platform/kernel/linux-starfive.git
12 months agoMerge tag 'perf-urgent-2023-10-21' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 21 Oct 2023 18:09:29 +0000 (11:09 -0700)]
Merge tag 'perf-urgent-2023-10-21' of git://git./linux/kernel/git/tip/tip

Pull perf events fix from Ingo Molnar:
 "Fix group event semantics"

* tag 'perf-urgent-2023-10-21' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf: Disallow mis-matched inherited group reads

12 months agoMerge tag 'probes-fixes-v6.6-rc6.2' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 21 Oct 2023 18:00:36 +0000 (11:00 -0700)]
Merge tag 'probes-fixes-v6.6-rc6.2' of git://git./linux/kernel/git/trace/linux-trace

Pull probes fixes from Masami Hiramatsu:

 - kprobe-events: Fix kprobe events to reject if the attached symbol is
   not unique name because it may not the function which the user want
   to attach to. (User can attach a probe to such symbol using the
   nearest unique symbol + offset.)

 - selftest: Add a testcase to ensure the kprobe event rejects non
   unique symbol correctly.

* tag 'probes-fixes-v6.6-rc6.2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  selftests/ftrace: Add new test case which checks non unique symbol
  tracing/kprobes: Return EADDRNOTAVAIL when func matches several symbols

12 months agoMerge tag 's390-6.6-4' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux
Linus Torvalds [Sat, 21 Oct 2023 17:11:11 +0000 (10:11 -0700)]
Merge tag 's390-6.6-4' of git://git./linux/kernel/git/s390/linux

Pull s390 fixes from Vasily Gorbik:

 - Fix IOMMU bitmap allocation in s390 PCI to avoid out of bounds access
   when IOMMU pages aren't a multiple of 64

 - Fix kasan crashes when accessing DCSS mapping in memory holes by
   adding corresponding kasan zero shadow mappings

 - Fix a memory leak in css_alloc_subchannel in case
   dma_set_coherent_mask fails

* tag 's390-6.6-4' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
  s390/pci: fix iommu bitmap allocation
  s390/kasan: handle DCSS mapping in memory holes
  s390/cio: fix a memleak in css_alloc_subchannel

12 months agoMerge tag 'platform-drivers-x86-v6.6-5' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 21 Oct 2023 17:02:46 +0000 (10:02 -0700)]
Merge tag 'platform-drivers-x86-v6.6-5' of git://git./linux/kernel/git/pdx86/platform-drivers-x86

Pull x86 platform driver fixes from Hans de Goede:

 -  Fix spurious brightness down presses on newer Asus laptop models

 -  Fix backlight control not working on T2 Mac Pro all-in-ones

 -  Add Armin Wolf as new maintainer for the WMI bus driver and change
    its status from orphaned to maintained

 -  A few other small fixes

* tag 'platform-drivers-x86-v6.6-5' of git://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86:
  platform/mellanox: mlxbf-tmfifo: Fix a warning message
  apple-gmux: Hard Code max brightness for MMIO gmux
  platform/surface: platform_profile: Propagate error if profile registration fails
  platform/x86: asus-wmi: Map 0x2a code, Ignore 0x2b and 0x2c events
  platform/x86: asus-wmi: Only map brightness codes when using asus-wmi backlight control
  platform/x86: asus-wmi: Change ASUS_WMI_BRN_DOWN code from 0x20 to 0x2e
  platform/x86: wmi: Update MAINTAINERS entry
  platform/x86: msi-ec: Fix the 3rd config
  platform/x86: intel-uncore-freq: Conditionally create attribute for read frequency
  platform: mellanox: Fix a resource leak in an error handling path in probing flow

12 months agoMerge tag 'usb-6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Linus Torvalds [Sat, 21 Oct 2023 16:57:34 +0000 (09:57 -0700)]
Merge tag 'usb-6.6-rc7' of git://git./linux/kernel/git/gregkh/usb

Pull USB / Thunderbolt fixes and ids from Greg KH:
 "Here are four small patches for USB and Thunderbolt for 6.6-rc7 that
  do the following:

   - new usb-serial device ids

   - thunderbolt driver fix for reported issue

  All of these have been in linux-next with no reported problems"

* tag 'usb-6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
  USB: serial: option: add Fibocom to DELL custom modem FM101R-GL
  USB: serial: option: add entry for Sierra EM9191 with new firmware
  USB: serial: option: add Telit LE910C4-WWX 0x1035 composition
  thunderbolt: Call tb_switch_put() once DisplayPort bandwidth request is finished

12 months agoMerge tag 'v6.6-p5' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6
Linus Torvalds [Sat, 21 Oct 2023 16:49:13 +0000 (09:49 -0700)]
Merge tag 'v6.6-p5' of git://git./linux/kernel/git/herbert/crypto-2.6

Pull crypto fix from Herbert Xu:
 "Fix a 6.5 regression in crypto/asymmetric_keys"

* tag 'v6.6-p5' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  KEYS: asymmetric: Fix sign/verify on pkcs1pad without a hash

12 months agoMerge tag 'iomap-6.6-fixes-5' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux
Linus Torvalds [Sat, 21 Oct 2023 16:43:09 +0000 (09:43 -0700)]
Merge tag 'iomap-6.6-fixes-5' of git://git./fs/xfs/xfs-linux

Pull iomap fix from Darrick Wong:

 - Fix a bug where a writev consisting of a bunch of sub-fsblock writes
   where the last buffer address is invalid could lead to an infinite
   loop

* tag 'iomap-6.6-fixes-5' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
  iomap: fix short copy in iomap_write_iter()

12 months agoMerge tag 'perf-tools-fixes-for-v6.6-2-2023-10-20' of git://git.kernel.org/pub/scm...
Linus Torvalds [Fri, 20 Oct 2023 21:49:24 +0000 (14:49 -0700)]
Merge tag 'perf-tools-fixes-for-v6.6-2-2023-10-20' of git://git./linux/kernel/git/perf/perf-tools

Pull perf tools fixes from Arnaldo Carvalho de Melo:

 - Fix regression in reading scale and unit files from sysfs for PMU
   events, so that we can use that info to pretty print instead of
   printing raw numbers:

     # perf stat -e power/energy-ram/,power/energy-gpu/ sleep 2

      Performance counter stats for 'system wide':

                 1.64 Joules power/energy-ram/
                 0.20 Joules power/energy-gpu/

          2.001228914 seconds time elapsed
     #
     # grep -m1 "model name" /proc/cpuinfo
     model name : Intel(R) Core(TM) i7-8650U CPU @ 1.90GHz
     #

 - The small llvm.cpp file used to check if the llvm devel files are
   present was incorrectly deleted when removing the BPF event in 'perf
   trace', put it back as it is also used by tools/bpf/bpftool, that
   uses llvm routines to do disassembly of BPF object files.

 - Fix use of addr_location__exit() in dlfilter__object_code(), making
   sure that it is only used to pair a previous addr_location__init()
   call.

* tag 'perf-tools-fixes-for-v6.6-2-2023-10-20' of git://git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools:
  tools build: Fix llvm feature detection, still used by bpftool
  perf dlfilter: Add a test for object_code()
  perf dlfilter: Fix use of addr_location__exit() in dlfilter__object_code()
  perf pmu: Fix perf stat output with correct scale and unit

12 months agoMerge tag 'linux_kselftest_active-fixes-6.6-rc7' of git://git.kernel.org/pub/scm...
Linus Torvalds [Fri, 20 Oct 2023 21:45:41 +0000 (14:45 -0700)]
Merge tag 'linux_kselftest_active-fixes-6.6-rc7' of git://git./linux/kernel/git/shuah/linux-kselftest

Pull Kselftest fix from Shuah Khan:
 "One single fix to assert check in user_events abi_test to properly
  check bit value on Big Endian architectures. The code treated the bit
  values as Little Endian and the check failed on Big Endian"

* tag 'linux_kselftest_active-fixes-6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest:
  selftests/user_events: Fix abi_test for BE archs

12 months agoMerge tag 'nfs-for-6.6-4' of git://git.linux-nfs.org/projects/anna/linux-nfs
Linus Torvalds [Fri, 20 Oct 2023 21:04:53 +0000 (14:04 -0700)]
Merge tag 'nfs-for-6.6-4' of git://git.linux-nfs.org/projects/anna/linux-nfs

Pull NFS client fixes from Anna Schumaker:
 "Stable Fix:
   - Fix a pNFS hang in nfs4_evict_inode()

  Fixes:
   - Force update of suid/sgid bits after an NFS v4.2 ALLOCATE op
   - Fix a potential oops in nfs_inode_remove_request()
   - Check the validity of the layout pointer in ff_layout_mirror_prepare_stats()
   - Fix incorrectly marking the pNFS MDS with USE_PNFS_DS in some cases"

* tag 'nfs-for-6.6-4' of git://git.linux-nfs.org/projects/anna/linux-nfs:
  NFSv4.1: fixup use EXCHGID4_FLAG_USE_PNFS_DS for DS server
  pNFS/flexfiles: Check the layout validity in ff_layout_mirror_prepare_stats
  pNFS: Fix a hang in nfs4_evict_inode()
  NFS: Fix potential oops in nfs_inode_remove_request()
  nfs42: client needs to strip file mode's suid/sgid bit after ALLOCATE op

12 months agoMerge tag 'fsnotify_for_v6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 20 Oct 2023 21:00:05 +0000 (14:00 -0700)]
Merge tag 'fsnotify_for_v6.6-rc7' of git://git./linux/kernel/git/jack/linux-fs

Pull fanotify fix from Jan Kara:
 "Disable superblock / mount marks for filesystems that can encode file
  handles but not open them (currently only overlayfs).

  It is not clear the functionality is useful in any way so let's better
  disable it before someone comes up with some creative misuse"

* tag 'fsnotify_for_v6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs:
  fanotify: limit reporting of event with non-decodeable file handles

12 months agoMerge tag 'acpi-6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Fri, 20 Oct 2023 20:47:05 +0000 (13:47 -0700)]
Merge tag 'acpi-6.6-rc7' of git://git./linux/kernel/git/rafael/linux-pm

Pull ACPI fixes from Rafael Wysocki:
 "These fix the ACPI initialization ordering on ARM and ACPI IRQ
  management in the cases when irq_create_fwspec_mapping() fails.

  Specifics:

   - Fix ACPI initialization ordering on ARM that was changed
     incorrectly during the 6.5 development cycle (Hanjun Guo)

   - Make acpi_register_gsi() return an error code as appropriate when
     irq_create_fwspec_mapping() returns 0 on failure (Sunil V L)"

* tag 'acpi-6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  ACPI: bus: Move acpi_arm_init() to the place of after acpi_ghes_init()
  ACPI: irq: Fix incorrect return value in acpi_register_gsi()

12 months agoMerge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Linus Torvalds [Fri, 20 Oct 2023 20:24:50 +0000 (13:24 -0700)]
Merge tag 'scsi-fixes' of git://git./linux/kernel/git/jejb/scsi

Pull SCSI fixes from James Bottomley:
 "Two small fixes, both in drivers.

  The mptsas one is really fixing an error path issue where it can leave
  the misc driver loaded even though the sas driver fails to initialize"

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  scsi: qla2xxx: Fix double free of dsd_list during driver load
  scsi: mpt3sas: Fix in error path

12 months agoMerge tag 'pinctrl-v6.6-3' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw...
Linus Torvalds [Fri, 20 Oct 2023 20:21:46 +0000 (13:21 -0700)]
Merge tag 'pinctrl-v6.6-3' of git://git./linux/kernel/git/linusw/linux-pinctrl

Pull pin control fixes from Linus Walleij:

 - Concurrent register updates in the Qualcomm LPASS pin controller gets
   a proper lock.

 - revert a mutex fix that was causing problems: contention on the mutex
   or something of the sort lead to probe reordering and MMC block
   devices start to register in a different order, which unsuspecting
   userspace is not ready to handle

* tag 'pinctrl-v6.6-3' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl:
  Revert "pinctrl: avoid unsafe code pattern in find_pinctrl()"
  pinctrl: qcom: lpass-lpi: fix concurrent register updates

12 months agoMerge tag 'mtd/fixes-for-6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 20 Oct 2023 20:12:34 +0000 (13:12 -0700)]
Merge tag 'mtd/fixes-for-6.6-rc7' of git://git./linux/kernel/git/mtd/linux

Pull MTD fixes from Miquel Raynal:
 "In the raw NAND subsystem, the major fix prevents using cached reads
  with devices not supporting it. There was two bug reports about this.

  Apart from that, three drivers (pl353, arasan and marvell) could
  sometimes hide page program failures due to their their own program
  page helper not being fully compliant with the specification (many
  drivers use the default helpers shared by the core). Adding a missing
  check prevents these situation.

  Finally, the Qualcomm driver had a broken error path.

  In the SPI-NAND subsystem one Micron device used a wrong bitmak
  reporting possibly corrupted ECC status.

