linux-2.6-block.git
6 months agoMerge tag 'mm-hotfixes-stable-2024-02-22-15-02' of git://git.kernel.org/pub/scm/linux...
Linus Torvalds [Fri, 23 Feb 2024 17:43:21 +0000 (09:43 -0800)]
Merge tag 'mm-hotfixes-stable-2024-02-22-15-02' of git://git./linux/kernel/git/akpm/mm

Pull misc fixes from Andrew Morton:
 "A batch of MM (and one non-MM) hotfixes.

  Ten are cc:stable and the remainder address post-6.7 issues or aren't
  considered appropriate for backporting"

* tag 'mm-hotfixes-stable-2024-02-22-15-02' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm:
  kasan: guard release_free_meta() shadow access with kasan_arch_is_ready()
  mm/damon/lru_sort: fix quota status loss due to online tunings
  mm/damon/reclaim: fix quota stauts loss due to online tunings
  MAINTAINERS: mailmap: update Shakeel's email address
  mm/damon/sysfs-schemes: handle schemes sysfs dir removal before commit_schemes_quota_goals
  mm: memcontrol: clarify swapaccount=0 deprecation warning
  mm/memblock: add MEMBLOCK_RSRV_NOINIT into flagname[] array
  mm/zswap: invalidate duplicate entry when !zswap_enabled
  lib/Kconfig.debug: TEST_IOV_ITER depends on MMU
  mm/swap: fix race when skipping swapcache
  mm/swap_state: update zswap LRU's protection range with the folio locked
  selftests/mm: uffd-unit-test check if huge page size is 0
  mm/damon/core: check apply interval in damon_do_apply_schemes()
  mm: zswap: fix missing folio cleanup in writeback race path

6 months agoMerge tag 'for-6.8/dm-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/devic...
Linus Torvalds [Fri, 23 Feb 2024 17:23:54 +0000 (09:23 -0800)]
Merge tag 'for-6.8/dm-fixes-2' of git://git./linux/kernel/git/device-mapper/linux-dm

Pull device mapper fixes from Mike Snitzer:

 - Stable fixes for 3 DM targets (integrity, verity and crypt) to
   address systemic failure that can occur if user provided pages map to
   the same block.

 - Fix DM crypt to not allow modifying data that being encrypted for
   authenticated encryption.

 - Fix DM crypt and verity targets to align their respective bvec_iter
   struct members to avoid the need for byte level access (due to
   __packed attribute) that is costly on some arches (like RISC).

* tag 'for-6.8/dm-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm:
  dm-crypt, dm-integrity, dm-verity: bump target version
  dm-verity, dm-crypt: align "struct bvec_iter" correctly
  dm-crypt: recheck the integrity tag after a failure
  dm-crypt: don't modify the data when using authenticated encryption
  dm-verity: recheck the hash after a failure
  dm-integrity: recheck the integrity tag after a failure

6 months agoMerge tag 'drm-fixes-2024-02-23' of git://anongit.freedesktop.org/drm/drm
Linus Torvalds [Fri, 23 Feb 2024 17:17:47 +0000 (09:17 -0800)]
Merge tag 'drm-fixes-2024-02-23' of git://anongit.freedesktop.org/drm/drm

Pull drm fixes from Dave Airlie:
 "This is the weekly drm fixes. Non-drivers there is a fbdev/sparc fix,
  syncobj, ttm and buddy fixes.

  On the driver side, ivpu, meson, i915 have a small fix each. Then
  amdgpu and xe have a bunch. Nouveau has some minor uapi additions to
  give userspace some useful info along with a Kconfig change to allow
  the new GSP firmware paths to be used by default on the GPUs it
  supports.

  Seems about the usual amount for this time of release cycle.

  fbdev:
   - fix sparc undefined reference

  syncobj:
   - fix sync obj fence waiting
   - handle NULL fence in syncobj eventfd code

  ttm:
   - fix invalid free

  buddy:
   - fix list handling
   - fix 32-bit build

  meson:
   - don't remove bridges from other drivers

  nouveau:
   - fix build warnings
   - add two minor info parameters
   - add a Kconfig to allow GSP by default on some GPUs

  ivpu:
   - allow fw to do initial tile config

  i915:
   - fix TV mode

  amdgpu:
   - Suspend/resume fixes
   - Backlight error fix
   - DCN 3.5 fixes
   - Misc fixes

  xe:
   - Remove support for persistent exec_queues
   - Drop a reduntant sysfs newline printout
   - A three-patch fix for a VM_BIND rebind optimization path
   - Fix a modpost warning on an xe KUNIT module"

* tag 'drm-fixes-2024-02-23' of git://anongit.freedesktop.org/drm/drm: (27 commits)
  nouveau: add an ioctl to report vram usage
  nouveau: add an ioctl to return vram bar size.
  nouveau/gsp: add kconfig option to enable GSP paths by default
  drm/amdgpu: Fix the runtime resume failure issue
  drm/amd/display: fix null-pointer dereference on edid reading
  drm/amd/display: Fix memory leak in dm_sw_fini()
  drm/amd/display: fix input states translation error for dcn35 & dcn351
  drm/amd/display: Fix potential null pointer dereference in dc_dmub_srv
  drm/amd/display: Only allow dig mapping to pwrseq in new asic
  drm/amd/display: adjust few initialization order in dm
  drm/syncobj: handle NULL fence in syncobj_eventfd_entry_func
  drm/syncobj: call drm_syncobj_fence_add_wait when WAIT_AVAILABLE flag is set
  drm/ttm: Fix an invalid freeing on already freed page in error path
  sparc: Fix undefined reference to fb_is_primary_device
  drm/xe: Fix modpost warning on xe_mocs kunit module
  drm/xe/xe_gt_idle: Drop redundant newline in name
  drm/xe: Return 2MB page size for compact 64k PTEs
  drm/xe: Add XE_VMA_PTE_64K VMA flag
  drm/xe: Fix xe_vma_set_pte_size
  drm/xe/uapi: Remove support for persistent exec_queues
  ...

6 months agoMerge tag 'ata-6.8-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/libata/linux
Linus Torvalds [Fri, 23 Feb 2024 17:05:56 +0000 (09:05 -0800)]
Merge tag 'ata-6.8-rc6' of git://git./linux/kernel/git/libata/linux

Pull ata fixes from Niklas Cassel:

 - Do not try to set a sleeping device to standby. Sleep is a deeper
   sleep state than standby, and needs a reset to wake up the drive. A
   system resume will reset the port. Sending a command other than reset
   to a sleeping device is not wise, as the command will timeout (Damien
   Le Moal)

 - Do not try to put a device to standby twice during system shutdown.
   ata_dev_power_set_standby() is currently called twice during
   shutdown, once after the scsi device is removed, and another when
   ata_pci_shutdown_one() executes. Modify ata_dev_power_set_standby()
   to do nothing if the device is already in standby (Damien Le Moal)

 - Add a quirk for ASM1064 to fixup the number of implemented ports. We
   probe all ports that the hardware reports to be implemented. Probing
   ports that are not implemented causes significantly increased boot
   time (Andrey Jr. Melnikov)

 - Fix error handling for the ahci_ceva driver. Ensure that the
   ahci_ceva driver does a proper cleanup of its resources in the error
   path (Radhey Shyam Pandey)

* tag 'ata-6.8-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/libata/linux:
  ata: libata-core: Do not call ata_dev_power_set_standby() twice
  ata: ahci_ceva: fix error handling for Xilinx GT PHY support
  ahci: asm1064: correct count of reported ports
  ata: libata-core: Do not try to set sleeping devices to standby

6 months agoMerge tag 'gpio-fixes-for-v6.8-rc6' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 23 Feb 2024 17:01:35 +0000 (09:01 -0800)]
Merge tag 'gpio-fixes-for-v6.8-rc6' of git://git./linux/kernel/git/brgl/linux

Pull gpio fix from Bartosz Golaszewski:

 - fix a use-case where no pins are mapped to GPIOs in
   gpiochip_generic_config()

* tag 'gpio-fixes-for-v6.8-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/brgl/linux:
  gpiolib: Handle no pin_ranges in gpiochip_generic_config()

6 months agoMerge tag 'hwmon-for-v6.8-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/groec...
Linus Torvalds [Fri, 23 Feb 2024 16:58:47 +0000 (08:58 -0800)]
Merge tag 'hwmon-for-v6.8-rc6' of git://git./linux/kernel/git/groeck/linux-staging

Pull hwmon fix from Guenter Roeck:
 "Fix a global-out-of-bounds bug in nct6775 driver"

* tag 'hwmon-for-v6.8-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
  hwmon: (nct6775) Fix access to temperature configuration registers

6 months agonouveau: add an ioctl to report vram usage
Dave Airlie [Wed, 24 Jan 2024 04:24:25 +0000 (14:24 +1000)]
nouveau: add an ioctl to report vram usage

This reports the currently used vram allocations.

userspace using this has been proposed for nvk, but
it's a rather trivial uapi addition.

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
6 months agonouveau: add an ioctl to return vram bar size.
Dave Airlie [Wed, 24 Jan 2024 03:50:58 +0000 (13:50 +1000)]
nouveau: add an ioctl to return vram bar size.

This returns the BAR resources size so userspace can make
decisions based on rebar support.

userspace using this has been proposed for nvk, but
it's a rather trivial uapi addition.

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
6 months agonouveau/gsp: add kconfig option to enable GSP paths by default
Dave Airlie [Wed, 14 Feb 2024 04:06:32 +0000 (14:06 +1000)]
nouveau/gsp: add kconfig option to enable GSP paths by default

Turing and Ampere will continue to use the old paths by default,
but we should allow distros to decide what the policy is.

Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240214040632.661069-1-airlied@gmail.com
6 months agoMerge tag 'drm-xe-fixes-2024-02-22' of https://gitlab.freedesktop.org/drm/xe/kernel...
Dave Airlie [Thu, 22 Feb 2024 23:44:44 +0000 (09:44 +1000)]
Merge tag 'drm-xe-fixes-2024-02-22' of https://gitlab.freedesktop.org/drm/xe/kernel into drm-fixes

UAPI Changes:
- Remove support for persistent exec_queues
- Drop a reduntant sysfs newline printout

Cross-subsystem Changes:

Core Changes:

Driver Changes:
- A three-patch fix for a VM_BIND rebind optimization path
- Fix a modpost warning on an xe KUNIT module

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Thomas Hellstrom <thomas.hellstrom@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/ZdcsNrxdWMMM417v@fedora
6 months agoMerge tag 'amd-drm-fixes-6.8-2024-02-22' of https://gitlab.freedesktop.org/agd5f...
Dave Airlie [Thu, 22 Feb 2024 22:36:46 +0000 (08:36 +1000)]
Merge tag 'amd-drm-fixes-6.8-2024-02-22' of https://gitlab.freedesktop.org/agd5f/linux into drm-fixes

amd-drm-fixes-6.8-2024-02-22:

amdgpu:
- Suspend/resume fixes
- Backlight error fix
- DCN 3.5 fixes
- Misc fixes

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240222195338.5809-1-alexander.deucher@amd.com
6 months agoMerge tag 'drm-intel-fixes-2024-02-22' of git://anongit.freedesktop.org/drm/drm-intel...
Dave Airlie [Thu, 22 Feb 2024 22:30:20 +0000 (08:30 +1000)]
Merge tag 'drm-intel-fixes-2024-02-22' of git://anongit.freedesktop.org/drm/drm-intel into drm-fixes

- Fixup for TV mode

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/ZdcwT9kltvEgJZZE@jlahtine-mobl.ger.corp.intel.com
6 months agoMerge tag 'drm-misc-fixes-2024-02-22' of git://anongit.freedesktop.org/drm/drm-misc...
Dave Airlie [Thu, 22 Feb 2024 22:09:45 +0000 (08:09 +1000)]
Merge tag 'drm-misc-fixes-2024-02-22' of git://anongit.freedesktop.org/drm/drm-misc into drm-fixes

A list handling fix and 64bit division on 32bit platform fix for the
drm/buddy allocator, a cast warning and an initialization fix for
nouveau, a bridge handling fix for meson, an initialisation fix for
ivpu, a SPARC build fix for fbdev, a double-free fix for ttm, and two
fence handling fixes for syncobj.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Maxime Ripard <mripard@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/gl2antuifidtzn3dfm426p7xwh5fxj23behagwh26owfnosh2w@gqoa7vj5prnh
6 months agoMerge tag 'block-6.8-2024-02-22' of git://git.kernel.dk/linux
Linus Torvalds [Thu, 22 Feb 2024 19:57:30 +0000 (11:57 -0800)]
Merge tag 'block-6.8-2024-02-22' of git://git.kernel.dk/linux

Pull block fixes from Jens Axboe:
 "Mostly just fixlets for md, but also a sed-opal parsing fix"

* tag 'block-6.8-2024-02-22' of git://git.kernel.dk/linux:
  block: sed-opal: handle empty atoms when parsing response
  md: Don't suspend the array for interrupted reshape
  md: Don't register sync_thread for reshape directly
  md: Make sure md_do_sync() will set MD_RECOVERY_DONE
  md: Don't ignore read-only array in md_check_recovery()
  md: Don't ignore suspended array in md_check_recovery()
  md: Fix missing release of 'active_io' for flush

6 months agoMerge tag 'for-linus-iommufd' of git://git.kernel.org/pub/scm/linux/kernel/git/jgg...
Linus Torvalds [Thu, 22 Feb 2024 19:53:09 +0000 (11:53 -0800)]
Merge tag 'for-linus-iommufd' of git://git./linux/kernel/git/jgg/iommufd

Pull iommufd fixes from Jason Gunthorpe:

 - Fix dirty tracking bitmap collection when using reporting bitmaps
   that are not neatly aligned to u64's or match the IO page table radix
   tree layout.

 - Add self tests to cover the cases that were found to be broken.

 - Add missing enforcement of invalidation type in the uapi.

 - Fix selftest config generation

* tag 'for-linus-iommufd' of git://git.kernel.org/pub/scm/linux/kernel/git/jgg/iommufd:
  selftests/iommu: fix the config fragment
  iommufd: Reject non-zero data_type if no data_len is provided
  iommufd/iova_bitmap: Consider page offset for the pages to be pinned
  iommufd/selftest: Add mock IO hugepages tests
  iommufd/selftest: Hugepage mock domain support
  iommufd/selftest: Refactor mock_domain_read_and_clear_dirty()
  iommufd/selftest: Refactor dirty bitmap tests
  iommufd/iova_bitmap: Handle recording beyond the mapped pages
  iommufd/selftest: Test u64 unaligned bitmaps
  iommufd/iova_bitmap: Switch iova_bitmap::bitmap to an u8 array
  iommufd/iova_bitmap: Bounds check mapped::pages access

6 months agoMerge tag 'platform-drivers-x86-v6.8-3' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 22 Feb 2024 19:47:07 +0000 (11:47 -0800)]
Merge tag 'platform-drivers-x86-v6.8-3' of git://git./linux/kernel/git/pdx86/platform-drivers-x86

Pull x86 platform driver fixes from Hans de Goede:
 "Regression fixes:

   - Fix INT0002 vGPIO events no longer working after 6.8 ACPI SCI
     changes

   - AMD-PMF: Fix laptops (e.g. Framework 13 AMD) hanging on suspend

   - x86-android-tablets: Fix touchscreen no longer working on Lenovo
     Yogabook

   - x86-android-tablets: Fix serdev instantiation regression

   - intel-vbtn: Fix ThinkPad X1 Tablet Gen2 no longer suspending

  Bug fixes:

   - think-lmi: Fix changing BIOS settings on Lenovo workstations

   - touchscreen_dmi: Fix Hi8 Air touchscreen data sometimes missing

   - AMD-PMF: Fix Smart PC support not working after suspend/resume

  Other misc small fixes"

* tag 'platform-drivers-x86-v6.8-3' of git://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86:
  platform/x86: thinkpad_acpi: Only update profile if successfully converted
  platform/x86: intel-vbtn: Stop calling "VBDL" from notify_handler
  platform/x86: x86-android-tablets: Fix acer_b1_750_goodix_gpios name
  platform/x86: x86-android-tablets: Fix serdev instantiation no longer working
  platform/x86: Add new get_serdev_controller() helper
  platform/x86: x86-android-tablets: Fix keyboard touchscreen on Lenovo Yogabook1 X90
  platform/x86/amd/pmf: Fix a potential race with policy binary sideload
  platform/x86/amd/pmf: Fixup error handling for amd_pmf_init_smart_pc()
  platform/x86/amd/pmf: Add debugging message for missing policy data
  platform/x86/amd/pmf: Fix a suspend hang on Framework 13
  platform/x86/amd/pmf: Fix TEE enact command failure after suspend and resume
  platform/x86/amd/pmf: Remove smart_pc_status enum
  platform/x86: touchscreen_dmi: Consolidate Goodix upside-down touchscreen data
  platform/x86: touchscreen_dmi: Allow partial (prefix) matches for ACPI names
  platform/x86: intel: int0002_vgpio: Pass IRQF_ONESHOT to request_irq()
  platform/x86: think-lmi: Fix password opcode ordering for workstations

6 months agoMerge tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Thu, 22 Feb 2024 19:44:20 +0000 (11:44 -0800)]
Merge tag 'clk-fixes-for-linus' of git://git./linux/kernel/git/clk/linux

Pull clk fixes from Stephen Boyd:
 "Here are some Samsung clk driver fixes I've been sitting on for far
  too long.

