Since the buffer cache for ntfs3 metadata is not released until the file
system is unmounted, allocating from the movable zone may result in cma
allocation failures. This is due to the page still being used by ntfs3,
leading to migration failures.
To address this, this commit use sb_bread_umovable() instead of
sb_bread(). This change prevents allocation from the movable zone,
ensuring compatibility with scenarios where the buffer head is not
released until unmount. This patch is inspired by commit a8ac900b8163("ext4: use non-movable memory for the ext4 superblock").
The issue is found when playing video files stored in NTFS on the
Android TV platform. During this process, the media parser reads the
video file, causing ntfs3 to allocate buffer cache from the CMA area.
Subsequently, the hardware decoder attempts to allocate memory from the
same CMA area. However, the page is still in use by ntfs3, resulting in
a migrate failure in alloc_contig_range().
The pinned page and allocating stacktrace reported by page owner shows
below:
page:ffffffff00b68880 refcount:3 mapcount:0 mapping:ffffff80046aa828
index:0xc0040 pfn:0x20fa4
aops:def_blk_aops ino:0
flags: 0x2020(active|private)
page dumped because: migration failure
page last allocated via order 0, migratetype Movable,
gfp_mask 0x108c48
(GFP_NOFS|__GFP_NOFAIL|__GFP_HARDWALL|__GFP_MOVABLE),
page_owner tracks the page as allocated
prep_new_page
get_page_from_freelist
__alloc_pages_nodemask
pagecache_get_page
__getblk_gfp
__bread_gfp
ntfs_read_run_nb
ntfs_read_bh
mi_read
ntfs_iget5
dir_search_u
ntfs_lookup
__lookup_slow
lookup_slow
walk_component
path_lookupat
Signed-off-by: Ism Hong <ism.hong@gmail.com> Signed-off-by: Konstantin Komarov <almaz.alexandrovich@paragon-software.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Unfortunately reparse attribute is used for many purposes (several dozens).
It is not possible here to know is this name symlink or not.
To get exactly the type of name we should to open inode (read mft).
getattr for opened file (fstat) correctly returns symlink.
Signed-off-by: Konstantin Komarov <almaz.alexandrovich@paragon-software.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
In the initial boot stage the integrated keyboard of Fujitsu Lifebook U728
refuses to work and it's not possible to type for example a dm-crypt
passphrase without the help of an external keyboard.
i8042.nomux kernel parameter resolves this issue but using that a PS/2
mouse is detected. This input device is unused even when the i2c-hid-acpi
kernel module is blacklisted making the integrated ELAN touchpad
(04F3:3092) not working at all.
So this notebook uses a hid-over-i2c touchpad which is managed by the
i2c_designware input driver. Since you can't find a PS/2 mouse port on this
computer and you can't connect a PS/2 mouse to it even with an official
port replicator I think it's safe to not use the PS/2 mouse port at all.
In ext4_map_blocks(), if we can't find a range of mapping in the
extents cache, we are calling ext4_ext_map_blocks() to search the real
path and ext4_ext_determine_hole() to determine the hole range. But if
the querying range was partially or completely overlaped by a delalloc
extent, we can't find it in the real extent path, so the returned hole
length could be incorrect.
Fortunately, ext4_ext_put_gap_in_cache() have already handle delalloc
extent, but it searches start from the expanded hole_start, doesn't
start from the querying range, so the delalloc extent found could not be
the one that overlaped the querying range, plus, it also didn't adjust
the hole length. Let's just remove ext4_ext_put_gap_in_cache(), handle
delalloc and insert adjusted hole extent in ext4_ext_determine_hole().
When deleting an association the shutdown path is deadlocking because we
try to flush the nvmet_wq nested. Avoid this by deadlock by deferring
the put work into its own work item.
Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Daniel Wagner <dwagner@suse.de> Signed-off-by: Keith Busch <kbusch@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
When the target port has not active port binding, there is no point in
trying to process the command as it has to fail anyway. Instead adding
checks to all commands abort the command early.
Reviewed-by: Hannes Reinecke <hare@suse.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Daniel Wagner <dwagner@suse.de> Signed-off-by: Keith Busch <kbusch@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The hostport data structure is shared between the association, this why
we keep track of the users via a refcount. So we should not decrement
the refcount on a match and free the hostport several times.
Reported by KASAN.
Reviewed-by: Hannes Reinecke <hare@suse.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Daniel Wagner <dwagner@suse.de> Signed-off-by: Keith Busch <kbusch@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
When the target executes a disconnect and the host triggers a reconnect
immediately, the reconnect command still finds an existing association.
The reconnect crashes later on because nvmet_fc_delete_target_assoc
blindly removes resources while the reconnect code wants to use it.
To address this, nvmet_fc_find_target_assoc should not be able to
lookup an association which is being removed. The association list
is already under RCU lifetime management, so let's properly use it
and remove the association from the list and wait for a grace period
before cleaning up all. This means we also can drop the RCU management
on the queues, because this is now handled via the association itself.
A second step split the execution context so that the initial disconnect
command can complete without running the reconnect code in the same
context. As usual, this is done by deferring the ->done to a workqueue.
Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hannes Reinecke <hare@suse.de> Signed-off-by: Daniel Wagner <dwagner@suse.de> Signed-off-by: Keith Busch <kbusch@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
In case we return early out of __nvmet_fc_finish_ls_req() we still have
to release the reference on the target port.
Reviewed-by: Hannes Reinecke <hare@suse.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Daniel Wagner <dwagner@suse.de> Signed-off-by: Keith Busch <kbusch@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The first argument of list_add_tail function is the new element which
should be added to the list which is the second argument. Swap the
arguments to allow processing more than one element at a time.
Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hannes Reinecke <hare@suse.de> Signed-off-by: Daniel Wagner <dwagner@suse.de> Signed-off-by: Keith Busch <kbusch@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The module exit path has race between deleting all controllers and
freeing 'left over IDs'. To prevent double free a synchronization
between nvme_delete_ctrl and ida_destroy has been added by the initial
commit.
There is some logic around trying to prevent from hanging forever in
wait_for_completion, though it does not handling all cases. E.g.
blktests is able to reproduce the situation where the module unload
hangs forever.
If we completely rely on the cleanup code executed from the
nvme_delete_ctrl path, all IDs will be freed eventually. This makes
calling ida_destroy unnecessary. We only have to ensure that all
nvme_delete_ctrl code has been executed before we leave
nvme_fc_exit_module. This is done by flushing the nvme_delete_wq
workqueue.
While at it, remove the unused nvme_fc_wq workqueue too.
Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hannes Reinecke <hare@suse.de> Signed-off-by: Daniel Wagner <dwagner@suse.de> Signed-off-by: Keith Busch <kbusch@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
For devices with multiple clock sources connected to a selector, we need
to check what a clock selector control request has returned. This is
needed to ensure that a requested clock source is indeed selected and for
autoclock feature to work.
For devices with single clock source connected, if we get an error there
is nothing else we can do about it. We can't skip clock selector setup as
it is required by some devices. So lets just ignore error in this case.