  Finally, the physmap-core got stripped from its map_rom fallback by
  mistake, this feature is added back"

* tag 'mtd/fixes-for-6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux:
  mtd: rawnand: Ensure the nand chip supports cached reads
  mtd: rawnand: qcom: Unmap the right resource upon probe failure
  mtd: rawnand: pl353: Ensure program page operations are successful
  mtd: rawnand: arasan: Ensure program page operations are successful
  mtd: spinand: micron: correct bitmask for ecc status
  mtd: physmap-core: Restore map_rom fallback
  mtd: rawnand: marvell: Ensure program page operations are successful

12 months agoMerge tag 'mmc-v6.6-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/ulfh/mmc
Linus Torvalds [Fri, 20 Oct 2023 20:09:19 +0000 (13:09 -0700)]
Merge tag 'mmc-v6.6-rc3' of git://git./linux/kernel/git/ulfh/mmc

Pull MMC fixes from Ulf Hansson:
 "MMC core:
   - Capture correct oemid-bits for eMMC cards
   - Fix error propagation for some ioctl commands
   - Hold retuning if SDIO is in 1-bit mode

  MMC host:
   - mtk-sd: Use readl_poll_timeout_atomic to not "schedule while atomic"
   - sdhci-msm: Correct minimum number of clocks
   - sdhci-pci-gli: Fix LPM negotiation so x86/S0ix SoCs can suspend
   - sdhci-sprd: Fix error code in sdhci_sprd_tuning()"

* tag 'mmc-v6.6-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/ulfh/mmc:
  mmc: core: Capture correct oemid-bits for eMMC cards
  mmc: mtk-sd: Use readl_poll_timeout_atomic in msdc_reset_hw
  mmc: core: Fix error propagation for some ioctl commands
  mmc: sdhci-sprd: Fix error code in sdhci_sprd_tuning()
  mmc: sdhci-pci-gli: fix LPM negotiation so x86/S0ix SoCs can suspend
  mmc: core: sdio: hold retuning if sdio in 1-bit mode
  dt-bindings: mmc: sdhci-msm: correct minimum number of clocks

12 months agoMerge tag 'block-6.6-2023-10-20' of git://git.kernel.dk/linux
Linus Torvalds [Fri, 20 Oct 2023 17:31:06 +0000 (10:31 -0700)]
Merge tag 'block-6.6-2023-10-20' of git://git.kernel.dk/linux

Pull block fixes from Jens Axboe:
 "A fix for a regression with sed-opal and saved keys, and outside of
  that an NVMe pull request fixing a few minor issues on that front"

* tag 'block-6.6-2023-10-20' of git://git.kernel.dk/linux:
  nvme-pci: add BOGUS_NID for Intel 0a54 device
  nvmet-auth: complete a request only after freeing the dhchap pointers
  nvme: sanitize metadata bounce buffer for reads
  block: Fix regression in sed-opal for a saved key.
  nvme-auth: use chap->s2 to indicate bidirectional authentication
  nvmet-tcp: Fix a possible UAF in queue intialization setup
  nvme-rdma: do not try to stop unallocated queues

12 months agoMerge tag 'io_uring-6.6-2023-10-20' of git://git.kernel.dk/linux
Linus Torvalds [Fri, 20 Oct 2023 17:28:46 +0000 (10:28 -0700)]
Merge tag 'io_uring-6.6-2023-10-20' of git://git.kernel.dk/linux

Pull io_uring fix from Jens Axboe:
 "Just a single fix for a bug report that came in, fixing a case where
  failure to init a ring with IORING_SETUP_NO_MMAP can trigger a NULL
  pointer dereference"

* tag 'io_uring-6.6-2023-10-20' of git://git.kernel.dk/linux:
  io_uring: fix crash with IORING_SETUP_NO_MMAP and invalid SQ ring address

12 months agoMerge tag 'sound-6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai...
Linus Torvalds [Fri, 20 Oct 2023 17:05:10 +0000 (10:05 -0700)]
Merge tag 'sound-6.6-rc7' of git://git./linux/kernel/git/tiwai/sound

Pull sound fixes from Takashi Iwai:
 "Still higher volume than wished, but all are driver-specific small
  fixes and look safe for this late RC.

  The majority of changes are for ASoC, especially for wcd938x driver
  and Cirrus codec drivers, while there are other random fixes including
  usual HD-audio quirks"

* tag 'sound-6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound: (22 commits)
  ASoC: da7219: Correct the process of setting up Gnd switch in AAD
  ALSA: hda/realtek - Fixed ASUS platform headset Mic issue
  ALSA: hda/realtek: Add quirk for ASUS ROG GU603ZV
  ALSA: hda/relatek: Enable Mute LED on HP Laptop 15s-fq5xxx
  ASoC: dwc: Fix non-DT instantiation
  ASoC: codecs: tas2780: Fix log of failed reset via I2C.
  ASoC: rt5650: fix the wrong result of key button
  ASoC: cs42l42: Fix missing include of gpio/consumer.h
  ASoC: cs42l43: Update values for bias sense
  ASoC: dt-bindings: cirrus,cs42l43: Update values for bias sense
  ASoC: cs35l56: ASP1 DOUT must default to Hi-Z when not transmitting
  ASoC: pxa: fix a memory leak in probe()
  ASoC: cs35l56: Fix illegal use of init_completion()
  ASoC: codecs: wcd938x-sdw: fix runtime PM imbalance on probe errors
  ASoC: codecs: wcd938x-sdw: fix use after free on driver unbind
  ASoC: codecs: wcd938x: fix runtime PM imbalance on remove
  ASoC: codecs: wcd938x: fix regulator leaks on probe errors
  ASoC: codecs: wcd938x: fix resource leaks on bind errors
  ASoC: codecs: wcd938x: fix unbind tear down order
  ASoC: codecs: wcd938x: drop bogus bind error handling
  ...

12 months agoMerge tag 'drm-fixes-2023-10-20' of git://anongit.freedesktop.org/drm/drm
Linus Torvalds [Fri, 20 Oct 2023 16:55:31 +0000 (09:55 -0700)]
Merge tag 'drm-fixes-2023-10-20' of git://anongit.freedesktop.org/drm/drm

Pull drm fixes from Dave Airlie:
 "Regular fixes for the week, amdgpu, i915, nouveau, with some other
  scattered around, nothing major.

  amdgpu:
   - Fix possible NULL pointer dereference
   - Avoid possible BUG_ON in GPUVM updates
   - Disable AMD_CTX_PRIORITY_UNSET

  i915:
   - Fix display issue that was blocking S0ix
   - Retry gtt fault when out of fence registers

  bridge:
   - ti-sn65dsi86: Fix device lifetime

  edid:
   - Add quirk for BenQ GW2765

  ivpu:
   - Extend address range for MMU mmap

  nouveau:
   - DP-connector fixes
   - Documentation fixes

  panel:
   - Move AUX B116XW03 into panel-simple

  scheduler:
   - Eliminate DRM_SCHED_PRIORITY_UNSET

  ttm:
   - Fix possible NULL-ptr deref in cleanup

  mediatek:
   - Correctly free sg_table in gem prime vmap"

* tag 'drm-fixes-2023-10-20' of git://anongit.freedesktop.org/drm/drm:
  drm/amdgpu: Reserve fences for VM update
  drm/amdgpu: Fix possible null pointer dereference
  accel/ivpu: Extend address range for MMU mmap
  Revert "accel/ivpu: Use cached buffers for FW loading"
  accel/ivpu: Don't enter d0i3 during FLR
  drm/i915: Retry gtt fault when out of fence registers
  drm/i915/cx0: Only clear/set the Pipe Reset bit of the PHY Lanes Owned
  gpu/drm: Eliminate DRM_SCHED_PRIORITY_UNSET
  drm/amdgpu: Unset context priority is now invalid
  drm/mediatek: Correctly free sg_table in gem prime vmap
  drm/edid: add 8 bpc quirk to the BenQ GW2765
  drm/ttm: Reorder sys manager cleanup step
  drm/nouveau/disp: fix DP capable DSM connectors
  drm/nouveau: exec: fix ioctl kernel-doc warning
  drm/panel: Move AUX B116XW03 out of panel-edp back to panel-simple
  drm/bridge: ti-sn65dsi86: Associate DSI device lifetime with auxiliary device

12 months agoMerge branch 'acpi-irq'
Rafael J. Wysocki [Fri, 20 Oct 2023 15:31:15 +0000 (17:31 +0200)]
Merge branch 'acpi-irq'

Merge ACPI IRQ management fix for 6.6-rc7 (Sunil V L).

* acpi-irq:
  ACPI: irq: Fix incorrect return value in acpi_register_gsi()

12 months agoselftests/ftrace: Add new test case which checks non unique symbol
Francis Laniel [Fri, 20 Oct 2023 10:42:50 +0000 (13:42 +0300)]
selftests/ftrace: Add new test case which checks non unique symbol

If name_show() is non unique, this test will try to install a kprobe on this
function which should fail returning EADDRNOTAVAIL.
On kernel where name_show() is not unique, this test is skipped.

Link: https://lore.kernel.org/all/20231020104250.9537-3-flaniel@linux.microsoft.com/
Cc: stable@vger.kernel.org
Signed-off-by: Francis Laniel <flaniel@linux.microsoft.com>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
12 months agotracing/kprobes: Return EADDRNOTAVAIL when func matches several symbols
Francis Laniel [Fri, 20 Oct 2023 10:42:49 +0000 (13:42 +0300)]
tracing/kprobes: Return EADDRNOTAVAIL when func matches several symbols

When a kprobe is attached to a function that's name is not unique (is
static and shares the name with other functions in the kernel), the
kprobe is attached to the first function it finds. This is a bug as the
function that it is attaching to is not necessarily the one that the
user wants to attach to.

Instead of blindly picking a function to attach to what is ambiguous,
error with EADDRNOTAVAIL to let the user know that this function is not
unique, and that the user must use another unique function with an
address offset to get to the function they want to attach to.

Link: https://lore.kernel.org/all/20231020104250.9537-2-flaniel@linux.microsoft.com/
Cc: stable@vger.kernel.org
Fixes: 413d37d1eb69 ("tracing: Add kprobe-based event tracer")
Suggested-by: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Francis Laniel <flaniel@linux.microsoft.com>
Link: https://lore.kernel.org/lkml/20230819101105.b0c104ae4494a7d1f2eea742@kernel.org/
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
12 months agoMerge tag 'usb-serial-6.6-rc7' of https://git.kernel.org/pub/scm/linux/kernel/git...
Greg Kroah-Hartman [Fri, 20 Oct 2023 05:52:44 +0000 (07:52 +0200)]
Merge tag 'usb-serial-6.6-rc7' of https://git./linux/kernel/git/johan/usb-serial into usb-linus

Johan writes:

USB-serial device ids for 6.6-rc7

Here are some new modem device ids, including an entry needed for Sierra
EM9191 which stopped working with recent firmware.

All have been in linux-next with no reported issues.

* tag 'usb-serial-6.6-rc7' of https://git.kernel.org/pub/scm/linux/kernel/git/johan/usb-serial:
  USB: serial: option: add Fibocom to DELL custom modem FM101R-GL
  USB: serial: option: add entry for Sierra EM9191 with new firmware
  USB: serial: option: add Telit LE910C4-WWX 0x1035 composition

12 months agoMerge tag 'mediatek-drm-fixes-20231017' of https://git.kernel.org/pub/scm/linux/kerne...
Dave Airlie [Fri, 20 Oct 2023 04:23:25 +0000 (14:23 +1000)]
Merge tag 'mediatek-drm-fixes-20231017' of https://git./linux/kernel/git/chunkuang.hu/linux into drm-fixes

Mediatek DRM Fixes - 20231017

1. Correctly free sg_table in gem prime vmap

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20231016233659.3639-1-chunkuang.hu@kernel.org
12 months agoMerge tag 'drm-intel-fixes-2023-10-19' of git://anongit.freedesktop.org/drm/drm-intel...
Dave Airlie [Fri, 20 Oct 2023 04:21:16 +0000 (14:21 +1000)]
Merge tag 'drm-intel-fixes-2023-10-19' of git://anongit.freedesktop.org/drm/drm-intel into drm-fixes

- Fix display issue that was blocking S0ix (Khaled)
- Retry gtt fault when out of fence registers (Ville)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/ZTFXbo6M5bWp/hTU@intel.com
12 months agoMerge tag 'drm-misc-fixes-2023-10-19' of git://anongit.freedesktop.org/drm/drm-misc...
Dave Airlie [Fri, 20 Oct 2023 04:07:41 +0000 (14:07 +1000)]
Merge tag 'drm-misc-fixes-2023-10-19' of git://anongit.freedesktop.org/drm/drm-misc into drm-fixes

Short summary of fixes pull:

amdgpu:
- Disable AMD_CTX_PRIORITY_UNSET

bridge:
- ti-sn65dsi86: Fix device lifetime

edid:
- Add quirk for BenQ GW2765

ivpu:
- Extend address range for MMU mmap

nouveau:
- DP-connector fixes
- Documentation fixes

panel:
- Move AUX B116XW03 into panel-simple

scheduler:
- Eliminate DRM_SCHED_PRIORITY_UNSET

ttm:
- Fix possible NULL-ptr deref in cleanup

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20231019114605.GA22540@linux-uq9g
12 months agoMerge tag 'amd-drm-fixes-6.6-2023-10-19' of https://gitlab.freedesktop.org/agd5f...
Dave Airlie [Fri, 20 Oct 2023 03:42:01 +0000 (13:42 +1000)]
Merge tag 'amd-drm-fixes-6.6-2023-10-19' of https://gitlab.freedesktop.org/agd5f/linux into drm-fixes

amd-drm-fixes-6.6-2023-10-19:

amdgpu:
- Fix possible NULL pointer dereference
- Avoid possible BUG_ON in GPUVM updates

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231020012417.4876-1-alexander.deucher@amd.com
12 months agotools build: Fix llvm feature detection, still used by bpftool
Arnaldo Carvalho de Melo [Thu, 19 Oct 2023 21:09:33 +0000 (18:09 -0300)]
tools build: Fix llvm feature detection, still used by bpftool

When removing the BPF event for perf a feature test that checks if the
llvm devel files are availabe was removed but that is also used by
bpftool.

bpftool uses it to decide what kind of disassembly it will use: llvm or
binutils based.

Removing the tools/build/feature/test-llvm.cpp file made bpftool to
always fallback to binutils disassembly, even with the llvm devel files
installed, fix it by restoring just that small test-llvm.cpp test file.

