platform/kernel/linux-starfive.git
2 years agoNFS: Don't request readdirplus when revalidation was forced
Trond Myklebust [Sat, 19 Feb 2022 15:06:05 +0000 (10:06 -0500)]
NFS: Don't request readdirplus when revalidation was forced

If the revalidation was forced, due to the presence of a LOOKUP_EXCL or
a LOOKUP_REVAL flag, then readdirplus won't help. It also can't help
when we're doing a path component lookup.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Readdirplus can't help lookup for case insensitive filesystems
Trond Myklebust [Sat, 19 Feb 2022 14:56:45 +0000 (09:56 -0500)]
NFS: Readdirplus can't help lookup for case insensitive filesystems

If the filesystem is case insensitive, then readdirplus can't help with
cache misses, since it won't return case folded variants of the filename.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFSv4: Ask for a full XDR buffer of readdir goodness
Trond Myklebust [Fri, 18 Feb 2022 17:04:06 +0000 (12:04 -0500)]
NFSv4: Ask for a full XDR buffer of readdir goodness

Instead of pretending that we know the ratio of directory info vs
readdirplus attribute info, just set the 'dircount' field to the same
value as the 'maxcount' field.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Don't ask for readdirplus unless it can help nfs_getattr()
Trond Myklebust [Thu, 17 Feb 2022 20:46:23 +0000 (15:46 -0500)]
NFS: Don't ask for readdirplus unless it can help nfs_getattr()

If attribute caching is turned off, then use of readdirplus is not going
to help stat() performance.
Readdirplus also doesn't help if a file is being written to, since we
will have to flush those writes in order to sync the mtime/ctime.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Improve heuristic for readdirplus
Trond Myklebust [Thu, 17 Feb 2022 16:08:24 +0000 (11:08 -0500)]
NFS: Improve heuristic for readdirplus

The heuristic for readdirplus is designed to try to detect 'ls -l' and
similar patterns. It does so by looking for cache hit/miss patterns in
both the attribute cache and in the dcache of the files in a given
directory, and then sets a flag for the readdirplus code to interpret.

The problem with this approach is that a single attribute or dcache miss
can cause the NFS code to force a refresh of the attributes for the
entire set of files contained in the directory.

To be able to make a more nuanced decision, let's sample the number of
hits and misses in the set of open directory descriptors. That allows us
to set thresholds at which we start preferring READDIRPLUS over regular
READDIR, or at which we start to force a re-read of the remaining
readdir cache using READDIRPLUS.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Reduce use of uncached readdir
Trond Myklebust [Thu, 17 Feb 2022 18:02:37 +0000 (13:02 -0500)]
NFS: Reduce use of uncached readdir

When reading a very large directory, we want to try to keep the page
cache up to date if doing so is inexpensive. With the change to allow
readdir to continue reading even when the cache is incomplete, we no
longer need to fall back to uncached readdir in order to scale to large
directories.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Simplify nfs_readdir_xdr_to_array()
Trond Myklebust [Mon, 7 Feb 2022 20:07:01 +0000 (15:07 -0500)]
NFS: Simplify nfs_readdir_xdr_to_array()

Recent changes to readdir mean that we can cope with partially filled
page cache entries, so we no longer need to rely on looping in
nfs_readdir_xdr_to_array().

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: If the cookie verifier changes, we must invalidate the page cache
Trond Myklebust [Tue, 22 Feb 2022 21:23:12 +0000 (16:23 -0500)]
NFS: If the cookie verifier changes, we must invalidate the page cache

Ensure that if the cookie verifier changes when we use the zero-valued
cookie, then we invalidate any cached pages.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Adjust the amount of readahead performed by NFS readdir
Trond Myklebust [Mon, 7 Feb 2022 18:37:00 +0000 (13:37 -0500)]
NFS: Adjust the amount of readahead performed by NFS readdir

The current NFS readdir code will always try to maximise the amount of
readahead it performs on the assumption that we can cache anything that
isn't immediately read by the process.
There are several cases where this assumption breaks down, including
when the 'ls -l' heuristic kicks in to try to force use of readdirplus
as a batch replacement for lookup/getattr.

This patch therefore tries to tone down the amount of readahead we
perform, and adjust it to try to match the amount of data being
requested by user space.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Don't advance the page pointer unless the page is full
Trond Myklebust [Sat, 26 Feb 2022 14:38:19 +0000 (09:38 -0500)]
NFS: Don't advance the page pointer unless the page is full

When we hit the end of the data in the readdir page, we don't want to
start filling a new page, unless this one is full.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Don't re-read the entire page cache to find the next cookie
Trond Myklebust [Tue, 22 Feb 2022 13:59:33 +0000 (08:59 -0500)]
NFS: Don't re-read the entire page cache to find the next cookie

If the page cache entry that was last read gets invalidated for some
reason, then make sure we can re-create it on the next call to readdir.
This, combined with the cache page validation, allows us to reuse the
cached value of page-index on successive calls to nfs_readdir.

Credit is due to Benjamin Coddington for showing that the concept works,
and that it allows for improved cache sharing between processes even in
the case where pages are lost due to LRU or active invalidation.

Suggested-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Store the change attribute in the directory page cache
Trond Myklebust [Tue, 22 Feb 2022 13:31:28 +0000 (08:31 -0500)]
NFS: Store the change attribute in the directory page cache

Use the change attribute and the first cookie in a directory page cache
entry to validate that the page is up to date.

Suggested-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Calculate page offsets algorithmically
Trond Myklebust [Tue, 22 Feb 2022 15:39:26 +0000 (10:39 -0500)]
NFS: Calculate page offsets algorithmically

Instead of relying on counting the page offsets as we walk through the
page cache, switch to calculating them algorithmically.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Use kzalloc() to avoid initialising the nfs_open_dir_context
Trond Myklebust [Tue, 22 Feb 2022 17:10:36 +0000 (12:10 -0500)]
NFS: Use kzalloc() to avoid initialising the nfs_open_dir_context

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Initialise the readdir verifier as best we can in nfs_opendir()
Trond Myklebust [Fri, 25 Feb 2022 15:22:30 +0000 (10:22 -0500)]
NFS: Initialise the readdir verifier as best we can in nfs_opendir()

For the purpose of ensuring that opendir() followed by seekdir() work as
correctly as possible, try to initialise the readdir verifier in
nfs_opendir().

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Trace lookup revalidation failure
Trond Myklebust [Sun, 20 Feb 2022 01:38:19 +0000 (20:38 -0500)]
NFS: Trace lookup revalidation failure

Enable tracing of lookup revalidation failures.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: constify nfs_server_capable() and nfs_have_writebacks()
Trond Myklebust [Fri, 18 Feb 2022 12:07:08 +0000 (07:07 -0500)]
NFS: constify nfs_server_capable() and nfs_have_writebacks()

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Return valid errors from nfs2/3_decode_dirent()
Trond Myklebust [Thu, 24 Feb 2022 15:59:37 +0000 (10:59 -0500)]
NFS: Return valid errors from nfs2/3_decode_dirent()

Valid return values for decode_dirent() callback functions are:
 0: Success
 -EBADCOOKIE: End of directory
 -EAGAIN: End of xdr_stream

All errors need to map into one of those three values.

Fixes: 573c4e1ef53a ("NFS: Simplify ->decode_dirent() calling sequence")
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoRevert "NFSv4: use unique client identifiers in network namespaces"
Trond Myklebust [Mon, 28 Feb 2022 15:09:23 +0000 (10:09 -0500)]
Revert "NFSv4: use unique client identifiers in network namespaces"

This reverts commit 50c790a0b69bdc420f00f30bdf348d6c90194c78.

The functionality is believed to be capable of causing regressions in
existing setups, so the author has requested that it be reverted.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Use of mapping_set_error() results in spurious errors
Trond Myklebust [Tue, 15 Feb 2022 20:58:38 +0000 (15:58 -0500)]
NFS: Use of mapping_set_error() results in spurious errors

The use of mapping_set_error() in conjunction with calls to
filemap_check_errors() is problematic because every error gets reported
as either an EIO or an ENOSPC by filemap_check_errors() in functions
such as filemap_write_and_wait() or filemap_write_and_wait_range().
In almost all cases, we prefer to use the more nuanced wb errors.

Fixes: b8946d7bfb94 ("NFS: Revalidate the file mapping on all fatal writeback errors")
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Clean up NFSv4.2 xattrs
Trond Myklebust [Wed, 23 Feb 2022 20:43:26 +0000 (15:43 -0500)]
NFS: Clean up NFSv4.2 xattrs

Add a helper for the xattr mask so that we can get rid of the inlined
ifdefs.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Remove unnecessary XATTR cache invalidation in nfs_fhget()
Trond Myklebust [Wed, 23 Feb 2022 20:46:20 +0000 (15:46 -0500)]
NFS: Remove unnecessary XATTR cache invalidation in nfs_fhget()

We should never expect the 'xattr_cache' to be non-null in that case,
hence nfs_set_cache_invalid() is just going to optimise it away.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: NFSv2/v3 clients should never be setting NFS_CAP_XATTR
Trond Myklebust [Tue, 22 Feb 2022 23:20:38 +0000 (18:20 -0500)]
NFS: NFSv2/v3 clients should never be setting NFS_CAP_XATTR

Ensure that we always initialise the 'xattr_support' field in struct
nfs_fsinfo, so that nfs_server_set_fsinfo() doesn't declare our NFSv2/v3
client to be capable of supporting the NFSv4.2 xattr protocol by setting
the NFS_CAP_XATTR capability.

This configuration can cause nfs_do_access() to set access mode bits
that are unsupported by the NFSv3 ACCESS call, which may confuse
spec-compliant servers.