  They fix the bindings and clk driver for the Google GS101 SoC"

* tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux:
  clk: samsung: clk-gs101: comply with the new dt cmu_misc clock names
  dt-bindings: clock: gs101: rename cmu_misc clock-names

6 months agoMerge tag 'vfs-6.8-rc6.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs
Linus Torvalds [Thu, 22 Feb 2024 18:06:29 +0000 (10:06 -0800)]
Merge tag 'vfs-6.8-rc6.fixes' of git://git./linux/kernel/git/vfs/vfs

Pull vfs fixes from Christian Brauner:

 - Fix a memory leak in cachefiles

 - Restrict aio cancellations to I/O submitted through the aio
   interfaces as this is otherwise causing issues for I/O submitted
   via io_uring

 - Increase buffer for afs volume status to avoid overflow

 - Fix a missing zero-length check in unbuffered writes in the
   netfs library. If generic_write_checks() returns zero make
   netfs_unbuffered_write_iter() return right away

 - Prevent a leak in i_dio_count caused by netfs_begin_read() operating
   past i_size. It will return early and leave i_dio_count incremented

 - Account for ipv4 addresses as well as ipv6 addresses when processing
   incoming callbacks in afs

* tag 'vfs-6.8-rc6.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs:
  fs/aio: Restrict kiocb_set_cancel_fn() to I/O submitted via libaio
  afs: Increase buffer size in afs_update_volume_status()
  afs: Fix ignored callbacks over ipv4
  cachefiles: fix memory leak in cachefiles_add_cache()
  netfs: Fix missing zero-length check in unbuffered write
  netfs: Fix i_dio_count leak on DIO read past i_size

6 months agoMerge tag 'net-6.8.0-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Linus Torvalds [Thu, 22 Feb 2024 17:57:58 +0000 (09:57 -0800)]
Merge tag 'net-6.8.0-rc6' of git://git./linux/kernel/git/netdev/net

Pull networking fixes from Paolo Abeni:
 "Including fixes from bpf and netfilter.

  Current release - regressions:

   - af_unix: fix another unix GC hangup

  Previous releases - regressions:

   - core: fix a possible AF_UNIX deadlock

   - bpf: fix NULL pointer dereference in sk_psock_verdict_data_ready()

   - netfilter: nft_flow_offload: release dst in case direct xmit path
     is used

   - bridge: switchdev: ensure MDB events are delivered exactly once

   - l2tp: pass correct message length to ip6_append_data

   - dccp/tcp: unhash sk from ehash for tb2 alloc failure after
     check_estalblished()

   - tls: fixes for record type handling with PEEK

   - devlink: fix possible use-after-free and memory leaks in
     devlink_init()

  Previous releases - always broken:

   - bpf: fix an oops when attempting to read the vsyscall page through
     bpf_probe_read_kernel

   - sched: act_mirred: use the backlog for mirred ingress

   - netfilter: nft_flow_offload: fix dst refcount underflow

   - ipv6: sr: fix possible use-after-free and null-ptr-deref

   - mptcp: fix several data races

   - phonet: take correct lock to peek at the RX queue

  Misc:

   - handful of fixes and reliability improvements for selftests"

* tag 'net-6.8.0-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (72 commits)
  l2tp: pass correct message length to ip6_append_data
  net: phy: realtek: Fix rtl8211f_config_init() for RTL8211F(D)(I)-VD-CG PHY
  selftests: ioam: refactoring to align with the fix
  Fix write to cloned skb in ipv6_hop_ioam()
  phonet/pep: fix racy skb_queue_empty() use
  phonet: take correct lock to peek at the RX queue
  net: sparx5: Add spinlock for frame transmission from CPU
  net/sched: flower: Add lock protection when remove filter handle
  devlink: fix port dump cmd type
  net: stmmac: Fix EST offset for dwmac 5.10
  tools: ynl: don't leak mcast_groups on init error
  tools: ynl: make sure we always pass yarg to mnl_cb_run
  net: mctp: put sock on tag allocation failure
  netfilter: nf_tables: use kzalloc for hook allocation
  netfilter: nf_tables: register hooks last when adding new chain/flowtable
  netfilter: nft_flow_offload: release dst in case direct xmit path is used
  netfilter: nft_flow_offload: reset dst in route object after setting up flow
  netfilter: nf_tables: set dormant flag on hook register failure
  selftests: tls: add test for peeking past a record of a different type
  selftests: tls: add test for merging of same-type control messages
  ...

6 months agodrm/amdgpu: Fix the runtime resume failure issue
Ma Jun [Wed, 21 Feb 2024 09:16:49 +0000 (17:16 +0800)]
drm/amdgpu: Fix the runtime resume failure issue

Don't set power state flag when system enter runtime suspend,
or it may cause runtime resume failure issue.

Fixes: 3a9626c816db ("drm/amd: Stop evicting resources on APUs in suspend")
Signed-off-by: Ma Jun <Jun.Ma2@amd.com>
Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
6 months agodrm/amd/display: fix null-pointer dereference on edid reading
Melissa Wen [Fri, 16 Feb 2024 12:23:19 +0000 (09:23 -0300)]
drm/amd/display: fix null-pointer dereference on edid reading

Use i2c adapter when there isn't aux_mode in dc_link to fix a
null-pointer derefence that happens when running
igt@kms_force_connector_basic in a system with DCN2.1 and HDMI connector
detected as below:

[  +0.178146] BUG: kernel NULL pointer dereference, address: 00000000000004c0
[  +0.000010] #PF: supervisor read access in kernel mode
[  +0.000005] #PF: error_code(0x0000) - not-present page
[  +0.000004] PGD 0 P4D 0
[  +0.000006] Oops: 0000 [#1] PREEMPT SMP NOPTI
[  +0.000006] CPU: 15 PID: 2368 Comm: kms_force_conne Not tainted 6.5.0-asdn+ #152
[  +0.000005] Hardware name: HP HP ENVY x360 Convertible 13-ay1xxx/8929, BIOS F.01 07/14/2021
[  +0.000004] RIP: 0010:i2c_transfer+0xd/0x100
[  +0.000011] Code: ea fc ff ff 66 0f 1f 84 00 00 00 00 00 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 f3 0f 1e fa 0f 1f 44 00 00 41 54 55 53 <48> 8b 47 10 48 89 fb 48 83 38 00 0f 84 b3 00 00 00 83 3d 2f 80 16
[  +0.000004] RSP: 0018:ffff9c4f89c0fad0 EFLAGS: 00010246
[  +0.000005] RAX: 0000000000000000 RBX: 0000000000000005 RCX: 0000000000000080
[  +0.000003] RDX: 0000000000000002 RSI: ffff9c4f89c0fb20 RDI: 00000000000004b0
[  +0.000003] RBP: ffff9c4f89c0fb80 R08: 0000000000000080 R09: ffff8d8e0b15b980
[  +0.000003] R10: 00000000000380e0 R11: 0000000000000000 R12: 0000000000000080
[  +0.000002] R13: 0000000000000002 R14: ffff9c4f89c0fb0e R15: ffff9c4f89c0fb0f
[  +0.000004] FS:  00007f9ad2176c40(0000) GS:ffff8d90fe9c0000(0000) knlGS:0000000000000000
[  +0.000003] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  +0.000004] CR2: 00000000000004c0 CR3: 0000000121bc4000 CR4: 0000000000750ee0
[  +0.000003] PKRU: 55555554
[  +0.000003] Call Trace:
[  +0.000006]  <TASK>
[  +0.000006]  ? __die+0x23/0x70
[  +0.000011]  ? page_fault_oops+0x17d/0x4c0
[  +0.000008]  ? preempt_count_add+0x6e/0xa0
[  +0.000008]  ? srso_alias_return_thunk+0x5/0x7f
[  +0.000011]  ? exc_page_fault+0x7f/0x180
[  +0.000009]  ? asm_exc_page_fault+0x26/0x30
[  +0.000013]  ? i2c_transfer+0xd/0x100
[  +0.000010]  drm_do_probe_ddc_edid+0xc2/0x140 [drm]
[  +0.000067]  ? srso_alias_return_thunk+0x5/0x7f
[  +0.000006]  ? _drm_do_get_edid+0x97/0x3c0 [drm]
[  +0.000043]  ? __pfx_drm_do_probe_ddc_edid+0x10/0x10 [drm]
[  +0.000042]  edid_block_read+0x3b/0xd0 [drm]
[  +0.000043]  _drm_do_get_edid+0xb6/0x3c0 [drm]
[  +0.000041]  ? __pfx_drm_do_probe_ddc_edid+0x10/0x10 [drm]
[  +0.000043]  drm_edid_read_custom+0x37/0xd0 [drm]
[  +0.000044]  amdgpu_dm_connector_mode_valid+0x129/0x1d0 [amdgpu]
[  +0.000153]  drm_connector_mode_valid+0x3b/0x60 [drm_kms_helper]
[  +0.000000]  __drm_helper_update_and_validate+0xfe/0x3c0 [drm_kms_helper]
[  +0.000000]  ? amdgpu_dm_connector_get_modes+0xb6/0x520 [amdgpu]
[  +0.000000]  ? srso_alias_return_thunk+0x5/0x7f
[  +0.000000]  drm_helper_probe_single_connector_modes+0x2ab/0x540 [drm_kms_helper]
[  +0.000000]  status_store+0xb2/0x1f0 [drm]
[  +0.000000]  kernfs_fop_write_iter+0x136/0x1d0
[  +0.000000]  vfs_write+0x24d/0x440
[  +0.000000]  ksys_write+0x6f/0xf0
[  +0.000000]  do_syscall_64+0x60/0xc0
[  +0.000000]  ? srso_alias_return_thunk+0x5/0x7f
[  +0.000000]  ? syscall_exit_to_user_mode+0x2b/0x40
[  +0.000000]  ? srso_alias_return_thunk+0x5/0x7f
[  +0.000000]  ? do_syscall_64+0x6c/0xc0
[  +0.000000]  ? do_syscall_64+0x6c/0xc0
[  +0.000000]  entry_SYSCALL_64_after_hwframe+0x6e/0xd8
[  +0.000000] RIP: 0033:0x7f9ad46b4b00
[  +0.000000] Code: 40 00 48 8b 15 19 b3 0d 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b7 0f 1f 00 80 3d e1 3a 0e 00 00 74 17 b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 58 c3 0f 1f 80 00 00 00 00 48 83 ec 28 48 89
[  +0.000000] RSP: 002b:00007ffcbd3bd6d8 EFLAGS: 00000202 ORIG_RAX: 0000000000000001
[  +0.000000] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f9ad46b4b00
[  +0.000000] RDX: 0000000000000002 RSI: 00007f9ad48a7417 RDI: 0000000000000009
[  +0.000000] RBP: 0000000000000002 R08: 0000000000000064 R09: 0000000000000000
[  +0.000000] R10: 0000000000000000 R11: 0000000000000202 R12: 00007f9ad48a7417
[  +0.000000] R13: 0000000000000009 R14: 00007ffcbd3bd760 R15: 0000000000000001
[  +0.000000]  </TASK>
[  +0.000000] Modules linked in: ctr ccm rfcomm snd_seq_dummy snd_hrtimer snd_seq snd_seq_device cmac algif_hash algif_skcipher af_alg bnep btusb btrtl btbcm btintel btmtk bluetooth uvcvideo videobuf2_vmalloc sha3_generic videobuf2_memops uvc jitterentropy_rng videobuf2_v4l2 videodev drbg videobuf2_common ansi_cprng mc ecdh_generic ecc qrtr binfmt_misc hid_sensor_accel_3d hid_sensor_magn_3d hid_sensor_gyro_3d hid_sensor_trigger industrialio_triggered_buffer kfifo_buf industrialio snd_ctl_led joydev hid_sensor_iio_common rtw89_8852ae rtw89_8852a rtw89_pci snd_hda_codec_realtek rtw89_core snd_hda_codec_generic intel_rapl_msr ledtrig_audio intel_rapl_common snd_hda_codec_hdmi mac80211 snd_hda_intel snd_intel_dspcfg kvm_amd snd_hda_codec snd_soc_dmic snd_acp3x_rn snd_acp3x_pdm_dma libarc4 snd_hwdep snd_soc_core kvm snd_hda_core cfg80211 snd_pci_acp6x snd_pcm nls_ascii snd_timer hp_wmi snd_pci_acp5x nls_cp437 snd_rn_pci_acp3x ucsi_acpi sparse_keymap ccp snd platform_profile snd_acp_config typec_ucsi irqbypass vfat sp5100_tco
[  +0.000000]  snd_soc_acpi fat rapl pcspkr wmi_bmof roles rfkill rng_core snd_pci_acp3x soundcore k10temp watchdog typec battery ac amd_pmc acpi_tad button hid_sensor_hub hid_multitouch evdev serio_raw msr parport_pc ppdev lp parport fuse loop efi_pstore configfs ip_tables x_tables autofs4 ext4 crc16 mbcache jbd2 btrfs blake2b_generic dm_crypt dm_mod efivarfs raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx libcrc32c crc32c_generic xor raid6_pq raid1 raid0 multipath linear md_mod amdgpu amdxcp i2c_algo_bit drm_ttm_helper ttm crc32_pclmul crc32c_intel drm_exec gpu_sched drm_suballoc_helper nvme ghash_clmulni_intel drm_buddy drm_display_helper sha512_ssse3 nvme_core ahci xhci_pci sha512_generic hid_generic xhci_hcd libahci rtsx_pci_sdmmc t10_pi i2c_hid_acpi drm_kms_helper i2c_hid mmc_core libata aesni_intel crc64_rocksoft_generic crypto_simd amd_sfh crc64_rocksoft scsi_mod usbcore cryptd crc_t10dif cec drm crct10dif_generic hid rtsx_pci crct10dif_pclmul scsi_common rc_core crc64 i2c_piix4
[  +0.000000]  usb_common crct10dif_common video wmi
[  +0.000000] CR2: 00000000000004c0
[  +0.000000] ---[ end trace 0000000000000000 ]---

Fixes: 0e859faf8670 ("drm/amd/display: Remove unwanted drm edid references")
Signed-off-by: Melissa Wen <mwen@igalia.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 months agodrm/amd/display: Fix memory leak in dm_sw_fini()
Armin Wolf [Tue, 13 Feb 2024 00:50:50 +0000 (01:50 +0100)]
drm/amd/display: Fix memory leak in dm_sw_fini()

After destroying dmub_srv, the memory associated with it is
not freed, causing a memory leak:

unreferenced object 0xffff896302b45800 (size 1024):
  comm "(udev-worker)", pid 222, jiffies 4294894636
  hex dump (first 32 bytes):
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
  backtrace (crc 6265fd77):
    [<ffffffff993495ed>] kmalloc_trace+0x29d/0x340
    [<ffffffffc0ea4a94>] dm_dmub_sw_init+0xb4/0x450 [amdgpu]
    [<ffffffffc0ea4e55>] dm_sw_init+0x15/0x2b0 [amdgpu]
    [<ffffffffc0ba8557>] amdgpu_device_init+0x1417/0x24e0 [amdgpu]
    [<ffffffffc0bab285>] amdgpu_driver_load_kms+0x15/0x190 [amdgpu]
    [<ffffffffc0ba09c7>] amdgpu_pci_probe+0x187/0x4e0 [amdgpu]
    [<ffffffff9968fd1e>] local_pci_probe+0x3e/0x90
    [<ffffffff996918a3>] pci_device_probe+0xc3/0x230
    [<ffffffff99805872>] really_probe+0xe2/0x480
    [<ffffffff99805c98>] __driver_probe_device+0x78/0x160
    [<ffffffff99805daf>] driver_probe_device+0x1f/0x90
    [<ffffffff9980601e>] __driver_attach+0xce/0x1c0
    [<ffffffff99803170>] bus_for_each_dev+0x70/0xc0
    [<ffffffff99804822>] bus_add_driver+0x112/0x210
    [<ffffffff99807245>] driver_register+0x55/0x100
    [<ffffffff990012d1>] do_one_initcall+0x41/0x300

Fix this by freeing dmub_srv after destroying it.

