linux-2.6-microblaze.git
2 years agoMerge tag 'pinctrl-v5.18-3' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw...
Linus Torvalds [Thu, 19 May 2022 16:02:41 +0000 (06:02 -1000)]
Merge tag 'pinctrl-v5.18-3' of git://git./linux/kernel/git/linusw/linux-pinctrl

Pull pin control fixes from Linus Walleij:

 - Fix an altmode in the Ocelot driver

 - Fix the IES control pins in the Mediatek MT8365 driver

 - Sunxi (AMLogic) driver:
    - Fix the UART2 function pin assignments
    - Fix the signal name of the PA2 SPI pin

* tag 'pinctrl-v5.18-3' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl:
  pinctrl: sunxi: f1c100s: Fix signal name comment for PA2 SPI pin
  pinctrl: sunxi: fix f1c100s uart2 function
  pinctrl: mediatek: mt8365: fix IES control pins
  pinctrl: ocelot: Fix for lan966x alt mode

2 years agoMerge tag 'net-5.18-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Linus Torvalds [Thu, 19 May 2022 15:50:29 +0000 (05:50 -1000)]
Merge tag 'net-5.18-rc8' of git://git./linux/kernel/git/netdev/net

Pull networking fixes from Paolo Abeni:
 "Including fixes from can, xfrm and netfilter subtrees.

  Notably this reverts a recent TCP/DCCP netns-related change to address
  a possible UaF.

  Current release - regressions:

   - tcp: revert "tcp/dccp: get rid of inet_twsk_purge()"

   - xfrm: set dst dev to blackhole_netdev instead of loopback_dev in
     ifdown

  Previous releases - regressions:

   - netfilter: flowtable: fix TCP flow teardown

   - can: revert "can: m_can: pci: use custom bit timings for Elkhart
     Lake"

   - xfrm: check encryption module availability consistency

   - eth: vmxnet3: fix possible use-after-free bugs in
     vmxnet3_rq_alloc_rx_buf()

   - eth: mlx5: initialize flow steering during driver probe

   - eth: ice: fix crash when writing timestamp on RX rings

  Previous releases - always broken:

   - mptcp: fix checksum byte order

   - eth: lan966x: fix assignment of the MAC address

   - eth: mlx5: remove HW-GRO from reported features

   - eth: ftgmac100: disable hardware checksum on AST2600"

* tag 'net-5.18-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (50 commits)
  net: bridge: Clear offload_fwd_mark when passing frame up bridge interface.
  ptp: ocp: change sysfs attr group handling
  selftests: forwarding: fix missing backslash
  netfilter: nf_tables: disable expression reduction infra
  netfilter: flowtable: move dst_check to packet path
  netfilter: flowtable: fix TCP flow teardown
  net: ftgmac100: Disable hardware checksum on AST2600
  igb: skip phy status check where unavailable
  nfc: pn533: Fix buggy cleanup order
  mptcp: Do TCP fallback on early DSS checksum failure
  mptcp: fix checksum byte order
  net: af_key: check encryption module availability consistency
  net: af_key: add check for pfkey_broadcast in function pfkey_process
  net/mlx5: Drain fw_reset when removing device
  net/mlx5e: CT: Fix setting flow_source for smfs ct tuples
  net/mlx5e: CT: Fix support for GRE tuples
  net/mlx5e: Remove HW-GRO from reported features
  net/mlx5e: Properly block HW GRO when XDP is enabled
  net/mlx5e: Properly block LRO when XDP is enabled
  net/mlx5e: Block rx-gro-hw feature in switchdev mode
  ...

2 years agommc: core: Fix busy polling for MMC_SEND_OP_COND again
Ulf Hansson [Tue, 17 May 2022 10:10:46 +0000 (12:10 +0200)]
mmc: core: Fix busy polling for MMC_SEND_OP_COND again

It turned out that polling period for MMC_SEND_OP_COND, that currently is
set to 1ms, still isn't sufficient. In particular a Micron eMMC on a
Beaglebone platform, is reported to sometimes fail to initialize.

Additional test, shows that extending the period to 4ms is working fine, so
let's make that change.

Reported-by: Jean Rene Dawin <jdawin@math.uni-bielefeld.de>
Tested-by: Jean Rene Dawin <jdawin@math.uni-bielefeld.de>
Fixes: 1760fdb6fe9f (mmc: core: Restore (almost) the busy polling for MMC_SEND_OP_COND")
Fixes: 76bfc7ccc2fa ("mmc: core: adjust polling interval for CMD1")
Cc: stable@vger.kernel.org
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Link: https://lore.kernel.org/r/20220517101046.27512-1-ulf.hansson@linaro.org
2 years agobfq: Remove bfq_requeue_request_body()
Jan Kara [Thu, 19 May 2022 10:52:32 +0000 (12:52 +0200)]
bfq: Remove bfq_requeue_request_body()

The function has only a single caller and two lines. Just remove it
since it is pointless and just harming readability.

Signed-off-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220519105235.31397-4-jack@suse.cz
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agobfq: Remove superfluous conversion from RQ_BIC()
Jan Kara [Thu, 19 May 2022 10:52:31 +0000 (12:52 +0200)]
bfq: Remove superfluous conversion from RQ_BIC()

We store struct bfq_io_cq pointer in rq->elv.priv[0] in bfq_init_rq().
Thus a call to icq_to_bic() in RQ_BIC() is wrong. Luckily it does no
harm currently because struct io_iq is the first one in struct
bfq_io_cq.

Signed-off-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220519105235.31397-3-jack@suse.cz
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agobfq: Allow current waker to defend against a tentative one
Jan Kara [Thu, 19 May 2022 10:52:30 +0000 (12:52 +0200)]
bfq: Allow current waker to defend against a tentative one

The code in bfq_check_waker() ignores wake up events from the current
waker. This makes it more likely we select a new tentative waker
although the current one is generating more wake up events. Treat
current waker the same way as any other process and allow it to reset
the waker detection logic.

Fixes: 71217df39dc6 ("block, bfq: make waker-queue detection more robust")
Signed-off-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220519105235.31397-2-jack@suse.cz
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agobfq: Relax waker detection for shared queues
Jan Kara [Thu, 19 May 2022 10:52:29 +0000 (12:52 +0200)]
bfq: Relax waker detection for shared queues

Currently we look for waker only if current queue has no requests. This
makes sense for bfq queues with a single process however for shared
queues when there is a larger number of processes the condition that
queue has no requests is difficult to meet because often at least one
process has some request in flight although all the others are waiting
for the waker to do the work and this harms throughput. Relax the "no
queued request for bfq queue" condition to "the current task has no
queued requests yet". For this, we also need to start tracking number of
requests in flight for each task.

This patch (together with the following one) restores the performance
for dbench with 128 clients that regressed with commit c65e6fd460b4
("bfq: Do not let waker requests skip proper accounting") because
this commit makes requests of wakers properly enter BFQ queues and thus
these queues become ineligible for the old waker detection logic.
Dbench results:

         Vanilla 5.18-rc3        5.18-rc3 + revert      5.18-rc3 patched
Mean     1237.36 (   0.00%)      950.16 *  23.21%*      988.35 *  20.12%*

Numbers are time to complete workload so lower is better.

Fixes: c65e6fd460b4 ("bfq: Do not let waker requests skip proper accounting")
Signed-off-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220519105235.31397-1-jack@suse.cz
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agofs-writeback: writeback_sb_inodes:Recalculate 'wrote' according skipped pages
Zhihao Cheng [Tue, 10 May 2022 13:38:05 +0000 (21:38 +0800)]
fs-writeback: writeback_sb_inodes:Recalculate 'wrote' according skipped pages

Commit 505a666ee3fc ("writeback: plug writeback in wb_writeback() and
writeback_inodes_wb()") has us holding a plug during wb_writeback, which
may cause a potential ABBA dead lock:

    wb_writeback fat_file_fsync
blk_start_plug(&plug)
for (;;) {
  iter i-1: some reqs have been added into plug->mq_list  // LOCK A
  iter i:
    progress = __writeback_inodes_wb(wb, work)
    . writeback_sb_inodes // fat's bdev
    .   __writeback_single_inode
    .   . generic_writepages
    .   .   __block_write_full_page
    .   .   . .      __generic_file_fsync
    .   .   . .        sync_inode_metadata
    .   .   . .          writeback_single_inode
    .   .   . .    __writeback_single_inode
    .   .   . .      fat_write_inode
    .   .   . .        __fat_write_inode
    .   .   . .          sync_dirty_buffer // fat's bdev
    .   .   . .    lock_buffer(bh) // LOCK B
    .   .   . .      submit_bh
    .   .   . .        blk_mq_get_tag // LOCK A
    .   .   . trylock_buffer(bh)  // LOCK B
    .   .   .   redirty_page_for_writepage
    .   .   .     wbc->pages_skipped++
    .   .   --wbc->nr_to_write
    .   wrote += write_chunk - wbc.nr_to_write  // wrote > 0
    .   requeue_inode
    .     redirty_tail_locked
    if (progress)    // progress > 0
      continue;
  iter i+1:
      queue_io
      // similar process with iter i, infinite for-loop !
}
blk_finish_plug(&plug)   // flush plug won't be called

Above process triggers a hungtask like:
[  399.044861] INFO: task bb:2607 blocked for more than 30 seconds.
[  399.046824]       Not tainted 5.18.0-rc1-00005-gefae4d9eb6a2-dirty
[  399.051539] task:bb              state:D stack:    0 pid: 2607 ppid:
2426 flags:0x00004000
[  399.051556] Call Trace:
[  399.051570]  __schedule+0x480/0x1050
[  399.051592]  schedule+0x92/0x1a0
[  399.051602]  io_schedule+0x22/0x50
[  399.051613]  blk_mq_get_tag+0x1d3/0x3c0
[  399.051640]  __blk_mq_alloc_requests+0x21d/0x3f0
[  399.051657]  blk_mq_submit_bio+0x68d/0xca0
[  399.051674]  __submit_bio+0x1b5/0x2d0
[  399.051708]  submit_bio_noacct+0x34e/0x720
[  399.051718]  submit_bio+0x3b/0x150
[  399.051725]  submit_bh_wbc+0x161/0x230
[  399.051734]  __sync_dirty_buffer+0xd1/0x420
[  399.051744]  sync_dirty_buffer+0x17/0x20
[  399.051750]  __fat_write_inode+0x289/0x310
[  399.051766]  fat_write_inode+0x2a/0xa0
[  399.051783]  __writeback_single_inode+0x53c/0x6f0
[  399.051795]  writeback_single_inode+0x145/0x200
[  399.051803]  sync_inode_metadata+0x45/0x70
[  399.051856]  __generic_file_fsync+0xa3/0x150
[  399.051880]  fat_file_fsync+0x1d/0x80
[  399.051895]  vfs_fsync_range+0x40/0xb0
[  399.051929]  __x64_sys_fsync+0x18/0x30

In my test, 'need_resched()' (which is imported by 590dca3a71 "fs-writeback:
unplug before cond_resched in writeback_sb_inodes") in function
'writeback_sb_inodes()' seldom comes true, unless cond_resched() is deleted
from write_cache_pages().

Fix it by correcting wrote number according number of skipped pages
in writeback_sb_inodes().

Goto Link to find a reproducer.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=215837
Cc: stable@vger.kernel.org # v4.3
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220510133805.1988292-1-chengzhihao1@huawei.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agox86/boot: Wrap literal addresses in absolute_pointer()
Kees Cook [Sun, 27 Feb 2022 19:59:18 +0000 (11:59 -0800)]
x86/boot: Wrap literal addresses in absolute_pointer()

GCC 11 (incorrectly[1]) assumes that literal values cast to (void *)
should be treated like a NULL pointer with an offset, and raises
diagnostics when doing bounds checking under -Warray-bounds. GCC 12
got "smarter" about finding these:

  In function 'rdfs8',
      inlined from 'vga_recalc_vertical' at /srv/code/arch/x86/boot/video-mode.c:124:29,
      inlined from 'set_mode' at /srv/code/arch/x86/boot/video-mode.c:163:3:
  /srv/code/arch/x86/boot/boot.h:114:9: warning: array subscript 0 is outside array bounds of 'u8[0]' {aka 'unsigned char[]'} [-Warray-bounds]
    114 |         asm volatile("movb %%fs:%1,%0" : "=q" (v) : "m" (*(u8 *)addr));
        |         ^~~

This has been solved in other places[2] already by using the recently
added absolute_pointer() macro. Do the same here.

  [1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99578
  [2] https://lore.kernel.org/all/20210912160149.2227137-1-linux@roeck-us.net/

Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Guenter Roeck <linux@roeck-us.net>
Link: https://lore.kernel.org/r/20220227195918.705219-1-keescook@chromium.org
2 years agodrm/i915: Use i915_gem_object_ggtt_pin_ww for reloc_iomap
Maarten Lankhorst [Wed, 11 May 2022 11:52:19 +0000 (13:52 +0200)]
drm/i915: Use i915_gem_object_ggtt_pin_ww for reloc_iomap

When removing short term pins, I've changed the the batch buffer
pinning for relocation to use __i915_vma_pin, because
i915_gem_object_ggtt_pin_ww was destroying the old vma. This
caused regressions, because the functions are not identical.

Fix the regressions by calling i915_gem_object_ggtt_pin_ww() again
on ggtt-only platforms, but only if the batch can be pinned without
being moved.