This should fix various buggy Mackie devices:
[ 649.109785] usb 1-1.3: parse_audio_format_rates_v2v3(): unable to find clock source (clock -32)
[ 649.111946] usb 1-1.3: parse_audio_format_rates_v2v3(): unable to find clock source (clock -32)
[ 649.113822] usb 1-1.3: parse_audio_format_rates_v2v3(): unable to find clock source (clock -32)
There is also interesting info from the Windows documentation [1] (this
is probably why manufacturers dont't even test this feature):
"The USB Audio 2.0 driver doesn't support clock selection. The driver
uses the Clock Source Entity, which is selected by default and never
issues a Clock Selector Control SET CUR request."
The annotation says in sctp_new(): "If it is a shutdown ack OOTB packet, we
expect a return shutdown complete, otherwise an ABORT Sec 8.4 (5) and (8)".
However, it does not check SCTP_CID_SHUTDOWN_ACK before setting vtag[REPLY]
in the conntrack entry(ct).
Because of that, if the ct in Router disappears for some reason in [1]
with the packet sequence like below:
when processing HB ACK packet in Router it calls sctp_new() to initialize
the new ct with vtag[REPLY] set to HB_ACK packet's vtag.
Later when sending DATA from Client, all the SACKs from Server will get
dropped in Router, as the SACK packet's vtag does not match vtag[REPLY]
in the ct. The worst thing is the vtag in this ct will never get fixed
by the upcoming packets from Server.
This patch fixes it by checking SCTP_CID_SHUTDOWN_ACK before setting
vtag[REPLY] in the ct in sctp_new() as the annotation says. With this
fix, it will leave vtag[REPLY] in ct to 0 in the case above, and the
next HB REQ/ACK from Server is able to fix the vtag as its value is 0
in nf_conntrack_sctp_packet().
Signed-off-by: Xin Long <lucien.xin@gmail.com> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The Lenovo Legion Go is a handheld gaming system, similar to a Steam Deck.
It has a gamepad (including rear paddles), 3 gyroscopes, a trackpad,
volume buttons, a power button, and 2 LED ring lights.
The Legion Go firmware presents these controls as a USB hub with various
devices attached. In its default state, the gamepad is presented as an
Xbox controller connected to this hub. (By holding a combination of
buttons, it can be changed to use the older DirectInput API.)
This patch teaches the existing Xbox controller module `xpad` to bind to
the controller in the Legion Go, which enables support for the:
- directional pad,
- analog sticks (including clicks),
- X, Y, A, B,
- start and select (or menu and capture),
- shoulder buttons, and
- rumble.
The trackpad, touchscreen, volume controls, and power button are already
supported via existing kernel modules. Two of the face buttons, the
gyroscopes, rear paddles, and LEDs are not.
After this patch lands, the Legion Go will be mostly functional in Linux,
out-of-the-box. The various components of the USB hub can be synthesized
into a single logical controller (including the additional buttons) in
userspace with [Handheld Daemon](https://github.com/hhd-dev/hhd), which
makes the Go fully functional.
Many devices with a single alternate setting do not have a Valid
Alternate Setting Control and validation performed by
validate_sample_rate_table_v2v3() doesn't work on them and is not
really needed. So check the presense of control before sending
altsetting validation requests.
MOTU Microbook IIc is suffering the most without this check. It
takes up to 40 seconds to bootup due to how slow it switches
sampling rates:
[ 2659.164824] usb 3-2: New USB device found, idVendor=07fd, idProduct=0004, bcdDevice= 0.60
[ 2659.164827] usb 3-2: New USB device strings: Mfr=1, Product=2, SerialNumber=0
[ 2659.164829] usb 3-2: Product: MicroBook IIc
[ 2659.164830] usb 3-2: Manufacturer: MOTU
[ 2659.166204] usb 3-2: Found last interface = 3
[ 2679.322298] usb 3-2: No valid sample rate available for 1:1, assuming a firmware bug
[ 2679.322306] usb 3-2: 1:1: add audio endpoint 0x3
[ 2679.322321] usb 3-2: Creating new data endpoint #3
[ 2679.322552] usb 3-2: 1:1 Set sample rate 96000, clock 1
[ 2684.362250] usb 3-2: 2:1: cannot get freq (v2/v3): err -110
[ 2694.444700] usb 3-2: No valid sample rate available for 2:1, assuming a firmware bug
[ 2694.444707] usb 3-2: 2:1: add audio endpoint 0x84
[ 2694.444721] usb 3-2: Creating new data endpoint #84
[ 2699.482103] usb 3-2: 2:1 Set sample rate 96000, clock 1
Continuous regulators can be configured to operate only in a certain
duty cycle range (for example from 0..91%). Add a check to error out if
the duty cycle translates to an unsupported (or out of range) voltage.
devm_kasprintf() returns a pointer to dynamically allocated memory
which can be NULL upon failure. Ensure the allocation was successful
by checking the pointer validity.
Places the logic for checking if the group's block bitmap is corrupt under
the protection of the group lock to avoid allocating blocks from the group
with a corrupted block bitmap.
Determine if the group block bitmap is corrupted before using ac_b_ex in
ext4_mb_try_best_found() to avoid allocating blocks from a group with a
corrupted block bitmap in the following concurrency and making the
situation worse.
ext4_mb_regular_allocator
ext4_lock_group(sb, group)
ext4_mb_good_group
// check if the group bbitmap is corrupted
ext4_mb_complex_scan_group
// Scan group gets ac_b_ex but doesn't use it
ext4_unlock_group(sb, group)
ext4_mark_group_bitmap_corrupted(group)
// The block bitmap was corrupted during
// the group unlock gap.
ext4_mb_try_best_found
ext4_lock_group(ac->ac_sb, group)
ext4_mb_use_best_found
mb_mark_used
// Allocating blocks in block bitmap corrupted group
Add touch screen info for TECLAST X16 Plus tablet.
Signed-off-by: Phoenix Chen <asbeltogf@gmail.com> Link: https://lore.kernel.org/r/20240126095308.5042-1-asbeltogf@gmail.com Reviewed-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
"cpu_probe" is called both by BP and APs, but reserving exception vector
(like 0x0-0x1000) called by "cpu_probe" need once and calling on APs is
too late since memblock is unavailable at that time.
So, reserve exception vector ONLY by BP.
Suggested-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Signed-off-by: Huang Pei <huangpei@loongson.cn> Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
With one of the on-board ASM1061 AHCI controllers (1b21:0612) on an
ASUSTeK Pro WS WRX80E-SAGE SE WIFI mainboard, a controller hang was
observed that was immediately preceded by the following kernel
messages:
The first message is produced by code in drivers/iommu/dma-iommu.c
which is accompanied by the following comment that seems to apply:
/*
* Try to use all the 32-bit PCI addresses first. The original SAC vs.
* DAC reasoning loses relevance with PCIe, but enough hardware and
* firmware bugs are still lurking out there that it's safest not to
* venture into the 64-bit space until necessary.