Fixes: 743b9786b14a ("drm/amd/display: Hook up the DMUB service in DM")
Signed-off-by: Armin Wolf <W_Armin@gmx.de>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 months agodrm/amd/display: fix input states translation error for dcn35 & dcn351
Swapnil Patel [Tue, 6 Feb 2024 16:40:20 +0000 (11:40 -0500)]
drm/amd/display: fix input states translation error for dcn35 & dcn351

[Why]
Currently there is an error while translating input clock sates into
output clock states. The highest fclk setting from output sates is
being dropped because of this error.

[How]
For dcn35 and dcn351, make output_states equal to input states.

Reviewed-by: Charlene Liu <charlene.liu@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Swapnil Patel <swapnil.patel@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 months agodrm/amd/display: Fix potential null pointer dereference in dc_dmub_srv
Srinivasan Shanmugam [Mon, 19 Feb 2024 06:13:16 +0000 (11:43 +0530)]
drm/amd/display: Fix potential null pointer dereference in dc_dmub_srv

Fixes potential null pointer dereference warnings in the
dc_dmub_srv_cmd_list_queue_execute() and dc_dmub_srv_is_hw_pwr_up()
functions.

In both functions, the 'dc_dmub_srv' variable was being dereferenced
before it was checked for null. This could lead to a null pointer
dereference if 'dc_dmub_srv' is null. The fix is to check if
'dc_dmub_srv' is null before dereferencing it.

Thus moving the null checks for 'dc_dmub_srv' to the beginning of the
functions to ensure that 'dc_dmub_srv' is not null when it is
dereferenced.

Found by smatch & thus fixing the below:
drivers/gpu/drm/amd/amdgpu/../display/dc/dc_dmub_srv.c:133 dc_dmub_srv_cmd_list_queue_execute() warn: variable dereferenced before check 'dc_dmub_srv' (see line 128)
drivers/gpu/drm/amd/amdgpu/../display/dc/dc_dmub_srv.c:1167 dc_dmub_srv_is_hw_pwr_up() warn: variable dereferenced before check 'dc_dmub_srv' (see line 1164)

Fixes: 028bac583449 ("drm/amd/display: decouple dmcub execution to reduce lock granularity")
Fixes: 65138eb72e1f ("drm/amd/display: Add DCN35 DMUB")
Cc: JinZe.Xu <jinze.xu@amd.com>
Cc: Hersen Wu <hersenxs.wu@amd.com>
Cc: Josip Pavic <josip.pavic@amd.com>
Cc: Roman Li <roman.li@amd.com>
Cc: Qingqing Zhuo <Qingqing.Zhuo@amd.com>
Cc: Harry Wentland <Harry.Wentland@amd.com>
Cc: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Cc: Aurabindo Pillai <aurabindo.pillai@amd.com>
Cc: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com>
Reviewed-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 months agoMerge tag 'trace-v6.8-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/trace...
Linus Torvalds [Thu, 22 Feb 2024 17:23:22 +0000 (09:23 -0800)]
Merge tag 'trace-v6.8-rc5' of git://git./linux/kernel/git/trace/linux-trace

Pull tracing fix from Steven Rostedt:

 - While working on the ring buffer I noticed that the counter used for
   knowing where the end of the data is on a sub-buffer was not a full
   "int" but just 20 bits. It was masked out to 0xfffff.

   With the new code that allows the user to change the size of the
   sub-buffer, it is theoretically possible to ask for a size bigger
   than 2^20. If that happens, unexpected results may occur as there's
   no code checking if the counter overflowed the 20 bits of the write
   mask. There are other checks to make sure events fit in the
   sub-buffer, but if the sub-buffer itself is too big, that is not
   checked.

   Add a check in the resize of the sub-buffer to make sure that it
   never goes beyond the size of the counter that holds how much data is
   on it.

* tag 'trace-v6.8-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  ring-buffer: Do not let subbuf be bigger than write mask

6 months agodrm/amd/display: Only allow dig mapping to pwrseq in new asic
Lewis Huang [Wed, 31 Jan 2024 09:20:17 +0000 (17:20 +0800)]
drm/amd/display: Only allow dig mapping to pwrseq in new asic

[Why]
The old asic only have 1 pwrseq hw.
We don't need to map the diginst to pwrseq inst in old asic.

[How]
1. Only mapping dig to pwrseq for new asic.
2. Move mapping function into dcn specific panel control component

Cc: Stable <stable@vger.kernel.org> # v6.6+
Cc: Mario Limonciello <mario.limonciello@amd.com>
Link: https://gitlab.freedesktop.org/drm/amd/-/issues/3122
Reviewed-by: Anthony Koo <anthony.koo@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Lewis Huang <lewis.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 months agodrm/amd/display: adjust few initialization order in dm
Wayne Lin [Fri, 2 Feb 2024 09:34:11 +0000 (17:34 +0800)]
drm/amd/display: adjust few initialization order in dm

[Why]
Observe error message "Can't retrieve aconnector in hpd_rx_irq_offload_work"
when boot up with a mst tbt4 dock connected. After analyzing, there are few
parts needed to be adjusted:

1. hpd_rx_offload_wq[].aconnector is not initialzed before the dmub outbox
hpd_irq handler get registered which causes the error message.

2. registeration of hpd and hpd_rx_irq event for usb4 dp tunneling is not
aligned with legacy interface sequence

[How]
Put DMUB_NOTIFICATION_HPD and DMUB_NOTIFICATION_HPD_IRQ handler
registration into register_hpd_handlers() to align other interfaces and
get hpd_rx_offload_wq[].aconnector initialized earlier than that.

Leave DMUB_NOTIFICATION_AUX_REPLY registered as it was since we need that
while calling dc_link_detect(). USB4 connection status will be proactively
detected by dc_link_detect_connection_type() in amdgpu_dm_initialize_drm_device()

Cc: Stable <stable@vger.kernel.org>
Reviewed-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 months agoselftests/iommu: fix the config fragment
Muhammad Usama Anjum [Thu, 22 Feb 2024 07:49:33 +0000 (12:49 +0500)]
selftests/iommu: fix the config fragment

The config fragment doesn't follow the correct format to enable those
config options which make the config options getting missed while
merging with other configs.

➜ merge_config.sh -m .config tools/testing/selftests/iommu/config
Using .config as base
Merging tools/testing/selftests/iommu/config
➜ make olddefconfig
.config:5295:warning: unexpected data: CONFIG_IOMMUFD
.config:5296:warning: unexpected data: CONFIG_IOMMUFD_TEST

While at it, add CONFIG_FAULT_INJECTION as well which is needed for
CONFIG_IOMMUFD_TEST. If CONFIG_FAULT_INJECTION isn't present in base
config (such as x86 defconfig), CONFIG_IOMMUFD_TEST doesn't get enabled.

Fixes: 57f0988706fe ("iommufd: Add a selftest")
Link: https://lore.kernel.org/r/20240222074934.71380-1-usama.anjum@collabora.com
Signed-off-by: Muhammad Usama Anjum <usama.anjum@collabora.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
6 months agodrm/syncobj: handle NULL fence in syncobj_eventfd_entry_func
Erik Kurzinger [Wed, 21 Feb 2024 18:44:28 +0000 (10:44 -0800)]
drm/syncobj: handle NULL fence in syncobj_eventfd_entry_func

During syncobj_eventfd_entry_func, dma_fence_chain_find_seqno may set
the fence to NULL if the given seqno is signaled and a later seqno has
already been submitted. In that case, the eventfd should be signaled
immediately which currently does not happen.

This is a similar issue to the one addressed by commit b19926d4f3a6
("drm/syncobj: Deal with signalled fences in drm_syncobj_find_fence.").

As a fix, if the return value of dma_fence_chain_find_seqno indicates
success but it sets the fence to NULL, we will assign a stub fence to
ensure the following code still signals the eventfd.

v1 -> v2: assign a stub fence instead of signaling the eventfd

Signed-off-by: Erik Kurzinger <ekurzinger@nvidia.com>
Fixes: c7a472297169 ("drm/syncobj: add IOCTL to register an eventfd")
Signed-off-by: Simon Ser <contact@emersion.fr>
Link: https://patchwork.freedesktop.org/patch/msgid/20240221184527.37667-1-ekurzinger@nvidia.com
6 months agodrm/syncobj: call drm_syncobj_fence_add_wait when WAIT_AVAILABLE flag is set
Erik Kurzinger [Fri, 19 Jan 2024 16:32:06 +0000 (08:32 -0800)]
drm/syncobj: call drm_syncobj_fence_add_wait when WAIT_AVAILABLE flag is set

When waiting for a syncobj timeline point whose fence has not yet been
submitted with the WAIT_FOR_SUBMIT flag, a callback is registered using
drm_syncobj_fence_add_wait and the thread is put to sleep until the
timeout expires. If the fence is submitted before then,
drm_syncobj_add_point will wake up the sleeping thread immediately which
will proceed to wait for the fence to be signaled.

However, if the WAIT_AVAILABLE flag is used instead,
drm_syncobj_fence_add_wait won't get called, meaning the waiting thread
will always sleep for the full timeout duration, even if the fence gets
submitted earlier. If it turns out that the fence *has* been submitted
by the time it eventually wakes up, it will still indicate to userspace
that the wait completed successfully (it won't return -ETIME), but it
will have taken much longer than it should have.

To fix this, we must call drm_syncobj_fence_add_wait if *either* the
WAIT_FOR_SUBMIT flag or the WAIT_AVAILABLE flag is set. The only
difference being that with WAIT_FOR_SUBMIT we will also wait for the
fence to be signaled after it has been submitted while with
WAIT_AVAILABLE we will return immediately.

IGT test patch: https://lists.freedesktop.org/archives/igt-dev/2024-January/067537.html

v1 -> v2: adjust lockdep_assert_none_held_once condition

(cherry picked from commit 8c44ea81634a4a337df70a32621a5f3791be23df)

Fixes: 01d6c3578379 ("drm/syncobj: add support for timeline point wait v8")
Signed-off-by: Erik Kurzinger <ekurzinger@nvidia.com>
Signed-off-by: Simon Ser <contact@emersion.fr>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Simon Ser <contact@emersion.fr>
Link: https://patchwork.freedesktop.org/patch/msgid/20240119163208.3723457-1-ekurzinger@nvidia.com
6 months agodrm/ttm: Fix an invalid freeing on already freed page in error path
Thomas Hellström [Wed, 21 Feb 2024 07:33:24 +0000 (08:33 +0100)]
drm/ttm: Fix an invalid freeing on already freed page in error path

If caching mode change fails due to, for example, OOM we
free the allocated pages in a two-step process. First the pages
for which the caching change has already succeeded. Secondly
the pages for which a caching change did not succeed.

However the second step was incorrectly freeing the pages already
freed in the first step.

Fix.

Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Fixes: 379989e7cbdc ("drm/ttm/pool: Fix ttm_pool_alloc error path")
Cc: Christian König <christian.koenig@amd.com>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Christian Koenig <christian.koenig@amd.com>
Cc: Huang Rui <ray.huang@amd.com>
Cc: dri-devel@lists.freedesktop.org
Cc: <stable@vger.kernel.org> # v6.4+
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240221073324.3303-1-thomas.hellstrom@linux.intel.com
6 months agol2tp: pass correct message length to ip6_append_data
Tom Parkin [Tue, 20 Feb 2024 12:21:56 +0000 (12:21 +0000)]
l2tp: pass correct message length to ip6_append_data

l2tp_ip6_sendmsg needs to avoid accounting for the transport header
twice when splicing more data into an already partially-occupied skbuff.

To manage this, we check whether the skbuff contains data using
skb_queue_empty when deciding how much data to append using
ip6_append_data.

However, the code which performed the calculation was incorrect:

     ulen = len + skb_queue_empty(&sk->sk_write_queue) ? transhdrlen : 0;

...due to C operator precedence, this ends up setting ulen to
transhdrlen for messages with a non-zero length, which results in
corrupted packets on the wire.

Add parentheses to correct the calculation in line with the original
intent.

Fixes: 9d4c75800f61 ("ipv4, ipv6: Fix handling of transhdrlen in __ip{,6}_append_data()")
Cc: David Howells <dhowells@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Tom Parkin <tparkin@katalix.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://lore.kernel.org/r/20240220122156.43131-1-tparkin@katalix.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
6 months agoMerge tag 'nf-24-02-22' of git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf
Paolo Abeni [Thu, 22 Feb 2024 09:20:50 +0000 (10:20 +0100)]
Merge tag 'nf-24-02-22' of git://git./linux/kernel/git/netfilter/nf

Pablo Neira Ayuso says:

====================
Netfilter fixes for net

The following patchset contains Netfilter fixes for net:

1) If user requests to wake up a table and hook fails, restore the
   dormant flag from the error path, from Florian Westphal.

2) Reset dst after transferring it to the flow object, otherwise dst
   gets released twice from the error path.

3) Release dst in case the flowtable selects a direct xmit path, eg.
   transmission to bridge port. Otherwise, dst is memleaked.

4) Register basechain and flowtable hooks at the end of the command.
   Error path releases these datastructure without waiting for the
   rcu grace period.

5) Use kzalloc() to initialize struct nft_hook to fix a KMSAN report
   on access to hook type, also from Florian Westphal.

netfilter pull request 24-02-22

* tag 'nf-24-02-22' of git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf:
  netfilter: nf_tables: use kzalloc for hook allocation
  netfilter: nf_tables: register hooks last when adding new chain/flowtable
  netfilter: nft_flow_offload: release dst in case direct xmit path is used
  netfilter: nft_flow_offload: reset dst in route object after setting up flow
  netfilter: nf_tables: set dormant flag on hook register failure
====================

Link: https://lore.kernel.org/r/20240222000843.146665-1-pablo@netfilter.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
6 months agoMerge tag 'for-netdev' of https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf
Paolo Abeni [Thu, 22 Feb 2024 09:04:46 +0000 (10:04 +0100)]
Merge tag 'for-netdev' of https://git./linux/kernel/git/bpf/bpf

Daniel Borkmann says:

====================
pull-request: bpf 2024-02-22

The following pull-request contains BPF updates for your *net* tree.

