All batches of the Pine64 Pinebook Pro, except the latest batch (as of 2024)
whose hardware design was revised due to the component shortage, use a 1S
lithium battery whose nominal/design capacity is 10,000 mAh, according to the
battery datasheet. [1][2] Let's correct the design full-charge value in the
Pinebook Pro board dts, to improve the accuracy of the hardware description,
and to hopefully improve the accuracy of the fuel gauge a bit on all units
that don't belong to the latest batch.
The above-mentioned latest batch uses a different 1S lithium battery with
a slightly lower capacity, more precisely 9,600 mAh. To make the fuel gauge
work reliably on the latest batch, a sample battery would need to be sent to
CellWise, to obtain its proprietary battery profile, whose data goes into
"cellwise,battery-profile" in the Pinebook Pro board dts. Without that data,
the fuel gauge reportedly works unreliably, so changing the design capacity
won't have any negative effects on the already unreliable operation of the
fuel gauge in the Pinebook Pros that belong to the latest batch.
According to the battery datasheet, its voltage can go as low as 2.75 V while
discharging, but it's better to leave the current 3.0 V value in the dts file,
because of the associated Pinebook Pro's voltage regulation issues.
Increase the frequency of the PWM signal that drives the LED backlight of
the Pinebook Pro's panel, from about 1.35 KHz (which equals to the PWM
period of 740,740 ns), to exactly 8 kHz (which equals to the PWM period of
125,000 ns). Using a higher PWM frequency for the panel backlight, which
reduces the flicker, can only be beneficial to the end users' eyes.
On top of that, increasing the backlight PWM signal frequency reportedly
eliminates the buzzing emitted from the Pinebook Pro's built-in speakers
when certain backlight levels are set, which cause some weird interference
with some of the components of the Pinebook Pro's audio chain.
The old value for the backlight PWM period, i.e. 740,740 ns, is pretty much
an arbitrary value that was selected during the very early bring-up of the
Pinebook Pro, only because that value seemed to minimize horizontal line
distortion on the display, which resulted from the old X.org drivers causing
screen tearing when dragging windows around. That's no longer an issue, so
there are no reasons to stick with the old PWM period value.
The lower and the upper backlight PWM frequency limits for the Pinebook Pro's
panel, according to its datasheet, are 200 Hz and 10 kHz, respectively. [1]
These changes still leave some headroom, which may have some positive effects
on the lifetime expectancy of the panel's backlight LEDs.
Add explicit casting to prevent expantion of 32th bit of
u32 into highest half of u64 in several places.
For example, in inject_abt64:
ESR_ELx_EC_DABT_LOW << ESR_ELx_EC_SHIFT = 0x24 << 26.
This operation's result is int with 1 in 32th bit.
While casting this value into u64 (esr is u64) 1
fills 32 highest bits.
Found by Linux Verification Center (linuxtesting.org) with SVACE.
Some f2fs ioctl interfaces like f2fs_ioc_set_pin_file(),
f2fs_move_file_range(), and f2fs_defragment_range() missed to
check atomic_write status, which may cause potential race issue,
fix it.
The F2FS ioctls for starting and committing atomic writes check for
inode_owner_or_capable(), but this does not give LSMs like SELinux or
Landlock an opportunity to deny the write access - if the caller's FSUID
matches the inode's UID, inode_owner_or_capable() immediately returns true.
There are scenarios where LSMs want to deny a process the ability to write
particular files, even files that the FSUID of the process owns; but this
can currently partially be bypassed using atomic write ioctls in two ways:
- F2FS_IOC_START_ATOMIC_REPLACE + F2FS_IOC_COMMIT_ATOMIC_WRITE can
truncate an inode to size 0
- F2FS_IOC_START_ATOMIC_WRITE + F2FS_IOC_ABORT_ATOMIC_WRITE can revert
changes another process concurrently made to a file
Fix it by requiring FMODE_WRITE for these operations, just like for
F2FS_IOC_MOVE_RANGE. Since any legitimate caller should only be using these
ioctls when intending to write into the file, that seems unlikely to break
anything.
Fixes: 88b88a667971 ("f2fs: support atomic writes") Cc: stable@vger.kernel.org Signed-off-by: Jann Horn <jannh@google.com> Reviewed-by: Chao Yu <chao@kernel.org> Reviewed-by: Eric Biggers <ebiggers@google.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The result of multiplication between values derived from functions
dir_buckets() and bucket_blocks() *could* technically reach
2^30 * 2^2 = 2^32.
While unlikely to happen, it is prudent to ensure that it will not
lead to integer overflow. Thus, use mul_u32_u32() as it's more
appropriate to mitigate the issue.
Found by Linux Verification Center (linuxtesting.org) with static
analysis tool SVACE.
When dealing with large extents and calculating file offsets by
summing up according extent offsets and lengths of unsigned int type,
one may encounter possible integer overflow if the values are
big enough.
Prevent this from happening by expanding one of the addends to
(pgoff_t) type.
Found by Linux Verification Center (linuxtesting.org) with static
analysis tool SVACE.
[BUG]
There are some reports about invalid data backref objectids, the report
looks like this:
BTRFS critical (device sda): corrupt leaf: block=333654787489792 slot=110 extent bytenr=333413935558656 len=65536 invalid data ref objectid value 2543
The data ref objectid is the inode number inside the subvolume.
But in above case, the value is completely sane, not really showing the
problem.
[CAUSE]
The root cause of the problem is the deprecated feature, inode cache.
This feature results a special inode number, -12ULL, and it's no longer
recognized by tree-checker, triggering the error.
The direct problem here is the output of data ref objectid. The value
shown is in fact the dref_root (subvolume id), not the dref_objectid
(inode number).
[FIX]
Fix the output to use dref_objectid instead.
fill_pool() uses 'obj_pool_min_free' to decide whether objects should be
handed back to the kmem cache. But 'obj_pool_min_free' records the lowest
historical value of the number of objects in the object pool and not the
minimum number of objects which should be kept in the pool.
Use 'debug_objects_pool_min_level' instead, which holds the minimum number
which was scaled to the number of CPUs at boot time.
[ tglx: Massage change log ]
Fixes: d26bf5056fc0 ("debugobjects: Reduce number of pool_lock acquisitions in fill_pool()") Fixes: 36c4ead6f6df ("debugobjects: Add global free list and the counter") Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/all/20240904133944.2124-3-thunder.leizhen@huawei.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
"iw dev wlp2s0 station dump" shows incorrect rx bitrate:
tx bitrate: 866.7 MBit/s VHT-MCS 9 80MHz short GI VHT-NSS 2
rx bitrate: 86.7 MBit/s VHT-MCS 9 VHT-NSS 1
This is because the RX band width is calculated incorrectly. Fix the
calculation according to the phydm_rxsc_2_bw() function from the
official drivers.
After:
tx bitrate: 866.7 MBit/s VHT-MCS 9 80MHz short GI VHT-NSS 2
rx bitrate: 390.0 MBit/s VHT-MCS 9 80MHz VHT-NSS 1
It also works correctly with the AP configured for 20 MHz and 40 MHz.
Remove VID/PID 0bda:c82c as it was inadvertently added to the device
list in driver rtw8821cu. This VID/PID is for the rtw8822cu device
and it is already in the appropriate place for that device.
pt_event_snapshot_aux() uses pt->handle_nmi to determine if tracing
needs to be stopped, however tracing can still be going because
pt->handle_nmi is set to zero before tracing is stopped in pt_event_stop,
whereas pt_event_snapshot_aux() requires that tracing must be stopped in
order to copy a sample of trace from the buffer.
Instead call pt_config_stop() always, which anyway checks config for
RTIT_CTL_TRACEEN and does nothing if it is already clear.
Note pt_event_snapshot_aux() can continue to use pt->handle_nmi to
determine if the trace needs to be restarted afterwards.
The TPM event log table is a Linux specific construct, where the data
produced by the GetEventLog() boot service is cached in memory, and
passed on to the OS using an EFI configuration table.
The use of EFI_LOADER_DATA here results in the region being left
unreserved in the E820 memory map constructed by the EFI stub, and this
is the memory description that is passed on to the incoming kernel by
kexec, which is therefore unaware that the region should be reserved.
Even though the utility of the TPM2 event log after a kexec is
questionable, any corruption might send the parsing code off into the
weeds and crash the kernel. So let's use EFI_ACPI_RECLAIM_MEMORY
instead, which is always treated as reserved by the E820 conversion
logic.
Internal documentation suggest that the TUXEDO Polaris 15 Gen5 AMD might
have GMxXGxX as the board name instead of GMxXGxx.
Adding both to be on the safe side.