*
* If your device goes wrong after seeing the notice then likely either
* its driver is not setting DMA masks accurately, the hardware has
* some inherent bug in handling >32-bit addresses, or not all the
* expected address bits are wired up between the device and the IOMMU.
*/
Asking the ASM1061 on a discrete PCIe card to DMA from I/O virtual
address 0xffffffff00000000 produces the following I/O page faults:
Note that the upper 21 bits of the logged DMA address are zero. (When
asking a different PCIe device in the same PCIe slot to DMA to the
same I/O virtual address, we do see all the upper 32 bits of the DMA
address as 1, so this is not an issue with the chipset or IOMMU
configuration on the test system.)
Also, hacking libahci to always set the upper 21 bits of all DMA
addresses to 1 produces no discernible effect on the behavior of the
ASM1061, and mkfs/mount/scrub/etc work as without this hack.
This all strongly suggests that the ASM1061 has a 43 bit DMA address
limit, and this commit therefore adds a quirk to deal with this limit.
This issue probably applies to (some of) the other supported ASMedia
parts as well, but we limit it to the PCI IDs known to refer to
ASM1061 parts, as that's the only part we know for sure to be affected
by this issue at this point.
Link: https://lore.kernel.org/linux-ide/ZaZ2PIpEId-rl6jv@wantstofly.org/ Signed-off-by: Lennert Buytenhek <kernel@wantstofly.org>
[cassel: drop date from error messages in commit log] Signed-off-by: Niklas Cassel <cassel@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The userspace program could pass any values to the driver through
ioctl() interface. If the driver doesn't check the value of pixclock,
it may cause divide-by-zero error.
In sisfb_check_var(), var->pixclock is used as a divisor to caculate
drate before it is checked against zero. Fix this by checking it
at the beginning.
This is similar to CVE-2022-3061 in i740fb which was fixed by
commit 15cf0b8.
The userspace program could pass any values to the driver through
ioctl() interface. If the driver doesn't check the value of pixclock,
it may cause divide-by-zero error.
Although pixclock is checked in savagefb_decode_var(), but it is not
checked properly in savagefb_probe(). Fix this by checking whether
pixclock is zero in the function savagefb_check_var() before
info->var.pixclock is used as the divisor.
This is similar to CVE-2022-3061 in i740fb which was fixed by
commit 15cf0b8.
fast-xmit must only be enabled after the sta has been uploaded to the driver,
otherwise it could end up passing the not-yet-uploaded sta via drv_tx calls
to the driver, leading to potential crashes because of uninitialized drv_priv
data.
Add a missing sta->uploaded check and re-check fast xmit after inserting a sta.
dump(cb=[0, 1]):
if_start=cb[1] (=1)
send rdev0.wdev1 -> ok
// since if_start=1 the rdev0.wdev0 got skipped
// through if_idx < if_start
send rdev1.wdev1 -> ok
The if_start needs to be reset back to 0 upon wdev
loop end.
The problem is actually hard to hit on a desktop,
and even on most routers. The prerequisites for
this manifesting was:
- more than 1 wiphy
- a few handful of interfaces
- dump without rdev or wdev filter
I was seeing this with 4 wiphys 9 interfaces each.
It'd miss 6 interfaces from the last wiphy
reported to userspace.
We seem to have hit warnings of 'output may be truncated' which is fixed
by increasing the size of 'irq_name'
drivers/dma/fsl-qdma.c: In function ‘fsl_qdma_irq_init’:
drivers/dma/fsl-qdma.c:824:46: error: ‘%d’ directive writing between 1 and 11 bytes into a region of size 10 [-Werror=format-overflow=]
824 | sprintf(irq_name, "qdma-queue%d", i);
| ^~
drivers/dma/fsl-qdma.c:824:35: note: directive argument in the range [-2147483641, 2147483646]
824 | sprintf(irq_name, "qdma-queue%d", i);
| ^~~~~~~~~~~~~~
drivers/dma/fsl-qdma.c:824:17: note: ‘sprintf’ output between 12 and 22 bytes into a destination of size 20
824 | sprintf(irq_name, "qdma-queue%d", i);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
We seem to have hit warnings of 'output may be truncated' which is fixed
by increasing the size of 'dev_id'
drivers/dma/sh/shdmac.c: In function ‘sh_dmae_probe’:
drivers/dma/sh/shdmac.c:541:34: error: ‘%d’ directive output may be truncated writing between 1 and 10 bytes into a region of size 9 [-Werror=format-truncation=]
541 | "sh-dmae%d.%d", pdev->id, id);
| ^~
In function ‘sh_dmae_chan_probe’,
inlined from ‘sh_dmae_probe’ at drivers/dma/sh/shdmac.c:845:9:
drivers/dma/sh/shdmac.c:541:26: note: directive argument in the range [0, 2147483647]
541 | "sh-dmae%d.%d", pdev->id, id);
| ^~~~~~~~~~~~~~
drivers/dma/sh/shdmac.c:541:26: note: directive argument in the range [0, 19]
drivers/dma/sh/shdmac.c:540:17: note: ‘snprintf’ output between 11 and 21 bytes into a destination of size 16
540 | snprintf(sh_chan->dev_id, sizeof(sh_chan->dev_id),
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
541 | "sh-dmae%d.%d", pdev->id, id);
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
An abort that is responded to by iSCSI itself is added to tmr_list but does
not go to target core. A LUN_RESET that goes through tmr_list takes a
refcounter on the abort and waits for completion. However, the abort will
be never complete because it was not started in target core.
Unable to locate ITT: 0x05000000 on CID: 0
Unable to locate RefTaskTag: 0x05000000 on CID: 0.
wait_for_tasks: Stopping tmf LUN_RESET with tag 0x0 ref_task_tag 0x0 i_state 34 t_state ISTATE_PROCESSING refcnt 2 transport_state active,stop,fabric_stop
wait for tasks: tmf LUN_RESET with tag 0x0 ref_task_tag 0x0 i_state 34 t_state ISTATE_PROCESSING refcnt 2 transport_state active,stop,fabric_stop
...
INFO: task kworker/0:2:49 blocked for more than 491 seconds.
task:kworker/0:2 state:D stack: 0 pid: 49 ppid: 2 flags:0x00000800
Workqueue: events target_tmr_work [target_core_mod]
Call Trace:
__switch_to+0x2c4/0x470
_schedule+0x314/0x1730
schedule+0x64/0x130
schedule_timeout+0x168/0x430
wait_for_completion+0x140/0x270
target_put_cmd_and_wait+0x64/0xb0 [target_core_mod]
core_tmr_lun_reset+0x30/0xa0 [target_core_mod]
target_tmr_work+0xc8/0x1b0 [target_core_mod]
process_one_work+0x2d4/0x5d0
worker_thread+0x78/0x6c0
To fix this, only add abort to tmr_list if it will be handled by target
core.
