platform/kernel/linux-starfive.git
17 months agothunderbolt: Disable XDomain lane 1 only in software connection manager
Mika Westerberg [Thu, 8 Sep 2022 06:45:22 +0000 (09:45 +0300)]
thunderbolt: Disable XDomain lane 1 only in software connection manager

commit 84ee211c83212f4d35b56e0603acdcc41f860f1b upstream.

When firmware connection manager is in use we should not touch the lane
adapter (well or any) configuration space so do this only when we know
that the software connection manager is active.

Fixes: 8e1de7042596 ("thunderbolt: Add support for XDomain lane bonding")
Cc: stable@vger.kernel.org
Acked-by: Yehezkel Bernat <YehezkelShB@gmail.com>
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agocifs: do not include page data when checking signature
Enzo Matsumiya [Wed, 18 Jan 2023 17:06:57 +0000 (14:06 -0300)]
cifs: do not include page data when checking signature

commit 30b2b2196d6e4cc24cbec633535a2404f258ce69 upstream.

On async reads, page data is allocated before sending.  When the
response is received but it has no data to fill (e.g.
STATUS_END_OF_FILE), __calc_signature() will still include the pages in
its computation, leading to an invalid signature check.

This patch fixes this by not setting the async read smb_rqst page data
(zeroed by default) if its got_bytes is 0.

This can be reproduced/verified with xfstests generic/465.

Cc: <stable@vger.kernel.org>
Signed-off-by: Enzo Matsumiya <ematsumiya@suse.de>
Reviewed-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agobtrfs: fix race between quota rescan and disable leading to NULL pointer deref
Filipe Manana [Thu, 12 Jan 2023 16:31:08 +0000 (16:31 +0000)]
btrfs: fix race between quota rescan and disable leading to NULL pointer deref

commit b7adbf9ada3513d2092362c8eac5cddc5b651f5c upstream.

If we have one task trying to start the quota rescan worker while another
one is trying to disable quotas, we can end up hitting a race that results
in the quota rescan worker doing a NULL pointer dereference. The steps for
this are the following:

1) Quotas are enabled;

2) Task A calls the quota rescan ioctl and enters btrfs_qgroup_rescan().
   It calls qgroup_rescan_init() which returns 0 (success) and then joins a
   transaction and commits it;

3) Task B calls the quota disable ioctl and enters btrfs_quota_disable().
   It clears the bit BTRFS_FS_QUOTA_ENABLED from fs_info->flags and calls
   btrfs_qgroup_wait_for_completion(), which returns immediately since the
   rescan worker is not yet running.
   Then it starts a transaction and locks fs_info->qgroup_ioctl_lock;

4) Task A queues the rescan worker, by calling btrfs_queue_work();

5) The rescan worker starts, and calls rescan_should_stop() at the start
   of its while loop, which results in 0 iterations of the loop, since
   the flag BTRFS_FS_QUOTA_ENABLED was cleared from fs_info->flags by
   task B at step 3);

6) Task B sets fs_info->quota_root to NULL;

7) The rescan worker tries to start a transaction and uses
   fs_info->quota_root as the root argument for btrfs_start_transaction().
   This results in a NULL pointer dereference down the call chain of
   btrfs_start_transaction(). The stack trace is something like the one
   reported in Link tag below:

   general protection fault, probably for non-canonical address 0xdffffc0000000041: 0000 [#1] PREEMPT SMP KASAN
   KASAN: null-ptr-deref in range [0x0000000000000208-0x000000000000020f]
   CPU: 1 PID: 34 Comm: kworker/u4:2 Not tainted 6.1.0-syzkaller-13872-gb6bb9676f216 #0
   Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 10/26/2022
   Workqueue: btrfs-qgroup-rescan btrfs_work_helper
   RIP: 0010:start_transaction+0x48/0x10f0 fs/btrfs/transaction.c:564
   Code: 48 89 fb 48 (...)
   RSP: 0018:ffffc90000ab7ab0 EFLAGS: 00010206
   RAX: 0000000000000041 RBX: 0000000000000208 RCX: ffff88801779ba80
   RDX: 0000000000000000 RSI: 0000000000000001 RDI: 0000000000000000
   RBP: dffffc0000000000 R08: 0000000000000001 R09: fffff52000156f5d
   R10: fffff52000156f5d R11: 1ffff92000156f5c R12: 0000000000000000
   R13: 0000000000000001 R14: 0000000000000001 R15: 0000000000000003
   FS:  0000000000000000(0000) GS:ffff8880b9900000(0000) knlGS:0000000000000000
   CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
   CR2: 00007f2bea75b718 CR3: 000000001d0cc000 CR4: 00000000003506e0
   DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
   DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
   Call Trace:
    <TASK>
    btrfs_qgroup_rescan_worker+0x3bb/0x6a0 fs/btrfs/qgroup.c:3402
    btrfs_work_helper+0x312/0x850 fs/btrfs/async-thread.c:280
    process_one_work+0x877/0xdb0 kernel/workqueue.c:2289
    worker_thread+0xb14/0x1330 kernel/workqueue.c:2436
    kthread+0x266/0x300 kernel/kthread.c:376
    ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:308
    </TASK>
   Modules linked in:

So fix this by having the rescan worker function not attempt to start a
transaction if it didn't do any rescan work.

Reported-by: syzbot+96977faa68092ad382c4@syzkaller.appspotmail.com
Link: https://lore.kernel.org/linux-btrfs/000000000000e5454b05f065a803@google.com/
Fixes: e804861bd4e6 ("btrfs: fix deadlock between quota disable and qgroup rescan worker")
CC: stable@vger.kernel.org # 5.4+
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agobtrfs: fix invalid leaf access due to inline extent during lseek
Filipe Manana [Thu, 12 Jan 2023 14:17:20 +0000 (14:17 +0000)]
btrfs: fix invalid leaf access due to inline extent during lseek

commit 1f55ee6d0901d915801618bda0af4e5b937e3db7 upstream.