We've added 11 non-merge commits during the last 24 day(s) which contain
a total of 15 files changed, 217 insertions(+), 17 deletions(-).

The main changes are:

1) Fix a syzkaller-triggered oops when attempting to read the vsyscall
   page through bpf_probe_read_kernel and friends, from Hou Tao.

2) Fix a kernel panic due to uninitialized iter position pointer in
   bpf_iter_task, from Yafang Shao.

3) Fix a race between bpf_timer_cancel_and_free and bpf_timer_cancel,
   from Martin KaFai Lau.

4) Fix a xsk warning in skb_add_rx_frag() (under CONFIG_DEBUG_NET)
   due to incorrect truesize accounting, from Sebastian Andrzej Siewior.

5) Fix a NULL pointer dereference in sk_psock_verdict_data_ready,
   from Shigeru Yoshida.

6) Fix a resolve_btfids warning when bpf_cpumask symbol cannot be
   resolved, from Hari Bathini.

bpf-for-netdev

* tag 'for-netdev' of https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf:
  bpf, sockmap: Fix NULL pointer dereference in sk_psock_verdict_data_ready()
  selftests/bpf: Add negtive test cases for task iter
  bpf: Fix an issue due to uninitialized bpf_iter_task
  selftests/bpf: Test racing between bpf_timer_cancel_and_free and bpf_timer_cancel
  bpf: Fix racing between bpf_timer_cancel_and_free and bpf_timer_cancel
  selftest/bpf: Test the read of vsyscall page under x86-64
  x86/mm: Disallow vsyscall page read for copy_from_kernel_nofault()
  x86/mm: Move is_vsyscall_vaddr() into asm/vsyscall.h
  bpf, scripts: Correct GPL license name
  xsk: Add truesize to skb_add_rx_frag().
  bpf: Fix warning for bpf_cpumask in verifier
====================

Link: https://lore.kernel.org/r/20240221231826.1404-1-daniel@iogearbox.net
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
6 months agonet: phy: realtek: Fix rtl8211f_config_init() for RTL8211F(D)(I)-VD-CG PHY
Siddharth Vadapalli [Tue, 20 Feb 2024 07:00:07 +0000 (12:30 +0530)]
net: phy: realtek: Fix rtl8211f_config_init() for RTL8211F(D)(I)-VD-CG PHY

Commit bb726b753f75 ("net: phy: realtek: add support for
RTL8211F(D)(I)-VD-CG") extended support of the driver from the existing
support for RTL8211F(D)(I)-CG PHY to the newer RTL8211F(D)(I)-VD-CG PHY.

While that commit indicated that the RTL8211F_PHYCR2 register is not
supported by the "VD-CG" PHY model and therefore updated the corresponding
section in rtl8211f_config_init() to be invoked conditionally, the call to
"genphy_soft_reset()" was left as-is, when it should have also been invoked
conditionally. This is because the call to "genphy_soft_reset()" was first
introduced by the commit 0a4355c2b7f8 ("net: phy: realtek: add dt property
to disable CLKOUT clock") since the RTL8211F guide indicates that a PHY
reset should be issued after setting bits in the PHYCR2 register.

As the PHYCR2 register is not applicable to the "VD-CG" PHY model, fix the
rtl8211f_config_init() function by invoking "genphy_soft_reset()"
conditionally based on the presence of the "PHYCR2" register.

Fixes: bb726b753f75 ("net: phy: realtek: add support for RTL8211F(D)(I)-VD-CG")
Signed-off-by: Siddharth Vadapalli <s-vadapalli@ti.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://lore.kernel.org/r/20240220070007.968762-1-s-vadapalli@ti.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
6 months agoMerge branch 'ioam6-fix-write-to-cloned-skb-s'
Paolo Abeni [Thu, 22 Feb 2024 08:28:06 +0000 (09:28 +0100)]
Merge branch 'ioam6-fix-write-to-cloned-skb-s'

Justin Iurman says:

====================
ioam6: fix write to cloned skb's

Make sure the IOAM data insertion is not applied on cloned skb's. As a
consequence, ioam selftests needed a refactoring.
====================

Link: https://lore.kernel.org/r/20240219135255.15429-1-justin.iurman@uliege.be
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
6 months agoselftests: ioam: refactoring to align with the fix
Justin Iurman [Mon, 19 Feb 2024 13:52:55 +0000 (14:52 +0100)]
selftests: ioam: refactoring to align with the fix

ioam6_parser uses a packet socket. After the fix to prevent writing to
cloned skb's, the receiver does not see its IOAM data anymore, which
makes input/forward ioam-selftests to fail. As a workaround,
ioam6_parser now uses an IPv6 raw socket and leverages ancillary data to
get hop-by-hop options. As a consequence, the hook is "after" the IOAM
data insertion by the receiver and all tests are working again.

Signed-off-by: Justin Iurman <justin.iurman@uliege.be>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
6 months agoFix write to cloned skb in ipv6_hop_ioam()
Justin Iurman [Mon, 19 Feb 2024 13:52:54 +0000 (14:52 +0100)]
Fix write to cloned skb in ipv6_hop_ioam()

ioam6_fill_trace_data() writes inside the skb payload without ensuring
it's writeable (e.g., not cloned). This function is called both from the
input and output path. The output path (ioam6_iptunnel) already does the
check. This commit provides a fix for the input path, inside
ipv6_hop_ioam(). It also updates ip6_parse_tlv() to refresh the network
header pointer ("nh") when returning from ipv6_hop_ioam().

Fixes: 9ee11f0fff20 ("ipv6: ioam: Data plane support for Pre-allocated Trace")
Reported-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Justin Iurman <justin.iurman@uliege.be>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
6 months agophonet/pep: fix racy skb_queue_empty() use
Rémi Denis-Courmont [Sun, 18 Feb 2024 08:12:14 +0000 (10:12 +0200)]
phonet/pep: fix racy skb_queue_empty() use

The receive queues are protected by their respective spin-lock, not
the socket lock. This could lead to skb_peek() unexpectedly
returning NULL or a pointer to an already dequeued socket buffer.

Fixes: 9641458d3ec4 ("Phonet: Pipe End Point for Phonet Pipes protocol")
Signed-off-by: Rémi Denis-Courmont <courmisch@gmail.com>
Link: https://lore.kernel.org/r/20240218081214.4806-2-remi@remlab.net
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
6 months agophonet: take correct lock to peek at the RX queue
Rémi Denis-Courmont [Sun, 18 Feb 2024 08:12:13 +0000 (10:12 +0200)]
phonet: take correct lock to peek at the RX queue

The receive queue is protected by its embedded spin-lock, not the
socket lock, so we need the former lock here (and only that one).

Fixes: 107d0d9b8d9a ("Phonet: Phonet datagram transport protocol")
Reported-by: Luosili <rootlab@huawei.com>
Signed-off-by: Rémi Denis-Courmont <courmisch@gmail.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20240218081214.4806-1-remi@remlab.net
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
6 months agonet: sparx5: Add spinlock for frame transmission from CPU
Horatiu Vultur [Mon, 19 Feb 2024 08:00:43 +0000 (09:00 +0100)]
net: sparx5: Add spinlock for frame transmission from CPU

Both registers used when doing manual injection or fdma injection are
shared between all the net devices of the switch. It was noticed that
when having two process which each of them trying to inject frames on
different ethernet ports, that the HW started to behave strange, by
sending out more frames then expected. When doing fdma injection it is
required to set the frame in the DCB and then make sure that the next
pointer of the last DCB is invalid. But because there is no locks for
this, then easily this pointer between the DCB can be broken and then it
would create a loop of DCBs. And that means that the HW will
continuously transmit these frames in a loop. Until the SW will break
this loop.
Therefore to fix this issue, add a spin lock for when accessing the
registers for manual or fdma injection.

Signed-off-by: Horatiu Vultur <horatiu.vultur@microchip.com>
Reviewed-by: Daniel Machon <daniel.machon@microchip.com>
Fixes: f3cad2611a77 ("net: sparx5: add hostmode with phylink support")
Link: https://lore.kernel.org/r/20240219080043.1561014-1-horatiu.vultur@microchip.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
6 months agonet/sched: flower: Add lock protection when remove filter handle
Jianbo Liu [Tue, 20 Feb 2024 08:59:28 +0000 (08:59 +0000)]
net/sched: flower: Add lock protection when remove filter handle

As IDR can't protect itself from the concurrent modification, place
idr_remove() under the protection of tp->lock.

Fixes: 08a0063df3ae ("net/sched: flower: Move filter handle initialization earlier")
Signed-off-by: Jianbo Liu <jianbol@nvidia.com>
Reviewed-by: Cosmin Ratiu <cratiu@nvidia.com>
Reviewed-by: Gal Pressman <gal@nvidia.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Link: https://lore.kernel.org/r/20240220085928.9161-1-jianbol@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
6 months agodevlink: fix port dump cmd type
Jiri Pirko [Tue, 20 Feb 2024 07:52:45 +0000 (08:52 +0100)]
devlink: fix port dump cmd type

Unlike other commands, due to a c&p error, port dump fills-up cmd with
wrong value, different from port-get request cmd, port-get doit reply
and port notification.

Fix it by filling cmd with value DEVLINK_CMD_PORT_NEW.

Skimmed through devlink userspace implementations, none of them cares
about this cmd value. Only ynl, for which, this is actually a fix, as it
expects doit and dumpit ops rsp_value to be the same.

Omit the fixes tag, even thought this is fix, better to target this for
next release.

Fixes: bfcd3a466172 ("Introduce devlink infrastructure")
Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Link: https://lore.kernel.org/r/20240220075245.75416-1-jiri@resnulli.us
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
6 months agonet: stmmac: Fix EST offset for dwmac 5.10
Kurt Kanzenbach [Tue, 20 Feb 2024 08:22:46 +0000 (09:22 +0100)]
net: stmmac: Fix EST offset for dwmac 5.10

Fix EST offset for dwmac 5.10.

Currently configuring Qbv doesn't work as expected. The schedule is
configured, but never confirmed:

|[  128.250219] imx-dwmac 428a0000.ethernet eth1: configured EST

The reason seems to be the refactoring of the EST code which set the wrong
EST offset for the dwmac 5.10. After fixing this it works as before:

|[  106.359577] imx-dwmac 428a0000.ethernet eth1: configured EST
|[  128.430715] imx-dwmac 428a0000.ethernet eth1: EST: SWOL has been switched

Tested on imx93.

Fixes: c3f3b97238f6 ("net: stmmac: Refactor EST implementation")
Signed-off-by: Kurt Kanzenbach <kurt@linutronix.de>
Reviewed-by: Serge Semin <fancer.lancer@gmail.com>
Link: https://lore.kernel.org/r/20240220-stmmac_est-v1-1-c41f9ae2e7b7@linutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
6 months agoMerge branch 'tools-ynl-fix-impossible-errors'
Jakub Kicinski [Thu, 22 Feb 2024 01:02:30 +0000 (17:02 -0800)]
Merge branch 'tools-ynl-fix-impossible-errors'

Jakub Kicinski says:

====================
tools: ynl: fix impossible errors

Fix bugs discovered while I was hacking in low level stuff in YNL
and kept breaking the socket, exercising the "impossible" error paths.

v1: https://lore.kernel.org/all/20240217001742.2466993-1-kuba@kernel.org/
====================

Link: https://lore.kernel.org/r/20240220161112.2735195-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
6 months agotools: ynl: don't leak mcast_groups on init error
Jakub Kicinski [Tue, 20 Feb 2024 16:11:12 +0000 (08:11 -0800)]
tools: ynl: don't leak mcast_groups on init error

Make sure to free the already-parsed mcast_groups if
we don't get an ack from the kernel when reading family info.
This is part of the ynl_sock_create() error path, so we won't
get a call to ynl_sock_destroy() to free them later.

Fixes: 86878f14d71a ("tools: ynl: user space helpers")
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Link: https://lore.kernel.org/r/20240220161112.2735195-3-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
6 months agotools: ynl: make sure we always pass yarg to mnl_cb_run
Jakub Kicinski [Tue, 20 Feb 2024 16:11:11 +0000 (08:11 -0800)]
tools: ynl: make sure we always pass yarg to mnl_cb_run

There is one common error handler in ynl - ynl_cb_error().
It expects priv to be a pointer to struct ynl_parse_arg AKA yarg.
To avoid potential crashes if we encounter a stray NLMSG_ERROR
always pass yarg as priv (or a struct which has it as the first
member).

ynl_cb_null() has a similar problem directly - it expects yarg
but priv passed by the caller is ys.

Found by code inspection.

Fixes: 86878f14d71a ("tools: ynl: user space helpers")
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Link: https://lore.kernel.org/r/20240220161112.2735195-2-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
6 months agonet: mctp: put sock on tag allocation failure
Jeremy Kerr [Thu, 15 Feb 2024 07:53:08 +0000 (15:53 +0800)]
net: mctp: put sock on tag allocation failure

We may hold an extra reference on a socket if a tag allocation fails: we
optimistically allocate the sk_key, and take a ref there, but do not
drop if we end up not using the allocated key.

Ensure we're dropping the sock on this failure by doing a proper unref
rather than directly kfree()ing.

Fixes: de8a6b15d965 ("net: mctp: add an explicit reference from a mctp_sk_key to sock")
Signed-off-by: Jeremy Kerr <jk@codeconstruct.com.au>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://lore.kernel.org/r/ce9b61e44d1cdae7797be0c5e3141baf582d23a0.1707983487.git.jk@codeconstruct.com.au
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
6 months agonetfilter: nf_tables: use kzalloc for hook allocation
Florian Westphal [Wed, 21 Feb 2024 17:38:45 +0000 (18:38 +0100)]
netfilter: nf_tables: use kzalloc for hook allocation

KMSAN reports unitialized variable when registering the hook,
   reg->hook_ops_type == NF_HOOK_OP_BPF)
        ~~~~~~~~~~~ undefined

This is a small structure, just use kzalloc to make sure this
won't happen again when new fields get added to nf_hook_ops.

Fixes: 7b4b2fa37587 ("netfilter: annotate nf_tables base hook ops")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
6 months agonetfilter: nf_tables: register hooks last when adding new chain/flowtable
Pablo Neira Ayuso [Mon, 19 Feb 2024 18:43:53 +0000 (19:43 +0100)]
netfilter: nf_tables: register hooks last when adding new chain/flowtable

Register hooks last when adding chain/flowtable to ensure that packets do
not walk over datastructure that is being released in the error path
without waiting for the rcu grace period.