Reported-by: Olga Kornievskaia <kolga@netapp.com>
Fixes: b78ef845c35d ("NFSv4.2: query the server for extended attribute support")
Cc: stable@vger.kernel.org
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Remove unused flag NFS_INO_REVAL_PAGECACHE
Trond Myklebust [Wed, 9 Feb 2022 18:26:19 +0000 (13:26 -0500)]
NFS: Remove unused flag NFS_INO_REVAL_PAGECACHE

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Replace last uses of NFS_INO_REVAL_PAGECACHE
Trond Myklebust [Wed, 9 Feb 2022 18:22:48 +0000 (13:22 -0500)]
NFS: Replace last uses of NFS_INO_REVAL_PAGECACHE

Now that we have more fine grained attribute revalidation, let's just
get rid of NFS_INO_REVAL_PAGECACHE.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFSv4: use unique client identifiers in network namespaces
Benjamin Coddington [Wed, 9 Feb 2022 14:07:01 +0000 (09:07 -0500)]
NFSv4: use unique client identifiers in network namespaces

In order to differentiate client state, assign a random uuid to the
uniquifing portion of the client identifier when a network namespace is
created.  Containers may still override this value if they wish to maintain
stable client identifiers by writing to /sys/fs/nfs/net/client/identifier,
either by udev rules or other means.

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFSv4.1 support for NFS4_RESULT_PRESERVER_UNLINKED
Olga Kornievskaia [Wed, 2 Feb 2022 22:55:02 +0000 (17:55 -0500)]
NFSv4.1 support for NFS4_RESULT_PRESERVER_UNLINKED

In 4.1+, the server is allowed to set a flag
NFS4_RESULT_PRESERVE_UNLINKED in reply to the OPEN, that tells
the client that it does not need to do a silly rename of an
opened file when it's being removed.

Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoSUNRPC/xprtrdma: Convert GFP_NOFS to GFP_KERNEL
Trond Myklebust [Sat, 29 Jan 2022 19:44:38 +0000 (14:44 -0500)]
SUNRPC/xprtrdma: Convert GFP_NOFS to GFP_KERNEL

Assume that the upper layers have set memalloc_nofs_save/restore as
appropriate.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoSUNRPC/auth_gss: Convert GFP_NOFS to GFP_KERNEL
Trond Myklebust [Sat, 29 Jan 2022 19:43:09 +0000 (14:43 -0500)]
SUNRPC/auth_gss: Convert GFP_NOFS to GFP_KERNEL

Assume that the upper layers have set memalloc_nofs_save/restore as
appropriate.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoSUNRPC: Convert GFP_NOFS to GFP_KERNEL
Trond Myklebust [Sat, 29 Jan 2022 17:49:44 +0000 (12:49 -0500)]
SUNRPC: Convert GFP_NOFS to GFP_KERNEL

The sections which should not re-enter the filesystem are already
protected with memalloc_nofs_save/restore calls, so it is better to use
GFP_KERNEL in these calls to allow better performance for synchronous
RPC calls.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFSv4.2/copyoffload: Convert GFP_NOFS to GFP_KERNEL
Trond Myklebust [Sat, 29 Jan 2022 19:19:15 +0000 (14:19 -0500)]
NFSv4.2/copyoffload: Convert GFP_NOFS to GFP_KERNEL

There doesn't seem to be any reason why the copy offload code can't use
GFP_KERNEL. It can't get called by direct reclaim.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFSv4/flexfiles: Convert GFP_NOFS to GFP_KERNEL
Trond Myklebust [Sat, 29 Jan 2022 19:16:55 +0000 (14:16 -0500)]
NFSv4/flexfiles: Convert GFP_NOFS to GFP_KERNEL

Assume that the higher layers will have set memalloc_nofs_save/restore
as appropriate.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Convert GFP_NOFS to GFP_KERNEL
Trond Myklebust [Sat, 29 Jan 2022 18:57:38 +0000 (13:57 -0500)]
NFS: Convert GFP_NOFS to GFP_KERNEL

Assume that sections that should not re-enter the filesystem are already
protected with memalloc_nofs_save/restore call, so relax those GFP_NOFS
instances which might be used by other contexts.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFSv4.2: Fix up an invalid combination of memory allocation flags
Trond Myklebust [Sat, 29 Jan 2022 19:03:06 +0000 (14:03 -0500)]
NFSv4.2: Fix up an invalid combination of memory allocation flags

We should use either GFP_KERNEL or GFP_NOFS, but not both. Also strip
GFP_KERNEL_ACCOUNT down to GFP_KERNEL. This memory is shrinkable, so
does not need to be limited by kmemcg.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFSv4: Charge NFSv4 open state trackers to kmemcg
Trond Myklebust [Sat, 29 Jan 2022 18:42:01 +0000 (13:42 -0500)]
NFSv4: Charge NFSv4 open state trackers to kmemcg

Allow kmemcg to limit the number of NFSv4 delegation, lock and open
state trackers.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: Charge open/lock file contexts to kmemcg
Trond Myklebust [Sat, 29 Jan 2022 18:04:10 +0000 (13:04 -0500)]
NFS: Charge open/lock file contexts to kmemcg

Allow kmemcg to limit the number of open/lock file contexts, in the same
way that it limits the parent file descriptors.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFSv4: Protect the state recovery thread against direct reclaim
Trond Myklebust [Sat, 29 Jan 2022 18:32:45 +0000 (13:32 -0500)]
NFSv4: Protect the state recovery thread against direct reclaim

If memory allocation triggers a direct reclaim from the state recovery
thread, then we can deadlock. Use memalloc_nofs_save/restore to ensure
that doesn't happen.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoSUNRPC: remove redundant pointer plainhdr
Colin Ian King [Sun, 30 Jan 2022 22:44:10 +0000 (22:44 +0000)]
SUNRPC: remove redundant pointer plainhdr

[You don't often get email from colin.i.king@gmail.com. Learn why this is important at http://aka.ms/LearnAboutSenderIdentification.]

Pointer plainhdr is being assigned a value that is never read, the
pointer is redundant and can be removed.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFSv4.2: fix reference count leaks in _nfs42_proc_copy_notify()
Xin Xiong [Tue, 25 Jan 2022 13:10:45 +0000 (21:10 +0800)]
NFSv4.2: fix reference count leaks in _nfs42_proc_copy_notify()

[You don't often get email from xiongx18@fudan.edu.cn. Learn why this is important at http://aka.ms/LearnAboutSenderIdentification.]

The reference counting issue happens in two error paths in the
function _nfs42_proc_copy_notify(). In both error paths, the function
simply returns the error code and forgets to balance the refcount of
object `ctx`, bumped by get_nfs_open_context() earlier, which may
cause refcount leaks.

Fix it by balancing refcount of the `ctx` object before the function
returns in both error paths.

Signed-off-by: Xin Xiong <xiongx18@fudan.edu.cn>
Signed-off-by: Xiyu Yang <xiyuyang19@fudan.edu.cn>
Signed-off-by: Xin Tan <tanxin.ctf@gmail.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoConvert NFS from readpages to readahead
Matthew Wilcox (Oracle) [Sat, 22 Jan 2022 20:54:52 +0000 (20:54 +0000)]
Convert NFS from readpages to readahead

NFS is one of the last two users of the deprecated ->readpages aop.
This conversion looks straightforward, but I have only compile-tested
it.

Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoNFS: simplify check for freeing cn_resp
Tom Rix [Mon, 17 Jan 2022 17:50:10 +0000 (09:50 -0800)]
NFS: simplify check for freeing cn_resp

nfs42_files_from_same_server() is called to check if freeing
cn_resp is required, just do the free.

Signed-off-by: Tom Rix <trix@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
2 years agoMerge tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 25 Feb 2022 01:35:22 +0000 (17:35 -0800)]
Merge tag 'clk-fixes-for-linus' of git://git./linux/kernel/git/clk/linux

Pull clk fixes from Stephen Boyd:
 "A couple driver fixes in the clk subsystem

   - Fix a hang due to bad clk parent in the Ingenic jz4725b driver

   - Fix SD controllers on Qualcomm MSM8994 SoCs by removing clks that
     shouldn't be touched"

* tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux:
  clk: jz4725b: fix mmc0 clock gating
  clk: qcom: gcc-msm8994: Remove NoC clocks

2 years agoMerge tag 'drm-fixes-2022-02-25' of git://anongit.freedesktop.org/drm/drm
Linus Torvalds [Fri, 25 Feb 2022 01:29:26 +0000 (17:29 -0800)]
Merge tag 'drm-fixes-2022-02-25' of git://anongit.freedesktop.org/drm/drm

Pull drm fixes from Dave Airlie:
 "Regular drm fixes pull, i915, amdgpu and tegra mostly, all pretty
  small.

  core:
   - edid: Always set RGB444

  tegra:
   - tegra186 suspend/resume fixes
   - syncpoint wait fix
   - build warning fix
   - eDP on older devices fix

  amdgpu:
   - Display FP fix
   - PCO powergating fix
   - RDNA2 OEM SKU stability fixes
   - Display PSR fix
   - PCI ASPM fix
   - Display link encoder fix for TEST_COMMIT
   - Raven2 suspend/resume fix
   - Fix a regression in virtual display support
   - GPUVM eviction fix

  i915:
   - Fix QGV handling on ADL-P+
   - Fix bw atomic check when switching between SAGV vs. no SAGV
   - Disconnect PHYs left connected by BIOS on disabled ports
   - Fix SAVG to no SAGV transitions on TGL+
   - Print PHY name properly on calibration error (DG2)

  imx:
   - dcss: Select GEM CMA helpers

  radeon:
   - Fix some variables's type

  vc4:
   - Fix codec cleanup
   - Fix PM reference counting"

* tag 'drm-fixes-2022-02-25' of git://anongit.freedesktop.org/drm/drm: (24 commits)
  drm/amdgpu: check vm ready by amdgpu_vm->evicting flag
  drm/amdgpu: bypass tiling flag check in virtual display case (v2)
  Revert "drm/amdgpu: add modifiers in amdgpu_vkms_plane_init()"
  drm/amdgpu: do not enable asic reset for raven2
  drm/amd/display: Fix stream->link_enc unassigned during stream removal
  drm/amd: Check if ASPM is enabled from PCIe subsystem
  drm/edid: Always set RGB444
  drm/tegra: dpaux: Populate AUX bus
  drm/radeon: fix variable type
  drm/amd/display: For vblank_disable_immediate, check PSR is really used
  drm/amd/pm: fix some OEM SKU specific stability issues
  drm/amdgpu: disable MMHUB PG for Picasso
  drm/amd/display: Protect update_bw_bounding_box FPU code.
  drm/i915/dg2: Print PHY name properly on calibration error
  drm/i915: Fix bw atomic check when switching between SAGV vs. no SAGV
  drm/i915: Correctly populate use_sagv_wm for all pipes
  drm/i915: Disconnect PHYs left connected by BIOS on disabled ports
  drm/i915: Widen the QGV point mask
  drm/imx/dcss: i.MX8MQ DCSS select DRM_GEM_CMA_HELPER
  drm/vc4: crtc: Fix runtime_pm reference counting
  ...