Fixes: 56b11a2126bf2f42 ("perf bpf: Remove support for embedding clang for compiling BPF events (-e foo.c)")
Reported-by: Manu Bretelle <chantr4@gmail.com>
Reviewed-by: Ian Rogers <irogers@google.com>
Reviewed-by: Manu Bretelle <chantr4@gmail.com>
Acked-by: Quentin Monnet <quentin@isovalent.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Andrii Nakryiko <andrii@kernel.org>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Carsten Haitzler <carsten.haitzler@arm.com>
Cc: Eduard Zingerman <eddyz87@gmail.com>
Cc: Fangrui Song <maskray@google.com>
Cc: He Kuang <hekuang@huawei.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Cc: Tiezhu Yang <yangtiezhu@loongson.cn>
Cc: Tom Rix <trix@redhat.com>
Cc: Wang Nan <wangnan0@huawei.com>
Cc: Wang ShaoBo <bobo.shaobowang@huawei.com>
Cc: Yang Jihong <yangjihong1@huawei.com>
Cc: Yonghong Song <yhs@fb.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Link: https://lore.kernel.org/lkml/ZTGa0Ukt7QyxWcVy@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
12 months agoMerge tag 'sev_fixes_for_v6.6' of //git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Linus Torvalds [Fri, 20 Oct 2023 01:12:08 +0000 (18:12 -0700)]
Merge tag 'sev_fixes_for_v6.6' of //git./linux/kernel/git/tip/tip

Pull x86 fixes from Borislav Petkov:
 "Take care of a race between when the #VC exception is raised and when
  the guest kernel gets to emulate certain instructions in SEV-{ES,SNP}
  guests by:

   - disabling emulation of MMIO instructions when coming from user mode

   - checking the IO permission bitmap before emulating IO instructions
     and verifying the memory operands of INS/OUTS insns"

* tag 'sev_fixes_for_v6.6' of //git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/sev: Check for user-space IOIO pointing to kernel space
  x86/sev: Check IOBM for IOIO exceptions from user-space
  x86/sev: Disable MMIO emulation from user mode

12 months agodrm/amdgpu: Reserve fences for VM update
Felix Kuehling [Mon, 17 Jul 2023 19:28:52 +0000 (15:28 -0400)]
drm/amdgpu: Reserve fences for VM update

In amdgpu_dma_buf_move_notify reserve fences for the page table updates
in amdgpu_vm_clear_freed and amdgpu_vm_handle_moved. This fixes a BUG_ON
in dma_resv_add_fence when using SDMA for page table updates.

Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
12 months agodrm/amdgpu: Fix possible null pointer dereference
Felix Kuehling [Tue, 17 Oct 2023 20:51:03 +0000 (16:51 -0400)]
drm/amdgpu: Fix possible null pointer dereference

abo->tbo.resource may be NULL in amdgpu_vm_bo_update.

Fixes: 180253782038 ("drm/ttm: stop allocating dummy resources during BO creation")
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
12 months agoMerge tag 'net-6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Linus Torvalds [Thu, 19 Oct 2023 19:08:18 +0000 (12:08 -0700)]
Merge tag 'net-6.6-rc7' of git://git./linux/kernel/git/netdev/net

Pull networking fixes from Jakub Kicinski:
 "Including fixes from bluetooth, netfilter, WiFi.

  Feels like an up-tick in regression fixes, mostly for older releases.
  The hfsc fix, tcp_disconnect() and Intel WWAN fixes stand out as
  fairly clear-cut user reported regressions. The mlx5 DMA bug was
  causing strife for 390x folks. The fixes themselves are not
  particularly scary, tho. No open investigations / outstanding reports
  at the time of writing.

  Current release - regressions:

   - eth: mlx5: perform DMA operations in the right locations, make
     devices usable on s390x, again

   - sched: sch_hfsc: upgrade 'rt' to 'sc' when it becomes a inner
     curve, previous fix of rejecting invalid config broke some scripts

   - rfkill: reduce data->mtx scope in rfkill_fop_open, avoid deadlock

   - revert "ethtool: Fix mod state of verbose no_mask bitset", needs
     more work

  Current release - new code bugs:

   - tcp: fix listen() warning with v4-mapped-v6 address

  Previous releases - regressions:

   - tcp: allow tcp_disconnect() again when threads are waiting, it was
     denied to plug a constant source of bugs but turns out .NET depends
     on it

   - eth: mlx5: fix double-free if buffer refill fails under OOM

   - revert "net: wwan: iosm: enable runtime pm support for 7560", it's
     causing regressions and the WWAN team at Intel disappeared

   - tcp: tsq: relax tcp_small_queue_check() when rtx queue contains a
     single skb, fix single-stream perf regression on some devices

  Previous releases - always broken:

   - Bluetooth:
      - fix issues in legacy BR/EDR PIN code pairing
      - correctly bounds check and pad HCI_MON_NEW_INDEX name

   - netfilter:
      - more fixes / follow ups for the large "commit protocol" rework,
        which went in as a fix to 6.5
      - fix null-derefs on netlink attrs which user may not pass in

   - tcp: fix excessive TLP and RACK timeouts from HZ rounding (bless
     Debian for keeping HZ=250 alive)

   - net: more strict VIRTIO_NET_HDR_GSO_UDP_L4 validation, prevent
     letting frankenstein UDP super-frames from getting into the stack

   - net: fix interface altnames when ifc moves to a new namespace

   - eth: qed: fix the size of the RX buffers

   - mptcp: avoid sending RST when closing the initial subflow"

* tag 'net-6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (94 commits)
  Revert "ethtool: Fix mod state of verbose no_mask bitset"
  selftests: mptcp: join: no RST when rm subflow/addr
  mptcp: avoid sending RST when closing the initial subflow
  mptcp: more conservative check for zero probes
  tcp: check mptcp-level constraints for backlog coalescing
  selftests: mptcp: join: correctly check for no RST
  net: ti: icssg-prueth: Fix r30 CMDs bitmasks
  selftests: net: add very basic test for netdev names and namespaces
  net: move altnames together with the netdevice
  net: avoid UAF on deleted altname
  net: check for altname conflicts when changing netdev's netns
  net: fix ifname in netlink ntf during netns move
  net: ethernet: ti: Fix mixed module-builtin object
  net: phy: bcm7xxx: Add missing 16nm EPHY statistics
  ipv4: fib: annotate races around nh->nh_saddr_genid and nh->nh_saddr
  tcp_bpf: properly release resources on error paths
  net/sched: sch_hfsc: upgrade 'rt' to 'sc' when it becomes a inner curve
  net: mdio-mux: fix C45 access returning -EIO after API change
  tcp: tsq: relax tcp_small_queue_check() when rtx queue contains a single skb
  octeon_ep: update BQL sent bytes before ringing doorbell
  ...

12 months agoMerge tag 'loongarch-fixes-6.6-3' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Thu, 19 Oct 2023 18:02:28 +0000 (11:02 -0700)]
Merge tag 'loongarch-fixes-6.6-3' of git://git./linux/kernel/git/chenhuacai/linux-loongson

Pull LoongArch fixes from Huacai ChenL
 "Fix 4-level pagetable building, disable WUC for pgprot_writecombine()
  like ioremap_wc(), use correct annotation for exception handlers, and
  a trivial cleanup"

* tag 'loongarch-fixes-6.6-3' of git://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai/linux-loongson:
  LoongArch: Disable WUC for pgprot_writecombine() like ioremap_wc()
  LoongArch: Replace kmap_atomic() with kmap_local_page() in copy_user_highpage()
  LoongArch: Export symbol invalid_pud_table for modules building
  LoongArch: Use SYM_CODE_* to annotate exception handlers

12 months agoMerge tag 'slab-fixes-for-6.6-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Thu, 19 Oct 2023 17:53:31 +0000 (10:53 -0700)]
Merge tag 'slab-fixes-for-6.6-rc6' of git://git./linux/kernel/git/vbabka/slab

Pull slab fix from Vlastimil Babka:

 - stable fix to prevent kernel warnings with KASAN_HW_TAGS on arm64
   due to improperly resolved kmalloc alignment restrictions (Catalin
   Marinas)

* tag 'slab-fixes-for-6.6-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/vbabka/slab:
  mm: slab: Do not create kmalloc caches smaller than arch_slab_minalign()

12 months agoMerge tag 'seccomp-v6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/kees...
Linus Torvalds [Thu, 19 Oct 2023 17:10:14 +0000 (10:10 -0700)]
Merge tag 'seccomp-v6.6-rc7' of git://git./linux/kernel/git/kees/linux

Pull seccomp fix from Kees Cook:

 - Fix seccomp_unotify perf benchmark for 32-bit (Jiri Slaby)

* tag 'seccomp-v6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux:
  perf/benchmark: fix seccomp_unotify benchmark for 32-bit

12 months agoiomap: fix short copy in iomap_write_iter()
Jan Stancek [Thu, 19 Oct 2023 16:41:36 +0000 (09:41 -0700)]
iomap: fix short copy in iomap_write_iter()

Starting with commit 5d8edfb900d5 ("iomap: Copy larger chunks from
userspace"), iomap_write_iter() can get into endless loop. This can
be reproduced with LTP writev07 which uses partially valid iovecs:
        struct iovec wr_iovec[] = {
                { buffer, 64 },
                { bad_addr, 64 },
                { buffer + 64, 64 },
                { buffer + 64 * 2, 64 },
        };

commit bc1bb416bbb9 ("generic_perform_write()/iomap_write_actor():
saner logics for short copy") previously introduced the logic, which
made short copy retry in next iteration with amount of "bytes" it
managed to copy:

                if (unlikely(status == 0)) {
                        /*
                         * A short copy made iomap_write_end() reject the
                         * thing entirely.  Might be memory poisoning
                         * halfway through, might be a race with munmap,
                         * might be severe memory pressure.
                         */
                        if (copied)
                                bytes = copied;

However, since 5d8edfb900d5 "bytes" is no longer carried into next
iteration, because it is now always initialized at the beginning of
the loop. And for iov_iter_count < PAGE_SIZE, "bytes" ends up with
same value as previous iteration, making the loop retry same copy
over and over, which leads to writev07 testcase hanging.

Make next iteration retry with amount of bytes we managed to copy.

Fixes: 5d8edfb900d5 ("iomap: Copy larger chunks from userspace")
Signed-off-by: Jan Stancek <jstancek@redhat.com>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
12 months agoMerge tag 'v6.6-rc7.vfs.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs
Linus Torvalds [Thu, 19 Oct 2023 16:37:41 +0000 (09:37 -0700)]
Merge tag 'v6.6-rc7.vfs.fixes' of git://git./linux/kernel/git/vfs/vfs

Pull vfs fix from Christian Brauner:
 "An openat() call from io_uring triggering an audit call can apparently
  cause the refcount of struct filename to be incremented from multiple
  threads concurrently during async execution, triggering a refcount
  underflow and hitting a BUG_ON(). That bug has been lurking around
  since at least v5.16 apparently.

  Switch to an atomic counter to fix that. The underflow check is
  downgraded from a BUG_ON() to a WARN_ON_ONCE() but we could easily
  remove that check altogether tbh"

* tag 'v6.6-rc7.vfs.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs:
  audit,io_uring: io_uring openat triggers audit reference count underflow

12 months agoRevert "ethtool: Fix mod state of verbose no_mask bitset"
Kory Maincent [Thu, 19 Oct 2023 13:16:41 +0000 (15:16 +0200)]
Revert "ethtool: Fix mod state of verbose no_mask bitset"

This reverts commit 108a36d07c01edbc5942d27c92494d1c6e4d45a0.

It was reported that this fix breaks the possibility to remove existing WoL
flags. For example:
~$ ethtool lan2
...
        Supports Wake-on: pg
        Wake-on: d
...
~$ ethtool -s lan2 wol gp
~$ ethtool lan2
...
        Wake-on: pg
...
~$ ethtool -s lan2 wol d
~$ ethtool lan2
...
        Wake-on: pg
...

This worked correctly before this commit because we were always updating
a zero bitmap (since commit 6699170376ab ("ethtool: fix application of
verbose no_mask bitset"), that is) so that the rest was left zero
naturally. But now the 1->0 change (old_val is true, bit not present in
netlink nest) no longer works.

Reported-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reported-by: Michal Kubecek <mkubecek@suse.cz>
Closes: https://lore.kernel.org/netdev/20231019095140.l6fffnszraeb6iiw@lion.mk-sys.cz/
Cc: stable@vger.kernel.org
Fixes: 108a36d07c01 ("ethtool: Fix mod state of verbose no_mask bitset")
Signed-off-by: Kory Maincent <kory.maincent@bootlin.com>
Reviewed-by: Michal Kubecek <mkubecek@suse.cz>
Link: https://lore.kernel.org/r/20231019-feature_ptp_bitset_fix-v1-1-70f3c429a221@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
12 months agoMerge tag 'ntfs3_for_6.6' of https://github.com/Paragon-Software-Group/linux-ntfs3
Linus Torvalds [Thu, 19 Oct 2023 16:10:18 +0000 (09:10 -0700)]
Merge tag 'ntfs3_for_6.6' of https://github.com/Paragon-Software-Group/linux-ntfs3

Pull ntfs3 fixes from Konstantin Komarov:

 - memory leak

 - some logic errors, NULL dereferences

 - some code was refactored

 - more sanity checks

* tag 'ntfs3_for_6.6' of https://github.com/Paragon-Software-Group/linux-ntfs3:
  fs/ntfs3: Avoid possible memory leak
  fs/ntfs3: Fix directory element type detection
  fs/ntfs3: Fix possible null-pointer dereference in hdr_find_e()
  fs/ntfs3: Fix OOB read in ntfs_init_from_boot
  fs/ntfs3: fix panic about slab-out-of-bounds caused by ntfs_list_ea()
  fs/ntfs3: Fix NULL pointer dereference on error in attr_allocate_frame()
  fs/ntfs3: Fix possible NULL-ptr-deref in ni_readpage_cmpr()
  fs/ntfs3: Do not allow to change label if volume is read-only
  fs/ntfs3: Add more info into /proc/fs/ntfs3/<dev>/volinfo
  fs/ntfs3: Refactoring and comments
  fs/ntfs3: Fix alternative boot searching
  fs/ntfs3: Allow repeated call to ntfs3_put_sbi
  fs/ntfs3: Use inode_set_ctime_to_ts instead of inode_set_ctime
  fs/ntfs3: Fix shift-out-of-bounds in ntfs_fill_super
  fs/ntfs3: fix deadlock in mark_as_free_ex
  fs/ntfs3: Add more attributes checks in mi_enum_attr()
  fs/ntfs3: Use kvmalloc instead of kmalloc(... __GFP_NOWARN)
  fs/ntfs3: Write immediately updated ntfs state
  fs/ntfs3: Add ckeck in ni_update_parent()

12 months agoMerge branch 'mptcp-fixes-for-v6-6'
Jakub Kicinski [Thu, 19 Oct 2023 16:10:02 +0000 (09:10 -0700)]
Merge branch 'mptcp-fixes-for-v6-6'

Mat Martineau says:

====================
mptcp: Fixes for v6.6

Patch 1 corrects the logic for MP_JOIN tests where 0 RSTs are expected.