During lseek, for SEEK_DATA and SEEK_HOLE modes, we access the disk_bytenr
of an extent without checking its type. However inline extents have their
data starting the offset of the disk_bytenr field, so accessing that field
when we have an inline extent can result in either of the following:

1) Interpret the inline extent's data as a disk_bytenr value;

2) In case the inline data is less than 8 bytes, we access part of some
   other item in the leaf, or unused space in the leaf;

3) In case the inline data is less than 8 bytes and the extent item is
   the first item in the leaf, we can access beyond the leaf's limit.

So fix this by not accessing the disk_bytenr field if we have an inline
extent.

Fixes: b6e833567ea1 ("btrfs: make hole and data seeking a lot more efficient")
Reported-by: Matthias Schoepfer <matthias.schoepfer@googlemail.com>
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=216908
Link: https://lore.kernel.org/linux-btrfs/7f25442f-b121-2a3a-5a3d-22bcaae83cd4@leemhuis.info/
CC: stable@vger.kernel.org # 6.1
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agobtrfs: qgroup: do not warn on record without old_roots populated
Qu Wenruo [Tue, 10 Jan 2023 07:14:17 +0000 (15:14 +0800)]
btrfs: qgroup: do not warn on record without old_roots populated

commit 75181406b4eafacc531ff2ee5fb032bd93317e2b upstream.

[BUG]
There are some reports from the mailing list that since v6.1 kernel, the
WARN_ON() inside btrfs_qgroup_account_extent() gets triggered during
rescan:

  WARNING: CPU: 3 PID: 6424 at fs/btrfs/qgroup.c:2756 btrfs_qgroup_account_extents+0x1ae/0x260 [btrfs]
  CPU: 3 PID: 6424 Comm: snapperd Tainted: P           OE      6.1.2-1-default #1 openSUSE Tumbleweed 05c7a1b1b61d5627475528f71f50444637b5aad7
  RIP: 0010:btrfs_qgroup_account_extents+0x1ae/0x260 [btrfs]
  Call Trace:
   <TASK>
  btrfs_commit_transaction+0x30c/0xb40 [btrfs c39c9c546c241c593f03bd6d5f39ea1b676250f6]
   ? start_transaction+0xc3/0x5b0 [btrfs c39c9c546c241c593f03bd6d5f39ea1b676250f6]
  btrfs_qgroup_rescan+0x42/0xc0 [btrfs c39c9c546c241c593f03bd6d5f39ea1b676250f6]
   btrfs_ioctl+0x1ab9/0x25c0 [btrfs c39c9c546c241c593f03bd6d5f39ea1b676250f6]
   ? __rseq_handle_notify_resume+0xa9/0x4a0
   ? mntput_no_expire+0x4a/0x240
   ? __seccomp_filter+0x319/0x4d0
   __x64_sys_ioctl+0x90/0xd0
   do_syscall_64+0x5b/0x80
   ? syscall_exit_to_user_mode+0x17/0x40
   ? do_syscall_64+0x67/0x80
  entry_SYSCALL_64_after_hwframe+0x63/0xcd
  RIP: 0033:0x7fd9b790d9bf
   </TASK>

[CAUSE]
Since commit e15e9f43c7ca ("btrfs: introduce
BTRFS_QGROUP_RUNTIME_FLAG_NO_ACCOUNTING to skip qgroup accounting"), if
our qgroup is already in inconsistent state, we will no longer do the
time-consuming backref walk.

This can leave some qgroup records without a valid old_roots ulist.
Normally this is fine, as btrfs_qgroup_account_extents() would also skip
those records if we have NO_ACCOUNTING flag set.

But there is a small window, if we have NO_ACCOUNTING flag set, and
inserted some qgroup_record without a old_roots ulist, but then the user
triggered a qgroup rescan.

During btrfs_qgroup_rescan(), we firstly clear NO_ACCOUNTING flag, then
commit current transaction.

And since we have a qgroup_record with old_roots = NULL, we trigger the
WARN_ON() during btrfs_qgroup_account_extents().

[FIX]
Unfortunately due to the introduction of NO_ACCOUNTING flag, the
assumption that every qgroup_record would have its old_roots populated
is no longer correct.

Fix the false alerts and drop the WARN_ON().

Reported-by: Lukas Straub <lukasstraub2@web.de>
Reported-by: HanatoK <summersnow9403@gmail.com>
Fixes: e15e9f43c7ca ("btrfs: introduce BTRFS_QGROUP_RUNTIME_FLAG_NO_ACCOUNTING to skip qgroup accounting")
CC: stable@vger.kernel.org # 6.1
Link: https://lore.kernel.org/linux-btrfs/2403c697-ddaf-58ad-3829-0335fc89df09@gmail.com/
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agobtrfs: do not abort transaction on failure to update log root
Filipe Manana [Tue, 10 Jan 2023 14:56:38 +0000 (14:56 +0000)]
btrfs: do not abort transaction on failure to update log root

commit 09e44868f1e03c7825ca4283256abedc95e249a3 upstream.

When syncing a log, if we fail to update a log root in the log root tree,
we are aborting the transaction if the failure was not -ENOSPC. This is
excessive because there is a chance that a transaction commit can succeed,
and therefore avoid to turn the filesystem into RO mode. All we need to be
careful about is to mark the log for a full commit, which we already do,
to make sure no one commits a super block pointing to an outdated log root
tree.

So don't abort the transaction if we fail to update a log root in the log
root tree, and log an error if the failure is not -ENOSPC, so that it does
not go completely unnoticed.

CC: stable@vger.kernel.org # 6.0+
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agobtrfs: do not abort transaction on failure to write log tree when syncing log
Filipe Manana [Tue, 10 Jan 2023 14:56:37 +0000 (14:56 +0000)]
btrfs: do not abort transaction on failure to write log tree when syncing log

commit 16199ad9eb6db60a6b10794a09fc1ac6d09312ff upstream.