Fixes: 91c7b38dc9f0 ("netfilter: nf_tables: use new transaction infrastructure to handle chain")
Fixes: 3b49e2e94e6e ("netfilter: nf_tables: add flow table netlink frontend")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
6 months agonetfilter: nft_flow_offload: release dst in case direct xmit path is used
Pablo Neira Ayuso [Tue, 20 Feb 2024 20:36:39 +0000 (21:36 +0100)]
netfilter: nft_flow_offload: release dst in case direct xmit path is used

Direct xmit does not use it since it calls dev_queue_xmit() to send
packets, hence it calls dst_release().

kmemleak reports:

unreferenced object 0xffff88814f440900 (size 184):
  comm "softirq", pid 0, jiffies 4294951896
  hex dump (first 32 bytes):
    00 60 5b 04 81 88 ff ff 00 e6 e8 82 ff ff ff ff  .`[.............
    21 0b 50 82 ff ff ff ff 00 00 00 00 00 00 00 00  !.P.............
  backtrace (crc cb2bf5d6):
    [<000000003ee17107>] kmem_cache_alloc+0x286/0x340
    [<0000000021a5de2c>] dst_alloc+0x43/0xb0
    [<00000000f0671159>] rt_dst_alloc+0x2e/0x190
    [<00000000fe5092c9>] __mkroute_output+0x244/0x980
    [<000000005fb96fb0>] ip_route_output_flow+0xc0/0x160
    [<0000000045367433>] nf_ip_route+0xf/0x30
    [<0000000085da1d8e>] nf_route+0x2d/0x60
    [<00000000d1ecd1cb>] nft_flow_route+0x171/0x6a0 [nft_flow_offload]
    [<00000000d9b2fb60>] nft_flow_offload_eval+0x4e8/0x700 [nft_flow_offload]
    [<000000009f447dbb>] expr_call_ops_eval+0x53/0x330 [nf_tables]
    [<00000000072e1be6>] nft_do_chain+0x17c/0x840 [nf_tables]
    [<00000000d0551029>] nft_do_chain_inet+0xa1/0x210 [nf_tables]
    [<0000000097c9d5c6>] nf_hook_slow+0x5b/0x160
    [<0000000005eccab1>] ip_forward+0x8b6/0x9b0
    [<00000000553a269b>] ip_rcv+0x221/0x230
    [<00000000412872e5>] __netif_receive_skb_one_core+0xfe/0x110

Fixes: fa502c865666 ("netfilter: flowtable: simplify route logic")
Reported-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
6 months agonetfilter: nft_flow_offload: reset dst in route object after setting up flow
Pablo Neira Ayuso [Wed, 21 Feb 2024 11:32:58 +0000 (12:32 +0100)]
netfilter: nft_flow_offload: reset dst in route object after setting up flow

dst is transferred to the flow object, route object does not own it
anymore.  Reset dst in route object, otherwise if flow_offload_add()
fails, error path releases dst twice, leading to a refcount underflow.

Fixes: a3c90f7a2323 ("netfilter: nf_tables: flow offload expression")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
6 months agonetfilter: nf_tables: set dormant flag on hook register failure
Florian Westphal [Mon, 19 Feb 2024 15:58:04 +0000 (16:58 +0100)]
netfilter: nf_tables: set dormant flag on hook register failure

We need to set the dormant flag again if we fail to register
the hooks.

During memory pressure hook registration can fail and we end up
with a table marked as active but no registered hooks.

On table/base chain deletion, nf_tables will attempt to unregister
the hook again which yields a warn splat from the nftables core.

Reported-and-tested-by: syzbot+de4025c006ec68ac56fc@syzkaller.appspotmail.com
Fixes: 179d9ba5559a ("netfilter: nf_tables: fix table flag updates")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
6 months agoMerge branch 'tls-fixes-for-record-type-handling-with-peek'
Jakub Kicinski [Wed, 21 Feb 2024 22:25:53 +0000 (14:25 -0800)]
Merge branch 'tls-fixes-for-record-type-handling-with-peek'

Sabrina Dubroca says:

====================
tls: fixes for record type handling with PEEK

There are multiple bugs in tls_sw_recvmsg's handling of record types
when MSG_PEEK flag is used, which can lead to incorrectly merging two
records:
 - consecutive non-DATA records shouldn't be merged, even if they're
   the same type (partly handled by the test at the end of the main
   loop)
 - records of the same type (even DATA) shouldn't be merged if one
   record of a different type comes in between
====================

Link: https://lore.kernel.org/r/cover.1708007371.git.sd@queasysnail.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
6 months agoselftests: tls: add test for peeking past a record of a different type
Sabrina Dubroca [Thu, 15 Feb 2024 16:17:33 +0000 (17:17 +0100)]
selftests: tls: add test for peeking past a record of a different type

If we queue 3 records:
 - record 1, type DATA
 - record 2, some other type
 - record 3, type DATA
the current code can look past the 2nd record and merge the 2 data
records.

Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Link: https://lore.kernel.org/r/4623550f8617c239581030c13402d3262f2bd14f.1708007371.git.sd@queasysnail.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
6 months agoselftests: tls: add test for merging of same-type control messages
Sabrina Dubroca [Thu, 15 Feb 2024 16:17:32 +0000 (17:17 +0100)]
selftests: tls: add test for merging of same-type control messages

Two consecutive control messages of the same type should never be
merged into one large received blob of data.

Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Link: https://lore.kernel.org/r/018f1633d5471684c65def5fe390de3b15c3d683.1708007371.git.sd@queasysnail.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
6 months agotls: don't skip over different type records from the rx_list
Sabrina Dubroca [Thu, 15 Feb 2024 16:17:31 +0000 (17:17 +0100)]
tls: don't skip over different type records from the rx_list

If we queue 3 records:
 - record 1, type DATA
 - record 2, some other type
 - record 3, type DATA
and do a recv(PEEK), the rx_list will contain the first two records.

The next large recv will walk through the rx_list and copy data from
record 1, then stop because record 2 is a different type. Since we
haven't filled up our buffer, we will process the next available
record. It's also DATA, so we can merge it with the current read.

We shouldn't do that, since there was a record in between that we
ignored.

Add a flag to let process_rx_list inform tls_sw_recvmsg that it had
more data available.

Fixes: 692d7b5d1f91 ("tls: Fix recvmsg() to be able to peek across multiple records")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Link: https://lore.kernel.org/r/f00c0c0afa080c60f016df1471158c1caf983c34.1708007371.git.sd@queasysnail.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
6 months agotls: stop recv() if initial process_rx_list gave us non-DATA
Sabrina Dubroca [Thu, 15 Feb 2024 16:17:30 +0000 (17:17 +0100)]
tls: stop recv() if initial process_rx_list gave us non-DATA

If we have a non-DATA record on the rx_list and another record of the
same type still on the queue, we will end up merging them:
 - process_rx_list copies the non-DATA record
 - we start the loop and process the first available record since it's
   of the same type
 - we break out of the loop since the record was not DATA

Just check the record type and jump to the end in case process_rx_list
did some work.

Fixes: 692d7b5d1f91 ("tls: Fix recvmsg() to be able to peek across multiple records")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Link: https://lore.kernel.org/r/bd31449e43bd4b6ff546f5c51cf958c31c511deb.1708007371.git.sd@queasysnail.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
6 months agotls: break out of main loop when PEEK gets a non-data record
Sabrina Dubroca [Thu, 15 Feb 2024 16:17:29 +0000 (17:17 +0100)]
tls: break out of main loop when PEEK gets a non-data record

PEEK needs to leave decrypted records on the rx_list so that we can
receive them later on, so it jumps back into the async code that
queues the skb. Unfortunately that makes us skip the
TLS_RECORD_TYPE_DATA check at the bottom of the main loop, so if two
records of the same (non-DATA) type are queued, we end up merging
them.

Add the same record type check, and make it unlikely to not penalize
the async fastpath. Async decrypt only applies to data record, so this
check is only needed for PEEK.

process_rx_list also has similar issues.

Fixes: 692d7b5d1f91 ("tls: Fix recvmsg() to be able to peek across multiple records")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Link: https://lore.kernel.org/r/3df2eef4fdae720c55e69472b5bea668772b45a2.1708007371.git.sd@queasysnail.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
6 months agogtp: fix use-after-free and null-ptr-deref in gtp_genl_dump_pdp()
Vasiliy Kovalev [Wed, 14 Feb 2024 16:27:33 +0000 (19:27 +0300)]
gtp: fix use-after-free and null-ptr-deref in gtp_genl_dump_pdp()

The gtp_net_ops pernet operations structure for the subsystem must be
registered before registering the generic netlink family.

Syzkaller hit 'general protection fault in gtp_genl_dump_pdp' bug:

general protection fault, probably for non-canonical address
0xdffffc0000000002: 0000 [#1] PREEMPT SMP KASAN NOPTI
KASAN: null-ptr-deref in range [0x0000000000000010-0x0000000000000017]
CPU: 1 PID: 5826 Comm: gtp Not tainted 6.8.0-rc3-std-def-alt1 #1
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.0-alt1 04/01/2014
RIP: 0010:gtp_genl_dump_pdp+0x1be/0x800 [gtp]
Code: c6 89 c6 e8 64 e9 86 df 58 45 85 f6 0f 85 4e 04 00 00 e8 c5 ee 86
      df 48 8b 54 24 18 48 b8 00 00 00 00 00 fc ff df 48 c1 ea 03 <80>
      3c 02 00 0f 85 de 05 00 00 48 8b 44 24 18 4c 8b 30 4c 39 f0 74
RSP: 0018:ffff888014107220 EFLAGS: 00010202
RAX: dffffc0000000000 RBX: 0000000000000000 RCX: 0000000000000000
RDX: 0000000000000002 RSI: 0000000000000000 RDI: 0000000000000000
RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
R13: ffff88800fcda588 R14: 0000000000000001 R15: 0000000000000000
FS:  00007f1be4eb05c0(0000) GS:ffff88806ce80000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f1be4e766cf CR3: 000000000c33e000 CR4: 0000000000750ef0
PKRU: 55555554
Call Trace:
 <TASK>
 ? show_regs+0x90/0xa0
 ? die_addr+0x50/0xd0
 ? exc_general_protection+0x148/0x220
 ? asm_exc_general_protection+0x22/0x30
 ? gtp_genl_dump_pdp+0x1be/0x800 [gtp]
 ? __alloc_skb+0x1dd/0x350
 ? __pfx___alloc_skb+0x10/0x10
 genl_dumpit+0x11d/0x230
 netlink_dump+0x5b9/0xce0
 ? lockdep_hardirqs_on_prepare+0x253/0x430
 ? __pfx_netlink_dump+0x10/0x10
 ? kasan_save_track+0x10/0x40
 ? __kasan_kmalloc+0x9b/0xa0
 ? genl_start+0x675/0x970
 __netlink_dump_start+0x6fc/0x9f0
 genl_family_rcv_msg_dumpit+0x1bb/0x2d0
 ? __pfx_genl_family_rcv_msg_dumpit+0x10/0x10
 ? genl_op_from_small+0x2a/0x440
 ? cap_capable+0x1d0/0x240
 ? __pfx_genl_start+0x10/0x10
 ? __pfx_genl_dumpit+0x10/0x10
 ? __pfx_genl_done+0x10/0x10
 ? security_capable+0x9d/0xe0

Cc: stable@vger.kernel.org
Signed-off-by: Vasiliy Kovalev <kovalev@altlinux.org>
Fixes: 459aa660eb1d ("gtp: add initial driver for datapath of GPRS Tunneling Protocol (GTP-U)")
Link: https://lore.kernel.org/r/20240214162733.34214-1-kovalev@altlinux.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
6 months agohwmon: (nct6775) Fix access to temperature configuration registers
Guenter Roeck [Wed, 21 Feb 2024 14:01:20 +0000 (06:01 -0800)]
hwmon: (nct6775) Fix access to temperature configuration registers

The number of temperature configuration registers does
not always match the total number of temperature registers.
This can result in access errors reported if KASAN is enabled.

BUG: KASAN: global-out-of-bounds in nct6775_probe+0x5654/0x6fe9 nct6775_core

Reported-by: Erhard Furtner <erhard_f@mailbox.org>
Closes: https://lore.kernel.org/linux-hwmon/d51181d1-d26b-42b2-b002-3f5a4037721f@roeck-us.net/
Fixes: b7f1f7b2523a ("hwmon: (nct6775) Additional TEMP registers for nct6799")
Cc: Ahmad Khalifa <ahmad@khalifa.ws>
Tested-by: Ahmad Khalifa <ahmad@khalifa.ws>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
6 months agoata: libata-core: Do not call ata_dev_power_set_standby() twice
Damien Le Moal [Mon, 19 Feb 2024 15:44:30 +0000 (16:44 +0100)]
ata: libata-core: Do not call ata_dev_power_set_standby() twice

For regular system shutdown, ata_dev_power_set_standby() will be
executed twice: once the scsi device is removed and another when
ata_pci_shutdown_one() executes and EH completes unloading the devices.

Make the second call to ata_dev_power_set_standby() do nothing by using
ata_dev_power_is_active() and return if the device is already in
standby.

Fixes: 2da4c5e24e86 ("ata: libata-core: Improve ata_dev_power_set_active()")
Cc: stable@vger.kernel.org
Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Niklas Cassel <cassel@kernel.org>
6 months agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Wed, 21 Feb 2024 17:13:27 +0000 (09:13 -0800)]
Merge tag 'for-linus' of git://git./virt/kvm/kvm

Pull kvm fixes from Paolo Bonzini:
 "Two fixes for ARM ITS emulation. Unmapped interrupts were used instead
  of ignored, causing NULL pointer dereferences"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: arm64: vgic-its: Test for valid IRQ in MOVALL handler
  KVM: arm64: vgic-its: Test for valid IRQ in its_sync_lpi_pending_table()

6 months agoMerge tag 'for-6.8-rc5-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave...
Linus Torvalds [Wed, 21 Feb 2024 16:45:07 +0000 (08:45 -0800)]
Merge tag 'for-6.8-rc5-tag' of git://git./linux/kernel/git/kdave/linux

Pull btrfs fixes from David Sterba:

 - Fix a deadlock in fiemap.

   There was a big lock around the whole operation that can interfere
   with a page fault and mkwrite.

   Reducing the lock scope can also speed up fiemap

 - Fix range condition for extent defragmentation which could lead to
   worse layout in some cases

* tag 'for-6.8-rc5-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
  btrfs: fix deadlock with fiemap and extent locking
  btrfs: defrag: avoid unnecessary defrag caused by incorrect extent size

6 months agoMerge tag 'v6.8-p4' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6
Linus Torvalds [Wed, 21 Feb 2024 16:37:49 +0000 (08:37 -0800)]
Merge tag 'v6.8-p4' of git://git./linux/kernel/git/herbert/crypto-2.6

Pull crypto fix from Herbert Xu:
 "Fix a stack overflow in virtio"

* tag 'v6.8-p4' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  crypto: virtio/akcipher - Fix stack overflow on memcpy

6 months agobpf, sockmap: Fix NULL pointer dereference in sk_psock_verdict_data_ready()
Shigeru Yoshida [Sun, 18 Feb 2024 15:09:33 +0000 (00:09 +0900)]
bpf, sockmap: Fix NULL pointer dereference in sk_psock_verdict_data_ready()

syzbot reported the following NULL pointer dereference issue [1]:

  BUG: kernel NULL pointer dereference, address: 0000000000000000
  [...]
  RIP: 0010:0x0
  [...]
  Call Trace:
   <TASK>
   sk_psock_verdict_data_ready+0x232/0x340 net/core/skmsg.c:1230
   unix_stream_sendmsg+0x9b4/0x1230 net/unix/af_unix.c:2293
   sock_sendmsg_nosec net/socket.c:730 [inline]
   __sock_sendmsg+0x221/0x270 net/socket.c:745
   ____sys_sendmsg+0x525/0x7d0 net/socket.c:2584
   ___sys_sendmsg net/socket.c:2638 [inline]
   __sys_sendmsg+0x2b0/0x3a0 net/socket.c:2667
   do_syscall_64+0xf9/0x240
   entry_SYSCALL_64_after_hwframe+0x6f/0x77

If sk_psock_verdict_data_ready() and sk_psock_stop_verdict() are called
concurrently, psock->saved_data_ready can be NULL, causing the above issue.

This patch fixes this issue by calling the appropriate data ready function
using the sk_psock_data_ready() helper and protecting it from concurrency
with sk->sk_callback_lock.

