]> git.proxmox.com Git - mirror_ubuntu-kernels.git/log
mirror_ubuntu-kernels.git
2 years agohugetlb: fix huge_pmd_unshare address update
Mike Kravetz [Tue, 24 May 2022 20:50:03 +0000 (13:50 -0700)]
hugetlb: fix huge_pmd_unshare address update

The routine huge_pmd_unshare() is passed a pointer to an address
associated with an area which may be unshared.  If unshare is successful
this address is updated to 'optimize' callers iterating over huge page
addresses.  For the optimization to work correctly, address should be
updated to the last huge page in the unmapped/unshared area.  However, in
the common case where the passed address is PUD_SIZE aligned, the address
is incorrectly updated to the address of the preceding huge page.  That
wastes CPU cycles as the unmapped/unshared range is scanned twice.

Link: https://lkml.kernel.org/r/20220524205003.126184-1-mike.kravetz@oracle.com
Fixes: 39dde65c9940 ("shared page table for hugetlb page")
Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com>
Acked-by: Muchun Song <songmuchun@bytedance.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 years agozsmalloc: fix races between asynchronous zspage free and page migration
Sultan Alsawaf [Fri, 13 May 2022 22:11:26 +0000 (15:11 -0700)]
zsmalloc: fix races between asynchronous zspage free and page migration

The asynchronous zspage free worker tries to lock a zspage's entire page
list without defending against page migration.  Since pages which haven't
yet been locked can concurrently migrate off the zspage page list while
lock_zspage() churns away, lock_zspage() can suffer from a few different
lethal races.

It can lock a page which no longer belongs to the zspage and unsafely
dereference page_private(), it can unsafely dereference a torn pointer to
the next page (since there's a data race), and it can observe a spurious
NULL pointer to the next page and thus not lock all of the zspage's pages
(since a single page migration will reconstruct the entire page list, and
create_page_chain() unconditionally zeroes out each list pointer in the
process).

Fix the races by using migrate_read_lock() in lock_zspage() to synchronize
with page migration.

Link: https://lkml.kernel.org/r/20220509024703.243847-1-sultan@kerneltoast.com
Fixes: 77ff465799c602 ("zsmalloc: zs_page_migrate: skip unnecessary loops but not return -EBUSY if zspage is not inuse")
Signed-off-by: Sultan Alsawaf <sultan@kerneltoast.com>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Nitin Gupta <ngupta@vflare.org>
Cc: Sergey Senozhatsky <senozhatsky@chromium.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 years agoRevert "mm/cma.c: remove redundant cma_mutex lock"
Dong Aisheng [Fri, 13 May 2022 22:11:26 +0000 (15:11 -0700)]
Revert "mm/cma.c: remove redundant cma_mutex lock"

This reverts commit a4efc174b382fcdb which introduced a regression issue
that when there're multiple processes allocating dma memory in parallel by
calling dma_alloc_coherent(), it may fail sometimes as follows:

Error log:
cma: cma_alloc: linux,cma: alloc failed, req-size: 148 pages, ret: -16
cma: number of available pages:
3@125+20@172+12@236+4@380+32@736+17@2287+23@2473+20@36076+99@40477+108@40852+44@41108+20@41196+108@41364+108@41620+
108@42900+108@43156+483@44061+1763@45341+1440@47712+20@49324+20@49388+5076@49452+2304@55040+35@58141+20@58220+20@58284+
7188@58348+84@66220+7276@66452+227@74525+6371@75549=> 33161 free of 81920 total pages

When issue happened, we saw there were still 33161 pages (129M) free CMA
memory and a lot available free slots for 148 pages in CMA bitmap that we
want to allocate.

When dumping memory info, we found that there was also ~342M normal
memory, but only 1352K CMA memory left in buddy system while a lot of
pageblocks were isolated.

Memory info log:
Normal free:351096kB min:30000kB low:37500kB high:45000kB reserved_highatomic:0KB
    active_anon:98060kB inactive_anon:98948kB active_file:60864kB inactive_file:31776kB
    unevictable:0kB writepending:0kB present:1048576kB managed:1018328kB mlocked:0kB
    bounce:0kB free_pcp:220kB local_pcp:192kB free_cma:1352kB lowmem_reserve[]: 0 0 0
Normal: 78*4kB (UECI) 1772*8kB (UMECI) 1335*16kB (UMECI) 360*32kB (UMECI) 65*64kB (UMCI)
36*128kB (UMECI) 16*256kB (UMCI) 6*512kB (EI) 8*1024kB (UEI) 4*2048kB (MI) 8*4096kB (EI)
8*8192kB (UI) 3*16384kB (EI) 8*32768kB (M) = 489288kB

The root cause of this issue is that since commit a4efc174b382 ("mm/cma.c:
remove redundant cma_mutex lock"), CMA supports concurrent memory
allocation.  It's possible that the memory range process A trying to alloc
has already been isolated by the allocation of process B during memory
migration.

The problem here is that the memory range isolated during one allocation
by start_isolate_page_range() could be much bigger than the real size we
want to alloc due to the range is aligned to MAX_ORDER_NR_PAGES.

Taking an ARMv7 platform with 1G memory as an example, when
MAX_ORDER_NR_PAGES is big (e.g.  32M with max_order 14) and CMA memory is
relatively small (e.g.  128M), there're only 4 MAX_ORDER slot, then it's
very easy that all CMA memory may have already been isolated by other
processes when one trying to allocate memory using dma_alloc_coherent().
Since current CMA code will only scan one time of whole available CMA
memory, then dma_alloc_coherent() may easy fail due to contention with
other processes.

This patch simply falls back to the original method that using cma_mutex
to make alloc_contig_range() run sequentially to avoid the issue.

Link: https://lkml.kernel.org/r/20220509094551.3596244-1-aisheng.dong@nxp.com
Link: https://lore.kernel.org/all/20220315144521.3810298-2-aisheng.dong@nxp.com/
Fixes: a4efc174b382 ("mm/cma.c: remove redundant cma_mutex lock")
Signed-off-by: Dong Aisheng <aisheng.dong@nxp.com>
Acked-by: Minchan Kim <minchan@kernel.org>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Marek Szyprowski <m.szyprowski@samsung.com>
Cc: Lecopzer Chen <lecopzer.chen@mediatek.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: <stable@vger.kernel.org> [5.11+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 years agoMAINTAINERS: add a mailing list for DAMON development
SeongJae Park [Tue, 10 May 2022 00:34:29 +0000 (17:34 -0700)]
MAINTAINERS: add a mailing list for DAMON development

This commit adds an open mailing list for DAMON in MAINTAINERS file.

Link: https://lkml.kernel.org/r/20220503180741.137079-1-sj@kernel.org
Signed-off-by: SeongJae Park <sj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 years agoselftests: vm: Makefile: rename TARGETS to VMTARGETS
Joel Savitz [Tue, 10 May 2022 00:34:29 +0000 (17:34 -0700)]
selftests: vm: Makefile: rename TARGETS to VMTARGETS

The tools/testing/selftests/vm/Makefile uses the variable TARGETS
internally to generate a list of platform-specific binary build targets
suffixed with _{32,64}.  When building the selftests using its own
Makefile directly, such as via the following command run in a kernel tree:

One receives an error such as the following:

make: Entering directory '/root/linux/tools/testing/selftests'
make --no-builtin-rules ARCH=x86 -C ../../.. headers_install
make[1]: Entering directory '/root/linux'
  INSTALL ./usr/include
make[1]: Leaving directory '/root/linux'
make[1]: Entering directory '/root/linux/tools/testing/selftests/vm'
make[1]: *** No rule to make target 'vm.c', needed by '/root/linux/tools/testing/selftests/vm/vm_64'.  Stop.
make[1]: Leaving directory '/root/linux/tools/testing/selftests/vm'
make: *** [Makefile:175: all] Error 2
make: Leaving directory '/root/linux/tools/testing/selftests'

The TARGETS variable passed to tools/testing/selftests/Makefile collides
with the TARGETS used in tools/testing/selftests/vm/Makefile, so rename
the latter to VMTARGETS, eliminating the collision with no functional
change.

Link: https://lkml.kernel.org/r/20220504213454.1282532-1-jsavitz@redhat.com
Fixes: f21fda8f6453 ("selftests: vm: pkeys: fix multilib builds for x86")
Signed-off-by: Joel Savitz <jsavitz@redhat.com>
Acked-by: Nico Pache <npache@redhat.com>
Cc: Joel Savitz <jsavitz@redhat.com>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Sandipan Das <sandipan@linux.ibm.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 years agomm/kfence: reset PG_slab and memcg_data before freeing __kfence_pool
Hyeonggon Yoo [Tue, 10 May 2022 00:34:29 +0000 (17:34 -0700)]
mm/kfence: reset PG_slab and memcg_data before freeing __kfence_pool

When kfence fails to initialize kfence pool, it frees the pool.  But it
does not reset memcg_data and PG_slab flag.

Below is a BUG because of this. Let's fix it by resetting memcg_data
and PG_slab flag before free.

[    0.089149] BUG: Bad page state in process swapper/0  pfn:3d8e06
[    0.089149] page:ffffea46cf638180 refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x3d8e06
[    0.089150] memcg:ffffffff94a475d1
[    0.089150] flags: 0x17ffffc0000200(slab|node=0|zone=2|lastcpupid=0x1fffff)
[    0.089151] raw: 0017ffffc0000200 ffffea46cf638188 ffffea46cf638188 0000000000000000
[    0.089152] raw: 0000000000000000 0000000000000000 00000000ffffffff ffffffff94a475d1
[    0.089152] page dumped because: page still charged to cgroup
[    0.089153] Modules linked in:
[    0.089153] CPU: 0 PID: 0 Comm: swapper/0 Tainted: G    B   W         5.18.0-rc1+ #965
[    0.089154] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-2 04/01/2014
[    0.089154] Call Trace:
[    0.089155]  <TASK>
[    0.089155]  dump_stack_lvl+0x49/0x5f
[    0.089157]  dump_stack+0x10/0x12
[    0.089158]  bad_page.cold+0x63/0x94
[    0.089159]  check_free_page_bad+0x66/0x70
[    0.089160]  __free_pages_ok+0x423/0x530
[    0.089161]  __free_pages_core+0x8e/0xa0
[    0.089162]  memblock_free_pages+0x10/0x12
[    0.089164]  memblock_free_late+0x8f/0xb9
[    0.089165]  kfence_init+0x68/0x92
[    0.089166]  start_kernel+0x789/0x992
[    0.089167]  x86_64_start_reservations+0x24/0x26
[    0.089168]  x86_64_start_kernel+0xa9/0xaf
[    0.089170]  secondary_startup_64_no_verify+0xd5/0xdb
[    0.089171]  </TASK>

Link: https://lkml.kernel.org/r/YnPG3pQrqfcgOlVa@hyeyoo
Fixes: 0ce20dd84089 ("mm: add Kernel Electric-Fence infrastructure")
Fixes: 8f0b36497303 ("mm: kfence: fix objcgs vector allocation")
Signed-off-by: Hyeonggon Yoo <42.hyeyoo@gmail.com>
Reviewed-by: Marco Elver <elver@google.com>
Reviewed-by: Muchun Song <songmuchun@bytedance.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 years agomailmap: add entry for martyna.szapar-mudlaw@intel.com
Martyna Szapar-Mudlaw [Tue, 10 May 2022 00:34:29 +0000 (17:34 -0700)]
mailmap: add entry for martyna.szapar-mudlaw@intel.com

Separate linux.intel.com account was created for submitting and reviewing
kernel patches, thus need to map previously used primary Intel e-mail
address.

Link: https://lkml.kernel.org/r/20220505132624.41802-1-martyna.szapar-mudlaw@linux.intel.com
Signed-off-by: Martyna Szapar-Mudlaw <martyna.szapar-mudlaw@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 years agoarm[64]/memremap: don't abuse pfn_valid() to ensure presence of linear map
Mike Rapoport [Tue, 10 May 2022 00:34:28 +0000 (17:34 -0700)]
arm[64]/memremap: don't abuse pfn_valid() to ensure presence of linear map

The semantics of pfn_valid() is to check presence of the memory map for a
PFN and not whether a PFN is covered by the linear map.  The memory map
may be present for NOMAP memory regions, but they won't be mapped in the
linear mapping.  Accessing such regions via __va() when they are
memremap()'ed will cause a crash.

On v5.4.y the crash happens on qemu-arm with UEFI [1]:

<1>[    0.084476] 8<--- cut here ---
<1>[    0.084595] Unable to handle kernel paging request at virtual address dfb76000
<1>[    0.084938] pgd = (ptrval)
<1>[    0.085038] [dfb76000] *pgd=5f7fe801, *pte=00000000, *ppte=00000000

...

<4>[    0.093923] [<c0ed6ce8>] (memcpy) from [<c16a06f8>] (dmi_setup+0x60/0x418)
<4>[    0.094204] [<c16a06f8>] (dmi_setup) from [<c16a38d4>] (arm_dmi_init+0x8/0x10)
<4>[    0.094408] [<c16a38d4>] (arm_dmi_init) from [<c0302e9c>] (do_one_initcall+0x50/0x228)
<4>[    0.094619] [<c0302e9c>] (do_one_initcall) from [<c16011e4>] (kernel_init_freeable+0x15c/0x1f8)
<4>[    0.094841] [<c16011e4>] (kernel_init_freeable) from [<c0f028cc>] (kernel_init+0x8/0x10c)
<4>[    0.095057] [<c0f028cc>] (kernel_init) from [<c03010e8>] (ret_from_fork+0x14/0x2c)

On kernels v5.10.y and newer the same crash won't reproduce on ARM because
commit b10d6bca8720 ("arch, drivers: replace for_each_membock() with
for_each_mem_range()") changed the way memory regions are registered in
the resource tree, but that merely covers up the problem.

On ARM64 memory resources registered in yet another way and there the
issue of wrong usage of pfn_valid() to ensure availability of the linear
map is also covered.

Implement arch_memremap_can_ram_remap() on ARM and ARM64 to prevent access
to NOMAP regions via the linear mapping in memremap().

Link: https://lore.kernel.org/all/Yl65zxGgFzF1Okac@sirena.org.uk
Link: https://lkml.kernel.org/r/20220426060107.7618-1-rppt@kernel.org
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Reported-by: "kernelci.org bot" <bot@kernelci.org>
Tested-by: Mark Brown <broonie@kernel.org>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Mark Brown <broonie@kernel.org>
Cc: Mark-PK Tsai <mark-pk.tsai@mediatek.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Will Deacon <will@kernel.org>
Cc: <stable@vger.kernel.org> [5.4+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 years agoprocfs: prevent unprivileged processes accessing fdinfo dir
Kalesh Singh [Tue, 10 May 2022 00:34:28 +0000 (17:34 -0700)]
procfs: prevent unprivileged processes accessing fdinfo dir

The file permissions on the fdinfo dir from were changed from
S_IRUSR|S_IXUSR to S_IRUGO|S_IXUGO, and a PTRACE_MODE_READ check was added
for opening the fdinfo files [1].  However, the ptrace permission check
was not added to the directory, allowing anyone to get the open FD numbers
by reading the fdinfo directory.

Add the missing ptrace permission check for opening the fdinfo directory.

[1] https://lkml.kernel.org/r/20210308170651.919148-1-kaleshsingh@google.com

Link: https://lkml.kernel.org/r/20210713162008.1056986-1-kaleshsingh@google.com
Fixes: 7bc3fa0172a4 ("procfs: allow reading fdinfo with PTRACE_MODE_READ")
Signed-off-by: Kalesh Singh <kaleshsingh@google.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Eric W. Biederman <ebiederm@xmission.com>
Cc: Christian Brauner <christian.brauner@ubuntu.com>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Hridya Valsaraju <hridya@google.com>
Cc: Jann Horn <jannh@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 years agomm: mremap: fix sign for EFAULT error return value
Niels Dossche [Tue, 10 May 2022 00:34:28 +0000 (17:34 -0700)]
mm: mremap: fix sign for EFAULT error return value

The mremap syscall is supposed to return a pointer to the new virtual
memory area on success, and a negative value of the error code in case of
failure.  Currently, EFAULT is returned when the VMA is not found, instead
of -EFAULT.  The users of this syscall will therefore believe the syscall
succeeded in case the VMA didn't exist, as it returns a pointer to address
0xe (0xe being the value of EFAULT).  Fix the sign of the error value.

Link: https://lkml.kernel.org/r/20220427224439.23828-2-dossche.niels@gmail.com
Fixes: 550a7d60bd5e ("mm, hugepages: add mremap() support for hugepage backed vma")
Signed-off-by: Niels Dossche <dossche.niels@gmail.com>
Cc: Mina Almasry <almasrymina@google.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 years agomm/hwpoison: use pr_err() instead of dump_page() in get_any_page()
Naoya Horiguchi [Fri, 29 Apr 2022 06:14:44 +0000 (23:14 -0700)]
mm/hwpoison: use pr_err() instead of dump_page() in get_any_page()

The following VM_BUG_ON_FOLIO() is triggered when memory error event
happens on the (thp/folio) pages which are about to be freed:

  [ 1160.232771] page:00000000b36a8a0f refcount:1 mapcount:0 mapping:0000000000000000 index:0x1 pfn:0x16a000
  [ 1160.236916] page:00000000b36a8a0f refcount:0 mapcount:0 mapping:0000000000000000 index:0x1 pfn:0x16a000
  [ 1160.240684] flags: 0x57ffffc0800000(hwpoison|node=1|zone=2|lastcpupid=0x1fffff)
  [ 1160.243458] raw: 0057ffffc0800000 dead000000000100 dead000000000122 0000000000000000
  [ 1160.246268] raw: 0000000000000001 0000000000000000 00000000ffffffff 0000000000000000
  [ 1160.249197] page dumped because: VM_BUG_ON_FOLIO(!folio_test_large(folio))
  [ 1160.251815] ------------[ cut here ]------------
  [ 1160.253438] kernel BUG at include/linux/mm.h:788!
  [ 1160.256162] invalid opcode: 0000 [#1] PREEMPT SMP PTI
  [ 1160.258172] CPU: 2 PID: 115368 Comm: mceinj.sh Tainted: G            E     5.18.0-rc1-v5.18-rc1-220404-2353-005-g83111+ #3
  [ 1160.262049] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1.fc35 04/01/2014
  [ 1160.265103] RIP: 0010:dump_page.cold+0x27e/0x2bd
  [ 1160.266757] Code: fe ff ff 48 c7 c6 81 f1 5a 98 e9 4c fe ff ff 48 c7 c6 a1 95 59 98 e9 40 fe ff ff 48 c7 c6 50 bf 5a 98 48 89 ef e8 9d 04 6d ff <0f> 0b 41 f7 c4 ff 0f 00 00 0f 85 9f fd ff ff 49 8b 04 24 a9 00 00
  [ 1160.273180] RSP: 0018:ffffaa2c4d59fd18 EFLAGS: 00010292
  [ 1160.274969] RAX: 000000000000003e RBX: 0000000000000001 RCX: 0000000000000000
  [ 1160.277263] RDX: 0000000000000001 RSI: ffffffff985995a1 RDI: 00000000ffffffff
  [ 1160.279571] RBP: ffffdc9c45a80000 R08: 0000000000000000 R09: 00000000ffffdfff
  [ 1160.281794] R10: ffffaa2c4d59fb08 R11: ffffffff98940d08 R12: ffffdc9c45a80000
  [ 1160.283920] R13: ffffffff985b6f94 R14: 0000000000000000 R15: ffffdc9c45a80000
  [ 1160.286641] FS:  00007eff54ce1740(0000) GS:ffff99c67bd00000(0000) knlGS:0000000000000000
  [ 1160.289498] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  [ 1160.291106] CR2: 00005628381a5f68 CR3: 0000000104712003 CR4: 0000000000170ee0
  [ 1160.293031] Call Trace:
  [ 1160.293724]  <TASK>
  [ 1160.294334]  get_hwpoison_page+0x47d/0x570
  [ 1160.295474]  memory_failure+0x106/0xaa0
  [ 1160.296474]  ? security_capable+0x36/0x50
  [ 1160.297524]  hard_offline_page_store+0x43/0x80
  [ 1160.298684]  kernfs_fop_write_iter+0x11c/0x1b0
  [ 1160.299829]  new_sync_write+0xf9/0x160
  [ 1160.300810]  vfs_write+0x209/0x290
  [ 1160.301835]  ksys_write+0x4f/0xc0
  [ 1160.302718]  do_syscall_64+0x3b/0x90
  [ 1160.303664]  entry_SYSCALL_64_after_hwframe+0x44/0xae
  [ 1160.304981] RIP: 0033:0x7eff54b018b7

As shown in the RIP address, this VM_BUG_ON in folio_entire_mapcount() is
called from dump_page("hwpoison: unhandlable page") in get_any_page().
The below explains the mechanism of the race:

  CPU 0                                       CPU 1

    memory_failure
      get_hwpoison_page
        get_any_page
          dump_page
            compound = PageCompound
                                                free_pages_prepare
                                                  page->flags &= ~PAGE_FLAGS_CHECK_AT_PREP
            folio_entire_mapcount
              VM_BUG_ON_FOLIO(!folio_test_large(folio))

So replace dump_page() with safer one, pr_err().

