcb_device_register() fails, make sure to give up the reference
in the same place the device was added.
Fixes: 728ac3389296 ("mcb: mcb-parse: fix error handing in chameleon_parse_gdd()") Cc: stable <stable@kernel.org> Reviewed-by: Jose Javier Rodriguez Barbarin <JoseJavier.Rodriguez@duagon.com> Signed-off-by: Jorge Sanjuan Garcia <jorge.sanjuangarcia@duagon.com> Link: https://lore.kernel.org/r/20231019141434.57971-2-jorge.sanjuangarcia@duagon.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This commit fixes a bug in commit 9ed9895370ae ("driver core: Functional
dependencies tracking support") where the device link status was
incorrectly updated in the driver unbind path before all the device's
resources were released.
Fixes: 9ed9895370ae ("driver core: Functional dependencies tracking support") Cc: stable <stable@kernel.org> Reported-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> Closes: https://lore.kernel.org/all/20231014161721.f4iqyroddkcyoefo@pengutronix.de/ Signed-off-by: Saravana Kannan <saravanak@google.com> Cc: Thierry Reding <thierry.reding@gmail.com> Cc: Yang Yingliang <yangyingliang@huawei.com> Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Cc: Mark Brown <broonie@kernel.org> Cc: Matti Vaittinen <mazziesaccount@gmail.com> Cc: James Clark <james.clark@arm.com> Acked-by: "Rafael J. Wysocki" <rafael@kernel.org> Tested-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> Acked-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> Link: https://lore.kernel.org/r/20231018013851.3303928-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
A synthetic event is created by the synthetic event interface that can
read both user or kernel address memory. In reality, it reads any
arbitrary memory location from within the kernel. If the address space is
in USER (where CONFIG_ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE is set) then
it uses strncpy_from_user_nofault() to copy strings otherwise it uses
strncpy_from_kernel_nofault().
But since both functions use the same variable there's no annotation to
what that variable is (ie. __user). This makes sparse complain.
Quiet sparse by typecasting the strncpy_from_user_nofault() variable to
a __user pointer.
Link: https://lore.kernel.org/linux-trace-kernel/20231031151033.73c42e23@gandalf.local.home Cc: stable@vger.kernel.org Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Mark Rutland <mark.rutland@arm.com> Fixes: 0934ae9977c2 ("tracing: Fix reading strings from synthetic events"); Reported-by: kernel test robot <lkp@intel.com> Closes: https://lore.kernel.org/oe-kbuild-all/202311010013.fm8WTxa5-lkp@intel.com/ Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When execute the following command to test clone3 under !CONFIG_TIME_NS:
# make headers && cd tools/testing/selftests/clone3 && make && ./clone3
we can see the following error info:
# [7538] Trying clone3() with flags 0x80 (size 0)
# Invalid argument - Failed to create new process
# [7538] clone3() with flags says: -22 expected 0
not ok 18 [7538] Result (-22) is different than expected (0)
...
# Totals: pass:18 fail:1 xfail:0 xpass:0 skip:0 error:0
This is because if CONFIG_TIME_NS is not set, but the flag
CLONE_NEWTIME (0x80) is used to clone a time namespace, it
will return -EINVAL in copy_time_ns().
If kernel does not support CONFIG_TIME_NS, /proc/self/ns/time
will be not exist, and then we should skip clone3() test with
CLONE_NEWTIME.
With this patch under !CONFIG_TIME_NS:
# make headers && cd tools/testing/selftests/clone3 && make && ./clone3
...
# Time namespaces are not supported
ok 18 # SKIP Skipping clone3() with CLONE_NEWTIME
...
# Totals: pass:18 fail:0 xfail:0 xpass:0 skip:1 error:0
Link: https://lkml.kernel.org/r/1689066814-13295-1-git-send-email-yangtiezhu@loongson.cn Fixes: 515bddf0ec41 ("selftests/clone3: test clone3 with CLONE_NEWTIME") Signed-off-by: Tiezhu Yang <yangtiezhu@loongson.cn> Suggested-by: Thomas Gleixner <tglx@linutronix.de> Cc: Christian Brauner <brauner@kernel.org> Cc: Shuah Khan <shuah@kernel.org> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since bae1d3a05a8b, i2c transfers are non-atomic if preemption is
disabled. However, non-atomic i2c transfers require preemption (e.g. in
wait_for_completion() while waiting for the DMA).
panic() calls preempt_disable_notrace() before calling
emergency_restart(). Therefore, if an i2c device is used for the
restart, the xfer should be atomic. This avoids warnings like:
[ 12.667612] WARNING: CPU: 1 PID: 1 at kernel/rcu/tree_plugin.h:318 rcu_note_context_switch+0x33c/0x6b0
[ 12.676926] Voluntary context switch within RCU read-side critical section!
...
[ 12.742376] schedule_timeout from wait_for_completion_timeout+0x90/0x114
[ 12.749179] wait_for_completion_timeout from tegra_i2c_wait_completion+0x40/0x70
...