2 years agoMerge tag 'perf-tools-fixes-for-v5.17-2022-02-24' of git://git.kernel.org/pub/scm...
Linus Torvalds [Thu, 24 Feb 2022 22:36:38 +0000 (14:36 -0800)]
Merge tag 'perf-tools-fixes-for-v5.17-2022-02-24' of git://git./linux/kernel/git/acme/linux

Pull perf tools fixes from Arnaldo Carvalho de Melo:

 - Fix double free in in the error path when opening perf.data from
   multiple files in a directory instead of from a single file

 - Sync the msr-index.h copy with the kernel sources

 - Fix error when printing 'weight' field in 'perf script'

 - Skip failing sigtrap test for arm+aarch64 in 'perf test'

 - Fix failure to use a cpu list for uncore events in hybrid systems,
   e.g. Intel Alder Lake

* tag 'perf-tools-fixes-for-v5.17-2022-02-24' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux:
  perf script: Fix error when printing 'weight' field
  tools arch x86: Sync the msr-index.h copy with the kernel sources
  perf data: Fix double free in perf_session__delete()
  perf evlist: Fix failed to use cpu list for uncore events
  perf test: Skip failing sigtrap test for arm+aarch64

2 years agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Thu, 24 Feb 2022 22:05:49 +0000 (14:05 -0800)]
Merge tag 'for-linus' of git://git./virt/kvm/kvm

Pull kvm fixes from Paolo Bonzini:
 "x86 host:

   - Expose KVM_CAP_ENABLE_CAP since it is supported

   - Disable KVM_HC_CLOCK_PAIRING in TSC catchup mode

   - Ensure async page fault token is nonzero

   - Fix lockdep false negative

   - Fix FPU migration regression from the AMX changes

  x86 guest:

   - Don't use PV TLB/IPI/yield on uniprocessor guests

  PPC:

   - reserve capability id (topic branch for ppc/kvm)"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: x86: nSVM: disallow userspace setting of MSR_AMD64_TSC_RATIO to non default value when tsc scaling disabled
  KVM: x86/mmu: make apf token non-zero to fix bug
  KVM: PPC: reserve capability 210 for KVM_CAP_PPC_AIL_MODE_3
  x86/kvm: Don't use pv tlb/ipi/sched_yield if on 1 vCPU
  x86/kvm: Fix compilation warning in non-x86_64 builds
  x86/kvm/fpu: Remove kvm_vcpu_arch.guest_supported_xcr0
  x86/kvm/fpu: Limit guest user_xfeatures to supported bits of XCR0
  kvm: x86: Disable KVM_HC_CLOCK_PAIRING if tsc is in always catchup mode
  KVM: Fix lockdep false negative during host resume
  KVM: x86: Add KVM_CAP_ENABLE_CAP to x86

2 years agoMerge tag 'pci-v5.17-fixes-5' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaa...
Linus Torvalds [Thu, 24 Feb 2022 21:19:57 +0000 (13:19 -0800)]
Merge tag 'pci-v5.17-fixes-5' of git://git./linux/kernel/git/helgaas/pci

Pull pci fixes from Bjorn Helgaas:

 - Fix a merge error that broke PCI device enumeration on mvebu
   platforms, including Turris Omnia (Armada 385) (Pali Rohár)

 - Avoid using ATS on all AMD Navi10 and Navi14 GPUs because some
   VBIOSes don't account for "harvested" (disabled) parts of the chip
   when initializing caches (Alex Deucher)

* tag 'pci-v5.17-fixes-5' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
  PCI: Mark all AMD Navi10 and Navi14 GPU ATS as broken
  PCI: mvebu: Fix device enumeration regression

2 years agoMerge tag 'net-5.17-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Linus Torvalds [Thu, 24 Feb 2022 20:45:32 +0000 (12:45 -0800)]
Merge tag 'net-5.17-rc6' of git://git./linux/kernel/git/netdev/net

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

  Current release - regressions:

   - bpf: fix crash due to out of bounds access into reg2btf_ids

   - mvpp2: always set port pcs ops, avoid null-deref

   - eth: marvell: fix driver load from initrd

   - eth: intel: revert "Fix reset bw limit when DCB enabled with 1 TC"

  Current release - new code bugs:

   - mptcp: fix race in overlapping signal events

  Previous releases - regressions:

   - xen-netback: revert hotplug-status changes causing devices to not
     be configured

   - dsa:
      - avoid call to __dev_set_promiscuity() while rtnl_mutex isn't
        held
      - fix panic when removing unoffloaded port from bridge

   - dsa: microchip: fix bridging with more than two member ports

  Previous releases - always broken:

   - bpf:
      - fix crash due to incorrect copy_map_value when both spin lock
        and timer are present in a single value
      - fix a bpf_timer initialization issue with clang
      - do not try bpf_msg_push_data with len 0
      - add schedule points in batch ops

   - nf_tables:
      - unregister flowtable hooks on netns exit
      - correct flow offload action array size
      - fix a couple of memory leaks

   - vsock: don't check owner in vhost_vsock_stop() while releasing

   - gso: do not skip outer ip header in case of ipip and net_failover

   - smc: use a mutex for locking "struct smc_pnettable"

   - openvswitch: fix setting ipv6 fields causing hw csum failure

   - mptcp: fix race in incoming ADD_ADDR option processing

   - sysfs: add check for netdevice being present to speed_show

   - sched: act_ct: fix flow table lookup after ct clear or switching
     zones

   - eth: intel: fixes for SR-IOV forwarding offloads

   - eth: broadcom: fixes for selftests and error recovery

   - eth: mellanox: flow steering and SR-IOV forwarding fixes

  Misc:

   - make __pskb_pull_tail() & pskb_carve_frag_list() drop_monitor
     friends not report freed skbs as drops

   - force inlining of checksum functions in net/checksum.h"

* tag 'net-5.17-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (85 commits)
  net: mv643xx_eth: process retval from of_get_mac_address
  ping: remove pr_err from ping_lookup
  Revert "i40e: Fix reset bw limit when DCB enabled with 1 TC"
  openvswitch: Fix setting ipv6 fields causing hw csum failure
  ipv6: prevent a possible race condition with lifetimes
  net/smc: Use a mutex for locking "struct smc_pnettable"
  bnx2x: fix driver load from initrd
  Revert "xen-netback: Check for hotplug-status existence before watching"
  Revert "xen-netback: remove 'hotplug-status' once it has served its purpose"
  net/mlx5e: Fix VF min/max rate parameters interchange mistake
  net/mlx5e: Add missing increment of count
  net/mlx5e: MPLSoUDP decap, fix check for unsupported matches
  net/mlx5e: Fix MPLSoUDP encap to use MPLS action information
  net/mlx5e: Add feature check for set fec counters
  net/mlx5e: TC, Skip redundant ct clear actions
  net/mlx5e: TC, Reject rules with forward and drop actions
  net/mlx5e: TC, Reject rules with drop and modify hdr action
  net/mlx5e: kTLS, Use CHECKSUM_UNNECESSARY for device-offloaded packets
  net/mlx5e: Fix wrong return value on ioctl EEPROM query failure
  net/mlx5: Fix possible deadlock on rule deletion
  ...

2 years agoMerge tag 'drm-intel-fixes-2022-02-24' of git://anongit.freedesktop.org/drm/drm-intel...
Dave Airlie [Thu, 24 Feb 2022 19:51:04 +0000 (05:51 +1000)]
Merge tag 'drm-intel-fixes-2022-02-24' of git://anongit.freedesktop.org/drm/drm-intel into drm-fixes

- Fix QGV handling on ADL-P+ (Ville Syrjälä)
- Fix bw atomic check when switching between SAGV vs. no SAGV (Ville Syrjälä)
- Disconnect PHYs left connected by BIOS on disabled ports (Imre Deak)
- Fix SAVG to no SAGV transitions on TGL+ (Ville Syrjälä)
- Print PHY name properly on calibration error (DG2) (Matt Roper)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/YhdyHwRWkOTWwlqi@tursulin-mobl2
2 years agoMerge tag 'block-5.17-2022-02-24' of git://git.kernel.dk/linux-block
Linus Torvalds [Thu, 24 Feb 2022 19:15:10 +0000 (11:15 -0800)]
Merge tag 'block-5.17-2022-02-24' of git://git.kernel.dk/linux-block

Pull block fixes from Jens Axboe:

 - NVMe pull request:
    - send H2CData PDUs based on MAXH2CDATA (Varun Prakash)
    - fix passthrough to namespaces with unsupported features (Christoph
      Hellwig)

 - Clear iocb->private at poll completion (Stefano)

* tag 'block-5.17-2022-02-24' of git://git.kernel.dk/linux-block:
  nvme-tcp: send H2CData PDUs based on MAXH2CDATA
  nvme: also mark passthrough-only namespaces ready in nvme_update_ns_info
  nvme: don't return an error from nvme_configure_metadata
  block: clear iocb->private in blkdev_bio_end_io_async()