Link: https://lkml.kernel.org/r/20220427053220.719866-1-naoya.horiguchi@linux.dev
Fixes: 74e8ee4708a8 ("mm: Turn head_compound_mapcount() into folio_entire_mapcount()")
Signed-off-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Reviewed-by: John Hubbard <jhubbard@nvidia.com>
Reviewed-by: Miaohe Lin <linmiaohe@huawei.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Cc: William Kucharski <william.kucharski@oracle.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 years agomm/huge_memory: do not overkill when splitting huge_zero_page
Xu Yu [Fri, 29 Apr 2022 06:14:43 +0000 (23:14 -0700)]
mm/huge_memory: do not overkill when splitting huge_zero_page

Kernel panic when injecting memory_failure for the global huge_zero_page,
when CONFIG_DEBUG_VM is enabled, as follows.

  Injecting memory failure for pfn 0x109ff9 at process virtual address 0x20ff9000
  page:00000000fb053fc3 refcount:2 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x109e00
  head:00000000fb053fc3 order:9 compound_mapcount:0 compound_pincount:0
  flags: 0x17fffc000010001(locked|head|node=0|zone=2|lastcpupid=0x1ffff)
  raw: 017fffc000010001 0000000000000000 dead000000000122 0000000000000000
  raw: 0000000000000000 0000000000000000 00000002ffffffff 0000000000000000
  page dumped because: VM_BUG_ON_PAGE(is_huge_zero_page(head))
  ------------[ cut here ]------------
  kernel BUG at mm/huge_memory.c:2499!
  invalid opcode: 0000 [#1] PREEMPT SMP PTI
  CPU: 6 PID: 553 Comm: split_bug Not tainted 5.18.0-rc1+ #11
  Hardware name: Alibaba Cloud Alibaba Cloud ECS, BIOS 3288b3c 04/01/2014
  RIP: 0010:split_huge_page_to_list+0x66a/0x880
  Code: 84 9b fb ff ff 48 8b 7c 24 08 31 f6 e8 9f 5d 2a 00 b8 b8 02 00 00 e9 e8 fb ff ff 48 c7 c6 e8 47 3c 82 4c b
  RSP: 0018:ffffc90000dcbdf8 EFLAGS: 00010246
  RAX: 000000000000003c RBX: 0000000000000001 RCX: 0000000000000000
  RDX: 0000000000000000 RSI: ffffffff823e4c4f RDI: 00000000ffffffff
  RBP: ffff88843fffdb40 R08: 0000000000000000 R09: 00000000fffeffff
  R10: ffffc90000dcbc48 R11: ffffffff82d68448 R12: ffffea0004278000
  R13: ffffffff823c6203 R14: 0000000000109ff9 R15: ffffea000427fe40
  FS:  00007fc375a26740(0000) GS:ffff88842fd80000(0000) knlGS:0000000000000000
  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  CR2: 00007fc3757c9290 CR3: 0000000102174006 CR4: 00000000003706e0
  DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
  DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
  Call Trace:
  try_to_split_thp_page+0x3a/0x130
  memory_failure+0x128/0x800
  madvise_inject_error.cold+0x8b/0xa1
  __x64_sys_madvise+0x54/0x60
  do_syscall_64+0x35/0x80
  entry_SYSCALL_64_after_hwframe+0x44/0xae
  RIP: 0033:0x7fc3754f8bf9
  Code: 01 00 48 81 c4 80 00 00 00 e9 f1 fe ff ff 0f 1f 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 8
  RSP: 002b:00007ffeda93a1d8 EFLAGS: 00000217 ORIG_RAX: 000000000000001c
  RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007fc3754f8bf9
  RDX: 0000000000000064 RSI: 0000000000003000 RDI: 0000000020ff9000
  RBP: 00007ffeda93a200 R08: 0000000000000000 R09: 0000000000000000
  R10: 00000000ffffffff R11: 0000000000000217 R12: 0000000000400490
  R13: 00007ffeda93a2e0 R14: 0000000000000000 R15: 0000000000000000

We think that raising BUG is overkilling for splitting huge_zero_page, the
huge_zero_page can't be met from normal paths other than memory failure,
but memory failure is a valid caller.  So we tend to replace the BUG to
WARN + returning -EBUSY, and thus the panic above won't happen again.

Link: https://lkml.kernel.org/r/f35f8b97377d5d3ede1bc5ac3114da888c57cbce.1651052574.git.xuyu@linux.alibaba.com
Fixes: d173d5417fb6 ("mm/memory-failure.c: skip huge_zero_page in memory_failure()")
Fixes: 6a46079cf57a ("HWPOISON: The high level memory error handler in the VM v7")
Signed-off-by: Xu Yu <xuyu@linux.alibaba.com>
Suggested-by: Yang Shi <shy828301@gmail.com>
Reported-by: kernel test robot <lkp@intel.com>
Reviewed-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Reviewed-by: Yang Shi <shy828301@gmail.com>
Reviewed-by: Miaohe Lin <linmiaohe@huawei.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 years agoRevert "mm/memory-failure.c: skip huge_zero_page in memory_failure()"
Xu Yu [Fri, 29 Apr 2022 06:14:43 +0000 (23:14 -0700)]
Revert "mm/memory-failure.c: skip huge_zero_page in memory_failure()"

Patch series "mm/memory-failure: rework fix on huge_zero_page splitting".

This patch (of 2):

This reverts commit d173d5417fb67411e623d394aab986d847e47dad.

The commit d173d5417fb6 ("mm/memory-failure.c: skip huge_zero_page in
memory_failure()") explicitly skips huge_zero_page in memory_failure(), in
order to avoid triggering VM_BUG_ON_PAGE on huge_zero_page in
split_huge_page_to_list().

This works, but Yang Shi thinks that,

    Raising BUG is overkilling for splitting huge_zero_page. The
    huge_zero_page can't be met from normal paths other than memory
    failure, but memory failure is a valid caller. So I tend to replace
    the BUG to WARN + returning -EBUSY. If we don't care about the
    reason code in memory failure, we don't have to touch memory
    failure.

And for the issue that huge_zero_page will be set PG_has_hwpoisoned,
Yang Shi comments that,

    The anonymous page fault doesn't check if the page is poisoned or
    not since it typically gets a fresh allocated page and assumes the
    poisoned page (isolated successfully) can't be reallocated again.
    But huge zero page and base zero page are reused every time. So no
    matter what fix we pick, the issue is always there.

Finally, Yang, David, Anshuman and Naoya all agree to fix the bug, i.e.,
to split huge_zero_page, in split_huge_page_to_list().

This reverts the commit d173d5417fb6 ("mm/memory-failure.c: skip
huge_zero_page in memory_failure()"), and the original bug will be fixed
by the next patch.

Link: https://lkml.kernel.org/r/872cefb182ba1dd686b0e7db1e6b2ebe5a4fff87.1651039624.git.xuyu@linux.alibaba.com
Fixes: d173d5417fb6 ("mm/memory-failure.c: skip huge_zero_page in memory_failure()")
Fixes: 6a46079cf57a ("HWPOISON: The high level memory error handler in the VM v7")
Signed-off-by: Xu Yu <xuyu@linux.alibaba.com>
Suggested-by: Yang Shi <shy828301@gmail.com>
Reviewed-by: Yang Shi <shy828301@gmail.com>
Reviewed-by: Miaohe Lin <linmiaohe@huawei.com>
Cc: Naoya Horiguchi <naoya.horiguchi@nec.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2 years agoMerge tag 'drm-fixes-2022-04-29' of git://anongit.freedesktop.org/drm/drm
Linus Torvalds [Fri, 29 Apr 2022 01:00:34 +0000 (18:00 -0700)]
Merge tag 'drm-fixes-2022-04-29' of git://anongit.freedesktop.org/drm/drm

Pull drm fixes from Dave Airlie:
 "Another relatively quiet week, amdgpu leads the way, some i915 display
  fixes, and a single sunxi fix.

  amdgpu:
   - Runtime pm fix
   - DCN memory leak fix in error path
   - SI DPM deadlock fix
   - S0ix fix

  amdkfd:
   - GWS fix
   - GWS support for CRIU

  i915:
   - Fix #5284: Backlight control regression on XMG Core 15 e21
   - Fix black display plane on Acer One AO532h
   - Two smaller display fixes

  sunxi:
   - Single fix removing applying PHYS_OFFSET twice"

* tag 'drm-fixes-2022-04-29' of git://anongit.freedesktop.org/drm/drm:
  drm/amdgpu: keep mmhub clock gating being enabled during s2idle suspend
  drm/amd/pm: fix the deadlock issue observed on SI
  drm/amd/display: Fix memory leak in dcn21_clock_source_create
  drm/amdgpu: don't runtime suspend if there are displays attached (v3)
  drm/amdkfd: CRIU add support for GWS queues
  drm/amdkfd: Fix GWS queue count
  drm/sun4i: Remove obsolete references to PHYS_OFFSET
  drm/i915/fbc: Consult hw.crtc instead of uapi.crtc
  drm/i915: Fix SEL_FETCH_PLANE_*(PIPE_B+) register addresses
  drm/i915: Check EDID for HDR static metadata when choosing blc
  drm/i915: Fix DISP_POS_Y and DISP_HEIGHT defines

2 years agoMerge tag 'amd-drm-fixes-5.18-2022-04-27' of https://gitlab.freedesktop.org/agd5f...
Dave Airlie [Fri, 29 Apr 2022 00:27:04 +0000 (10:27 +1000)]
Merge tag 'amd-drm-fixes-5.18-2022-04-27' of https://gitlab.freedesktop.org/agd5f/linux into drm-fixes

amd-drm-fixes-5.18-2022-04-27:

amdgpu:
- Runtime pm fix
- DCN memory leak fix in error path
- SI DPM deadlock fix
- S0ix fix

amdkfd:
- GWS fix
- GWS support for CRIU

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220428023232.5794-1-alexander.deucher@amd.com
2 years agoMerge tag 'drm-intel-fixes-2022-04-28' of git://anongit.freedesktop.org/drm/drm-intel...
Dave Airlie [Fri, 29 Apr 2022 00:17:46 +0000 (10:17 +1000)]
Merge tag 'drm-intel-fixes-2022-04-28' of git://anongit.freedesktop.org/drm/drm-intel into drm-fixes

- Fix #5284: Backlight control regression on XMG Core 15 e21
- Fix black display plane on Acer One AO532h
- Two smaller display fixes

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/Ymotel5VfZUrJahf@jlahtine-mobl.ger.corp.intel.com
2 years agoMerge tag 'drm-misc-fixes-2022-04-27' of git://anongit.freedesktop.org/drm/drm-misc...
Dave Airlie [Fri, 29 Apr 2022 00:02:04 +0000 (10:02 +1000)]
Merge tag 'drm-misc-fixes-2022-04-27' of git://anongit.freedesktop.org/drm/drm-misc into drm-fixes

drm-misc-fixes for v5.18-rc5:
- Single fix removing applying PHYS_OFFSET twice in sunxi.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/f692bb62-5620-1868-91b7-dffb8d6f9175@linux.intel.com
2 years agoMerge tag 'net-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Linus Torvalds [Thu, 28 Apr 2022 19:34:50 +0000 (12:34 -0700)]
Merge tag 'net-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net

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

  Current release - new code bugs:

   - bridge: switchdev: check br_vlan_group() return value

   - use this_cpu_inc() to increment net->core_stats, fix preempt-rt

  Previous releases - regressions:

   - eth: stmmac: fix write to sgmii_adapter_base

  Previous releases - always broken:

   - netfilter: nf_conntrack_tcp: re-init for syn packets only,
     resolving issues with TCP fastopen

   - tcp: md5: fix incorrect tcp_header_len for incoming connections

   - tcp: fix F-RTO may not work correctly when receiving DSACK

   - tcp: ensure use of most recently sent skb when filling rate samples

   - tcp: fix potential xmit stalls caused by TCP_NOTSENT_LOWAT

   - virtio_net: fix wrong buf address calculation when using xdp

   - xsk: fix forwarding when combining copy mode with busy poll

   - xsk: fix possible crash when multiple sockets are created

   - bpf: lwt: fix crash when using bpf_skb_set_tunnel_key() from
     bpf_xmit lwt hook

   - sctp: null-check asoc strreset_chunk in sctp_generate_reconf_event

   - wireguard: device: check for metadata_dst with skb_valid_dst()

   - netfilter: update ip6_route_me_harder to consider L3 domain

   - gre: make o_seqno start from 0 in native mode

   - gre: switch o_seqno to atomic to prevent races in collect_md mode

  Misc:

   - add Eric Dumazet to networking maintainers

   - dt: dsa: realtek: remove realtek,rtl8367s string

   - netfilter: flowtable: Remove the empty file"

* tag 'net-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (65 commits)
  tcp: fix F-RTO may not work correctly when receiving DSACK
  Revert "ibmvnic: Add ethtool private flag for driver-defined queue limits"
  net: enetc: allow tc-etf offload even with NETIF_F_CSUM_MASK
  ixgbe: ensure IPsec VF<->PF compatibility
  MAINTAINERS: Update BNXT entry with firmware files
  netfilter: nft_socket: only do sk lookups when indev is available
  net: fec: add missing of_node_put() in fec_enet_init_stop_mode()
  bnx2x: fix napi API usage sequence
  tls: Skip tls_append_frag on zero copy size
  Add Eric Dumazet to networking maintainers
  netfilter: conntrack: fix udp offload timeout sysctl
  netfilter: nf_conntrack_tcp: re-init for syn packets only
  net: dsa: lantiq_gswip: Don't set GSWIP_MII_CFG_RMII_CLK
  net: Use this_cpu_inc() to increment net->core_stats
  Bluetooth: hci_sync: Cleanup hci_conn if it cannot be aborted
  Bluetooth: hci_event: Fix creating hci_conn object on error status
  Bluetooth: hci_event: Fix checking for invalid handle on error status
  ice: fix use-after-free when deinitializing mailbox snapshot
  ice: wait 5 s for EMP reset after firmware flash
  ice: Protect vf_state check by cfg_lock in ice_vc_process_vf_msg()
  ...

2 years agoMerge tag 'thermal-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Thu, 28 Apr 2022 18:57:00 +0000 (11:57 -0700)]
Merge tag 'thermal-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull thermal control fixes from Rafael Wysocki:
 "These take back recent chages that started to confuse users and fix up
  an attr.show callback prototype in a driver.

  Specifics:

   - Stop warning about deprecation of the userspace thermal governor
     and cooling device status interface, because there are cases in
     which user space has to drive thermal management with the help of
     them (Daniel Lezcano)

   - Fix attr.show callback prototype in the int340x thermal driver
     (Kees Cook)"

* tag 'thermal-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  thermal/governor: Remove deprecated information
  Revert "thermal/core: Deprecate changing cooling device state from userspace"
  thermal: int340x: Fix attr.show callback prototype

2 years agoMerge tag 'pm-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Linus Torvalds [Thu, 28 Apr 2022 18:50:21 +0000 (11:50 -0700)]
Merge tag 'pm-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull power management fixes from Rafael Wysocki:
 "These fix up recent intel_idle driver changes and fix some ARM cpufreq
  driver issues.

  Specifics:

   - Fix issues with the Qualcomm's cpufreq driver (Dmitry Baryshkov,
     Vladimir Zapolskiy).

   - Fix memory leak with the Sun501 driver (Xiaobing Luo).

   - Make intel_idle enable C1E promotion on all CPUs when C1E is
     preferred to C1 (Artem Bityutskiy).

   - Make C6 optimization on Sapphire Rapids added recently work as
     expected if both C1E and C1 are "preferred" (Artem Bityutskiy)"

* tag 'pm-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  intel_idle: Fix SPR C6 optimization
  intel_idle: Fix the 'preferred_cstates' module parameter
  cpufreq: qcom-cpufreq-hw: Clear dcvs interrupts
  cpufreq: fix memory leak in sun50i_cpufreq_nvmem_probe
  cpufreq: qcom-cpufreq-hw: Fix throttle frequency value on EPSS platforms
  cpufreq: qcom-hw: provide online/offline operations
  cpufreq: qcom-hw: fix the opp entries refcounting
  cpufreq: qcom-hw: fix the race between LMH worker and cpuhp
  cpufreq: qcom-hw: drop affinity hint before freeing the IRQ

2 years agoMerge tag 'acpi-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Thu, 28 Apr 2022 18:37:20 +0000 (11:37 -0700)]
Merge tag 'acpi-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull ACPI fixes from Rafael WysockiL
 "These fix up the ACPI processor driver after a change made during the
  5.16 cycle that inadvertently broke falling back to shallower C-states
  when C3 cannot be used.

  Specifics:

   - Make the ACPI processor driver avoid falling back to C3 type of
     C-states when C3 cannot be requested (Ville Syrjälä)

   - Revert a quirk that is not necessary any more after fixing the
     underlying issue properly (Ville Syrjälä)"

* tag 'acpi-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  Revert "ACPI: processor: idle: fix lockup regression on 32-bit ThinkPad T40"
  ACPI: processor: idle: Avoid falling back to C3 type C-states

2 years agoMerge tag 'platform-drivers-x86-v5.18-3' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 28 Apr 2022 18:13:00 +0000 (11:13 -0700)]
Merge tag 'platform-drivers-x86-v5.18-3' of git://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86

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

   - asus-wmi bug-fixes

   - intel-sdsu bug-fixes

   - build (warning) fixes

   - couple of hw-id additions"

* tag 'platform-drivers-x86-v5.18-3' of git://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86:
  platform/x86/intel: pmc/core: change pmc_lpm_modes to static
  platform/x86/intel/sdsi: Fix bug in multi packet reads
  platform/x86/intel/sdsi: Poll on ready bit for writes
  platform/x86/intel/sdsi: Handle leaky bucket
  platform/x86: intel-uncore-freq: Prevent driver loading in guests
  platform/x86: gigabyte-wmi: added support for B660 GAMING X DDR4 motherboard
  platform/x86: dell-laptop: Add quirk entry for Latitude 7520
  platform/x86: asus-wmi: Fix driver not binding when fan curve control probe fails
  platform/x86: asus-wmi: Potential buffer overflow in asus_wmi_evaluate_method_buf()
  tools/power/x86/intel-speed-select: fix build failure when using -Wl,--as-needed

2 years agoMerge tag 'regulator-fix-v5.18-rc4' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 28 Apr 2022 18:07:49 +0000 (11:07 -0700)]
Merge tag 'regulator-fix-v5.18-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator

Pull regulator fix from Mark Brown:
 "A minor fix for the DT binding documentation of the rt5190a driver"

* tag 'regulator-fix-v5.18-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator:
  regulator: dt-bindings: Revise the rt5190a buck/ldo description

2 years agotcp: fix F-RTO may not work correctly when receiving DSACK
Pengcheng Yang [Tue, 26 Apr 2022 10:03:39 +0000 (18:03 +0800)]
tcp: fix F-RTO may not work correctly when receiving DSACK

Currently DSACK is regarded as a dupack, which may cause
F-RTO to incorrectly enter "loss was real" when receiving
DSACK.