Signed-off-by: Werner Sembach <wse@tuxedocomputers.com> Cc: All applicable <stable@vger.kernel.org> Link: https://patch.msgid.link/20240910094008.1601230-1-wse@tuxedocomputers.com Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Only buffer objects are valid return values of _STR.
If something else is returned description_show() will access invalid
memory.
Fixes: d1efe3c324ea ("ACPI: Add new sysfs interface to export device description") Cc: All applicable <stable@vger.kernel.org> Signed-off-by: Thomas Weißschuh <linux@weissschuh.net> Link: https://patch.msgid.link/20240709-acpi-sysfs-groups-v2-1-058ab0667fa8@weissschuh.net Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If the net_conf pointer is NULL and the code attempts to access its
fields without a check, it will lead to a null pointer dereference.
Add a NULL check before dereferencing the pointer.
Found by Linux Verification Center (linuxtesting.org) with SVACE.
The violation of atomicity occurs when the drbd_uuid_set_bm function is
executed simultaneously with modifying the value of
device->ldev->md.uuid[UI_BITMAP]. Consider a scenario where, while
device->ldev->md.uuid[UI_BITMAP] passes the validity check when its
value is not zero, the value of device->ldev->md.uuid[UI_BITMAP] is
written to zero. In this case, the check in drbd_uuid_set_bm might refer
to the old value of device->ldev->md.uuid[UI_BITMAP] (before locking),
which allows an invalid value to pass the validity check, resulting in
inconsistency.
To address this issue, it is recommended to include the data validity
check within the locked section of the function. This modification
ensures that the value of device->ldev->md.uuid[UI_BITMAP] does not
change during the validation process, thereby maintaining its integrity.
This possible bug is found by an experimental static analysis tool
developed by our team. This tool analyzes the locking APIs to extract
function pairs that can be concurrently executed, and then analyzes the
instructions in the paired functions to identify possible concurrency
bugs including data races and atomicity violations.
Fixes: 9f2247bb9b75 ("drbd: Protect accesses to the uuid set with a spinlock") Cc: stable@vger.kernel.org Signed-off-by: Qiu-ji Chen <chenqiuji666@gmail.com> Reviewed-by: Philipp Reisner <philipp.reisner@linbit.com> Link: https://lore.kernel.org/r/20240913083504.10549-1-chenqiuji666@gmail.com Signed-off-by: Jens Axboe <axboe@kernel.dk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In case of sev PLATFORM_STATUS failure, sev_get_api_version() fails
resulting in sev_data field of psp_master nulled out. This later becomes
a problem when unloading the ccp module because the device has not been
unregistered (via misc_deregister()) before clearing the sev_data field
of psp_master. As a result, on reloading the ccp module, a duplicate
device issue is encountered as can be seen from the dmesg log below.
on reloading ccp module via modprobe ccp
Call Trace:
<TASK>
dump_stack_lvl+0xd7/0xf0
dump_stack+0x10/0x20
sysfs_warn_dup+0x5c/0x70
sysfs_create_dir_ns+0xbc/0xd
kobject_add_internal+0xb1/0x2f0
kobject_add+0x7a/0xe0
? srso_alias_return_thunk+0x5/0xfbef5
? get_device_parent+0xd4/0x1e0
? __pfx_klist_children_get+0x10/0x10
device_add+0x121/0x870
? srso_alias_return_thunk+0x5/0xfbef5
device_create_groups_vargs+0xdc/0x100
device_create_with_groups+0x3f/0x60
misc_register+0x13b/0x1c0
sev_dev_init+0x1d4/0x290 [ccp]
psp_dev_init+0x136/0x300 [ccp]
sp_init+0x6f/0x80 [ccp]
sp_pci_probe+0x2a6/0x310 [ccp]
? srso_alias_return_thunk+0x5/0xfbef5
local_pci_probe+0x4b/0xb0
work_for_cpu_fn+0x1a/0x30
process_one_work+0x203/0x600
worker_thread+0x19e/0x350
? __pfx_worker_thread+0x10/0x10
kthread+0xeb/0x120
? __pfx_kthread+0x10/0x10
ret_from_fork+0x3c/0x60
? __pfx_kthread+0x10/0x10
ret_from_fork_asm+0x1a/0x30
</TASK>
kobject: kobject_add_internal failed for sev with -EEXIST, don't try to register things with the same name in the same directory.
ccp 0000:22:00.1: sev initialization failed
ccp 0000:22:00.1: psp initialization failed
ccp 0000:a2:00.1: no command queues available
ccp 0000:a2:00.1: psp enabled
Address this issue by unregistering the /dev/sev before clearing out
sev_data in case of PLATFORM_STATUS failure.
The qcom_geni_serial_poll_bit() can be used to wait for events like
command completion and is supposed to wait for the time it takes to
clear a full fifo before timing out.
As noted by Doug, the current implementation does not account for start,
stop and parity bits when determining the timeout. The helper also does
not currently account for the shift register and the two-word
intermediate transfer register.
A too short timeout can specifically lead to lost characters when
waiting for a transfer to complete as the transfer is cancelled on
timeout.
Instead of determining the poll timeout on every call, store the fifo
timeout when updating it in set_termios() and make sure to take the
shift and intermediate registers into account. Note that serial core has
already added a 20 ms margin to the fifo timeout.
Also note that the current uart_fifo_timeout() interface does
unnecessary calculations on every call and did not exist in earlier
kernels so only store its result once. This facilitates backports too as
earlier kernels can derive the timeout from uport->timeout, which has
since been removed.
Fixes: c4f528795d1a ("tty: serial: msm_geni_serial: Add serial driver support for GENI based QUP") Cc: stable@vger.kernel.org # 4.17 Reported-by: Douglas Anderson <dianders@chromium.org> Tested-by: NÃcolas F. R. A. Prado <nfraprado@collabora.com> Signed-off-by: Johan Hovold <johan+linaro@kernel.org> Link: https://lore.kernel.org/r/20240906131336.23625-2-johan+linaro@kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
PCI xHC host should be stopped and xhci driver memory freed before putting
host to PCI D3 state during PCI remove callback.
Hosts with XHCI_SPURIOUS_WAKEUP quirk did this the wrong way around
and set the host to D3 before calling usb_hcd_pci_remove(dev), which will
access the host to stop it, and then free xhci.
The write to RP2_GLOBAL_CMD followed by an immediate read of
RP2_GLOBAL_CMD in rp2_reset_asic() is intented to flush out the write,
however by then the device is already in reset and cannot respond to a
memory cycle access.
On platforms such as the Raspberry Pi 4 and others using the
pcie-brcmstb.c driver, any memory access to a device that cannot respond
is met with a fatal system error, rather than being substituted with all
1s as is usually the case on PC platforms.
Swapping the delay and the read ensures that the device has finished
resetting before we attempt to read from it.
Fixes: 7d9f49afa451 ("serial: rp2: New driver for Comtrol RocketPort 2 cards") Cc: stable <stable@kernel.org> Suggested-by: Jim Quinlan <james.quinlan@broadcom.com> Signed-off-by: Florian Fainelli <florian.fainelli@broadcom.com> Link: https://lore.kernel.org/r/20240906225435.707837-1-florian.fainelli@broadcom.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Most firmware names are hardcoded strings, or are constructed from fairly
constrained format strings where the dynamic parts are just some hex
numbers or such.
However, there are a couple codepaths in the kernel where firmware file
names contain string components that are passed through from a device or
semi-privileged userspace; the ones I could find (not counting interfaces
that require root privileges) are:
- lpfc_sli4_request_firmware_update() seems to construct the firmware
filename from "ModelName", a string that was previously parsed out of
some descriptor ("Vital Product Data") in lpfc_fill_vpd()
- nfp_net_fw_find() seems to construct a firmware filename from a model
name coming from nfp_hwinfo_lookup(pf->hwinfo, "nffw.partno"), which I
think parses some descriptor that was read from the device.
(But this case likely isn't exploitable because the format string looks
like "netronome/nic_%s", and there shouldn't be any *folders* starting
with "netronome/nic_". The previous case was different because there,
the "%s" is *at the start* of the format string.)
- module_flash_fw_schedule() is reachable from the
ETHTOOL_MSG_MODULE_FW_FLASH_ACT netlink command, which is marked as
GENL_UNS_ADMIN_PERM (meaning CAP_NET_ADMIN inside a user namespace is
enough to pass the privilege check), and takes a userspace-provided
firmware name.
(But I think to reach this case, you need to have CAP_NET_ADMIN over a
network namespace that a special kind of ethernet device is mapped into,
so I think this is not a viable attack path in practice.)
Fix it by rejecting any firmware names containing ".." path components.
For what it's worth, I went looking and haven't found any USB device
drivers that use the firmware loader dangerously.