2 years agoMerge tag 'io_uring-5.17-2022-02-23' of git://git.kernel.dk/linux-block
Linus Torvalds [Thu, 24 Feb 2022 19:08:15 +0000 (11:08 -0800)]
Merge tag 'io_uring-5.17-2022-02-23' of git://git.kernel.dk/linux-block

Pull io_uring fixes from Jens Axboe:

 - Add a conditional schedule point in io_add_buffers() (Eric)

 - Fix for a quiesce speedup merged in this release (Dylan)

 - Don't convert to jiffies for event timeout waiting, it's way too
   coarse when we accept a timespec as input (me)

* tag 'io_uring-5.17-2022-02-23' of git://git.kernel.dk/linux-block:
  io_uring: disallow modification of rsrc_data during quiesce
  io_uring: don't convert to jiffies for waiting on timeouts
  io_uring: add a schedule point in io_add_buffers()

2 years agoMerge tag 'platform-drivers-x86-v5.17-4' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 24 Feb 2022 18:42:20 +0000 (10:42 -0800)]
Merge tag 'platform-drivers-x86-v5.17-4' of git://git./linux/kernel/git/pdx86/platform-drivers-x86

Pull more x86 platform driver fixes from Hans de Goede:
 "Two more fixes:

   - Fix suspend/resume regression on AMD Cezanne APUs in >= 5.16

   - Fix Microsoft Surface 3 battery readings"

* tag 'platform-drivers-x86-v5.17-4' of git://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86:
  surface: surface3_power: Fix battery readings on batteries without a serial number
  platform/x86: amd-pmc: Set QOS during suspend on CZN w/ timer wakeup

2 years agonet: mv643xx_eth: process retval from of_get_mac_address
Mauri Sandberg [Wed, 23 Feb 2022 14:23:37 +0000 (16:23 +0200)]
net: mv643xx_eth: process retval from of_get_mac_address

Obtaining a MAC address may be deferred in cases when the MAC is stored
in an NVMEM block, for example, and it may not be ready upon the first
retrieval attempt and return EPROBE_DEFER.

It is also possible that a port that does not rely on NVMEM has been
already created when getting the defer request. Thus, also the resources
allocated previously must be freed when doing a roll-back.

Fixes: 76723bca2802 ("net: mv643xx_eth: add DT parsing support")
Signed-off-by: Mauri Sandberg <maukka@ext.kapsi.fi>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/20220223142337.41757-1-maukka@ext.kapsi.fi
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoKVM: x86: nSVM: disallow userspace setting of MSR_AMD64_TSC_RATIO to non default...
Maxim Levitsky [Wed, 23 Feb 2022 11:56:49 +0000 (13:56 +0200)]
KVM: x86: nSVM: disallow userspace setting of MSR_AMD64_TSC_RATIO to non default value when tsc scaling disabled

If nested tsc scaling is disabled, MSR_AMD64_TSC_RATIO should
never have non default value.

Due to way nested tsc scaling support was implmented in qemu,
it would set this msr to 0 when nested tsc scaling was disabled.
Ignore that value for now, as it causes no harm.

Fixes: 5228eb96a487 ("KVM: x86: nSVM: implement nested TSC scaling")
Cc: stable@vger.kernel.org
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20220223115649.319134-1-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoKVM: x86/mmu: make apf token non-zero to fix bug
Liang Zhang [Tue, 22 Feb 2022 03:12:39 +0000 (11:12 +0800)]
KVM: x86/mmu: make apf token non-zero to fix bug

In current async pagefault logic, when a page is ready, KVM relies on
kvm_arch_can_dequeue_async_page_present() to determine whether to deliver
a READY event to the Guest. This function test token value of struct
kvm_vcpu_pv_apf_data, which must be reset to zero by Guest kernel when a
READY event is finished by Guest. If value is zero meaning that a READY
event is done, so the KVM can deliver another.
But the kvm_arch_setup_async_pf() may produce a valid token with zero
value, which is confused with previous mention and may lead the loss of
this READY event.

This bug may cause task blocked forever in Guest:
 INFO: task stress:7532 blocked for more than 1254 seconds.
       Not tainted 5.10.0 #16
 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
 task:stress          state:D stack:    0 pid: 7532 ppid:  1409
 flags:0x00000080
 Call Trace:
  __schedule+0x1e7/0x650
  schedule+0x46/0xb0
  kvm_async_pf_task_wait_schedule+0xad/0xe0
  ? exit_to_user_mode_prepare+0x60/0x70
  __kvm_handle_async_pf+0x4f/0xb0
  ? asm_exc_page_fault+0x8/0x30
  exc_page_fault+0x6f/0x110
  ? asm_exc_page_fault+0x8/0x30
  asm_exc_page_fault+0x1e/0x30
 RIP: 0033:0x402d00
 RSP: 002b:00007ffd31912500 EFLAGS: 00010206
 RAX: 0000000000071000 RBX: ffffffffffffffff RCX: 00000000021a32b0
 RDX: 000000000007d011 RSI: 000000000007d000 RDI: 00000000021262b0
 RBP: 00000000021262b0 R08: 0000000000000003 R09: 0000000000000086
 R10: 00000000000000eb R11: 00007fefbdf2baa0 R12: 0000000000000000
 R13: 0000000000000002 R14: 000000000007d000 R15: 0000000000001000

Signed-off-by: Liang Zhang <zhangliang5@huawei.com>
Message-Id: <20220222031239.1076682-1-zhangliang5@huawei.com>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoping: remove pr_err from ping_lookup
Xin Long [Thu, 24 Feb 2022 03:41:08 +0000 (22:41 -0500)]
ping: remove pr_err from ping_lookup

As Jakub noticed, prints should be avoided on the datapath.
Also, as packets would never come to the else branch in
ping_lookup(), remove pr_err() from ping_lookup().

Fixes: 35a79e64de29 ("ping: fix the dif and sdif check in ping_lookup")
Reported-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Link: https://lore.kernel.org/r/1ef3f2fcd31bd681a193b1fcf235eee1603819bd.1645674068.git.lucien.xin@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoRevert "i40e: Fix reset bw limit when DCB enabled with 1 TC"
Mateusz Palczewski [Wed, 23 Feb 2022 17:53:47 +0000 (09:53 -0800)]
Revert "i40e: Fix reset bw limit when DCB enabled with 1 TC"

Revert of a patch that instead of fixing a AQ error when trying
to reset BW limit introduced several regressions related to
creation and managing TC. Currently there are errors when creating
a TC on both PF and VF.

Error log:
[17428.783095] i40e 0000:3b:00.1: AQ command Config VSI BW allocation per TC failed = 14
[17428.783107] i40e 0000:3b:00.1: Failed configuring TC map 0 for VSI 391
[17428.783254] i40e 0000:3b:00.1: AQ command Config VSI BW allocation per TC failed = 14
[17428.783259] i40e 0000:3b:00.1: Unable to  configure TC map 0 for VSI 391

This reverts commit 3d2504663c41104b4359a15f35670cfa82de1bbf.

Fixes: 3d2504663c41 (i40e: Fix reset bw limit when DCB enabled with 1 TC)
Signed-off-by: Mateusz Palczewski <mateusz.palczewski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Link: https://lore.kernel.org/r/20220223175347.1690692-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoopenvswitch: Fix setting ipv6 fields causing hw csum failure
Paul Blakey [Wed, 23 Feb 2022 16:34:16 +0000 (18:34 +0200)]
openvswitch: Fix setting ipv6 fields causing hw csum failure

Ipv6 ttl, label and tos fields are modified without first
pulling/pushing the ipv6 header, which would have updated
the hw csum (if available). This might cause csum validation
when sending the packet to the stack, as can be seen in
the trace below.

Fix this by updating skb->csum if available.

Trace resulted by ipv6 ttl dec and then sending packet
to conntrack [actions: set(ipv6(hlimit=63)),ct(zone=99)]:
[295241.900063] s_pf0vf2: hw csum failure
[295241.923191] Call Trace:
[295241.925728]  <IRQ>
[295241.927836]  dump_stack+0x5c/0x80
[295241.931240]  __skb_checksum_complete+0xac/0xc0
[295241.935778]  nf_conntrack_tcp_packet+0x398/0xba0 [nf_conntrack]
[295241.953030]  nf_conntrack_in+0x498/0x5e0 [nf_conntrack]
[295241.958344]  __ovs_ct_lookup+0xac/0x860 [openvswitch]
[295241.968532]  ovs_ct_execute+0x4a7/0x7c0 [openvswitch]
[295241.979167]  do_execute_actions+0x54a/0xaa0 [openvswitch]
[295242.001482]  ovs_execute_actions+0x48/0x100 [openvswitch]
[295242.006966]  ovs_dp_process_packet+0x96/0x1d0 [openvswitch]
[295242.012626]  ovs_vport_receive+0x6c/0xc0 [openvswitch]
[295242.028763]  netdev_frame_hook+0xc0/0x180 [openvswitch]
[295242.034074]  __netif_receive_skb_core+0x2ca/0xcb0
[295242.047498]  netif_receive_skb_internal+0x3e/0xc0
[295242.052291]  napi_gro_receive+0xba/0xe0
[295242.056231]  mlx5e_handle_rx_cqe_mpwrq_rep+0x12b/0x250 [mlx5_core]
[295242.062513]  mlx5e_poll_rx_cq+0xa0f/0xa30 [mlx5_core]
[295242.067669]  mlx5e_napi_poll+0xe1/0x6b0 [mlx5_core]
[295242.077958]  net_rx_action+0x149/0x3b0
[295242.086762]  __do_softirq+0xd7/0x2d6
[295242.090427]  irq_exit+0xf7/0x100
[295242.093748]  do_IRQ+0x7f/0xd0
[295242.096806]  common_interrupt+0xf/0xf
[295242.100559]  </IRQ>
[295242.102750] RIP: 0033:0x7f9022e88cbd
[295242.125246] RSP: 002b:00007f9022282b20 EFLAGS: 00000246 ORIG_RAX: ffffffffffffffda
[295242.132900] RAX: 0000000000000005 RBX: 0000000000000010 RCX: 0000000000000000
[295242.140120] RDX: 00007f9022282ba8 RSI: 00007f9022282a30 RDI: 00007f9014005c30
[295242.147337] RBP: 00007f9014014d60 R08: 0000000000000020 R09: 00007f90254a8340
[295242.154557] R10: 00007f9022282a28 R11: 0000000000000246 R12: 0000000000000000
[295242.161775] R13: 00007f902308c000 R14: 000000000000002b R15: 00007f9022b71f40

Fixes: 3fdbd1ce11e5 ("openvswitch: add ipv6 'set' action")
Signed-off-by: Paul Blakey <paulb@nvidia.com>
Link: https://lore.kernel.org/r/20220223163416.24096-1-paulb@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoipv6: prevent a possible race condition with lifetimes
Niels Dossche [Wed, 23 Feb 2022 13:19:56 +0000 (14:19 +0100)]
ipv6: prevent a possible race condition with lifetimes

valid_lft, prefered_lft and tstamp are always accessed under the lock
"lock" in other places. Reading these without taking the lock may result
in inconsistencies regarding the calculation of the valid and preferred
variables since decisions are taken on these fields for those variables.