Signed-off-by: Dmitry Bogdanov <d.bogdanov@yadro.com> Link: https://lore.kernel.org/r/20240111125941.8688-1-d.bogdanov@yadro.com Reviewed-by: Mike Christie <michael.christie@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The validation of the value written to sched_rt_period_us was broken
because:
- the sysclt_sched_rt_period is declared as unsigned int
- parsed by proc_do_intvec()
- the range is asserted after the value parsed by proc_do_intvec()
Because of this negative values written to the file were written into a
unsigned integer that were later on interpreted as large positive
integers which did passed the check:
if (sysclt_sched_rt_period <= 0)
return EINVAL;
This commit fixes the parsing by setting explicit range for both
perid_us and runtime_us into the sched_rt_sysctls table and processes
the values with proc_dointvec_minmax() instead.
Alternatively if we wanted to use full range of unsigned int for the
period value we would have to split the proc_handler and use
proc_douintvec() for it however even the
Documentation/scheduller/sched-rt-group.rst describes the range as 1 to
INT_MAX.
As far as I can tell the only problem this causes is that the sysctl
file allows writing negative values which when read back may confuse
userspace.
There is also a LTP test being submitted for these sysctl files at:
What this test does is to compare the return value from the
sched_rr_get_interval() and the sched_rr_timeslice_ms sysctl file and
fails if they do not match.
The problem it found is the intial sysctl file value which was computed as:
static int sysctl_sched_rr_timeslice = (MSEC_PER_SEC / HZ) * RR_TIMESLICE;
which works fine as long as MSEC_PER_SEC is multiple of HZ, however it
introduces 10% rounding error for CONFIG_HZ_300:
(MSEC_PER_SEC / HZ) * (100 * HZ / 1000)
(1000 / 300) * (100 * 300 / 1000)
3 * 30 = 90
This can be easily fixed by reversing the order of the multiplication
and division. After this fix we get:
(MSEC_PER_SEC * (100 * HZ / 1000)) / HZ
(1000 * (100 * 300 / 1000)) / 300
(1000 * 30) / 300 = 100
Fixes: 975e155ed873 ("sched/rt: Show the 'sched_rr_timeslice' SCHED_RR timeslice tuning knob in milliseconds") Signed-off-by: Cyril Hrubis <chrubis@suse.cz> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Petr Vorel <pvorel@suse.cz> Acked-by: Mel Gorman <mgorman@suse.de> Tested-by: Petr Vorel <pvorel@suse.cz> Link: https://lore.kernel.org/r/20230802151906.25258-2-chrubis@suse.cz
[ pvorel: rebased for 5.15, 5.10 ] Signed-off-by: Petr Vorel <pvorel@suse.cz> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Before this change, the expected size of the user space buffer was
taken from fx_sw->xstate_size. fx_sw->xstate_size can be changed
from user-space, so it is possible construct a sigreturn frame where:
* fx_sw->xstate_size is smaller than the size required by valid bits in
fx_sw->xfeatures.
* user-space unmaps parts of the sigrame fpu buffer so that not all of
the buffer required by xrstor is accessible.
In this case, xrstor tries to restore and accesses the unmapped area
which results in a fault. But fault_in_readable succeeds because buf +
fx_sw->xstate_size is within the still mapped area, so it goes back and
tries xrstor again. It will spin in this loop forever.
Instead, fault in the maximum size which can be touched by XRSTOR (taken
from fpstate->user_size).
[ dhansen: tweak subject / changelog ]
Fixes: fcb3635f5018 ("x86/fpu/signal: Handle #PF in the direct restore path") Reported-by: Konstantin Bogomolov <bogomolov@google.com> Suggested-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Andrei Vagin <avagin@google.com> Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Cc:stable@vger.kernel.org Link: https://lore.kernel.org/all/20240130063603.3392627-1-avagin%40google.com Link: https://lore.kernel.org/all/20240130063603.3392627-1-avagin%40google.com Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Write error handling is racy and can sometime lead to the error recovery
path wrongly changing the inode size of a sequential zone file to an
incorrect value which results in garbage data being readable at the end
of a file. There are 2 problems:
1) zonefs_file_dio_write() updates a zone file write pointer offset
after issuing a direct IO with iomap_dio_rw(). This update is done
only if the IO succeed for synchronous direct writes. However, for
asynchronous direct writes, the update is done without waiting for
the IO completion so that the next asynchronous IO can be
immediately issued. However, if an asynchronous IO completes with a
failure right before the i_truncate_mutex lock protecting the update,
the update may change the value of the inode write pointer offset
that was corrected by the error path (zonefs_io_error() function).
2) zonefs_io_error() is called when a read or write error occurs. This
function executes a report zone operation using the callback function
zonefs_io_error_cb(), which does all the error recovery handling
based on the current zone condition, write pointer position and
according to the mount options being used. However, depending on the
zoned device being used, a report zone callback may be executed in a
context that is different from the context of __zonefs_io_error(). As
a result, zonefs_io_error_cb() may be executed without the inode
truncate mutex lock held, which can lead to invalid error processing.
Fix both problems as follows:
- Problem 1: Perform the inode write pointer offset update before a
direct write is issued with iomap_dio_rw(). This is safe to do as
partial direct writes are not supported (IOMAP_DIO_PARTIAL is not
set) and any failed IO will trigger the execution of zonefs_io_error()
which will correct the inode write pointer offset to reflect the
current state of the one on the device.
- Problem 2: Change zonefs_io_error_cb() into zonefs_handle_io_error()
and call this function directly from __zonefs_io_error() after
obtaining the zone information using blkdev_report_zones() with a
simple callback function that copies to a local stack variable the
struct blk_zone obtained from the device. This ensures that error
handling is performed holding the inode truncate mutex.
This change also simplifies error handling for conventional zone files
by bypassing the execution of report zones entirely. This is safe to
do because the condition of conventional zones cannot be read-only or
offline and conventional zone files are always fully mapped with a
constant file size.
Reported-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com> Fixes: 8dcc1a9d90c1 ("fs: New zonefs file system") Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <dlemoal@kernel.org> Tested-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com> Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In mfill_atomic_hugetlb(), mmap_changing isn't being checked
again if we drop mmap_lock and reacquire it. When the lock is not held,
mmap_changing could have been incremented. This is also inconsistent
with the behavior in mfill_atomic().
Link: https://lkml.kernel.org/r/20240117223729.1444522-1-lokeshgidra@google.com Fixes: df2cc96e77011 ("userfaultfd: prevent non-cooperative events vs mcopy_atomic races") Signed-off-by: Lokesh Gidra <lokeshgidra@google.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Mike Rapoport <rppt@kernel.org> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Brian Geffon <bgeffon@google.com> Cc: David Hildenbrand <david@redhat.com> Cc: Jann Horn <jannh@google.com> Cc: Kalesh Singh <kaleshsingh@google.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Nicolas Geoffray <ngeoffray@google.com> Cc: Peter Xu <peterx@redhat.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Mike Rapoport (IBM) <rppt@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Both bpf_trace_printk and bpf_trace_vprintk helpers use static buffer guarded
with trace_printk_lock spin lock.