Fixes: b5cfe6f7a6e1 ("drm/i915: Remove short-term pins from execbuf, v6.")
Cc: Matthew Auld <matthew.auld@intel.com>
Reported-by: Mateusz Jończyk <mat.jonczyk@o2.pl>
Tested-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Acked-by: Matthew Auld <matthew.auld@intel.com>
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/5806
Link: https://patchwork.freedesktop.org/patch/msgid/20220511115219.46507-1-maarten.lankhorst@linux.intel.com
(cherry picked from commit 451374eef622fca6f00eeeda89aaccb45a30a149)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
2 years agox86/sev: Mark the code returning to user space as syscall gap
Lai Jiangshan [Tue, 12 Apr 2022 12:49:08 +0000 (20:49 +0800)]
x86/sev: Mark the code returning to user space as syscall gap

When returning to user space, %rsp is user-controlled value.

If it is a SNP-guest and the hypervisor decides to mess with the
code-page for this path while a CPU is executing it, a potential #VC
could hit in the syscall return path and mislead the #VC handler.

So make ip_within_syscall_gap() return true in this case.

Signed-off-by: Lai Jiangshan <jiangshan.ljs@antgroup.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Joerg Roedel <jroedel@suse.de>
Link: https://lore.kernel.org/r/20220412124909.10467-1-jiangshanlai@gmail.com
2 years agoriscv/efi_stub: Add support for RISCV_EFI_BOOT_PROTOCOL
Sunil V L [Thu, 19 May 2022 05:15:12 +0000 (10:45 +0530)]
riscv/efi_stub: Add support for RISCV_EFI_BOOT_PROTOCOL

Add support for getting the boot hart ID from the Linux EFI stub using
RISCV_EFI_BOOT_PROTOCOL. This method is preferred over the existing DT
based approach since it works irrespective of DT or ACPI.

The specification of the protocol is hosted at:
https://github.com/riscv-non-isa/riscv-uefi

Signed-off-by: Sunil V L <sunilvl@ventanamicro.com>
Acked-by: Palmer Dabbelt <palmer@rivosinc.com>
Reviewed-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Link: https://lore.kernel.org/r/20220519051512.136724-2-sunilvl@ventanamicro.com
[ardb: minor tweaks for coding style and whitespace]
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
2 years agonet: bridge: Clear offload_fwd_mark when passing frame up bridge interface.
Andrew Lunn [Wed, 18 May 2022 00:58:40 +0000 (02:58 +0200)]
net: bridge: Clear offload_fwd_mark when passing frame up bridge interface.

It is possible to stack bridges on top of each other. Consider the
following which makes use of an Ethernet switch:

       br1
     /    \
    /      \
   /        \
 br0.11    wlan0
   |
   br0
 /  |  \
p1  p2  p3

br0 is offloaded to the switch. Above br0 is a vlan interface, for
vlan 11. This vlan interface is then a slave of br1. br1 also has a
wireless interface as a slave. This setup trunks wireless lan traffic
over the copper network inside a VLAN.

A frame received on p1 which is passed up to the bridge has the
skb->offload_fwd_mark flag set to true, indicating that the switch has
dealt with forwarding the frame out ports p2 and p3 as needed. This
flag instructs the software bridge it does not need to pass the frame
back down again. However, the flag is not getting reset when the frame
is passed upwards. As a result br1 sees the flag, wrongly interprets
it, and fails to forward the frame to wlan0.

When passing a frame upwards, clear the flag. This is the Rx
equivalent of br_switchdev_frame_unmark() in br_dev_xmit().

Fixes: f1c2eddf4cb6 ("bridge: switchdev: Use an helper to clear forward mark")
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Tested-by: Ido Schimmel <idosch@nvidia.com>
Acked-by: Nikolay Aleksandrov <razor@blackwall.org>
Link: https://lore.kernel.org/r/20220518005840.771575-1-andrew@lunn.ch
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agoptp: ocp: change sysfs attr group handling
Jonathan Lemon [Tue, 17 May 2022 21:46:00 +0000 (14:46 -0700)]
ptp: ocp: change sysfs attr group handling

In the detach path, the driver calls sysfs_remove_group() for the
groups it believes has been registered.  However, if the group was
never previously registered, then this causes a splat.

Instead, compute the groups that should be registered in advance,
and then call sysfs_create_groups(), which registers them all at once.

Update the error handling appropriately.

Fixes: c205d53c4923 ("ptp: ocp: Add firmware capability bits for feature gating")
Reported-by: Zheyu Ma <zheyuma97@gmail.com>
Signed-off-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Link: https://lore.kernel.org/r/20220517214600.10606-1-jonathan.lemon@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge tag 'amd-drm-fixes-5.18-2022-05-18' of https://gitlab.freedesktop.org/agd5f...
Dave Airlie [Thu, 19 May 2022 04:03:02 +0000 (14:03 +1000)]
Merge tag 'amd-drm-fixes-5.18-2022-05-18' of https://gitlab.freedesktop.org/agd5f/linux into drm-fixes

amd-drm-fixes-5.18-2022-05-18:

amdgpu:
- Suspend/resume regression fix

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220518202045.9123-1-alexander.deucher@amd.com
2 years agoselftests: forwarding: fix missing backslash
Joachim Wiberg [Wed, 18 May 2022 15:16:30 +0000 (17:16 +0200)]
selftests: forwarding: fix missing backslash

Fix missing backslash, introduced in f62c5acc800ee.  Causes all tests to
not be installed.

Fixes: f62c5acc800e ("selftests/net/forwarding: add missing tests to Makefile")
Signed-off-by: Joachim Wiberg <troglobit@gmail.com>
Acked-by: Hangbin Liu <liuhangbin@gmail.com>
Link: https://lore.kernel.org/r/20220518151630.2747773-1-troglobit@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf
Jakub Kicinski [Thu, 19 May 2022 02:34:25 +0000 (19:34 -0700)]
Merge git://git./linux/kernel/git/netfilter/nf

Pablo Neira Ayuso says:

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

1) Reduce number of hardware offload retries from flowtable datapath
   which might hog system with retries, from Felix Fietkau.

2) Skip neighbour lookup for PPPoE device, fill_forward_path() already
   provides this and set on destination address from fill_forward_path for
   PPPoE device, also from Felix.

4) When combining PPPoE on top of a VLAN device, set info->outdev to the
   PPPoE device so software offload works, from Felix.

5) Fix TCP teardown flowtable state, races with conntrack gc might result
   in resetting the state to ESTABLISHED and the time to one day. Joint
   work with Oz Shlomo and Sven Auhagen.

6) Call dst_check() from flowtable datapath to check if dst is stale
   instead of doing it from garbage collector path.

7) Disable register tracking infrastructure, either user-space or
   kernel need to pre-fetch keys inconditionally, otherwise register
   tracking assumes data is already available in register that might
   not well be there, leading to incorrect reductions.

* git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf:
  netfilter: nf_tables: disable expression reduction infra
  netfilter: flowtable: move dst_check to packet path
  netfilter: flowtable: fix TCP flow teardown
  netfilter: nft_flow_offload: fix offload with pppoe + vlan
  net: fix dev_fill_forward_path with pppoe + bridge
  netfilter: nft_flow_offload: skip dst neigh lookup for ppp devices
  netfilter: flowtable: fix excessive hw offload attempts after failure
====================

Link: https://lore.kernel.org/r/20220518213841.359653-1-pablo@netfilter.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge tag 'block-5.18-2022-05-18' of git://git.kernel.dk/linux-block
Linus Torvalds [Thu, 19 May 2022 00:32:27 +0000 (14:32 -1000)]
Merge tag 'block-5.18-2022-05-18' of git://git.kernel.dk/linux-block

Pull block fix from Jens Axboe:
 "Just a small fix for a missing fifo time assigment for the head
  insertion case in mq-deadline"

* tag 'block-5.18-2022-05-18' of git://git.kernel.dk/linux-block:
  block/mq-deadline: Set the fifo_time member also if inserting at head

2 years agoMerge tag 'io_uring-5.18-2022-05-18' of git://git.kernel.dk/linux-block
Linus Torvalds [Thu, 19 May 2022 00:21:30 +0000 (14:21 -1000)]
Merge tag 'io_uring-5.18-2022-05-18' of git://git.kernel.dk/linux-block

Pull io_uring fixes from Jens Axboe:
 "Two small changes fixing issues from the 5.18 merge window:

   - Fix wrong ordering of a tracepoint (Dylan)

   - Fix MSG_RING on IOPOLL rings (me)"

* tag 'io_uring-5.18-2022-05-18' of git://git.kernel.dk/linux-block:
  io_uring: don't attempt to IOPOLL for MSG_RING requests
  io_uring: fix ordering of args in io_uring_queue_async_work

2 years agoMerge tag 'audit-pr-20220518' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoor...
Linus Torvalds [Thu, 19 May 2022 00:19:26 +0000 (14:19 -1000)]
Merge tag 'audit-pr-20220518' of git://git./linux/kernel/git/pcmoore/audit

Pull audit fix from Paul Moore:
 "A single audit patch to fix a problem where a task's audit_context was
  not being properly reset with io_uring"

* tag 'audit-pr-20220518' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/audit:
  audit,io_uring,io-wq: call __audit_uring_exit for dummy contexts

2 years agoMerge tag 'selinux-pr-20220518' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Thu, 19 May 2022 00:15:35 +0000 (14:15 -1000)]
Merge tag 'selinux-pr-20220518' of git://git./linux/kernel/git/pcmoore/selinux

Pull selinux fix from Paul Moore:
 "A single SELinux patch to fix an error path that was doing the wrong
  thing with respect to freeing memory"

* tag 'selinux-pr-20220518' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux:
  selinux: fix bad cleanup on error in hashtab_duplicate()

2 years agoMerge branch 'arm/fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc
Linus Torvalds [Thu, 19 May 2022 00:07:43 +0000 (14:07 -1000)]
Merge branch 'arm/fixes' of git://git./linux/kernel/git/soc/soc

Pull ARM SoC fixes from Arnd Bergmann:
 "The SoC bug fixes have calmed down sufficiently, there is one minor
  update for the MAINTAINERS file, and few bug fixes for dts
  descriptions:

   - Updates to the BananaPi R2-Pro (rk3568) dts to match production
     hardware rather than the prototype version.

   - Qualcomm sm8250 soundwire gets disabled on some machines to avoid
     crashes

   - A number of aspeed SoC specific fixes, addressing incorrect pin
     cotrol settings, some values in the romed8hm board, and a revert
     for an accidental removal of a DT node"

* 'arm/fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc:
  MAINTAINERS: omap: remove me as a maintainer
  ARM: dts: aspeed: Add video engine to g6
  ARM: dts: aspeed: romed8hm3: Fix GPIOB0 name
  ARM: dts: aspeed: romed8hm3: Add lm25066 sense resistor values
  ARM: dts: aspeed-g6: fix SPI1/SPI2 quad pin group
  ARM: dts: aspeed-g6: add FWQSPI group in pinctrl dtsi
  dt-bindings: pinctrl: aspeed-g6: add FWQSPI function/group
  pinctrl: pinctrl-aspeed-g6: add FWQSPI function-group
  dt-bindings: pinctrl: aspeed-g6: remove FWQSPID group
  pinctrl: pinctrl-aspeed-g6: remove FWQSPID group in pinctrl
  ARM: dts: aspeed-g6: remove FWQSPID group in pinctrl dtsi
  arm64: dts: qcom: sm8250: don't enable rx/tx macro by default
  arm64: dts: rockchip: Add gmac1 and change network settings of bpi-r2-pro
  arm64: dts: rockchip: Change io-domains of bpi-r2-pro

2 years agoMerge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Thu, 19 May 2022 00:02:25 +0000 (14:02 -1000)]
Merge branch 'fixes' of git://git./linux/kernel/git/viro/vfs

Pull misc fixes from Al Viro:
 "vhost race fix and a percpu_ref_init-caused cgroup double-free fix.

  The latter had manifested as buggered struct mount refcounting - those
  are also using percpu data structures, but anything that does percpu
  allocations could be hit"

* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  Fix double fget() in vhost_net_set_backend()
  percpu_ref_init(): clean ->percpu_count_ref on failure

2 years agoMerge tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost
Linus Torvalds [Wed, 18 May 2022 23:53:53 +0000 (13:53 -1000)]
Merge tag 'for_linus' of git://git./linux/kernel/git/mst/vhost

Pull mlx5 fix from Michael Tsirkin:
 "One last minute fixup

  The patch has been on list for a while but as it was posted as part of
  a thread it was missed"

* tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost:
  vdpa/mlx5: Use consistent RQT size

2 years agoblk-cgroup: delete rcu_read_lock_held() WARN_ON_ONCE()
Jens Axboe [Wed, 18 May 2022 22:32:00 +0000 (16:32 -0600)]
blk-cgroup: delete rcu_read_lock_held() WARN_ON_ONCE()

A previous commit got rid of unnecessary rcu_read_lock() inside the
IRQ disabling queue_lock, but this debug statement was left. It's now
firing since we are indeed not inside a RCU read lock, but we don't
need to be as we're still preempt safe.

Get rid of the check, as we have a lockdep assert for holding the
queue lock right after it anyway.

Link: https://lore.kernel.org/linux-block/46253c48-81cb-0787-20ad-9133afdd9e21@samsung.com/
Reported-by: Marek Szyprowski <m.szyprowski@samsung.com>
Fixes: 77c570a1ea85 ("blk-cgroup: Remove unnecessary rcu_read_lock/unlock()")
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agoio_uring: disallow mixed provided buffer group registrations
Jens Axboe [Wed, 18 May 2022 21:34:37 +0000 (15:34 -0600)]
io_uring: disallow mixed provided buffer group registrations

It's nonsensical to register a provided buffer ring, if a classic
provided buffer group with the same ID exists. Depending on the order of
which we decide what type to pick, the other type will never get used.
Explicitly disallow it and return an error if this is attempted.