Fixes: 6df7f764cd3c ("bpf, sockmap: Wake up polling after data copy")
Reported-by: syzbot+fd7b34375c1c8ce29c93@syzkaller.appspotmail.com
Signed-off-by: Shigeru Yoshida <syoshida@redhat.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Tested-by: syzbot+fd7b34375c1c8ce29c93@syzkaller.appspotmail.com
Acked-by: John Fastabend <john.fastabend@gmail.com>
Closes: https://syzkaller.appspot.com/bug?extid=fd7b34375c1c8ce29c93 [1]
Link: https://lore.kernel.org/bpf/20240218150933.6004-1-syoshida@redhat.com
6 months agofs/aio: Restrict kiocb_set_cancel_fn() to I/O submitted via libaio
Bart Van Assche [Thu, 15 Feb 2024 20:47:38 +0000 (12:47 -0800)]
fs/aio: Restrict kiocb_set_cancel_fn() to I/O submitted via libaio

If kiocb_set_cancel_fn() is called for I/O submitted via io_uring, the
following kernel warning appears:

WARNING: CPU: 3 PID: 368 at fs/aio.c:598 kiocb_set_cancel_fn+0x9c/0xa8
Call trace:
 kiocb_set_cancel_fn+0x9c/0xa8
 ffs_epfile_read_iter+0x144/0x1d0
 io_read+0x19c/0x498
 io_issue_sqe+0x118/0x27c
 io_submit_sqes+0x25c/0x5fc
 __arm64_sys_io_uring_enter+0x104/0xab0
 invoke_syscall+0x58/0x11c
 el0_svc_common+0xb4/0xf4
 do_el0_svc+0x2c/0xb0
 el0_svc+0x2c/0xa4
 el0t_64_sync_handler+0x68/0xb4
 el0t_64_sync+0x1a4/0x1a8

Fix this by setting the IOCB_AIO_RW flag for read and write I/O that is
submitted by libaio.

Suggested-by: Jens Axboe <axboe@kernel.dk>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Avi Kivity <avi@scylladb.com>
Cc: Sandeep Dhavale <dhavale@google.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Kent Overstreet <kent.overstreet@linux.dev>
Cc: stable@vger.kernel.org
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240215204739.2677806-2-bvanassche@acm.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
6 months agoring-buffer: Do not let subbuf be bigger than write mask
Steven Rostedt (Google) [Tue, 20 Feb 2024 14:51:12 +0000 (09:51 -0500)]
ring-buffer: Do not let subbuf be bigger than write mask

The data on the subbuffer is measured by a write variable that also
contains status flags. The counter is just 20 bits in length. If the
subbuffer is bigger than then counter, it will fail.

Make sure that the subbuffer can not be set to greater than the counter
that keeps track of the data on the subbuffer.

Link: https://lore.kernel.org/linux-trace-kernel/20240220095112.77e9cb81@gandalf.local.home
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Fixes: 2808e31ec12e5 ("ring-buffer: Add interface for configuring trace sub buffer size")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
6 months agoiommufd: Reject non-zero data_type if no data_len is provided
Jason Gunthorpe [Tue, 20 Feb 2024 18:43:54 +0000 (14:43 -0400)]
iommufd: Reject non-zero data_type if no data_len is provided

Since the current design doesn't forward the data_type to the driver to
check unless there is a data_len/uptr for a driver specific struct we
should check and ensure that data_type is 0 if data_len is 0. Otherwise
any value is permitted.

Fixes: bd529dbb661d ("iommufd: Add a nested HW pagetable object")
Link: https://lore.kernel.org/r/0-v1-9b1ea6869554+110c60-iommufd_ck_data_type_jgg@nvidia.com
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
6 months agosparc: Fix undefined reference to fb_is_primary_device
Javier Martinez Canillas [Tue, 20 Feb 2024 09:54:12 +0000 (10:54 +0100)]
sparc: Fix undefined reference to fb_is_primary_device

Commit 55bffc8170bb ("fbdev: Split frame buffer support in FB and FB_CORE
symbols") added a new FB_CORE Kconfig symbol, that can be enabled to only
have fbcon/VT and DRM fbdev emulation, but without support for any legacy
fbdev driver.

Unfortunately, it missed to change the CONFIG_FB in arch/sparc makefiles,
which leads to the following linking error in some sparc64 configurations:

   sparc64-linux-ld: drivers/video/fbdev/core/fbcon.o: in function `fbcon_fb_registered':
>> fbcon.c:(.text+0x4f60): undefined reference to `fb_is_primary_device'

Fixes: 55bffc8170bb ("fbdev: Split frame buffer support in FB and FB_CORE symbols")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/r/202401290306.IV8rhJ02-lkp@intel.com/
Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Cc: <stable@vger.kernel.org> # v6.6+
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20240220095428.3341195-1-javierm@redhat.com
6 months agoMAINTAINERS: Add framer headers to NETWORKING [GENERAL]
Simon Horman [Mon, 19 Feb 2024 17:55:31 +0000 (17:55 +0000)]
MAINTAINERS: Add framer headers to NETWORKING [GENERAL]

The cited commit [1] added framer support under drivers/net/wan,
which is covered by NETWORKING [GENERAL]. And it is implied
that framer-provider.h and framer.h, which were also added
buy the same patch, are also maintained as part of NETWORKING [GENERAL].

Make this explicit by adding these files to the corresponding
section in MAINTAINERS.

[1] 82c944d05b1a ("net: wan: Add framer framework support")

Signed-off-by: Simon Horman <horms@kernel.org>
Reviewed-by: Herve Codina <herve.codina@bootlin.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 months agoaf_unix: Drop oob_skb ref before purging queue in GC.
Kuniyuki Iwashima [Mon, 19 Feb 2024 17:46:57 +0000 (09:46 -0800)]
af_unix: Drop oob_skb ref before purging queue in GC.

syzbot reported another task hung in __unix_gc().  [0]

The current while loop assumes that all of the left candidates
have oob_skb and calling kfree_skb(oob_skb) releases the remaining
candidates.

However, I missed a case that oob_skb has self-referencing fd and
another fd and the latter sk is placed before the former in the
candidate list.  Then, the while loop never proceeds, resulting
the task hung.

__unix_gc() has the same loop just before purging the collected skb,
so we can call kfree_skb(oob_skb) there and let __skb_queue_purge()
release all inflight sockets.

[0]:
Sending NMI from CPU 0 to CPUs 1:
NMI backtrace for cpu 1
CPU: 1 PID: 2784 Comm: kworker/u4:8 Not tainted 6.8.0-rc4-syzkaller-01028-g71b605d32017 #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/25/2024
Workqueue: events_unbound __unix_gc
RIP: 0010:__sanitizer_cov_trace_pc+0x0/0x70 kernel/kcov.c:200
Code: 89 fb e8 23 00 00 00 48 8b 3d 84 f5 1a 0c 48 89 de 5b e9 43 26 57 00 0f 1f 00 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 <f3> 0f 1e fa 48 8b 04 24 65 48 8b 0d 90 52 70 7e 65 8b 15 91 52 70
RSP: 0018:ffffc9000a17fa78 EFLAGS: 00000287
RAX: ffffffff8a0a6108 RBX: ffff88802b6c2640 RCX: ffff88802c0b3b80
RDX: 0000000000000000 RSI: 0000000000000002 RDI: 0000000000000000
RBP: ffffc9000a17fbf0 R08: ffffffff89383f1d R09: 1ffff1100ee5ff84
R10: dffffc0000000000 R11: ffffed100ee5ff85 R12: 1ffff110056d84ee
R13: ffffc9000a17fae0 R14: 0000000000000000 R15: ffffffff8f47b840
FS:  0000000000000000(0000) GS:ffff8880b9500000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007ffef5687ff8 CR3: 0000000029b34000 CR4: 00000000003506f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
 <NMI>
 </NMI>
 <TASK>
 __unix_gc+0xe69/0xf40 net/unix/garbage.c:343
 process_one_work kernel/workqueue.c:2633 [inline]
 process_scheduled_works+0x913/0x1420 kernel/workqueue.c:2706
 worker_thread+0xa5f/0x1000 kernel/workqueue.c:2787
 kthread+0x2ef/0x390 kernel/kthread.c:388
 ret_from_fork+0x4b/0x80 arch/x86/kernel/process.c:147
 ret_from_fork_asm+0x1b/0x30 arch/x86/entry/entry_64.S:242
 </TASK>

Reported-and-tested-by: syzbot+ecab4d36f920c3574bf9@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=ecab4d36f920c3574bf9
Fixes: 25236c91b5ab ("af_unix: Fix task hung while purging oob_skb in GC.")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 months agonet: ipa: don't overrun IPA suspend interrupt registers
Alex Elder [Mon, 19 Feb 2024 14:40:15 +0000 (08:40 -0600)]
net: ipa: don't overrun IPA suspend interrupt registers

In newer hardware, IPA supports more than 32 endpoints.  Some
registers--such as IPA interrupt registers--represent endpoints
as bits in a 4-byte register, and such registers are repeated as
needed to represent endpoints beyond the first 32.

In ipa_interrupt_suspend_clear_all(), we clear all pending IPA
suspend interrupts by reading all status register(s) and writing
corresponding registers to clear interrupt conditions.

Unfortunately the number of registers to read/write is calculated
incorrectly, and as a result we access *many* more registers than
intended.  This bug occurs only when the IPA hardware signals a
SUSPEND interrupt, which happens when a packet is received for an
endpoint (or its underlying GSI channel) that is suspended.  This
situation is difficult to reproduce, but possible.

Fix this by correctly computing the number of interrupt registers to
read and write.  This is the only place in the code where registers
that map endpoints or channels this way perform this calculation.

Fixes: f298ba785e2d ("net: ipa: add a parameter to suspend registers")
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 months agonet: implement lockless setsockopt(SO_PEEK_OFF)
Eric Dumazet [Mon, 19 Feb 2024 14:12:20 +0000 (14:12 +0000)]
net: implement lockless setsockopt(SO_PEEK_OFF)

syzbot reported a lockdep violation [1] involving af_unix
support of SO_PEEK_OFF.

Since SO_PEEK_OFF is inherently not thread safe (it uses a per-socket
sk_peek_off field), there is really no point to enforce a pointless
thread safety in the kernel.

After this patch :

- setsockopt(SO_PEEK_OFF) no longer acquires the socket lock.

- skb_consume_udp() no longer has to acquire the socket lock.

- af_unix no longer needs a special version of sk_set_peek_off(),
  because it does not lock u->iolock anymore.

As a followup, we could replace prot->set_peek_off to be a boolean
and avoid an indirect call, since we always use sk_set_peek_off().

[1]

WARNING: possible circular locking dependency detected
6.8.0-rc4-syzkaller-00267-g0f1dd5e91e2b #0 Not tainted

syz-executor.2/30025 is trying to acquire lock:
 ffff8880765e7d80 (&u->iolock){+.+.}-{3:3}, at: unix_set_peek_off+0x26/0xa0 net/unix/af_unix.c:789

but task is already holding lock:
 ffff8880765e7930 (sk_lock-AF_UNIX){+.+.}-{0:0}, at: lock_sock include/net/sock.h:1691 [inline]
 ffff8880765e7930 (sk_lock-AF_UNIX){+.+.}-{0:0}, at: sockopt_lock_sock net/core/sock.c:1060 [inline]
 ffff8880765e7930 (sk_lock-AF_UNIX){+.+.}-{0:0}, at: sk_setsockopt+0xe52/0x3360 net/core/sock.c:1193

which lock already depends on the new lock.

the existing dependency chain (in reverse order) is:

-> #1 (sk_lock-AF_UNIX){+.+.}-{0:0}:
        lock_acquire+0x1e3/0x530 kernel/locking/lockdep.c:5754
        lock_sock_nested+0x48/0x100 net/core/sock.c:3524
        lock_sock include/net/sock.h:1691 [inline]
        __unix_dgram_recvmsg+0x1275/0x12c0 net/unix/af_unix.c:2415
        sock_recvmsg_nosec+0x18e/0x1d0 net/socket.c:1046
        ____sys_recvmsg+0x3c0/0x470 net/socket.c:2801
        ___sys_recvmsg net/socket.c:2845 [inline]
        do_recvmmsg+0x474/0xae0 net/socket.c:2939
        __sys_recvmmsg net/socket.c:3018 [inline]
        __do_sys_recvmmsg net/socket.c:3041 [inline]
        __se_sys_recvmmsg net/socket.c:3034 [inline]
        __x64_sys_recvmmsg+0x199/0x250 net/socket.c:3034
       do_syscall_64+0xf9/0x240
       entry_SYSCALL_64_after_hwframe+0x6f/0x77

-> #0 (&u->iolock){+.+.}-{3:3}:
        check_prev_add kernel/locking/lockdep.c:3134 [inline]
        check_prevs_add kernel/locking/lockdep.c:3253 [inline]
        validate_chain+0x18ca/0x58e0 kernel/locking/lockdep.c:3869
        __lock_acquire+0x1345/0x1fd0 kernel/locking/lockdep.c:5137
        lock_acquire+0x1e3/0x530 kernel/locking/lockdep.c:5754
        __mutex_lock_common kernel/locking/mutex.c:608 [inline]
        __mutex_lock+0x136/0xd70 kernel/locking/mutex.c:752
        unix_set_peek_off+0x26/0xa0 net/unix/af_unix.c:789
       sk_setsockopt+0x207e/0x3360
        do_sock_setsockopt+0x2fb/0x720 net/socket.c:2307
        __sys_setsockopt+0x1ad/0x250 net/socket.c:2334
        __do_sys_setsockopt net/socket.c:2343 [inline]
        __se_sys_setsockopt net/socket.c:2340 [inline]
        __x64_sys_setsockopt+0xb5/0xd0 net/socket.c:2340
       do_syscall_64+0xf9/0x240
       entry_SYSCALL_64_after_hwframe+0x6f/0x77

other info that might help us debug this:

 Possible unsafe locking scenario:

       CPU0                    CPU1
       ----                    ----
  lock(sk_lock-AF_UNIX);
                               lock(&u->iolock);
                               lock(sk_lock-AF_UNIX);
  lock(&u->iolock);

 *** DEADLOCK ***

1 lock held by syz-executor.2/30025:
  #0: ffff8880765e7930 (sk_lock-AF_UNIX){+.+.}-{0:0}, at: lock_sock include/net/sock.h:1691 [inline]
  #0: ffff8880765e7930 (sk_lock-AF_UNIX){+.+.}-{0:0}, at: sockopt_lock_sock net/core/sock.c:1060 [inline]
  #0: ffff8880765e7930 (sk_lock-AF_UNIX){+.+.}-{0:0}, at: sk_setsockopt+0xe52/0x3360 net/core/sock.c:1193

stack backtrace:
CPU: 0 PID: 30025 Comm: syz-executor.2 Not tainted 6.8.0-rc4-syzkaller-00267-g0f1dd5e91e2b #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/25/2024
Call Trace:
 <TASK>
  __dump_stack lib/dump_stack.c:88 [inline]
  dump_stack_lvl+0x1e7/0x2e0 lib/dump_stack.c:106
  check_noncircular+0x36a/0x4a0 kernel/locking/lockdep.c:2187
  check_prev_add kernel/locking/lockdep.c:3134 [inline]
  check_prevs_add kernel/locking/lockdep.c:3253 [inline]
  validate_chain+0x18ca/0x58e0 kernel/locking/lockdep.c:3869
  __lock_acquire+0x1345/0x1fd0 kernel/locking/lockdep.c:5137
  lock_acquire+0x1e3/0x530 kernel/locking/lockdep.c:5754
  __mutex_lock_common kernel/locking/mutex.c:608 [inline]
  __mutex_lock+0x136/0xd70 kernel/locking/mutex.c:752
  unix_set_peek_off+0x26/0xa0 net/unix/af_unix.c:789
 sk_setsockopt+0x207e/0x3360
  do_sock_setsockopt+0x2fb/0x720 net/socket.c:2307
  __sys_setsockopt+0x1ad/0x250 net/socket.c:2334
  __do_sys_setsockopt net/socket.c:2343 [inline]
  __se_sys_setsockopt net/socket.c:2340 [inline]
  __x64_sys_setsockopt+0xb5/0xd0 net/socket.c:2340
 do_syscall_64+0xf9/0x240
 entry_SYSCALL_64_after_hwframe+0x6f/0x77
RIP: 0033:0x7f78a1c7dda9
Code: 28 00 00 00 75 05 48 83 c4 28 c3 e8 e1 20 00 00 90 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b0 ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007f78a0fde0c8 EFLAGS: 00000246 ORIG_RAX: 0000000000000036
RAX: ffffffffffffffda RBX: 00007f78a1dac050 RCX: 00007f78a1c7dda9
RDX: 000000000000002a RSI: 0000000000000001 RDI: 0000000000000006
RBP: 00007f78a1cca47a R08: 0000000000000004 R09: 0000000000000000
R10: 0000000020000180 R11: 0000000000000246 R12: 0000000000000000
R13: 000000000000006e R14: 00007f78a1dac050 R15: 00007ffe5cd81ae8

Fixes: 859051dd165e ("bpf: Implement cgroup sockaddr hooks for unix sockets")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Willem de Bruijn <willemdebruijn.kernel@gmail.com>
Cc: Daan De Meyer <daan.j.demeyer@gmail.com>
Cc: Kuniyuki Iwashima <kuniyu@amazon.com>
Cc: Martin KaFai Lau <martin.lau@kernel.org>
Cc: David Ahern <dsahern@kernel.org>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 months agoocteontx2-af: Consider the action set by PF
Subbaraya Sundeep [Mon, 19 Feb 2024 12:55:14 +0000 (18:25 +0530)]
octeontx2-af: Consider the action set by PF

AF reserves MCAM entries for each PF, VF present in the
system and populates the entry with DMAC and action with
default RSS so that basic packet I/O works. Since PF/VF is
not aware of the RSS action installed by AF, AF only fixup
the actions of the rules installed by PF/VF with corresponding
default RSS action. This worked well for rules installed by
PF/VF for features like RX VLAN offload and DMAC filters but
rules involving action like drop/forward to queue are also
getting modified by AF. Hence fix it by setting the default
RSS action only if requested by PF/VF.