The spin lock contention causes issues with bpf programs attached to
contention_begin tracepoint [1][2].
Andrii suggested we could get rid of the contention by using trylock, but we
could actually get rid of the spinlock completely by using percpu buffers the
same way as for bin_args in bpf_bprintf_prepare function.
Adding new return 'buf' argument to struct bpf_bprintf_data and making
bpf_bprintf_prepare to return also the buffer for printk helpers.
Reported-by: Hao Sun <sunhao.th@gmail.com> Suggested-by: Andrii Nakryiko <andrii@kernel.org> Signed-off-by: Jiri Olsa <jolsa@kernel.org> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Yonghong Song <yhs@fb.com> Link: https://lore.kernel.org/bpf/20221215214430.1336195-4-jolsa@kernel.org
[cascardo: there is no bpf_trace_vprintk in 5.15] Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@igalia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently we always cleanup/decrement bpf_bprintf_nest_level variable
in bpf_bprintf_cleanup if it's > 0.
There's possible scenario where this could cause a problem, when
bpf_bprintf_prepare does not get bin_args buffer (because num_args is 0)
and following bpf_bprintf_cleanup call decrements bpf_bprintf_nest_level
variable, like:
in task context:
bpf_bprintf_prepare(num_args != 0) increments 'bpf_bprintf_nest_level = 1'
-> first irq :
bpf_bprintf_prepare(num_args == 0)
bpf_bprintf_cleanup decrements 'bpf_bprintf_nest_level = 0'
-> second irq:
bpf_bprintf_prepare(num_args != 0) bpf_bprintf_nest_level = 1
gets same buffer as task context above
Adding check to bpf_bprintf_cleanup and doing the real cleanup only if we
got bin_args data in the first place.
Signed-off-by: Jiri Olsa <jolsa@kernel.org> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Yonghong Song <yhs@fb.com> Link: https://lore.kernel.org/bpf/20221215214430.1336195-3-jolsa@kernel.org
[cascardo: there is no bpf_trace_vprintk in 5.15] Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@igalia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Adding struct bpf_bprintf_data to hold bin_args argument for
bpf_bprintf_prepare function.
We will add another return argument to bpf_bprintf_prepare and
pass the struct to bpf_bprintf_cleanup for proper cleanup in
following changes.
Signed-off-by: Jiri Olsa <jolsa@kernel.org> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Yonghong Song <yhs@fb.com> Link: https://lore.kernel.org/bpf/20221215214430.1336195-2-jolsa@kernel.org
[cascardo: there is no bpf_trace_vprintk in 5.15] Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@igalia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
MAX_SNPRINTF_VARARGS and MAX_SEQ_PRINTF_VARARGS are used by bpf helpers
bpf_snprintf and bpf_seq_printf to limit their varargs. Both call into
bpf_bprintf_prepare for print formatting logic and have convenience
macros in libbpf (BPF_SNPRINTF, BPF_SEQ_PRINTF) which use the same
helper macros to convert varargs to a byte array.
Changing shared functionality to support more varargs for either bpf
helper would affect the other as well, so let's combine the _VARARGS
macros to make this more obvious.
Signed-off-by: Dave Marchevsky <davemarchevsky@fb.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org> Acked-by: Andrii Nakryiko <andrii@kernel.org> Link: https://lore.kernel.org/bpf/20210917182911.2426606-2-davemarchevsky@fb.com Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@igalia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The sched_rr_timeslice can be reset to default by writing value that is
<= 0. However after reading from this file we always got the last value
written, which is not useful at all.
The data offset for the SMB3.1.1 POSIX create context will always be
8-byte aligned so having the check 'noff + nlen >= doff' in
smb2_parse_contexts() is wrong as it will lead to -EINVAL because noff
+ nlen == doff.
Fix the sanity check to correctly handle aligned create context data.
Fixes: af1689a9b770 ("smb: client: fix potential OOBs in smb2_parse_contexts()") Signed-off-by: Paulo Alcantara <pc@manguebit.com> Signed-off-by: Steve French <stfrench@microsoft.com>
[Guru:smb2_parse_contexts() is present in file smb2ops.c,
smb2ops.c file location is changed, modified patch accordingly.] Signed-off-by: Guruswamy Basavaiah <guruswamy.basavaiah@broadcom.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reported-by: Robert Morris <rtm@csail.mit.edu> Cc: stable@vger.kernel.org Signed-off-by: Paulo Alcantara (SUSE) <pc@manguebit.com> Signed-off-by: Steve French <stfrench@microsoft.com>
[Guru: Removed changes to cached_dir.c and checking return value
of smb2_parse_contexts in smb2ops.c] Signed-off-by: Guruswamy Basavaiah <guruswamy.basavaiah@broadcom.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The dsmark qdisc has served us well over the years for diffserv but has not
been getting much attention due to other more popular approaches to do diffserv
services. Most recently it has become a shooting target for syzkaller. For this
reason, we are retiring it.
Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com> Acked-by: Jiri Pirko <jiri@nvidia.com> Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The ATM qdisc has served us well over the years but has not been getting much
TLC due to lack of known users. Most recently it has become a shooting target
for syzkaller. For this reason, we are retiring it.
Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com> Acked-by: Jiri Pirko <jiri@nvidia.com> Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
While this amazing qdisc has served us well over the years it has not been
getting any tender love and care and has bitrotted over time.
It has become mostly a shooting target for syzkaller lately.
For this reason, we are retiring it. Goodbye CBQ - we loved you.
Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com> Acked-by: Jiri Pirko <jiri@nvidia.com> Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If we delay sending End Transfer for Setup TRB to be prepared, we need
to check if the End Transfer was in preparation for a driver
teardown/soft-disconnect. In those cases, just send the End Transfer
command without delay.
In the case of soft-disconnect, there's a very small chance the command
may not go through immediately. But should it happen, the Setup TRB will
be prepared during the polling of the controller halted state, allowing
the command to go through then.
In the case of disabling endpoint due to reconfiguration (e.g.
set_interface(alt-setting) or usb reset), then it's driven by the host.
Typically the host wouldn't immediately cancel the control request and
send another control transfer to trigger the End Transfer command
timeout.
The rkisp1 does share interrupt lines on some platforms, after all. Thus
we need to revert this, and implement a fix for the rkisp1 shared irq
handling in a follow-up patch.
Do not call gadget stop until the poll for controller halt is
completed. DEVTEN is cleared as part of gadget stop, so the intention to
allow ep0 events to continue while waiting for controller halt is not
happening.
The gadget driver may wait on the request completion when it sets the
USB_GADGET_DELAYED_STATUS. Make sure that the End Transfer command can
go through if the dwc->delayed_status is set so that the request can
complete. When the delayed_status is set, the Setup packet is already
processed, and the next phase should be either Data or Status. It's
unlikely that the host would cancel the control transfer and send a new
Setup packet during End Transfer command. But if that's the case, we can
try again when ep0state returns to EP0_SETUP_PHASE.