Packetdrill to demonstrate:

// Enable F-RTO and TLP
    0 `sysctl -q net.ipv4.tcp_frto=2`
    0 `sysctl -q net.ipv4.tcp_early_retrans=3`
    0 `sysctl -q net.ipv4.tcp_congestion_control=cubic`

// Establish a connection
   +0 socket(..., SOCK_STREAM, IPPROTO_TCP) = 3
   +0 setsockopt(3, SOL_SOCKET, SO_REUSEADDR, [1], 4) = 0
   +0 bind(3, ..., ...) = 0
   +0 listen(3, 1) = 0

// RTT 10ms, RTO 210ms
  +.1 < S 0:0(0) win 32792 <mss 1000,sackOK,nop,nop,nop,wscale 7>
   +0 > S. 0:0(0) ack 1 <...>
 +.01 < . 1:1(0) ack 1 win 257
   +0 accept(3, ..., ...) = 4

// Send 2 data segments
   +0 write(4, ..., 2000) = 2000
   +0 > P. 1:2001(2000) ack 1

// TLP
+.022 > P. 1001:2001(1000) ack 1

// Continue to send 8 data segments
   +0 write(4, ..., 10000) = 10000
   +0 > P. 2001:10001(8000) ack 1

// RTO
+.188 > . 1:1001(1000) ack 1

// The original data is acked and new data is sent(F-RTO step 2.b)
   +0 < . 1:1(0) ack 2001 win 257
   +0 > P. 10001:12001(2000) ack 1

// D-SACK caused by TLP is regarded as a dupack, this results in
// the incorrect judgment of "loss was real"(F-RTO step 3.a)
+.022 < . 1:1(0) ack 2001 win 257 <sack 1001:2001,nop,nop>

// Never-retransmitted data(3001:4001) are acked and
// expect to switch to open state(F-RTO step 3.b)
   +0 < . 1:1(0) ack 4001 win 257
+0 %{ assert tcpi_ca_state == 0, tcpi_ca_state }%

Fixes: e33099f96d99 ("tcp: implement RFC5682 F-RTO")
Signed-off-by: Pengcheng Yang <yangpc@wangsu.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Tested-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/1650967419-2150-1-git-send-email-yangpc@wangsu.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, 28 Apr 2022 16:55:59 +0000 (09:55 -0700)]
Merge git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf

Pablo Neira Ayuso says:

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

1) Fix incorrect TCP connection tracking window reset for non-syn
   packets, from Florian Westphal.

2) Incorrect dependency on CONFIG_NFT_FLOW_OFFLOAD, from Volodymyr Mytnyk.

3) Fix nft_socket from the output path, from Florian Westphal.

* git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf:
  netfilter: nft_socket: only do sk lookups when indev is available
  netfilter: conntrack: fix udp offload timeout sysctl
  netfilter: nf_conntrack_tcp: re-init for syn packets only
====================

Link: https://lore.kernel.org/r/20220428142109.38726-1-pablo@netfilter.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge tag 'gfs2-v5.18-rc4-fix2' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Thu, 28 Apr 2022 16:50:29 +0000 (09:50 -0700)]
Merge tag 'gfs2-v5.18-rc4-fix2' of git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2

Pull gfs2 fix from Andreas Gruenbacher:

 - No short reads or writes upon glock contention

* tag 'gfs2-v5.18-rc4-fix2' of git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2:
  gfs2: No short reads or writes upon glock contention

2 years agoRevert "ibmvnic: Add ethtool private flag for driver-defined queue limits"
Dany Madden [Wed, 27 Apr 2022 23:51:46 +0000 (18:51 -0500)]
Revert "ibmvnic: Add ethtool private flag for driver-defined queue limits"

This reverts commit 723ad916134784b317b72f3f6cf0f7ba774e5dae

When client requests channel or ring size larger than what the server
can support the server will cap the request to the supported max. So,
the client would not be able to successfully request resources that
exceed the server limit.

Fixes: 723ad9161347 ("ibmvnic: Add ethtool private flag for driver-defined queue limits")
Signed-off-by: Dany Madden <drt@linux.ibm.com>
Link: https://lore.kernel.org/r/20220427235146.23189-1-drt@linux.ibm.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: enetc: allow tc-etf offload even with NETIF_F_CSUM_MASK
Vladimir Oltean [Wed, 27 Apr 2022 20:30:17 +0000 (23:30 +0300)]
net: enetc: allow tc-etf offload even with NETIF_F_CSUM_MASK

The Time-Specified Departure feature is indeed mutually exclusive with
TX IP checksumming in ENETC, but TX checksumming in itself is broken and
was removed from this driver in commit 82728b91f124 ("enetc: Remove Tx
checksumming offload code").

The blamed commit declared NETIF_F_HW_CSUM in dev->features to comply
with software TSO's expectations, and still did the checksumming in
software by calling skb_checksum_help(). So there isn't any restriction
for the Time-Specified Departure feature.

However, enetc_setup_tc_txtime() doesn't understand that, and blindly
looks for NETIF_F_CSUM_MASK.

Instead of checking for things which can literally never happen in the
current code base, just remove the check and let the driver offload
tc-etf qdiscs.

Fixes: acede3c5dad5 ("net: enetc: declare NETIF_F_HW_CSUM and do it in software")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Link: https://lore.kernel.org/r/20220427203017.1291634-1-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoixgbe: ensure IPsec VF<->PF compatibility
Leon Romanovsky [Wed, 27 Apr 2022 17:31:52 +0000 (10:31 -0700)]
ixgbe: ensure IPsec VF<->PF compatibility

The VF driver can forward any IPsec flags and such makes the function
is not extendable and prone to backward/forward incompatibility.

If new software runs on VF, it won't know that PF configured something
completely different as it "knows" only XFRM_OFFLOAD_INBOUND flag.

Fixes: eda0333ac293 ("ixgbe: add VF IPsec management")
Reviewed-by: Raed Salem <raeds@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Reviewed-by: Shannon Nelson <snelson@pensando.io>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Link: https://lore.kernel.org/r/20220427173152.443102-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge tag 'xfs-5.18-fixes-1' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux
Linus Torvalds [Thu, 28 Apr 2022 16:37:56 +0000 (09:37 -0700)]
Merge tag 'xfs-5.18-fixes-1' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux

Pull xfs fixes from Dave Chinner:

 - define buffer bit flags as unsigned to fix gcc-5 + c11 warnings

 - remove redundant XFS fields from MAINTAINERS

 - fix inode buffer locking order regression

* tag 'xfs-5.18-fixes-1' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
  xfs: reorder iunlink remove operation in xfs_ifree
  MAINTAINERS: update IOMAP FILESYSTEM LIBRARY and XFS FILESYSTEM
  xfs: convert buffer flags to unsigned.

2 years agoMAINTAINERS: Update BNXT entry with firmware files
Florian Fainelli [Wed, 27 Apr 2022 16:36:06 +0000 (09:36 -0700)]
MAINTAINERS: Update BNXT entry with firmware files

There appears to be a maintainer gap for BNXT TEE firmware files which
causes some patches to be missed. Update the entry for the BNXT Ethernet
controller with its companion firmware files.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Reviewed-by: Michael Chan <michael.chan@broadcom.com>
Link: https://lore.kernel.org/r/20220427163606.126154-1-f.fainelli@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge branch 'thermal-int340x'
Rafael J. Wysocki [Thu, 28 Apr 2022 14:51:24 +0000 (16:51 +0200)]
Merge branch 'thermal-int340x'

Merge a fix for the attr.show callback prototype in the int340x thermal
driver (Kees Cook).

* thermal-int340x:
  thermal: int340x: Fix attr.show callback prototype

2 years agonetfilter: nft_socket: only do sk lookups when indev is available
Florian Westphal [Thu, 28 Apr 2022 07:39:21 +0000 (09:39 +0200)]
netfilter: nft_socket: only do sk lookups when indev is available

Check if the incoming interface is available and NFT_BREAK
in case neither skb->sk nor input device are set.

Because nf_sk_lookup_slow*() assume packet headers are in the
'in' direction, use in postrouting is not going to yield a meaningful
result.  Same is true for the forward chain, so restrict the use
to prerouting, input and output.

Use in output work if a socket is already attached to the skb.

Fixes: 554ced0a6e29 ("netfilter: nf_tables: add support for native socket matching")
Reported-and-tested-by: Topi Miettinen <toiwoton@gmail.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2 years agoMerge branch 'pm-cpuidle'
Rafael J. Wysocki [Thu, 28 Apr 2022 14:09:50 +0000 (16:09 +0200)]
Merge branch 'pm-cpuidle'

Merge cpuidle fixes for 5.18-rc5:

 - Make intel_idle enable C1E promotion on all CPUs when C1E is
   preferred to C1 (Artem Bityutskiy).

 - Make C6 optimization on Sapphire Rapids added recently work as
   expected if both C1E and C1 are "preferred" (Artem Bityutskiy).

* pm-cpuidle:
  intel_idle: Fix SPR C6 optimization
  intel_idle: Fix the 'preferred_cstates' module parameter

2 years agogfs2: No short reads or writes upon glock contention
Andreas Gruenbacher [Thu, 28 Apr 2022 12:51:33 +0000 (14:51 +0200)]
gfs2: No short reads or writes upon glock contention

Commit 00bfe02f4796 ("gfs2: Fix mmap + page fault deadlocks for buffered
I/O") changed gfs2_file_read_iter() and gfs2_file_buffered_write() to
allow dropping the inode glock while faulting in user buffers.  When the
lock was dropped, a short result was returned to indicate that the
operation was interrupted.

As pointed out by Linus (see the link below), this behavior is broken
and the operations should always re-acquire the inode glock and resume
the operation instead.

Link: https://lore.kernel.org/lkml/CAHk-=whaz-g_nOOoo8RRiWNjnv2R+h6_xk2F1J4TuSRxk1MtLw@mail.gmail.com/
Fixes: 00bfe02f4796 ("gfs2: Fix mmap + page fault deadlocks for buffered I/O")
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
2 years agoMerge tag 'for-net-2022-04-27' of git://git.kernel.org/pub/scm/linux/kernel/git/bluet...
Paolo Abeni [Thu, 28 Apr 2022 08:18:51 +0000 (10:18 +0200)]
Merge tag 'for-net-2022-04-27' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetooth/bluetooth

Luiz Augusto von Dentz says:

====================
bluetooth pull request for net:

 - Fix regression causing some HCI events to be discarded when they
   shouldn't.

* tag 'for-net-2022-04-27' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetooth/bluetooth:
  Bluetooth: hci_sync: Cleanup hci_conn if it cannot be aborted
  Bluetooth: hci_event: Fix creating hci_conn object on error status
  Bluetooth: hci_event: Fix checking for invalid handle on error status
====================

Link: https://lore.kernel.org/r/20220427234031.1257281-1-luiz.dentz@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agonet: fec: add missing of_node_put() in fec_enet_init_stop_mode()
Yang Yingliang [Tue, 26 Apr 2022 12:52:31 +0000 (20:52 +0800)]
net: fec: add missing of_node_put() in fec_enet_init_stop_mode()

Put device node in error path in fec_enet_init_stop_mode().

Fixes: 8a448bf832af ("net: ethernet: fec: move GPR register offset and bit into DT")
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Link: https://lore.kernel.org/r/20220426125231.375688-1-yangyingliang@huawei.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agobnx2x: fix napi API usage sequence
Manish Chopra [Tue, 26 Apr 2022 15:39:13 +0000 (08:39 -0700)]
bnx2x: fix napi API usage sequence

While handling PCI errors (AER flow) driver tries to
disable NAPI [napi_disable()] after NAPI is deleted
[__netif_napi_del()] which causes unexpected system
hang/crash.

System message log shows the following:
=======================================
[ 3222.537510] EEH: Detected PCI bus error on PHB#384-PE#800000 [ 3222.537511] EEH: This PCI device has failed 2 times in the last hour and will be permanently disabled after 5 failures.
[ 3222.537512] EEH: Notify device drivers to shutdown [ 3222.537513] EEH: Beginning: 'error_detected(IO frozen)'
[ 3222.537514] EEH: PE#800000 (PCI 0384:80:00.0): Invoking
bnx2x->error_detected(IO frozen)
[ 3222.537516] bnx2x: [bnx2x_io_error_detected:14236(eth14)]IO error detected [ 3222.537650] EEH: PE#800000 (PCI 0384:80:00.0): bnx2x driver reports:
'need reset'
[ 3222.537651] EEH: PE#800000 (PCI 0384:80:00.1): Invoking
bnx2x->error_detected(IO frozen)
[ 3222.537651] bnx2x: [bnx2x_io_error_detected:14236(eth13)]IO error detected [ 3222.537729] EEH: PE#800000 (PCI 0384:80:00.1): bnx2x driver reports:
'need reset'
[ 3222.537729] EEH: Finished:'error_detected(IO frozen)' with aggregate recovery state:'need reset'
[ 3222.537890] EEH: Collect temporary log [ 3222.583481] EEH: of node=0384:80:00.0 [ 3222.583519] EEH: PCI device/vendor: 168e14e4 [ 3222.583557] EEH: PCI cmd/status register: 00100140 [ 3222.583557] EEH: PCI-E capabilities and status follow:
[ 3222.583744] EEH: PCI-E 00: 00020010 012c8da2 00095d5e 00455c82 [ 3222.583892] EEH: PCI-E 10: 10820000 00000000 00000000 00000000 [ 3222.583893] EEH: PCI-E 20: 00000000 [ 3222.583893] EEH: PCI-E AER capability register set follows:
[ 3222.584079] EEH: PCI-E AER 00: 13c10001 00000000 00000000 00062030 [ 3222.584230] EEH: PCI-E AER 10: 00002000 000031c0 000001e0 00000000 [ 3222.584378] EEH: PCI-E AER 20: 00000000 00000000 00000000 00000000 [ 3222.584416] EEH: PCI-E AER 30: 00000000 00000000 [ 3222.584416] EEH: of node=0384:80:00.1 [ 3222.584454] EEH: PCI device/vendor: 168e14e4 [ 3222.584491] EEH: PCI cmd/status register: 00100140 [ 3222.584492] EEH: PCI-E capabilities and status follow:
[ 3222.584677] EEH: PCI-E 00: 00020010 012c8da2 00095d5e 00455c82 [ 3222.584825] EEH: PCI-E 10: 10820000 00000000 00000000 00000000 [ 3222.584826] EEH: PCI-E 20: 00000000 [ 3222.584826] EEH: PCI-E AER capability register set follows:
[ 3222.585011] EEH: PCI-E AER 00: 13c10001 00000000 00000000 00062030 [ 3222.585160] EEH: PCI-E AER 10: 00002000 000031c0 000001e0 00000000 [ 3222.585309] EEH: PCI-E AER 20: 00000000 00000000 00000000 00000000 [ 3222.585347] EEH: PCI-E AER 30: 00000000 00000000 [ 3222.586872] RTAS: event: 5, Type: Platform Error (224), Severity: 2 [ 3222.586873] EEH: Reset without hotplug activity [ 3224.762767] EEH: Beginning: 'slot_reset'
[ 3224.762770] EEH: PE#800000 (PCI 0384:80:00.0): Invoking
bnx2x->slot_reset()
[ 3224.762771] bnx2x: [bnx2x_io_slot_reset:14271(eth14)]IO slot reset initializing...
[ 3224.762887] bnx2x 0384:80:00.0: enabling device (0140 -> 0142) [ 3224.768157] bnx2x: [bnx2x_io_slot_reset:14287(eth14)]IO slot reset
--> driver unload

Uninterruptible tasks
=====================
crash> ps | grep UN
     213      2  11  c000000004c89e00  UN   0.0       0      0  [eehd]
     215      2   0  c000000004c80000  UN   0.0       0      0
[kworker/0:2]
    2196      1  28  c000000004504f00  UN   0.1   15936  11136  wickedd
    4287      1   9  c00000020d076800  UN   0.0    4032   3008  agetty
    4289      1  20  c00000020d056680  UN   0.0    7232   3840  agetty
   32423      2  26  c00000020038c580  UN   0.0       0      0
[kworker/26:3]
   32871   4241  27  c0000002609ddd00  UN   0.1   18624  11648  sshd
   32920  10130  16  c00000027284a100  UN   0.1   48512  12608  sendmail
   33092  32987   0  c000000205218b00  UN   0.1   48512  12608  sendmail
   33154   4567  16  c000000260e51780  UN   0.1   48832  12864  pickup
   33209   4241  36  c000000270cb6500  UN   0.1   18624  11712  sshd
   33473  33283   0  c000000205211480  UN   0.1   48512  12672  sendmail
   33531   4241  37  c00000023c902780  UN   0.1   18624  11648  sshd

EEH handler hung while bnx2x sleeping and holding RTNL lock
===========================================================
crash> bt 213
PID: 213    TASK: c000000004c89e00  CPU: 11  COMMAND: "eehd"
  #0 [c000000004d477e0] __schedule at c000000000c70808
  #1 [c000000004d478b0] schedule at c000000000c70ee0
  #2 [c000000004d478e0] schedule_timeout at c000000000c76dec
  #3 [c000000004d479c0] msleep at c0000000002120cc
  #4 [c000000004d479f0] napi_disable at c000000000a06448
                                        ^^^^^^^^^^^^^^^^
  #5 [c000000004d47a30] bnx2x_netif_stop at c0080000018dba94 [bnx2x]
  #6 [c000000004d47a60] bnx2x_io_slot_reset at c0080000018a551c [bnx2x]
  #7 [c000000004d47b20] eeh_report_reset at c00000000004c9bc
  #8 [c000000004d47b90] eeh_pe_report at c00000000004d1a8
  #9 [c000000004d47c40] eeh_handle_normal_event at c00000000004da64

And the sleeping source code
============================
crash> dis -ls c000000000a06448
FILE: ../net/core/dev.c
LINE: 6702

   6697  {
   6698          might_sleep();
   6699          set_bit(NAPI_STATE_DISABLE, &n->state);
   6700
   6701          while (test_and_set_bit(NAPI_STATE_SCHED, &n->state))
* 6702                  msleep(1);
   6703          while (test_and_set_bit(NAPI_STATE_NPSVC, &n->state))
   6704                  msleep(1);
   6705
   6706          hrtimer_cancel(&n->timer);
   6707
   6708          clear_bit(NAPI_STATE_DISABLE, &n->state);
   6709  }

EEH calls into bnx2x twice based on the system log above, first through
bnx2x_io_error_detected() and then bnx2x_io_slot_reset(), and executes
the following call chains:

bnx2x_io_error_detected()
  +-> bnx2x_eeh_nic_unload()
       +-> bnx2x_del_all_napi()
            +-> __netif_napi_del()

bnx2x_io_slot_reset()
  +-> bnx2x_netif_stop()
       +-> bnx2x_napi_disable()
            +->napi_disable()

Fix this by correcting the sequence of NAPI APIs usage,
that is delete the NAPI after disabling it.