When syncing the log, if we fail to write log tree extent buffers, we mark
the log for a full commit and abort the transaction. However we don't need
to abort the transaction, all we really need to do is to make sure no one
can commit a superblock pointing to new log tree roots. Just because we
got a failure writing extent buffers for a log tree, it does not mean we
will also fail to do a transaction commit.

One particular case is if due to a bug somewhere, when writing log tree
extent buffers, the tree checker detects some corruption and the writeout
fails because of that. Aborting the transaction can be very disruptive for
a user, specially if the issue happened on a root filesystem. One example
is the scenario in the Link tag below, where an isolated corruption on log
tree leaves was causing transaction aborts when syncing the log.

Link: https://lore.kernel.org/linux-btrfs/ae169fc6-f504-28f0-a098-6fa6a4dfb612@leemhuis.info/
CC: stable@vger.kernel.org # 5.15+
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agobtrfs: add missing setup of log for full commit at add_conflicting_inode()
Filipe Manana [Tue, 10 Jan 2023 14:56:36 +0000 (14:56 +0000)]
btrfs: add missing setup of log for full commit at add_conflicting_inode()

commit 94cd63ae679973edeb5ea95ec25a54467c3e54c8 upstream.

When logging conflicting inodes, if we reach the maximum limit of inodes,
we return BTRFS_LOG_FORCE_COMMIT to force a transaction commit. However
we don't mark the log for full commit (with btrfs_set_log_full_commit()),
which means that once we leave the log transaction and before we commit
the transaction, some other task may sync the log, which is incomplete
as we have not logged all conflicting inodes, leading to some inconsistent
in case that log ends up being replayed.

So also call btrfs_set_log_full_commit() at add_conflicting_inode().

Fixes: e09d94c9e448 ("btrfs: log conflicting inodes without holding log mutex of the initial inode")
CC: stable@vger.kernel.org # 6.1
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agobtrfs: fix directory logging due to race with concurrent index key deletion
Filipe Manana [Tue, 10 Jan 2023 14:56:35 +0000 (14:56 +0000)]
btrfs: fix directory logging due to race with concurrent index key deletion

commit 8bb6898da6271d82d8e76d8088d66b971a7dcfa6 upstream.

Sometimes we log a directory without holding its VFS lock, so while we
logging it, dir index entries may be added or removed. This typically
happens when logging a dentry from a parent directory that points to a
new directory, through log_new_dir_dentries(), or when while logging
some other inode we also need to log its parent directories (through
btrfs_log_all_parents()).

This means that while we are at log_dir_items(), we may not find a dir
index key we found before, because it was deleted in the meanwhile, so
a call to btrfs_search_slot() may return 1 (key not found). In that case
we return from log_dir_items() with a success value (the variable 'err'
has a value of 0). This can lead to a few problems, specially in the case
where the variable 'last_offset' has a value of (u64)-1 (and it's
initialized to that when it was declared):

1) By returning from log_dir_items() with success (0) and a value of
   (u64)-1 for '*last_offset_ret', we end up not logging any other dir
   index keys that follow the missing, just deleted, index key. The
   (u64)-1 value makes log_directory_changes() not call log_dir_items()
   again;

2) Before returning with success (0), log_dir_items(), will log a dir
   index range item covering a range from the last old dentry index
   (stored in the variable 'last_old_dentry_offset') to the value of
   'last_offset'. If 'last_offset' has a value of (u64)-1, then it means
   if the log is persisted and replayed after a power failure, it will
   cause deletion of all the directory entries that have an index number
   between last_old_dentry_offset + 1 and (u64)-1;

3) We can end up returning from log_dir_items() with
   ctx->last_dir_item_offset having a lower value than
   inode->last_dir_index_offset, because the former is set to the current
   key we are processing at process_dir_items_leaf(), and at the end of
   log_directory_changes() we set inode->last_dir_index_offset to the
   current value of ctx->last_dir_item_offset. So if for example a
   deletion of a lower dir index key happened, we set
   ctx->last_dir_item_offset to that index value, then if we return from
   log_dir_items() because btrfs_search_slot() returned 1, we end up
   returning from log_dir_items() with success (0) and then
   log_directory_changes() sets inode->last_dir_index_offset to a lower
   value than it had before.
   This can result in unpredictable and unexpected behaviour when we
   need to log again the directory in the same transaction, and can result
   in ending up with a log tree leaf that has duplicated keys, as we do
   batch insertions of dir index keys into a log tree.

So fix this by making log_dir_items() move on to the next dir index key
if it does not find the one it was looking for.

Reported-by: David Arendt <admin@prnet.org>
Link: https://lore.kernel.org/linux-btrfs/ae169fc6-f504-28f0-a098-6fa6a4dfb612@leemhuis.info/
CC: stable@vger.kernel.org # 4.14+
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agobtrfs: fix missing error handling when logging directory items
Filipe Manana [Tue, 10 Jan 2023 14:56:34 +0000 (14:56 +0000)]
btrfs: fix missing error handling when logging directory items

commit 6d3d970b2735b967650d319be27268fedc5598d1 upstream.

When logging a directory, at log_dir_items(), if we get an error when
attempting to search the subvolume tree for a dir index item, we end up
returning 0 (success) from log_dir_items() because 'err' is left with a
value of 0.