The fbtft_framebuffer_alloc() calls fb_deferred_io_init() before
initializing info->fix.smem_len. It is set to zero by the
framebuffer_alloc() function. It will trigger a WARN_ON() at the
start of fb_deferred_io_init() and the function will not do anything.
The kernel is globally removing the ambiguous 0-length and 1-element
arrays in favor of flexible arrays, so that we can gain both compile-time
and run-time array bounds checking[1].
Replace the trailing 1-element array with a flexible array in the
following structures:
Replace the trailing 1-element array with a flexible array, but leave
the existing structure padding:
struct smb2_file_all_info
struct smb2_lock_req
Adjust all related size calculations to match the changes to sizeof().
No machine code output or .data section differences are produced after
these changes.
[1] For lots of details, see both:
https://docs.kernel.org/process/deprecated.html#zero-length-and-one-element-arrays
https://people.kernel.org/kees/bounded-flexible-arrays-in-c
Cc: Steve French <sfrench@samba.org> Cc: Paulo Alcantara <pc@cjr.nz> Cc: Ronnie Sahlberg <lsahlber@redhat.com> Cc: Shyam Prasad N <sprasad@microsoft.com> Cc: Tom Talpey <tom@talpey.com> Cc: Namjae Jeon <linkinjeon@kernel.org> Cc: Sergey Senozhatsky <senozhatsky@chromium.org> Cc: linux-cifs@vger.kernel.org Cc: samba-technical@lists.samba.org Reviewed-by: Namjae Jeon <linkinjeon@kernel.org> Signed-off-by: Kees Cook <keescook@chromium.org> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Vasiliy Kovalev <kovalev@altlinux.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
libbpf support for "tc" progs doesn't exist for the linux-5.15.y tree.
This commit was backported too far back in upstream, to a kernel where
the libbpf support was not there for the test.
Drivers in ->remove() callback should undo the runtime PM changes done
in ->probe(). Usually this means calling pm_runtime_disable(),
pm_runtime_dont_use_autosuspend() etc.
>From grepping code, it's clear that many people aren't aware of the
need to call pm_runtime_dont_use_autosuspend().
When brainstorming solutions, one idea that came up was to leverage
the new-ish devm_pm_runtime_enable() function. The idea here is that:
* When the devm action is called we know that the driver is being
removed. It's the perfect time to undo the use_autosuspend.
* The code of pm_runtime_dont_use_autosuspend() already handles the
case of being called when autosuspend wasn't enabled.
Suggested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Signed-off-by: Douglas Anderson <dianders@chromium.org> Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Stable-dep-of: 3d07a411b4fa ("drm/msm/dsi: Use pm_runtime_resume_and_get to prevent refcnt leaks") Signed-off-by: Amit Pundir <amit.pundir@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add the MIDR value of Microsoft Azure Cobalt 100, which is a Microsoft
implemented CPU based on r0p0 of the ARM Neoverse N2 CPU, and therefore
suffers from all the same errata.
CC: stable@vger.kernel.org # 5.15+ Signed-off-by: Easwar Hariharan <eahariha@linux.microsoft.com> Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com> Acked-by: Mark Rutland <mark.rutland@arm.com> Acked-by: Marc Zyngier <maz@kernel.org> Reviewed-by: Oliver Upton <oliver.upton@linux.dev> Link: https://lore.kernel.org/r/20240214175522.2457857-1-eahariha@linux.microsoft.com Signed-off-by: Will Deacon <will@kernel.org> Signed-off-by: Easwar Hariharan <eahariha@linux.microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If DAT metadata file block access fails due to corruption of the DAT file
or abnormal virtual block numbers held by b-trees or inodes, a kernel
warning is generated.
This replaces the WARN_ONs by error output, so that a kernel, booted with
panic_on_warn, does not panic. This patch also replaces the detected
return code -ENOENT with another internal code -EINVAL to notify the bmap
layer of metadata corruption. When the bmap layer sees -EINVAL, it
handles the abnormal situation with nilfs_bmap_convert_error() and finally
returns code -EIO as it should.
According to a syzbot report, end_buffer_async_write(), which handles the
completion of block device writes, may detect abnormal condition of the
buffer async_write flag and cause a BUG_ON failure when using nilfs2.
Nilfs2 itself does not use end_buffer_async_write(). But, the async_write
flag is now used as a marker by commit 7f42ec394156 ("nilfs2: fix issue
with race condition of competition between segments for dirty blocks") as
a means of resolving double list insertion of dirty blocks in
nilfs_lookup_dirty_data_buffers() and nilfs_lookup_node_buffers() and the
resulting crash.
This modification is safe as long as it is used for file data and b-tree
node blocks where the page caches are independent. However, it was
irrelevant and redundant to also introduce async_write for segment summary
and super root blocks that share buffers with the backing device. This
led to the possibility that the BUG_ON check in end_buffer_async_write
would fail as described above, if independent writebacks of the backing
device occurred in parallel.
The use of async_write for segment summary buffers has already been
removed in a previous change.
Fix this issue by removing the manipulation of the async_write flag for
the remaining super root block buffer.
Link: https://lkml.kernel.org/r/20240203161645.4992-1-konishi.ryusuke@gmail.com Fixes: 7f42ec394156 ("nilfs2: fix issue with race condition of competition between segments for dirty blocks") Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com> Reported-by: syzbot+5c04210f7c7f897c1e7f@syzkaller.appspotmail.com Closes: https://lkml.kernel.org/r/00000000000019a97c05fd42f8c8@google.com Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
On some systems, sys_membarrier can be very expensive, causing overall
slowdowns for everything. So put a lock on the path in order to
serialize the accesses to prevent the ability for this to be called at
too high of a frequency and saturate the machine.
Reviewed-and-tested-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com> Acked-by: Borislav Petkov <bp@alien8.de> Fixes: 22e4ebb97582 ("membarrier: Provide expedited private command") Fixes: c5f58bd58f43 ("membarrier: Provide GLOBAL_EXPEDITED command") Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[ converted to explicit mutex_*() calls - cleanup.h is not in this stable
branch - gregkh ] Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The patch fdb8e12cc2cc ("netfilter: ipset: fix performance regression
in swap operation") missed to add the calls to gc cancellations
at the error path of create operations and at module unload. Also,
because the half of the destroy operations now executed by a
function registered by call_rcu(), neither NFNL_SUBSYS_IPSET mutex
or rcu read lock is held and therefore the checking of them results
false warnings.
While in theory the timer can be triggered before expires + delta, for the
cases of RT tasks they really have no business giving any lenience for
extra slack time, so override any passed value by the user and always use
zero for schedule_hrtimeout_range() calls. Furthermore, this is similar to
what the nanosleep(2) family already does with current->timer_slack_ns.
Signed-off-by: Davidlohr Bueso <dave@stgolabs.net> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Link: https://lore.kernel.org/r/20230123173206.6764-3-dave@stgolabs.net Signed-off-by: Felix Moessbauer <felix.moessbauer@siemens.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The patch "netfilter: ipset: fix race condition between swap/destroy
and kernel side add/del/test", commit 28628fa9 fixes a race condition.