Patch 2 ensures MPTCP packets are not incorrectly coalesced in the TCP
backlog queue.

Patch 3 avoids a zero-window probe and associated WARN_ON_ONCE() in an
expected MPTCP reinjection scenario.

Patches 4 & 5 allow an initial MPTCP subflow to be closed cleanly
instead of always sending RST. Associated selftest is updated.
====================

Link: https://lore.kernel.org/r/20231018-send-net-20231018-v1-0-17ecb002e41d@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
12 months agoselftests: mptcp: join: no RST when rm subflow/addr
Matthieu Baerts [Wed, 18 Oct 2023 18:23:56 +0000 (11:23 -0700)]
selftests: mptcp: join: no RST when rm subflow/addr

Recently, we noticed that some RST were wrongly generated when removing
the initial subflow.

This patch makes sure RST are not sent when removing any subflows or any
addresses.

Fixes: c2b2ae3925b6 ("mptcp: handle correctly disconnect() failures")
Cc: stable@vger.kernel.org
Acked-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Matthieu Baerts <matttbe@kernel.org>
Signed-off-by: Mat Martineau <martineau@kernel.org>
Link: https://lore.kernel.org/r/20231018-send-net-20231018-v1-5-17ecb002e41d@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
12 months agomptcp: avoid sending RST when closing the initial subflow
Geliang Tang [Wed, 18 Oct 2023 18:23:55 +0000 (11:23 -0700)]
mptcp: avoid sending RST when closing the initial subflow

When closing the first subflow, the MPTCP protocol unconditionally
calls tcp_disconnect(), which in turn generates a reset if the subflow
is established.

That is unexpected and different from what MPTCP does with MPJ
subflows, where resets are generated only on FASTCLOSE and other edge
scenarios.

We can't reuse for the first subflow the same code in place for MPJ
subflows, as MPTCP clean them up completely via a tcp_close() call,
while must keep the first subflow socket alive for later re-usage, due
to implementation constraints.

This patch adds a new helper __mptcp_subflow_disconnect() that
encapsulates, a logic similar to tcp_close, issuing a reset only when
the MPTCP_CF_FASTCLOSE flag is set, and performing a clean shutdown
otherwise.

Fixes: c2b2ae3925b6 ("mptcp: handle correctly disconnect() failures")
Cc: stable@vger.kernel.org
Reviewed-by: Matthieu Baerts <matttbe@kernel.org>
Co-developed-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Geliang Tang <geliang.tang@suse.com>
Signed-off-by: Mat Martineau <martineau@kernel.org>
Link: https://lore.kernel.org/r/20231018-send-net-20231018-v1-4-17ecb002e41d@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
12 months agomptcp: more conservative check for zero probes
Paolo Abeni [Wed, 18 Oct 2023 18:23:54 +0000 (11:23 -0700)]
mptcp: more conservative check for zero probes

Christoph reported that the MPTCP protocol can find the subflow-level
write queue unexpectedly not empty while crafting a zero-window probe,
hitting a warning:

------------[ cut here ]------------
WARNING: CPU: 0 PID: 188 at net/mptcp/protocol.c:1312 mptcp_sendmsg_frag+0xc06/0xe70
Modules linked in:
CPU: 0 PID: 188 Comm: kworker/0:2 Not tainted 6.6.0-rc2-g1176aa719d7a #47
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.11.0-2.el7 04/01/2014
Workqueue: events mptcp_worker
RIP: 0010:mptcp_sendmsg_frag+0xc06/0xe70 net/mptcp/protocol.c:1312
RAX: 47d0530de347ff6a RBX: 47d0530de347ff6b RCX: ffff8881015d3c00
RDX: ffff8881015d3c00 RSI: 47d0530de347ff6b RDI: 47d0530de347ff6b
RBP: 47d0530de347ff6b R08: ffffffff8243c6a8 R09: ffffffff82042d9c
R10: 0000000000000002 R11: ffffffff82056850 R12: ffff88812a13d580
R13: 0000000000000001 R14: ffff88812b375e50 R15: ffff88812bbf3200
FS:  0000000000000000(0000) GS:ffff88813bc00000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000695118 CR3: 0000000115dfc001 CR4: 0000000000170ef0
Call Trace:
 <TASK>
 __subflow_push_pending+0xa4/0x420 net/mptcp/protocol.c:1545
 __mptcp_push_pending+0x128/0x3b0 net/mptcp/protocol.c:1614
 mptcp_release_cb+0x218/0x5b0 net/mptcp/protocol.c:3391
 release_sock+0xf6/0x100 net/core/sock.c:3521
 mptcp_worker+0x6e8/0x8f0 net/mptcp/protocol.c:2746
 process_scheduled_works+0x341/0x690 kernel/workqueue.c:2630
 worker_thread+0x3a7/0x610 kernel/workqueue.c:2784
 kthread+0x143/0x180 kernel/kthread.c:388
 ret_from_fork+0x4d/0x60 arch/x86/kernel/process.c:147
 ret_from_fork_asm+0x1b/0x30 arch/x86/entry/entry_64.S:304
 </TASK>

The root cause of the issue is that expectations are wrong: e.g. due
to MPTCP-level re-injection we can hit the critical condition.

Explicitly avoid the zero-window probe when the subflow write queue
is not empty and drop the related warnings.

Reported-by: Christoph Paasch <cpaasch@apple.com>
Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/444
Fixes: f70cad1085d1 ("mptcp: stop relying on tcp_tx_skb_cache")
Cc: stable@vger.kernel.org
Reviewed-by: Mat Martineau <martineau@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <martineau@kernel.org>
Link: https://lore.kernel.org/r/20231018-send-net-20231018-v1-3-17ecb002e41d@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
12 months agotcp: check mptcp-level constraints for backlog coalescing
Paolo Abeni [Wed, 18 Oct 2023 18:23:53 +0000 (11:23 -0700)]
tcp: check mptcp-level constraints for backlog coalescing

The MPTCP protocol can acquire the subflow-level socket lock and
cause the tcp backlog usage. When inserting new skbs into the
backlog, the stack will try to coalesce them.

Currently, we have no check in place to ensure that such coalescing
will respect the MPTCP-level DSS, and that may cause data stream
corruption, as reported by Christoph.

Address the issue by adding the relevant admission check for coalescing
in tcp_add_backlog().

Note the issue is not easy to reproduce, as the MPTCP protocol tries
hard to avoid acquiring the subflow-level socket lock.

Fixes: 648ef4b88673 ("mptcp: Implement MPTCP receive path")
Cc: stable@vger.kernel.org
Reported-by: Christoph Paasch <cpaasch@apple.com>
Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/420
Reviewed-by: Mat Martineau <martineau@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <martineau@kernel.org>
Link: https://lore.kernel.org/r/20231018-send-net-20231018-v1-2-17ecb002e41d@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
12 months agoselftests: mptcp: join: correctly check for no RST
Matthieu Baerts [Wed, 18 Oct 2023 18:23:52 +0000 (11:23 -0700)]
selftests: mptcp: join: correctly check for no RST

The commit mentioned below was more tolerant with the number of RST seen
during a test because in some uncontrollable situations, multiple RST
can be generated.

But it was not taking into account the case where no RST are expected:
this validation was then no longer reporting issues for the 0 RST case
because it is not possible to have less than 0 RST in the counter. This
patch fixes the issue by adding a specific condition.

Fixes: 6bf41020b72b ("selftests: mptcp: update and extend fastclose test-cases")
Cc: stable@vger.kernel.org
Reviewed-by: Mat Martineau <martineau@kernel.org>
Signed-off-by: Matthieu Baerts <matttbe@kernel.org>
Signed-off-by: Mat Martineau <martineau@kernel.org>
Link: https://lore.kernel.org/r/20231018-send-net-20231018-v1-1-17ecb002e41d@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
12 months agonet: ti: icssg-prueth: Fix r30 CMDs bitmasks
MD Danish Anwar [Wed, 18 Oct 2023 15:07:15 +0000 (20:37 +0530)]
net: ti: icssg-prueth: Fix r30 CMDs bitmasks

The bitmasks for EMAC_PORT_DISABLE and EMAC_PORT_FORWARD r30 commands are
wrong in the driver.

Update the bitmasks of these commands to the correct ones as used by the
ICSSG firmware. These bitmasks are backwards compatible and work with
any ICSSG firmware version.

Fixes: e9b4ece7d74b ("net: ti: icssg-prueth: Add Firmware config and classification APIs.")
Signed-off-by: MD Danish Anwar <danishanwar@ti.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/20231018150715.3085380-1-danishanwar@ti.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
12 months agoMerge tag 'for-6.6-rc6-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave...
Linus Torvalds [Thu, 19 Oct 2023 15:56:01 +0000 (08:56 -0700)]
Merge tag 'for-6.6-rc6-tag' of git://git./linux/kernel/git/kdave/linux

Pull btrfs fix from David Sterba:
 "Fix a bug in chunk size decision that could lead to suboptimal
  placement and filling patterns"

* tag 'for-6.6-rc6-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
  btrfs: fix stripe length calculation for non-zoned data chunk allocation

12 months agos390/pci: fix iommu bitmap allocation
Niklas Schnelle [Tue, 17 Oct 2023 13:37:29 +0000 (15:37 +0200)]
s390/pci: fix iommu bitmap allocation

Since the fixed commits both zdev->iommu_bitmap and zdev->lazy_bitmap
are allocated as vzalloc(zdev->iommu_pages / 8). The problem is that
zdev->iommu_bitmap is a pointer to unsigned long but the above only
yields an allocation that is a multiple of sizeof(unsigned long) which
is 8 on s390x if the number of IOMMU pages is a multiple of 64.
This in turn is the case only if the effective IOMMU aperture is
a multiple of 64 * 4K = 256K. This is usually the case and so didn't
cause visible issues since both the virt_to_phys(high_memory) reduced
limit and hardware limits use nice numbers.

Under KVM, and in particular with QEMU limiting the IOMMU aperture to
the vfio DMA limit (default 65535), it is possible for the reported
aperture not to be a multiple of 256K however. In this case we end up
with an iommu_bitmap whose allocation is not a multiple of
8 causing bitmap operations to access it out of bounds.

Sadly we can't just fix this in the obvious way and use bitmap_zalloc()
because for large RAM systems (tested on 8 TiB) the zdev->iommu_bitmap
grows too large for kmalloc(). So add our own bitmap_vzalloc() wrapper.
This might be a candidate for common code, but this area of code will
be replaced by the upcoming conversion to use the common code DMA API on
s390 so just add a local routine.

Fixes: 224593215525 ("s390/pci: use virtual memory for iommu bitmap")
Fixes: 13954fd6913a ("s390/pci_dma: improve lazy flush for unmap")
Cc: stable@vger.kernel.org
Reviewed-by: Matthew Rosato <mjrosato@linux.ibm.com>
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
12 months agofanotify: limit reporting of event with non-decodeable file handles
Amir Goldstein [Wed, 18 Oct 2023 09:59:56 +0000 (12:59 +0300)]
fanotify: limit reporting of event with non-decodeable file handles

Commit a95aef69a740 ("fanotify: support reporting non-decodeable file
handles") merged in v6.5-rc1, added the ability to use an fanotify group
with FAN_REPORT_FID mode to watch filesystems that do not support nfs
export, but do know how to encode non-decodeable file handles, with the
newly introduced AT_HANDLE_FID flag.

At the time that this commit was merged, there were no filesystems
in-tree with those traits.

Commit 16aac5ad1fa9 ("ovl: support encoding non-decodable file handles"),
merged in v6.6-rc1, added this trait to overlayfs, thus allowing fanotify
watching of overlayfs with FAN_REPORT_FID mode.

In retrospect, allowing an fanotify filesystem/mount mark on such
filesystem in FAN_REPORT_FID mode will result in getting events with
file handles, without the ability to resolve the filesystem objects from
those file handles (i.e. no open_by_handle_at() support).

For v6.6, the safer option would be to allow this mode for inode marks
only, where the caller has the opportunity to use name_to_handle_at() at
the time of setting the mark. In the future we can revise this decision.

Fixes: a95aef69a740 ("fanotify: support reporting non-decodeable file handles")
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Message-Id: <20231018100000.2453965-2-amir73il@gmail.com>

12 months agoMerge branch 'net-fix-bugs-in-device-netns-move-and-rename'
Paolo Abeni [Thu, 19 Oct 2023 13:51:18 +0000 (15:51 +0200)]
Merge branch 'net-fix-bugs-in-device-netns-move-and-rename'

Jakub Kicinski says:

====================
net: fix bugs in device netns-move and rename

Daniel reported issues with the uevents generated during netdev
namespace move, if the netdev is getting renamed at the same time.