Fixes: c7fb19428d67 ("io_uring: add support for ring mapped supplied buffers")
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agoio_uring: initialize io_buffer_list head when shared ring is unregistered
Jens Axboe [Wed, 18 May 2022 20:36:18 +0000 (14:36 -0600)]
io_uring: initialize io_buffer_list head when shared ring is unregistered

We use ->buf_pages != 0 to tell if this is a shared buffer ring or a
classic provided buffer group. If we unregister the shared ring and
then attempt to use it, buf_pages is zero yet the classic list head
isn't properly initialized. This causes io_buffer_select() to think
that we have classic buffers available, but then we crash when we try
and get one from the list.

Just initialize the list if we unregister a shared buffer ring, leaving
it in a sane state for either re-registration or for attempting to use
it. And do the same for the initial setup from the classic path.

Fixes: c7fb19428d67 ("io_uring: add support for ring mapped supplied buffers")
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agoInput: ili210x - use one common reset implementation
Marek Vasut [Wed, 18 May 2022 21:30:09 +0000 (14:30 -0700)]
Input: ili210x - use one common reset implementation

Rename ili251x_hardware_reset() to ili210x_hardware_reset(), change its
parameter from struct device * to struct gpio_desc *, and use it as one
single consistent reset implementation all over the driver. Also increase
the minimum reset duration to 12ms, to make sure the reset is really
within the spec.

Signed-off-by: Marek Vasut <marex@denx.de>
Link: https://lore.kernel.org/r/20220518210423.106555-1-marex@denx.de
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
2 years agoInput: ili210x - fix reset timing
Marek Vasut [Wed, 18 May 2022 21:28:32 +0000 (14:28 -0700)]
Input: ili210x - fix reset timing

According to Ilitek "231x & ILI251x Programming Guide" Version: 2.30
"2.1. Power Sequence", "T4 Chip Reset and discharge time" is minimum
10ms and "T2 Chip initial time" is maximum 150ms. Adjust the reset
timings such that T4 is 12ms and T2 is 160ms to fit those figures.

This prevents sporadic touch controller start up failures when some
systems with at least ILI251x controller boot, without this patch
the systems sometimes fail to communicate with the touch controller.

Fixes: 201f3c803544c ("Input: ili210x - add reset GPIO support")
Signed-off-by: Marek Vasut <marex@denx.de>
Link: https://lore.kernel.org/r/20220518204901.93534-1-marex@denx.de
Cc: stable@vger.kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
2 years agodrm/amd: Don't reset dGPUs if the system is going to s2idle
Mario Limonciello [Tue, 17 May 2022 17:00:37 +0000 (12:00 -0500)]
drm/amd: Don't reset dGPUs if the system is going to s2idle

An A+A configuration on ASUS ROG Strix G513QY proves that the ASIC
reset for handling aborted suspend can't work with s2idle.

This functionality was introduced in commit daf8de0874ab5b ("drm/amdgpu:
always reset the asic in suspend (v2)").  A few other commits have
gone on top of the ASIC reset, but this still doesn't work on the A+A
configuration in s2idle.

Avoid doing the reset on dGPUs specifically when using s2idle.

Fixes: daf8de0874ab5b ("drm/amdgpu: always reset the asic in suspend (v2)")
Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2008
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
2 years agolibceph: fix misleading ceph_osdc_cancel_request() comment
Ilya Dryomov [Mon, 16 May 2022 15:17:54 +0000 (17:17 +0200)]
libceph: fix misleading ceph_osdc_cancel_request() comment

cancel_request() never guaranteed that after its return the OSD
client would be completely done with the OSD request.  The callback
(if specified) can still be invoked and a ref can still be held.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Xiubo Li <xiubli@redhat.com>
2 years agolibceph: fix potential use-after-free on linger ping and resends
Ilya Dryomov [Sat, 14 May 2022 10:16:47 +0000 (12:16 +0200)]
libceph: fix potential use-after-free on linger ping and resends

request_reinit() is not only ugly as the comment rightfully suggests,
but also unsafe.  Even though it is called with osdc->lock held for
write in all cases, resetting the OSD request refcount can still race
with handle_reply() and result in use-after-free.  Taking linger ping
as an example:

    handle_timeout thread                     handle_reply thread

                                              down_read(&osdc->lock)
                                              req = lookup_request(...)
                                              ...
                                              finish_request(req)  # unregisters
                                              up_read(&osdc->lock)
                                              __complete_request(req)
                                                linger_ping_cb(req)

      # req->r_kref == 2 because handle_reply still holds its ref

    down_write(&osdc->lock)
    send_linger_ping(lreq)
      req = lreq->ping_req  # same req
      # cancel_linger_request is NOT
      # called - handle_reply already
      # unregistered
      request_reinit(req)
        WARN_ON(req->r_kref != 1)  # fires
        request_init(req)
          kref_init(req->r_kref)

                   # req->r_kref == 1 after kref_init

                                              ceph_osdc_put_request(req)
                                                kref_put(req->r_kref)

            # req->r_kref == 0 after kref_put, req is freed

        <further req initialization/use> !!!

This happens because send_linger_ping() always (re)uses the same OSD
request for watch ping requests, relying on cancel_linger_request() to
unregister it from the OSD client and rip its messages out from the
messenger.  send_linger() does the same for watch/notify registration
and watch reconnect requests.  Unfortunately cancel_request() doesn't
guarantee that after it returns the OSD client would be completely done
with the OSD request -- a ref could still be held and the callback (if
specified) could still be invoked too.

The original motivation for request_reinit() was inability to deal with
allocation failures in send_linger() and send_linger_ping().  Switching
to using osdc->req_mempool (currently only used by CephFS) respects that
and allows us to get rid of request_reinit().

Cc: stable@vger.kernel.org
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Xiubo Li <xiubli@redhat.com>
Acked-by: Jeff Layton <jlayton@kernel.org>
2 years agoio_uring: add fully sparse buffer registration
Pavel Begunkov [Wed, 18 May 2022 18:13:49 +0000 (19:13 +0100)]
io_uring: add fully sparse buffer registration

Honour IORING_RSRC_REGISTER_SPARSE not only for direct files but fixed
buffers as well. It makes the rsrc API more consistent.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/66f429e4912fe39fb3318217ff33a2853d4544be.1652879898.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agox86/sev: Annotate stack change in the #VC handler
Lai Jiangshan [Wed, 16 Mar 2022 04:16:12 +0000 (12:16 +0800)]
x86/sev: Annotate stack change in the #VC handler

In idtentry_vc(), vc_switch_off_ist() determines a safe stack to
switch to, off of the IST stack. Annotate the new stack switch with
ENCODE_FRAME_POINTER in case UNWINDER_FRAME_POINTER is used.

A stack walk before looks like this:

  CPU: 0 PID: 0 Comm: swapper Not tainted 5.18.0-rc7+ #2
  Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015
  Call Trace:
   <TASK>
   dump_stack_lvl
   dump_stack
   kernel_exc_vmm_communication
   asm_exc_vmm_communication
   ? native_read_msr
   ? __x2apic_disable.part.0
   ? x2apic_setup
   ? cpu_init
   ? trap_init
   ? start_kernel
   ? x86_64_start_reservations
   ? x86_64_start_kernel
   ? secondary_startup_64_no_verify
   </TASK>

and with the fix, the stack dump is exact:

  CPU: 0 PID: 0 Comm: swapper Not tainted 5.18.0-rc7+ #3
  Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015
  Call Trace:
   <TASK>
   dump_stack_lvl
   dump_stack
   kernel_exc_vmm_communication
   asm_exc_vmm_communication
  RIP: 0010:native_read_msr
  Code: ...
  < snipped regs >
   ? __x2apic_disable.part.0
   x2apic_setup
   cpu_init
   trap_init
   start_kernel
   x86_64_start_reservations
   x86_64_start_kernel
   secondary_startup_64_no_verify
   </TASK>

  [ bp: Test in a SEV-ES guest and rewrite the commit message to
    explain what exactly this does. ]

Fixes: a13644f3a53d ("x86/entry/64: Add entry code for #VC handler")
Signed-off-by: Lai Jiangshan <jiangshan.ljs@antgroup.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Josh Poimboeuf <jpoimboe@redhat.com>
Link: https://lore.kernel.org/r/20220316041612.71357-1-jiangshanlai@gmail.com
2 years agonvme: add support for TP4084 - Time-to-Ready Enhancements
Christoph Hellwig [Mon, 16 May 2022 13:09:21 +0000 (15:09 +0200)]
nvme: add support for TP4084 - Time-to-Ready Enhancements

Add support for using longer timeouts during controller initialization
and letting the controller come up with namespaces that are not ready
for I/O yet.  We skip these not ready namespaces during scanning and
only bring them online once anoter scan is kicked off by the AEN that
is set when the NRDY bit gets set in the  I/O Command Set Independent
Identify Namespace Data Structure.   This asynchronous probing avoids
blocking the kernel boot when controllers take a very long time to
recover after unclean shutdowns (up to minutes).

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Keith Busch <kbusch@kernel.org>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
2 years agoFix double fget() in vhost_net_set_backend()
Al Viro [Mon, 16 May 2022 08:42:13 +0000 (16:42 +0800)]
Fix double fget() in vhost_net_set_backend()

Descriptor table is a shared resource; two fget() on the same descriptor
may return different struct file references.  get_tap_ptr_ring() is
called after we'd found (and pinned) the socket we'll be using and it
tries to find the private tun/tap data structures associated with it.
Redoing the lookup by the same file descriptor we'd used to get the
socket is racy - we need to same struct file.

Thanks to Jason for spotting a braino in the original variant of patch -
I'd missed the use of fd == -1 for disabling backend, and in that case
we can end up with sock == NULL and sock != oldsock.

Cc: stable@kernel.org
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2 years agovdpa/mlx5: Use consistent RQT size
Eli Cohen [Mon, 16 May 2022 08:47:35 +0000 (11:47 +0300)]
vdpa/mlx5: Use consistent RQT size

The current code evaluates RQT size based on the configured number of
virtqueues. This can raise an issue in the following scenario:

Assume MQ was negotiated.
1. mlx5_vdpa_set_map() gets called.
2. handle_ctrl_mq() is called setting cur_num_vqs to some value, lower
   than the configured max VQs.
3. A second set_map gets called, but now a smaller number of VQs is used
   to evaluate the size of the RQT.
4. handle_ctrl_mq() is called with a value larger than what the RQT can
   hold. This will emit errors and the driver state is compromised.

To fix this, we use a new field in struct mlx5_vdpa_net to hold the
required number of entries in the RQT. This value is evaluated in
mlx5_vdpa_set_driver_features() where we have the negotiated features
all set up.

In addition to that, we take into consideration the max capability of RQT
entries early when the device is added so we don't need to take consider
it when creating the RQT.

Last, we remove the use of mlx5_vdpa_max_qps() which just returns the
max_vas / 2 and make the code clearer.

Fixes: 52893733f2c5 ("vdpa/mlx5: Add multiqueue support")
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Eli Cohen <elic@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agoMerge tag 'sound-5.18' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound
Linus Torvalds [Wed, 18 May 2022 15:53:43 +0000 (05:53 -1000)]
Merge tag 'sound-5.18' of git://git./linux/kernel/git/tiwai/sound

Pull sound fixes from Takashi Iwai:
 "A collection of last-minute HD- an USB-audio quirks in addition to a
  fix for the legacy ISA wavefront driver.

  All look small and easy"

* tag 'sound-5.18' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
  ALSA: usb-audio: Restore Rane SL-1 quirk
  ALSA: hda/realtek: fix right sounds and mute/micmute LEDs for HP machine
  ALSA: hda/realtek: Add quirk for TongFang devices with pop noise
  ALSA: hda/realtek: Add quirk for the Framework Laptop
  ALSA: wavefront: Proper check of get_user() error
  ALSA: hda/realtek: Add quirk for Dell Latitude 7520
  ALSA: hda - fix unused Realtek function when PM is not enabled
  ALSA: usb-audio: Don't get sample rate for MCT Trigger 5 USB-to-HDMI

2 years agonetfilter: nf_tables: disable expression reduction infra
Pablo Neira Ayuso [Wed, 18 May 2022 12:51:34 +0000 (14:51 +0200)]
netfilter: nf_tables: disable expression reduction infra

Either userspace or kernelspace need to pre-fetch keys inconditionally
before comparisons for this to work. Otherwise, register tracking data
is misleading and it might result in reducing expressions which are not
yet registers.

First expression is also guaranteed to be evaluated always, however,
certain expressions break before writing data to registers, before
comparing the data, leaving the register in undetermined state.

This patch disables this infrastructure by now.

Fixes: b2d306542ff9 ("netfilter: nf_tables: do not reduce read-only expressions")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2 years agonetfilter: flowtable: move dst_check to packet path
Ritaro Takenaka [Tue, 17 May 2022 10:55:30 +0000 (12:55 +0200)]
netfilter: flowtable: move dst_check to packet path

Fixes sporadic IPv6 packet loss when flow offloading is enabled.

IPv6 route GC and flowtable GC are not synchronized.
When dst_cache becomes stale and a packet passes through the flow before
the flowtable GC teardowns it, the packet can be dropped.
So, it is necessary to check dst every time in packet path.