This can lead to a few problems, specially in the case the variable
'last_offset' has a value of (u64)-1 (and it's initialized to that when
it was declared):

1) By returning from log_dir_items() with success (0) and a value of
   (u64)-1 for '*last_offset_ret', we end up not logging any other dir
   index keys that follow the missing, just deleted, index key. The
   (u64)-1 value makes log_directory_changes() not call log_dir_items()
   again;

2) Before returning with success (0), log_dir_items(), will log a dir
   index range item covering a range from the last old dentry index
   (stored in the variable 'last_old_dentry_offset') to the value of
   'last_offset'. If 'last_offset' has a value of (u64)-1, then it means
   if the log is persisted and replayed after a power failure, it will
   cause deletion of all the directory entries that have an index number
   between last_old_dentry_offset + 1 and (u64)-1;

3) We can end up returning from log_dir_items() with
   ctx->last_dir_item_offset having a lower value than
   inode->last_dir_index_offset, because the former is set to the current
   key we are processing at process_dir_items_leaf(), and at the end of
   log_directory_changes() we set inode->last_dir_index_offset to the
   current value of ctx->last_dir_item_offset. So if for example a
   deletion of a lower dir index key happened, we set
   ctx->last_dir_item_offset to that index value, then if we return from
   log_dir_items() because btrfs_search_slot() returned an error, we end up
   returning without any error from log_dir_items() and then
   log_directory_changes() sets inode->last_dir_index_offset to a lower
   value than it had before.
   This can result in unpredictable and unexpected behaviour when we
   need to log again the directory in the same transaction, and can result
   in ending up with a log tree leaf that has duplicated keys, as we do
   batch insertions of dir index keys into a log tree.

Fix this by setting 'err' to the value of 'ret' in case
btrfs_search_slot() or btrfs_previous_item() returned an error. That will
result in falling back to a full transaction commit.

Reported-by: David Arendt <admin@prnet.org>
Link: https://lore.kernel.org/linux-btrfs/ae169fc6-f504-28f0-a098-6fa6a4dfb612@leemhuis.info/
Fixes: e02119d5a7b4 ("Btrfs: Add a write ahead tree log to optimize synchronous operations")
CC: stable@vger.kernel.org # 4.14+
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agobtrfs: add extra error messages to cover non-ENOMEM errors from device_add_list()
Qu Wenruo [Mon, 12 Dec 2022 02:19:37 +0000 (10:19 +0800)]
btrfs: add extra error messages to cover non-ENOMEM errors from device_add_list()

commit ed02363fbbed52a3f5ea0d188edd09045a806eb5 upstream.

[BUG]
When test case btrfs/219 (aka, mount a registered device but with a lower
generation) failed, there is not any useful information for the end user
to find out what's going wrong.

The mount failure just looks like this:

  #  mount -o loop /tmp/219.img2 /mnt/btrfs/
  mount: /mnt/btrfs: mount(2) system call failed: File exists.
         dmesg(1) may have more information after failed mount system call.

While the dmesg contains nothing but the loop device change:

  loop1: detected capacity change from 0 to 524288

[CAUSE]
In device_list_add() we have a lot of extra checks to reject invalid
cases.

That function also contains the regular device scan result like the
following prompt:

  BTRFS: device fsid 6222333e-f9f1-47e6-b306-55ddd4dcaef4 devid 1 transid 8 /dev/loop0 scanned by systemd-udevd (3027)

But unfortunately not all errors have their own error messages, thus if
we hit something wrong in device_add_list(), there may be no error
messages at all.

[FIX]
Add errors message for all non-ENOMEM errors.

For ENOMEM, I'd say we're in a much worse situation, and there should be
some OOM messages way before our call sites.

CC: stable@vger.kernel.org # 6.0+
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agomm/MADV_COLLAPSE: don't expand collapse when vm_end is past requested end
Zach O'Keefe [Sat, 24 Dec 2022 08:20:34 +0000 (00:20 -0800)]
mm/MADV_COLLAPSE: don't expand collapse when vm_end is past requested end

commit 52dc031088f00e323140ece4004e70c33153c6dd upstream.

MADV_COLLAPSE acts on one hugepage-aligned/sized region at a time, until
it has collapsed all eligible memory contained within the bounds supplied
by the user.

At the top of each hugepage iteration we (re)lock mmap_lock and
(re)validate the VMA for eligibility and update variables that might have
changed while mmap_lock was dropped.  One thing that might occur is that
the VMA could be resized, and as such, we refetch vma->vm_end to make sure
we don't collapse past the end of the VMA's new end.

However, it's possible that when refetching vma->vm_end that we expand the
region acted on by MADV_COLLAPSE if vma->vm_end is greater than size+len
supplied by the user.

The consequence here is that we may attempt to collapse more memory than
requested, possibly yielding either "too much success" or "false failure"
user-visible results.  An example of the former is if we MADV_COLLAPSE the
first 4MiB of a 2TiB mmap()'d file, the incorrect refetch would cause the
operation to block for much longer than anticipated as we attempt to
collapse the entire TiB region.  An example of the latter is that applying
MADV_COLLPSE to a 4MiB file mapped to the start of a 6MiB VMA will
successfully collapse the first 4MiB, then incorrectly attempt to collapse
the last hugepage-aligned/sized region -- fail (since readahead/page cache
lookup will fail) -- and report a failure to the user.

I don't believe there is a kernel stability concern here as we always
(re)validate the VMA / region accordingly.  Also as Hugh mentions, the
user-visible effects are: we try to collapse more memory than requested
by the user, and/or failing an operation that should have otherwise
succeeded.  An example is trying to collapse a 4MiB file contained
within a 12MiB VMA.

Don't expand the acted-on region when refetching vma->vm_end.

Link: https://lkml.kernel.org/r/20221224082035.3197140-1-zokeefe@google.com
Fixes: 4d24de9425f7 ("mm: MADV_COLLAPSE: refetch vm_end after reacquiring mmap_lock")
Signed-off-by: Zach O'Keefe <zokeefe@google.com>
Reported-by: Hugh Dickins <hughd@google.com>
Cc: Yang Shi <shy828301@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agomm/userfaultfd: enable writenotify while userfaultfd-wp is enabled for a VMA
David Hildenbrand [Fri, 9 Dec 2022 08:09:12 +0000 (09:09 +0100)]
mm/userfaultfd: enable writenotify while userfaultfd-wp is enabled for a VMA

commit 51d3d5eb74ff53b92dcff48b30ae2ed8edd85a32 upstream.