While the issue that he actually cares about is not fixed here,
there is a bunch of seemingly obvious other bugs in this code.
Fix the purely networking bugs while the discussion around
the uevent fix is still ongoing.
====================

Link: https://lore.kernel.org/r/20231018013817.2391509-1-kuba@kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
12 months agoselftests: net: add very basic test for netdev names and namespaces
Jakub Kicinski [Wed, 18 Oct 2023 01:38:17 +0000 (18:38 -0700)]
selftests: net: add very basic test for netdev names and namespaces

Add selftest for fixes around naming netdevs and namespaces.

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
12 months agonet: move altnames together with the netdevice
Jakub Kicinski [Wed, 18 Oct 2023 01:38:16 +0000 (18:38 -0700)]
net: move altnames together with the netdevice

The altname nodes are currently not moved to the new netns
when netdevice itself moves:

  [ ~]# ip netns add test
  [ ~]# ip -netns test link add name eth0 type dummy
  [ ~]# ip -netns test link property add dev eth0 altname some-name
  [ ~]# ip -netns test link show dev some-name
  2: eth0: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
      link/ether 1e:67:ed:19:3d:24 brd ff:ff:ff:ff:ff:ff
      altname some-name
  [ ~]# ip -netns test link set dev eth0 netns 1
  [ ~]# ip link
  ...
  3: eth0: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
      link/ether 02:40:88:62:ec:b8 brd ff:ff:ff:ff:ff:ff
      altname some-name
  [ ~]# ip li show dev some-name
  Device "some-name" does not exist.

Remove them from the hash table when device is unlisted
and add back when listed again.

Fixes: 36fbf1e52bd3 ("net: rtnetlink: add linkprop commands to add and delete alternative ifnames")
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
12 months agonet: avoid UAF on deleted altname
Jakub Kicinski [Wed, 18 Oct 2023 01:38:15 +0000 (18:38 -0700)]
net: avoid UAF on deleted altname

Altnames are accessed under RCU (dev_get_by_name_rcu())
but freed by kfree() with no synchronization point.

Each node has one or two allocations (node and a variable-size
name, sometimes the name is netdev->name). Adding rcu_heads
here is a bit tedious. Besides most code which unlists the names
already has rcu barriers - so take the simpler approach of adding
synchronize_rcu(). Note that the one on the unregistration path
(which matters more) is removed by the next fix.

Fixes: ff92741270bf ("net: introduce name_node struct to be used in hashlist")
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
12 months agonet: check for altname conflicts when changing netdev's netns
Jakub Kicinski [Wed, 18 Oct 2023 01:38:14 +0000 (18:38 -0700)]
net: check for altname conflicts when changing netdev's netns

It's currently possible to create an altname conflicting
with an altname or real name of another device by creating
it in another netns and moving it over:

 [ ~]$ ip link add dev eth0 type dummy

 [ ~]$ ip netns add test
 [ ~]$ ip -netns test link add dev ethX netns test type dummy
 [ ~]$ ip -netns test link property add dev ethX altname eth0
 [ ~]$ ip -netns test link set dev ethX netns 1

 [ ~]$ ip link
 ...
 3: eth0: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
     link/ether 02:40:88:62:ec:b8 brd ff:ff:ff:ff:ff:ff
 ...
 5: ethX: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
     link/ether 26:b7:28:78:38:0f brd ff:ff:ff:ff:ff:ff
     altname eth0

Create a macro for walking the altnames, this hopefully makes
it clearer that the list we walk contains only altnames.
Which is otherwise not entirely intuitive.

Fixes: 36fbf1e52bd3 ("net: rtnetlink: add linkprop commands to add and delete alternative ifnames")
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
12 months agonet: fix ifname in netlink ntf during netns move
Jakub Kicinski [Wed, 18 Oct 2023 01:38:13 +0000 (18:38 -0700)]
net: fix ifname in netlink ntf during netns move

dev_get_valid_name() overwrites the netdev's name on success.
This makes it hard to use in prepare-commit-like fashion,
where we do validation first, and "commit" to the change
later.

Factor out a helper which lets us save the new name to a buffer.
Use it to fix the problem of notification on netns move having
incorrect name:

 5: eth0: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN group default
     link/ether be:4d:58:f9:d5:40 brd ff:ff:ff:ff:ff:ff
 6: eth1: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN group default
     link/ether 1e:4a:34:36:e3:cd brd ff:ff:ff:ff:ff:ff

 [ ~]# ip link set dev eth0 netns 1 name eth1

ip monitor inside netns:
 Deleted inet eth0
 Deleted inet6 eth0
 Deleted 5: eth1: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN group default
     link/ether be:4d:58:f9:d5:40 brd ff:ff:ff:ff:ff:ff new-netnsid 0 new-ifindex 7

Name is reported as eth1 in old netns for ifindex 5, already renamed.

Fixes: d90310243fd7 ("net: device name allocation cleanups")
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
12 months agonet: ethernet: ti: Fix mixed module-builtin object
MD Danish Anwar [Wed, 18 Oct 2023 06:49:36 +0000 (12:19 +0530)]
net: ethernet: ti: Fix mixed module-builtin object

With CONFIG_TI_K3_AM65_CPSW_NUSS=y and CONFIG_TI_ICSSG_PRUETH=m,
k3-cppi-desc-pool.o is linked to a module and also to vmlinux even though
the expected CFLAGS are different between builtins and modules.

The build system is complaining about the following:

k3-cppi-desc-pool.o is added to multiple modules: icssg-prueth
ti-am65-cpsw-nuss

Introduce the new module, k3-cppi-desc-pool, to provide the common
functions to ti-am65-cpsw-nuss and icssg-prueth.

Fixes: 128d5874c082 ("net: ti: icssg-prueth: Add ICSSG ethernet driver")
Signed-off-by: MD Danish Anwar <danishanwar@ti.com>
Link: https://lore.kernel.org/r/20231018064936.3146846-1-danishanwar@ti.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
12 months agoRevert "pinctrl: avoid unsafe code pattern in find_pinctrl()"
Andy Shevchenko [Tue, 17 Oct 2023 14:18:06 +0000 (17:18 +0300)]
Revert "pinctrl: avoid unsafe code pattern in find_pinctrl()"

The commit breaks MMC enumeration on the Intel Merrifield
plaform.

Before:
[   36.439057] mmc0: SDHCI controller on PCI [0000:00:01.0] using ADMA
[   36.450924] mmc2: SDHCI controller on PCI [0000:00:01.3] using ADMA
[   36.459355] mmc1: SDHCI controller on PCI [0000:00:01.2] using ADMA
[   36.706399] mmc0: new DDR MMC card at address 0001
[   37.058972] mmc2: new ultra high speed DDR50 SDIO card at address 0001
[   37.278977] mmcblk0: mmc0:0001 H4G1d 3.64 GiB
[   37.297300]  mmcblk0: p1 p2 p3 p4 p5 p6 p7 p8 p9 p10

After:
[   36.436704] mmc2: SDHCI controller on PCI [0000:00:01.3] using ADMA
[   36.436720] mmc1: SDHCI controller on PCI [0000:00:01.0] using ADMA
[   36.463685] mmc0: SDHCI controller on PCI [0000:00:01.2] using ADMA
[   36.720627] mmc1: new DDR MMC card at address 0001
[   37.068181] mmc2: new ultra high speed DDR50 SDIO card at address 0001
[   37.279998] mmcblk1: mmc1:0001 H4G1d 3.64 GiB
[   37.302670]  mmcblk1: p1 p2 p3 p4 p5 p6 p7 p8 p9 p10

This reverts commit c153a4edff6ab01370fcac8e46f9c89cca1060c2.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20231017141806.535191-1-andriy.shevchenko@linux.intel.com
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
12 months agoperf: Disallow mis-matched inherited group reads
Peter Zijlstra [Wed, 18 Oct 2023 11:56:54 +0000 (13:56 +0200)]
perf: Disallow mis-matched inherited group reads

Because group consistency is non-atomic between parent (filedesc) and children
(inherited) events, it is possible for PERF_FORMAT_GROUP read() to try and sum
non-matching counter groups -- with non-sensical results.

Add group_generation to distinguish the case where a parent group removes and
adds an event and thus has the same number, but a different configuration of
events as inherited groups.

This became a problem when commit fa8c269353d5 ("perf/core: Invert
perf_read_group() loops") flipped the order of child_list and sibling_list.
Previously it would iterate the group (sibling_list) first, and for each
sibling traverse the child_list. In this order, only the group composition of
the parent is relevant. By flipping the order the group composition of the
child (inherited) events becomes an issue and the mis-match in group
composition becomes evident.

That said; even prior to this commit, while reading of a group that is not
equally inherited was not broken, it still made no sense.

(Ab)use ECHILD as error return to indicate issues with child process group
composition.

Fixes: fa8c269353d5 ("perf/core: Invert perf_read_group() loops")
Reported-by: Budimir Markovic <markovicbudimir@gmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20231018115654.GK33217@noisy.programming.kicks-ass.net
12 months agoaccel/ivpu: Extend address range for MMU mmap
Wludzik, Jozef [Wed, 18 Oct 2023 11:01:13 +0000 (13:01 +0200)]
accel/ivpu: Extend address range for MMU mmap

Allow to use whole address range in MMU context mmap which is up to 48
bits. Return invalid argument from MMU context mmap in case address is
not aligned to MMU page size, address is below MMU page size or address
is greater then 47 bits.

This fixes problem disallowing to run large models on VPU4

Signed-off-by: Wludzik, Jozef <jozef.wludzik@intel.com>
Reviewed-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231018110113.547208-1-stanislaw.gruszka@linux.intel.com
12 months agoRevert "accel/ivpu: Use cached buffers for FW loading"
Stanislaw Gruszka [Tue, 17 Oct 2023 12:13:53 +0000 (14:13 +0200)]
Revert "accel/ivpu: Use cached buffers for FW loading"

This reverts commit 645d694559cab36fe6a57c717efcfa27d9321396.

The commit cause issues with memory access from the device side.
Switch back to write-combined memory mappings until the issues
will be properly addressed.

Add extra wmb() needed when boot_params->save_restore_ret_address() is
modified.

Reviewed-by: Karol Wachowski <karol.wachowski@linux.intel.com>
Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231017121353.532466-1-stanislaw.gruszka@linux.intel.com
12 months agoaccel/ivpu: Don't enter d0i3 during FLR
Jacek Lawrynowicz [Tue, 3 Oct 2023 06:42:13 +0000 (08:42 +0200)]
accel/ivpu: Don't enter d0i3 during FLR

Avoid HW bug on some platforms where we enter D0i3 state
and CPU is in low power states (C8 or above).

Fixes: 852be13f3bd3 ("accel/ivpu: Add PM support")
Cc: stable@vger.kernel.org
Signed-off-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
Reviewed-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231003064213.1527327-1-stanislaw.gruszka@linux.intel.com
12 months agoMerge tag 'nf-23-10-18' of https://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf
Jakub Kicinski [Thu, 19 Oct 2023 01:17:50 +0000 (18:17 -0700)]
Merge tag 'nf-23-10-18' of https://git./linux/kernel/git/netfilter/nf

Florian Westphal says:

====================
netfilter: updates for net

First patch, from Phil Sutter, reduces number of audit notifications
when userspace requests to re-set stateful objects.
This change also comes with a selftest update.

Second patch, also from Phil, moves the nftables audit selftest
to its own netns to avoid interference with the init netns.

Third patch, from Pablo Neira, fixes an inconsistency with the "rbtree"
set backend: When set element X has expired, a request to delete element
X should fail (like with all other backends).

Finally, patch four, also from Pablo, reverts a recent attempt to speed
up abort of a large pending update with the "pipapo" set backend.

It could cause stray references to remain in the set, which then
results in a double-free.

* tag 'nf-23-10-18' of https://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf:
  netfilter: nf_tables: revert do not remove elements if set backend implements .abort
  netfilter: nft_set_rbtree: .deactivate fails if element has expired
  selftests: netfilter: Run nft_audit.sh in its own netns
  netfilter: nf_tables: audit log object reset once per table
====================

Link: https://lore.kernel.org/r/20231018125605.27299-1-fw@strlen.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
12 months agoMerge tag 'wireless-2023-10-18' of git://git.kernel.org/pub/scm/linux/kernel/git...
Jakub Kicinski [Thu, 19 Oct 2023 01:14:25 +0000 (18:14 -0700)]
Merge tag 'wireless-2023-10-18' of git://git./linux/kernel/git/wireless/wireless

Johannes Berg says:

====================
A few more fixes:
 * prevent value bounce/glitch in rfkill GPIO probe
 * fix lockdep report in rfkill
 * fix error path leak in mac80211 key handling
 * use system_unbound_wq for wiphy work since it
   can take longer

* tag 'wireless-2023-10-18' of git://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless:
  net: rfkill: reduce data->mtx scope in rfkill_fop_open
  net: rfkill: gpio: prevent value glitch during probe
  wifi: mac80211: fix error path key leak
  wifi: cfg80211: use system_unbound_wq for wiphy work
====================

Link: https://lore.kernel.org/r/20231018071041.8175-2-johannes@sipsolutions.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
12 months agonet: phy: bcm7xxx: Add missing 16nm EPHY statistics
Florian Fainelli [Tue, 17 Oct 2023 20:51:19 +0000 (13:51 -0700)]
net: phy: bcm7xxx: Add missing 16nm EPHY statistics

The .probe() function would allocate the necessary space and ensure that
the library call sizes the number of statistics but the callbacks
necessary to fetch the name and values were not wired up.

Reported-by: Justin Chen <justin.chen@broadcom.com>
Fixes: f68d08c437f9 ("net: phy: bcm7xxx: Add EPHY entry for 72165")
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Florian Fainelli <florian.fainelli@broadcom.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://lore.kernel.org/r/20231017205119.416392-1-florian.fainelli@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
12 months agoipv4: fib: annotate races around nh->nh_saddr_genid and nh->nh_saddr
Eric Dumazet [Tue, 17 Oct 2023 19:23:04 +0000 (19:23 +0000)]
ipv4: fib: annotate races around nh->nh_saddr_genid and nh->nh_saddr

syzbot reported a data-race while accessing nh->nh_saddr_genid [1]

Add annotations, but leave the code lazy as intended.