Signed-off-by: Niels Dossche <dossche.niels@gmail.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: Niels Dossche <niels.dossche@ugent.be>
Link: https://lore.kernel.org/r/20220223131954.6570-1-niels.dossche@ugent.be
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet/smc: Use a mutex for locking "struct smc_pnettable"
Fabio M. De Francesco [Wed, 23 Feb 2022 10:02:52 +0000 (11:02 +0100)]
net/smc: Use a mutex for locking "struct smc_pnettable"

smc_pnetid_by_table_ib() uses read_lock() and then it calls smc_pnet_apply_ib()
which, in turn, calls mutex_lock(&smc_ib_devices.mutex).

read_lock() disables preemption. Therefore, the code acquires a mutex while in
atomic context and it leads to a SAC bug.

Fix this bug by replacing the rwlock with a mutex.

Reported-and-tested-by: syzbot+4f322a6d84e991c38775@syzkaller.appspotmail.com
Fixes: 64e28b52c7a6 ("net/smc: add pnet table namespace support")
Confirmed-by: Tony Lu <tonylu@linux.alibaba.com>
Signed-off-by: Fabio M. De Francesco <fmdefrancesco@gmail.com>
Acked-by: Karsten Graul <kgraul@linux.ibm.com>
Link: https://lore.kernel.org/r/20220223100252.22562-1-fmdefrancesco@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agobnx2x: fix driver load from initrd
Manish Chopra [Wed, 23 Feb 2022 08:57:20 +0000 (00:57 -0800)]
bnx2x: fix driver load from initrd

Commit b7a49f73059f ("bnx2x: Utilize firmware 7.13.21.0") added
new firmware support in the driver with maintaining older firmware
compatibility. However, older firmware was not added in MODULE_FIRMWARE()
which caused missing firmware files in initrd image leading to driver load
failure from initrd. This patch adds MODULE_FIRMWARE() for older firmware
version to have firmware files included in initrd.

Fixes: b7a49f73059f ("bnx2x: Utilize firmware 7.13.21.0")
Link: https://bugzilla.kernel.org/show_bug.cgi?id=215627
Signed-off-by: Manish Chopra <manishc@marvell.com>
Signed-off-by: Alok Prasad <palok@marvell.com>
Signed-off-by: Ariel Elior <aelior@marvell.com>
Link: https://lore.kernel.org/r/20220223085720.12021-1-manishc@marvell.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoRevert "xen-netback: Check for hotplug-status existence before watching"
Marek Marczykowski-Górecki [Tue, 22 Feb 2022 00:18:17 +0000 (01:18 +0100)]
Revert "xen-netback: Check for hotplug-status existence before watching"

This reverts commit 2afeec08ab5c86ae21952151f726bfe184f6b23d.

The reasoning in the commit was wrong - the code expected to setup the
watch even if 'hotplug-status' didn't exist. In fact, it relied on the
watch being fired the first time - to check if maybe 'hotplug-status' is
already set to 'connected'. Not registering a watch for non-existing
path (which is the case if hotplug script hasn't been executed yet),
made the backend not waiting for the hotplug script to execute. This in
turns, made the netfront think the interface is fully operational, while
in fact it was not (the vif interface on xen-netback side might not be
configured yet).

This was a workaround for 'hotplug-status' erroneously being removed.
But since that is reverted now, the workaround is not necessary either.

More discussion at
https://lore.kernel.org/xen-devel/afedd7cb-a291-e773-8b0d-4db9b291fa98@ipxe.org/T/#u

Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Reviewed-by: Paul Durrant <paul@xen.org>
Reviewed-by: Michael Brown <mbrown@fensystems.co.uk>
Link: https://lore.kernel.org/r/20220222001817.2264967-2-marmarek@invisiblethingslab.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoRevert "xen-netback: remove 'hotplug-status' once it has served its purpose"
Marek Marczykowski-Górecki [Tue, 22 Feb 2022 00:18:16 +0000 (01:18 +0100)]
Revert "xen-netback: remove 'hotplug-status' once it has served its purpose"

This reverts commit 1f2565780e9b7218cf92c7630130e82dcc0fe9c2.

The 'hotplug-status' node should not be removed as long as the vif
device remains configured. Otherwise the xen-netback would wait for
re-running the network script even if it was already called (in case of
the frontent re-connecting). But also, it _should_ be removed when the
vif device is destroyed (for example when unbinding the driver) -
otherwise hotplug script would not configure the device whenever it
re-appear.

Moving removal of the 'hotplug-status' node was a workaround for nothing
calling network script after xen-netback module is reloaded. But when
vif interface is re-created (on xen-netback unbind/bind for example),
the script should be called, regardless of who does that - currently
this case is not handled by the toolstack, and requires manual
script call. Keeping hotplug-status=connected to skip the call is wrong
and leads to not configured interface.

More discussion at
https://lore.kernel.org/xen-devel/afedd7cb-a291-e773-8b0d-4db9b291fa98@ipxe.org/T/#u

Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Reviewed-by: Paul Durrant <paul@xen.org>
Link: https://lore.kernel.org/r/20220222001817.2264967-1-marmarek@invisiblethingslab.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge tag 'nvme-5.17-2022-02-24' of git://git.infradead.org/nvme into block-5.17
Jens Axboe [Thu, 24 Feb 2022 14:02:15 +0000 (07:02 -0700)]
Merge tag 'nvme-5.17-2022-02-24' of git://git.infradead.org/nvme into block-5.17

Pull NVMe fixes from Christoph:

"nvme fixes for Linux 5.17

 - send H2CData PDUs based on MAXH2CDATA (Varun Prakash)
 - fix passthrough to namespaces with unsupported features (me)"

* tag 'nvme-5.17-2022-02-24' of git://git.infradead.org/nvme:
  nvme-tcp: send H2CData PDUs based on MAXH2CDATA
  nvme: also mark passthrough-only namespaces ready in nvme_update_ns_info
  nvme: don't return an error from nvme_configure_metadata

2 years agosurface: surface3_power: Fix battery readings on batteries without a serial number
Hans de Goede [Thu, 24 Feb 2022 10:18:48 +0000 (11:18 +0100)]
surface: surface3_power: Fix battery readings on batteries without a serial number

The battery on the 2nd hand Surface 3 which I recently bought appears to
not have a serial number programmed in. This results in any I2C reads from
the registers containing the serial number failing with an I2C NACK.

This was causing mshw0011_bix() to fail causing the battery readings to
not work at all.

Ignore EREMOTEIO (I2C NACK) errors when retrieving the serial number and
continue with an empty serial number to fix this.

Fixes: b1f81b496b0d ("platform/x86: surface3_power: MSHW0011 rev-eng implementation")
BugLink: https://github.com/linux-surface/linux-surface/issues/608
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Reviewed-by: Maximilian Luz <luzmaximilian@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20220224101848.7219-1-hdegoede@redhat.com
2 years agoplatform/x86: amd-pmc: Set QOS during suspend on CZN w/ timer wakeup
Mario Limonciello [Wed, 23 Feb 2022 17:52:37 +0000 (11:52 -0600)]
platform/x86: amd-pmc: Set QOS during suspend on CZN w/ timer wakeup

commit 59348401ebed ("platform/x86: amd-pmc: Add special handling for
timer based S0i3 wakeup") adds support for using another platform timer
in lieu of the RTC which doesn't work properly on some systems. This path
was validated and worked well before submission. During the 5.16-rc1 merge
window other patches were merged that caused this to stop working properly.

When this feature was used with 5.16-rc1 or later some OEM laptops with the
matching firmware requirements from that commit would shutdown instead of
program a timer based wakeup.

This was bisected to commit 8d89835b0467 ("PM: suspend: Do not pause
cpuidle in the suspend-to-idle path").  This wasn't supposed to cause any
negative impacts and also tested well on both Intel and ARM platforms.
However this changed the semantics of when CPUs are allowed to be in the
deepest state. For the AMD systems in question it appears this causes a
firmware crash for timer based wakeup.

It's hypothesized to be caused by the `amd-pmc` driver sending `OS_HINT`
and all the CPUs going into a deep state while the timer is still being
programmed. It's likely a firmware bug, but to avoid it don't allow setting
CPUs into the deepest state while using CZN timer wakeup path.

If later it's discovered that this also occurs from "regular" suspends
without a timer as well or on other silicon, this may be later expanded to
run in the suspend path for more scenarios.