Fixes: 227e1e4d0d6c ("netfilter: nf_flowtable: skip device lookup from interface index")
Signed-off-by: Ritaro Takenaka <ritarot634@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2 years agonetfilter: flowtable: fix TCP flow teardown
Pablo Neira Ayuso [Tue, 17 May 2022 08:44:14 +0000 (10:44 +0200)]
netfilter: flowtable: fix TCP flow teardown

This patch addresses three possible problems:

1. ct gc may race to undo the timeout adjustment of the packet path, leaving
   the conntrack entry in place with the internal offload timeout (one day).

2. ct gc removes the ct because the IPS_OFFLOAD_BIT is not set and the CLOSE
   timeout is reached before the flow offload del.

3. tcp ct is always set to ESTABLISHED with a very long timeout
   in flow offload teardown/delete even though the state might be already
   CLOSED. Also as a remark we cannot assume that the FIN or RST packet
   is hitting flow table teardown as the packet might get bumped to the
   slow path in nftables.

This patch resets IPS_OFFLOAD_BIT from flow_offload_teardown(), so
conntrack handles the tcp rst/fin packet which triggers the CLOSE/FIN
state transition.

Moreover, teturn the connection's ownership to conntrack upon teardown
by clearing the offload flag and fixing the established timeout value.
The flow table GC thread will asynchonrnously free the flow table and
hardware offload entries.

Before this patch, the IPS_OFFLOAD_BIT remained set for expired flows on
which is also misleading since the flow is back to classic conntrack
path.

If nf_ct_delete() removes the entry from the conntrack table, then it
calls nf_ct_put() which decrements the refcnt. This is not a problem
because the flowtable holds a reference to the conntrack object from
flow_offload_alloc() path which is released via flow_offload_free().

This patch also updates nft_flow_offload to skip packets in SYN_RECV
state. Since we might miss or bump packets to slow path, we do not know
what will happen there while we are still in SYN_RECV, this patch
postpones offload up to the next packet which also aligns to the
existing behaviour in tc-ct.

flow_offload_teardown() does not reset the existing tcp state from
flow_offload_fixup_tcp() to ESTABLISHED anymore, packets bump to slow
path might have already update the state to CLOSE/FIN.

Joint work with Oz and Sven.

Fixes: 1e5b2471bcc4 ("netfilter: nf_flow_table: teardown flow timeout race")
Signed-off-by: Oz Shlomo <ozsh@nvidia.com>
Signed-off-by: Sven Auhagen <sven.auhagen@voleatech.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2 years agonet: ftgmac100: Disable hardware checksum on AST2600
Joel Stanley [Tue, 17 May 2022 09:22:17 +0000 (18:52 +0930)]
net: ftgmac100: Disable hardware checksum on AST2600

The AST2600 when using the i210 NIC over NC-SI has been observed to
produce incorrect checksum results with specific MTU values. This was
first observed when sending data across a long distance set of networks.

On a local network, the following test was performed using a 1MB file of
random data.

On the receiver run this script:

 #!/bin/bash
 while [ 1 ]; do
        # Zero the stats
        nstat -r  > /dev/null
        nc -l 9899 > test-file
        # Check for checksum errors
        TcpInCsumErrors=$(nstat | grep TcpInCsumErrors)
        if [ -z "$TcpInCsumErrors" ]; then
                echo No TcpInCsumErrors
        else
                echo TcpInCsumErrors = $TcpInCsumErrors
        fi
 done

On an AST2600 system:

 # nc <IP of  receiver host> 9899 < test-file

The test was repeated with various MTU values:

 # ip link set mtu 1410 dev eth0

The observed results:

 1500 - good
 1434 - bad
 1400 - good
 1410 - bad
 1420 - good

The test was repeated after disabling tx checksumming:

 # ethtool -K eth0 tx-checksumming off

And all MTU values tested resulted in transfers without error.

An issue with the driver cannot be ruled out, however there has been no
bug discovered so far.

David has done the work to take the original bug report of slow data
transfer between long distance connections and triaged it down to this
test case.

The vendor suspects this this is a hardware issue when using NC-SI. The
fixes line refers to the patch that introduced AST2600 support.

Reported-by: David Wilder <wilder@us.ibm.com>
Reviewed-by: Dylan Hung <dylan_hung@aspeedtech.com>
Signed-off-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoigb: skip phy status check where unavailable
Kevin Mitchell [Tue, 17 May 2022 18:01:05 +0000 (11:01 -0700)]
igb: skip phy status check where unavailable

igb_read_phy_reg() will silently return, leaving phy_data untouched, if
hw->ops.read_reg isn't set. Depending on the uninitialized value of
phy_data, this led to the phy status check either succeeding immediately
or looping continuously for 2 seconds before emitting a noisy err-level
timeout. This message went out to the console even though there was no
actual problem.

Instead, first check if there is read_reg function pointer. If not,
proceed without trying to check the phy status register.

Fixes: b72f3f72005d ("igb: When GbE link up, wait for Remote receiver status condition")
Signed-off-by: Kevin Mitchell <kevmitch@arista.com>
Tested-by: Gurucharan <gurucharanx.g@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonfc: pn533: Fix buggy cleanup order
Lin Ma [Wed, 18 May 2022 10:53:21 +0000 (18:53 +0800)]
nfc: pn533: Fix buggy cleanup order

When removing the pn533 device (i2c or USB), there is a logic error. The
original code first cancels the worker (flush_delayed_work) and then
destroys the workqueue (destroy_workqueue), leaving the timer the last
one to be deleted (del_timer). This result in a possible race condition
in a multi-core preempt-able kernel. That is, if the cleanup
(pn53x_common_clean) is concurrently run with the timer handler
(pn533_listen_mode_timer), the timer can queue the poll_work to the
already destroyed workqueue, causing use-after-free.

This patch reorder the cleanup: it uses the del_timer_sync to make sure
the handler is finished before the routine will destroy the workqueue.
Note that the timer cannot be activated by the worker again.

static void pn533_wq_poll(struct work_struct *work)
...
 rc = pn533_send_poll_frame(dev);
 if (rc)
   return;

 if (cur_mod->len == 0 && dev->poll_mod_count > 1)
   mod_timer(&dev->listen_timer, ...);

That is, the mod_timer can be called only when pn533_send_poll_frame()
returns no error, which is impossible because the device is detaching
and the lower driver should return ENODEV code.

Signed-off-by: Lin Ma <linma@zju.edu.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge tag 'nvme-5.19-2022-05-18' of git://git.infradead.org/nvme into for-5.19/drivers
Jens Axboe [Wed, 18 May 2022 12:28:04 +0000 (06:28 -0600)]
Merge tag 'nvme-5.19-2022-05-18' of git://git.infradead.org/nvme into for-5.19/drivers

Pull NVMe updates from Christoph:

"nvme updates for Linux 5.19

 - tighten the PCI presence check (Stefan Roese):
 - fix a potential NULL pointer dereference in an error path
   (Kyle Miller Smith)
 - fix interpretation of the DMRSL field (Tom Yan)
 - relax the data transfer alignment (Keith Busch)
 - verbose error logging improvements (Max Gurtovoy, Chaitanya Kulkarni)
 - misc cleanups (Chaitanya Kulkarni, me)"

* tag 'nvme-5.19-2022-05-18' of git://git.infradead.org/nvme:
  nvme: split the enum used for various register constants
  nvme-fabrics: add a request timeout helper
  nvme-pci: harden drive presence detect in nvme_dev_disable()
  nvme-pci: fix a NULL pointer dereference in nvme_alloc_admin_tags
  nvme: mark internal passthru request RQF_QUIET
  nvme: remove unneeded include from constants file
  nvme: add missing status values to verbose logging
  nvme: set dma alignment to dword
  nvme: fix interpretation of DMRSL

2 years agoio_uring: use rcu_dereference in io_close
Christoph Hellwig [Wed, 18 May 2022 08:40:05 +0000 (10:40 +0200)]
io_uring: use rcu_dereference in io_close

Accessing the file table needs a rcu_dereference_protected().

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220518084005.3255380-7-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agoio_uring: consistently use the EPOLL* defines
Christoph Hellwig [Wed, 18 May 2022 08:40:04 +0000 (10:40 +0200)]
io_uring: consistently use the EPOLL* defines

POLL* are unannotated values for the userspace ABI, while everything
in-kernel should use EPOLL* and the __poll_t type.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220518084005.3255380-6-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agoio_uring: make apoll_events a __poll_t
Christoph Hellwig [Wed, 18 May 2022 08:40:03 +0000 (10:40 +0200)]
io_uring: make apoll_events a __poll_t

apoll_events is fed to vfs_poll and the poll tables, so it should be
a __poll_t.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220518084005.3255380-5-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agoio_uring: drop a spurious inline on a forward declaration
Christoph Hellwig [Wed, 18 May 2022 08:40:02 +0000 (10:40 +0200)]
io_uring: drop a spurious inline on a forward declaration

io_file_get_normal isn't marked inline, so don't claim it as such in the
forward declaration.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220518084005.3255380-4-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agoio_uring: don't use ERR_PTR for user pointers
Christoph Hellwig [Wed, 18 May 2022 08:40:01 +0000 (10:40 +0200)]
io_uring: don't use ERR_PTR for user pointers

ERR_PTR abuses the high bits of a pointer to transport error information.
This is only safe for kernel pointers and not user pointers.  Fix
io_buffer_select and its helpers to just return NULL for failure and get
rid of this abuse.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220518084005.3255380-3-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agoio_uring: use a rwf_t for io_rw.flags
Christoph Hellwig [Wed, 18 May 2022 08:40:00 +0000 (10:40 +0200)]
io_uring: use a rwf_t for io_rw.flags

Use the proper type.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220518084005.3255380-2-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agoio_uring: add support for ring mapped supplied buffers
Jens Axboe [Sat, 30 Apr 2022 20:38:53 +0000 (14:38 -0600)]
io_uring: add support for ring mapped supplied buffers

Provided buffers allow an application to supply io_uring with buffers
that can then be grabbed for a read/receive request, when the data
source is ready to deliver data. The existing scheme relies on using
IORING_OP_PROVIDE_BUFFERS to do that, but it can be difficult to use
in real world applications. It's pretty efficient if the application
is able to supply back batches of provided buffers when they have been
consumed and the application is ready to recycle them, but if
fragmentation occurs in the buffer space, it can become difficult to
supply enough buffers at the time. This hurts efficiency.

Add a register op, IORING_REGISTER_PBUF_RING, which allows an application
to setup a shared queue for each buffer group of provided buffers. The
application can then supply buffers simply by adding them to this ring,
and the kernel can consume then just as easily. The ring shares the head
with the application, the tail remains private in the kernel.

Provided buffers setup with IORING_REGISTER_PBUF_RING cannot use
IORING_OP_{PROVIDE,REMOVE}_BUFFERS for adding or removing entries to the
ring, they must use the mapped ring. Mapped provided buffer rings can
co-exist with normal provided buffers, just not within the same group ID.

To gauge overhead of the existing scheme and evaluate the mapped ring
approach, a simple NOP benchmark was written. It uses a ring of 128
entries, and submits/completes 32 at the time. 'Replenish' is how
many buffers are provided back at the time after they have been
consumed:

Test Replenish NOPs/sec
================================================================
No provided buffers NA ~30M
Provided buffers 32 ~16M
Provided buffers  1 ~10M
Ring buffers 32 ~27M
Ring buffers  1 ~27M

The ring mapped buffers perform almost as well as not using provided
buffers at all, and they don't care if you provided 1 or more back at
the same time. This means application can just replenish as they go,
rather than need to batch and compact, further reducing overhead in the
application. The NOP benchmark above doesn't need to do any compaction,
so that overhead isn't even reflected in the above test.

Co-developed-by: Dylan Yudaken <dylany@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agoio_uring: add io_pin_pages() helper
Jens Axboe [Thu, 28 Apr 2022 19:02:27 +0000 (13:02 -0600)]
io_uring: add io_pin_pages() helper

Abstract this out from io_sqe_buffer_register() so we can use it
elsewhere too without duplicating this code.

No intended functional changes in this patch.

Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agoio_uring: add buffer selection support to IORING_OP_NOP
Jens Axboe [Fri, 6 May 2022 03:47:55 +0000 (21:47 -0600)]
io_uring: add buffer selection support to IORING_OP_NOP

Obviously not really useful since it's not transferring data, but it
is helpful in benchmarking overhead of provided buffers.

Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agoio_uring: fix locking state for empty buffer group
Jens Axboe [Sun, 15 May 2022 17:19:43 +0000 (11:19 -0600)]
io_uring: fix locking state for empty buffer group

io_provided_buffer_select() must drop the submit lock, if needed, even
in the error handling case. Failure to do so will leave us with the
ctx->uring_lock held, causing spew like:

====================================
WARNING: iou-wrk-366/368 still has locks held!
5.18.0-rc6-00294-gdf8dc7004331 #994 Not tainted
------------------------------------
1 lock held by iou-wrk-366/368:
 #0: ffff0000c72598a8 (&ctx->uring_lock){+.+.}-{3:3}, at: io_ring_submit_lock+0x20/0x48

stack backtrace:
CPU: 4 PID: 368 Comm: iou-wrk-366 Not tainted 5.18.0-rc6-00294-gdf8dc7004331 #994
Hardware name: linux,dummy-virt (DT)
Call trace:
 dump_backtrace.part.0+0xa4/0xd4
 show_stack+0x14/0x5c
 dump_stack_lvl+0x88/0xb0
 dump_stack+0x14/0x2c
 debug_check_no_locks_held+0x84/0x90
 try_to_freeze.isra.0+0x18/0x44
 get_signal+0x94/0x6ec
 io_wqe_worker+0x1d8/0x2b4
 ret_from_fork+0x10/0x20

and triggering later hangs off get_signal() because we attempt to
re-grab the lock.