[1]
BUG: KCSAN: data-race in fib_select_path / fib_select_path

write to 0xffff8881387166f0 of 4 bytes by task 6778 on cpu 1:
fib_info_update_nhc_saddr net/ipv4/fib_semantics.c:1334 [inline]
fib_result_prefsrc net/ipv4/fib_semantics.c:1354 [inline]
fib_select_path+0x292/0x330 net/ipv4/fib_semantics.c:2269
ip_route_output_key_hash_rcu+0x659/0x12c0 net/ipv4/route.c:2810
ip_route_output_key_hash net/ipv4/route.c:2644 [inline]
__ip_route_output_key include/net/route.h:134 [inline]
ip_route_output_flow+0xa6/0x150 net/ipv4/route.c:2872
send4+0x1f5/0x520 drivers/net/wireguard/socket.c:61
wg_socket_send_skb_to_peer+0x94/0x130 drivers/net/wireguard/socket.c:175
wg_socket_send_buffer_to_peer+0xd6/0x100 drivers/net/wireguard/socket.c:200
wg_packet_send_handshake_initiation drivers/net/wireguard/send.c:40 [inline]
wg_packet_handshake_send_worker+0x10c/0x150 drivers/net/wireguard/send.c:51
process_one_work kernel/workqueue.c:2630 [inline]
process_scheduled_works+0x5b8/0xa30 kernel/workqueue.c:2703
worker_thread+0x525/0x730 kernel/workqueue.c:2784
kthread+0x1d7/0x210 kernel/kthread.c:388
ret_from_fork+0x48/0x60 arch/x86/kernel/process.c:147
ret_from_fork_asm+0x11/0x20 arch/x86/entry/entry_64.S:304

read to 0xffff8881387166f0 of 4 bytes by task 6759 on cpu 0:
fib_result_prefsrc net/ipv4/fib_semantics.c:1350 [inline]
fib_select_path+0x1cb/0x330 net/ipv4/fib_semantics.c:2269
ip_route_output_key_hash_rcu+0x659/0x12c0 net/ipv4/route.c:2810
ip_route_output_key_hash net/ipv4/route.c:2644 [inline]
__ip_route_output_key include/net/route.h:134 [inline]
ip_route_output_flow+0xa6/0x150 net/ipv4/route.c:2872
send4+0x1f5/0x520 drivers/net/wireguard/socket.c:61
wg_socket_send_skb_to_peer+0x94/0x130 drivers/net/wireguard/socket.c:175
wg_socket_send_buffer_to_peer+0xd6/0x100 drivers/net/wireguard/socket.c:200
wg_packet_send_handshake_initiation drivers/net/wireguard/send.c:40 [inline]
wg_packet_handshake_send_worker+0x10c/0x150 drivers/net/wireguard/send.c:51
process_one_work kernel/workqueue.c:2630 [inline]
process_scheduled_works+0x5b8/0xa30 kernel/workqueue.c:2703
worker_thread+0x525/0x730 kernel/workqueue.c:2784
kthread+0x1d7/0x210 kernel/kthread.c:388
ret_from_fork+0x48/0x60 arch/x86/kernel/process.c:147
ret_from_fork_asm+0x11/0x20 arch/x86/entry/entry_64.S:304

value changed: 0x959d3217 -> 0x959d3218

Reported by Kernel Concurrency Sanitizer on:
CPU: 0 PID: 6759 Comm: kworker/u4:15 Not tainted 6.6.0-rc4-syzkaller-00029-gcbf3a2cb156a #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/06/2023
Workqueue: wg-kex-wg1 wg_packet_handshake_send_worker

Fixes: 436c3b66ec98 ("ipv4: Invalidate nexthop cache nh_saddr more correctly.")
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: David Ahern <dsahern@kernel.org>
Link: https://lore.kernel.org/r/20231017192304.82626-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
12 months agotcp_bpf: properly release resources on error paths
Paolo Abeni [Tue, 17 Oct 2023 15:49:51 +0000 (17:49 +0200)]
tcp_bpf: properly release resources on error paths

In the blamed commit below, I completely forgot to release the acquired
resources before erroring out in the TCP BPF code, as reported by Dan.

Address the issues by replacing the bogus return with a jump to the
relevant cleanup code.

Fixes: 419ce133ab92 ("tcp: allow again tcp_disconnect() when threads are waiting")
Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Acked-by: Jakub Sitnicki <jakub@cloudflare.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/r/8f99194c698bcef12666f0a9a999c58f8b1cb52c.1697557782.git.pabeni@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
12 months agonet/sched: sch_hfsc: upgrade 'rt' to 'sc' when it becomes a inner curve
Pedro Tammela [Tue, 17 Oct 2023 14:36:02 +0000 (11:36 -0300)]
net/sched: sch_hfsc: upgrade 'rt' to 'sc' when it becomes a inner curve

Christian Theune says:
   I upgraded from 6.1.38 to 6.1.55 this morning and it broke my traffic shaping script,
   leaving me with a non-functional uplink on a remote router.

A 'rt' curve cannot be used as a inner curve (parent class), but we were
allowing such configurations since the qdisc was introduced. Such
configurations would trigger a UAF as Budimir explains:
   The parent will have vttree_insert() called on it in init_vf(),
   but will not have vttree_remove() called on it in update_vf()
   because it does not have the HFSC_FSC flag set.

The qdisc always assumes that inner classes have the HFSC_FSC flag set.
This is by design as it doesn't make sense 'qdisc wise' for an 'rt'
curve to be an inner curve.

Budimir's original patch disallows users to add classes with a 'rt'
parent, but this is too strict as it breaks users that have been using
'rt' as a inner class. Another approach, taken by this patch, is to
upgrade the inner 'rt' into a 'sc', warning the user in the process.
It avoids the UAF reported by Budimir while also being more permissive
to bad scripts/users/code using 'rt' as a inner class.

Users checking the `tc class ls [...]` or `tc class get [...]` dumps would
observe the curve change and are potentially breaking with this change.

v1->v2: https://lore.kernel.org/all/20231013151057.2611860-1-pctammela@mojatatu.com/
- Correct 'Fixes' tag and merge with revert (Jakub)

Cc: Christian Theune <ct@flyingcircus.io>
Cc: Budimir Markovic <markovicbudimir@gmail.com>
Fixes: b3d26c5702c7 ("net/sched: sch_hfsc: Ensure inner classes have fsc curve")
Signed-off-by: Pedro Tammela <pctammela@mojatatu.com>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Link: https://lore.kernel.org/r/20231017143602.3191556-1-pctammela@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
12 months agonet: mdio-mux: fix C45 access returning -EIO after API change
Vladimir Oltean [Tue, 17 Oct 2023 14:31:44 +0000 (17:31 +0300)]
net: mdio-mux: fix C45 access returning -EIO after API change

The mii_bus API conversion to read_c45() and write_c45() did not cover
the mdio-mux driver before read() and write() were made C22-only.

This broke arch/arm64/boot/dts/freescale/fsl-ls1028a-qds-13bb.dtso.
The -EOPNOTSUPP from mdiobus_c45_read() is transformed by
get_phy_c45_devs_in_pkg() into -EIO, is further propagated to
of_mdiobus_register() and this makes the mdio-mux driver fail to probe
the entire child buses, not just the PHYs that cause access errors.

Fix the regression by introducing special c45 read and write accessors
to mdio-mux which forward the operation to the parent MDIO bus.

Fixes: db1a63aed89c ("net: phy: Remove fallback to old C45 method")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <florian.fainelli@broadcom.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://lore.kernel.org/r/20231017143144.3212657-1-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
12 months agotcp: tsq: relax tcp_small_queue_check() when rtx queue contains a single skb
Eric Dumazet [Tue, 17 Oct 2023 12:45:26 +0000 (12:45 +0000)]
tcp: tsq: relax tcp_small_queue_check() when rtx queue contains a single skb

In commit 75eefc6c59fd ("tcp: tsq: add a shortcut in tcp_small_queue_check()")
we allowed to send an skb regardless of TSQ limits being hit if rtx queue
was empty or had a single skb, in order to better fill the pipe
when/if TX completions were slow.

Then later, commit 75c119afe14f ("tcp: implement rb-tree based
retransmit queue") accidentally removed the special case for
one skb in rtx queue.

Stefan Wahren reported a regression in single TCP flow throughput
using a 100Mbit fec link, starting from commit 65466904b015 ("tcp: adjust
TSO packet sizes based on min_rtt"). This last commit only made the
regression more visible, because it locked the TCP flow on a particular
behavior where TSQ prevented two skbs being pushed downstream,
adding silences on the wire between each TSO packet.

Many thanks to Stefan for his invaluable help !

Fixes: 75c119afe14f ("tcp: implement rb-tree based retransmit queue")
Link: https://lore.kernel.org/netdev/7f31ddc8-9971-495e-a1f6-819df542e0af@gmx.net/
Reported-by: Stefan Wahren <wahrenst@gmx.net>
Tested-by: Stefan Wahren <wahrenst@gmx.net>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Link: https://lore.kernel.org/r/20231017124526.4060202-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
12 months agoocteon_ep: update BQL sent bytes before ringing doorbell
Shinas Rasheed [Tue, 17 Oct 2023 10:50:30 +0000 (03:50 -0700)]
octeon_ep: update BQL sent bytes before ringing doorbell

Sometimes Tx is completed immediately after doorbell is updated, which
causes Tx completion routing to update completion bytes before the
same packet bytes are updated in sent bytes in transmit function, hence
hitting BUG_ON() in dql_completed(). To avoid this, update BQL
sent bytes before ringing doorbell.

Fixes: 37d79d059606 ("octeon_ep: add Tx/Rx processing and interrupt support")
Signed-off-by: Shinas Rasheed <srasheed@marvell.com>
Link: https://lore.kernel.org/r/20231017105030.2310966-1-srasheed@marvell.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
12 months agoperf/benchmark: fix seccomp_unotify benchmark for 32-bit
Jiri Slaby (SUSE) [Tue, 17 Oct 2023 08:30:19 +0000 (10:30 +0200)]
perf/benchmark: fix seccomp_unotify benchmark for 32-bit

Commit 7d5cb68af638 (perf/benchmark: add a new benchmark for
seccom_unotify) added a reference to __NR_seccomp into perf. This is
fine as it added also a definition of __NR_seccomp for 64-bit. But it
failed to do so for 32-bit as instead of ifndef, ifdef was used.

Fix this typo (so fix the build of perf on 32-bit).

Fixes: 7d5cb68af638 (perf/benchmark: add a new benchmark for seccom_unotify)
Cc: Andrei Vagin <avagin@google.com>
Cc: "Peter Zijlstra (Intel)" <peterz@infradead.org>
Cc: Kees Cook <keescook@chromium.org>
Signed-off-by: "Jiri Slaby (SUSE)" <jirislaby@kernel.org>
Link: https://lore.kernel.org/r/20231017083019.31733-1-jirislaby@kernel.org
Signed-off-by: Kees Cook <keescook@chromium.org>
12 months agoMerge tag 'nvme-6.6-2023-10-18' of git://git.infradead.org/nvme into block-6.6
Jens Axboe [Wed, 18 Oct 2023 21:32:51 +0000 (15:32 -0600)]
Merge tag 'nvme-6.6-2023-10-18' of git://git.infradead.org/nvme into block-6.6

Pull NVMe fixes from Keith:

"nvme fixes for Linux 6.6

 - nvme-rdma queue fix (Maurizio)
 - nvmet-auth double free fix (Maurizio)
 - nvme-tcp use-after-free fix (Sagi)
 - nvme-auth data direction fix (Martin)
 - nvme passthrough metadata sanitization (Keith)
 - nvme bogus identifiers for multi-controller ssd (Keith)"

* tag 'nvme-6.6-2023-10-18' of git://git.infradead.org/nvme:
  nvme-pci: add BOGUS_NID for Intel 0a54 device
  nvmet-auth: complete a request only after freeing the dhchap pointers
  nvme: sanitize metadata bounce buffer for reads
  nvme-auth: use chap->s2 to indicate bidirectional authentication
  nvmet-tcp: Fix a possible UAF in queue intialization setup
  nvme-rdma: do not try to stop unallocated queues

12 months agonvme-pci: add BOGUS_NID for Intel 0a54 device
Keith Busch [Thu, 12 Oct 2023 18:13:51 +0000 (11:13 -0700)]
nvme-pci: add BOGUS_NID for Intel 0a54 device

These ones claim cmic and nmic capable, so need special consideration to ignore
their duplicate identifiers.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=217981
Reported-by: welsh@cassens.com
Signed-off-by: Keith Busch <kbusch@kernel.org>
12 months agonvmet-auth: complete a request only after freeing the dhchap pointers
Maurizio Lombardi [Tue, 17 Oct 2023 08:28:45 +0000 (10:28 +0200)]
nvmet-auth: complete a request only after freeing the dhchap pointers

It may happen that the work to destroy a queue
(for example nvmet_tcp_release_queue_work()) is started while
an auth-send or auth-receive command is still completing.

nvmet_sq_destroy() will block, waiting for all the references
to the sq to be dropped, the last reference is then
dropped when nvmet_req_complete() is called.

When this happens, both nvmet_sq_destroy() and
nvmet_execute_auth_send()/_receive() will free the dhchap pointers by
calling nvmet_auth_sq_free().
Since there isn't any lock, the two threads may race against each other,
causing double frees and memory corruptions, as reported by KASAN.