Fixes: 7fa6f34081f1 ("bnx2x: AER revised")
Reported-by: David Christensen <drc@linux.vnet.ibm.com>
Tested-by: David Christensen <drc@linux.vnet.ibm.com>
Signed-off-by: Manish Chopra <manishc@marvell.com>
Signed-off-by: Ariel Elior <aelior@marvell.com>
Link: https://lore.kernel.org/r/20220426153913.6966-1-manishc@marvell.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agotls: Skip tls_append_frag on zero copy size
Maxim Mikityanskiy [Tue, 26 Apr 2022 15:49:49 +0000 (18:49 +0300)]
tls: Skip tls_append_frag on zero copy size

Calling tls_append_frag when max_open_record_len == record->len might
add an empty fragment to the TLS record if the call happens to be on the
page boundary. Normally tls_append_frag coalesces the zero-sized
fragment to the previous one, but not if it's on page boundary.

If a resync happens then, the mlx5 driver posts dump WQEs in
tx_post_resync_dump, and the empty fragment may become a data segment
with byte_count == 0, which will confuse the NIC and lead to a CQE
error.

This commit fixes the described issue by skipping tls_append_frag on
zero size to avoid adding empty fragments. The fix is not in the driver,
because an empty fragment is hardly the desired behavior.

Fixes: e8f69799810c ("net/tls: Add generic NIC offload infrastructure")
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Link: https://lore.kernel.org/r/20220426154949.159055-1-maximmi@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf
Jakub Kicinski [Wed, 27 Apr 2022 22:18:39 +0000 (15:18 -0700)]
Merge https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf

Daniel Borkmann says:

====================
pull-request: bpf 2022-04-27

We've added 5 non-merge commits during the last 20 day(s) which contain
a total of 6 files changed, 34 insertions(+), 12 deletions(-).

The main changes are:

1) Fix xsk sockets when rx and tx are separately bound to the same umem, also
   fix xsk copy mode combined with busy poll, from Maciej Fijalkowski.

2) Fix BPF tunnel/collect_md helpers with bpf_xmit lwt hook usage which triggered
   a crash due to invalid metadata_dst access, from Eyal Birger.

3) Fix release of page pool in XDP live packet mode, from Toke Høiland-Jørgensen.

4) Fix potential NULL pointer dereference in kretprobes, from Adam Zabrocki.

   (Masami & Steven preferred this small fix to be routed via bpf tree given it's
    follow-up fix to Masami's rethook work that went via bpf earlier, too.)

* https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf:
  xsk: Fix possible crash when multiple sockets are created
  kprobes: Fix KRETPROBES when CONFIG_KRETPROBE_ON_RETHOOK is set
  bpf, lwt: Fix crash when using bpf_skb_set_tunnel_key() from bpf_xmit lwt hook
  bpf: Fix release of page_pool in BPF_PROG_RUN in test runner
  xsk: Fix l2fwd for copy mode + busy poll combo
====================

Link: https://lore.kernel.org/r/20220427212748.9576-1-daniel@iogearbox.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agodrm/amdgpu: keep mmhub clock gating being enabled during s2idle suspend
Prike Liang [Tue, 19 Apr 2022 09:22:34 +0000 (17:22 +0800)]
drm/amdgpu: keep mmhub clock gating being enabled during s2idle suspend

Without MMHUB clock gating being enabled then MMHUB will not disconnect
from DF and will result in DF C-state entry can't be accessed during S2idle
suspend, and eventually s0ix entry will be blocked.

Signed-off-by: Prike Liang <Prike.Liang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/pm: fix the deadlock issue observed on SI
Evan Quan [Fri, 8 Apr 2022 11:51:34 +0000 (19:51 +0800)]
drm/amd/pm: fix the deadlock issue observed on SI

The adev->pm.mutx is already held at the beginning of
amdgpu_dpm_compute_clocks/amdgpu_dpm_enable_uvd/amdgpu_dpm_enable_vce.
But on their calling path, amdgpu_display_bandwidth_update will be
called and thus its sub functions amdgpu_dpm_get_sclk/mclk. They
will then try to acquire the same adev->pm.mutex and deadlock will
occur.

By placing amdgpu_display_bandwidth_update outside of adev->pm.mutex
protection(considering logically they do not need such protection) and
restructuring the call flow accordingly, we can eliminate the deadlock
issue. This comes with no real logics change.

Fixes: 3712e7a49459 ("drm/amd/pm: unified lock protections in amdgpu_dpm.c")
Reported-by: Paul Menzel <pmenzel@molgen.mpg.de>
Reported-by: Arthur Marsh <arthur.marsh@internode.on.net>
Link: https://lore.kernel.org/all/9e689fea-6c69-f4b0-8dee-32c4cf7d8f9c@molgen.mpg.de/
BugLink: https://gitlab.freedesktop.org/drm/amd/-/issues/1957
Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amd/display: Fix memory leak in dcn21_clock_source_create
Miaoqian Lin [Thu, 21 Apr 2022 09:03:09 +0000 (17:03 +0800)]
drm/amd/display: Fix memory leak in dcn21_clock_source_create

When dcn20_clk_src_construct() fails, we need to release clk_src.

Fixes: 6f4e6361c3ff ("drm/amd/display: Add Renoir resource (v2)")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdgpu: don't runtime suspend if there are displays attached (v3)
Alex Deucher [Tue, 28 Dec 2021 22:26:24 +0000 (17:26 -0500)]
drm/amdgpu: don't runtime suspend if there are displays attached (v3)

We normally runtime suspend when there are displays attached if they
are in the DPMS off state, however, if something wakes the GPU
we send a hotplug event on resume (in case any displays were connected
while the GPU was in suspend) which can cause userspace to light
up the displays again soon after they were turned off.

Prior to
commit 087451f372bf76 ("drm/amdgpu: use generic fb helpers instead of setting up AMD own's."),
the driver took a runtime pm reference when the fbdev emulation was
enabled because we didn't implement proper shadowing support for
vram access when the device was off so the device never runtime
suspended when there was a console bound.  Once that commit landed,
we now utilize the core fb helper implementation which properly
handles the emulation, so runtime pm now suspends in cases where it did
not before.  Ultimately, we need to sort out why runtime suspend in not
working in this case for some users, but this should restore similar
behavior to before.

v2: move check into runtime_suspend
v3: wake ups -> wakeups in comment, retain pm_runtime behavior in
    runtime_idle callback

Fixes: 087451f372bf76 ("drm/amdgpu: use generic fb helpers instead of setting up AMD own's.")
Link: https://lore.kernel.org/r/20220403132322.51c90903@darkstar.example.org/
Tested-by: Michele Ballabio <ballabio.m@gmail.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
2 years agodrm/amdkfd: CRIU add support for GWS queues
David Yat Sin [Wed, 13 Apr 2022 15:37:53 +0000 (11:37 -0400)]
drm/amdkfd: CRIU add support for GWS queues

Add support to checkpoint/restore GWS (Global Wave Sync) queues.

Signed-off-by: David Yat Sin <david.yatsin@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdkfd: Fix GWS queue count
David Yat Sin [Mon, 18 Apr 2022 15:55:58 +0000 (11:55 -0400)]
drm/amdkfd: Fix GWS queue count

dqm->gws_queue_count and pdd->qpd.mapped_gws_queue need to be updated
each time the queue gets evicted.

Fixes: b8020b0304c8 ("drm/amdkfd: Enable over-subscription with >1 GWS queue")
Signed-off-by: David Yat Sin <david.yatsin@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agoMerge branch 'akpm' (patches from Andrew)
Linus Torvalds [Wed, 27 Apr 2022 20:44:37 +0000 (13:44 -0700)]
Merge branch 'akpm' (patches from Andrew)

Merge fixes from Andrew Morton:
 "Two patches.

  Subsystems affected by this patch series: mm/kasan and mm/debug"

* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
  docs: vm/page_owner: use literal blocks for param description
  kasan: prevent cpu_quarantine corruption when CPU offline and cache shrink occur at same time

2 years agodocs: vm/page_owner: use literal blocks for param description
Akira Yokosawa [Wed, 27 Apr 2022 19:41:59 +0000 (12:41 -0700)]
docs: vm/page_owner: use literal blocks for param description

Sphinx generates hard-to-read lists of parameters at the bottom of the
page.  Fix them by putting literal-block markers of "::" in front of
them.

Link: https://lkml.kernel.org/r/cfd3bcc0-b51d-0c68-c065-ca1c4c202447@gmail.com
Signed-off-by: Akira Yokosawa <akiyks@gmail.com>
Fixes: 57f2b54a9379 ("Documentation/vm/page_owner.rst: update the documentation")
Cc: Shenghong Han <hanshenghong2019@email.szu.edu.cn>
Cc: Haowen Bai <baihaowen@meizu.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Alex Shi <seakeel@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agokasan: prevent cpu_quarantine corruption when CPU offline and cache shrink occur...
Zqiang [Wed, 27 Apr 2022 19:41:56 +0000 (12:41 -0700)]
kasan: prevent cpu_quarantine corruption when CPU offline and cache shrink occur at same time

kasan_quarantine_remove_cache() is called in kmem_cache_shrink()/
destroy().  The kasan_quarantine_remove_cache() call is protected by
cpuslock in kmem_cache_destroy() to ensure serialization with
kasan_cpu_offline().

However the kasan_quarantine_remove_cache() call is not protected by
cpuslock in kmem_cache_shrink().  When a CPU is going offline and cache
shrink occurs at same time, the cpu_quarantine may be corrupted by
interrupt (per_cpu_remove_cache operation).

So add a cpu_quarantine offline flags check in per_cpu_remove_cache().

[akpm@linux-foundation.org: add comment, per Zqiang]

Link: https://lkml.kernel.org/r/20220414025925.2423818-1-qiang1.zhang@intel.com
Signed-off-by: Zqiang <qiang1.zhang@intel.com>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Andrey Konovalov <andreyknvl@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agointel_idle: Fix SPR C6 optimization
Artem Bityutskiy [Wed, 27 Apr 2022 06:08:53 +0000 (09:08 +0300)]
intel_idle: Fix SPR C6 optimization

The Sapphire Rapids (SPR) C6 optimization was added to the end of the
'spr_idle_state_table_update()' function. However, the function has a
'return' which may happen before the optimization has a chance to run.
And this may prevent the optimization from happening.

This is an unlikely scenario, but possible if user boots with, say,
the 'intel_idle.preferred_cstates=6' kernel boot option.

This patch fixes the issue by eliminating the problematic 'return'
statement.

Fixes: 3a9cf77b60dc ("intel_idle: add core C6 optimization for SPR")
Suggested-by: Jan Beulich <jbeulich@suse.com>
Reported-by: Jan Beulich <jbeulich@suse.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
[ rjw: Minor changelog edits ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2 years agointel_idle: Fix the 'preferred_cstates' module parameter
Artem Bityutskiy [Wed, 27 Apr 2022 06:08:52 +0000 (09:08 +0300)]
intel_idle: Fix the 'preferred_cstates' module parameter

Problem description.

When user boots kernel up with the 'intel_idle.preferred_cstates=4' option,
we enable C1E and disable C1 states on Sapphire Rapids Xeon (SPR). In order
for C1E to work on SPR, we have to enable the C1E promotion bit on all
CPUs.  However, we enable it only on one CPU.

Fix description.

The 'intel_idle' driver already has the infrastructure for disabling C1E
promotion on every CPU. This patch uses the same infrastructure for
enabling C1E promotion on every CPU. It changes the boolean
'disable_promotion_to_c1e' variable to a tri-state 'c1e_promotion'
variable.

Tested on a 2-socket SPR system. I verified the following combinations:

 * C1E promotion enabled and disabled in BIOS.
 * Booted with and without the 'intel_idle.preferred_cstates=4' kernel
   argument.

In all 4 cases C1E promotion was correctly set on all CPUs.

Also tested on an old Broadwell system, just to make sure it does not cause
a regression. C1E promotion was correctly disabled on that system, both C1
and C1E were exposed (as expected).

Fixes: da0e58c038e6 ("intel_idle: add 'preferred_cstates' module argument")
Reported-by: Jan Beulich <jbeulich@suse.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
[ rjw: Minor changelog edits ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2 years agoMerge tag 'cpufreq-arm-fixes-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel...
Rafael J. Wysocki [Wed, 27 Apr 2022 18:20:03 +0000 (20:20 +0200)]
Merge tag 'cpufreq-arm-fixes-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/vireshk/pm

Pull ARM cpufreq fixes for 5.18-rc5 from Viresh Kumar:

"- Fix issues with the Qualcomm's cpufreq driver (Dmitry Baryshkov and
   Vladimir Zapolskiy).
 - Fix memory leak with the Sun501 driver (Xiaobing Luo)."

* tag 'cpufreq-arm-fixes-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/vireshk/pm:
  cpufreq: qcom-cpufreq-hw: Clear dcvs interrupts
  cpufreq: fix memory leak in sun50i_cpufreq_nvmem_probe
  cpufreq: qcom-cpufreq-hw: Fix throttle frequency value on EPSS platforms
  cpufreq: qcom-hw: provide online/offline operations
  cpufreq: qcom-hw: fix the opp entries refcounting
  cpufreq: qcom-hw: fix the race between LMH worker and cpuhp
  cpufreq: qcom-hw: drop affinity hint before freeing the IRQ

2 years agohex2bin: fix access beyond string end
Mikulas Patocka [Wed, 27 Apr 2022 15:26:40 +0000 (11:26 -0400)]
hex2bin: fix access beyond string end

If we pass too short string to "hex2bin" (and the string size without
the terminating NUL character is even), "hex2bin" reads one byte after
the terminating NUL character.  This patch fixes it.

Note that hex_to_bin returns -1 on error and hex2bin return -EINVAL on
error - so we can't just return the variable "hi" or "lo" on error.
This inconsistency may be fixed in the next merge window, but for the
purpose of fixing this bug, we just preserve the existing behavior and
return -1 and -EINVAL.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Fixes: b78049831ffe ("lib: add error checking to hex2bin")
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agohex2bin: make the function hex_to_bin constant-time
Mikulas Patocka [Mon, 25 Apr 2022 12:07:48 +0000 (08:07 -0400)]
hex2bin: make the function hex_to_bin constant-time

The function hex2bin is used to load cryptographic keys into device
mapper targets dm-crypt and dm-integrity.  It should take constant time
independent on the processed data, so that concurrently running
unprivileged code can't infer any information about the keys via
microarchitectural convert channels.

This patch changes the function hex_to_bin so that it contains no
branches and no memory accesses.

Note that this shouldn't cause performance degradation because the size
of the new function is the same as the size of the old function (on
x86-64) - and the new function causes no branch misprediction penalties.

I compile-tested this function with gcc on aarch64 alpha arm hppa hppa64
i386 ia64 m68k mips32 mips64 powerpc powerpc64 riscv sh4 s390x sparc32
sparc64 x86_64 and with clang on aarch64 arm hexagon i386 mips32 mips64
powerpc powerpc64 s390x sparc32 sparc64 x86_64 to verify that there are
no branches in the generated code.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agoMerge tag 'zonefs-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/dlemoal...
Linus Torvalds [Wed, 27 Apr 2022 17:30:29 +0000 (10:30 -0700)]
Merge tag 'zonefs-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/dlemoal/zonefs

Pull zonefs fixes from Damien Le Moal:
 "Two fixes for rc5:

   - Fix inode initialization to make sure that the inode flags are all
     cleared.

   - Use zone reset operation instead of close to make sure that the
     zone of an empty sequential file in never in an active state after
     closing the file"

* tag 'zonefs-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/dlemoal/zonefs:
  zonefs: Fix management of open zones
  zonefs: Clear inode information flags on inode creation

2 years agoMerge tag 'mtd/fixes-for-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 27 Apr 2022 17:14:52 +0000 (10:14 -0700)]
Merge tag 'mtd/fixes-for-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux

Pull MTD fixes from Miquel Raynal:
 "Core fix:

   - Fix a possible data corruption of the 'part' field in mtd_info

  Rawnand fixes:

   - Fix the check on the return value of wait_for_completion_timeout

   - Fix wrong ECC parameters for mt7622

   - Fix a possible memory corruption that might panic in the Qcom
     driver"

* tag 'mtd/fixes-for-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux:
  mtd: rawnand: qcom: fix memory corruption that causes panic
  mtd: fix 'part' field data corruption in mtd_info
  mtd: rawnand: Fix return value check of wait_for_completion_timeout
  mtd: rawnand: fix ecc parameters for mt7622

2 years agoAdd Eric Dumazet to networking maintainers
Jakub Kicinski [Tue, 26 Apr 2022 17:57:23 +0000 (10:57 -0700)]
Add Eric Dumazet to networking maintainers

Welcome Eric!

Acked-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Link: https://lore.kernel.org/r/20220426175723.417614-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agofloppy: disable FDRAWCMD by default
Willy Tarreau [Tue, 26 Apr 2022 20:41:05 +0000 (23:41 +0300)]
floppy: disable FDRAWCMD by default

Minh Yuan reported a concurrency use-after-free issue in the floppy code
between raw_cmd_ioctl and seek_interrupt.

[ It turns out this has been around, and that others have reported the
  KASAN splats over the years, but Minh Yuan had a reproducer for it and
  so gets primary credit for reporting it for this fix   - Linus ]

The problem is, this driver tends to break very easily and nowadays,
nobody is expected to use FDRAWCMD anyway since it was used to
manipulate non-standard formats.  The risk of breaking the driver is
higher than the risk presented by this race, and accessing the device
requires privileges anyway.

Let's just add a config option to completely disable this ioctl and
leave it disabled by default.  Distros shouldn't use it, and only those
running on antique hardware might need to enable it.

Link: https://lore.kernel.org/all/000000000000b71cdd05d703f6bf@google.com/
Link: https://lore.kernel.org/lkml/CAKcFiNC=MfYVW-Jt9A3=FPJpTwCD2PL_ULNCpsCVE5s8ZeBQgQ@mail.gmail.com
Link: https://lore.kernel.org/all/CAEAjamu1FRhz6StCe_55XY5s389ZP_xmCF69k987En+1z53=eg@mail.gmail.com
Reported-by: Minh Yuan <yuanmingbuaa@gmail.com>
Reported-by: syzbot+8e8958586909d62b6840@syzkaller.appspotmail.com
Reported-by: cruise k <cruise4k@gmail.com>
Reported-by: Kyungtae Kim <kt0755@gmail.com>
Suggested-by: Linus Torvalds <torvalds@linuxfoundation.org>
Tested-by: Denis Efremov <efremov@linux.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agoplatform/x86/intel: pmc/core: change pmc_lpm_modes to static
Tom Rix [Sat, 23 Apr 2022 12:30:48 +0000 (08:30 -0400)]
platform/x86/intel: pmc/core: change pmc_lpm_modes to static

Sparse reports this issue
core.c: note: in included file:
core.h:239:12: warning: symbol 'pmc_lpm_modes' was not declared. Should it be static?

Global variables should not be defined in headers.  This only works
because core.h is only included by core.c. Single file use
variables should be static, so change its storage-class specifier
to static.