But the synchronize_rcu() added to the swap function unnecessarily slows
it down: it can safely be moved to destroy and use call_rcu() instead.
Eric Dumazet pointed out that simply calling the destroy functions as
rcu callback does not work: sets with timeout use garbage collectors
which need cancelling at destroy which can wait. Therefore the destroy
functions are split into two: cancelling garbage collectors safely at
executing the command received by netlink and moving the remaining
part only into the rcu callback.
Link: https://lore.kernel.org/lkml/C0829B10-EAA6-4809-874E-E1E9C05A8D84@automattic.com/ Fixes: 28628fa952fe ("netfilter: ipset: fix race condition between swap/destroy and kernel side add/del/test") Reported-by: Ale Crismani <ale.crismani@automattic.com> Reported-by: David Wang <00107082@163.com> Tested-by: David Wang <00107082@163.com> Signed-off-by: Jozsef Kadlecsik <kadlec@netfilter.org> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
GNU's addr2line can have problems parsing a vmlinux built with LLVM,
particularly when LTO was used. In order to decode the traces correctly
this patch adds the ability to switch to LLVM's utilities readelf and
addr2line. The same approach is followed by Will in [1].
Note that one could set CROSS_COMPILE=llvm- instead to hack around this
issue. However, doing so can break the decodecode routine as it will
force the selection of other LLVM utilities down the line e.g. llvm-as.
When framebuffer gets closed, the queued deferred IO gets cancelled. This
can cause some last display data to vanish. This is problematic for users
who send a still image to the framebuffer, then close the file: the image
may never appear.
To ensure none of display data get lost, flush the queued deferred IO
first before closing.
Another possible solution is to delete the cancel_delayed_work_sync()
instead. The difference is that the display may appear some time after
closing. However, the clearing of page mapping after this needs to be
removed too, because the page mapping is used by the deferred work. It is
not completely obvious whether it is okay to not clear the page mapping.
For a patch intended for stable trees, go with the simple and obvious
solution.
Fixes: 60b59beafba8 ("fbdev: mm: Deferred IO support") Cc: stable@vger.kernel.org Signed-off-by: Nam Cao <namcao@linutronix.de> Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Signed-off-by: Helge Deller <deller@gmx.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
The recent fix for the deferred I/O by the commit 3efc61d95259 ("fbdev: Fix invalid page access after closing deferred I/O devices")
caused a regression when the same fb device is opened/closed while
it's being used. It resulted in a frozen screen even if something
is redrawn there after the close. The breakage is because the patch
was made under a wrong assumption of a single open; in the current
code, fb_deferred_io_release() cleans up the page mapping of the
pageref list and it calls cancel_delayed_work_sync() unconditionally,
where both are no correct behavior for multiple opens.
This patch adds a refcount for the opens of the device, and applies
the cleanup only when all files get closed.
As both fb_deferred_io_open() and _close() are called always in the
fb_info lock (mutex), it's safe to use the normal int for the
refcounting.
When a fbdev with deferred I/O is once opened and closed, the dirty
pages still remain queued in the pageref list, and eventually later
those may be processed in the delayed work. This may lead to a
corruption of pages, hitting an Oops.
This patch makes sure to cancel the delayed work and clean up the
pageref list at closing the device for addressing the bug. A part of
the cleanup code is factored out as a new helper function that is
called from the common fb_release().
Rename various instances of pagelist to pagereflist. The list now
stores pageref structures, so the new name is more appropriate.
In their write-back helpers, several fbdev drivers refer to the
pageref list in struct fb_deferred_io instead of using the one
supplied as argument to the function. Convert them over to the
supplied one. It's the same instance, so no change of behavior
occurs.
Store the per-page state for fbdev's deferred I/O in struct
fb_deferred_io_pageref. Maintain a list of pagerefs for the pages
that have to be written back to video memory. Update all affected
drivers.
As with pages before, fbdev acquires a pageref when an mmaped page
of the framebuffer is being written to. It holds the pageref in a
list of all currently written pagerefs until it flushes the written
pages to video memory. Writeback occurs periodically. After writeback
fbdev releases all pagerefs and builds up a new dirty list until the
next writeback occurs.
Using pagerefs has a number of benefits.
For pages of the framebuffer, the deferred I/O code used struct
page.lru as an entry into the list of dirty pages. The lru field is
owned by the page cache, which makes deferred I/O incompatible with
some memory pages (e.g., most notably DRM's GEM SHMEM allocator).
struct fb_deferred_io_pageref now provides an entry into a list of
dirty framebuffer pages, freeing lru for use with the page cache.
Drivers also assumed that struct page.index is the page offset into
the framebuffer. This is not true for DRM buffers, which are located
at various offset within a mapped area. struct fb_deferred_io_pageref
explicitly stores an offset into the framebuffer. struct page.index
is now only the page offset into the mapped area.
These changes will allow DRM to use fbdev deferred I/O without an
intermediate shadow buffer.
v3:
* use pageref->offset for sorting
* fix grammar in comment
v2:
* minor fixes in commit message
Easily hit the below list corruption:
==
list_add corruption. prev->next should be next (ffffffffc0ceb090), but
was ffffec604507edc8. (prev=ffffec604507edc8).
WARNING: CPU: 65 PID: 3959 at lib/list_debug.c:26
__list_add_valid+0x53/0x80
CPU: 65 PID: 3959 Comm: fbdev Tainted: G U
RIP: 0010:__list_add_valid+0x53/0x80
Call Trace:
<TASK>
fb_deferred_io_mkwrite+0xea/0x150
do_page_mkwrite+0x57/0xc0
do_wp_page+0x278/0x2f0
__handle_mm_fault+0xdc2/0x1590
handle_mm_fault+0xdd/0x2c0
do_user_addr_fault+0x1d3/0x650
exc_page_fault+0x77/0x180
? asm_exc_page_fault+0x8/0x30
asm_exc_page_fault+0x1e/0x30
RIP: 0033:0x7fd98fc8fad1
==
Figure out the race happens when one process is adding &page->lru into
the pagelist tail in fb_deferred_io_mkwrite(), another process is
re-initializing the same &page->lru in fb_deferred_io_fault(), which is
not protected by the lock.
This fix is to init all the page lists one time during initialization,
it not only fixes the list corruption, but also avoids INIT_LIST_HEAD()
redundantly.
V2: change "int i" to "unsigned int i" (Geert Uytterhoeven)
Signed-off-by: Chuansheng Liu <chuansheng.liu@intel.com> Fixes: 105a940416fc ("fbdev/defio: Early-out if page is already enlisted") Cc: Thomas Zimmermann <tzimmermann@suse.de> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Reviewed-by: Javier Martinez Canillas <javierm@redhat.com> Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de> Link: https://patchwork.freedesktop.org/patch/msgid/20220318005003.51810-1-chuansheng.liu@intel.com
Stable-dep-of: 33cd6ea9c067 ("fbdev: flush deferred IO before closing") Signed-off-by: Sasha Levin <sashal@kernel.org>
Fbdev's deferred I/O sorts all dirty pages by default, which incurs a
significant overhead. Make the sorting step optional and update the few
drivers that require it. Use a FIFO list by default.