[ 12.994527] atomic_notifier_call_chain from machine_restart+0x34/0x58
[ 13.001050] machine_restart from panic+0x2a8/0x32c
Use !preemptible() instead, which is basically the same check as
pre-v5.2.
Fixes: bae1d3a05a8b ("i2c: core: remove use of in_atomic()") Cc: stable@vger.kernel.org # v5.2+ Suggested-by: Dmitry Osipenko <dmitry.osipenko@collabora.com> Acked-by: Wolfram Sang <wsa@kernel.org> Reviewed-by: Dmitry Osipenko <dmitry.osipenko@collabora.com> Tested-by: Nishanth Menon <nm@ti.com> Signed-off-by: Benjamin Bara <benjamin.bara@skidata.com> Link: https://lore.kernel.org/r/20230327-tegra-pmic-reboot-v7-2-18699d5dcd76@skidata.com Signed-off-by: Lee Jones <lee@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As the emergency restart does not call kernel_restart_prepare(), the
system_state stays in SYSTEM_RUNNING.
Since bae1d3a05a8b, this hinders i2c_in_atomic_xfer_mode() from becoming
active, and therefore might lead to avoidable warnings in the restart
handlers, e.g.:
[ 12.667612] WARNING: CPU: 1 PID: 1 at kernel/rcu/tree_plugin.h:318 rcu_note_context_switch+0x33c/0x6b0
[ 12.676926] Voluntary context switch within RCU read-side critical section!
...
[ 12.742376] schedule_timeout from wait_for_completion_timeout+0x90/0x114
[ 12.749179] wait_for_completion_timeout from tegra_i2c_wait_completion+0x40/0x70
...
[ 12.994527] atomic_notifier_call_chain from machine_restart+0x34/0x58
[ 13.001050] machine_restart from panic+0x2a8/0x32c
Avoid these by setting the correct system_state.
Fixes: bae1d3a05a8b ("i2c: core: remove use of in_atomic()") Cc: stable@vger.kernel.org # v5.2+ Reviewed-by: Dmitry Osipenko <dmitry.osipenko@collabora.com> Tested-by: Nishanth Menon <nm@ti.com> Signed-off-by: Benjamin Bara <benjamin.bara@skidata.com> Link: https://lore.kernel.org/r/20230327-tegra-pmic-reboot-v7-1-18699d5dcd76@skidata.com Signed-off-by: Lee Jones <lee@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since commit d7e7b9af104c ("fscrypt: stop using keyrings subsystem for
fscrypt_master_key"), xfstest generic/270 causes a WARNING when run on
f2fs with test_dummy_encryption in the mount options:
The cause of the WARNING is that not all encrypted inodes have been
evicted before fscrypt_destroy_keyring() is called, which violates an
assumption. This happens because the test uses an external quota file,
which gets automatically encrypted due to test_dummy_encryption.
Encryption of quota files has never really been supported. On ext4,
ext4_quota_read() does not decrypt the data, so encrypted quota files
are always considered invalid on ext4. On f2fs, f2fs_quota_read() uses
the pagecache, so trying to use an encrypted quota file gets farther,
resulting in the issue described above being possible. But this was
never intended to be possible, and there is no use case for it.
Therefore, make the quota support layer explicitly reject using
IS_ENCRYPTED inodes when quotaon is attempted.
Cc: stable@vger.kernel.org Signed-off-by: Eric Biggers <ebiggers@google.com> Signed-off-by: Jan Kara <jack@suse.cz>
Message-Id: <20230905003227.326998-1-ebiggers@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
JBD2 makes sure journal data is fallen on fs device by sync_blockdev(),
however, other process could intercept the EIO information from bdev's
mapping, which leads journal recovering successful even EIO occurs during
data written back to fs device.
We found this problem in our product, iscsi + multipath is chosen for block
device of ext4. Unstable network may trigger kpartx to rescan partitions in
device mapper layer. Detailed process is shown as following:
mount kpartx irq
jbd2_journal_recover
do_one_pass
memcpy(nbh->b_data, obh->b_data) // copy data to fs dev from journal
mark_buffer_dirty // mark bh dirty
vfs_read
generic_file_read_iter // dio
filemap_write_and_wait_range
__filemap_fdatawrite_range
do_writepages
block_write_full_folio
submit_bh_wbc
>> EIO occurs in disk <<
end_buffer_async_write
mark_buffer_write_io_error
mapping_set_error
set_bit(AS_EIO, &mapping->flags) // set!
filemap_check_errors
test_and_clear_bit(AS_EIO, &mapping->flags) // clear!
err2 = sync_blockdev
filemap_write_and_wait
filemap_check_errors
test_and_clear_bit(AS_EIO, &mapping->flags) // false
err2 = 0
Filesystem is mounted successfully even data from journal is failed written
into disk, and ext4/ocfs2 could become corrupted.
Fix it by comparing the wb_err state in fs block device before recovering
and after recovering.
A reproducer can be found in the kernel bugzilla referenced below.
projects / linux / log