Cc: stable@vger.kernel.org Reviewed-by: Danilo Krummrich <dakr@kernel.org> Fixes: abb139e75c2c ("firmware: teach the kernel to load firmware files directly from the filesystem") Signed-off-by: Jann Horn <jannh@google.com> Acked-by: Luis Chamberlain <mcgrof@kernel.org> Link: https://lore.kernel.org/r/20240828-firmware-traversal-v3-1-c76529c63b5f@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The dwc2_handle_usb_suspend_intr() function disables gadget clocks in USB
peripheral mode when no other power-down mode is available (introduced by
commit 0112b7ce68ea ("usb: dwc2: Update dwc2_handle_usb_suspend_intr function.")).
However, the dwc2_drd_role_sw_set() USB role update handler attempts to
read DWC2 registers if the USB role has changed while the USB is in suspend
mode (when the clocks are gated). This causes the system to hang.
Release the gadget clocks before handling the USB role update.
Fixes: 0112b7ce68ea ("usb: dwc2: Update dwc2_handle_usb_suspend_intr function.") Cc: stable@vger.kernel.org Signed-off-by: Tomas Marek <tomas.marek@elrest.cz> Link: https://lore.kernel.org/r/20240906055025.25057-1-tomas.marek@elrest.cz Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fix changes incorrect usb_request->status returned during disabling
endpoints. Before fix the status returned during dequeuing requests
while disabling endpoint was ECONNRESET.
Patch change it to ESHUTDOWN.
Patch fixes issue detected during testing UVC gadget.
During stopping streaming the class starts dequeuing usb requests and
controller driver returns the -ECONNRESET status. After completion
requests the class or application "uvc-gadget" try to queue this
request again. Changing this status to ESHUTDOWN cause that UVC assumes
that endpoint is disabled, or device is disconnected and stops
re-queuing usb requests.
TIOCGSERIAL is an ioctl. Thus it must be atomic. It returns
two values. Racing with set_serial it can return an inconsistent
result. The mutex must be taken.
In terms of logic the bug is as old as the driver. In terms of
code it goes back to the conversion to the get_serial and
set_serial methods.
Signed-off-by: Oliver Neukum <oneukum@suse.com> Cc: stable <stable@kernel.org> Fixes: 99f75a1fcd865 ("cdc-acm: switch to ->[sg]et_serial()") Link: https://lore.kernel.org/r/20240912141916.1044393-1-oneukum@suse.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There is a small window during probing when IO is running
but the backlight is not registered. Processing events
during that time will crash. The completion handler
needs to check for a backlight before scheduling work.
The work can submit URBs and the URBs can schedule the work.
This cycle needs to be broken, when a device is to be stopped.
Use a flag to do so.
This is a design issue as old as the driver.
Signed-off-by: Oliver Neukum <oneukum@suse.com> Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") CC: stable@vger.kernel.org Link: https://patch.msgid.link/20240919123525.688065-1-oneukum@suse.com Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
All USB devices supported by rtw88 have the same problem: they don't
transmit beacons in AP mode. (Some?) SDIO devices are also affected.
The cause appears to be clearing BIT_EN_BCNQ_DL of REG_FWHW_TXQ_CTRL
before uploading the beacon reserved page, so don't clear the bit for
USB and SDIO devices.
Remove the CAN_CTRLMODE_3_SAMPLES announcement for CAN-USB/3-FD devices
because these devices don't support it.
The hardware has a Microchip SAM E70 microcontroller that uses a Bosch
MCAN IP core as CAN FD controller. But this MCAN core doesn't support
triple sampling.
Fixes: 80662d943075 ("can: esd_usb: Add support for esd CAN-USB/3") Cc: stable@vger.kernel.org Signed-off-by: Stefan Mätje <stefan.maetje@esd.eu> Reviewed-by: Vincent Mailhol <mailhol.vincent@wanadoo.fr> Link: https://patch.msgid.link/20240904222740.2985864-2-stefan.maetje@esd.eu Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
SD cards can produce write latency spikes on the order of a hundred
milliseconds. If the target firmware does not hide that latency during DATA
IN and OUT phases it can cause the PDMA circuitry to raise a processor bus
fault which in turn leads to an unreliable byte count and a DMA overrun.
The Last Byte Sent flag is used to detect the overrun but this mechanism is
unreliable on some systems. Instead, set a DID_ERROR result whenever there
is a bus fault during a PDMA send, unless the cause was a phase mismatch.
Cc: stable@vger.kernel.org # 5.15+ Reported-and-tested-by: Stan Johnson <userm57@yahoo.com> Fixes: 7c1f3e3447a1 ("scsi: mac_scsi: Treat Last Byte Sent time-out as failure") Signed-off-by: Finn Thain <fthain@linux-m68k.org> Link: https://lore.kernel.org/r/cc38df687ace2c4ffc375a683b2502fc476b600d.1723001788.git.fthain@linux-m68k.org Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Before the error handling can be revised, some preparation is needed.
Refactor the polling loop with a new function, macscsi_wait_for_drq().
This function will gain more call sites in the next patch.
After a bus fault, capture and log the chip registers immediately, if the
NDEBUG_PSEUDO_DMA macro is defined. Remove some printk(KERN_DEBUG ...)
messages that aren't needed any more. Don't skip the debug message when
bytes == 0. Show all of the byte counters in the debug messages.
Commit 8db8f6ce556a ("scsi: ufs: qcom: Add missing interconnect bandwidth
values for Gear 5") updated the ufs_qcom_bw_table for Gear 5. However, it
missed updating the cfg_bw value for the max mode.
Hence update the cfg_bw value for the max mode for UFS 4.x devices.
Ff the device returns page 0xb1 with length 8 (happens with qemu v2.x, for
example), sd_read_block_characteristics() may attempt an out-of-bounds
memory access when accessing the zoned field at offset 8.
Fixes: 7fb019c46eee ("scsi: sd: Switch to using scsi_device VPD pages") Cc: stable@vger.kernel.org Signed-off-by: Martin Wilck <mwilck@suse.com> Link: https://lore.kernel.org/r/20240912134308.282824-1-mwilck@suse.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When the user requests the ALL_SUB_MPAGES mode sense page,
ata_msense_control() adds the CDL_T2A_SUB_MPAGE twice instead of adding
the CDL_T2A_SUB_MPAGE and CDL_T2B_SUB_MPAGE pages information. Correct
the second call to ata_msense_control_spgt2() to report the
CDL_T2B_SUB_MPAGE page.
Fixes: 673b2fe6ff1d ("scsi: ata: libata-scsi: Add support for CDL pages mode sense") Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <dlemoal@kernel.org> Reviewed-by: Hannes Reinecke <hare@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Windows client write with FILE_APPEND_DATA when using git.
ksmbd should allow write it with this flags.
Z:\test>git commit -m "test"
fatal: cannot update the ref 'HEAD': unable to append to
'.git/logs/HEAD': Bad file descriptor
Fixes: 0626e6641f6b ("cifsd: add server handler for central processing and tranport layers") Cc: stable@vger.kernel.org # v5.15+ Signed-off-by: Namjae Jeon <linkinjeon@kernel.org> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Some file systems may not provide dot (.) and dot-dot (..) as they are
optional in POSIX. ksmbd can misjudge emptiness of a directory in those
file systems, since it assumes there are always at least two entries:
dot and dot-dot.
Just don't count dot and dot-dot.
Cc: stable@vger.kernel.org # v6.1+ Signed-off-by: Hobin Woo <hobin.woo@samsung.com> Acked-by: Namjae Jeon <linkinjeon@kernel.org> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
De-duplicate the same functionality in several places by hoisting
the is_dot_dotdot() utility function into linux/fs.h.
Suggested-by: Amir Goldstein <amir73il@gmail.com> Reviewed-by: Jeff Layton <jlayton@kernel.org> Reviewed-by: Amir Goldstein <amir73il@gmail.com> Acked-by: Christian Brauner <brauner@kernel.org> Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The 'ld' and 'std' instructions require a 4-byte aligned displacement
because they are DS-form instructions. But the "m" asm constraint
doesn't enforce that.
That can lead to build errors if the compiler chooses a non-aligned
displacement, as seen with GCC 14:
/tmp/ccuSzwiR.s: Assembler messages:
/tmp/ccuSzwiR.s:2579: Error: operand out of domain (39 is not a multiple of 4)
make[5]: *** [scripts/Makefile.build:229: net/core/page_pool.o] Error 1
Dumping the generated assembler shows:
ld 8,39(8) # MEM[(const struct atomic64_t *)_29].counter, t
Use the YZ constraints to tell the compiler either to generate a DS-form
displacement, or use an X-form instruction, either of which prevents the
build error.
See commit 2d43cc701b96 ("powerpc/uaccess: Fix build errors seen with
GCC 13/14") for more details on the constraint letters.