Cc: stable@vger.kernel.org # 5.16+
Suggested-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://lore.kernel.org/linux-acpi/BL1PR12MB51570F5BD05980A0DCA1F3F4E23A9@BL1PR12MB5157.namprd12.prod.outlook.com/T/#mee35f39c41a04b624700ab2621c795367f19c90e
Fixes: 8d89835b0467 ("PM: suspend: Do not pause cpuidle in the suspend-to-idle path")
Fixes: 23f62d7ab25b ("PM: sleep: Pause cpuidle later and resume it earlier during system transitions")
Fixes: 59348401ebed ("platform/x86: amd-pmc: Add special handling for timer based S0i3 wakeup"
Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Link: https://lore.kernel.org/r/20220223175237.6209-1-mario.limonciello@amd.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
2 years agoMerge tag 'mlx5-fixes-2022-02-23' of git://git.kernel.org/pub/scm/linux/kernel/git...
Jakub Kicinski [Thu, 24 Feb 2022 04:30:00 +0000 (20:30 -0800)]
Merge tag 'mlx5-fixes-2022-02-23' of git://git./linux/kernel/git/saeed/linux

Saeed Mahameed says:

====================
mlx5 fixes 2022-02-22

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

* tag 'mlx5-fixes-2022-02-23' of git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux:
  net/mlx5e: Fix VF min/max rate parameters interchange mistake
  net/mlx5e: Add missing increment of count
  net/mlx5e: MPLSoUDP decap, fix check for unsupported matches
  net/mlx5e: Fix MPLSoUDP encap to use MPLS action information
  net/mlx5e: Add feature check for set fec counters
  net/mlx5e: TC, Skip redundant ct clear actions
  net/mlx5e: TC, Reject rules with forward and drop actions
  net/mlx5e: TC, Reject rules with drop and modify hdr action
  net/mlx5e: kTLS, Use CHECKSUM_UNNECESSARY for device-offloaded packets
  net/mlx5e: Fix wrong return value on ioctl EEPROM query failure
  net/mlx5: Fix possible deadlock on rule deletion
  net/mlx5: Fix tc max supported prio for nic mode
  net/mlx5: Fix wrong limitation of metadata match on ecpf
  net/mlx5: Update log_max_qp value to be 17 at most
  net/mlx5: DR, Fix the threshold that defines when pool sync is initiated
  net/mlx5: DR, Don't allow match on IP w/o matching on full ethertype/ip_version
  net/mlx5: DR, Fix slab-out-of-bounds in mlx5_cmd_dr_create_fte
  net/mlx5: DR, Cache STE shadow memory
  net/mlx5: Update the list of the PCI supported devices
====================

Link: https://lore.kernel.org/r/20220224001123.365265-1-saeed@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge tag 'amd-drm-fixes-5.17-2022-02-23' of https://gitlab.freedesktop.org/agd5f...
Dave Airlie [Thu, 24 Feb 2022 04:27:35 +0000 (14:27 +1000)]
Merge tag 'amd-drm-fixes-5.17-2022-02-23' of https://gitlab.freedesktop.org/agd5f/linux into drm-fixes

amd-drm-fixes-5.17-2022-02-23:

amdgpu:
- Display FP fix
- PCO powergating fix
- RDNA2 OEM SKU stability fixes
- Display PSR fix
- PCI ASPM fix
- Display link encoder fix for TEST_COMMIT
- Raven2 suspend/resume fix
- Fix a regression in virtual display support
- GPUVM eviction fix

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220223214623.28823-1-alexander.deucher@amd.com
2 years agoMerge tag 'drm/tegra/for-5.17-rc6' of https://gitlab.freedesktop.org/drm/tegra into...
Dave Airlie [Thu, 24 Feb 2022 04:21:57 +0000 (14:21 +1000)]
Merge tag 'drm/tegra/for-5.17-rc6' of https://gitlab.freedesktop.org/drm/tegra into drm-fixes

drm/tegra: Fixes for v5.17-rc6

Contains a couple of fixes for Tegra186 suspend/resume, syncpoint
waiting, a build warning and eDP on older Tegra devices.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Thierry Reding <thierry.reding@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220223161903.293392-1-thierry.reding@gmail.com
2 years agoMerge tag 'drm-misc-fixes-2022-02-23' of git://anongit.freedesktop.org/drm/drm-misc...
Dave Airlie [Thu, 24 Feb 2022 03:51:46 +0000 (13:51 +1000)]
Merge tag 'drm-misc-fixes-2022-02-23' of git://anongit.freedesktop.org/drm/drm-misc into drm-fixes

 * edid: Always set RGB444
 * imx/dcss: Select GEM CMA helpers
 * radeon: Fix some variables's type
 * vc4: Fix codec cleanup; Fix PM reference counting

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/YhaKj4zWJ42YWRts@linux-uq9g.fritz.box
2 years agoMerge tag 'devicetree-fixes-for-5.17-2' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 24 Feb 2022 01:25:22 +0000 (17:25 -0800)]
Merge tag 'devicetree-fixes-for-5.17-2' of git://git./linux/kernel/git/robh/linux

Pull devicetree fixes from Rob Herring:

 - Update some maintainers email addresses

 - Fix handling of elfcorehdr reservation for crash dump kernel

 - Fix unittest expected warnings text

* tag 'devicetree-fixes-for-5.17-2' of git://git.kernel.org/pub/scm/linux/kernel/git/robh/linux:
  dt-bindings: update Roger Quadros email
  MAINTAINERS: sifive: drop Yash Shah
  of/fdt: move elfcorehdr reservation early for crash dump kernel
  of: unittest: update text of expected warnings

2 years agoMerge tag 'selinux-pr-20220223' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Thu, 24 Feb 2022 01:19:55 +0000 (17:19 -0800)]
Merge tag 'selinux-pr-20220223' of git://git./linux/kernel/git/pcmoore/selinux

Pull selinux fix from Paul Moore:
 "A second small SELinux fix which addresses an incorrect
  mutex_is_locked() check"

* tag 'selinux-pr-20220223' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux:
  selinux: fix misuse of mutex_is_locked()

2 years agonet/mlx5e: Fix VF min/max rate parameters interchange mistake
Gal Pressman [Mon, 21 Feb 2022 15:54:34 +0000 (17:54 +0200)]
net/mlx5e: Fix VF min/max rate parameters interchange mistake

The VF min and max rate were passed incorrectly and resulted in wrongly
interchanging them. Fix the order of parameters in
mlx5_esw_qos_set_vport_rate().

Fixes: d7df09f5e7b4 ("net/mlx5: E-switch, Enable vport QoS on demand")
Signed-off-by: Gal Pressman <gal@nvidia.com>
Reviewed-by: Aya Levin <ayal@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: Add missing increment of count
Lama Kayal [Mon, 21 Feb 2022 10:26:11 +0000 (12:26 +0200)]
net/mlx5e: Add missing increment of count

Add mistakenly missing increment of count variable when looping over
output buffer in mlx5e_self_test().

This resolves the issue of garbage values output when querying with self
test via ethtool.

before:
$ ethtool -t eth2
The test result is PASS
The test extra info:
Link Test        0
Speed Test       1768697188
Health Test      758528120
Loopback Test    3288687

after:
$ ethtool -t eth2
The test result is PASS
The test extra info:
Link Test        0
Speed Test       0
Health Test      0
Loopback Test    0

Fixes: 7990b1b5e8bd ("net/mlx5e: loopback test is not supported in switchdev mode")
Signed-off-by: Lama Kayal <lkayal@nvidia.com>
Reviewed-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: MPLSoUDP decap, fix check for unsupported matches
Maor Dickman [Thu, 6 Jan 2022 12:46:24 +0000 (14:46 +0200)]
net/mlx5e: MPLSoUDP decap, fix check for unsupported matches

Currently offload of rule on bareudp device require tunnel key
in order to match on mpls fields and without it the mpls fields
are ignored, this is incorrect due to the fact udp tunnel doesn't
have key to match on.

Fix by returning error in case flow is matching on tunnel key.

Fixes: 72046a91d134 ("net/mlx5e: Allow to match on mpls parameters")
Signed-off-by: Maor Dickman <maord@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: Fix MPLSoUDP encap to use MPLS action information
Maor Dickman [Thu, 6 Jan 2022 12:10:18 +0000 (14:10 +0200)]
net/mlx5e: Fix MPLSoUDP encap to use MPLS action information

Currently the MPLSoUDP encap builds the MPLS header using encap action
information (tunnel id, ttl and tos) instead of the MPLS action
information (label, ttl, tc and bos) which is wrong.

Fix by storing the MPLS action information during the flow action
parse and later using it to create the encap MPLS header.

Fixes: f828ca6a2fb6 ("net/mlx5e: Add support for hw encapsulation of MPLS over UDP")
Signed-off-by: Maor Dickman <maord@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: Add feature check for set fec counters
Lama Kayal [Tue, 1 Feb 2022 09:24:41 +0000 (11:24 +0200)]
net/mlx5e: Add feature check for set fec counters

Fec counters support is checked via the PCAM feature_cap_mask,
bit 0: PPCNT_counter_group_Phy_statistical_counter_group.
Add feature check to avoid faulty behavior.

Fixes: 0a1498ebfa55 ("net/mlx5e: Expose FEC counters via ethtool")
Signed-off-by: Lama Kayal <lkayal@nvidia.com>
Reviewed-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: TC, Skip redundant ct clear actions
Roi Dayan [Thu, 3 Feb 2022 07:42:19 +0000 (09:42 +0200)]
net/mlx5e: TC, Skip redundant ct clear actions

Offload of ct clear action is just resetting the reg_c register.
It's done by allocating modify hdr resources which is limited.
Doing it multiple times is redundant and wasting modify hdr resources
and if resources depleted the driver will fail offloading the rule.
Ignore redundant ct clear actions after the first one.

Fixes: 806401c20a0f ("net/mlx5e: CT, Fix multiple allocations and memleak of mod acts")
Signed-off-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Ariel Levkovich <lariel@nvidia.com>
Reviewed-by: Maor Dickman <maord@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: TC, Reject rules with forward and drop actions
Roi Dayan [Mon, 17 Jan 2022 13:00:30 +0000 (15:00 +0200)]
net/mlx5e: TC, Reject rules with forward and drop actions

Such rules are redundant but allowed and passed to the driver.
The driver does not support offloading such rules so return an error.