Fixes: 967db3529eca ("octeontx2-af: add support for multicast/promisc packet replication feature")
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 months agoMerge tag 'kvmarm-fixes-6.8-3' of git://git.kernel.org/pub/scm/linux/kernel/git/kvmar...
Paolo Bonzini [Wed, 21 Feb 2024 10:18:56 +0000 (05:18 -0500)]
Merge tag 'kvmarm-fixes-6.8-3' of git://git./linux/kernel/git/kvmarm/kvmarm into HEAD

KVM/arm64 fixes for 6.8, take #3

- Check for the validity of interrupts handled by a MOVALL
  command

- Check for the validity of interrupts while reading the
  pending state on enabling LPIs.

6 months agodrm/xe: Fix modpost warning on xe_mocs kunit module
Ashutosh Dixit [Tue, 13 Feb 2024 03:35:48 +0000 (19:35 -0800)]
drm/xe: Fix modpost warning on xe_mocs kunit module

$ make W=1 -j100 M=drivers/gpu/drm/xe
MODPOST drivers/gpu/drm/xe/Module.symvers
WARNING: modpost: missing MODULE_DESCRIPTION() in drivers/gpu/drm/xe/tests/xe_mocs_test.o

Fix is identical to '1d425066f15f ("drm/xe: Fix modpost warning on kunit
modules")'.

Fixes: a6a4ea6d7d37 ("drm/xe: Add mocs kunit")
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
(cherry picked from commit bb619d71224ea85ec94e0a83b2bb82ebe7df2a41)
Link:  https://patchwork.freedesktop.org/patch/msgid/20240213033548.76219-1-ashutosh.dixit@intel.com
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
6 months agoKVM: arm64: vgic-its: Test for valid IRQ in MOVALL handler
Oliver Upton [Wed, 21 Feb 2024 09:27:32 +0000 (09:27 +0000)]
KVM: arm64: vgic-its: Test for valid IRQ in MOVALL handler

It is possible that an LPI mapped in a different ITS gets unmapped while
handling the MOVALL command. If that is the case, there is no state that
can be migrated to the destination. Silently ignore it and continue
migrating other LPIs.

Cc: stable@vger.kernel.org
Fixes: ff9c114394aa ("KVM: arm/arm64: GICv4: Handle MOVALL applied to a vPE")
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
Link: https://lore.kernel.org/r/20240221092732.4126848-3-oliver.upton@linux.dev
Signed-off-by: Marc Zyngier <maz@kernel.org>
6 months agoKVM: arm64: vgic-its: Test for valid IRQ in its_sync_lpi_pending_table()
Oliver Upton [Wed, 21 Feb 2024 09:27:31 +0000 (09:27 +0000)]
KVM: arm64: vgic-its: Test for valid IRQ in its_sync_lpi_pending_table()

vgic_get_irq() may not return a valid descriptor if there is no ITS that
holds a valid translation for the specified INTID. If that is the case,
it is safe to silently ignore it and continue processing the LPI pending
table.

Cc: stable@vger.kernel.org
Fixes: 33d3bc9556a7 ("KVM: arm64: vgic-its: Read initial LPI pending table")
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
Link: https://lore.kernel.org/r/20240221092732.4126848-2-oliver.upton@linux.dev
Signed-off-by: Marc Zyngier <maz@kernel.org>
6 months agodrm/xe/xe_gt_idle: Drop redundant newline in name
Ashutosh Dixit [Tue, 6 Feb 2024 19:27:31 +0000 (11:27 -0800)]
drm/xe/xe_gt_idle: Drop redundant newline in name

Newline in name is redunant and produces an unnecessary empty line during
'cat name'. Newline is added during sysfs_emit. See '27a1a1e2e47d ("drm/xe:
stringify the argument to avoid potential vulnerability")'.

v2: Add Fixes tag (Riana)

Fixes: 7b076d14f21a ("drm/xe/mtl: Add support to get C6 residency/status of MTL")
Reviewed-by: Riana Tauro <riana.tauro@intel.com>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
(cherry picked from commit e5626eb80026c4b63f8682cdeca1456303c65791)
Link: https://patchwork.freedesktop.org/patch/msgid/20240206192731.3533608-1-ashutosh.dixit@intel.com
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
6 months agodrm/xe: Return 2MB page size for compact 64k PTEs
Matthew Brost [Mon, 19 Feb 2024 21:19:42 +0000 (13:19 -0800)]
drm/xe: Return 2MB page size for compact 64k PTEs

Compact 64k PTEs are only intended to be used within a single VMA which
covers the entire 2MB range of the compact 64k PTEs. Add
XE_VMA_PTE_COMPACT VMA flag to indicate compact 64k PTEs are used and
update xe_vma_max_pte_size to return at least 2MB if set.

v2: Include missing changes

Fixes: 8f33b4f054fc ("drm/xe: Avoid doing rebinds")
Fixes: c47794bdd63d ("drm/xe: Set max pte size when skipping rebinds")
Reported-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/758
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240219211942.3633795-4-matthew.brost@intel.com
(cherry picked from commit 0f688c0eb63a643ef0568b29b12cefbb23181e1a)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
6 months agodrm/xe: Add XE_VMA_PTE_64K VMA flag
Matthew Brost [Mon, 19 Feb 2024 21:19:41 +0000 (13:19 -0800)]
drm/xe: Add XE_VMA_PTE_64K VMA flag

Add XE_VMA_PTE_64K VMA flag to ensure skipping rebinds does not cross
64k page boundaries.

Fixes: 8f33b4f054fc ("drm/xe: Avoid doing rebinds")
Fixes: c47794bdd63d ("drm/xe: Set max pte size when skipping rebinds")
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240219211942.3633795-3-matthew.brost@intel.com
(cherry picked from commit 15f0e0c2c46dddd8ee56d9b3db679fd302cc4b91)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
6 months agodrm/xe: Fix xe_vma_set_pte_size
Matthew Brost [Mon, 19 Feb 2024 21:19:40 +0000 (13:19 -0800)]
drm/xe: Fix xe_vma_set_pte_size

xe_vma_set_pte_size had a return value and did not set the 4k VMA flag.
Both of these were incorrect. Fix these.

Fixes: c47794bdd63d ("drm/xe: Set max pte size when skipping rebinds")
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240219211942.3633795-2-matthew.brost@intel.com
(cherry picked from commit 19adaccef8b246182dc89a7470aa7758245efd5d)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
6 months agodrm/xe/uapi: Remove support for persistent exec_queues
Thomas Hellström [Fri, 9 Feb 2024 11:34:44 +0000 (12:34 +0100)]
drm/xe/uapi: Remove support for persistent exec_queues

Persistent exec_queues delays explicit destruction of exec_queues
until they are done executing, but destruction on process exit
is still immediate. It turns out no UMD is relying on this
functionality, so remove it. If there turns out to be a use-case
in the future, let's re-add.

Persistent exec_queues were never used for LR VMs

v2:
- Don't add an "UNUSED" define for the missing property
  (Lucas, Rodrigo)
v3:
- Remove the remaining struct xe_exec_queue::persistent state
  (Niranjana, Lucas)

Fixes: dd08ebf6c352 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: David Airlie <airlied@gmail.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Francois Dugast <francois.dugast@intel.com>
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Acked-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240209113444.8396-1-thomas.hellstrom@linux.intel.com
(cherry picked from commit f1a9abc0cf311375695bede1590364864c05976d)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
6 months agodrm/i915/tv: Fix TV mode
Maxime Ripard [Tue, 20 Feb 2024 13:12:51 +0000 (14:12 +0100)]
drm/i915/tv: Fix TV mode

Commit 1fd4a5a36f9f ("drm/connector: Rename legacy TV property") failed
to update all the users of the struct drm_tv_connector_state mode field,
which resulted in a build failure in i915.

However, a subsequent commit in the same series reintroduced a mode
field in that structure, with a different semantic but the same type,
with the assumption that all previous users were updated.

Since that didn't happen, the i915 driver now compiles, but mixes
accesses to the legacy_mode field and the newer mode field, but with the
previous semantics.

This obviously doesn't work very well, so we need to update the accesses
that weren't in the legacy renaming commit.

Fixes: 1fd4a5a36f9f ("drm/connector: Rename legacy TV property")
Reported-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240220131251.453060-1-mripard@kernel.org
(cherry picked from commit bf7626f19d6ff14b9722273e23700400cc4d78ba)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
6 months agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma
Linus Torvalds [Wed, 21 Feb 2024 01:00:26 +0000 (17:00 -0800)]
Merge tag 'for-linus' of git://git./linux/kernel/git/rdma/rdma

Pull rdma fixes from Jason Gunthorpe:
 "Mostly irdma and bnxt_re fixes:

   - Missing error unwind in hf1

   - For bnxt - fix fenching behavior to work on new chips, fail
     unsupported SRQ resize back to userspace, propogate SRQ FW failure
     back to userspace.

   - Correctly fail unsupported SRQ resize back to userspace in bnxt

   - Adjust a memcpy in mlx5 to not overflow a struct field.

   - Prevent userspace from triggering mlx5 fw syndrome logging from
     sysfs

   - Use the correct access mode for MLX5_IB_METHOD_DEVX_OBJ_MODIFY to
     avoid a userspace failure on modify

   - For irdma - Don't UAF a concurrent tasklet during destroy, prevent
     userspace from issuing invalid QP attrs, fix a possible CQ
     overflow, capture a missing HW async error event

   - sendmsg() triggerable memory access crash in hfi1

   - Fix the srpt_service_guid parameter to not crash due to missing
     function pointer

   - Don't leak objects in error unwind in qedr

   - Don't weirdly cast function pointers in srpt"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma:
  RDMA/srpt: fix function pointer cast warnings
  RDMA/qedr: Fix qedr_create_user_qp error flow
  RDMA/srpt: Support specifying the srpt_service_guid parameter
  IB/hfi1: Fix sdma.h tx->num_descs off-by-one error
  RDMA/irdma: Add AE for too many RNRS
  RDMA/irdma: Set the CQ read threshold for GEN 1
  RDMA/irdma: Validate max_send_wr and max_recv_wr
  RDMA/irdma: Fix KASAN issue with tasklet
  RDMA/mlx5: Relax DEVX access upon modify commands
  IB/mlx5: Don't expose debugfs entries for RRoCE general parameters if not supported
  RDMA/mlx5: Fix fortify source warning while accessing Eth segment
  RDMA/bnxt_re: Add a missing check in bnxt_qplib_query_srq
  RDMA/bnxt_re: Return error for SRQ resize
  RDMA/bnxt_re: Fix unconditional fence for newer adapters
  RDMA/bnxt_re: Remove a redundant check inside bnxt_re_vf_res_config
  RDMA/bnxt_re: Avoid creating fence MR for newer adapters
  IB/hfi1: Fix a memleak in init_credit_return

6 months agokasan: guard release_free_meta() shadow access with kasan_arch_is_ready()
Benjamin Gray [Tue, 13 Feb 2024 03:39:58 +0000 (14:39 +1100)]
kasan: guard release_free_meta() shadow access with kasan_arch_is_ready()

release_free_meta() accesses the shadow directly through the path

  kasan_slab_free
    __kasan_slab_free
      kasan_release_object_meta
        release_free_meta
          kasan_mem_to_shadow

There are no kasan_arch_is_ready() guards here, allowing an oops when the
shadow is not initialized.  The oops can be seen on a Power8 KVM guest.

This patch adds the guard to release_free_meta(), as it's the first level
that specifically requires the shadow.

It is safe to put the guard at the start of this function, before the
stack put: only kasan_save_free_info() can initialize the saved stack,
which itself is guarded with kasan_arch_is_ready() by its caller
poison_slab_object().  If the arch becomes ready before
release_free_meta() then we will not observe KASAN_SLAB_FREE_META in the
object's shadow, so we will not put an uninitialized stack either.

Link: https://lkml.kernel.org/r/20240213033958.139383-1-bgray@linux.ibm.com
Fixes: 63b85ac56a64 ("kasan: stop leaking stack trace handles")
Signed-off-by: Benjamin Gray <bgray@linux.ibm.com>
Reviewed-by: Andrey Konovalov <andreyknvl@gmail.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Vincenzo Frascino <vincenzo.frascino@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
6 months agomm/damon/lru_sort: fix quota status loss due to online tunings
SeongJae Park [Fri, 16 Feb 2024 19:40:25 +0000 (11:40 -0800)]
mm/damon/lru_sort: fix quota status loss due to online tunings

For online parameters change, DAMON_LRU_SORT creates new schemes based on
latest values of the parameters and replaces the old schemes with the new
one.  When creating it, the internal status of the quotas of the old
schemes is not preserved.  As a result, charging of the quota starts from
zero after the online tuning.  The data that collected to estimate the
throughput of the scheme's action is also reset, and therefore the
estimation should start from the scratch again.  Because the throughput
estimation is being used to convert the time quota to the effective size
quota, this could result in temporal time quota inaccuracy.  It would be
recovered over time, though.  In short, the quota accuracy could be
temporarily degraded after online parameters update.

Fix the problem by checking the case and copying the internal fields for
the status.

Link: https://lkml.kernel.org/r/20240216194025.9207-3-sj@kernel.org
Fixes: 40e983cca927 ("mm/damon: introduce DAMON-based LRU-lists Sorting")
Signed-off-by: SeongJae Park <sj@kernel.org>
Cc: <stable@vger.kernel.org> [6.0+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
6 months agomm/damon/reclaim: fix quota stauts loss due to online tunings
SeongJae Park [Fri, 16 Feb 2024 19:40:24 +0000 (11:40 -0800)]
mm/damon/reclaim: fix quota stauts loss due to online tunings

Patch series "mm/damon: fix quota status loss due to online tunings".