Fixes: 9f0cbea0d8cc ("[POWERPC] Implement atomic{, 64}_{read, write}() without volatile") Cc: stable@vger.kernel.org # v2.6.24+ Reported-by: Stephen Rothwell <sfr@canb.auug.org.au> Closes: https://lore.kernel.org/all/20240913125302.0a06b4c7@canb.auug.org.au Tested-by: Mina Almasry <almasrymina@google.com> Reviewed-by: Segher Boessenkool <segher@kernel.crashing.org> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://msgid.link/20240916120510.2017749-1-mpe@ellerman.id.au Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In find_asymmetric_key(), if all NULLs are passed in the id_{0,1,2}
arguments, the kernel will first emit WARN but then have an oops
because id_2 gets dereferenced anyway.
Add the missing id_2 check and move WARN_ON() to the final else branch
to avoid duplicate NULL checks.
Found by Linux Verification Center (linuxtesting.org) with Svace static
analysis tool.
Cc: stable@vger.kernel.org # v5.17+ Fixes: 7d30198ee24f ("keys: X.509 public key issuer lookup without AKID") Suggested-by: Sergey Shtylyov <s.shtylyov@omp.ru> Signed-off-by: Roman Smirnov <r.smirnov@omp.ru> Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru> Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org> Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently amdgpu takes backlight caps provided by the ACPI tables
on systems as is. If the firmware sets maximums that are too low
this means that users don't get a good experience.
To avoid having to maintain a quirk list of such systems, do a sanity
check on the values. Check that the spread is at least half of the
values that amdgpu would use if no ACPI table was found and if not
use the amdgpu defaults.
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3020 Reviewed-by: Harry Wentland <harry.wentland@amd.com> Signed-off-by: Mario Limonciello <mario.limonciello@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Cc: stable@vger.kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[WHY]
The calculated vtotal may has 1 line deviation. To get precisely
vtotal number, round the vtotal result.
Cc: Mario Limonciello <mario.limonciello@amd.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: stable@vger.kernel.org Reviewed-by: Anthony Koo <anthony.koo@amd.com> Signed-off-by: Robin Chen <robin.chen@amd.com> Signed-off-by: Alex Hung <alex.hung@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[WHY && HOW]
For some HDMI OVT timing, YCbCr422 encoding fails at the DSC
bandwidth check. The root cause is our DSC policy for timing
doesn't account for HDMI YCbCr422 native support.
Cc: Mario Limonciello <mario.limonciello@amd.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: stable@vger.kernel.org Reviewed-by: Chris Park <chris.park@amd.com> Signed-off-by: Leo Ma <hanghong.ma@amd.com> Signed-off-by: Alex Hung <alex.hung@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[how]
dsc recompute should be skipped if no mode change detected on the new
request. If detected, keep checking whether the stream is already on
current state or not.
Cc: Mario Limonciello <mario.limonciello@amd.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: stable@vger.kernel.org Reviewed-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com> Signed-off-by: Fangzhi Zuo <Jerry.Zuo@amd.com> Signed-off-by: Wayne Lin <wayne.lin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Use a dedicated mutex to guard kvm_usage_count to fix a potential deadlock
on x86 due to a chain of locks and SRCU synchronizations. Translating the
below lockdep splat, CPU1 #6 will wait on CPU0 #1, CPU0 #8 will wait on
CPU2 #3, and CPU2 #7 will wait on CPU1 #4 (if there's a writer, due to the
fairness of r/w semaphores).
Note, there are likely more potential deadlocks in KVM x86, e.g. the same
pattern of taking cpu_hotplug_lock outside of kvm_lock likely exists with
__kvmclock_cpufreq_notifier():
But, actually triggering such deadlocks is beyond rare due to the
combination of dependencies and timings involved. E.g. the cpufreq
notifier is only used on older CPUs without a constant TSC, mucking with
the NX hugepage mitigation while VMs are running is very uncommon, and
doing so while also onlining/offlining a CPU (necessary to generate
contention on cpu_hotplug_lock) would be even more unusual.
The most robust solution to the general cpu_hotplug_lock issue is likely
to switch vm_list to be an RCU-protected list, e.g. so that x86's cpufreq
notifier doesn't to take kvm_lock. For now, settle for fixing the most
blatant deadlock, as switching to an RCU-protected list is a much more
involved change, but add a comment in locking.rst to call out that care
needs to be taken when walking holding kvm_lock and walking vm_list.
======================================================
WARNING: possible circular locking dependency detected
6.10.0-smp--c257535a0c9d-pip #330 Tainted: G S O
------------------------------------------------------
tee/35048 is trying to acquire lock: ff6a80eced71e0a8 (&kvm->slots_lock){+.+.}-{3:3}, at: set_nx_huge_pages+0x179/0x1e0 [kvm]
but task is already holding lock: ffffffffc07abb08 (kvm_lock){+.+.}-{3:3}, at: set_nx_huge_pages+0x14a/0x1e0 [kvm]
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
Hoist kvm_x2apic_icr_write() above kvm_apic_write_nodecode() so that a
local helper to _read_ the x2APIC ICR can be added and used in the
nodecode path without needing a forward declaration.
Inject a #GP on a WRMSR(ICR) that attempts to set any reserved bits that
are must-be-zero on both Intel and AMD, i.e. any reserved bits other than
the BUSY bit, which Intel ignores and basically says is undefined.
KVM's xapic_state_test selftest has been fudging the bug since commit 4b88b1a518b3 ("KVM: selftests: Enhance handling WRMSR ICR register in
x2APIC mode"), which essentially removed the testcase instead of fixing
the bug.
WARN if the nodecode path triggers a #GP, as the CPU is supposed to check
reserved bits for ICR when it's partially virtualized.
When we share memory through FF-A and the description of the buffers
exceeds the size of the mapped buffer, the fragmentation API is used.
The fragmentation API allows specifying chunks of descriptors in subsequent
FF-A fragment calls and no upper limit has been established for this.
The entire memory region transferred is identified by a handle which can be
used to reclaim the transferred memory.
To be able to reclaim the memory, the description of the buffers has to fit
in the ffa_desc_buf.
Add a bounds check on the FF-A sharing path to prevent the memory reclaim
from failing.
Also do_ffa_mem_xfer() does not need __always_inline, except for the
BUILD_BUG_ON() aspect, which gets moved to a macro.
[maz: fixed the BUILD_BUG_ON() breakage with LLVM, thanks to Wei-Lin Chang
for the timely report]
Fixes: 634d90cf0ac65 ("KVM: arm64: Handle FFA_MEM_LEND calls from the host") Cc: stable@vger.kernel.org Reviewed-by: Sebastian Ene <sebastianene@google.com> Signed-off-by: Snehal Koukuntla <snehalreddy@google.com> Reviewed-by: Oliver Upton <oliver.upton@linux.dev> Link: https://lore.kernel.org/r/20240909180154.3267939-1-snehalreddy@google.com Signed-off-by: Marc Zyngier <maz@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The Gen6 devices have the same problem and the same Solution as the Gen5
ones.
Some TongFang barebones have touchpad and/or keyboard issues after
suspend, fixable with nomux + reset + noloop + nopnp. Luckily, none of
them have an external PS/2 port so this can safely be set for all of
them.
I'm not entirely sure if every device listed really needs all four quirks,
but after testing and production use, no negative effects could be
observed when setting all four.
Some TongFang barebones have touchpad and/or keyboard issues after
suspend, fixable with nomux + reset + noloop + nopnp. Luckily, none of
them have an external PS/2 port so this can safely be set for all of
them.
I'm not entirely sure if every device listed really needs all four quirks,
but after testing and production use, no negative effects could be
observed when setting all four.
During adp5588_setup(), we read all the events to clear the event FIFO.
However, adp5588_read() just calls i2c_smbus_read_byte_data() which
returns the byte read in case everything goes well. Hence, we need to
explicitly check for a negative error code instead of checking for
something different than 0.
Fixes: e960309ce318 ("Input: adp5588-keys - bail out on returned error") Cc: stable@vger.kernel.org Signed-off-by: Nuno Sa <nuno.sa@analog.com> Link: https://lore.kernel.org/r/20240920-fix-adp5588-err-check-v1-1-81f6e957ef24@analog.com Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Userspace can supply an iova and uptr such that the target iova alignment
becomes really big and ALIGN() overflows which corrupts the selected area
range during allocation. CONFIG_IOMMUFD_TEST can detect this:
Cap the automatic alignment to the huge page size, which is probably a
better idea overall. Huge automatic alignments can fragment and chew up
the available IOVA space without any reason.