Reproduced by stress blktests nvme/041 nvme/042 nvme/043

 nvme nvme2: qid 0: authenticated with hash hmac(sha512) dhgroup ffdhe4096
 ==================================================================
 BUG: KASAN: double-free in kfree+0xec/0x4b0

 Call Trace:
  <TASK>
  kfree+0xec/0x4b0
  nvmet_auth_sq_free+0xe1/0x160 [nvmet]
  nvmet_execute_auth_send+0x482/0x16d0 [nvmet]
  process_one_work+0x8e5/0x1510

 Allocated by task 191846:
  __kasan_kmalloc+0x81/0xa0
  nvmet_auth_ctrl_sesskey+0xf6/0x380 [nvmet]
  nvmet_auth_reply+0x119/0x990 [nvmet]

 Freed by task 143270:
  kfree+0xec/0x4b0
  nvmet_auth_sq_free+0xe1/0x160 [nvmet]
  process_one_work+0x8e5/0x1510

Fix this bug by calling nvmet_req_complete() only after freeing the
pointers, so we will prevent the race by holding the sq reference.

V2: remove redundant code

Fixes: db1312dd9548 ("nvmet: implement basic In-Band Authentication")
Signed-off-by: Maurizio Lombardi <mlombard@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Keith Busch <kbusch@kernel.org>
12 months agonvme: sanitize metadata bounce buffer for reads
Keith Busch [Mon, 16 Oct 2023 20:12:47 +0000 (13:12 -0700)]
nvme: sanitize metadata bounce buffer for reads

User can request more metadata bytes than the device will write. Ensure
kernel buffer is initialized so we're not leaking unsanitized memory on
the copy-out.

Fixes: 0b7f1f26f95a51a ("nvme: use the block layer for userspace passthrough metadata")
Reviewed-by: Jens Axboe <axboe@kernel.dk>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Kanchan Joshi <joshi.k@samsung.com>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Signed-off-by: Keith Busch <kbusch@kernel.org>
12 months agoNFSv4.1: fixup use EXCHGID4_FLAG_USE_PNFS_DS for DS server
Olga Kornievskaia [Mon, 9 Oct 2023 14:59:01 +0000 (10:59 -0400)]
NFSv4.1: fixup use EXCHGID4_FLAG_USE_PNFS_DS for DS server

This patches fixes commit 51d674a5e488 "NFSv4.1: use
EXCHGID4_FLAG_USE_PNFS_DS for DS server", purpose of that
commit was to mark EXCHANGE_ID to the DS with the appropriate
flag.

However, connection to MDS can return both EXCHGID4_FLAG_USE_PNFS_DS
and EXCHGID4_FLAG_USE_PNFS_MDS set but previous patch would only
remember the USE_PNFS_DS and for the 2nd EXCHANGE_ID send that
to the MDS.

Instead, just mark the pnfs path exclusively.

Fixes: 51d674a5e488 ("NFSv4.1: use EXCHGID4_FLAG_USE_PNFS_DS for DS server")
Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
12 months agopNFS/flexfiles: Check the layout validity in ff_layout_mirror_prepare_stats
Trond Myklebust [Sun, 8 Oct 2023 18:28:46 +0000 (14:28 -0400)]
pNFS/flexfiles: Check the layout validity in ff_layout_mirror_prepare_stats

Ensure that we check the layout pointer and validity after dereferencing
it in ff_layout_mirror_prepare_stats.

Fixes: 08e2e5bc6c9a ("pNFS/flexfiles: Clean up layoutstats")
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
12 months agopNFS: Fix a hang in nfs4_evict_inode()
Trond Myklebust [Sun, 8 Oct 2023 18:20:19 +0000 (14:20 -0400)]
pNFS: Fix a hang in nfs4_evict_inode()

We are not allowed to call pnfs_mark_matching_lsegs_return() without
also holding a reference to the layout header, since doing so could lead
to the reference count going to zero when we call
pnfs_layout_remove_lseg(). This again can lead to a hang when we get to
nfs4_evict_inode() and are unable to clear the layout pointer.

pnfs_layout_return_unused_byserver() is guilty of this behaviour, and
has been seen to trigger the refcount warning prior to a hang.

Fixes: b6d49ecd1081 ("NFSv4: Fix a pNFS layout related use-after-free race when freeing the inode")
Cc: stable@vger.kernel.org
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
12 months agoMerge tag 'asoc-fix-v6.6-rc6' of https://git.kernel.org/pub/scm/linux/kernel/git...
Takashi Iwai [Wed, 18 Oct 2023 18:02:46 +0000 (20:02 +0200)]
Merge tag 'asoc-fix-v6.6-rc6' of https://git./linux/kernel/git/broonie/sound into for-linus

ASoC: Fixes for v6.6

A fairly large set of fixes here but all driver specific, the biggest
block is Johan's work shaking out issues with device setup and teardown
for the wcd938x driver which is a relatively large but clearly broken
down set of changes.

There is one core helper function added as part of a fix for wsa-macro.

12 months agoMerge tag 'spi-fix-v6-6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi
Linus Torvalds [Wed, 18 Oct 2023 16:37:36 +0000 (09:37 -0700)]
Merge tag 'spi-fix-v6-6-rc4' of git://git./linux/kernel/git/broonie/spi

Pull spi fix from Mark Brown:
 "A fix for the npcm-fiu driver in cases where there are no dummy bytes
  during reads"

* tag 'spi-fix-v6-6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi:
  spi: npcm-fiu: Fix UMA reads when dummy.nbytes == 0

12 months agoMerge tag 'regmap-fix-v6.6-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 18 Oct 2023 16:30:03 +0000 (09:30 -0700)]
Merge tag 'regmap-fix-v6.6-rc6' of git://git./linux/kernel/git/broonie/regmap

Pull regmap fix from Mark Brown:
 "A straightforward fix from Johan for a long standing bug in cases
  where we both have regmaps without devices and something is using
  dev_get_regmap()"

* tag 'regmap-fix-v6.6-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap:
  regmap: fix NULL deref on lookup

12 months agoio_uring: fix crash with IORING_SETUP_NO_MMAP and invalid SQ ring address
Jens Axboe [Wed, 18 Oct 2023 14:09:27 +0000 (08:09 -0600)]
io_uring: fix crash with IORING_SETUP_NO_MMAP and invalid SQ ring address

If we specify a valid CQ ring address but an invalid SQ ring address,
we'll correctly spot this and free the allocated pages and clear them
to NULL. However, we don't clear the ring page count, and hence will
attempt to free the pages again. We've already cleared the address of
the page array when freeing them, but we don't check for that. This
causes the following crash:

Unable to handle kernel NULL pointer dereference at virtual address 0000000000000000
Oops [#1]
Modules linked in:
CPU: 0 PID: 20 Comm: kworker/u2:1 Not tainted 6.6.0-rc5-dirty #56
Hardware name: ucbbar,riscvemu-bare (DT)
Workqueue: events_unbound io_ring_exit_work
epc : io_pages_free+0x2a/0x58
 ra : io_rings_free+0x3a/0x50
 epc : ffffffff808811a2 ra : ffffffff80881406 sp : ffff8f80000c3cd0
 status: 0000000200000121 badaddr: 0000000000000000 cause: 000000000000000d
 [<ffffffff808811a2>] io_pages_free+0x2a/0x58
 [<ffffffff80881406>] io_rings_free+0x3a/0x50
 [<ffffffff80882176>] io_ring_exit_work+0x37e/0x424
 [<ffffffff80027234>] process_one_work+0x10c/0x1f4
 [<ffffffff8002756e>] worker_thread+0x252/0x31c
 [<ffffffff8002f5e4>] kthread+0xc4/0xe0
 [<ffffffff8000332a>] ret_from_fork+0xa/0x1c

Check for a NULL array in io_pages_free(), but also clear the page counts
when we free them to be on the safer side.

Reported-by: rtm@csail.mit.edu
Fixes: 03d89a2de25b ("io_uring: support for user allocated memory for rings/sqes")
Cc: stable@vger.kernel.org
Reviewed-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
12 months agoplatform/mellanox: mlxbf-tmfifo: Fix a warning message
Liming Sun [Thu, 12 Oct 2023 23:02:35 +0000 (19:02 -0400)]
platform/mellanox: mlxbf-tmfifo: Fix a warning message

This commit fixes the smatch static checker warning in function
mlxbf_tmfifo_rxtx_word() which complains data not initialized at
line 634 when IS_VRING_DROP() is TRUE.

Signed-off-by: Liming Sun <limings@nvidia.com>
Link: https://lore.kernel.org/r/20231012230235.219861-1-limings@nvidia.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
12 months agoapple-gmux: Hard Code max brightness for MMIO gmux
Orlando Chamberlain [Tue, 17 Oct 2023 11:14:45 +0000 (22:14 +1100)]
apple-gmux: Hard Code max brightness for MMIO gmux

The data in the max brightness port for iMacs with MMIO gmux incorrectly
reports 0x03ff, but it should be 0xffff. As all other MMIO gmux models
have 0xffff, hard code this for all MMIO gmux's so they all have the
proper brightness range accessible.

Fixes: 0c18184de990 ("platform/x86: apple-gmux: support MMIO gmux on T2 Macs")
Reported-by: Karsten Leipold <poldi@dfn.de>
Signed-off-by: Orlando Chamberlain <orlandoch.dev@gmail.com>
Link: https://lore.kernel.org/r/20231017111444.19304-2-orlandoch.dev@gmail.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
12 months agoplatform/surface: platform_profile: Propagate error if profile registration fails
Armin Wolf [Sat, 14 Oct 2023 23:54:49 +0000 (01:54 +0200)]
platform/surface: platform_profile: Propagate error if profile registration fails

If platform_profile_register() fails, the driver does not propagate
the error, but instead probes successfully. This means when the driver
unbinds, the a warning might be issued by platform_profile_remove().

Fix this by propagating the error back to the caller of
surface_platform_profile_probe().

Compile-tested only.

Fixes: b78b4982d763 ("platform/surface: Add platform profile driver")
Signed-off-by: Armin Wolf <W_Armin@gmx.de>
Reviewed-by: Maximilian Luz <luzmaximilian@gmail.com>
Tested-by: Maximilian Luz <luzmaximilian@gmail.com>
Link: https://lore.kernel.org/r/20231014235449.288702-1-W_Armin@gmx.de
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
12 months agonetfilter: nf_tables: revert do not remove elements if set backend implements .abort
Pablo Neira Ayuso [Wed, 18 Oct 2023 11:18:39 +0000 (13:18 +0200)]
netfilter: nf_tables: revert do not remove elements if set backend implements .abort

nf_tables_abort_release() path calls nft_set_elem_destroy() for
NFT_MSG_NEWSETELEM which releases the element, however, a reference to
the element still remains in the working copy.

Fixes: ebd032fa8818 ("netfilter: nf_tables: do not remove elements if set backend implements .abort")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
12 months agonetfilter: nft_set_rbtree: .deactivate fails if element has expired
Pablo Neira Ayuso [Tue, 17 Oct 2023 10:28:27 +0000 (12:28 +0200)]
netfilter: nft_set_rbtree: .deactivate fails if element has expired

This allows to remove an expired element which is not possible in other
existing set backends, this is more noticeable if gc-interval is high so
expired elements remain in the tree. On-demand gc also does not help in
this case, because this is delete element path. Return NULL if element
has expired.

Fixes: 8d8540c4f5e0 ("netfilter: nft_set_rbtree: add timeout support")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
12 months agoselftests: netfilter: Run nft_audit.sh in its own netns
Phil Sutter [Fri, 13 Oct 2023 20:02:24 +0000 (22:02 +0200)]
selftests: netfilter: Run nft_audit.sh in its own netns

Don't mess with the host's firewall ruleset. Since audit logging is not
per-netns, add an initial delay of a second so other selftests' netns
cleanups have a chance to finish.

Fixes: e8dbde59ca3f ("selftests: netfilter: Test nf_tables audit logging")
Signed-off-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: Florian Westphal <fw@strlen.de>
12 months agonetfilter: nf_tables: audit log object reset once per table
Phil Sutter [Wed, 11 Oct 2023 15:06:59 +0000 (17:06 +0200)]
netfilter: nf_tables: audit log object reset once per table

When resetting multiple objects at once (via dump request), emit a log
message per table (or filled skb) and resurrect the 'entries' parameter
to contain the number of objects being logged for.

To test the skb exhaustion path, perform some bulk counter and quota
adds in the kselftest.

Signed-off-by: Phil Sutter <phil@nwl.cc>
Reviewed-by: Richard Guy Briggs <rgb@redhat.com>
Acked-by: Paul Moore <paul@paul-moore.com> (Audit)
Signed-off-by: Florian Westphal <fw@strlen.de>
12 months agoACPI: bus: Move acpi_arm_init() to the place of after acpi_ghes_init()
Hanjun Guo [Tue, 10 Oct 2023 08:21:23 +0000 (16:21 +0800)]
ACPI: bus: Move acpi_arm_init() to the place of after acpi_ghes_init()

acpi_agdi_init() in acpi_arm_init() will register a SDEI event, so
it needs the SDEI subsystem to be initialized (which is done in
acpi_ghes_init()) before the AGDI driver probing.

In commit fcea0ccf4fd7 ("ACPI: bus: Consolidate all arm specific
initialisation into acpi_arm_init()"), the acpi_agdi_init() was
called before acpi_ghes_init() and it causes following failure:

| [    0.515864] sdei: Failed to create event 1073741825: -5
| [    0.515866] agdi agdi.0: Failed to register for SDEI event 1073741825
| [    0.515867] agdi: probe of agdi.0 failed with error -5
| ...
| [    0.516022] sdei: SDEIv1.0 (0x0) detected in firmware.

Fix it by moving acpi_arm_init() to the place of after
acpi_ghes_init().

Fixes: fcea0ccf4fd7 ("ACPI: bus: Consolidate all arm specific initialisation into acpi_arm_init()")
Reported-by: D Scott Phillips <scott@os.amperecomputing.com>
Signed-off-by: Hanjun Guo <guohanjun@huawei.com>
Reviewed-by: Sudeep Holla <sudeep.holla@arm.com>
Tested-by: D Scott Phillips <scott@os.amperecomputing.com>
Cc: 6.5+ <stable@vger.kernel.org> # 6.5+
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
12 months agoACPI: irq: Fix incorrect return value in acpi_register_gsi()
Sunil V L [Mon, 16 Oct 2023 17:09:39 +0000 (22:39 +0530)]
ACPI: irq: Fix incorrect return value in acpi_register_gsi()

acpi_register_gsi() should return a negative value in case of failure.