Reported-by: syzbot+987d7bb19195ae45208c@syzkaller.appspotmail.com
Fixes: 149c69b04a90 ("io_uring: abstract out provided buffer list selection")
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agoMerge branch 'mptcp-checksums'
David S. Miller [Wed, 18 May 2022 12:05:43 +0000 (13:05 +0100)]
Merge branch 'mptcp-checksums'

Mat Martineau says:

====================
mptcp: Fix checksum byte order on little-endian

These patches address a bug in the byte ordering of MPTCP checksums on
little-endian architectures. The __sum16 type is always big endian, but
was being cast to u16 and then byte-swapped (on little-endian archs)
when reading/writing the checksum field in MPTCP option headers.

MPTCP checksums are off by default, but are enabled if one or both peers
request it in the SYN/SYNACK handshake.

The corrected code is verified to interoperate between big-endian and
little-endian machines.

Patch 1 fixes the checksum byte order, patch 2 partially mitigates
interoperation with peers sending bad checksums by falling back to TCP
instead of resetting the connection.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agomptcp: Do TCP fallback on early DSS checksum failure
Mat Martineau [Tue, 17 May 2022 18:02:12 +0000 (11:02 -0700)]
mptcp: Do TCP fallback on early DSS checksum failure

RFC 8684 section 3.7 describes several opportunities for a MPTCP
connection to "fall back" to regular TCP early in the connection
process, before it has been confirmed that MPTCP options can be
successfully propagated on all SYN, SYN/ACK, and data packets. If a peer
acknowledges the first received data packet with a regular TCP header
(no MPTCP options), fallback is allowed.

If the recipient of that first data packet finds a MPTCP DSS checksum
error, this provides an opportunity to fail gracefully with a TCP
fallback rather than resetting the connection (as might happen if a
checksum failure were detected later).

This commit modifies the checksum failure code to attempt fallback on
the initial subflow of a MPTCP connection, only if it's a failure in the
first data mapping. In cases where the peer initiates the connection,
requests checksums, is the first to send data, and the peer is sending
incorrect checksums (see
https://github.com/multipath-tcp/mptcp_net-next/issues/275), this allows
the connection to proceed as TCP rather than reset.

Fixes: dd8bcd1768ff ("mptcp: validate the data checksum")
Acked-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agomptcp: fix checksum byte order
Paolo Abeni [Tue, 17 May 2022 18:02:11 +0000 (11:02 -0700)]
mptcp: fix checksum byte order

The MPTCP code typecasts the checksum value to u16 and
then converts it to big endian while storing the value into
the MPTCP option.

As a result, the wire encoding for little endian host is
wrong, and that causes interoperabilty interoperability
issues with other implementation or host with different endianness.

Address the issue writing in the packet the unmodified __sum16 value.

MPTCP checksum is disabled by default, interoperating with systems
with bad mptcp-level csum encoding should cause fallback to TCP.

Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/275
Fixes: c5b39e26d003 ("mptcp: send out checksum for DSS")
Fixes: 390b95a5fb84 ("mptcp: receive checksum for DSS")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net...
David S. Miller [Wed, 18 May 2022 11:59:36 +0000 (12:59 +0100)]
Merge branch '100GbE' of git://git./linux/kernel/git/tnguy/net-queue

Tony Nguyen says:

====================
Intel Wired LAN Driver Updates 2022-05-17

This series contains updates to ice driver only.

Arkadiusz prevents writing of timestamps when rings are being
configured to resolve null pointer dereference.

Paul changes a delayed call to baseline statistics to occur immediately
which was causing misreporting of statistics due to the delay.

Michal fixes incorrect restoration of interrupt moderation settings.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec
David S. Miller [Wed, 18 May 2022 11:47:36 +0000 (12:47 +0100)]
Merge branch 'master' of git://git./linux/kernel/git/klassert/ipsec

Steffen Klassert says:

====================
pull request (net): ipsec 2022-05-18

1) Fix "disable_policy" flag use when arriving from different devices.
   From Eyal Birger.

2) Fix error handling of pfkey_broadcast in function pfkey_process.
   From Jiasheng Jiang.

3) Check the encryption module availability consistency in pfkey.
   From Thomas Bartschies.

Please pull or let me know if there are problems.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoARM: 9197/1: spectre-bhb: fix loop8 sequence for Thumb2
Ard Biesheuvel [Wed, 20 Apr 2022 08:46:17 +0000 (09:46 +0100)]
ARM: 9197/1: spectre-bhb: fix loop8 sequence for Thumb2

In Thumb2, 'b . + 4' produces a branch instruction that uses a narrow
encoding, and so it does not jump to the following instruction as
expected. So use W(b) instead.

Fixes: 6c7cb60bff7a ("ARM: fix Thumb2 regression with Spectre BHB")
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
2 years agoARM: 9196/1: spectre-bhb: enable for Cortex-A15
Ard Biesheuvel [Wed, 20 Apr 2022 08:44:51 +0000 (09:44 +0100)]
ARM: 9196/1: spectre-bhb: enable for Cortex-A15

The Spectre-BHB mitigations were inadvertently left disabled for
Cortex-A15, due to the fact that cpu_v7_bugs_init() is not called in
that case. So fix that.

Fixes: b9baf5c8c5c3 ("ARM: Spectre-BHB workaround")
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
2 years agoMerge tag 'mlx5-fixes-2022-05-17' of git://git.kernel.org/pub/scm/linux/kernel/git...
David S. Miller [Wed, 18 May 2022 10:33:44 +0000 (11:33 +0100)]
Merge tag 'mlx5-fixes-2022-05-17' of git://git./linux/kernel/git/saeed/linux

Saeed Mahameed says:

====================
mlx5 fixes 2022-05-17

This series provides bug fixes to mlx5 driver.
Please pull and let me know if there is any problem.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: af_key: check encryption module availability consistency
Thomas Bartschies [Wed, 18 May 2022 06:32:18 +0000 (08:32 +0200)]
net: af_key: check encryption module availability consistency

Since the recent introduction supporting the SM3 and SM4 hash algos for IPsec, the kernel
produces invalid pfkey acquire messages, when these encryption modules are disabled. This
happens because the availability of the algos wasn't checked in all necessary functions.
This patch adds these checks.

Signed-off-by: Thomas Bartschies <thomas.bartschies@cvk.de>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2 years agonet: af_key: add check for pfkey_broadcast in function pfkey_process
Jiasheng Jiang [Tue, 17 May 2022 09:42:31 +0000 (17:42 +0800)]
net: af_key: add check for pfkey_broadcast in function pfkey_process

If skb_clone() returns null pointer, pfkey_broadcast() will
return error.
Therefore, it should be better to check the return value of
pfkey_broadcast() and return error if fails.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Jiasheng Jiang <jiasheng@iscas.ac.cn>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2 years agopercpu_ref_init(): clean ->percpu_count_ref on failure
Al Viro [Wed, 18 May 2022 06:13:40 +0000 (02:13 -0400)]
percpu_ref_init(): clean ->percpu_count_ref on failure

That way percpu_ref_exit() is safe after failing percpu_ref_init().
At least one user (cgroup_create()) had a double-free that way;
there might be other similar bugs.  Easier to fix in percpu_ref_init(),
rather than playing whack-a-mole in sloppy users...

Usual symptoms look like a messed refcounting in one of subsystems
that use percpu allocations (might be percpu-refcount, might be
something else).  Having refcounts for two different objects share
memory is Not Nice(tm)...

Reported-by: syzbot+5b1e53987f858500ec00@syzkaller.appspotmail.com
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2 years agonet/mlx5: Drain fw_reset when removing device
Shay Drory [Mon, 4 Apr 2022 07:47:36 +0000 (10:47 +0300)]
net/mlx5: Drain fw_reset when removing device

In case fw sync reset is called in parallel to device removal, device
might stuck in the following deadlock:
         CPU 0                        CPU 1
         -----                        -----
                                  remove_one
                                   uninit_one (locks intf_state_mutex)
mlx5_sync_reset_now_event()
work in fw_reset->wq.
 mlx5_enter_error_state()
  mutex_lock (intf_state_mutex)
                                   cleanup_once
                                    fw_reset_cleanup()
                                     destroy_workqueue(fw_reset->wq)

Drain the fw_reset WQ, and make sure no new work is being queued, before
entering uninit_one().
The Drain is done before devlink_unregister() since fw_reset, in some
flows, is using devlink API devlink_remote_reload_actions_performed().

Fixes: 38b9f903f22b ("net/mlx5: Handle sync reset request event")
Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: CT: Fix setting flow_source for smfs ct tuples
Paul Blakey [Thu, 7 Apr 2022 10:37:32 +0000 (13:37 +0300)]
net/mlx5e: CT: Fix setting flow_source for smfs ct tuples

Cited patch sets flow_source to ANY overriding the provided spec
flow_source, avoiding the optimization done by commit c9c079b4deaa
("net/mlx5: CT: Set flow source hint from provided tuple device").

To fix the above, set the dr_rule flow_source from provided flow spec.

Fixes: 3ee61ebb0df1 ("net/mlx5: CT: Add software steering ct flow steering provider")
Signed-off-by: Paul Blakey <paulb@nvidia.com>
Reviewed-by: Oz Shlomo <ozsh@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: CT: Fix support for GRE tuples
Paul Blakey [Sun, 20 Mar 2022 12:02:24 +0000 (14:02 +0200)]
net/mlx5e: CT: Fix support for GRE tuples

cited commit removed support for GRE tuples when software steering was enabled.

To bring back support for GRE tuples, add GRE ipv4/ipv6 matchers.

Fixes: 3ee61ebb0df1 ("net/mlx5: CT: Add software steering ct flow steering provider")
Signed-off-by: Paul Blakey <paulb@nvidia.com>
Reviewed-by: Oz Shlomo <ozsh@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: Remove HW-GRO from reported features
Gal Pressman [Wed, 13 Apr 2022 12:50:42 +0000 (15:50 +0300)]
net/mlx5e: Remove HW-GRO from reported features

We got reports of certain HW-GRO flows causing kernel call traces, which
might be related to firmware. To be on the safe side, disable the
feature for now and re-enable it once a driver/firmware fix is found.

Fixes: 83439f3c37aa ("net/mlx5e: Add HW-GRO offload")
Signed-off-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: Properly block HW GRO when XDP is enabled
Maxim Mikityanskiy [Tue, 12 Apr 2022 15:54:26 +0000 (18:54 +0300)]
net/mlx5e: Properly block HW GRO when XDP is enabled

HW GRO is incompatible and mutually exclusive with XDP and XSK. However,
the needed checks are only made when enabling XDP. If HW GRO is enabled
when XDP is already active, the command will succeed, and XDP will be
skipped in the data path, although still enabled.

This commit fixes the bug by checking the XDP and XSK status in
mlx5e_fix_features and disabling HW GRO if XDP is enabled.

Fixes: 83439f3c37aa ("net/mlx5e: Add HW-GRO offload")
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: Properly block LRO when XDP is enabled
Maxim Mikityanskiy [Tue, 12 Apr 2022 15:37:03 +0000 (18:37 +0300)]
net/mlx5e: Properly block LRO when XDP is enabled

LRO is incompatible and mutually exclusive with XDP. However, the needed
checks are only made when enabling XDP. If LRO is enabled when XDP is
already active, the command will succeed, and XDP will be skipped in the
data path, although still enabled.

This commit fixes the bug by checking the XDP status in
mlx5e_fix_features and disabling LRO if XDP is enabled.

Fixes: 86994156c736 ("net/mlx5e: XDP fast RX drop bpf programs support")
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: Block rx-gro-hw feature in switchdev mode
Aya Levin [Mon, 11 Apr 2022 14:29:08 +0000 (17:29 +0300)]
net/mlx5e: Block rx-gro-hw feature in switchdev mode

When the driver is in switchdev mode and rx-gro-hw is set, the RQ needs
special CQE handling. Till then, block setting of rx-gro-hw feature in
switchdev mode, to avoid failure while setting the feature due to
failure while opening the RQ.

Fixes: f97d5c2a453e ("net/mlx5e: Add handle SHAMPO cqe support")
Signed-off-by: Aya Levin <ayal@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: Wrap mlx5e_trap_napi_poll into rcu_read_lock
Maxim Mikityanskiy [Mon, 4 Apr 2022 18:51:15 +0000 (21:51 +0300)]
net/mlx5e: Wrap mlx5e_trap_napi_poll into rcu_read_lock

The body of mlx5e_napi_poll is wrapped into rcu_read_lock to be able to
read the XDP program pointer using rcu_dereference. However, the trap RQ
NAPI doesn't use rcu_read_lock, because the trap RQ works only in the
non-linear mode, and mlx5e_skb_from_cqe_nonlinear, until recently,
didn't support XDP and didn't call rcu_dereference.

Starting from the cited commit, mlx5e_skb_from_cqe_nonlinear supports
XDP and calls rcu_dereference, but mlx5e_trap_napi_poll doesn't wrap it
into rcu_read_lock. It leads to RCU-lockdep warnings like this:

    WARNING: suspicious RCU usage

This commit fixes the issue by adding an rcu_read_lock to
mlx5e_trap_napi_poll, similarly to mlx5e_napi_poll.