Link: https://patch.msgid.link/r/0-v1-8009738b9891+1f7-iommufd_align_overflow_jgg@nvidia.com Cc: stable@vger.kernel.org Fixes: 51fe6141f0f6 ("iommufd: Data structure to provide IOVA to PFN mapping") Reviewed-by: Nicolin Chen <nicolinc@nvidia.com> Reported-by: syzbot+16073ebbc4c64b819b47@syzkaller.appspotmail.com Closes: https://lore.kernel.org/r/000000000000388410061a74f014@google.com Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As Fedor Pchelkin pointed out, this commit violates the
convention of using the macro return value, which causes errors.
For example, in functions tda18271_attach(), xc5000_attach(),
simple_tuner_attach().
Link: https://lore.kernel.org/linux-media/20240424202031.syigrtrtipbq5f2l@fpc/ Suggested-by: Fedor Pchelkin <pchelkin@ispras.ru> Signed-off-by: Roman Smirnov <r.smirnov@omp.ru> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Return devm_of_clk_add_hw_provider() in order to transfer the error, if it
fails due to resource allocation failure or device tree clock provider
registration failure.
Cc: stable@vger.kernel.org Fixes: ebbfabc16d23 ("ASoC: rt5682: Add CCF usage for providing I2S clks") Signed-off-by: Ma Ke <make24@iscas.ac.cn> Link: https://patch.msgid.link/20240830143154.3448004-1-make24@iscas.ac.cn Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The rpm_requests device nodes have the compatible node. As such the
rpmsg core uses OF modalias instead of a native rpmsg modalias. Thus if
smd-rpm is built as a module, it doesn't get autoloaded for the device.
Revert the commit bcabe1e09135 ("soc: qcom: smd-rpm: Match rpmsg channel
instead of compatible")
MSGF_LEG_MASK is laid out with INTA in bit 0, INTB in bit 1, INTC in bit 2,
and INTD in bit 3. Hardware IRQ numbers start at 0, and we register
PCI_NUM_INTX IRQs. So to enable INTA (aka hwirq 0) we should set bit 0.
Remove the subtraction of one.
This bug would cause INTx interrupts not to be delivered, as enabling INTB
would actually enable INTA, and enabling INTA wouldn't enable anything at
all. It is likely that this got overlooked for so long since most PCIe
hardware uses MSIs. This fixes the following UBSAN error:
UBSAN: shift-out-of-bounds in ../drivers/pci/controller/pcie-xilinx-nwl.c:389:11
shift exponent 18446744073709551615 is too large for 32-bit type 'int'
CPU: 1 PID: 61 Comm: kworker/u10:1 Not tainted 6.6.20+ #268
Hardware name: xlnx,zynqmp (DT)
Workqueue: events_unbound deferred_probe_work_func
Call trace:
dump_backtrace (arch/arm64/kernel/stacktrace.c:235)
show_stack (arch/arm64/kernel/stacktrace.c:242)
dump_stack_lvl (lib/dump_stack.c:107)
dump_stack (lib/dump_stack.c:114)
__ubsan_handle_shift_out_of_bounds (lib/ubsan.c:218 lib/ubsan.c:387)
nwl_unmask_leg_irq (drivers/pci/controller/pcie-xilinx-nwl.c:389 (discriminator 1))
irq_enable (kernel/irq/internals.h:234 kernel/irq/chip.c:170 kernel/irq/chip.c:439 kernel/irq/chip.c:432 kernel/irq/chip.c:345)
__irq_startup (kernel/irq/internals.h:239 kernel/irq/chip.c:180 kernel/irq/chip.c:250)
irq_startup (kernel/irq/chip.c:270)
__setup_irq (kernel/irq/manage.c:1800)
request_threaded_irq (kernel/irq/manage.c:2206)
pcie_pme_probe (include/linux/interrupt.h:168 drivers/pci/pcie/pme.c:348)
Fixes: 9a181e1093af ("PCI: xilinx-nwl: Modify IRQ chip for legacy interrupts") Link: https://lore.kernel.org/r/20240531161337.864994-3-sean.anderson@linux.dev Signed-off-by: Sean Anderson <sean.anderson@linux.dev> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Cc: stable@vger.kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Given how the call place in pcie_wait_for_link_delay() got structured now,
and that pcie_retrain_link() returns a potentially useful error code,
convert pcie_failed_link_retrain() to return an error code rather than a
boolean status, fixing handling at the call site mentioned. Update the
other call site accordingly.
Only return successful completion status from pcie_failed_link_retrain() if
retraining has actually been done, preventing excessive delays from being
triggered at call sites in a hope that communication will finally be
established with the downstream device where in fact nothing has been done
about the link in question that would justify such a hope.
Fix missing call to phy_power_off() in the error path of
imx6_pcie_host_init(). Remove unnecessary check for imx6_pcie->phy
as the PHY API already handles NULL pointers.
Fixes: cbcf8722b523 ("phy: freescale: imx8m-pcie: Fix the wrong order of phy_init() and phy_power_on()") Link: https://lore.kernel.org/linux-pci/20240729-pci2_upstream-v8-3-b68ee5ef2b4d@nxp.com Signed-off-by: Frank Li <Frank.Li@nxp.com>
[kwilczynski: commit log] Signed-off-by: Krzysztof Wilczyński <kwilczynski@kernel.org> Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org> Cc: <stable@vger.kernel.org> # 6.1+ Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit da87d35a6e51 ("PCI: dra7xx: Use threaded IRQ handler for
"dra7xx-pcie-main" IRQ") switched from devm_request_irq() to
devm_request_threaded_irq() for the "dra7xx-pcie-main" interrupt.
Since the primary handler was set to NULL, the "IRQF_ONESHOT" flag
should have also been set. Fix this.
The LBMS bit, where implemented, is set by hardware either in response
to the completion of retraining caused by writing 1 to the Retrain Link
bit or whenever hardware has changed the link speed or width in attempt
to correct unreliable link operation. It is never cleared by hardware
other than by software writing 1 to the bit position in the Link Status
register and we never do such a write.
We currently have two places, namely apply_bad_link_workaround() and
pcie_failed_link_retrain() in drivers/pci/controller/dwc/pcie-tegra194.c
and drivers/pci/quirks.c respectively where we check the state of the LBMS
bit and neither is interested in the state of the bit resulting from the
completion of retraining, both check for a link fault.
And in particular pcie_failed_link_retrain() causes issues consequently, by
trying to retrain a link where there's no downstream device anymore and the
state of 1 in the LBMS bit has been retained from when there was a device
downstream that has since been removed.
Clear the LBMS bit then at the conclusion of pcie_retrain_link(), so that
we have a single place that controls it and that our code can track link
speed or width changes resulting from unreliable link operation.
When `pcie_failed_link_retrain' has failed to retrain the link by hand
it leaves the link speed restricted to 2.5GT/s, which will then affect
any device that has been plugged in later on, which may not suffer from
the problem that caused the speed restriction to have been attempted.
Consequently such a downstream device will suffer from an unnecessary
communication throughput limitation and therefore performance loss.
Remove the speed restriction then and revert the Link Control 2 register
to its original state if link retraining with the speed restriction in
place has failed. Retrain the link again afterwards so as to remove any
residual state, waiting on LT rather than DLLLA to avoid an excessive
delay and ignoring the result as this training is supposed to fail
anyway.
Commit 3f1b0e1f2875 (".gitignore update") added *.orig and *.rej
patterns to .gitignore in v2.6.23. The commit message didn't give a
rationale. Later on, commit 1f5d3a6b6532 ("Remove *.rej pattern from
.gitignore") removed the *.rej pattern in v2.6.26, on the rationale that
*.rej files indicated something went really wrong and should not be
ignored.
The *.rej files are now shown by `git status`, which helps located
conflicts when applying patches and lowers the probability that they
will go unnoticed. It is however still easy to overlook the *.orig files
which slowly polute the source tree. That's not as big of a deal as not
noticing a conflict, but it's still not nice.
Drop the *.orig pattern from .gitignore to avoid this and help keep the
source tree clean.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
[masahiroy@kernel.org:
I do not have a strong opinion about this. Perhaps some people may have
a different opinion.
If you are someone who wants to ignore *.orig, it is likely you would
want to do so across all projects. Then, $XDG_CONFIG_HOME/git/ignore
would be more suitable for your needs. gitignore(5) suggests, "Patterns
which a user wants Git to ignore in all situations generally go into a
file specified by core.excludesFile in the user's ~/.gitconfig".
Please note that you cannot do the opposite; if *.orig is ignored by
the project's .gitignore, you cannot override the decision because
$XDG_CONFIG_HOME/git/ignore has a lower priority.
If *.orig is sitting on the fence, I'd leave it to the users. ] Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Putting the cpumask on the stack is deprecated for a long time (since 2d3854a37e8), as these can be big. Given that, change the on-stack
allocation of allowed_mask to be dynamically allocated.