Currently, we don't enable writenotify when enabling userfaultfd-wp on a
shared writable mapping (for now only shmem and hugetlb).  The consequence
is that vma->vm_page_prot will still include write permissions, to be set
as default for all PTEs that get remapped (e.g., mprotect(), NUMA hinting,
page migration, ...).

So far, vma->vm_page_prot is assumed to be a safe default, meaning that we
only add permissions (e.g., mkwrite) but not remove permissions (e.g.,
wrprotect).  For example, when enabling softdirty tracking, we enable
writenotify.  With uffd-wp on shared mappings, that changed.  More details
on vma->vm_page_prot semantics were summarized in [1].

This is problematic for uffd-wp: we'd have to manually check for a uffd-wp
PTEs/PMDs and manually write-protect PTEs/PMDs, which is error prone.
Prone to such issues is any code that uses vma->vm_page_prot to set PTE
permissions: primarily pte_modify() and mk_pte().

Instead, let's enable writenotify such that PTEs/PMDs/...  will be mapped
write-protected as default and we will only allow selected PTEs that are
definitely safe to be mapped without write-protection (see
can_change_pte_writable()) to be writable.  In the future, we might want
to enable write-bit recovery -- e.g., can_change_pte_writable() -- at more
locations, for example, also when removing uffd-wp protection.

This fixes two known cases:

(a) remove_migration_pte() mapping uffd-wp'ed PTEs writable, resulting
    in uffd-wp not triggering on write access.
(b) do_numa_page() / do_huge_pmd_numa_page() mapping uffd-wp'ed PTEs/PMDs
    writable, resulting in uffd-wp not triggering on write access.

Note that do_numa_page() / do_huge_pmd_numa_page() can be reached even
without NUMA hinting (which currently doesn't seem to be applicable to
shmem), for example, by using uffd-wp with a PROT_WRITE shmem VMA.  On
such a VMA, userfaultfd-wp is currently non-functional.

Note that when enabling userfaultfd-wp, there is no need to walk page
tables to enforce the new default protection for the PTEs: we know that
they cannot be uffd-wp'ed yet, because that can only happen after enabling
uffd-wp for the VMA in general.

Also note that this makes mprotect() on ranges with uffd-wp'ed PTEs not
accidentally set the write bit -- which would result in uffd-wp not
triggering on later write access.  This commit makes uffd-wp on shmem
behave just like uffd-wp on anonymous memory in that regard, even though,
mixing mprotect with uffd-wp is controversial.

[1] https://lkml.kernel.org/r/92173bad-caa3-6b43-9d1e-9a471fdbc184@redhat.com

Link: https://lkml.kernel.org/r/20221209080912.7968-1-david@redhat.com
Fixes: b1f9e876862d ("mm/uffd: enable write protection for shmem & hugetlbfs")
Signed-off-by: David Hildenbrand <david@redhat.com>
Reported-by: Ives van Hoorne <ives@codesandbox.io>
Debugged-by: Peter Xu <peterx@redhat.com>
Acked-by: Peter Xu <peterx@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Mike Rapoport <rppt@linux.vnet.ibm.com>
Cc: Nadav Amit <nadav.amit@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agomm/hugetlb: pre-allocate pgtable pages for uffd wr-protects
Peter Xu [Wed, 4 Jan 2023 22:52:05 +0000 (17:52 -0500)]
mm/hugetlb: pre-allocate pgtable pages for uffd wr-protects

commit fed15f1345dc8a7fc8baa81e8b55c3ba010d7f4b upstream.

Userfaultfd-wp uses pte markers to mark wr-protected pages for both shmem
and hugetlb.  Shmem has pre-allocation ready for markers, but hugetlb path
was overlooked.

Doing so by calling huge_pte_alloc() if the initial pgtable walk fails to
find the huge ptep.  It's possible that huge_pte_alloc() can fail with
high memory pressure, in that case stop the loop immediately and fail
silently.  This is not the most ideal solution but it matches with what we
do with shmem meanwhile it avoids the splat in dmesg.

Link: https://lkml.kernel.org/r/20230104225207.1066932-2-peterx@redhat.com
Fixes: 60dfaad65aa9 ("mm/hugetlb: allow uffd wr-protect none ptes")
Signed-off-by: Peter Xu <peterx@redhat.com>
Reported-by: James Houghton <jthoughton@google.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Acked-by: David Hildenbrand <david@redhat.com>
Acked-by: James Houghton <jthoughton@google.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: Muchun Song <songmuchun@bytedance.com>
Cc: Nadav Amit <nadav.amit@gmail.com>
Cc: <stable@vger.kernel.org> [5.19+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agomm/hugetlb: fix uffd-wp handling for migration entries in hugetlb_change_protection()
David Hildenbrand [Thu, 22 Dec 2022 20:55:11 +0000 (21:55 +0100)]
mm/hugetlb: fix uffd-wp handling for migration entries in hugetlb_change_protection()

commit 44f86392bdd165da7e43d3c772aeb1e128ffd6c8 upstream.

We have to update the uffd-wp SWP PTE bit independent of the type of
migration entry.  Currently, if we're unlucky and we want to install/clear
the uffd-wp bit just while we're migrating a read-only mapped hugetlb
page, we would miss to set/clear the uffd-wp bit.

Further, if we're processing a readable-exclusive migration entry and
neither want to set or clear the uffd-wp bit, we could currently end up
losing the uffd-wp bit.  Note that the same would hold for writable
migrating entries, however, having a writable migration entry with the
uffd-wp bit set would already mean that something went wrong.

Note that the change from !is_readable_migration_entry ->
writable_migration_entry is harmless and actually cleaner, as raised by
Miaohe Lin and discussed in [1].