Fixes: ea5d49bdae8b ("net/mlx5e: Add XDP multi buffer support to the non-linear legacy RQ")
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: DR, Ignore modify TTL on RX if device doesn't support it
Yevgeny Kliteynik [Sun, 3 Apr 2022 20:18:10 +0000 (23:18 +0300)]
net/mlx5: DR, Ignore modify TTL on RX if device doesn't support it

When modifying TTL, packet's csum has to be recalculated.
Due to HW issue in ConnectX-5, csum recalculation for modify
TTL on RX is supported through a work-around that is specifically
enabled by configuration.
If the work-around isn't enabled, rather than adding an unsupported
action the modify TTL action on RX should be ignored.
Ignoring modify TTL action might result in zero actions, so in such
cases we will not convert the match STE to modify STE, as it is done
by FW in DMFS.

This patch fixes an issue where modify TTL action was ignored both
on RX and TX instead of only on RX.

Fixes: 4ff725e1d4ad ("net/mlx5: DR, Ignore modify TTL if device doesn't support it")
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Reviewed-by: Alex Vesker <valex@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Initialize flow steering during driver probe
Shay Drory [Wed, 9 Mar 2022 12:45:58 +0000 (14:45 +0200)]
net/mlx5: Initialize flow steering during driver probe

Currently, software objects of flow steering are created and destroyed
during reload flow. In case a device is unloaded, the following error
is printed during grace period:

 mlx5_core 0000:00:0b.0: mlx5_fw_fatal_reporter_err_work:690:(pid 95):
    Driver is in error state. Unloading

As a solution to fix use-after-free bugs, where we try to access
these objects, when reading the value of flow_steering_mode devlink
param[1], let's split flow steering creation and destruction into two
routines:
    * init and cleanup: memory, cache, and pools allocation/free.
    * create and destroy: namespaces initialization and cleanup.

While at it, re-order the cleanup function to mirror the init function.

[1]
Kasan trace:

[  385.119849 ] BUG: KASAN: use-after-free in mlx5_devlink_fs_mode_get+0x3b/0xa0
[  385.119849 ] Read of size 4 at addr ffff888104b79308 by task bash/291
[  385.119849 ]
[  385.119849 ] CPU: 1 PID: 291 Comm: bash Not tainted 5.17.0-rc1+ #2
[  385.119849 ] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.13.0-2.fc32 04/01/2014
[  385.119849 ] Call Trace:
[  385.119849 ]  <TASK>
[  385.119849 ]  dump_stack_lvl+0x6e/0x91
[  385.119849 ]  print_address_description.constprop.0+0x1f/0x160
[  385.119849 ]  ? mlx5_devlink_fs_mode_get+0x3b/0xa0
[  385.119849 ]  ? mlx5_devlink_fs_mode_get+0x3b/0xa0
[  385.119849 ]  kasan_report.cold+0x83/0xdf
[  385.119849 ]  ? devlink_param_notify+0x20/0x190
[  385.119849 ]  ? mlx5_devlink_fs_mode_get+0x3b/0xa0
[  385.119849 ]  mlx5_devlink_fs_mode_get+0x3b/0xa0
[  385.119849 ]  devlink_nl_param_fill+0x18a/0xa50
[  385.119849 ]  ? _raw_spin_lock_irqsave+0x8d/0xe0
[  385.119849 ]  ? devlink_flash_update_timeout_notify+0xf0/0xf0
[  385.119849 ]  ? __wake_up_common+0x4b/0x1e0
[  385.119849 ]  ? preempt_count_sub+0x14/0xc0
[  385.119849 ]  ? _raw_spin_unlock_irqrestore+0x28/0x40
[  385.119849 ]  ? __wake_up_common_lock+0xe3/0x140
[  385.119849 ]  ? __wake_up_common+0x1e0/0x1e0
[  385.119849 ]  ? __sanitizer_cov_trace_const_cmp8+0x27/0x80
[  385.119849 ]  ? __rcu_read_unlock+0x48/0x70
[  385.119849 ]  ? kasan_unpoison+0x23/0x50
[  385.119849 ]  ? __kasan_slab_alloc+0x2c/0x80
[  385.119849 ]  ? memset+0x20/0x40
[  385.119849 ]  ? __sanitizer_cov_trace_const_cmp4+0x25/0x80
[  385.119849 ]  devlink_param_notify+0xce/0x190
[  385.119849 ]  devlink_unregister+0x92/0x2b0
[  385.119849 ]  remove_one+0x41/0x140
[  385.119849 ]  pci_device_remove+0x68/0x140
[  385.119849 ]  ? pcibios_free_irq+0x10/0x10
[  385.119849 ]  __device_release_driver+0x294/0x3f0
[  385.119849 ]  device_driver_detach+0x82/0x130
[  385.119849 ]  unbind_store+0x193/0x1b0
[  385.119849 ]  ? subsys_interface_unregister+0x270/0x270
[  385.119849 ]  drv_attr_store+0x4e/0x70
[  385.119849 ]  ? drv_attr_show+0x60/0x60
[  385.119849 ]  sysfs_kf_write+0xa7/0xc0
[  385.119849 ]  kernfs_fop_write_iter+0x23a/0x2f0
[  385.119849 ]  ? sysfs_kf_bin_read+0x160/0x160
[  385.119849 ]  new_sync_write+0x311/0x430
[  385.119849 ]  ? new_sync_read+0x480/0x480
[  385.119849 ]  ? _raw_spin_lock+0x87/0xe0
[  385.119849 ]  ? __sanitizer_cov_trace_cmp4+0x25/0x80
[  385.119849 ]  ? security_file_permission+0x94/0xa0
[  385.119849 ]  vfs_write+0x4c7/0x590
[  385.119849 ]  ksys_write+0xf6/0x1e0
[  385.119849 ]  ? __x64_sys_read+0x50/0x50
[  385.119849 ]  ? fpregs_assert_state_consistent+0x99/0xa0
[  385.119849 ]  do_syscall_64+0x3d/0x90
[  385.119849 ]  entry_SYSCALL_64_after_hwframe+0x44/0xae
[  385.119849 ] RIP: 0033:0x7fc36ef38504
[  385.119849 ] Code: 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b3 0f 1f
80 00 00 00 00 48 8d 05 f9 61 0d 00 8b 00 85 c0 75 13 b8 01 00 00 00 0f
05 <48> 3d 00 f0 ff ff 77 54 c3 0f 1f 00 41 54 49 89 d4 55 48 89 f5 53
[  385.119849 ] RSP: 002b:00007ffde0ff3d08 EFLAGS: 00000246 ORIG_RAX: 0000000000000001
[  385.119849 ] RAX: ffffffffffffffda RBX: 000000000000000c RCX: 00007fc36ef38504
[  385.119849 ] RDX: 000000000000000c RSI: 00007fc370521040 RDI: 0000000000000001
[  385.119849 ] RBP: 00007fc370521040 R08: 00007fc36f00b8c0 R09: 00007fc36ee4b740
[  385.119849 ] R10: 0000000000000000 R11: 0000000000000246 R12: 00007fc36f00a760
[  385.119849 ] R13: 000000000000000c R14: 00007fc36f005760 R15: 000000000000000c
[  385.119849 ]  </TASK>
[  385.119849 ]
[  385.119849 ] Allocated by task 65:
[  385.119849 ]  kasan_save_stack+0x1e/0x40
[  385.119849 ]  __kasan_kmalloc+0x81/0xa0
[  385.119849 ]  mlx5_init_fs+0x11b/0x1160
[  385.119849 ]  mlx5_load+0x13c/0x220
[  385.119849 ]  mlx5_load_one+0xda/0x160
[  385.119849 ]  mlx5_recover_device+0xb8/0x100
[  385.119849 ]  mlx5_health_try_recover+0x2f9/0x3a1
[  385.119849 ]  devlink_health_reporter_recover+0x75/0x100
[  385.119849 ]  devlink_health_report+0x26c/0x4b0
[  385.275909 ]  mlx5_fw_fatal_reporter_err_work+0x11e/0x1b0
[  385.275909 ]  process_one_work+0x520/0x970
[  385.275909 ]  worker_thread+0x378/0x950
[  385.275909 ]  kthread+0x1bb/0x200
[  385.275909 ]  ret_from_fork+0x1f/0x30
[  385.275909 ]
[  385.275909 ] Freed by task 65:
[  385.275909 ]  kasan_save_stack+0x1e/0x40
[  385.275909 ]  kasan_set_track+0x21/0x30
[  385.275909 ]  kasan_set_free_info+0x20/0x30
[  385.275909 ]  __kasan_slab_free+0xfc/0x140
[  385.275909 ]  kfree+0xa5/0x3b0
[  385.275909 ]  mlx5_unload+0x2e/0xb0
[  385.275909 ]  mlx5_unload_one+0x86/0xb0
[  385.275909 ]  mlx5_fw_fatal_reporter_err_work.cold+0xca/0xcf
[  385.275909 ]  process_one_work+0x520/0x970
[  385.275909 ]  worker_thread+0x378/0x950
[  385.275909 ]  kthread+0x1bb/0x200
[  385.275909 ]  ret_from_fork+0x1f/0x30
[  385.275909 ]
[  385.275909 ] The buggy address belongs to the object at ffff888104b79300
[  385.275909 ]  which belongs to the cache kmalloc-128 of size 128
[  385.275909 ] The buggy address is located 8 bytes inside of
[  385.275909 ]  128-byte region [ffff888104b79300ffff888104b79380)
[  385.275909 ] The buggy address belongs to the page:
[  385.275909 ] page:00000000de44dd39 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x104b78
[  385.275909 ] head:00000000de44dd39 order:1 compound_mapcount:0
[  385.275909 ] flags: 0x8000000000010200(slab|head|zone=2)
[  385.275909 ] raw: 8000000000010200 0000000000000000 dead000000000122 ffff8881000428c0
[  385.275909 ] raw: 0000000000000000 0000000080200020 00000001ffffffff 0000000000000000
[  385.275909 ] page dumped because: kasan: bad access detected
[  385.275909 ]
[  385.275909 ] Memory state around the buggy address:
[  385.275909 ]  ffff888104b79200: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 fc fc
[  385.275909 ]  ffff888104b79280: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
[  385.275909 ] >ffff888104b79300: fa fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[  385.275909 ]                       ^
[  385.275909 ]  ffff888104b79380: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
[  385.275909 ]  ffff888104b79400: fa fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[  385.275909 ]]

Fixes: e890acd5ff18 ("net/mlx5: Add devlink flow_steering_mode parameter")
Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: DR, Fix missing flow_source when creating multi-destination FW table
Maor Dickman [Mon, 21 Mar 2022 08:07:44 +0000 (10:07 +0200)]
net/mlx5: DR, Fix missing flow_source when creating multi-destination FW table

In order to support multiple destination FTEs with SW steering
FW table is created with single FTE with multiple actions and
SW steering rule forward to it. When creating this table, flow
source isn't set according to the original FTE.

Fix this by passing the original FTE flow source to the created
FW table.

Fixes: 34583beea4b7 ("net/mlx5: DR, Create multi-destination table for SW-steering use")
Signed-off-by: Maor Dickman <maord@nvidia.com>
Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agoscsi: target: Fix incorrect use of cpumask_t
Mingzhe Zou [Mon, 16 May 2022 05:47:21 +0000 (13:47 +0800)]
scsi: target: Fix incorrect use of cpumask_t

In commit d72d827f2f26, I used 'cpumask_t' incorrectly:

    void iscsit_thread_get_cpumask(struct iscsi_conn *conn)
    {
            int ord, cpu;
            cpumask_t conn_allowed_cpumask;
            ......
    }

    static ssize_t lio_target_wwn_cpus_allowed_list_store(
                   struct config_item *item, const char *page, size_t count)
    {
            int ret;
            char *orig;
            cpumask_t new_allowed_cpumask;
            ......
    }

The correct pattern should be as follows:

    cpumask_var_t mask;

    if (!zalloc_cpumask_var(&mask, GFP_KERNEL))
            return -ENOMEM;
    ... use 'mask' here ...
    free_cpumask_var(mask);

Link: https://lore.kernel.org/r/20220516054721.1548-1-mingzhe.zou@easystack.cn
Fixes: d72d827f2f26 ("scsi: target: Add iscsi/cpus_allowed_list in configfs")
Reported-by: Test Bot <zgrieee@gmail.com>
Reviewed-by: Mike Christie <michael.christie@oracle.com>
Signed-off-by: Mingzhe Zou <mingzhe.zou@easystack.cn>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2 years agoblk-throttle: Set BIO_THROTTLED when bio has been throttled
Laibin Qiu [Tue, 1 Mar 2022 12:39:19 +0000 (20:39 +0800)]
blk-throttle: Set BIO_THROTTLED when bio has been throttled

1.In current process, all bio will set the BIO_THROTTLED flag
after __blk_throtl_bio().

2.If bio needs to be throttled, it will start the timer and
stop submit bio directly. Bio will submit in
blk_throtl_dispatch_work_fn() when the timer expires.But in
the current process, if bio is throttled. The BIO_THROTTLED
will be set to bio after timer start. If the bio has been
completed, it may cause use-after-free blow.