Fixes: 03a9d11e6eeb ("net/mlx5e: Add TC drop and mirred/redirect action parsing for SRIOV offloads")
Signed-off-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Oz Shlomo <ozsh@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: TC, Reject rules with drop and modify hdr action
Roi Dayan [Tue, 4 Jan 2022 08:38:02 +0000 (10:38 +0200)]
net/mlx5e: TC, Reject rules with drop and modify hdr action

This kind of action is not supported by firmware and generates a
syndrome.

kernel: mlx5_core 0000:08:00.0: mlx5_cmd_check:777:(pid 102063): SET_FLOW_TABLE_ENTRY(0x936) op_mod(0x0) failed, status bad parameter(0x3), syndrome (0x8708c3)

Fixes: d7e75a325cb2 ("net/mlx5e: Add offloading of E-Switch TC pedit (header re-write) actions")
Signed-off-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Maor Dickman <maord@nvidia.com>
Reviewed-by: Oz Shlomo <ozsh@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: kTLS, Use CHECKSUM_UNNECESSARY for device-offloaded packets
Tariq Toukan [Mon, 31 Jan 2022 08:26:19 +0000 (10:26 +0200)]
net/mlx5e: kTLS, Use CHECKSUM_UNNECESSARY for device-offloaded packets

For RX TLS device-offloaded packets, the HW spec guarantees checksum
validation for the offloaded packets, but does not define whether the
CQE.checksum field matches the original packet (ciphertext) or
the decrypted one (plaintext). This latitude allows architetctural
improvements between generations of chips, resulting in different decisions
regarding the value type of CQE.checksum.

Hence, for these packets, the device driver should not make use of this CQE
field. Here we block CHECKSUM_COMPLETE usage for RX TLS device-offloaded
packets, and use CHECKSUM_UNNECESSARY instead.

Value of the packet's tcp_hdr.csum is not modified by the HW, and it always
matches the original ciphertext.

Fixes: 1182f3659357 ("net/mlx5e: kTLS, Add kTLS RX HW offload support")
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: Fix wrong return value on ioctl EEPROM query failure
Gal Pressman [Wed, 2 Feb 2022 14:07:21 +0000 (16:07 +0200)]
net/mlx5e: Fix wrong return value on ioctl EEPROM query failure

The ioctl EEPROM query wrongly returns success on read failures, fix
that by returning the appropriate error code.

Fixes: bb64143eee8c ("net/mlx5e: Add ethtool support for dump module EEPROM")
Signed-off-by: Gal Pressman <gal@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Fix possible deadlock on rule deletion
Maor Gottlieb [Mon, 24 Jan 2022 19:25:04 +0000 (21:25 +0200)]
net/mlx5: Fix possible deadlock on rule deletion

Add missing call to up_write_ref_node() which releases the semaphore
in case the FTE doesn't have destinations, such in drop rule case.

Fixes: 465e7baab6d9 ("net/mlx5: Fix deletion of duplicate rules")
Signed-off-by: Maor Gottlieb <maorg@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Fix tc max supported prio for nic mode
Chris Mi [Tue, 14 Dec 2021 01:52:53 +0000 (03:52 +0200)]
net/mlx5: Fix tc max supported prio for nic mode

Only prio 1 is supported if firmware doesn't support ignore flow
level for nic mode. The offending commit removed the check wrongly.
Add it back.

Fixes: 9a99c8f1253a ("net/mlx5e: E-Switch, Offload all chain 0 priorities when modify header and forward action is not supported")
Signed-off-by: Chris Mi <cmi@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Fix wrong limitation of metadata match on ecpf
Ariel Levkovich [Fri, 28 Jan 2022 23:39:24 +0000 (01:39 +0200)]
net/mlx5: Fix wrong limitation of metadata match on ecpf

Match metadata support check returns false for ecpf device.
However, this support does exist for ecpf and therefore this
limitation should be removed to allow feature such as stacked
devices and internal port offloaded to be supported.

Fixes: 92ab1eb392c6 ("net/mlx5: E-Switch, Enable vport metadata matching if firmware supports it")
Signed-off-by: Ariel Levkovich <lariel@nvidia.com>
Reviewed-by: Maor Dickman <maord@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Update log_max_qp value to be 17 at most
Maher Sanalla [Wed, 16 Feb 2022 09:01:04 +0000 (11:01 +0200)]
net/mlx5: Update log_max_qp value to be 17 at most

Currently, log_max_qp value is dependent on what FW reports as its max capability.
In reality, due to a bug, some FWs report a value greater than 17, even though they
don't support log_max_qp > 17.

This FW issue led the driver to exhaust memory on startup.
Thus, log_max_qp value is set to be no more than 17 regardless
of what FW reports, as it was before the cited commit.

Fixes: f79a609ea6bf ("net/mlx5: Update log_max_qp value to FW max capability")
Signed-off-by: Maher Sanalla <msanalla@nvidia.com>
Reviewed-by: Avihai Horon <avihaih@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: DR, Fix the threshold that defines when pool sync is initiated
Yevgeny Kliteynik [Wed, 29 Dec 2021 20:22:05 +0000 (22:22 +0200)]
net/mlx5: DR, Fix the threshold that defines when pool sync is initiated

When deciding whether to start syncing and actually free all the "hot"
ICM chunks, we need to consider the type of the ICM chunks that we're
dealing with. For instance, the amount of available ICM for MODIFY_ACTION
is significantly lower than the usual STE ICM, so the threshold should
account for that - otherwise we can deplete MODIFY_ACTION memory just by
creating and deleting the same modify header action in a continuous loop.

This patch replaces the hard-coded threshold with a dynamic value.

Fixes: 1c58651412bb ("net/mlx5: DR, ICM memory pools sync optimization")
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Reviewed-by: Alex Vesker <valex@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: DR, Don't allow match on IP w/o matching on full ethertype/ip_version
Yevgeny Kliteynik [Thu, 13 Jan 2022 12:52:48 +0000 (14:52 +0200)]
net/mlx5: DR, Don't allow match on IP w/o matching on full ethertype/ip_version

Currently SMFS allows adding rule with matching on src/dst IP w/o matching
on full ethertype or ip_version, which is not supported by HW.
This patch fixes this issue and adds the check as it is done in DMFS.

Fixes: 26d688e33f88 ("net/mlx5: DR, Add Steering entry (STE) utilities")
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Reviewed-by: Alex Vesker <valex@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: DR, Fix slab-out-of-bounds in mlx5_cmd_dr_create_fte
Yevgeny Kliteynik [Tue, 11 Jan 2022 01:00:03 +0000 (03:00 +0200)]
net/mlx5: DR, Fix slab-out-of-bounds in mlx5_cmd_dr_create_fte

When adding a rule with 32 destinations, we hit the following out-of-band
access issue:

  BUG: KASAN: slab-out-of-bounds in mlx5_cmd_dr_create_fte+0x18ee/0x1e70

This patch fixes the issue by both increasing the allocated buffers to
accommodate for the needed actions and by checking the number of actions
to prevent this issue when a rule with too many actions is provided.

Fixes: 1ffd498901c1 ("net/mlx5: DR, Increase supported num of actions to 32")
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Reviewed-by: Alex Vesker <valex@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: DR, Cache STE shadow memory
Yevgeny Kliteynik [Thu, 23 Dec 2021 23:07:30 +0000 (01:07 +0200)]
net/mlx5: DR, Cache STE shadow memory

During rule insertion on each ICM memory chunk we also allocate shadow memory
used for management. This includes the hw_ste, dr_ste and miss list per entry.
Since the scale of these allocations is large we noticed a performance hiccup
that happens once malloc and free are stressed.
In extreme usecases when ~1M chunks are freed at once, it might take up to 40
seconds to complete this, up to the point the kernel sees this as self-detected
stall on CPU:

 rcu: INFO: rcu_sched self-detected stall on CPU

To resolve this we will increase the reuse of shadow memory.
Doing this we see that a time in the aforementioned usecase dropped from ~40
seconds to ~8-10 seconds.

Fixes: 29cf8febd185 ("net/mlx5: DR, ICM pool memory allocator")
Signed-off-by: Alex Vesker <valex@nvidia.com>
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Update the list of the PCI supported devices
Meir Lichtinger [Mon, 10 Jan 2022 08:14:41 +0000 (10:14 +0200)]
net/mlx5: Update the list of the PCI supported devices

Add the upcoming BlueField-4 and ConnectX-8 device IDs.

Fixes: 2e9d3e83ab82 ("net/mlx5: Update the list of the PCI supported devices")
Signed-off-by: Meir Lichtinger <meirl@nvidia.com>
Reviewed-by: Gal Pressman <gal@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agodrm/amdgpu: check vm ready by amdgpu_vm->evicting flag
Qiang Yu [Mon, 21 Feb 2022 09:53:56 +0000 (17:53 +0800)]
drm/amdgpu: check vm ready by amdgpu_vm->evicting flag

Workstation application ANSA/META v21.1.4 get this error dmesg when
running CI test suite provided by ANSA/META:
[drm:amdgpu_gem_va_ioctl [amdgpu]] *ERROR* Couldn't update BO_VA (-16)

This is caused by:
1. create a 256MB buffer in invisible VRAM
2. CPU map the buffer and access it causes vm_fault and try to move
   it to visible VRAM
3. force visible VRAM space and traverse all VRAM bos to check if
   evicting this bo is valuable
4. when checking a VM bo (in invisible VRAM), amdgpu_vm_evictable()
   will set amdgpu_vm->evicting, but latter due to not in visible
   VRAM, won't really evict it so not add it to amdgpu_vm->evicted
5. before next CS to clear the amdgpu_vm->evicting, user VM ops
   ioctl will pass amdgpu_vm_ready() (check amdgpu_vm->evicted)
   but fail in amdgpu_vm_bo_update_mapping() (check
   amdgpu_vm->evicting) and get this error log

This error won't affect functionality as next CS will finish the
waiting VM ops. But we'd better clear the error log by checking
the amdgpu_vm->evicting flag in amdgpu_vm_ready() to stop calling
amdgpu_vm_bo_update_mapping() later.