[1] https://lkml.kernel.org/r/90dd6a93-4500-e0de-2bf0-bf522c311b0c@huawei.com

Link: https://lkml.kernel.org/r/20221222205511.675832-3-david@redhat.com
Fixes: 60dfaad65aa9 ("mm/hugetlb: allow uffd wr-protect none ptes")
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Peter Xu <peterx@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agomm/hugetlb: fix PTE marker handling in hugetlb_change_protection()
David Hildenbrand [Thu, 22 Dec 2022 20:55:10 +0000 (21:55 +0100)]
mm/hugetlb: fix PTE marker handling in hugetlb_change_protection()

commit 0e678153f5be7e6c8d28835f5a678618da4b7a9c upstream.

Patch series "mm/hugetlb: uffd-wp fixes for hugetlb_change_protection()".

Playing with virtio-mem and background snapshots (using uffd-wp) on
hugetlb in QEMU, I managed to trigger a VM_BUG_ON().  Looking into the
details, hugetlb_change_protection() seems to not handle uffd-wp correctly
in all cases.

Patch #1 fixes my test case.  I don't have reproducers for patch #2, as it
requires running into migration entries.

I did not yet check in detail yet if !hugetlb code requires similar care.

This patch (of 2):

There are two problematic cases when stumbling over a PTE marker in
hugetlb_change_protection():

(1) We protect an uffd-wp PTE marker a second time using uffd-wp: we will
    end up in the "!huge_pte_none(pte)" case and mess up the PTE marker.

(2) We unprotect a uffd-wp PTE marker: we will similarly end up in the
    "!huge_pte_none(pte)" case even though we cleared the PTE, because
    the "pte" variable is stale. We'll mess up the PTE marker.

For example, if we later stumble over such a "wrongly modified" PTE marker,
we'll treat it like a present PTE that maps some garbage page.

This can, for example, be triggered by mapping a memfd backed by huge
pages, registering uffd-wp, uffd-wp'ing an unmapped page and (a)
uffd-wp'ing it a second time; or (b) uffd-unprotecting it; or (c)
unregistering uffd-wp. Then, ff we trigger fallocate(FALLOC_FL_PUNCH_HOLE)
on that file range, we will run into a VM_BUG_ON:

[  195.039560] page:00000000ba1f2987 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x0
[  195.039565] flags: 0x7ffffc0001000(reserved|node=0|zone=0|lastcpupid=0x1fffff)
[  195.039568] raw: 0007ffffc0001000 ffffe742c0000008 ffffe742c0000008 0000000000000000
[  195.039569] raw: 0000000000000000 0000000000000000 00000001ffffffff 0000000000000000
[  195.039569] page dumped because: VM_BUG_ON_PAGE(compound && !PageHead(page))
[  195.039573] ------------[ cut here ]------------
[  195.039574] kernel BUG at mm/rmap.c:1346!
[  195.039579] invalid opcode: 0000 [#1] PREEMPT SMP NOPTI
[  195.039581] CPU: 7 PID: 4777 Comm: qemu-system-x86 Not tainted 6.0.12-200.fc36.x86_64 #1
[  195.039583] Hardware name: LENOVO 20WNS1F81N/20WNS1F81N, BIOS N35ET50W (1.50 ) 09/15/2022
[  195.039584] RIP: 0010:page_remove_rmap+0x45b/0x550
[  195.039588] Code: [...]
[  195.039589] RSP: 0018:ffffbc03c3633ba8 EFLAGS: 00010292
[  195.039591] RAX: 0000000000000040 RBX: ffffe742c0000000 RCX: 0000000000000000
[  195.039592] RDX: 0000000000000002 RSI: ffffffff8e7aac1a RDI: 00000000ffffffff
[  195.039592] RBP: 0000000000000001 R08: 0000000000000000 R09: ffffbc03c3633a08
[  195.039593] R10: 0000000000000003 R11: ffffffff8f146328 R12: ffff9b04c42754b0
[  195.039594] R13: ffffffff8fcc6328 R14: ffffbc03c3633c80 R15: ffff9b0484ab9100
[  195.039595] FS:  00007fc7aaf68640(0000) GS:ffff9b0bbf7c0000(0000) knlGS:0000000000000000
[  195.039596] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  195.039597] CR2: 000055d402c49110 CR3: 0000000159392003 CR4: 0000000000772ee0
[  195.039598] PKRU: 55555554
[  195.039599] Call Trace:
[  195.039600]  <TASK>
[  195.039602]  __unmap_hugepage_range+0x33b/0x7d0
[  195.039605]  unmap_hugepage_range+0x55/0x70
[  195.039608]  hugetlb_vmdelete_list+0x77/0xa0
[  195.039611]  hugetlbfs_fallocate+0x410/0x550
[  195.039612]  ? _raw_spin_unlock_irqrestore+0x23/0x40
[  195.039616]  vfs_fallocate+0x12e/0x360
[  195.039618]  __x64_sys_fallocate+0x40/0x70
[  195.039620]  do_syscall_64+0x58/0x80
[  195.039623]  ? syscall_exit_to_user_mode+0x17/0x40
[  195.039624]  ? do_syscall_64+0x67/0x80
[  195.039626]  entry_SYSCALL_64_after_hwframe+0x63/0xcd
[  195.039628] RIP: 0033:0x7fc7b590651f
[  195.039653] Code: [...]
[  195.039654] RSP: 002b:00007fc7aaf66e70 EFLAGS: 00000293 ORIG_RAX: 000000000000011d
[  195.039655] RAX: ffffffffffffffda RBX: 0000558ef4b7f370 RCX: 00007fc7b590651f
[  195.039656] RDX: 0000000018000000 RSI: 0000000000000003 RDI: 000000000000000c
[  195.039657] RBP: 0000000008000000 R08: 0000000000000000 R09: 0000000000000073
[  195.039658] R10: 0000000008000000 R11: 0000000000000293 R12: 0000000018000000
[  195.039658] R13: 00007fb8bbe00000 R14: 000000000000000c R15: 0000000000001000
[  195.039661]  </TASK>

Fix it by not going into the "!huge_pte_none(pte)" case if we stumble over
an exclusive marker.  spin_unlock() + continue would get the job done.