BUG: KASAN: use-after-free in blk_throtl_bio+0x12f0/0x2c70
Read of size 2 at addr ffff88801b8902d4 by task fio/26380

 dump_stack+0x9b/0xce
 print_address_description.constprop.6+0x3e/0x60
 kasan_report.cold.9+0x22/0x3a
 blk_throtl_bio+0x12f0/0x2c70
 submit_bio_checks+0x701/0x1550
 submit_bio_noacct+0x83/0xc80
 submit_bio+0xa7/0x330
 mpage_readahead+0x380/0x500
 read_pages+0x1c1/0xbf0
 page_cache_ra_unbounded+0x471/0x6f0
 do_page_cache_ra+0xda/0x110
 ondemand_readahead+0x442/0xae0
 page_cache_async_ra+0x210/0x300
 generic_file_buffered_read+0x4d9/0x2130
 generic_file_read_iter+0x315/0x490
 blkdev_read_iter+0x113/0x1b0
 aio_read+0x2ad/0x450
 io_submit_one+0xc8e/0x1d60
 __se_sys_io_submit+0x125/0x350
 do_syscall_64+0x2d/0x40
 entry_SYSCALL_64_after_hwframe+0x44/0xa9

Allocated by task 26380:
 kasan_save_stack+0x19/0x40
 __kasan_kmalloc.constprop.2+0xc1/0xd0
 kmem_cache_alloc+0x146/0x440
 mempool_alloc+0x125/0x2f0
 bio_alloc_bioset+0x353/0x590
 mpage_alloc+0x3b/0x240
 do_mpage_readpage+0xddf/0x1ef0
 mpage_readahead+0x264/0x500
 read_pages+0x1c1/0xbf0
 page_cache_ra_unbounded+0x471/0x6f0
 do_page_cache_ra+0xda/0x110
 ondemand_readahead+0x442/0xae0
 page_cache_async_ra+0x210/0x300
 generic_file_buffered_read+0x4d9/0x2130
 generic_file_read_iter+0x315/0x490
 blkdev_read_iter+0x113/0x1b0
 aio_read+0x2ad/0x450
 io_submit_one+0xc8e/0x1d60
 __se_sys_io_submit+0x125/0x350
 do_syscall_64+0x2d/0x40
 entry_SYSCALL_64_after_hwframe+0x44/0xa9

Freed by task 0:
 kasan_save_stack+0x19/0x40
 kasan_set_track+0x1c/0x30
 kasan_set_free_info+0x1b/0x30
 __kasan_slab_free+0x111/0x160
 kmem_cache_free+0x94/0x460
 mempool_free+0xd6/0x320
 bio_free+0xe0/0x130
 bio_put+0xab/0xe0
 bio_endio+0x3a6/0x5d0
 blk_update_request+0x590/0x1370
 scsi_end_request+0x7d/0x400
 scsi_io_completion+0x1aa/0xe50
 scsi_softirq_done+0x11b/0x240
 blk_mq_complete_request+0xd4/0x120
 scsi_mq_done+0xf0/0x200
 virtscsi_vq_done+0xbc/0x150
 vring_interrupt+0x179/0x390
 __handle_irq_event_percpu+0xf7/0x490
 handle_irq_event_percpu+0x7b/0x160
 handle_irq_event+0xcc/0x170
 handle_edge_irq+0x215/0xb20
 common_interrupt+0x60/0x120
 asm_common_interrupt+0x1e/0x40

Fix this by move BIO_THROTTLED set into the queue_lock.

Signed-off-by: Laibin Qiu <qiulaibin@huawei.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20220301123919.2381579-1-qiulaibin@huawei.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agoNFC: nci: fix sleep in atomic context bugs caused by nci_skb_alloc
Duoming Zhou [Tue, 17 May 2022 01:25:30 +0000 (09:25 +0800)]
NFC: nci: fix sleep in atomic context bugs caused by nci_skb_alloc

There are sleep in atomic context bugs when the request to secure
element of st-nci is timeout. The root cause is that nci_skb_alloc
with GFP_KERNEL parameter is called in st_nci_se_wt_timeout which is
a timer handler. The call paths that could trigger bugs are shown below:

    (interrupt context 1)
st_nci_se_wt_timeout
  nci_hci_send_event
    nci_hci_send_data
      nci_skb_alloc(..., GFP_KERNEL) //may sleep

   (interrupt context 2)
st_nci_se_wt_timeout
  nci_hci_send_event
    nci_hci_send_data
      nci_send_data
        nci_queue_tx_data_frags
          nci_skb_alloc(..., GFP_KERNEL) //may sleep

This patch changes allocation mode of nci_skb_alloc from GFP_KERNEL to
GFP_ATOMIC in order to prevent atomic context sleeping. The GFP_ATOMIC
flag makes memory allocation operation could be used in atomic context.

Fixes: ed06aeefdac3 ("nfc: st-nci: Rename st21nfcb to st-nci")
Signed-off-by: Duoming Zhou <duoming@zju.edu.cn>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://lore.kernel.org/r/20220517012530.75714-1-duoming@zju.edu.cn
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet/qla3xxx: Fix a test in ql_reset_work()
Christophe JAILLET [Sun, 15 May 2022 18:07:02 +0000 (20:07 +0200)]
net/qla3xxx: Fix a test in ql_reset_work()

test_bit() tests if one bit is set or not.
Here the logic seems to check of bit QL_RESET_PER_SCSI (i.e. 4) OR bit
QL_RESET_START (i.e. 3) is set.

In fact, it checks if bit 7 (4 | 3 = 7) is set, that is to say
QL_ADAPTER_UP.

This looks harmless, because this bit is likely be set, and when the
ql_reset_work() delayed work is scheduled in ql3xxx_isr() (the only place
that schedule this work), QL_RESET_START or QL_RESET_PER_SCSI is set.

This has been spotted by smatch.

Fixes: 5a4faa873782 ("[PATCH] qla3xxx NIC driver")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/80e73e33f390001d9c0140ffa9baddf6466a41a2.1652637337.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge tag 'pci-v5.18-fixes-1' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaa...
Linus Torvalds [Tue, 17 May 2022 23:46:22 +0000 (13:46 -1000)]
Merge tag 'pci-v5.18-fixes-1' of git://git./linux/kernel/git/helgaas/pci

Pull PCI fixes from Bjorn Helgaas:

 - Avoid putting Elo i2 PCIe Ports in D3cold because downstream devices
   are inaccessible after going back to D0 (Rafael J. Wysocki)

 - Qualcomm SM8250 has a ddrss_sf_tbu clock but SC8180X does not; make a
   SC8180X-specific config without the clock so it probes correctly
   (Bjorn Andersson)

 - Revert aardvark chained IRQ handler rewrite because it broke
   interrupt affinity (Pali Rohár)

* tag 'pci-v5.18-fixes-1' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
  Revert "PCI: aardvark: Rewrite IRQ code to chained IRQ handler"
  PCI: qcom: Remove ddrss_sf_tbu clock from SC8180X
  PCI/PM: Avoid putting Elo i2 PCIe Ports in D3cold

2 years agoMerge tag 'thermal-5.18-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Tue, 17 May 2022 23:40:44 +0000 (13:40 -1000)]
Merge tag 'thermal-5.18-rc8' of git://git./linux/kernel/git/rafael/linux-pm

Pull thermal control fix from Rafael Wysocki:
 "Fix up a recent change in the int340x thermal driver that
  inadvertently broke thermal zone handling on some systems
  (Srinivas Pandruvada)"

* tag 'thermal-5.18-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  thermal: int340x: Mode setting with new OS handshake

2 years agoselinux: fix bad cleanup on error in hashtab_duplicate()
Ondrej Mosnacek [Tue, 17 May 2022 12:08:16 +0000 (14:08 +0200)]
selinux: fix bad cleanup on error in hashtab_duplicate()

The code attempts to free the 'new' pointer using kmem_cache_free(),
which is wrong because this function isn't responsible of freeing it.
Instead, the function should free new->htable and clear the contents of
*new (to prevent double-free).

Cc: stable@vger.kernel.org
Fixes: c7c556f1e81b ("selinux: refactor changing booleans")
Reported-by: Wander Lairson Costa <wander@redhat.com>
Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com>
Signed-off-by: Paul Moore <paul@paul-moore.com>
2 years agodrm/dp/mst: fix a possible memory leak in fetch_monitor_name()
Hangyu Hua [Mon, 16 May 2022 03:20:42 +0000 (11:20 +0800)]
drm/dp/mst: fix a possible memory leak in fetch_monitor_name()

drm_dp_mst_get_edid call kmemdup to create mst_edid. So mst_edid need to be
freed after use.

Signed-off-by: Hangyu Hua <hbh25y@gmail.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
Cc: stable@vger.kernel.org
Link: https://patchwork.freedesktop.org/patch/msgid/20220516032042.13166-1-hbh25y@gmail.com
2 years agoparisc: Fix patch code locking and flushing
John David Anglin [Mon, 16 May 2022 15:32:00 +0000 (15:32 +0000)]
parisc: Fix patch code locking and flushing

This change fixes the following:

1) The flags variable is not initialized. Always use raw_spin_lock_irqsave
and raw_spin_unlock_irqrestore to serialize patching.

2) flush_kernel_vmap_range is primarily intended for DMA flushes.
The whole cache flush in flush_kernel_vmap_range is only possible
when interrupts are enabled on SMP machines. Since __patch_text_multiple
calls flush_kernel_vmap_range with interrupts disabled, it is better
to directly call flush_kernel_dcache_range_asm and
flush_kernel_icache_range_asm.

3) The final call to flush_icache_range is unnecessary.

Tested with `[PATCH, V3] parisc: Rewrite cache flush code for
PA8800/PA8900' change on rp3440, c8000 and c3750 (32 and 64-bit).

Note by Helge:
This patch had been temporarily reverted shortly before v5.18-rc6 in order
to fix boot issues. Now it can be re-applied.

Signed-off-by: John David Anglin <dave.anglin@bell.net>
Signed-off-by: Helge Deller <deller@gmx.de>
2 years agoparisc: Rewrite cache flush code for PA8800/PA8900
John David Anglin [Mon, 16 May 2022 15:14:47 +0000 (15:14 +0000)]
parisc: Rewrite cache flush code for PA8800/PA8900

Originally, I was convinced that we needed to use tmpalias flushes
everwhere, for both user and kernel flushes. However, when I modified
flush_kernel_dcache_page_addr, to use a tmpalias flush, my c8000
would crash quite early when booting.

The PDC returns alias values of 0 for the icache and dcache. This
indicates that either the alias boundary is greater than 16MB or
equivalent aliasing doesn't work. I modified the tmpalias code to
make it easy to try alternate boundaries. I tried boundaries up to
128MB but still kernel tmpalias flushes didn't work on c8000.

This led me to conclude that tmpalias flushes don't work on PA8800
and PA8900 machines, and that we needed to flush directly using the
virtual address of user and kernel pages. This is likely the major
cause of instability on the c8000 and rp34xx machines.

Flushing user pages requires doing a temporary context switch as we
have to flush pages that don't belong to the current context. Further,
we have to deal with pages that aren't present. If a page isn't
present, the flush instructions fault on every line.

Other code has been rearranged and simplified based on testing. For
example, I introduced a flush_cache_dup_mm routine. flush_cache_mm
and flush_cache_dup_mm differ in that flush_cache_mm calls
purge_cache_pages and flush_cache_dup_mm calls flush_cache_pages.
In some implementations, pdc is more efficient than fdc. Based on
my testing, I don't believe there's any performance benefit on the
c8000.

Signed-off-by: John David Anglin <dave.anglin@bell.net>
Signed-off-by: Helge Deller <deller@gmx.de>
2 years agoparisc: Disable debug code regarding cache flushes in handle_nadtlb_fault()
John David Anglin [Mon, 16 May 2022 15:14:47 +0000 (15:14 +0000)]
parisc: Disable debug code regarding cache flushes in handle_nadtlb_fault()

Change the "BUG" to "WARNING" and disable the message because it triggers
occasionally in spite of the check in flush_cache_page_if_present.

The pte value extracted for the "from" page in copy_user_highpage is racy and
occasionally the pte is cleared before the flush is complete.  I assume that
the page is simultaneously flushed by flush_cache_mm before the pte is cleared
as nullifying the fdc doesn't seem to cause problems.

I investigated various locking scenarios but I wasn't able to find a way to
sequence the flushes.  This code is called for every COW break and locks impact
performance.

This patch is related to the bigger cache flush patch because we need the pte
on PA8800/PA8900 to flush using the vma context.
I have also seen this from copy_to_user_page and copy_from_user_page.

The messages appear infrequently when enabled.

Signed-off-by: John David Anglin <dave.anglin@bell.net>
Signed-off-by: Helge Deller <deller@gmx.de>
2 years agoclk: at91: generated: consider range when calculating best rate
Codrin Ciubotariu [Wed, 13 Apr 2022 07:13:18 +0000 (10:13 +0300)]
clk: at91: generated: consider range when calculating best rate

clk_generated_best_diff() helps in finding the parent and the divisor to
compute a rate closest to the required one. However, it doesn't take into
account the request's range for the new rate. Make sure the new rate
is within the required range.

Fixes: 8a8f4bf0c480 ("clk: at91: clk-generated: create function to find best_diff")
Signed-off-by: Codrin Ciubotariu <codrin.ciubotariu@microchip.com>
Link: https://lore.kernel.org/r/20220413071318.244912-1-codrin.ciubotariu@microchip.com
Reviewed-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
2 years agoaudit,io_uring,io-wq: call __audit_uring_exit for dummy contexts
Julian Orth [Tue, 17 May 2022 10:32:53 +0000 (12:32 +0200)]
audit,io_uring,io-wq: call __audit_uring_exit for dummy contexts

Not calling the function for dummy contexts will cause the context to
not be reset. During the next syscall, this will cause an error in
__audit_syscall_entry:

WARN_ON(context->context != AUDIT_CTX_UNUSED);
WARN_ON(context->name_count);
if (context->context != AUDIT_CTX_UNUSED || context->name_count) {
audit_panic("unrecoverable error in audit_syscall_entry()");
return;
}

These problematic dummy contexts are created via the following call
chain:

       exit_to_user_mode_prepare
    -> arch_do_signal_or_restart
    -> get_signal
    -> task_work_run
    -> tctx_task_work
    -> io_req_task_submit
    -> io_issue_sqe
    -> audit_uring_entry

Cc: stable@vger.kernel.org
Fixes: 5bd2182d58e9 ("audit,io_uring,io-wq: add some basic audit support to io_uring")
Signed-off-by: Julian Orth <ju.orth@gmail.com>
[PM: subject line tweaks]
Signed-off-by: Paul Moore <paul@paul-moore.com>
2 years agoio_uring: don't attempt to IOPOLL for MSG_RING requests
Jens Axboe [Tue, 17 May 2022 18:32:05 +0000 (12:32 -0600)]
io_uring: don't attempt to IOPOLL for MSG_RING requests

We gate whether to IOPOLL for a request on whether the opcode is allowed
on a ring setup for IOPOLL and if it's got a file assigned. MSG_RING
is the only one that allows a file yet isn't pollable, it's merely
supported to allow communication on an IOPOLL ring, not because we can
poll for completion of it.