Most fbdev drivers with deferred I/O build a bounding rectangle around
the dirty pages or simply flush the whole screen. The only two affected
DRM drivers, generic fbdev and vmwgfx, both use a bounding rectangle.
In those cases, the exact order of the pages doesn't matter. The other
drivers look at the page index or handle pages one-by-one. The patch
sets the sort_pagelist flag for those, even though some of them would
probably work correctly without sorting. Driver maintainers should update
their driver accordingly.
Sorting pages by memory offset for deferred I/O performs an implicit
bubble-sort step on the list of dirty pages. The algorithm goes through
the list of dirty pages and inserts each new page according to its
index field. Even worse, list traversal always starts at the first
entry. As video memory is most likely updated scanline by scanline, the
algorithm traverses through the complete list for each updated page.
For example, with 1024x768x32bpp each page covers exactly one scanline.
Writing a single screen update from top to bottom requires updating
768 pages. With an average list length of 384 entries, a screen update
creates (768 * 384 =) 294912 compare operation.
Fix this by making the sorting step opt-in and update the few drivers
that require it. All other drivers work with unsorted page lists. Pages
are appended to the list. Therefore, in the common case of writing the
framebuffer top to bottom, pages are still sorted by offset, which may
have a positive effect on performance.
Playing a video [1] in mplayer's benchmark mode shows the difference
(i7-4790, FullHD, simpledrm, kernel with debugging).
VC shows the overhead of video decoding, VO shows the overhead of the
video output. Using unsorted page lists reduces the benchmark's run time
by ~32s/~25%.
v2:
* Make sorted pagelists the special case (Sam)
* Comment on drivers' use of pagelist (Sam)
* Warn about the overhead in comment
Return early if a page is already in the list of dirty pages for
deferred I/O. This can be detected if the page's list head is not
empty. Keep the list head initialized while the page is not enlisted
to make this work reliably.
The UART supports an auto-RTS mode in which the RTS pin is automatically
activated during transmission. So mark this mode as being supported even
if RTS is not controlled by the driver but the UART.
Also the serial core expects now at least one of both modes rts-on-send or
rts-after-send to be supported. This is since during sanitization
unsupported flags are deleted from a RS485 configuration set by userspace.
However if the configuration ends up with both flags unset, the core prints
a warning since it considers such a configuration invalid (see
uart_sanitize_serial_rs485()).
There is a scenario where DWC3 runtime suspend is blocked due to the
dwc->connected flag still being true while PM usage_count is zero after
DWC3 giveback is completed and the USB gadget session is being terminated.
This leads to a case where nothing schedules a PM runtime idle for the
device.
The exact condition is seen with the following sequence:
1. USB bus reset is issued by the host
2. Shortly after, or concurrently, a USB PD DR SWAP request is received
(sink->source)
3. USB bus reset event handler runs and issues
dwc3_stop_active_transfers(), and pending transfer are stopped
4. DWC3 usage_count decremented to 0, and runtime idle occurs while
dwc->connected == true, returns -EBUSY
5. DWC3 disconnect event seen, dwc->connected set to false due to DR
swap handling
6. No runtime idle after this point
Address this by issuing an asynchronous PM runtime idle call after the
disconnect event is completed, as it modifies the dwc->connected flag,
which is what blocks the initial runtime idle.
Current EP0 dequeue path will share the same as other EPs. However, there
are some special considerations that need to be made for EP0 transfers:
- EP0 transfers never transition into the started_list
- EP0 only has one active request at a time
In case there is a vendor specific control message for a function over USB
FFS, then there is no guarantee on the timeline which the DATA/STATUS stage
is responded to. While this occurs, any attempt to end transfers on
non-control EPs will end up having the DWC3_EP_DELAY_STOP flag set, and
defer issuing of the end transfer command. If the USB FFS application
decides to timeout the control transfer, or if USB FFS AIO path exits, the
USB FFS driver will issue a call to usb_ep_dequeue() for the ep0 request.
In case of the AIO exit path, the AIO FS blocks until all pending USB
requests utilizing the AIO path is completed. However, since the dequeue
of ep0 req does not happen properly, all non-control EPs with the
DWC3_EP_DELAY_STOP flag set will not be handled, and the AIO exit path will
be stuck waiting for the USB FFS data endpoints to receive a completion
callback.
Fix is to utilize dwc3_ep0_reset_state() in the dequeue API to ensure EP0
is brought back to the SETUP state, and ensures that any deferred end
transfer commands are handled. This also will end any active transfers
on EP0, compared to the previous implementation which directly called
giveback only.
Several sequences utilize the same routine for forcing the control endpoint
back into the SETUP phase. This is required, because those operations need
to ensure that EP0 is back in the default state.
It was observed that there are hosts that may complete pending SETUP
transactions before the stop active transfers and controller halt occurs,
leading to lingering endxfer commands on DEPs on subsequent pullup/gadget
start iterations.
The sequence shows that the USB gadget disconnect (dwc3_gadget_pullup(0))
routine completed successfully, allowing for the USB gadget to proceed with
a USB gadget connect. However, if this occurs the system runs into an
issue where:
BUG: spinlock already unlocked on CPU
spin_bug+0x0
dwc3_remove_requests+0x278
dwc3_ep0_out_start+0xb0
__dwc3_gadget_start+0x25c
This is due to the pending endxfers, leading to gadget start (w/o lock
held) to execute the remove requests, which will unlock the dwc3
spinlock as part of giveback.
To mitigate this, resolve the pending endxfers on the pullup disable
path by re-locating the SETUP phase check after stop active transfers, since
that is where the DWC3_EP_DELAY_STOP is potentially set. This also allows
for handling of a host that may be unresponsive by using the completion
timeout to trigger the stall and restart for EP0.
During a cable disconnect sequence, if ep0state is not in the SETUP phase,
then nothing will trigger any pending end transfer commands. Force
stopping of any pending SETUP transaction, and move back to the SETUP
phase.
If any function drivers request for a delayed status phase, this leads to a
SETUP transfer timeout error, since the function may take longer to process
the DATA stage. This eventually results in end transfer timeouts, as there
is a pending SETUP transaction.
In addition, allow the DWC3_EP_DELAY_STOP to be set for if there is a
delayed status requested. Ocasionally, a host may abort the current SETUP
transaction, by issuing a subsequent SETUP token. In those situations, it
would result in an endxfer timeout as well.
According to the databook ep0 should be in setup phase during reset.
If host issues reset between control transfers, ep0 will be in an
invalid state. Fix this by issuing stall and restart on ep0 if it
is not in setup phase.
Also SW needs to complete pending control transfer and setup core for
next setup stage as per data book. Hence check ep0 state during reset
interrupt handling and make sure active transfers on ep0 out/in
endpoint are stopped by queuing ENDXFER command for that endpoint and
restart ep0 out again to receive next setup packet.
projects / thirdparty / kernel / stable.git / log