Signed-off-by: Tom Rix <trix@redhat.com>
Reviewed-by: David E. Box <david.e.box@linux.intel.com>
Link: https://lore.kernel.org/r/20220423123048.591405-1-trix@redhat.com
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
2 years agoplatform/x86/intel/sdsi: Fix bug in multi packet reads
David E. Box [Wed, 20 Apr 2022 15:56:22 +0000 (08:56 -0700)]
platform/x86/intel/sdsi: Fix bug in multi packet reads

Fix bug that added an offset to the mailbox addr during multi-packet
reads. Did not affect current ABI since it doesn't support multi-packet
transactions.

Fixes: 2546c6000430 ("platform/x86: Add Intel Software Defined Silicon driver")
Signed-off-by: David E. Box <david.e.box@linux.intel.com>
Link: https://lore.kernel.org/r/20220420155622.1763633-4-david.e.box@linux.intel.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
2 years agoplatform/x86/intel/sdsi: Poll on ready bit for writes
David E. Box [Wed, 20 Apr 2022 15:56:21 +0000 (08:56 -0700)]
platform/x86/intel/sdsi: Poll on ready bit for writes

Due to change in firmware flow, update mailbox writes to poll on ready bit
instead of run_busy bit. This change makes the polling method consistent
for both writes and reads, which also uses the ready bit.

Fixes: 2546c6000430 ("platform/x86: Add Intel Software Defined Silicon driver")
Signed-off-by: David E. Box <david.e.box@linux.intel.com>
Link: https://lore.kernel.org/r/20220420155622.1763633-3-david.e.box@linux.intel.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
2 years agoplatform/x86/intel/sdsi: Handle leaky bucket
David E. Box [Wed, 20 Apr 2022 15:56:20 +0000 (08:56 -0700)]
platform/x86/intel/sdsi: Handle leaky bucket

To prevent an agent from indefinitely holding the mailbox firmware has
implemented a leaky bucket algorithm. Repeated access to the mailbox may
now incur a delay of up to 2.1 seconds. Add a retry loop that tries for
up to 2.5 seconds to acquire the mailbox.

Fixes: 2546c6000430 ("platform/x86: Add Intel Software Defined Silicon driver")
Signed-off-by: David E. Box <david.e.box@linux.intel.com>
Link: https://lore.kernel.org/r/20220420155622.1763633-2-david.e.box@linux.intel.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
2 years agoplatform/x86: intel-uncore-freq: Prevent driver loading in guests
Srinivas Pandruvada [Wed, 27 Apr 2022 10:03:04 +0000 (03:03 -0700)]
platform/x86: intel-uncore-freq: Prevent driver loading in guests

Loading this driver in guests results in unchecked MSR access error for
MSR 0x620.

There is no use of reading and modifying package/die scope uncore MSRs
in guests. So check for CPU feature X86_FEATURE_HYPERVISOR to prevent
loading of this driver in guests.

Fixes: dbce412a7733 ("platform/x86/intel-uncore-freq: Split common and enumeration part")
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=215870
Suggested-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Link: https://lore.kernel.org/r/20220427100304.2562990-1-srinivas.pandruvada@linux.intel.com
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
2 years agoplatform/x86: gigabyte-wmi: added support for B660 GAMING X DDR4 motherboard
Darryn Anton Jordan [Thu, 14 Apr 2022 14:24:43 +0000 (16:24 +0200)]
platform/x86: gigabyte-wmi: added support for B660 GAMING X DDR4 motherboard

This works on my system.

Signed-off-by: Darryn Anton Jordan <darrynjordan@icloud.com>
Acked-by: Thomas Weißschuh <thomas@weissschuh.net>
Link: https://lore.kernel.org/r/Ylguq87YG+9L3foV@hark
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
2 years agoplatform/x86: dell-laptop: Add quirk entry for Latitude 7520
Gabriele Mazzotta [Tue, 26 Apr 2022 12:08:27 +0000 (14:08 +0200)]
platform/x86: dell-laptop: Add quirk entry for Latitude 7520

The Latitude 7520 supports AC timeouts, but it has no KBD_LED_AC_TOKEN
and so changes to stop_timeout appear to have no effect if the laptop
is plugged in.

Signed-off-by: Gabriele Mazzotta <gabriele.mzt@gmail.com>
Acked-by: Pali Rohár <pali@kernel.org>
Link: https://lore.kernel.org/r/20220426120827.12363-1-gabriele.mzt@gmail.com
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
2 years agoplatform/x86: asus-wmi: Fix driver not binding when fan curve control probe fails
Hans de Goede [Wed, 27 Apr 2022 11:49:56 +0000 (13:49 +0200)]
platform/x86: asus-wmi: Fix driver not binding when fan curve control probe fails

Before this commit fan_curve_check_present() was trying to not cause
the probe to fail on devices without fan curve control by testing for
known error codes returned by asus_wmi_evaluate_method_buf().

Checking for ENODATA or ENODEV, with the latter being returned by this
function when an ACPI integer with a value of ASUS_WMI_UNSUPPORTED_METHOD
is returned. But for other ACPI integer returns this function just returns
them as is, including the ASUS_WMI_DSTS_UNKNOWN_BIT value of 2.

On the Asus U36SD ASUS_WMI_DSTS_UNKNOWN_BIT gets returned, leading to:

  asus-nb-wmi: probe of asus-nb-wmi failed with error 2

Instead of playing whack a mole with error codes here, simply treat all
errors as there not being any fan curves, fixing the driver no longer
loading on the Asus U36SD laptop.

Fixes: e3d13da7f77d ("platform/x86: asus-wmi: Fix regression when probing for fan curve control")
BugLink: https://bugzilla.redhat.com/show_bug.cgi?id=2079125
Cc: Luke D. Jones <luke@ljones.dev>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20220427114956.332919-1-hdegoede@redhat.com
2 years agoplatform/x86: asus-wmi: Potential buffer overflow in asus_wmi_evaluate_method_buf()
Dan Carpenter [Wed, 13 Apr 2022 07:37:44 +0000 (10:37 +0300)]
platform/x86: asus-wmi: Potential buffer overflow in asus_wmi_evaluate_method_buf()

This code tests for if the obj->buffer.length is larger than the buffer
but then it just does the memcpy() anyway.

Fixes: 0f0ac158d28f ("platform/x86: asus-wmi: Add support for custom fan curves")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Link: https://lore.kernel.org/r/20220413073744.GB8812@kili
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
2 years agonetfilter: conntrack: fix udp offload timeout sysctl
Volodymyr Mytnyk [Wed, 27 Apr 2022 11:09:00 +0000 (14:09 +0300)]
netfilter: conntrack: fix udp offload timeout sysctl

`nf_flowtable_udp_timeout` sysctl option is available only
if CONFIG_NFT_FLOW_OFFLOAD enabled. But infra for this flow
offload UDP timeout was added under CONFIG_NF_FLOW_TABLE
config option. So, if you have CONFIG_NFT_FLOW_OFFLOAD
disabled and CONFIG_NF_FLOW_TABLE enabled, the
`nf_flowtable_udp_timeout` is not present in sysfs.
Please note, that TCP flow offload timeout sysctl option
is present even CONFIG_NFT_FLOW_OFFLOAD is disabled.

I suppose it was a typo in commit that adds UDP flow offload
timeout and CONFIG_NF_FLOW_TABLE should be used instead.

Fixes: 975c57504da1 ("netfilter: conntrack: Introduce udp offload timeout configuration")
Signed-off-by: Volodymyr Mytnyk <volodymyr.mytnyk@plvision.eu>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2 years agonetfilter: nf_conntrack_tcp: re-init for syn packets only
Florian Westphal [Mon, 25 Apr 2022 09:47:11 +0000 (11:47 +0200)]
netfilter: nf_conntrack_tcp: re-init for syn packets only

Jaco Kroon reported tcp problems that Eric Dumazet and Neal Cardwell
pinpointed to nf_conntrack tcp_in_window() bug.

tcp trace shows following sequence:

I > R Flags [S], seq 3451342529, win 62580, options [.. tfo [|tcp]>
R > I Flags [S.], seq 2699962254, ack 3451342530, win 65535, options [..]
R > I Flags [P.], seq 1:89, ack 1, [..]

Note 3rd ACK is from responder to initiator so following branch is taken:
    } else if (((state->state == TCP_CONNTRACK_SYN_SENT
               && dir == IP_CT_DIR_ORIGINAL)
               || (state->state == TCP_CONNTRACK_SYN_RECV
               && dir == IP_CT_DIR_REPLY))
               && after(end, sender->td_end)) {

... because state == TCP_CONNTRACK_SYN_RECV and dir is REPLY.
This causes the scaling factor to be reset to 0: window scale option
is only present in syn(ack) packets.  This in turn makes nf_conntrack
mark valid packets as out-of-window.

This was always broken, it exists even in original commit where
window tracking was added to ip_conntrack (nf_conntrack predecessor)
in 2.6.9-rc1 kernel.

Restrict to 'tcph->syn', just like the 3rd condtional added in
commit 82b72cb94666 ("netfilter: conntrack: re-init state for retransmitted syn-ack").

Upon closer look, those conditionals/branches can be merged:

Because earlier checks prevent syn-ack from showing up in
original direction, the 'dir' checks in the conditional quoted above are
redundant, remove them. Return early for pure syn retransmitted in reply
direction (simultaneous open).

Fixes: 9fb9cbb1082d ("[NETFILTER]: Add nf_conntrack subsystem.")
Reported-by: Jaco Kroon <jaco@uls.co.za>
Signed-off-by: Florian Westphal <fw@strlen.de>
Acked-by: Jozsef Kadlecsik <kadlec@netfilter.org>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2 years agoMerge branch '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net
David S. Miller [Wed, 27 Apr 2022 09:58:39 +0000 (10:58 +0100)]
Merge branch '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net
-queue

Tony Nguyen says:

====================
Intel Wired LAN Driver Updates 2022-04-26

This series contains updates to ice driver only.

Ivan Vecera removes races related to VF message processing by changing
mutex_trylock() call to mutex_lock() and moving additional operations
to occur under mutex.

Petr Oros increases wait time after firmware flash as current time is
not sufficient.

Jake resolves a use-after-free issue for mailbox snapshot.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: dsa: lantiq_gswip: Don't set GSWIP_MII_CFG_RMII_CLK
Martin Blumenstingl [Mon, 25 Apr 2022 15:20:27 +0000 (17:20 +0200)]
net: dsa: lantiq_gswip: Don't set GSWIP_MII_CFG_RMII_CLK

Commit 4b5923249b8fa4 ("net: dsa: lantiq_gswip: Configure all remaining
GSWIP_MII_CFG bits") added all known bits in the GSWIP_MII_CFGp
register. It helped bring this register into a well-defined state so the
driver has to rely less on the bootloader to do things right.
Unfortunately it also sets the GSWIP_MII_CFG_RMII_CLK bit without any
possibility to configure it. Upon further testing it turns out that all
boards which are supported by the GSWIP driver in OpenWrt which use an
RMII PHY have a dedicated oscillator on the board which provides the
50MHz RMII reference clock.

Don't set the GSWIP_MII_CFG_RMII_CLK bit (but keep the code which always
clears it) to fix support for the Fritz!Box 7362 SL in OpenWrt. This is
a board with two Atheros AR8030 RMII PHYs. With the "RMII clock" bit set
the MAC also generates the RMII reference clock whose signal then
conflicts with the signal from the oscillator on the board. This results
in a constant cycle of the PHY detecting link up/down (and as a result
of that: the two ports using the AR8030 PHYs are not working).

At the time of writing this patch there's no known board where the MAC
(GSWIP) has to generate the RMII reference clock. If needed this can be
implemented in future by providing a device-tree flag so the
GSWIP_MII_CFG_RMII_CLK bit can be toggled per port.

Fixes: 4b5923249b8fa4 ("net: dsa: lantiq_gswip: Configure all remaining GSWIP_MII_CFG bits")
Tested-by: Jan Hoffmann <jan@3e8.eu>
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Acked-by: Hauke Mehrtens <hauke@hauke-m.de>
Link: https://lore.kernel.org/r/20220425152027.2220750-1-martin.blumenstingl@googlemail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: Use this_cpu_inc() to increment net->core_stats
Sebastian Andrzej Siewior [Mon, 25 Apr 2022 16:39:46 +0000 (18:39 +0200)]
net: Use this_cpu_inc() to increment net->core_stats

The macro dev_core_stats_##FIELD##_inc() disables preemption and invokes
netdev_core_stats_alloc() to return a per-CPU pointer.
netdev_core_stats_alloc() will allocate memory on its first invocation
which breaks on PREEMPT_RT because it requires non-atomic context for
memory allocation.

This can be avoided by enabling preemption in netdev_core_stats_alloc()
assuming the caller always disables preemption.

It might be better to replace local_inc() with this_cpu_inc() now that
dev_core_stats_##FIELD##_inc() gained a preempt-disable section and does
not rely on already disabled preemption. This results in less
instructions on x86-64:
local_inc:
|          incl %gs:__preempt_count(%rip)  # __preempt_count
|          movq    488(%rdi), %rax # _1->core_stats, _22
|          testq   %rax, %rax      # _22
|          je      .L585   #,
|          add %gs:this_cpu_off(%rip), %rax        # this_cpu_off, tcp_ptr__
|  .L586:
|          testq   %rax, %rax      # _27
|          je      .L587   #,
|          incq (%rax)            # _6->a.counter
|  .L587:
|          decl %gs:__preempt_count(%rip)  # __preempt_count

this_cpu_inc(), this patch:
|         movq    488(%rdi), %rax # _1->core_stats, _5
|         testq   %rax, %rax      # _5
|         je      .L591   #,
| .L585:
|         incq %gs:(%rax) # _18->rx_dropped

Use unsigned long as type for the counter. Use this_cpu_inc() to
increment the counter. Use a plain read of the counter.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/YmbO0pxgtKpCw4SY@linutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge tag 'pinctrl-v5.18-2' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw...
Linus Torvalds [Tue, 26 Apr 2022 23:34:11 +0000 (16:34 -0700)]
Merge tag 'pinctrl-v5.18-2' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl

Pull pin control fixes from Linus Walleij:

 - Fix some register offsets on Intel Alderlake

 - Fix the order the UFS and SDC pins on Qualcomm SM6350

 - Fix a build error in Mediatek Moore.

 - Fix a pin function table in the Sunplus SP7021.

 - Fix some Kconfig and static keywords on the Samsung Tesla FSD SoC.

 - Fix up the EOI function for edge triggered IRQs and keep the block
   clock enabled for level IRQs in the STM32 driver.

 - Fix some bits and order in the Rockchip RK3308 driver.

 - Handle the errorpath in the Pistachio driver probe() properly.

* tag 'pinctrl-v5.18-2' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl:
  pinctrl: pistachio: fix use of irq_of_parse_and_map()
  pinctrl: stm32: Keep pinctrl block clock enabled when LEVEL IRQ requested
  pinctrl: rockchip: sort the rk3308_mux_recalced_data entries
  pinctrl: rockchip: fix RK3308 pinmux bits
  pinctrl: stm32: Do not call stm32_gpio_get() for edge triggered IRQs in EOI
  pinctrl: Fix an error in pin-function table of SP7021
  pinctrl: samsung: fix missing GPIOLIB on ARM64 Exynos config
  pinctrl: mediatek: moore: Fix build error
  pinctrl: qcom: sm6350: fix order of UFS & SDC pins
  pinctrl: alderlake: Fix register offsets for ADL-N variant
  pinctrl: samsung: staticize fsd_pin_ctrl

2 years agoMerge tag 'for-5.18/fbdev-2' of git://git.kernel.org/pub/scm/linux/kernel/git/deller...
Linus Torvalds [Tue, 26 Apr 2022 18:32:01 +0000 (11:32 -0700)]
Merge tag 'for-5.18/fbdev-2' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/linux-fbdev

Pull fbdev fixes and updates from Helge Deller:
 "A bunch of outstanding fbdev patches - all trivial and small"

* tag 'for-5.18/fbdev-2' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/linux-fbdev:
  video: fbdev: clps711x-fb: Use syscon_regmap_lookup_by_phandle
  video: fbdev: mmp: replace usage of found with dedicated list iterator variable
  video: fbdev: sh_mobile_lcdcfb: Remove sh_mobile_lcdc_check_var() declaration
  video: fbdev: i740fb: Error out if 'pixclock' equals zero
  video: fbdev: i740fb: use memset_io() to clear screen
  video: fbdev: s3fb: Error out if 'pixclock' equals zero
  video: fbdev: arkfb: Error out if 'pixclock' equals zero
  video: fbdev: tridentfb: Error out if 'pixclock' equals zero
  video: fbdev: vt8623fb: Error out if 'pixclock' equals zero
  video: fbdev: kyro: Error out if 'lineclock' equals zero
  video: fbdev: neofb: Fix the check of 'var->pixclock'
  video: fbdev: imxfb: Fix missing of_node_put in imxfb_probe
  video: fbdev: omap: Make it CCF clk API compatible
  video: fbdev: aty/matrox/...: Prepare cleanup of powerpc's asm/prom.h
  video: fbdev: pm2fb: Fix a kernel-doc formatting issue
  linux/fb.h: Spelling s/palette/palette/
  video: fbdev: sis: fix potential NULL dereference in sisfb_post_sis300()
  video: fbdev: pxafb: use if else instead
  video: fbdev: udlfb: properly check endpoint type
  video: fbdev: of: display_timing: Remove a redundant zeroing of memory

2 years agoMerge tag 'gfs2-v5.18-rc4-fix' of git://git.kernel.org/pub/scm/linux/kernel/git/gfs2...
Linus Torvalds [Tue, 26 Apr 2022 18:17:18 +0000 (11:17 -0700)]
Merge tag 'gfs2-v5.18-rc4-fix' of git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2

Pull gfs2 fix from Andreas Gruenbacher:

 - Only re-check for direct I/O writes past the end of the file after
   re-acquiring the inode glock.

* tag 'gfs2-v5.18-rc4-fix' of git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2:
  gfs2: Don't re-check for write past EOF unnecessarily

2 years agoBluetooth: hci_sync: Cleanup hci_conn if it cannot be aborted
Luiz Augusto von Dentz [Fri, 22 Apr 2022 19:58:18 +0000 (12:58 -0700)]
Bluetooth: hci_sync: Cleanup hci_conn if it cannot be aborted

This attempts to cleanup the hci_conn if it cannot be aborted as
otherwise it would likely result in having the controller and host
stack out of sync with respect to connection handle.

Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
2 years agoMerge tag 'for-5.18-rc4-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave...
Linus Torvalds [Tue, 26 Apr 2022 18:10:42 +0000 (11:10 -0700)]
Merge tag 'for-5.18-rc4-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux

Pull btrfs fixes from David Sterba:

 - direct IO fixes:

      - restore passing file offset to correctly calculate checksums
        when repairing on read and bio split happens

      - use correct bio when sumitting IO on zoned filesystem

 - zoned mode fixes:

      - fix selection of device to correctly calculate device
        capabilities when allocating a new bio

      - use a dedicated lock for exclusion during relocation

      - fix leaked plug after failure syncing log

 - fix assertion during scrub and relocation

* tag 'for-5.18-rc4-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
  btrfs: zoned: use dedicated lock for data relocation
  btrfs: fix assertion failure during scrub due to block group reallocation
  btrfs: fix direct I/O writes for split bios on zoned devices
  btrfs: fix direct I/O read repair for split bios
  btrfs: fix and document the zoned device choice in alloc_new_bio
  btrfs: fix leaked plug after failure syncing log on zoned filesystems

2 years agoBluetooth: hci_event: Fix creating hci_conn object on error status
Luiz Augusto von Dentz [Fri, 22 Apr 2022 19:58:17 +0000 (12:58 -0700)]
Bluetooth: hci_event: Fix creating hci_conn object on error status

It is useless to create a hci_conn object if on error status as the
result would be it being freed in the process and anyway it is likely
the result of controller and host stack being out of sync.

Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
2 years agoBluetooth: hci_event: Fix checking for invalid handle on error status
Luiz Augusto von Dentz [Fri, 22 Apr 2022 19:58:16 +0000 (12:58 -0700)]
Bluetooth: hci_event: Fix checking for invalid handle on error status

Commit d5ebaa7c5f6f6 introduces checks for handle range
(e.g HCI_CONN_HANDLE_MAX) but controllers like Intel AX200 don't seem
to respect the valid range int case of error status:

> HCI Event: Connect Complete (0x03) plen 11
        Status: Page Timeout (0x04)
        Handle: 65535
        Address: 94:DB:56:XX:XX:XX (Sony Home Entertainment&
Sound Products Inc)
        Link type: ACL (0x01)
        Encryption: Disabled (0x00)
[1644965.827560] Bluetooth: hci0: Ignoring HCI_Connection_Complete for invalid handle

Because of it is impossible to cleanup the connections properly since
the stack would attempt to cancel the connection which is no longer in
progress causing the following trace:

< HCI Command: Create Connection Cancel (0x01|0x0008) plen 6
        Address: 94:DB:56:XX:XX:XX (Sony Home Entertainment&
Sound Products Inc)
= bluetoothd: src/profile.c:record_cb() Unable to get Hands-Free Voice
gateway SDP record: Connection timed out
> HCI Event: Command Complete (0x0e) plen 10
      Create Connection Cancel (0x01|0x0008) ncmd 1
        Status: Unknown Connection Identifier (0x02)
        Address: 94:DB:56:XX:XX:XX (Sony Home Entertainment&
Sound Products Inc)
< HCI Command: Create Connection Cancel (0x01|0x0008) plen 6
        Address: 94:DB:56:XX:XX:XX (Sony Home Entertainment&
Sound Products Inc)

Fixes: d5ebaa7c5f6f6 ("Bluetooth: hci_event: Ignore multiple conn complete events")
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
2 years agoice: fix use-after-free when deinitializing mailbox snapshot
Jacob Keller [Tue, 12 Apr 2022 17:34:22 +0000 (10:34 -0700)]
ice: fix use-after-free when deinitializing mailbox snapshot

During ice_sriov_configure, if num_vfs is 0, we are being asked by the
kernel to remove all VFs.

The driver first de-initializes the snapshot before freeing all the VFs.
This results in a use-after-free BUG detected by KASAN. The bug occurs
because the snapshot can still be accessed until all VFs are removed.

Fix this by freeing all the VFs first before calling
ice_mbx_deinit_snapshot.

[  +0.032591] ==================================================================
[  +0.000021] BUG: KASAN: use-after-free in ice_mbx_vf_state_handler+0x1c3/0x410 [ice]
[  +0.000315] Write of size 28 at addr ffff889908eb6f28 by task kworker/55:2/1530996

[  +0.000029] CPU: 55 PID: 1530996 Comm: kworker/55:2 Kdump: loaded Tainted: G S        I       5.17.0-dirty #1
[  +0.000022] Hardware name: Dell Inc. PowerEdge R740/0923K0, BIOS 1.6.13 12/17/2018
[  +0.000013] Workqueue: ice ice_service_task [ice]
[  +0.000279] Call Trace:
[  +0.000012]  <TASK>
[  +0.000011]  dump_stack_lvl+0x33/0x42
[  +0.000030]  print_report.cold.13+0xb2/0x6b3
[  +0.000028]  ? ice_mbx_vf_state_handler+0x1c3/0x410 [ice]
[  +0.000295]  kasan_report+0xa5/0x120
[  +0.000026]  ? __switch_to_asm+0x21/0x70
[  +0.000024]  ? ice_mbx_vf_state_handler+0x1c3/0x410 [ice]
[  +0.000298]  kasan_check_range+0x183/0x1e0
[  +0.000019]  memset+0x1f/0x40
[  +0.000018]  ice_mbx_vf_state_handler+0x1c3/0x410 [ice]
[  +0.000304]  ? ice_conv_link_speed_to_virtchnl+0x160/0x160 [ice]
[  +0.000297]  ? ice_vsi_dis_spoofchk+0x40/0x40 [ice]
[  +0.000305]  ice_is_malicious_vf+0x1aa/0x250 [ice]
[  +0.000303]  ? ice_restore_all_vfs_msi_state+0x160/0x160 [ice]
[  +0.000297]  ? __mutex_unlock_slowpath.isra.15+0x410/0x410
[  +0.000022]  ? ice_debug_cq+0xb7/0x230 [ice]
[  +0.000273]  ? __kasan_slab_alloc+0x2f/0x90
[  +0.000022]  ? memset+0x1f/0x40
[  +0.000017]  ? do_raw_spin_lock+0x119/0x1d0
[  +0.000022]  ? rwlock_bug.part.2+0x60/0x60
[  +0.000024]  __ice_clean_ctrlq+0x3a6/0xd60 [ice]
[  +0.000273]  ? newidle_balance+0x5b1/0x700
[  +0.000026]  ? ice_print_link_msg+0x2f0/0x2f0 [ice]
[  +0.000271]  ? update_cfs_group+0x1b/0x140
[  +0.000018]  ? load_balance+0x1260/0x1260
[  +0.000022]  ? ice_process_vflr_event+0x27/0x130 [ice]
[  +0.000301]  ice_service_task+0x136e/0x1470 [ice]
[  +0.000281]  process_one_work+0x3b4/0x6c0
[  +0.000030]  worker_thread+0x65/0x660
[  +0.000023]  ? __kthread_parkme+0xe4/0x100
[  +0.000021]  ? process_one_work+0x6c0/0x6c0
[  +0.000020]  kthread+0x179/0x1b0
[  +0.000018]  ? kthread_complete_and_exit+0x20/0x20
[  +0.000022]  ret_from_fork+0x22/0x30
[  +0.000026]  </TASK>

[  +0.000018] Allocated by task 10742:
[  +0.000013]  kasan_save_stack+0x1c/0x40
[  +0.000018]  __kasan_kmalloc+0x84/0xa0
[  +0.000016]  kmem_cache_alloc_trace+0x16c/0x2e0
[  +0.000015]  intel_iommu_probe_device+0xeb/0x860
[  +0.000015]  __iommu_probe_device+0x9a/0x2f0
[  +0.000016]  iommu_probe_device+0x43/0x270
[  +0.000015]  iommu_bus_notifier+0xa7/0xd0
[  +0.000015]  blocking_notifier_call_chain+0x90/0xc0
[  +0.000017]  device_add+0x5f3/0xd70
[  +0.000014]  pci_device_add+0x404/0xa40
[  +0.000015]  pci_iov_add_virtfn+0x3b0/0x550
[  +0.000016]  sriov_enable+0x3bb/0x600
[  +0.000013]  ice_ena_vfs+0x113/0xa79 [ice]
[  +0.000293]  ice_sriov_configure.cold.17+0x21/0xe0 [ice]
[  +0.000291]  sriov_numvfs_store+0x160/0x200
[  +0.000015]  kernfs_fop_write_iter+0x1db/0x270
[  +0.000018]  new_sync_write+0x21d/0x330
[  +0.000013]  vfs_write+0x376/0x410
[  +0.000013]  ksys_write+0xba/0x150
[  +0.000012]  do_syscall_64+0x3a/0x80
[  +0.000012]  entry_SYSCALL_64_after_hwframe+0x44/0xae

[  +0.000028] Freed by task 10742:
[  +0.000011]  kasan_save_stack+0x1c/0x40
[  +0.000015]  kasan_set_track+0x21/0x30
[  +0.000016]  kasan_set_free_info+0x20/0x30
[  +0.000012]  __kasan_slab_free+0x104/0x170
[  +0.000016]  kfree+0x9b/0x470
[  +0.000013]  devres_destroy+0x1c/0x20
[  +0.000015]  devm_kfree+0x33/0x40
[  +0.000012]  ice_mbx_deinit_snapshot+0x39/0x70 [ice]
[  +0.000295]  ice_sriov_configure+0xb0/0x260 [ice]
[  +0.000295]  sriov_numvfs_store+0x1bc/0x200
[  +0.000015]  kernfs_fop_write_iter+0x1db/0x270
[  +0.000016]  new_sync_write+0x21d/0x330
[  +0.000012]  vfs_write+0x376/0x410
[  +0.000012]  ksys_write+0xba/0x150
[  +0.000012]  do_syscall_64+0x3a/0x80
[  +0.000012]  entry_SYSCALL_64_after_hwframe+0x44/0xae

[  +0.000024] Last potentially related work creation:
[  +0.000010]  kasan_save_stack+0x1c/0x40
[  +0.000016]  __kasan_record_aux_stack+0x98/0xa0
[  +0.000013]  insert_work+0x34/0x160
[  +0.000015]  __queue_work+0x20e/0x650
[  +0.000016]  queue_work_on+0x4c/0x60
[  +0.000015]  nf_nat_masq_schedule+0x297/0x2e0 [nf_nat]
[  +0.000034]  masq_device_event+0x5a/0x60 [nf_nat]
[  +0.000031]  raw_notifier_call_chain+0x5f/0x80
[  +0.000017]  dev_close_many+0x1d6/0x2c0
[  +0.000015]  unregister_netdevice_many+0x4e3/0xa30
[  +0.000015]  unregister_netdevice_queue+0x192/0x1d0
[  +0.000014]  iavf_remove+0x8f9/0x930 [iavf]
[  +0.000058]  pci_device_remove+0x65/0x110
[  +0.000015]  device_release_driver_internal+0xf8/0x190
[  +0.000017]  pci_stop_bus_device+0xb5/0xf0
[  +0.000014]  pci_stop_and_remove_bus_device+0xe/0x20
[  +0.000016]  pci_iov_remove_virtfn+0x19c/0x230
[  +0.000015]  sriov_disable+0x4f/0x170
[  +0.000014]  ice_free_vfs+0x9a/0x490 [ice]
[  +0.000306]  ice_sriov_configure+0xb8/0x260 [ice]
[  +0.000294]  sriov_numvfs_store+0x1bc/0x200
[  +0.000015]  kernfs_fop_write_iter+0x1db/0x270
[  +0.000016]  new_sync_write+0x21d/0x330
[  +0.000012]  vfs_write+0x376/0x410
[  +0.000012]  ksys_write+0xba/0x150
[  +0.000012]  do_syscall_64+0x3a/0x80
[  +0.000012]  entry_SYSCALL_64_after_hwframe+0x44/0xae

[  +0.000025] The buggy address belongs to the object at ffff889908eb6f00
               which belongs to the cache kmalloc-96 of size 96
[  +0.000016] The buggy address is located 40 bytes inside of
               96-byte region [ffff889908eb6f00ffff889908eb6f60)

[  +0.000026] The buggy address belongs to the physical page:
[  +0.000010] page:00000000b7e99a2e refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x1908eb6
[  +0.000016] flags: 0x57ffffc0000200(slab|node=1|zone=2|lastcpupid=0x1fffff)
[  +0.000024] raw: 0057ffffc0000200 ffffea0069d9fd80 dead000000000002 ffff88810004c780
[  +0.000015] raw: 0000000000000000 0000000000200020 00000001ffffffff 0000000000000000
[  +0.000009] page dumped because: kasan: bad access detected

[  +0.000016] Memory state around the buggy address:
[  +0.000012]  ffff889908eb6e00: fa fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc
[  +0.000014]  ffff889908eb6e80: fa fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc
[  +0.000014] >ffff889908eb6f00: fa fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc
[  +0.000011]                                   ^
[  +0.000013]  ffff889908eb6f80: fa fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc
[  +0.000013]  ffff889908eb7000: fa fb fb fb fb fb fb fb fc fc fc fc fa fb fb fb
[  +0.000012] ==================================================================

Fixes: 0891c89674e8 ("ice: warn about potentially malicious VFs")
Reported-by: Slawomir Laba <slawomirx.laba@intel.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
2 years agoice: wait 5 s for EMP reset after firmware flash
Petr Oros [Wed, 13 Apr 2022 15:37:45 +0000 (17:37 +0200)]
ice: wait 5 s for EMP reset after firmware flash

We need to wait 5 s for EMP reset after firmware flash. Code was extracted
from OOT driver (ice v1.8.3 downloaded from sourceforge). Without this
wait, fw_activate let card in inconsistent state and recoverable only
by second flash/activate. Flash was tested on these fw's:
From -> To
 3.00 -> 3.10/3.20
 3.10 -> 3.00/3.20
 3.20 -> 3.00/3.10

Reproducer:
[root@host ~]# devlink dev flash pci/0000:ca:00.0 file E810_XXVDA4_FH_O_SEC_FW_1p6p1p9_NVM_3p10_PLDMoMCTP_0.11_8000AD7B.bin
Preparing to flash
[fw.mgmt] Erasing
[fw.mgmt] Erasing done
[fw.mgmt] Flashing 100%
[fw.mgmt] Flashing done 100%
[fw.undi] Erasing
[fw.undi] Erasing done
[fw.undi] Flashing 100%
[fw.undi] Flashing done 100%
[fw.netlist] Erasing
[fw.netlist] Erasing done
[fw.netlist] Flashing 100%
[fw.netlist] Flashing done 100%
Activate new firmware by devlink reload
[root@host ~]# devlink dev reload pci/0000:ca:00.0 action fw_activate
reload_actions_performed:
    fw_activate
[root@host ~]# ip link show ens7f0
71: ens7f0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc mq state DOWN mode DEFAULT group default qlen 1000
    link/ether b4:96:91:dc:72:e0 brd ff:ff:ff:ff:ff:ff
    altname enp202s0f0

dmesg after flash:
[   55.120788] ice: Copyright (c) 2018, Intel Corporation.
[   55.274734] ice 0000:ca:00.0: Get PHY capabilities failed status = -5, continuing anyway
[   55.569797] ice 0000:ca:00.0: The DDP package was successfully loaded: ICE OS Default Package version 1.3.28.0
[   55.603629] ice 0000:ca:00.0: Get PHY capability failed.
[   55.608951] ice 0000:ca:00.0: ice_init_nvm_phy_type failed: -5
[   55.647348] ice 0000:ca:00.0: PTP init successful
[   55.675536] ice 0000:ca:00.0: DCB is enabled in the hardware, max number of TCs supported on this port are 8
[   55.685365] ice 0000:ca:00.0: FW LLDP is disabled, DCBx/LLDP in SW mode.
[   55.692179] ice 0000:ca:00.0: Commit DCB Configuration to the hardware
[   55.701382] ice 0000:ca:00.0: 126.024 Gb/s available PCIe bandwidth, limited by 16.0 GT/s PCIe x8 link at 0000:c9:02.0 (capable of 252.048 Gb/s with 16.0 GT/s PCIe x16 link)
Reboot doesn’t help, only second flash/activate with OOT or patched
driver put card back in consistent state.

After patch:
[root@host ~]# devlink dev flash pci/0000:ca:00.0 file E810_XXVDA4_FH_O_SEC_FW_1p6p1p9_NVM_3p10_PLDMoMCTP_0.11_8000AD7B.bin
Preparing to flash
[fw.mgmt] Erasing
[fw.mgmt] Erasing done
[fw.mgmt] Flashing 100%
[fw.mgmt] Flashing done 100%
[fw.undi] Erasing
[fw.undi] Erasing done
[fw.undi] Flashing 100%
[fw.undi] Flashing done 100%
[fw.netlist] Erasing
[fw.netlist] Erasing done
[fw.netlist] Flashing 100%
[fw.netlist] Flashing done 100%
Activate new firmware by devlink reload
[root@host ~]# devlink dev reload pci/0000:ca:00.0 action fw_activate
reload_actions_performed:
    fw_activate
[root@host ~]# ip link show ens7f0
19: ens7f0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP mode DEFAULT group default qlen 1000
    link/ether b4:96:91:dc:72:e0 brd ff:ff:ff:ff:ff:ff
    altname enp202s0f0

Fixes: 399e27dbbd9e94 ("ice: support immediate firmware activation via devlink reload")
Signed-off-by: Petr Oros <poros@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 agoice: Protect vf_state check by cfg_lock in ice_vc_process_vf_msg()
Ivan Vecera [Tue, 19 Apr 2022 14:22:21 +0000 (16:22 +0200)]
ice: Protect vf_state check by cfg_lock in ice_vc_process_vf_msg()

Previous patch labelled "ice: Fix incorrect locking in
ice_vc_process_vf_msg()"  fixed an issue with ignored messages
sent by VF driver but a small race window still left.

Recently caught trace during 'ip link set ... vf 0 vlan ...' operation:

[ 7332.995625] ice 0000:3b:00.0: Clearing port VLAN on VF 0
[ 7333.001023] iavf 0000:3b:01.0: Reset indication received from the PF
[ 7333.007391] iavf 0000:3b:01.0: Scheduling reset task
[ 7333.059575] iavf 0000:3b:01.0: PF returned error -5 (IAVF_ERR_PARAM) to our request 3
[ 7333.059626] ice 0000:3b:00.0: Invalid message from VF 0, opcode 3, len 4, error -1

Setting of VLAN for VF causes a reset of the affected VF using
ice_reset_vf() function that runs with cfg_lock taken:

1. ice_notify_vf_reset() informs IAVF driver that reset is needed and
   IAVF schedules its own reset procedure
2. Bit ICE_VF_STATE_DIS is set in vf->vf_state
3. Misc initialization steps
4. ice_sriov_post_vsi_rebuild() -> ice_vf_set_initialized() and that
   clears ICE_VF_STATE_DIS in vf->vf_state

Step 3 is mentioned race window because IAVF reset procedure runs in
parallel and one of its step is sending of VIRTCHNL_OP_GET_VF_RESOURCES
message (opcode==3). This message is handled in ice_vc_process_vf_msg()
and if it is received during the mentioned race window then it's
marked as invalid and error is returned to VF driver.

Protect vf_state check in ice_vc_process_vf_msg() by cfg_lock to avoid
this race condition.

Fixes: e6ba5273d4ed ("ice: Fix race conditions between virtchnl handling and VF ndo ops")
Tested-by: Fei Liu <feliu@redhat.com>
Signed-off-by: Ivan Vecera <ivecera@redhat.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
2 years agoice: Fix incorrect locking in ice_vc_process_vf_msg()
Ivan Vecera [Fri, 1 Apr 2022 10:40:52 +0000 (12:40 +0200)]
ice: Fix incorrect locking in ice_vc_process_vf_msg()

Usage of mutex_trylock() in ice_vc_process_vf_msg() is incorrect
because message sent from VF is ignored and never processed.

Use mutex_lock() instead to fix the issue. It is safe because this
mutex is used to prevent races between VF related NDOs and
handlers processing request messages from VF and these handlers
are running in ice_service_task() context. Additionally move this
mutex lock prior ice_vc_is_opcode_allowed() call to avoid potential
races during allowlist access.

Fixes: e6ba5273d4ed ("ice: Fix race conditions between virtchnl handling and VF ndo ops")
Signed-off-by: Ivan Vecera <ivecera@redhat.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
2 years agoxsk: Fix possible crash when multiple sockets are created
Maciej Fijalkowski [Mon, 25 Apr 2022 15:37:45 +0000 (17:37 +0200)]
xsk: Fix possible crash when multiple sockets are created

Fix a crash that happens if an Rx only socket is created first, then a
second socket is created that is Tx only and bound to the same umem as
the first socket and also the same netdev and queue_id together with the
XDP_SHARED_UMEM flag. In this specific case, the tx_descs array page
pool was not created by the first socket as it was an Rx only socket.
When the second socket is bound it needs this tx_descs array of this
shared page pool as it has a Tx component, but unfortunately it was
never allocated, leading to a crash. Note that this array is only used
for zero-copy drivers using the batched Tx APIs, currently only ice and
i40e.