Fixes: f011c9cf04c0 ("io_uring/sqpoll: do not allow pinning outside of cpuset") Signed-off-by: Felix Moessbauer <felix.moessbauer@siemens.com> Link: https://lore.kernel.org/r/20240916111150.1266191-1-felix.moessbauer@siemens.com Signed-off-by: Jens Axboe <axboe@kernel.dk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
A recent commit ensured that SQPOLL cannot be setup with a CPU that
isn't in the current tasks cpuset, but it also dropped testing whether
the CPU is valid in the first place. Without that, if a task passes in
a CPU value that is too high, the following KASAN splat can get
triggered:
BUG: KASAN: stack-out-of-bounds in io_sq_offload_create+0x858/0xaa4
Read of size 8 at addr ffff800089bc7b90 by task wq-aff.t/1391
When running as a Xen PV dom0 the system needs to map ACPI data of the
host using host physical addresses, while those addresses can conflict
with the guest physical addresses of the loaded linux kernel. The same
problem might apply in case a PV guest is configured to use the host
memory map.
This conflict can be solved by mapping the ACPI data to a different
guest physical address, but mapping the data via acpi_os_ioremap()
must still be possible using the host physical address, as this
address might be generated by AML when referencing some of the ACPI
data.
When configured to support running as a Xen PV domain, have an
implementation of acpi_os_ioremap() being aware of the possibility to
need above mentioned translation of a host physical address to the
guest physical address.
This modification requires to #include linux/acpi.h in some sources
which need to include asm/acpi.h directly.
Move the checks for e820 memory map conflicts using the
xen_chk_is_e820_usable() helper further up in order to prepare
resolving some of the possible conflicts by doing some e820 map
modifications, which must happen before evaluating the RAM layout.
Signed-off-by: Juergen Gross <jgross@suse.com> Tested-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com> Reviewed-by: Jan Beulich <jbeulich@suse.com> Signed-off-by: Juergen Gross <jgross@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This reverts commit 69197dfc64007b5292cc960581548f41ccd44828.
commit 937d46ecc5f9 ("net: wangxun: add ethtool_ops for
channel number") changed NIC misc irq from most significant
bit to least significant bit, the former condition is not
required to apply this patch, because we only need to set
irq affinity for NIC queue irq vectors.
this patch is required after commit 937d46ecc5f9 ("net: wangxun:
add ethtool_ops for channel number") was applied, so this is only
relevant to 6.6.y branch.
The kms paths keep a persistent map active to read and compare the cursor
buffer. These maps can race with each other in simple scenario where:
a) buffer "a" mapped for update
b) buffer "a" mapped for compare
c) do the compare
d) unmap "a" for compare
e) update the cursor
f) unmap "a" for update
At step "e" the buffer has been unmapped and the read contents is bogus.
Prevent unmapping of active read buffers by simply keeping a count of
how many paths have currently active maps and unmap only when the count
reaches 0.
Fixes: 485d98d472d5 ("drm/vmwgfx: Add support for CursorMob and CursorBypass 4") Cc: Broadcom internal kernel review list <bcm-kernel-feedback-list@broadcom.com> Cc: dri-devel@lists.freedesktop.org Cc: <stable@vger.kernel.org> # v5.19+ Signed-off-by: Zack Rusin <zack.rusin@broadcom.com> Link: https://patchwork.freedesktop.org/patch/msgid/20240816183332.31961-2-zack.rusin@broadcom.com Reviewed-by: Martin Krastev <martin.krastev@broadcom.com> Reviewed-by: Maaz Mombasawala <maaz.mombasawala@broadcom.com>
[Shivani: Modified to apply on v6.6.y] Signed-off-by: Shivani Agarwal <shivani.agarwal@broadcom.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The remap_file_pages syscall handler calls do_mmap() directly, which
doesn't contain the LSM security check. And if the process has called
personality(READ_IMPLIES_EXEC) before and remap_file_pages() is called for
RW pages, this will actually result in remapping the pages to RWX,
bypassing a W^X policy enforced by SELinux.
So we should check prot by security_mmap_file LSM hook in the
remap_file_pages syscall handler before do_mmap() is called. Otherwise, it
potentially permits an attacker to bypass a W^X policy enforced by
SELinux.
The bypass is similar to CVE-2016-10044, which bypass the same thing via
AIO and can be found in [1].
The PoC:
$ cat > test.c
int main(void) {
size_t pagesz = sysconf(_SC_PAGE_SIZE);
int mfd = syscall(SYS_memfd_create, "test", 0);
const char *buf = mmap(NULL, 4 * pagesz, PROT_READ | PROT_WRITE,
MAP_SHARED, mfd, 0);
unsigned int old = syscall(SYS_personality, 0xffffffff);
syscall(SYS_personality, READ_IMPLIES_EXEC | old);
syscall(SYS_remap_file_pages, buf, pagesz, 0, 2, 0);
syscall(SYS_personality, old);
// show the RWX page exists even if W^X policy is enforced
int fd = open("/proc/self/maps", O_RDONLY);
unsigned char buf2[1024];
while (1) {
int ret = read(fd, buf2, 1024);
if (ret <= 0) break;
write(1, buf2, ret);
}
close(fd);
}
Link: https://project-zero.issues.chromium.org/issues/42452389 Cc: stable@vger.kernel.org Signed-off-by: Shu Han <ebpqwerty472123@gmail.com> Acked-by: Stephen Smalley <stephen.smalley.work@gmail.com>
[PM: subject line tweaks] Signed-off-by: Paul Moore <paul@paul-moore.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If some part of the kernel adds task_work that needs executing, in terms
of signaling it'll generally use TWA_SIGNAL or TWA_RESUME. Those two
directly translate to TIF_NOTIFY_SIGNAL or TIF_NOTIFY_RESUME, and can
be used for a variety of use case outside of task_work.
However, io_cqring_wait_schedule() only tests explicitly for
TIF_NOTIFY_SIGNAL. This means it can miss if task_work got added for
the task, but used a different kind of signaling mechanism (or none at
all). Normally this doesn't matter as any task_work will be run once
the task exits to userspace, except if:
1) The ring is setup with DEFER_TASKRUN
2) The local work item may generate normal task_work
For condition 2, this can happen when closing a file and it's the final
put of that file, for example. This can cause stalls where a task is
waiting to make progress inside io_cqring_wait(), but there's nothing else
that will wake it up. Hence change the "should we schedule or loop around"
check to check for the presence of task_work explicitly, rather than just
TIF_NOTIFY_SIGNAL as the mechanism. While in there, also change the
ordering of what type of task_work first in terms of ordering, to both
make it consistent with other task_work runs in io_uring, but also to
better handle the case of defer task_work generating normal task_work,
like in the above example.
The submit queue polling threads are userland threads that just never
exit to the userland. When creating the thread with IORING_SETUP_SQ_AFF,
the affinity of the poller thread is set to the cpu specified in
sq_thread_cpu. However, this CPU can be outside of the cpuset defined
by the cgroup cpuset controller. This violates the rules defined by the
cpuset controller and is a potential issue for realtime applications.
In b7ed6d8ffd6 we fixed the default affinity of the poller thread, in
case no explicit pinning is required by inheriting the one of the
creating task. In case of explicit pinning, the check is more
complicated, as also a cpu outside of the parent cpumask is allowed.
We implemented this by using cpuset_cpus_allowed (that has support for
cgroup cpusets) and testing if the requested cpu is in the set.
Lockless iteration over hook list is possible from netlink dump path,
use rcu variant to iterate over the hook list as is done with flowtable
hooks.
Fixes: b9703ed44ffb ("netfilter: nf_tables: support for adding new devices to an existing netdev chain") Reported-by: Phil Sutter <phil@nwl.cc> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Only provide ctnetlink_label_size when it is used,
which is when CONFIG_NF_CONNTRACK_EVENTS is configured.
Flagged by clang-18 W=1 builds as:
.../nf_conntrack_netlink.c:385:19: warning: unused function 'ctnetlink_label_size' [-Wunused-function]
385 | static inline int ctnetlink_label_size(const struct nf_conn *ct)
| ^~~~~~~~~~~~~~~~~~~~
The condition on CONFIG_NF_CONNTRACK_LABELS being removed by
this patch guards compilation of non-trivial implementations
of ctnetlink_dump_labels() and ctnetlink_label_size().
However, this is not necessary as each of these functions
will always return 0 if CONFIG_NF_CONNTRACK_LABELS is not defined
as each function starts with the equivalent of:
And nf_ct_labels_find always returns NULL if CONFIG_NF_CONNTRACK_LABELS
is not enabled. So I believe that the compiler optimises the code away
in such cases anyway.
Found by inspection.
Compile tested only.
Originally splitted in two patches, Pablo Neira Ayuso collapsed them and
added Fixes: tag.