However, instead, make it clearer that there are no fall-through
statements: we process each case (hwpoison, migration, marker, !none,
none) and then unlock the page table to continue with the next PTE.  Let's
avoid "continue" statements and use a single spin_unlock() at the end.

Link: https://lkml.kernel.org/r/20221222205511.675832-1-david@redhat.com
Link: https://lkml.kernel.org/r/20221222205511.675832-2-david@redhat.com
Fixes: 60dfaad65aa9 ("mm/hugetlb: allow uffd wr-protect none ptes")
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agommc: sdhci-esdhc-imx: correct the tuning start tap and step setting
Haibo Chen [Wed, 7 Dec 2022 11:23:15 +0000 (19:23 +0800)]
mmc: sdhci-esdhc-imx: correct the tuning start tap and step setting

commit 1e336aa0c0250ec84c6f16efac40c9f0138e367d upstream.

Current code logic may be impacted by the setting of ROM/Bootloader,
so unmask these bits first, then setting these bits accordingly.

Fixes: 2b16cf326b70 ("mmc: sdhci-esdhc-imx: move tuning static configuration into hwinit function")
Signed-off-by: Haibo Chen <haibo.chen@nxp.com>
Acked-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20221207112315.1812222-1-haibo.chen@nxp.com
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agommc: sunxi-mmc: Fix clock refcount imbalance during unbind
Samuel Holland [Wed, 10 Aug 2022 02:25:09 +0000 (21:25 -0500)]
mmc: sunxi-mmc: Fix clock refcount imbalance during unbind

commit 8509419758f2cc28dd05370385af0d91573b76b4 upstream.

If the controller is suspended by runtime PM, the clock is already
disabled, so do not try to disable it again during removal. Use
pm_runtime_disable() to flush any pending runtime PM transitions.

Fixes: 9a8e1e8cc2c0 ("mmc: sunxi: Add runtime_pm support")
Signed-off-by: Samuel Holland <samuel@sholland.org>
Acked-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20220810022509.43743-1-samuel@sholland.org
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoACPI: PRM: Check whether EFI runtime is available
Ard Biesheuvel [Thu, 12 Jan 2023 13:33:19 +0000 (14:33 +0100)]
ACPI: PRM: Check whether EFI runtime is available

commit 182da6f2b81a78709c58021542fb694f8ed80774 upstream.

The ACPI PRM address space handler calls efi_call_virt_pointer() to
execute PRM firmware code, but doing so is only permitted when the EFI
runtime environment is available. Otherwise, such calls are guaranteed
to result in a crash, and must therefore be avoided.

Given that the EFI runtime services may become unavailable after a crash
occurring in the firmware, we need to check this each time the PRM
address space handler is invoked. If the EFI runtime services were not
available at registration time to being with, don't install the address
space handler at all.

Fixes: cefc7ca46235 ("ACPI: PRM: implement OperationRegion handler for the PlatformRtMechanism subtype")
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Cc: All applicable <stable@vger.kernel.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agocomedi: adv_pci1760: Fix PWM instruction handling
Ian Abbott [Tue, 3 Jan 2023 14:37:54 +0000 (14:37 +0000)]
comedi: adv_pci1760: Fix PWM instruction handling

commit 2efb6edd52dc50273f5e68ad863dd1b1fb2f2d1c upstream.

(Actually, this is fixing the "Read the Current Status" command sent to
the device's outgoing mailbox, but it is only currently used for the PWM
instructions.)

The PCI-1760 is operated mostly by sending commands to a set of Outgoing
Mailbox registers, waiting for the command to complete, and reading the
result from the Incoming Mailbox registers.  One of these commands is
the "Read the Current Status" command.  The number of this command is
0x07 (see the User's Manual for the PCI-1760 at
<https://advdownload.advantech.com/productfile/Downloadfile2/1-11P6653/PCI-1760.pdf>.
The `PCI1760_CMD_GET_STATUS` macro defined in the driver should expand
to this command number 0x07, but unfortunately it currently expands to
0x03.  (Command number 0x03 is not defined in the User's Manual.)
Correct the definition of the `PCI1760_CMD_GET_STATUS` macro to fix it.

This is used by all the PWM subdevice related instructions handled by
`pci1760_pwm_insn_config()` which are probably all broken.  The effect
of sending the undefined command number 0x03 is not known.

Fixes: 14b93bb6bbf0 ("staging: comedi: adv_pci_dio: separate out PCI-1760 support")
Cc: <stable@vger.kernel.org> # v4.5+
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Link: https://lore.kernel.org/r/20230103143754.17564-1-abbotti@mev.co.uk
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agousb: core: hub: disable autosuspend for TI TUSB8041
Flavio Suligoi [Mon, 19 Dec 2022 12:47:59 +0000 (13:47 +0100)]
usb: core: hub: disable autosuspend for TI TUSB8041

commit 7171b0e261b17de96490adf053b8bb4b00061bcf upstream.

The Texas Instruments TUSB8041 has an autosuspend problem at high
temperature.

If there is not USB traffic, after a couple of ms, the device enters in
autosuspend mode. In this condition the external clock stops working, to
save energy. When the USB activity turns on, ther hub exits the
autosuspend state, the clock starts running again and all works fine.

At ambient temperature all works correctly, but at high temperature,
when the USB activity turns on, the external clock doesn't restart and
the hub disappears from the USB bus.

Disabling the autosuspend mode for this hub solves the issue.

Signed-off-by: Flavio Suligoi <f.suligoi@asem.it>
Cc: stable <stable@kernel.org>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Link: https://lore.kernel.org/r/20221219124759.3207032-1-f.suligoi@asem.it
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agomisc: fastrpc: Fix use-after-free race condition for maps
Ola Jeppsson [Thu, 24 Nov 2022 17:49:41 +0000 (17:49 +0000)]
misc: fastrpc: Fix use-after-free race condition for maps

commit 96b328d119eca7563c1edcc4e1039a62e6370ecb upstream.