[ 5511.150360] BUG: kernel NULL pointer dereference, address: 0000000000000008
[ 5511.158419] #PF: supervisor write access in kernel mode
[ 5511.164472] #PF: error_code(0x0002) - not-present page
[ 5511.170416] PGD 0 P4D 0
[ 5511.173347] Oops: 0002 [#1] PREEMPT SMP PTI
[ 5511.178186] CPU: 0 PID: 0 Comm: swapper/0 Tainted: G            E     5.18.0-rc1+ #97
[ 5511.187245] Hardware name: Intel Corp. GRANTLEY/GRANTLEY, BIOS GRRFCRB1.86B.0276.D07.1605190235 05/19/2016
[ 5511.198418] RIP: 0010:xsk_tx_peek_release_desc_batch+0x198/0x310
[ 5511.205375] Code: c0 83 c6 01 84 c2 74 6d 8d 46 ff 23 07 44 89 e1 48 83 c0 14 48 c1 e1 04 48 c1 e0 04 48 03 47 10 4c 01 c1 48 8b 50 08 48 8b 00 <48> 89 51 08 48 89 01 41 80 bd d7 00 00 00 00 75 82 48 8b 19 49 8b
[ 5511.227091] RSP: 0018:ffffc90000003dd0 EFLAGS: 00010246
[ 5511.233135] RAX: 0000000000000000 RBX: ffff88810c8da600 RCX: 0000000000000000
[ 5511.241384] RDX: 000000000000003c RSI: 0000000000000001 RDI: ffff888115f555c0
[ 5511.249634] RBP: ffffc90000003e08 R08: 0000000000000000 R09: ffff889092296b48
[ 5511.257886] R10: 0000ffffffffffff R11: ffff889092296800 R12: 0000000000000000
[ 5511.266138] R13: ffff88810c8db500 R14: 0000000000000040 R15: 0000000000000100
[ 5511.274387] FS:  0000000000000000(0000) GS:ffff88903f800000(0000) knlGS:0000000000000000
[ 5511.283746] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 5511.290389] CR2: 0000000000000008 CR3: 00000001046e2001 CR4: 00000000003706f0
[ 5511.298640] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 5511.306892] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[ 5511.315142] Call Trace:
[ 5511.317972]  <IRQ>
[ 5511.320301]  ice_xmit_zc+0x68/0x2f0 [ice]
[ 5511.324977]  ? ktime_get+0x38/0xa0
[ 5511.328913]  ice_napi_poll+0x7a/0x6a0 [ice]
[ 5511.333784]  __napi_poll+0x2c/0x160
[ 5511.337821]  net_rx_action+0xdd/0x200
[ 5511.342058]  __do_softirq+0xe6/0x2dd
[ 5511.346198]  irq_exit_rcu+0xb5/0x100
[ 5511.350339]  common_interrupt+0xa4/0xc0
[ 5511.354777]  </IRQ>
[ 5511.357201]  <TASK>
[ 5511.359625]  asm_common_interrupt+0x1e/0x40
[ 5511.364466] RIP: 0010:cpuidle_enter_state+0xd2/0x360
[ 5511.370211] Code: 49 89 c5 0f 1f 44 00 00 31 ff e8 e9 00 7b ff 45 84 ff 74 12 9c 58 f6 c4 02 0f 85 72 02 00 00 31 ff e8 02 0c 80 ff fb 45 85 f6 <0f> 88 11 01 00 00 49 63 c6 4c 2b 2c 24 48 8d 14 40 48 8d 14 90 49
[ 5511.391921] RSP: 0018:ffffffff82a03e60 EFLAGS: 00000202
[ 5511.397962] RAX: ffff88903f800000 RBX: 0000000000000001 RCX: 000000000000001f
[ 5511.406214] RDX: 0000000000000000 RSI: ffffffff823400b9 RDI: ffffffff8234c046
[ 5511.424646] RBP: ffff88810a384800 R08: 000005032a28c046 R09: 0000000000000008
[ 5511.443233] R10: 000000000000000b R11: 0000000000000006 R12: ffffffff82bcf700
[ 5511.461922] R13: 000005032a28c046 R14: 0000000000000001 R15: 0000000000000000
[ 5511.480300]  cpuidle_enter+0x29/0x40
[ 5511.494329]  do_idle+0x1c7/0x250
[ 5511.507610]  cpu_startup_entry+0x19/0x20
[ 5511.521394]  start_kernel+0x649/0x66e
[ 5511.534626]  secondary_startup_64_no_verify+0xc3/0xcb
[ 5511.549230]  </TASK>

Detect such case during bind() and allocate this memory region via newly
introduced xp_alloc_tx_descs(). Also, use kvcalloc instead of kcalloc as
for other buffer pool allocations, so that it matches the kvfree() from
xp_destroy().

Fixes: d1bc532e99be ("i40e: xsk: Move tmp desc array from driver to pool")
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Magnus Karlsson <magnus.karlsson@intel.com>
Link: https://lore.kernel.org/bpf/20220425153745.481322-1-maciej.fijalkowski@intel.com
2 years agokprobes: Fix KRETPROBES when CONFIG_KRETPROBE_ON_RETHOOK is set
Adam Zabrocki [Fri, 22 Apr 2022 16:40:27 +0000 (18:40 +0200)]
kprobes: Fix KRETPROBES when CONFIG_KRETPROBE_ON_RETHOOK is set

The recent kernel change in 73f9b911faa7 ("kprobes: Use rethook for kretprobe
if possible"), introduced a potential NULL pointer dereference bug in the
KRETPROBE mechanism. The official Kprobes documentation defines that "Any or
all handlers can be NULL". Unfortunately, there is a missing return handler
verification to fulfill these requirements and can result in a NULL pointer
dereference bug.

This patch adds such verification in kretprobe_rethook_handler() function.

Fixes: 73f9b911faa7 ("kprobes: Use rethook for kretprobe if possible")
Signed-off-by: Adam Zabrocki <pi3@pi3.com.pl>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Naveen N. Rao <naveen.n.rao@linux.ibm.com>
Cc: Anil S. Keshavamurthy <anil.s.keshavamurthy@intel.com>
Link: https://lore.kernel.org/bpf/20220422164027.GA7862@pi3.com.pl
2 years agogfs2: Don't re-check for write past EOF unnecessarily
Andreas Gruenbacher [Tue, 19 Apr 2022 18:51:50 +0000 (20:51 +0200)]
gfs2: Don't re-check for write past EOF unnecessarily

Only re-check for direct I/O writes past the end of the file after
re-acquiring the inode glock.

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
2 years agodrm/sun4i: Remove obsolete references to PHYS_OFFSET
Samuel Holland [Sun, 24 Apr 2022 16:26:21 +0000 (11:26 -0500)]
drm/sun4i: Remove obsolete references to PHYS_OFFSET

commit b4bdc4fbf8d0 ("soc: sunxi: Deal with the MBUS DMA offsets in a
central place") added a platform device notifier that sets the DMA
offset for all of the display engine frontend and backend devices.

The code applying the offset to DMA buffer physical addresses was then
removed from the backend driver in commit 756668ba682e ("drm/sun4i:
backend: Remove the MBUS quirks"), but the code subtracting PHYS_OFFSET
was left in the frontend driver.

As a result, the offset was applied twice in the frontend driver. This
likely went unnoticed because it only affects specific configurations
(scaling or certain pixel formats) where the frontend is used, on boards
with both one of these older SoCs and more than 1 GB of DRAM.

In addition, the references to PHYS_OFFSET prevent compiling the driver
on architectures where PHYS_OFFSET is not defined.

Fixes: b4bdc4fbf8d0 ("soc: sunxi: Deal with the MBUS DMA offsets in a central place")
Reviewed-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Signed-off-by: Samuel Holland <samuel@sholland.org>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20220424162633.12369-4-samuel@sholland.org
2 years agovirtio_net: fix wrong buf address calculation when using xdp
Nikolay Aleksandrov [Mon, 25 Apr 2022 10:37:03 +0000 (13:37 +0300)]
virtio_net: fix wrong buf address calculation when using xdp

We received a report[1] of kernel crashes when Cilium is used in XDP
mode with virtio_net after updating to newer kernels. After
investigating the reason it turned out that when using mergeable bufs
with an XDP program which adjusts xdp.data or xdp.data_meta page_to_buf()
calculates the build_skb address wrong because the offset can become less
than the headroom so it gets the address of the previous page (-X bytes
depending on how lower offset is):
 page_to_skb: page addr ffff9eb2923e2000 buf ffff9eb2923e1ffc offset 252 headroom 256

This is a pr_err() I added in the beginning of page_to_skb which clearly
shows offset that is less than headroom by adding 4 bytes of metadata
via an xdp prog. The calculations done are:
 receive_mergeable():
 headroom = VIRTIO_XDP_HEADROOM; // VIRTIO_XDP_HEADROOM == 256 bytes
 offset = xdp.data - page_address(xdp_page) -
          vi->hdr_len - metasize;

 page_to_skb():
 p = page_address(page) + offset;
 ...
 buf = p - headroom;

Now buf goes -4 bytes from the page's starting address as can be seen
above which is set as skb->head and skb->data by build_skb later. Depending
on what's done with the skb (when it's freed most often) we get all kinds
of corruptions and BUG_ON() triggers in mm[2]. We have to recalculate
the new headroom after the xdp program has run, similar to how offset
and len are recalculated. Headroom is directly related to
data_hard_start, data and data_meta, so we use them to get the new size.
The result is correct (similar pr_err() in page_to_skb, one case of
xdp_page and one case of virtnet buf):
 a) Case with 4 bytes of metadata
 [  115.949641] page_to_skb: page addr ffff8b4dcfad2000 offset 252 headroom 252
 [  121.084105] page_to_skb: page addr ffff8b4dcf018000 offset 20732 headroom 252
 b) Case of pushing data +32 bytes
 [  153.181401] page_to_skb: page addr ffff8b4dd0c4d000 offset 288 headroom 288
 [  158.480421] page_to_skb: page addr ffff8b4dd00b0000 offset 24864 headroom 288
 c) Case of pushing data -33 bytes
 [  835.906830] page_to_skb: page addr ffff8b4dd3270000 offset 223 headroom 223
 [  840.839910] page_to_skb: page addr ffff8b4dcdd68000 offset 12511 headroom 223

Offset and headroom are equal because offset points to the start of
reserved bytes for the virtio_net header which are at buf start +
headroom, while data points at buf start + vnet hdr size + headroom so
when data or data_meta are adjusted by the xdp prog both the headroom size
and the offset change equally. We can use data_hard_start to compute the
new headroom after the xdp prog (linearized / page start case, the
virtnet buf case is similar just with bigger base offset):
 xdp.data_hard_start = page_address + vnet_hdr
 xdp.data = page_address + vnet_hdr + headroom
 new headroom after xdp prog = xdp.data - xdp.data_hard_start - metasize

An example reproducer xdp prog[3] is below.

[1] https://github.com/cilium/cilium/issues/19453

[2] Two of the many traces:
 [   40.437400] BUG: Bad page state in process swapper/0  pfn:14940
 [   40.916726] BUG: Bad page state in process systemd-resolve  pfn:053b7
 [   41.300891] kernel BUG at include/linux/mm.h:720!
 [   41.301801] invalid opcode: 0000 [#1] PREEMPT SMP NOPTI
 [   41.302784] CPU: 1 PID: 1181 Comm: kubelet Kdump: loaded Tainted: G    B   W         5.18.0-rc1+ #37
 [   41.304458] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.15.0-1.fc35 04/01/2014
 [   41.306018] RIP: 0010:page_frag_free+0x79/0xe0
 [   41.306836] Code: 00 00 75 ea 48 8b 07 a9 00 00 01 00 74 e0 48 8b 47 48 48 8d 50 ff a8 01 48 0f 45 fa eb d0 48 c7 c6 18 b8 30 a6 e8 d7 f8 fc ff <0f> 0b 48 8d 78 ff eb bc 48 8b 07 a9 00 00 01 00 74 3a 66 90 0f b6
 [   41.310235] RSP: 0018:ffffac05c2a6bc78 EFLAGS: 00010292
 [   41.311201] RAX: 000000000000003e RBX: 0000000000000000 RCX: 0000000000000000
 [   41.312502] RDX: 0000000000000001 RSI: ffffffffa6423004 RDI: 00000000ffffffff
 [   41.313794] RBP: ffff993c98823600 R08: 0000000000000000 R09: 00000000ffffdfff
 [   41.315089] R10: ffffac05c2a6ba68 R11: ffffffffa698ca28 R12: ffff993c98823600
 [   41.316398] R13: ffff993c86311ebc R14: 0000000000000000 R15: 000000000000005c
 [   41.317700] FS:  00007fe13fc56740(0000) GS:ffff993cdd900000(0000) knlGS:0000000000000000
 [   41.319150] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 [   41.320152] CR2: 000000c00008a000 CR3: 0000000014908000 CR4: 0000000000350ee0
 [   41.321387] Call Trace:
 [   41.321819]  <TASK>
 [   41.322193]  skb_release_data+0x13f/0x1c0
 [   41.322902]  __kfree_skb+0x20/0x30
 [   41.343870]  tcp_recvmsg_locked+0x671/0x880
 [   41.363764]  tcp_recvmsg+0x5e/0x1c0
 [   41.384102]  inet_recvmsg+0x42/0x100
 [   41.406783]  ? sock_recvmsg+0x1d/0x70
 [   41.428201]  sock_read_iter+0x84/0xd0
 [   41.445592]  ? 0xffffffffa3000000
 [   41.462442]  new_sync_read+0x148/0x160
 [   41.479314]  ? 0xffffffffa3000000
 [   41.496937]  vfs_read+0x138/0x190
 [   41.517198]  ksys_read+0x87/0xc0
 [   41.535336]  do_syscall_64+0x3b/0x90
 [   41.551637]  entry_SYSCALL_64_after_hwframe+0x44/0xae
 [   41.568050] RIP: 0033:0x48765b
 [   41.583955] Code: e8 4a 35 fe ff eb 88 cc cc cc cc cc cc cc cc e8 fb 7a fe ff 48 8b 7c 24 10 48 8b 74 24 18 48 8b 54 24 20 48 8b 44 24 08 0f 05 <48> 3d 01 f0 ff ff 76 20 48 c7 44 24 28 ff ff ff ff 48 c7 44 24 30
 [   41.632818] RSP: 002b:000000c000a2f5b8 EFLAGS: 00000212 ORIG_RAX: 0000000000000000
 [   41.664588] RAX: ffffffffffffffda RBX: 000000c000062000 RCX: 000000000048765b
 [   41.681205] RDX: 0000000000005e54 RSI: 000000c000e66000 RDI: 0000000000000016
 [   41.697164] RBP: 000000c000a2f608 R08: 0000000000000001 R09: 00000000000001b4
 [   41.713034] R10: 00000000000000b6 R11: 0000000000000212 R12: 00000000000000e9
 [   41.728755] R13: 0000000000000001 R14: 000000c000a92000 R15: ffffffffffffffff
 [   41.744254]  </TASK>
 [   41.758585] Modules linked in: br_netfilter bridge veth netconsole virtio_net

 and

 [   33.524802] BUG: Bad page state in process systemd-network  pfn:11e60
 [   33.528617] page ffffe05dc0147b00 ffffe05dc04e7a00 ffff8ae9851ec000 (1) len 82 offset 252 metasize 4 hroom 0 hdr_len 12 data ffff8ae9851ec10c data_meta ffff8ae9851ec108 data_end ffff8ae9851ec14e
 [   33.529764] page:000000003792b5ba refcount:0 mapcount:-512 mapping:0000000000000000 index:0x0 pfn:0x11e60
 [   33.532463] flags: 0xfffffc0000000(node=0|zone=1|lastcpupid=0x1fffff)
 [   33.532468] raw: 000fffffc0000000 0000000000000000 dead000000000122 0000000000000000
 [   33.532470] raw: 0000000000000000 0000000000000000 00000000fffffdff 0000000000000000
 [   33.532471] page dumped because: nonzero mapcount
 [   33.532472] Modules linked in: br_netfilter bridge veth netconsole virtio_net
 [   33.532479] CPU: 0 PID: 791 Comm: systemd-network Kdump: loaded Not tainted 5.18.0-rc1+ #37
 [   33.532482] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.15.0-1.fc35 04/01/2014
 [   33.532484] Call Trace:
 [   33.532496]  <TASK>
 [   33.532500]  dump_stack_lvl+0x45/0x5a
 [   33.532506]  bad_page.cold+0x63/0x94
 [   33.532510]  free_pcp_prepare+0x290/0x420
 [   33.532515]  free_unref_page+0x1b/0x100
 [   33.532518]  skb_release_data+0x13f/0x1c0
 [   33.532524]  kfree_skb_reason+0x3e/0xc0
 [   33.532527]  ip6_mc_input+0x23c/0x2b0
 [   33.532531]  ip6_sublist_rcv_finish+0x83/0x90
 [   33.532534]  ip6_sublist_rcv+0x22b/0x2b0

[3] XDP program to reproduce(xdp_pass.c):
 #include <linux/bpf.h>
 #include <bpf/bpf_helpers.h>

 SEC("xdp_pass")
 int xdp_pkt_pass(struct xdp_md *ctx)
 {
          bpf_xdp_adjust_head(ctx, -(int)32);
          return XDP_PASS;
 }

 char _license[] SEC("license") = "GPL";

 compile: clang -O2 -g -Wall -target bpf -c xdp_pass.c -o xdp_pass.o
 load on virtio_net: ip link set enp1s0 xdpdrv obj xdp_pass.o sec xdp_pass

CC: stable@vger.kernel.org
CC: Jason Wang <jasowang@redhat.com>
CC: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
CC: Daniel Borkmann <daniel@iogearbox.net>
CC: "Michael S. Tsirkin" <mst@redhat.com>
CC: virtualization@lists.linux-foundation.org
Fixes: 8fb7da9e9907 ("virtio_net: get build_skb() buf by data ptr")
Signed-off-by: Nikolay Aleksandrov <razor@blackwall.org>
Reviewed-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Link: https://lore.kernel.org/r/20220425103703.3067292-1-razor@blackwall.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agonet: dsa: mv88e6xxx: Fix port_hidden_wait to account for port_base_addr
Nathan Rossi [Mon, 25 Apr 2022 07:04:54 +0000 (07:04 +0000)]
net: dsa: mv88e6xxx: Fix port_hidden_wait to account for port_base_addr

The other port_hidden functions rely on the port_read/port_write
functions to access the hidden control port. These functions apply the
offset for port_base_addr where applicable. Update port_hidden_wait to
use the port_wait_bit so that port_base_addr offsets are accounted for
when waiting for the busy bit to change.

Without the offset the port_hidden_wait function would timeout on
devices that have a non-zero port_base_addr (e.g. MV88E6141), however
devices that have a zero port_base_addr would operate correctly (e.g.
MV88E6390).

Fixes: 609070133aff ("net: dsa: mv88e6xxx: update code operating on hidden registers")
Signed-off-by: Nathan Rossi <nathan@nathanrossi.com>
Reviewed-by: Marek Behún <kabel@kernel.org>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/20220425070454.348584-1-nathan@nathanrossi.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agonet: phy: marvell10g: fix return value on error
Baruch Siach [Mon, 25 Apr 2022 06:27:38 +0000 (09:27 +0300)]
net: phy: marvell10g: fix return value on error

Return back the error value that we get from phy_read_mmd().