Fixes: 0ceabd83875b ("netfilter: ctnetlink: deliver labels to userspace") Link: https://lore.kernel.org/netfilter-devel/20240909151712.GZ2097826@kernel.org/ Signed-off-by: Simon Horman <horms@kernel.org> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Documentation of list_del_rcu() warns callers to not immediately free
the deleted list item. While it seems not necessary to use the
RCU-variant of list_del() here in the first place, doing so seems to
require calling kfree_rcu() on the deleted item as well.
Fixes: 3f0465a9ef02 ("netfilter: nf_tables: dynamically allocate hooks per net_device in flowtables") Signed-off-by: Phil Sutter <phil@nwl.cc> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Commit 5fabb01207a2 ("net: stmmac: Add initial XDP support") sets
PP_FLAG_DMA_SYNC_DEV flag for page_pool unconditionally,
page_pool_recycle_direct() will call page_pool_dma_sync_for_device()
on every page even the page is not going to be reused by XDP program.
When XDP is not enabled, the page which holds the received buffer
will be recycled once the buffer is copied into new SKB by
skb_copy_to_linear_data(), then the MAC core will never reuse this
page any longer. Always setting PP_FLAG_DMA_SYNC_DEV wastes CPU cycles
on unnecessary calling of page_pool_dma_sync_for_device().
After this patch, up to 9% noticeable performance improvement was observed
on certain platforms.
Currently, the virtio-net driver will perform a pre-dma-mapping for
small or mergeable RX buffer. But for small packets, a mismatched address
without VIRTNET_RX_PAD and xdp_headroom is used for unmapping.
That will result in unsynchronized buffers when SWIOTLB is enabled, for
example, when running as a TDX guest.
This patch unifies the address passed to the virtio core as the address of
the virtnet header and fixes the mismatched buffer address.
Changes from v2: unify the buf that passed to the virtio core in small
and merge mode.
Changes from v1: Use ctx to get xdp_headroom.
Fixes: 295525e29a5b ("virtio_net: merge dma operations when filling mergeable buffers") Signed-off-by: Wenbo Li <liwenbo.martin@bytedance.com> Signed-off-by: Jiahui Cen <cenjiahui@bytedance.com> Signed-off-by: Ying Fang <fangying.tommy@bytedance.com> Reviewed-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com> Link: https://patch.msgid.link/20240919081351.51772-1-liwenbo.martin@bytedance.com Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
syzbot reported a WARNING in bond_xdp_get_xmit_slave. To reproduce
this[1], one bond device (bond1) has xdpdrv, which increases
bpf_master_redirect_enabled_key. Another bond device (bond0) which is
unsupported by XDP but its slave (veth3) has xdpgeneric that returns
XDP_TX. This triggers WARN_ON_ONCE() from the xdp_master_redirect().
To reduce unnecessary warnings and improve log management, we need to
delete the WARN_ON_ONCE() and add ratelimit to the netdev_err().
[1] Steps to reproduce:
# Needs tx_xdp with return XDP_TX;
ip l add veth0 type veth peer veth1
ip l add veth3 type veth peer veth4
ip l add bond0 type bond mode 6 # BOND_MODE_ALB, unsupported by XDP
ip l add bond1 type bond # BOND_MODE_ROUNDROBIN by default
ip l set veth0 master bond1
ip l set bond1 up
# Increases bpf_master_redirect_enabled_key
ip l set dev bond1 xdpdrv object tx_xdp.o section xdp_tx
ip l set veth3 master bond0
ip l set bond0 up
ip l set veth4 up
# Triggers WARN_ON_ONCE() from the xdp_master_redirect()
ip l set veth3 xdpgeneric object tx_xdp.o section xdp_tx
Reported-by: syzbot+c187823a52ed505b2257@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=c187823a52ed505b2257 Fixes: 9e2ee5c7e7c3 ("net, bonding: Add XDP support to the bonding driver") Signed-off-by: Jiwon Kim <jiwonaid0@gmail.com> Signed-off-by: Nikolay Aleksandrov <razor@blackwall.org> Link: https://patch.msgid.link/20240918140602.18644-1-jiwonaid0@gmail.com Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
When broadcasting data to multiple nodes via MHI, using skb_clone()
causes all nodes to receive the same header data. This can result in
packets being discarded by endpoints, leading to lost data.
This issue occurs when a socket is closed, and a QRTR_TYPE_DEL_CLIENT
packet is broadcasted. All nodes receive the same destination node ID,
causing the node connected to the client to discard the packet and
remain unaware of the client's deletion.
Replace skb_clone() with pskb_copy(), to create a separate copy of
the header for each sk_buff.
Fixes: bdabad3e363d ("net: Add Qualcomm IPC router") Signed-off-by: Youssef Samir <quic_yabdulra@quicinc.com> Reviewed-by: Jeffery Hugo <quic_jhugo@quicinc.com> Reviewed-by: Carl Vanderlip <quic_carlv@quicinc.com> Reviewed-by: Chris Lew <quic_clew@quicinc.com> Link: https://patch.msgid.link/20240916170858.2382247-1-quic_yabdulra@quicinc.com Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
We have some machines running stock Ubuntu 20.04.6 which is their 5.4.0-174-generic
kernel that are running ceph and recently hit a null ptr dereference in
tcp_rearm_rto(). Initially hitting it from the TLP path, but then later we also
saw it getting hit from the RACK case as well. Here are examples of the oops
messages we saw in each of those cases:
After we hit this we disabled TLP by setting tcp_early_retrans to 0 and then hit the crash in the RACK case:
Aug 7 07:26:16 rx [1006006.265582] BUG: kernel NULL pointer dereference, address: 0000000000000020
Aug 7 07:26:16 rx [1006006.272719] #PF: supervisor read access in kernel mode
Aug 7 07:26:16 rx [1006006.278030] #PF: error_code(0x0000) - not-present page
Aug 7 07:26:16 rx [1006006.283343] PGD 0 P4D 0
Aug 7 07:26:16 rx [1006006.286057] Oops: 0000 [#1] SMP NOPTI
Aug 7 07:26:16 rx [1006006.289896] CPU: 5 PID: 0 Comm: swapper/5 Tainted: G W 5.4.0-174-generic #193-Ubuntu
Aug 7 07:26:16 rx [1006006.299107] Hardware name: Supermicro SMC 2x26 os-gen8 64C NVME-Y 256G/H12SSW-NTR, BIOS 2.5.V1.2U.NVMe.UEFI 05/09/2023
Aug 7 07:26:16 rx [1006006.309970] RIP: 0010:tcp_rearm_rto+0xe4/0x160
Aug 7 07:26:16 rx [1006006.314584] Code: 87 ca 04 00 00 00 5b 41 5c 41 5d 5d c3 c3 49 8b bc 24 40 06 00 00 eb 8d 48 bb cf f7 53 e3 a5 9b c4 20 4c 89 ef e8 0c fe 0e 00 <48> 8b 78 20 48 c1 ef 03 48 89 f8 41 8b bc 24 80 04 00 00 48 f7 e3
Aug 7 07:26:16 rx [1006006.333499] RSP: 0018:ffffb42600a50960 EFLAGS: 00010246
Aug 7 07:26:16 rx [1006006.338895] RAX: 0000000000000000 RBX: 20c49ba5e353f7cf RCX: 0000000000000000
Aug 7 07:26:16 rx [1006006.346193] RDX: 0000000000000000 RSI: 0000000000000001 RDI: ffff92d687ed8160
Aug 7 07:26:16 rx [1006006.353489] RBP: ffffb42600a50978 R08: 0000000000000000 R09: 00000000cd896dcc
Aug 7 07:26:16 rx [1006006.360786] R10: ffff92dc3404f400 R11: 0000000000000001 R12: ffff92d687ed8000
Aug 7 07:26:16 rx [1006006.368084] R13: ffff92d687ed8160 R14: 00000000cd896dcc R15: 00000000cd8fca81
Aug 7 07:26:16 rx [1006006.375381] FS: 0000000000000000(0000) GS:ffff93158ad40000(0000) knlGS:0000000000000000