Currently, it returns the return value from irq_create_fwspec_mapping().
However, irq_create_fwspec_mapping() returns 0 for failure. Fix the
issue by returning -EINVAL if irq_create_fwspec_mapping() returns zero.

Fixes: d44fa3d46079 ("ACPI: Add support for ResourceSource/IRQ domain mapping")
Cc: 4.11+ <stable@vger.kernel.org> # 4.11+
Signed-off-by: Sunil V L <sunilvl@ventanamicro.com>
[ rjw: Rename a new local variable ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
12 months agoneighbor: tracing: Move pin6 inside CONFIG_IPV6=y section
Geert Uytterhoeven [Mon, 16 Oct 2023 12:49:04 +0000 (14:49 +0200)]
neighbor: tracing: Move pin6 inside CONFIG_IPV6=y section

When CONFIG_IPV6=n, and building with W=1:

    In file included from include/trace/define_trace.h:102,
     from include/trace/events/neigh.h:255,
     from net/core/net-traces.c:51:
    include/trace/events/neigh.h: In function ‘trace_event_raw_event_neigh_create’:
    include/trace/events/neigh.h:42:34: error: variable ‘pin6’ set but not used [-Werror=unused-but-set-variable]
       42 |                 struct in6_addr *pin6;
  |                                  ^~~~
    include/trace/trace_events.h:402:11: note: in definition of macro ‘DECLARE_EVENT_CLASS’
      402 |         { assign; }                                                     \
  |           ^~~~~~
    include/trace/trace_events.h:44:30: note: in expansion of macro ‘PARAMS’
       44 |                              PARAMS(assign),                   \
  |                              ^~~~~~
    include/trace/events/neigh.h:23:1: note: in expansion of macro ‘TRACE_EVENT’
       23 | TRACE_EVENT(neigh_create,
  | ^~~~~~~~~~~
    include/trace/events/neigh.h:41:9: note: in expansion of macro ‘TP_fast_assign’
       41 |         TP_fast_assign(
  |         ^~~~~~~~~~~~~~
    In file included from include/trace/define_trace.h:103,
     from include/trace/events/neigh.h:255,
     from net/core/net-traces.c:51:
    include/trace/events/neigh.h: In function ‘perf_trace_neigh_create’:
    include/trace/events/neigh.h:42:34: error: variable ‘pin6’ set but not used [-Werror=unused-but-set-variable]
       42 |                 struct in6_addr *pin6;
  |                                  ^~~~
    include/trace/perf.h:51:11: note: in definition of macro ‘DECLARE_EVENT_CLASS’
       51 |         { assign; }                                                     \
  |           ^~~~~~
    include/trace/trace_events.h:44:30: note: in expansion of macro ‘PARAMS’
       44 |                              PARAMS(assign),                   \
  |                              ^~~~~~
    include/trace/events/neigh.h:23:1: note: in expansion of macro ‘TRACE_EVENT’
       23 | TRACE_EVENT(neigh_create,
  | ^~~~~~~~~~~
    include/trace/events/neigh.h:41:9: note: in expansion of macro ‘TP_fast_assign’
       41 |         TP_fast_assign(
  |         ^~~~~~~~~~~~~~

Indeed, the variable pin6 is declared and initialized unconditionally,
while it is only used and needlessly re-initialized when support for
IPv6 is enabled.

Fix this by dropping the unused variable initialization, and moving the
variable declaration inside the existing section protected by a check
for CONFIG_IPV6.

Fixes: fc651001d2c5ca4f ("neighbor: Add tracepoint to __neigh_create")
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Simon Horman <horms@kernel.org> # build-tested
Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 months agoRevert "net: wwan: iosm: enable runtime pm support for 7560"
Bagas Sanjaya [Tue, 17 Oct 2023 08:08:12 +0000 (15:08 +0700)]
Revert "net: wwan: iosm: enable runtime pm support for 7560"

Runtime power management support breaks Intel LTE modem where dmesg dump
showes timeout errors:

```
[   72.027442] iosm 0000:01:00.0: msg timeout
[   72.531638] iosm 0000:01:00.0: msg timeout
[   73.035414] iosm 0000:01:00.0: msg timeout
[   73.540359] iosm 0000:01:00.0: msg timeout
```

Furthermore, when shutting down with `poweroff` and modem attached, the
system rebooted instead of powering down as expected. The modem works
again only after power cycling.

Revert runtime power management support for IOSM driver as introduced by
commit e4f5073d53be6c ("net: wwan: iosm: enable runtime pm support for
7560").

Fixes: e4f5073d53be ("net: wwan: iosm: enable runtime pm support for 7560")
Reported-by: Martin <mwolf@adiumentum.com>
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=217996
Link: https://lore.kernel.org/r/267abf02-4b60-4a2e-92cd-709e3da6f7d3@gmail.com/
Signed-off-by: Bagas Sanjaya <bagasdotme@gmail.com>
Reviewed-by: Loic Poulain <loic.poulain@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 months agoplatform/x86: asus-wmi: Map 0x2a code, Ignore 0x2b and 0x2c events
Hans de Goede [Tue, 17 Oct 2023 09:07:25 +0000 (11:07 +0200)]
platform/x86: asus-wmi: Map 0x2a code, Ignore 0x2b and 0x2c events

Newer Asus laptops send the following new WMI event codes when some
of the F1 - F12 "media" hotkeys are pressed:

0x2a Screen Capture
0x2b PrintScreen
0x2c CapsLock

Map 0x2a to KEY_SELECTIVE_SCREENSHOT mirroring how similar hotkeys
are mapped on other laptops.

PrintScreem and CapsLock are also reported as normal PS/2 keyboard events,
map these event codes to KE_IGNORE to avoid "Unknown key code 0x%x\n" log
messages.

Reported-by: James John <me@donjajo.com>
Closes: https://lore.kernel.org/platform-driver-x86/a2c441fe-457e-44cf-a146-0ecd86b037cf@donjajo.com/
Closes: https://bbs.archlinux.org/viewtopic.php?pid=2123716
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20231017090725.38163-4-hdegoede@redhat.com
12 months agoplatform/x86: asus-wmi: Only map brightness codes when using asus-wmi backlight control
Hans de Goede [Tue, 17 Oct 2023 09:07:24 +0000 (11:07 +0200)]
platform/x86: asus-wmi: Only map brightness codes when using asus-wmi backlight control

Older Asus laptops change the backlight level themselves and then send
WMI events with different codes for different backlight levels.

The asus-wmi.c code maps the entire range of codes reported on
brightness down keypresses to an internal ASUS_WMI_BRN_DOWN code:

define NOTIFY_BRNUP_MIN                0x11
define NOTIFY_BRNUP_MAX                0x1f
define NOTIFY_BRNDOWN_MIN              0x20
define NOTIFY_BRNDOWN_MAX              0x2e

        if (code >= NOTIFY_BRNUP_MIN && code <= NOTIFY_BRNUP_MAX)
                code = ASUS_WMI_BRN_UP;
        else if (code >= NOTIFY_BRNDOWN_MIN && code <= NOTIFY_BRNDOWN_MAX)
                code = ASUS_WMI_BRN_DOWN;

This mapping is causing issues on new laptop models which actually
send 0x2b events for printscreen presses and 0x2c events for
capslock presses, which get translated into spurious brightness-down
presses.

This mapping is really only necessary when asus-wmi has registered
a backlight-device for backlight control. In this case the mapping
was used to decide to filter out the keypresss since in this case
the firmware has already modified the brightness itself and instead
of reporting a keypress asus-wmi will just report the new brightness
value to userspace.

OTOH when the firmware does not adjust the brightness itself then
it seems to always report 0x2e for brightness-down presses and
0x2f for brightness up presses independent of the actual brightness
level. So in this case the mapping of the code is not necessary
and this translation actually leads to spurious brightness-down
presses being send to userspace when pressing printscreen or capslock.

Modify asus_wmi_handle_event_code() to only do the mapping
when using asus-wmi backlight control to fix the spurious
brightness-down presses.

Reported-by: James John <me@donjajo.com>
Closes: https://lore.kernel.org/platform-driver-x86/a2c441fe-457e-44cf-a146-0ecd86b037cf@donjajo.com/
Closes: https://bbs.archlinux.org/viewtopic.php?pid=2123716
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20231017090725.38163-3-hdegoede@redhat.com
12 months agoplatform/x86: asus-wmi: Change ASUS_WMI_BRN_DOWN code from 0x20 to 0x2e
Hans de Goede [Tue, 17 Oct 2023 09:07:23 +0000 (11:07 +0200)]
platform/x86: asus-wmi: Change ASUS_WMI_BRN_DOWN code from 0x20 to 0x2e

Older Asus laptops change the backlight level themselves and then send
WMI events with different codes for different backlight levels.

The asus-wmi.c code maps the entire range of codes reported on
brightness down keypresses to an internal ASUS_WMI_BRN_DOWN code:

define NOTIFY_BRNUP_MIN                0x11
define NOTIFY_BRNUP_MAX                0x1f
define NOTIFY_BRNDOWN_MIN              0x20
define NOTIFY_BRNDOWN_MAX              0x2e

        if (code >= NOTIFY_BRNUP_MIN && code <= NOTIFY_BRNUP_MAX)
                code = ASUS_WMI_BRN_UP;
        else if (code >= NOTIFY_BRNDOWN_MIN && code <= NOTIFY_BRNDOWN_MAX)
                code = ASUS_WMI_BRN_DOWN;

Before this commit all the NOTIFY_BRNDOWN_MIN - NOTIFY_BRNDOWN_MAX
aka 0x20 - 0x2e events were mapped to 0x20.

This mapping is causing issues on new laptop models which actually
send 0x2b events for printscreen presses and 0x2c events for
capslock presses, which get translated into spurious brightness-down
presses.

The plan is disable the 0x11-0x2e special mapping on laptops
where asus-wmi does not register a backlight-device to avoid
the spurious brightness-down keypresses. New laptops always send
0x2e for brightness-down presses, change the special internal
ASUS_WMI_BRN_DOWN value from 0x20 to 0x2e to match this in
preparation for fixing the spurious brightness-down presses.

This change does not have any functional impact since all
of 0x20 - 0x2e is mapped to ASUS_WMI_BRN_DOWN first and only
then checked against the keymap code and the new 0x2e
value is still in the 0x20 - 0x2e range.

Reported-by: James John <me@donjajo.com>
Closes: https://lore.kernel.org/platform-driver-x86/a2c441fe-457e-44cf-a146-0ecd86b037cf@donjajo.com/
Closes: https://bbs.archlinux.org/viewtopic.php?pid=2123716
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20231017090725.38163-2-hdegoede@redhat.com
12 months agonet: pktgen: Fix interface flags printing
Gavrilov Ilia [Mon, 16 Oct 2023 14:08:59 +0000 (14:08 +0000)]
net: pktgen: Fix interface flags printing

Device flags are displayed incorrectly:
1) The comparison (i == F_FLOW_SEQ) is always false, because F_FLOW_SEQ
is equal to (1 << FLOW_SEQ_SHIFT) == 2048, and the maximum value
of the 'i' variable is (NR_PKT_FLAG - 1) == 17. It should be compared
with FLOW_SEQ_SHIFT.

2) Similarly to the F_IPSEC flag.

3) Also add spaces to the print end of the string literal "spi:%u"
to prevent the output from merging with the flag that follows.

Found by InfoTeCS on behalf of Linux Verification Center
(linuxtesting.org) with SVACE.

Fixes: 99c6d3d20d62 ("pktgen: Remove brute-force printing of flags")
Signed-off-by: Gavrilov Ilia <Ilia.Gavrilov@infotecs.ru>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 months agoKEYS: asymmetric: Fix sign/verify on pkcs1pad without a hash
Herbert Xu [Mon, 16 Oct 2023 08:35:36 +0000 (16:35 +0800)]
KEYS: asymmetric: Fix sign/verify on pkcs1pad without a hash

The new sign/verify code broke the case of pkcs1pad without a
hash algorithm.  Fix it by setting issig correctly for this case.

Fixes: 63ba4d67594a ("KEYS: asymmetric: Use new crypto interface without scatterlists")
Cc: stable@vger.kernel.org # v6.5
Reported-by: Denis Kenzior <denkenz@gmail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Tested-by: Denis Kenzior <denkenz@gmail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
12 months agodrm/i915: Retry gtt fault when out of fence registers
Ville Syrjälä [Thu, 12 Oct 2023 13:28:01 +0000 (16:28 +0300)]
drm/i915: Retry gtt fault when out of fence registers

If we can't find a free fence register to handle a fault in the GMADR
range just return VM_FAULT_NOPAGE without populating the PTE so that
userspace will retry the access and trigger another fault. Eventually
we should find a free fence and the fault will get properly handled.

A further improvement idea might be to reserve a fence (or one per CPU?)
for the express purpose of handling faults without having to retry. But
that would require some additional work.

Looks like this may have gotten broken originally by
commit 39965b376601 ("drm/i915: don't trash the gtt when running out of fences")
as that changed the errno to -EDEADLK which wasn't handle by the gtt
fault code either. But later in commit 2feeb52859fc ("drm/i915/gt: Fix
-EDEADLK handling regression") I changed it again to -ENOBUFS as -EDEADLK
was now getting used for the ww mutex dance. So this fix only makes
sense after that last commit.

Cc: stable@vger.kernel.org
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/9479
Fixes: 2feeb52859fc ("drm/i915/gt: Fix -EDEADLK handling regression")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231012132801.16292-1-ville.syrjala@linux.intel.com
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
(cherry picked from commit 7f403caabe811b88ab0de3811ff3f4782c415761)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>