Fixes: c84786fa8f91 ("net: phy: marvell10g: read copper results from CSSR1")
Signed-off-by: Baruch Siach <baruch.siach@siklu.com>
Reviewed-by: Marek Behún <kabel@kernel.org>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://lore.kernel.org/r/f47cb031aeae873bb008ba35001607304a171a20.1650868058.git.baruch@tkos.co.il
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agonet: bcmgenet: hide status block before TX timestamping
Jonathan Lemon [Sun, 24 Apr 2022 16:53:07 +0000 (09:53 -0700)]
net: bcmgenet: hide status block before TX timestamping

The hardware checksum offloading requires use of a transmit
status block inserted before the outgoing frame data, this was
updated in '9a9ba2a4aaaa ("net: bcmgenet: always enable status blocks")'

However, skb_tx_timestamp() assumes that it is passed a raw frame
and PTP parsing chokes on this status block.

Fix this by calling __skb_pull(), which hides the TSB before calling
skb_tx_timestamp(), so an outgoing PTP packet is parsed correctly.

As the data in the skb has already been set up for DMA, and the
dma_unmap_* calls use a separately stored address, there is no
no effective change in the data transmission.

Signed-off-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20220424165307.591145-1-jonathan.lemon@gmail.com
Fixes: d03825fba459 ("net: bcmgenet: add skb_tx_timestamp call")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agomctp: defer the kfree of object mdev->addrs
Lin Ma [Fri, 22 Apr 2022 11:43:40 +0000 (19:43 +0800)]
mctp: defer the kfree of object mdev->addrs

The function mctp_unregister() reclaims the device's relevant resource
when a netcard detaches. However, a running routine may be unaware of
this and cause the use-after-free of the mdev->addrs object.

The race condition can be demonstrated below

 cleanup thread               another thread
                          |
unregister_netdev()       |  mctp_sendmsg()
...                       |    ...
  mctp_unregister()       |    rt = mctp_route_lookup()
    ...                   |    mctl_local_output()
    kfree(mdev->addrs)    |      ...
                          |      saddr = rt->dev->addrs[0];
                          |

An attacker can adopt the (recent provided) mtcpserial driver with pty
to fake the device detaching and use the userfaultfd to increase the
race success chance (in mctp_sendmsg). The KASan report for such a POC
is shown below:

[   86.051955] ==================================================================
[   86.051955] BUG: KASAN: use-after-free in mctp_local_output+0x4e9/0xb7d
[   86.051955] Read of size 1 at addr ffff888005f298c0 by task poc/295
[   86.051955]
[   86.051955] Call Trace:
[   86.051955]  <TASK>
[   86.051955]  dump_stack_lvl+0x33/0x42
[   86.051955]  print_report.cold.13+0xb2/0x6b3
[   86.051955]  ? preempt_schedule_irq+0x57/0x80
[   86.051955]  ? mctp_local_output+0x4e9/0xb7d
[   86.051955]  kasan_report+0xa5/0x120
[   86.051955]  ? mctp_local_output+0x4e9/0xb7d
[   86.051955]  mctp_local_output+0x4e9/0xb7d
[   86.051955]  ? mctp_dev_set_key+0x79/0x79
[   86.051955]  ? copyin+0x38/0x50
[   86.051955]  ? _copy_from_iter+0x1b6/0xf20
[   86.051955]  ? sysvec_apic_timer_interrupt+0x97/0xb0
[   86.051955]  ? asm_sysvec_apic_timer_interrupt+0x12/0x20
[   86.051955]  ? mctp_local_output+0x1/0xb7d
[   86.051955]  mctp_sendmsg+0x64d/0xdb0
[   86.051955]  ? mctp_sk_close+0x20/0x20
[   86.051955]  ? __fget_light+0x2fd/0x4f0
[   86.051955]  ? mctp_sk_close+0x20/0x20
[   86.051955]  sock_sendmsg+0xdd/0x110
[   86.051955]  __sys_sendto+0x1cc/0x2a0
[   86.051955]  ? __ia32_sys_getpeername+0xa0/0xa0
[   86.051955]  ? new_sync_write+0x335/0x550
[   86.051955]  ? alloc_file+0x22f/0x500
[   86.051955]  ? __ip_do_redirect+0x820/0x1820
[   86.051955]  ? vfs_write+0x44d/0x7b0
[   86.051955]  ? vfs_write+0x44d/0x7b0
[   86.051955]  ? fput_many+0x15/0x120
[   86.051955]  ? ksys_write+0x155/0x1b0
[   86.051955]  ? __ia32_sys_read+0xa0/0xa0
[   86.051955]  __x64_sys_sendto+0xd8/0x1b0
[   86.051955]  ? exit_to_user_mode_prepare+0x2f/0x120
[   86.051955]  ? syscall_exit_to_user_mode+0x12/0x20
[   86.051955]  do_syscall_64+0x3a/0x80
[   86.051955]  entry_SYSCALL_64_after_hwframe+0x44/0xae
[   86.051955] RIP: 0033:0x7f82118a56b3
[   86.051955] RSP: 002b:00007ffdb154b110 EFLAGS: 00000293 ORIG_RAX: 000000000000002c
[   86.051955] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f82118a56b3
[   86.051955] RDX: 0000000000000010 RSI: 00007f8211cd4000 RDI: 0000000000000007
[   86.051955] RBP: 00007ffdb154c1d0 R08: 00007ffdb154b164 R09: 000000000000000c
[   86.051955] R10: 0000000000000000 R11: 0000000000000293 R12: 000055d779800db0
[   86.051955] R13: 00007ffdb154c2b0 R14: 0000000000000000 R15: 0000000000000000
[   86.051955]  </TASK>
[   86.051955]
[   86.051955] Allocated by task 295:
[   86.051955]  kasan_save_stack+0x1c/0x40
[   86.051955]  __kasan_kmalloc+0x84/0xa0
[   86.051955]  mctp_rtm_newaddr+0x242/0x610
[   86.051955]  rtnetlink_rcv_msg+0x2fd/0x8b0
[   86.051955]  netlink_rcv_skb+0x11c/0x340
[   86.051955]  netlink_unicast+0x439/0x630
[   86.051955]  netlink_sendmsg+0x752/0xc00
[   86.051955]  sock_sendmsg+0xdd/0x110
[   86.051955]  __sys_sendto+0x1cc/0x2a0
[   86.051955]  __x64_sys_sendto+0xd8/0x1b0
[   86.051955]  do_syscall_64+0x3a/0x80
[   86.051955]  entry_SYSCALL_64_after_hwframe+0x44/0xae
[   86.051955]
[   86.051955] Freed by task 301:
[   86.051955]  kasan_save_stack+0x1c/0x40
[   86.051955]  kasan_set_track+0x21/0x30
[   86.051955]  kasan_set_free_info+0x20/0x30
[   86.051955]  __kasan_slab_free+0x104/0x170
[   86.051955]  kfree+0x8c/0x290
[   86.051955]  mctp_dev_notify+0x161/0x2c0
[   86.051955]  raw_notifier_call_chain+0x8b/0xc0
[   86.051955]  unregister_netdevice_many+0x299/0x1180
[   86.051955]  unregister_netdevice_queue+0x210/0x2f0
[   86.051955]  unregister_netdev+0x13/0x20
[   86.051955]  mctp_serial_close+0x6d/0xa0
[   86.051955]  tty_ldisc_kill+0x31/0xa0
[   86.051955]  tty_ldisc_hangup+0x24f/0x560
[   86.051955]  __tty_hangup.part.28+0x2ce/0x6b0
[   86.051955]  tty_release+0x327/0xc70
[   86.051955]  __fput+0x1df/0x8b0
[   86.051955]  task_work_run+0xca/0x150
[   86.051955]  exit_to_user_mode_prepare+0x114/0x120
[   86.051955]  syscall_exit_to_user_mode+0x12/0x20
[   86.051955]  do_syscall_64+0x46/0x80
[   86.051955]  entry_SYSCALL_64_after_hwframe+0x44/0xae
[   86.051955]
[   86.051955] The buggy address belongs to the object at ffff888005f298c0
[   86.051955]  which belongs to the cache kmalloc-8 of size 8
[   86.051955] The buggy address is located 0 bytes inside of
[   86.051955]  8-byte region [ffff888005f298c0ffff888005f298c8)
[   86.051955]
[   86.051955] The buggy address belongs to the physical page:
[   86.051955] flags: 0x100000000000200(slab|node=0|zone=1)
[   86.051955] raw: 0100000000000200 dead000000000100 dead000000000122 ffff888005c42280
[   86.051955] raw: 0000000000000000 0000000080660066 00000001ffffffff 0000000000000000
[   86.051955] page dumped because: kasan: bad access detected
[   86.051955]
[   86.051955] Memory state around the buggy address:
[   86.051955]  ffff888005f29780: 00 fc fc fc fc 00 fc fc fc fc 00 fc fc fc fc 00
[   86.051955]  ffff888005f29800: fc fc fc fc 00 fc fc fc fc 00 fc fc fc fc 00 fc
[   86.051955] >ffff888005f29880: fc fc fc fb fc fc fc fc fa fc fc fc fc fa fc fc
[   86.051955]                                            ^
[   86.051955]  ffff888005f29900: fc fc 00 fc fc fc fc 00 fc fc fc fc 00 fc fc fc
[   86.051955]  ffff888005f29980: fc 00 fc fc fc fc 00 fc fc fc fc 00 fc fc fc fc
[   86.051955] ==================================================================

To this end, just like the commit e04480920d1e ("Bluetooth: defer
cleanup of resources in hci_unregister_dev()")  this patch defers the
destructive kfree(mdev->addrs) in mctp_unregister to the mctp_dev_put,
where the refcount of mdev is zero and the entire device is reclaimed.
This prevents the use-after-free because the sendmsg thread holds the
reference of mdev in the mctp_route object.

Fixes: 583be982d934 (mctp: Add device handling and netlink interface)
Signed-off-by: Lin Ma <linma@zju.edu.cn>
Acked-by: Jeremy Kerr <jk@codeconstruct.com.au>
Link: https://lore.kernel.org/r/20220422114340.32346-1-linma@zju.edu.cn
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agodrm/i915/fbc: Consult hw.crtc instead of uapi.crtc
Ville Syrjälä [Wed, 13 Apr 2022 15:28:51 +0000 (18:28 +0300)]
drm/i915/fbc: Consult hw.crtc instead of uapi.crtc

plane_state->uapi.crtc is not what we want to be looking at.
If bigjoiner is used hw.crtc is what tells us what crtc the plane
is supposedly using.

Not an actual problem on current hardware as the only FBC capable
pipe (A) can't be a bigjoiner slave and thus uapi.crtc==hw.crtc
always here. But when we get more FBC instances this will become
actually important.

Fixes: 2e6c99f88679 ("drm/i915/fbc: Nuke lots of crap from intel_fbc_state_cache")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220413152852.7336-1-ville.syrjala@linux.intel.com
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>
(cherry picked from commit 3e1faae3398789abe8d4797255bfe28d95d81308)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
2 years agodrm/i915: Fix SEL_FETCH_PLANE_*(PIPE_B+) register addresses
Imre Deak [Thu, 21 Apr 2022 16:22:21 +0000 (19:22 +0300)]
drm/i915: Fix SEL_FETCH_PLANE_*(PIPE_B+) register addresses

Fix typo in the _SEL_FETCH_PLANE_BASE_1_B register base address.

Fixes: a5523e2ff074a5 ("drm/i915: Add PSR2 selective fetch registers")
References: https://gitlab.freedesktop.org/drm/intel/-/issues/5400
Cc: José Roberto de Souza <jose.souza@intel.com>
Cc: <stable@vger.kernel.org> # v5.9+
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220421162221.2261895-1-imre.deak@intel.com
(cherry picked from commit af2cbc6ef967f61711a3c40fca5366ea0bc7fecc)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
2 years agocpufreq: qcom-cpufreq-hw: Clear dcvs interrupts
Vladimir Zapolskiy [Thu, 7 Apr 2022 20:09:19 +0000 (23:09 +0300)]
cpufreq: qcom-cpufreq-hw: Clear dcvs interrupts

It's noted that dcvs interrupts are not self-clearing, thus an interrupt
handler runs constantly, which leads to a severe regression in runtime.
To fix the problem an explicit write to clear interrupt register is
required, note that on OSM platforms the register may not be present.

Fixes: 275157b367f4 ("cpufreq: qcom-cpufreq-hw: Add dcvs interrupt support")
Signed-off-by: Vladimir Zapolskiy <vladimir.zapolskiy@linaro.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
2 years agovideo: fbdev: clps711x-fb: Use syscon_regmap_lookup_by_phandle
Alexander Shiyan [Wed, 20 Apr 2022 07:06:39 +0000 (10:06 +0300)]
video: fbdev: clps711x-fb: Use syscon_regmap_lookup_by_phandle

Since version 5.13, the standard syscon bindings have been added
to all clps711x DT nodes, so we can now use the more general
syscon_regmap_lookup_by_phandle function to get the syscon pointer.

Signed-off-by: Alexander Shiyan <eagle.alexander923@gmail.com>
Signed-off-by: Helge Deller <deller@gmx.de>
2 years agoMerge branch 'net-smc-two-fixes-for-smc-fallback'
Jakub Kicinski [Mon, 25 Apr 2022 18:03:51 +0000 (11:03 -0700)]
Merge branch 'net-smc-two-fixes-for-smc-fallback'

Wen Gu says:

====================
net/smc: Two fixes for smc fallback

This patch set includes two fixes for smc fallback:

Patch 1/2 introduces some simple helpers to wrap the replacement
and restore of clcsock's callback functions. Make sure that only
the original callbacks will be saved and not overwritten.

Patch 2/2 fixes a syzbot reporting slab-out-of-bound issue where
smc_fback_error_report() accesses the already freed smc sock (see
https://lore.kernel.org/r/00000000000013ca8105d7ae3ada@google.com/).
The patch fixes it by resetting sk_user_data and restoring clcsock
callback functions timely in fallback situation.

But it should be noted that although patch 2/2 can fix the issue
of 'slab-out-of-bounds/use-after-free in smc_fback_error_report',
it can't pass the syzbot reproducer test. Because after applying
these two patches in upstream, syzbot reproducer triggered another
known issue like this:

==================================================================
BUG: KASAN: use-after-free in tcp_retransmit_timer+0x2ef3/0x3360 net/ipv4/tcp_timer.c:511
Read of size 8 at addr ffff888020328380 by task udevd/4158

CPU: 1 PID: 4158 Comm: udevd Not tainted 5.18.0-rc3-syzkaller-00074-gb05a5683eba6-dirty #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
 <IRQ>
  __dump_stack lib/dump_stack.c:88 [inline]
  dump_stack_lvl+0xcd/0x134 lib/dump_stack.c:106
  print_address_description.constprop.0.cold+0xeb/0x467 mm/kasan/report.c:313
  print_report mm/kasan/report.c:429 [inline]
  kasan_report.cold+0xf4/0x1c6 mm/kasan/report.c:491
  tcp_retransmit_timer+0x2ef3/0x3360 net/ipv4/tcp_timer.c:511
  tcp_write_timer_handler+0x5e6/0xbc0 net/ipv4/tcp_timer.c:622
  tcp_write_timer+0xa2/0x2b0 net/ipv4/tcp_timer.c:642
  call_timer_fn+0x1a5/0x6b0 kernel/time/timer.c:1421
  expire_timers kernel/time/timer.c:1466 [inline]
  __run_timers.part.0+0x679/0xa80 kernel/time/timer.c:1737
  __run_timers kernel/time/timer.c:1715 [inline]
  run_timer_softirq+0xb3/0x1d0 kernel/time/timer.c:1750
  __do_softirq+0x29b/0x9c2 kernel/softirq.c:558
  invoke_softirq kernel/softirq.c:432 [inline]
  __irq_exit_rcu+0x123/0x180 kernel/softirq.c:637
  irq_exit_rcu+0x5/0x20 kernel/softirq.c:649
  sysvec_apic_timer_interrupt+0x93/0xc0 arch/x86/kernel/apic/apic.c:1097
 </IRQ>
 ...
(detail report can be found in https://syzkaller.appspot.com/text?tag=CrashReport&x=15406b44f00000)

IMHO, the above issue is the same as this known one: https://syzkaller.appspot.com/bug?extid=694120e1002c117747ed,
and it doesn't seem to be related with SMC. The discussion about this known issue is ongoing and can be found in
https://lore.kernel.org/bpf/000000000000f75af905d3ba0716@google.com/T/.

And I added the temporary solution mentioned in the above discussion on
top of my two patches, the syzbot reproducer of 'slab-out-of-bounds/
use-after-free in smc_fback_error_report' no longer triggers any issue.
====================

Link: https://lore.kernel.org/r/1650614179-11529-1-git-send-email-guwen@linux.alibaba.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet/smc: Fix slab-out-of-bounds issue in fallback
Wen Gu [Fri, 22 Apr 2022 07:56:19 +0000 (15:56 +0800)]
net/smc: Fix slab-out-of-bounds issue in fallback

syzbot reported a slab-out-of-bounds/use-after-free issue,
which was caused by accessing an already freed smc sock in
fallback-specific callback functions of clcsock.

This patch fixes the issue by restoring fallback-specific
callback functions to original ones and resetting clcsock
sk_user_data to NULL before freeing smc sock.

Meanwhile, this patch introduces sk_callback_lock to make
the access and assignment to sk_user_data mutually exclusive.

Reported-by: syzbot+b425899ed22c6943e00b@syzkaller.appspotmail.com
Fixes: 341adeec9ada ("net/smc: Forward wakeup to smc socket waitqueue after fallback")
Link: https://lore.kernel.org/r/00000000000013ca8105d7ae3ada@google.com/
Signed-off-by: Wen Gu <guwen@linux.alibaba.com>
Acked-by: Karsten Graul <kgraul@linux.ibm.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet/smc: Only save the original clcsock callback functions
Wen Gu [Fri, 22 Apr 2022 07:56:18 +0000 (15:56 +0800)]
net/smc: Only save the original clcsock callback functions

Both listen and fallback process will save the current clcsock
callback functions and establish new ones. But if both of them
happen, the saved callback functions will be overwritten.

So this patch introduces some helpers to ensure that only save
the original callback functions of clcsock.

Fixes: 341adeec9ada ("net/smc: Forward wakeup to smc socket waitqueue after fallback")
Signed-off-by: Wen Gu <guwen@linux.alibaba.com>
Acked-by: Karsten Graul <kgraul@linux.ibm.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge tag 'f2fs-fix-5.18' of git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk...
Linus Torvalds [Mon, 25 Apr 2022 17:53:56 +0000 (10:53 -0700)]
Merge tag 'f2fs-fix-5.18' of git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs

Pull f2fs fixes from Jaegeuk Kim:
 "This includes major bug fixes introduced in 5.18-rc1 and 5.17+:

   - Remove obsolete whint_mode (5.18-rc1)

   - Fix IO split issue caused by op_flags change in f2fs (5.18-rc1)

   - Fix a wrong condition check to detect IO failure loop (5.18-rc1)

   - Fix wrong data truncation during roll-forward (5.17+)"

* tag 'f2fs-fix-5.18' of git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs:
  f2fs: should not truncate blocks during roll-forward recovery
  f2fs: fix wrong condition check when failing metapage read
  f2fs: keep io_flags to avoid IO split due to different op_flags in two fio holders
  f2fs: remove obsolete whint_mode