Aug 7 07:26:16 rx [1006006.383632] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
Aug 7 07:26:16 rx [1006006.389544] CR2: 0000000000000020 CR3: 0000003e775ce006 CR4: 0000000000760ee0
Aug 7 07:26:16 rx [1006006.396839] PKRU: 55555554
Aug 7 07:26:16 rx [1006006.399717] Call Trace:
Aug 7 07:26:16 rx [1006006.402335]
Aug 7 07:26:16 rx [1006006.404525] ? show_regs.cold+0x1a/0x1f
Aug 7 07:26:16 rx [1006006.408532] ? __die+0x90/0xd9
Aug 7 07:26:16 rx [1006006.411760] ? no_context+0x196/0x380
Aug 7 07:26:16 rx [1006006.415599] ? __bad_area_nosemaphore+0x50/0x1a0
Aug 7 07:26:16 rx [1006006.420392] ? _raw_spin_lock+0x1e/0x30
Aug 7 07:26:16 rx [1006006.424401] ? bad_area_nosemaphore+0x16/0x20
Aug 7 07:26:16 rx [1006006.428927] ? do_user_addr_fault+0x267/0x450
Aug 7 07:26:16 rx [1006006.433450] ? __do_page_fault+0x58/0x90
Aug 7 07:26:16 rx [1006006.437542] ? do_page_fault+0x2c/0xe0
Aug 7 07:26:16 rx [1006006.441470] ? page_fault+0x34/0x40
Aug 7 07:26:16 rx [1006006.445134] ? tcp_rearm_rto+0xe4/0x160
Aug 7 07:26:16 rx [1006006.449145] tcp_ack+0xa32/0xb30
Aug 7 07:26:16 rx [1006006.452542] tcp_rcv_established+0x13c/0x670
Aug 7 07:26:16 rx [1006006.456981] ? sk_filter_trim_cap+0x48/0x220
Aug 7 07:26:16 rx [1006006.461419] tcp_v6_do_rcv+0xdb/0x450
Aug 7 07:26:16 rx [1006006.465257] tcp_v6_rcv+0xc2b/0xd10
Aug 7 07:26:16 rx [1006006.468918] ip6_protocol_deliver_rcu+0xd3/0x4e0
Aug 7 07:26:16 rx [1006006.473706] ip6_input_finish+0x15/0x20
Aug 7 07:26:16 rx [1006006.477710] ip6_input+0xa2/0xb0
Aug 7 07:26:16 rx [1006006.481109] ? ip6_protocol_deliver_rcu+0x4e0/0x4e0
Aug 7 07:26:16 rx [1006006.486151] ip6_sublist_rcv_finish+0x3d/0x50
Aug 7 07:26:16 rx [1006006.490679] ip6_sublist_rcv+0x1aa/0x250
Aug 7 07:26:16 rx [1006006.494779] ? ip6_rcv_finish_core.isra.0+0xa0/0xa0
Aug 7 07:26:16 rx [1006006.499828] ipv6_list_rcv+0x112/0x140
Aug 7 07:26:16 rx [1006006.503748] __netif_receive_skb_list_core+0x1a4/0x250
Aug 7 07:26:16 rx [1006006.509057] netif_receive_skb_list_internal+0x1a1/0x2b0
Aug 7 07:26:16 rx [1006006.514538] gro_normal_list.part.0+0x1e/0x40
Aug 7 07:26:16 rx [1006006.519068] napi_complete_done+0x91/0x130
Aug 7 07:26:16 rx [1006006.523352] mlx5e_napi_poll+0x18e/0x610 [mlx5_core]
Aug 7 07:26:16 rx [1006006.528481] net_rx_action+0x142/0x390
Aug 7 07:26:16 rx [1006006.532398] __do_softirq+0xd1/0x2c1
Aug 7 07:26:16 rx [1006006.536142] irq_exit+0xae/0xb0
Aug 7 07:26:16 rx [1006006.539452] do_IRQ+0x5a/0xf0
Aug 7 07:26:16 rx [1006006.542590] common_interrupt+0xf/0xf
Aug 7 07:26:16 rx [1006006.546421]
Aug 7 07:26:16 rx [1006006.548695] RIP: 0010:native_safe_halt+0xe/0x10
Aug 7 07:26:16 rx [1006006.553399] Code: 7b ff ff ff eb bd 90 90 90 90 90 90 e9 07 00 00 00 0f 00 2d 36 2c 50 00 f4 c3 66 90 e9 07 00 00 00 0f 00 2d 26 2c 50 00 fb f4 90 0f 1f 44 00 00 55 48 89 e5 41 55 41 54 53 e8 dd 5e 61 ff 65
Aug 7 07:26:16 rx [1006006.572309] RSP: 0018:ffffb42600177e70 EFLAGS: 00000246 ORIG_RAX: ffffffffffffffc2
Aug 7 07:26:16 rx [1006006.580040] RAX: ffffffff8ed08b20 RBX: 0000000000000005 RCX: 0000000000000001
Aug 7 07:26:16 rx [1006006.587337] RDX: 00000000f48eeca2 RSI: 0000000000000082 RDI: 0000000000000082
Aug 7 07:26:16 rx [1006006.594635] RBP: ffffb42600177e90 R08: 0000000000000000 R09: 000000000000020f
Aug 7 07:26:16 rx [1006006.601931] R10: 0000000000100000 R11: 0000000000000000 R12: 0000000000000005
Aug 7 07:26:16 rx [1006006.609229] R13: ffff93157deb5f00 R14: 0000000000000000 R15: 0000000000000000
Aug 7 07:26:16 rx [1006006.616530] ? __cpuidle_text_start+0x8/0x8
Aug 7 07:26:16 rx [1006006.620886] ? default_idle+0x20/0x140
Aug 7 07:26:16 rx [1006006.624804] arch_cpu_idle+0x15/0x20
Aug 7 07:26:16 rx [1006006.628545] default_idle_call+0x23/0x30
Aug 7 07:26:16 rx [1006006.632640] do_idle+0x1fb/0x270
Aug 7 07:26:16 rx [1006006.636035] cpu_startup_entry+0x20/0x30
Aug 7 07:26:16 rx [1006006.640126] start_secondary+0x178/0x1d0
Aug 7 07:26:16 rx [1006006.644218] secondary_startup_64+0xa4/0xb0
Aug 7 07:26:17 rx [1006006.648568] Modules linked in: vrf bridge stp llc vxlan ip6_udp_tunnel udp_tunnel nls_iso8859_1 nft_ct amd64_edac_mod edac_mce_amd kvm_amd kvm crct10dif_pclmul ghash_clmulni_intel aesni_intel crypto_simd cryptd glue_helper wmi_bmof ipmi_ssif input_leds joydev rndis_host cdc_ether usbnet ast mii drm_vram_helper ttm drm_kms_helper i2c_algo_bit fb_sys_fops syscopyarea sysfillrect sysimgblt ccp mac_hid ipmi_si ipmi_devintf ipmi_msghandler sch_fq_codel nf_tables_set nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 nf_tables nfnetlink ramoops reed_solomon efi_pstore drm ip_tables x_tables autofs4 raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx xor raid6_pq libcrc32c raid0 multipath linear mlx5_ib ib_uverbs ib_core raid1 hid_generic mlx5_core pci_hyperv_intf crc32_pclmul usbhid ahci tls mlxfw bnxt_en hid libahci nvme i2c_piix4 nvme_core wmi [last unloaded: cpuid]
Aug 7 07:26:17 rx [1006006.726180] CR2: 0000000000000020
Aug 7 07:26:17 rx [1006006.729718] ---[ end trace e0e2e37e4e612984 ]---
Prior to seeing the first crash and on other machines we also see the warning in
tcp_send_loss_probe() where packets_out is non-zero, but both transmit and retrans
queues are empty so we know the box is seeing some accounting issue in this area:
The NULL ptr deref is coming from tcp_rto_delta_us() attempting to pull an skb
off the head of the retransmit queue and then dereferencing that skb to get the
skb_mstamp_ns value via tcp_skb_timestamp_us(skb).
The crash is the same one that was reported a # of years ago here:
https://lore.kernel.org/netdev/86c0f836-9a7c-438b-d81a-839be45f1f58@gmail.com/T/#t
and the kernel we're running has the fix which was added to resolve this issue.
Unfortunately we've been unsuccessful so far in reproducing this problem in the
lab and do not have the luxury of pushing out a new kernel to try and test if
newer kernels resolve this issue at the moment. I realize this is a report
against both an Ubuntu kernel and also an older 5.4 kernel. I have reported this
issue to Ubuntu here: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2077657
however I feel like since this issue has possibly cropped up again it makes
sense to build in some protection in this path (even on the latest kernel
versions) since the code in question just blindly assumes there's a valid skb
without testing if it's NULL b/f it looks at the timestamp.
Given we have seen crashes in this path before and now this case it seems like
we should protect ourselves for when packets_out accounting is incorrect.
While we should fix that root cause we should also just make sure the skb
is not NULL before dereferencing it. Also add a warn once here to capture
some information if/when the problem case is hit again.
Fixes: e1a10ef7fa87 ("tcp: introduce tcp_rto_delta_us() helper for xmit timer fix") Signed-off-by: Josh Hunt <johunt@akamai.com> Acked-by: Neal Cardwell <ncardwell@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
projects / thirdparty / kernel / stable.git / log