Another reason is amdgpu_vm->evicted list holds all BOs (both
user buffer and page table), but only page table BOs' eviction
prevent VM ops. amdgpu_vm->evicting flag is set only for page
table BOs, so we should use evicting flag instead of evicted list
in amdgpu_vm_ready().

The side effect of this change is: previously blocked VM op (user
buffer in "evicted" list but no page table in it) gets done
immediately.

v2: update commit comments.

Acked-by: Paul Menzel <pmenzel@molgen.mpg.de>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Qiang Yu <qiang.yu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
2 years agodrm/amdgpu: bypass tiling flag check in virtual display case (v2)
Guchun Chen [Fri, 18 Feb 2022 05:05:26 +0000 (13:05 +0800)]
drm/amdgpu: bypass tiling flag check in virtual display case (v2)

vkms leverages common amdgpu framebuffer creation, and
also as it does not support FB modifier, there is no need
to check tiling flags when initing framebuffer when virtual
display is enabled.

This can fix below calltrace:

amdgpu 0000:00:08.0: GFX9+ requires FB check based on format modifier
WARNING: CPU: 0 PID: 1023 at drivers/gpu/drm/amd/amdgpu/amdgpu_display.c:1150 amdgpu_display_framebuffer_init+0x8e7/0xb40 [amdgpu]

v2: check adev->enable_virtual_display instead as vkms can be
enabled in bare metal as well.

Signed-off-by: Leslie Shi <Yuliang.Shi@amd.com>
Signed-off-by: Guchun Chen <guchun.chen@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agoRevert "drm/amdgpu: add modifiers in amdgpu_vkms_plane_init()"
Guchun Chen [Fri, 18 Feb 2022 04:57:52 +0000 (12:57 +0800)]
Revert "drm/amdgpu: add modifiers in amdgpu_vkms_plane_init()"

This reverts commit 4046afcebfc3c8c0dd5666c2671b2c192b344f78.

No need to support modifier in virtual kms, otherwise, in SRIOV
mode, when lanuching X server, set crtc will fail due to mismatch
between primary plane modifier and framebuffer modifier.

Signed-off-by: Guchun Chen <guchun.chen@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2 years agodrm/amdgpu: do not enable asic reset for raven2
Chen Gong [Thu, 17 Feb 2022 07:29:41 +0000 (15:29 +0800)]
drm/amdgpu: do not enable asic reset for raven2

The GPU reset function of raven2 is not maintained or tested, so it should be
very unstable.

Now the amdgpu_asic_reset function is added to amdgpu_pmops_suspend, which
causes the S3 test of raven2 to fail, so the asic_reset of raven2 is ignored
here.

Fixes: daf8de0874ab5b ("drm/amdgpu: always reset the asic in suspend (v2)")
Signed-off-by: Chen Gong <curry.gong@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
2 years agodrm/amd/display: Fix stream->link_enc unassigned during stream removal
Nicholas Kazlauskas [Tue, 25 Jan 2022 17:04:34 +0000 (12:04 -0500)]
drm/amd/display: Fix stream->link_enc unassigned during stream removal

[Why]
Found when running igt@kms_atomic.

Userspace attempts to do a TEST_COMMIT when 0 streams which calls
dc_remove_stream_from_ctx. This in turn calls link_enc_unassign
which ends up modifying stream->link = NULL directly, causing the
global link_enc to be removed preventing further link activity
and future link validation from passing.

[How]
We take care of link_enc unassignment at the start of
link_enc_cfg_link_encs_assign so this call is no longer necessary.

Fixes global state from being modified while unlocked.

Reviewed-by: Jimmy Kizito <Jimmy.Kizito@amd.com>
Acked-by: Jasdeep Dhillon <jdhillon@amd.com>
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
2 years agodrm/amd: Check if ASPM is enabled from PCIe subsystem
Mario Limonciello [Tue, 1 Feb 2022 16:26:33 +0000 (10:26 -0600)]
drm/amd: Check if ASPM is enabled from PCIe subsystem

commit 0064b0ce85bb ("drm/amd/pm: enable ASPM by default") enabled ASPM
by default but a variety of hardware configurations it turns out that this
caused a regression.

* PPC64LE hardware does not support ASPM at a hardware level.
  CONFIG_PCIEASPM is often disabled on these architectures.
* Some dGPUs on ALD platforms don't work with ASPM enabled and PCIe subsystem
  disables it

Check with the PCIe subsystem to see that ASPM has been enabled
or not.

Fixes: 0064b0ce85bb ("drm/amd/pm: enable ASPM by default")
Link: https://wiki.raptorcs.com/w/images/a/ad/P9_PHB_version1.0_27July2018_pub.pdf
Link: https://gitlab.freedesktop.org/drm/amd/-/issues/1723
Link: https://gitlab.freedesktop.org/drm/amd/-/issues/1739
Link: https://gitlab.freedesktop.org/drm/amd/-/issues/1885
Link: https://gitlab.freedesktop.org/drm/amd/-/issues/1907
Tested-by: koba.ko@canonical.com
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
2 years agoMerge tag 'for-5.17/parisc-4' of git://git.kernel.org/pub/scm/linux/kernel/git/deller...
Linus Torvalds [Wed, 23 Feb 2022 20:06:23 +0000 (12:06 -0800)]
Merge tag 'for-5.17/parisc-4' of git://git./linux/kernel/git/deller/parisc-linux

Pull parisc unaligned handler fixes from Helge Deller:
 "Two patches which fix a few bugs in the unalignment handlers.

  The fldd and fstd instructions weren't handled at all on 32-bit
  kernels, the stw instruction didn't check for fault errors and the
  fldw_l and ldw_m were handled wrongly as integer vs floating point
  instructions.

  Both patches are tagged for stable series"

* tag 'for-5.17/parisc-4' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
  parisc/unaligned: Fix ldw() and stw() unalignment handlers
  parisc/unaligned: Fix fldd and fstd unaligned handlers on 32-bit kernel

2 years agoMerge tag 'hwmon-for-v5.17-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 23 Feb 2022 19:51:35 +0000 (11:51 -0800)]
Merge tag 'hwmon-for-v5.17-rc6' of git://git./linux/kernel/git/groeck/linux-staging

Pull hwmon fixes from Guenter Roeck:
 "Fix two old bugs and one new bug in the hwmon subsystem:

   - In pmbus core, clear pmbus fault/warning status bits after read to
     follow PMBus standard

   - In hwmon core, handle failure to register sensor with thermal zone
     correctly

   - In ntc_thermal driver, use valid thermistor names for Samsung
     thermistors"

* tag 'hwmon-for-v5.17-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
  hwmon: (pmbus) Clear pmbus fault/warning bits after read
  hwmon: Handle failure to register sensor with thermal zone correctly
  hwmon: (ntc_thermistor) Underscore Samsung thermistor

2 years agoMerge tag 'slab-for-5.17-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/vbabka...
Linus Torvalds [Wed, 23 Feb 2022 19:33:12 +0000 (11:33 -0800)]
Merge tag 'slab-for-5.17-rc6' of git://git./linux/kernel/git/vbabka/slab

Pull slab fixes from Vlastimil Babka:

 - Build fix (workaround) for clang.

 - Fix a /proc/kcore based slabinfo script broken by struct slab changes
   in 5.17-rc1.

* tag 'slab-for-5.17-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/vbabka/slab:
  tools/cgroup/slabinfo: update to work with struct slab
  slab: remove __alloc_size attribute from __kmalloc_track_caller

2 years agoPCI: Mark all AMD Navi10 and Navi14 GPU ATS as broken
Alex Deucher [Tue, 22 Feb 2022 16:08:01 +0000 (11:08 -0500)]
PCI: Mark all AMD Navi10 and Navi14 GPU ATS as broken

There are enough VBIOS escapes without the proper workaround that some
users still hit this.  Microsoft never productized ATS on Windows so OEM
platforms that were Windows-only didn't always validate ATS.

The advantages of ATS are not worth it compared to the potential
instabilities on harvested boards.  Disable ATS on all Navi10 and Navi14
boards.

Symptoms include:

  amdgpu 0000:07:00.0: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0007 address=0xffffc02000 flags=0x0000]
  AMD-Vi: Event logged [IO_PAGE_FAULT device=07:00.0 domain=0x0007 address=0xffffc02000 flags=0x0000]
  [drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring sdma0 timeout, signaled seq=6047, emitted seq=6049
  amdgpu 0000:07:00.0: amdgpu: GPU reset begin!
  amdgpu 0000:07:00.0: amdgpu: GPU reset succeeded, trying to resume
  amdgpu 0000:07:00.0: [drm:amdgpu_ring_test_helper [amdgpu]] *ERROR* ring sdma0 test failed (-110)
  [drm:amdgpu_device_ip_resume_phase2 [amdgpu]] *ERROR* resume of IP block <sdma_v4_0> failed -110
  amdgpu 0000:07:00.0: amdgpu: GPU reset(1) failed

Related commits:

  e8946a53e2a6 ("PCI: Mark AMD Navi14 GPU ATS as broken")
  a2da5d8cc0b0 ("PCI: Mark AMD Raven iGPU ATS as broken in some platforms")
  45beb31d3afb ("PCI: Mark AMD Navi10 GPU rev 0x00 ATS as broken")
  5e89cd303e3a ("PCI: Mark AMD Navi14 GPU rev 0xc5 ATS as broken")
  d28ca864c493 ("PCI: Mark AMD Stoney Radeon R7 GPU ATS as broken")
  9b44b0b09dec ("PCI: Mark AMD Stoney GPU ATS as broken")

[bhelgaas: add symptoms and related commits]
Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1760
Link: https://lore.kernel.org/r/20220222160801.841643-1-alexander.deucher@amd.com
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Christian König <christian.koenig@amd.com>
Acked-by: Guchun Chen <guchun.chen@amd.com>