DAMON_RECLAIM and DAMON_LRU_SORT is not preserving internal quota status
when applying new user parameters, and hence could cause temporal quota
accuracy degradation.  Fix it by preserving the status.

This patch (of 2):

For online parameters change, DAMON_RECLAIM creates new scheme based on
latest values of the parameters and replaces the old scheme with the new
one.  When creating it, the internal status of the quota of the old
scheme is not preserved.  As a result, charging of the quota starts from
zero after the online tuning.  The data that collected to estimate the
throughput of the scheme's action is also reset, and therefore the
estimation should start from the scratch again.  Because the throughput
estimation is being used to convert the time quota to the effective size
quota, this could result in temporal time quota inaccuracy.  It would be
recovered over time, though.  In short, the quota accuracy could be
temporarily degraded after online parameters update.

Fix the problem by checking the case and copying the internal fields for
the status.

Link: https://lkml.kernel.org/r/20240216194025.9207-1-sj@kernel.org
Link: https://lkml.kernel.org/r/20240216194025.9207-2-sj@kernel.org
Fixes: e035c280f6df ("mm/damon/reclaim: support online inputs update")
Signed-off-by: SeongJae Park <sj@kernel.org>
Cc: <stable@vger.kernel.org> [5.19+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
6 months agoMAINTAINERS: mailmap: update Shakeel's email address
Shakeel Butt [Mon, 19 Feb 2024 20:50:50 +0000 (12:50 -0800)]
MAINTAINERS: mailmap: update Shakeel's email address

Moving to linux.dev based email for kernel work.

Link: https://lkml.kernel.org/r/20240219205050.887810-1-shakeel.butt@linux.dev
Signed-off-by: Shakeel Butt <shakeel.butt@linux.dev>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
6 months agomm/damon/sysfs-schemes: handle schemes sysfs dir removal before commit_schemes_quota_...
SeongJae Park [Tue, 13 Feb 2024 02:36:32 +0000 (18:36 -0800)]
mm/damon/sysfs-schemes: handle schemes sysfs dir removal before commit_schemes_quota_goals

'commit_schemes_quota_goals' command handler,
damos_sysfs_set_quota_scores() assumes the number of schemes sysfs
directory will be same to the number of schemes of the DAMON context.  The
assumption is wrong since users can remove schemes sysfs directories while
DAMON is running.  In the case, illegal memory accesses can happen.  Fix
it by checking the case.

Link: https://lkml.kernel.org/r/20240213023633.124928-1-sj@kernel.org
Fixes: d91beaa505a0 ("mm/damon/sysfs-schemes: implement a command for scheme quota goals only commit")
Signed-off-by: SeongJae Park <sj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
6 months agomm: memcontrol: clarify swapaccount=0 deprecation warning
Johannes Weiner [Tue, 13 Feb 2024 08:16:34 +0000 (03:16 -0500)]
mm: memcontrol: clarify swapaccount=0 deprecation warning

The swapaccount deprecation warning is throwing false positives.  Since we
deprecated the knob and defaulted to enabling, the only reports we've been
getting are from folks that set swapaccount=1.  While this is a nice
affirmation that always-enabling was the right choice, we certainly don't
want to warn when users request the supported mode.

Only warn when disabling is requested, and clarify the warning.

[colin.i.king@gmail.com: spelling: "commdandline" -> "commandline"]
Link: https://lkml.kernel.org/r/20240215090544.1649201-1-colin.i.king@gmail.com
Link: https://lkml.kernel.org/r/20240213081634.3652326-1-hannes@cmpxchg.org
Fixes: b25806dcd3d5 ("mm: memcontrol: deprecate swapaccounting=0 mode")
Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Reported-by: "Jonas Schäfer" <jonas@wielicki.name>
Reported-by: Narcis Garcia <debianlists@actiu.net>
Suggested-by: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Yosry Ahmed <yosryahmed@google.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Shakeel Butt <shakeelb@google.com>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
6 months agomm/memblock: add MEMBLOCK_RSRV_NOINIT into flagname[] array
Anshuman Khandual [Fri, 9 Feb 2024 03:09:12 +0000 (08:39 +0530)]
mm/memblock: add MEMBLOCK_RSRV_NOINIT into flagname[] array

The commit 77e6c43e137c ("memblock: introduce MEMBLOCK_RSRV_NOINIT flag")
skipped adding this newly introduced memblock flag into flagname[] array,
thus preventing a correct memblock flags output for applicable memblock
regions.

Link: https://lkml.kernel.org/r/20240209030912.1382251-1-anshuman.khandual@arm.com
Fixes: 77e6c43e137c ("memblock: introduce MEMBLOCK_RSRV_NOINIT flag")
Signed-off-by: Anshuman Khandual <anshuman.khandual@arm.com>
Reviewed-by: Mike Rapoport <rppt@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
6 months agomm/zswap: invalidate duplicate entry when !zswap_enabled
Chengming Zhou [Thu, 8 Feb 2024 02:32:54 +0000 (02:32 +0000)]
mm/zswap: invalidate duplicate entry when !zswap_enabled

We have to invalidate any duplicate entry even when !zswap_enabled since
zswap can be disabled anytime.  If the folio store success before, then
got dirtied again but zswap disabled, we won't invalidate the old
duplicate entry in the zswap_store().  So later lru writeback may
overwrite the new data in swapfile.

Link: https://lkml.kernel.org/r/20240208023254.3873823-1-chengming.zhou@linux.dev
Fixes: 42c06a0e8ebe ("mm: kill frontswap")
Signed-off-by: Chengming Zhou <zhouchengming@bytedance.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
6 months agolib/Kconfig.debug: TEST_IOV_ITER depends on MMU
Guenter Roeck [Thu, 8 Feb 2024 15:30:10 +0000 (07:30 -0800)]
lib/Kconfig.debug: TEST_IOV_ITER depends on MMU

Trying to run the iov_iter unit test on a nommu system such as the qemu
kc705-nommu emulation results in a crash.

    KTAP version 1
    # Subtest: iov_iter
    # module: kunit_iov_iter
    1..9
BUG: failure at mm/nommu.c:318/vmap()!
Kernel panic - not syncing: BUG!

The test calls vmap() directly, but vmap() is not supported on nommu
systems, causing the crash.  TEST_IOV_ITER therefore needs to depend on
MMU.

Link: https://lkml.kernel.org/r/20240208153010.1439753-1-linux@roeck-us.net
Fixes: 2d71340ff1d4 ("iov_iter: Kunit tests for copying to/from an iterator")
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Cc: David Howells <dhowells@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
6 months agomm/swap: fix race when skipping swapcache
Kairui Song [Tue, 6 Feb 2024 18:25:59 +0000 (02:25 +0800)]
mm/swap: fix race when skipping swapcache

When skipping swapcache for SWP_SYNCHRONOUS_IO, if two or more threads
swapin the same entry at the same time, they get different pages (A, B).
Before one thread (T0) finishes the swapin and installs page (A) to the
PTE, another thread (T1) could finish swapin of page (B), swap_free the
entry, then swap out the possibly modified page reusing the same entry.
It breaks the pte_same check in (T0) because PTE value is unchanged,
causing ABA problem.  Thread (T0) will install a stalled page (A) into the
PTE and cause data corruption.

One possible callstack is like this:

CPU0                                 CPU1
----                                 ----
do_swap_page()                       do_swap_page() with same entry
<direct swapin path>                 <direct swapin path>
<alloc page A>                       <alloc page B>
swap_read_folio() <- read to page A  swap_read_folio() <- read to page B
<slow on later locks or interrupt>   <finished swapin first>
...                                  set_pte_at()
                                     swap_free() <- entry is free
                                     <write to page B, now page A stalled>
                                     <swap out page B to same swap entry>
pte_same() <- Check pass, PTE seems
              unchanged, but page A
              is stalled!
swap_free() <- page B content lost!
set_pte_at() <- staled page A installed!

And besides, for ZRAM, swap_free() allows the swap device to discard the
entry content, so even if page (B) is not modified, if swap_read_folio()
on CPU0 happens later than swap_free() on CPU1, it may also cause data
loss.

To fix this, reuse swapcache_prepare which will pin the swap entry using
the cache flag, and allow only one thread to swap it in, also prevent any
parallel code from putting the entry in the cache.  Release the pin after
PT unlocked.

Racers just loop and wait since it's a rare and very short event.  A
schedule_timeout_uninterruptible(1) call is added to avoid repeated page
faults wasting too much CPU, causing livelock or adding too much noise to
perf statistics.  A similar livelock issue was described in commit
029c4628b2eb ("mm: swap: get rid of livelock in swapin readahead")

Reproducer:

This race issue can be triggered easily using a well constructed
reproducer and patched brd (with a delay in read path) [1]:

With latest 6.8 mainline, race caused data loss can be observed easily:
$ gcc -g -lpthread test-thread-swap-race.c && ./a.out
  Polulating 32MB of memory region...
  Keep swapping out...
  Starting round 0...
  Spawning 65536 workers...
  32746 workers spawned, wait for done...
  Round 0: Error on 0x5aa00, expected 32746, got 32743, 3 data loss!
  Round 0: Error on 0x395200, expected 32746, got 32743, 3 data loss!
  Round 0: Error on 0x3fd000, expected 32746, got 32737, 9 data loss!
  Round 0 Failed, 15 data loss!

This reproducer spawns multiple threads sharing the same memory region
using a small swap device.  Every two threads updates mapped pages one by
one in opposite direction trying to create a race, with one dedicated
thread keep swapping out the data out using madvise.

The reproducer created a reproduce rate of about once every 5 minutes, so
the race should be totally possible in production.

After this patch, I ran the reproducer for over a few hundred rounds and
no data loss observed.

Performance overhead is minimal, microbenchmark swapin 10G from 32G
zram:

Before:     10934698 us
After:      11157121 us
Cached:     13155355 us (Dropping SWP_SYNCHRONOUS_IO flag)

[kasong@tencent.com: v4]
Link: https://lkml.kernel.org/r/20240219082040.7495-1-ryncsn@gmail.com
Link: https://lkml.kernel.org/r/20240206182559.32264-1-ryncsn@gmail.com
Fixes: 0bcac06f27d7 ("mm, swap: skip swapcache for swapin of synchronous device")
Reported-by: "Huang, Ying" <ying.huang@intel.com>
Closes: https://lore.kernel.org/lkml/87bk92gqpx.fsf_-_@yhuang6-desk2.ccr.corp.intel.com/
Link: https://github.com/ryncsn/emm-test-project/tree/master/swap-stress-race
Signed-off-by: Kairui Song <kasong@tencent.com>
Reviewed-by: "Huang, Ying" <ying.huang@intel.com>
Acked-by: Yu Zhao <yuzhao@google.com>
Acked-by: David Hildenbrand <david@redhat.com>
Acked-by: Chris Li <chrisl@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Yosry Ahmed <yosryahmed@google.com>
Cc: Yu Zhao <yuzhao@google.com>
Cc: Barry Song <21cnbao@gmail.com>
Cc: SeongJae Park <sj@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
6 months agomm/swap_state: update zswap LRU's protection range with the folio locked
Nhat Pham [Mon, 5 Feb 2024 23:24:42 +0000 (15:24 -0800)]
mm/swap_state: update zswap LRU's protection range with the folio locked

When a folio is swapped in, the protection size of the corresponding zswap
LRU is incremented, so that the zswap shrinker is more conservative with
its reclaiming action.  This field is embedded within the struct lruvec,
so updating it requires looking up the folio's memcg and lruvec.  However,
currently this lookup can happen after the folio is unlocked, for instance
if a new folio is allocated, and swap_read_folio() unlocks the folio
before returning.  In this scenario, there is no stability guarantee for
the binding between a folio and its memcg and lruvec:

* A folio's memcg and lruvec can be freed between the lookup and the
  update, leading to a UAF.
* Folio migration can clear the now-unlocked folio's memcg_data, which
  directs the zswap LRU protection size update towards the root memcg
  instead of the original memcg. This was recently picked up by the
  syzbot thanks to a warning in the inlined folio_lruvec() call.

Move the zswap LRU protection range update above the swap_read_folio()
call, and only when a new page is allocated, to prevent this.

[nphamcs@gmail.com: add VM_WARN_ON_ONCE() to zswap_folio_swapin()]
Link: https://lkml.kernel.org/r/20240206180855.3987204-1-nphamcs@gmail.com
[nphamcs@gmail.com: remove unneeded if (folio) checks]
Link: https://lkml.kernel.org/r/20240206191355.83755-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20240205232442.3240571-1-nphamcs@gmail.com
Fixes: b5ba474f3f51 ("zswap: shrink zswap pool based on memory pressure")
Reported-by: syzbot+17a611d10af7d18a7092@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/all/000000000000ae47f90610803260@google.com/
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Reviewed-by: Chengming Zhou <zhouchengming@bytedance.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
6 months agoselftests/mm: uffd-unit-test check if huge page size is 0
Terry Tritton [Mon, 5 Feb 2024 14:50:56 +0000 (14:50 +0000)]
selftests/mm: uffd-unit-test check if huge page size is 0

If HUGETLBFS is not enabled then the default_huge_page_size function will
return 0 and cause a divide by 0 error. Add a check to see if the huge page
size is 0 and skip the hugetlb tests if it is.

Link: https://lkml.kernel.org/r/20240205145055.3545806-2-terry.tritton@linaro.org
Fixes: 16a45b57cbf2 ("selftests/mm: add framework for uffd-unit-test")
Signed-off-by: Terry Tritton <terry.tritton@linaro.org>
Cc: Peter Griffin <peter.griffin@linaro.org>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Peter Xu <peterx@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
6 months agomm/damon/core: check apply interval in damon_do_apply_schemes()
SeongJae Park [Mon, 5 Feb 2024 20:13:06 +0000 (12:13 -0800)]
mm/damon/core: check apply interval in damon_do_apply_schemes()

kdamond_apply_schemes() checks apply intervals of schemes and avoid
further applying any schemes if no scheme passed its apply interval.
However, the following schemes applying function, damon_do_apply_schemes()
iterates all schemes without the apply interval check.  As a result, the
shortest apply interval is applied to all schemes.  Fix the problem by
checking the apply interval in damon_do_apply_schemes().

Link: https://lkml.kernel.org/r/20240205201306.88562-1-sj@kernel.org
Fixes: 42f994b71404 ("mm/damon/core: implement scheme-specific apply interval")
Signed-off-by: SeongJae Park <sj@kernel.org>
Cc: <stable@vger.kernel.org> [6.7.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
6 months agomm: zswap: fix missing folio cleanup in writeback race path
Yosry Ahmed [Thu, 25 Jan 2024 08:51:27 +0000 (08:51 +0000)]
mm: zswap: fix missing folio cleanup in writeback race path

In zswap_writeback_entry(), after we get a folio from
__read_swap_cache_async(), we grab the tree lock again to check that the
swap entry was not invalidated and recycled.  If it was, we delete the
folio we just added to the swap cache and exit.

However, __read_swap_cache_async() returns the folio locked when it is
newly allocated, which is always true for this path, and the folio is
ref'd.  Make sure to unlock and put the folio before returning.

This was discovered by code inspection, probably because this path handles
a race condition that should not happen often, and the bug would not crash
the system, it will only strand the folio indefinitely.

Link: https://lkml.kernel.org/r/20240125085127.1327013-1-yosryahmed@google.com
Fixes: 04fc7816089c ("mm: fix zswap writeback race condition")
Signed-off-by: Yosry Ahmed <yosryahmed@google.com>
Reviewed-by: Chengming Zhou <zhouchengming@bytedance.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Nhat Pham <nphamcs@gmail.com>
Cc: Domenico Cerasuolo <cerasuolodomenico@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>