Put the assigned file early and clear it, so we don't attempt to poll
for it.

Reported-by: syzbot+1a0a53300ce782f8b3ad@syzkaller.appspotmail.com
Fixes: 3f1d52abf098 ("io_uring: defer msg-ring file validity check until command issue")
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agoice: Fix interrupt moderation settings getting cleared
Michal Wilczynski [Sun, 8 May 2022 23:33:48 +0000 (19:33 -0400)]
ice: Fix interrupt moderation settings getting cleared

Adaptive-rx and Adaptive-tx are interrupt moderation settings
that can be enabled/disabled using ethtool:
ethtool -C ethX adaptive-rx on/off adaptive-tx on/off

Unfortunately those settings are getting cleared after
changing number of queues, or in ethtool world 'channels':
ethtool -L ethX rx 1 tx 1

Clearing was happening due to introduction of bit fields
in ice_ring_container struct. This way only itr_setting
bits were rebuilt during ice_vsi_rebuild_set_coalesce().

Introduce an anonymous struct of bitfields and create a
union to refer to them as a single variable.
This way variable can be easily saved and restored.

Fixes: 61dc79ced7aa ("ice: Restore interrupt throttle settings after VSI rebuild")
Signed-off-by: Michal Wilczynski <michal.wilczynski@intel.com>
Tested-by: Gurucharan <gurucharanx.g@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
2 years agoice: fix possible under reporting of ethtool Tx and Rx statistics
Paul Greenwalt [Thu, 28 Apr 2022 21:11:42 +0000 (14:11 -0700)]
ice: fix possible under reporting of ethtool Tx and Rx statistics

The hardware statistics counters are not cleared during resets so the
drivers first access is to initialize the baseline and then subsequent
reads are for reporting the counters. The statistics counters are read
during the watchdog subtask when the interface is up. If the baseline
is not initialized before the interface is up, then there can be a brief
window in which some traffic can be transmitted/received before the
initial baseline reading takes place.

Directly initialize ethtool statistics in driver open so the baseline will
be initialized when the interface is up, and any dropped packets
incremented before the interface is up won't be reported.

Fixes: 28dc1b86f8ea9 ("ice: ignore dropped packets during init")
Signed-off-by: Paul Greenwalt <paul.greenwalt@intel.com>
Tested-by: Gurucharan <gurucharanx.g@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
2 years agoice: fix crash when writing timestamp on RX rings
Arkadiusz Kubalewski [Thu, 28 Apr 2022 08:33:50 +0000 (10:33 +0200)]
ice: fix crash when writing timestamp on RX rings

Do not allow to write timestamps on RX rings if PF is being configured.
When PF is being configured RX rings can be freed or rebuilt. If at the
same time timestamps are updated, the kernel will crash by dereferencing
null RX ring pointer.

PID: 1449   TASK: ff187d28ed658040  CPU: 34  COMMAND: "ice-ptp-0000:51"
 #0 [ff1966a94a713bb0] machine_kexec at ffffffff9d05a0be
 #1 [ff1966a94a713c08] __crash_kexec at ffffffff9d192e9d
 #2 [ff1966a94a713cd0] crash_kexec at ffffffff9d1941bd
 #3 [ff1966a94a713ce8] oops_end at ffffffff9d01bd54
 #4 [ff1966a94a713d08] no_context at ffffffff9d06bda4
 #5 [ff1966a94a713d60] __bad_area_nosemaphore at ffffffff9d06c10c
 #6 [ff1966a94a713da8] do_page_fault at ffffffff9d06cae4
 #7 [ff1966a94a713de0] page_fault at ffffffff9da0107e
    [exception RIP: ice_ptp_update_cached_phctime+91]
    RIP: ffffffffc076db8b  RSP: ff1966a94a713e98  RFLAGS: 00010246
    RAX: 16e3db9c6b7ccae4  RBX: ff187d269dd3c180  RCX: ff187d269cd4d018
    RDX: 0000000000000000  RSI: 0000000000000000  RDI: 0000000000000000
    RBP: ff187d269cfcc644   R8: ff187d339b9641b0   R9: 0000000000000000
    R10: 0000000000000002  R11: 0000000000000000  R12: ff187d269cfcc648
    R13: ffffffff9f128784  R14: ffffffff9d101b70  R15: ff187d269cfcc640
    ORIG_RAX: ffffffffffffffff  CS: 0010  SS: 0018
 #8 [ff1966a94a713ea0] ice_ptp_periodic_work at ffffffffc076dbef [ice]
 #9 [ff1966a94a713ee0] kthread_worker_fn at ffffffff9d101c1b
 #10 [ff1966a94a713f10] kthread at ffffffff9d101b4d
 #11 [ff1966a94a713f50] ret_from_fork at ffffffff9da0023f

Fixes: 77a781155a65 ("ice: enable receive hardware timestamping")
Signed-off-by: Arkadiusz Kubalewski <arkadiusz.kubalewski@intel.com>
Reviewed-by: Michal Schmidt <mschmidt@redhat.com>
Tested-by: Dave Cain <dcain@redhat.com>
Tested-by: Gurucharan <gurucharanx.g@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
2 years agoMerge tag 'kvmarm-fixes-5.18-3' of git://git.kernel.org/pub/scm/linux/kernel/git...
Paolo Bonzini [Tue, 17 May 2022 17:26:33 +0000 (13:26 -0400)]
Merge tag 'kvmarm-fixes-5.18-3' of git://git./linux/kernel/git/kvmarm/kvmarm into HEAD

KVM/arm64 fixes for 5.18, take #3

- Correctly expose GICv3 support even if no irqchip is created
  so that userspace doesn't observe it changing pointlessly
  (fixing a regression with QEMU)

- Don't issue a hypercall to set the id-mapped vectors when
  protected mode is enabled (fix for pKVM in combination with
  CPUs affected by Spectre-v3a)

2 years agoEDAC/i5100: Remove unused inline function i5100_nrecmema_dm_buf_id()
YueHaibing [Sat, 14 May 2022 08:04:33 +0000 (16:04 +0800)]
EDAC/i5100: Remove unused inline function i5100_nrecmema_dm_buf_id()

Commit

  a4972b1b9a04 ("edac: i5100_edac: Remove unused i5100_recmema_dm_buf_id")

left this function unused. Remove it.

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Link: https://lore.kernel.org/r/20220514080433.29944-1-yuehaibing@huawei.com
2 years agoarm64: mte: Ensure the cleared tags are visible before setting the PTE
Catalin Marinas [Tue, 17 May 2022 09:35:32 +0000 (10:35 +0100)]
arm64: mte: Ensure the cleared tags are visible before setting the PTE

As an optimisation, only pages mapped with PROT_MTE in user space have
the MTE tags zeroed. This is done lazily at the set_pte_at() time via
mte_sync_tags(). However, this function is missing a barrier and another
CPU may see the PTE updated before the zeroed tags are visible. Add an
smp_wmb() barrier if the mapping is Normal Tagged.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Fixes: 34bfeea4a9e9 ("arm64: mte: Clear the tags when a page is mapped in user-space with PROT_MTE")
Cc: <stable@vger.kernel.org> # 5.10.x
Reported-by: Vladimir Murzin <vladimir.murzin@arm.com>
Cc: Will Deacon <will@kernel.org>
Reviewed-by: Steven Price <steven.price@arm.com>
Tested-by: Vladimir Murzin <vladimir.murzin@arm.com>
Link: https://lore.kernel.org/r/20220517093532.127095-1-catalin.marinas@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
2 years agoarm64: kexec: load from kimage prior to clobbering
Mark Rutland [Mon, 16 May 2022 16:07:35 +0000 (17:07 +0100)]
arm64: kexec: load from kimage prior to clobbering

In arm64_relocate_new_kernel() we load some fields out of the kimage
structure after relocation has occurred. As the kimage structure isn't
allocated to be relocation-safe, it may be clobbered during relocation,
and we may load junk values out of the structure.

Due to this, kexec may fail when the kimage allocation happens to fall
within a PA range that an object will be relocated to. This has been
observed to occur for regular kexec on a QEMU TCG 'virt' machine with
2GiB of RAM, where the PA range of the new kernel image overlaps the
kimage structure.

Avoid this by ensuring we load all values from the kimage structure
prior to relocation.

I've tested this atop v5.16 and v5.18-rc6.

Fixes: 878fdbd70486 ("arm64: kexec: pass kimage as the only argument to relocation function")
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: James Morse <james.morse@arm.com>
Cc: Pasha Tatashin <pasha.tatashin@soleen.com>
Cc: Will Deacon <will@kernel.org>
Reviewed-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Link: https://lore.kernel.org/r/20220516160735.731404-1-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
2 years agoarm64: paravirt: Use RCU read locks to guard stolen_time
Prakruthi Deepak Heragu [Fri, 13 May 2022 17:46:54 +0000 (10:46 -0700)]
arm64: paravirt: Use RCU read locks to guard stolen_time

During hotplug, the stolen time data structure is unmapped and memset.
There is a possibility of the timer IRQ being triggered before memset
and stolen time is getting updated as part of this timer IRQ handler. This
causes the below crash in timer handler -

  [ 3457.473139][    C5] Unable to handle kernel paging request at virtual address ffffffc03df05148
  ...
  [ 3458.154398][    C5] Call trace:
  [ 3458.157648][    C5]  para_steal_clock+0x30/0x50
  [ 3458.162319][    C5]  irqtime_account_process_tick+0x30/0x194
  [ 3458.168148][    C5]  account_process_tick+0x3c/0x280
  [ 3458.173274][    C5]  update_process_times+0x5c/0xf4
  [ 3458.178311][    C5]  tick_sched_timer+0x180/0x384
  [ 3458.183164][    C5]  __run_hrtimer+0x160/0x57c
  [ 3458.187744][    C5]  hrtimer_interrupt+0x258/0x684
  [ 3458.192698][    C5]  arch_timer_handler_virt+0x5c/0xa0
  [ 3458.198002][    C5]  handle_percpu_devid_irq+0xdc/0x414
  [ 3458.203385][    C5]  handle_domain_irq+0xa8/0x168
  [ 3458.208241][    C5]  gic_handle_irq.34493+0x54/0x244
  [ 3458.213359][    C5]  call_on_irq_stack+0x40/0x70
  [ 3458.218125][    C5]  do_interrupt_handler+0x60/0x9c
  [ 3458.223156][    C5]  el1_interrupt+0x34/0x64
  [ 3458.227560][    C5]  el1h_64_irq_handler+0x1c/0x2c
  [ 3458.232503][    C5]  el1h_64_irq+0x7c/0x80
  [ 3458.236736][    C5]  free_vmap_area_noflush+0x108/0x39c
  [ 3458.242126][    C5]  remove_vm_area+0xbc/0x118
  [ 3458.246714][    C5]  vm_remove_mappings+0x48/0x2a4
  [ 3458.251656][    C5]  __vunmap+0x154/0x278
  [ 3458.255796][    C5]  stolen_time_cpu_down_prepare+0xc0/0xd8
  [ 3458.261542][    C5]  cpuhp_invoke_callback+0x248/0xc34
  [ 3458.266842][    C5]  cpuhp_thread_fun+0x1c4/0x248
  [ 3458.271696][    C5]  smpboot_thread_fn+0x1b0/0x400
  [ 3458.276638][    C5]  kthread+0x17c/0x1e0
  [ 3458.280691][    C5]  ret_from_fork+0x10/0x20

As a fix, introduce rcu lock to update stolen time structure.

Fixes: 75df529bec91 ("arm64: paravirt: Initialize steal time when cpu is online")
Cc: stable@vger.kernel.org
Suggested-by: Will Deacon <will@kernel.org>
Signed-off-by: Prakruthi Deepak Heragu <quic_pheragu@quicinc.com>
Signed-off-by: Elliot Berman <quic_eberman@quicinc.com>
Reviewed-by: Srivatsa S. Bhat (VMware) <srivatsa@csail.mit.edu>
Link: https://lore.kernel.org/r/20220513174654.362169-1-quic_eberman@quicinc.com
Signed-off-by: Will Deacon <will@kernel.org>
2 years agoblk-cgroup: Remove unnecessary rcu_read_lock/unlock()
Fanjun Kong [Mon, 16 May 2022 17:39:30 +0000 (01:39 +0800)]
blk-cgroup: Remove unnecessary rcu_read_lock/unlock()

spin_lock_irq/spin_unlock_irq contains preempt_disable/enable().
Which can serve as RCU read-side critical region, so remove
rcu_read_lock/unlock().

Signed-off-by: Fanjun Kong <bh1scw@gmail.com>
Reviewed-by: Muchun Song <songmuchun@bytedance.com>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20220516173930.159535-1-bh1scw@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>