It is possible that in between calling fastrpc_map_get() until
map->fl->lock is taken in fastrpc_free_map(), another thread can call
fastrpc_map_lookup() and get a reference to a map that is about to be
deleted.

Rewrite fastrpc_map_get() to only increase the reference count of a map
if it's non-zero. Propagate this to callers so they can know if a map is
about to be deleted.

Fixes this warning:
refcount_t: addition on 0; use-after-free.
WARNING: CPU: 5 PID: 10100 at lib/refcount.c:25 refcount_warn_saturate
...
Call trace:
 refcount_warn_saturate
 [fastrpc_map_get inlined]
 [fastrpc_map_lookup inlined]
 fastrpc_map_create
 fastrpc_internal_invoke
 fastrpc_device_ioctl
 __arm64_sys_ioctl
 invoke_syscall

Fixes: c68cfb718c8f ("misc: fastrpc: Add support for context Invoke method")
Cc: stable <stable@kernel.org>
Signed-off-by: Ola Jeppsson <ola@snap.com>
Signed-off-by: Abel Vesa <abel.vesa@linaro.org>
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Link: https://lore.kernel.org/r/20221124174941.418450-4-srinivas.kandagatla@linaro.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agomisc: fastrpc: Don't remove map on creater_process and device_release
Abel Vesa [Thu, 24 Nov 2022 17:49:40 +0000 (17:49 +0000)]
misc: fastrpc: Don't remove map on creater_process and device_release

commit 5bb96c8f9268e2fdb0e5321cbc358ee5941efc15 upstream.

Do not remove the map from the list on error path in
fastrpc_init_create_process, instead call fastrpc_map_put, to avoid
use-after-free. Do not remove it on fastrpc_device_release either,
call fastrpc_map_put instead.

The fastrpc_free_map is the only proper place to remove the map.
This is called only after the reference count is 0.

Fixes: b49f6d83e290 ("misc: fastrpc: Fix a possible double free")
Cc: stable <stable@kernel.org>
Co-developed-by: Ola Jeppsson <ola@snap.com>
Signed-off-by: Ola Jeppsson <ola@snap.com>
Signed-off-by: Abel Vesa <abel.vesa@linaro.org>
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Link: https://lore.kernel.org/r/20221124174941.418450-3-srinivas.kandagatla@linaro.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agomisc: fastrpc: Fix use-after-free and race in fastrpc_map_find
Abel Vesa [Thu, 24 Nov 2022 17:49:39 +0000 (17:49 +0000)]
misc: fastrpc: Fix use-after-free and race in fastrpc_map_find

commit 9446fa1683a7e3937d9970248ced427c1983a1c5 upstream.

Currently, there is a race window between the point when the mutex is
unlocked in fastrpc_map_lookup and the reference count increasing
(fastrpc_map_get) in fastrpc_map_find, which can also lead to
use-after-free.

So lets merge fastrpc_map_find into fastrpc_map_lookup which allows us
to both protect the maps list by also taking the &fl->lock spinlock and
the reference count, since the spinlock will be released only after.
Add take_ref argument to make this suitable for all callers.

Fixes: 8f6c1d8c4f0c ("misc: fastrpc: Add fdlist implementation")
Cc: stable <stable@kernel.org>
Co-developed-by: Ola Jeppsson <ola@snap.com>
Signed-off-by: Ola Jeppsson <ola@snap.com>
Signed-off-by: Abel Vesa <abel.vesa@linaro.org>
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Link: https://lore.kernel.org/r/20221124174941.418450-2-srinivas.kandagatla@linaro.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agousb: misc: onboard_hub: Move 'attach' work to the driver
Matthias Kaehlcke [Tue, 10 Jan 2023 17:32:53 +0000 (17:32 +0000)]
usb: misc: onboard_hub: Move 'attach' work to the driver

commit cde37881e2e14590675d0acdfbad408300d9ca95 upstream.

Currently each onboard_hub platform device owns an 'attach' work,
which is scheduled when the device probes. With this deadlocks
have been reported on a Raspberry Pi 3 B+ [1], which has nested
onboard hubs.

The flow of the deadlock is something like this (with the onboard_hub
driver built as a module) [2]:

- USB root hub is instantiated
- core hub driver calls onboard_hub_create_pdevs(), which creates the
  'raw' platform device for the 1st level hub
- 1st level hub is probed by the core hub driver
- core hub driver calls onboard_hub_create_pdevs(), which creates
  the 'raw' platform device for the 2nd level hub

- onboard_hub platform driver is registered
- platform device for 1st level hub is probed
  - schedules 'attach' work
- platform device for 2nd level hub is probed
  - schedules 'attach' work

- onboard_hub USB driver is registered
- device (and parent) lock of hub is held while the device is
  re-probed with the onboard_hub driver

- 'attach' work (running in another thread) calls driver_attach(), which
   blocks on one of the hub device locks

- onboard_hub_destroy_pdevs() is called by the core hub driver when one
  of the hubs is detached
- destroying the pdevs invokes onboard_hub_remove(), which waits for the
  'attach' work to complete
  - waits forever, since the 'attach' work can't acquire the device lock

Use a single work struct for the driver instead of having a work struct
per onboard hub platform driver instance. With that it isn't necessary
to cancel the work in onboard_hub_remove(), which fixes the deadlock.
The work is only cancelled when the driver is unloaded.

[1] https://lore.kernel.org/r/d04bcc45-3471-4417-b30b-5cf9880d785d@i2se.com/
[2] https://lore.kernel.org/all/Y6OrGbqaMy2iVDWB@google.com/

Cc: stable@vger.kernel.org
Fixes: 8bc063641ceb ("usb: misc: Add onboard_usb_hub driver")
Link: https://lore.kernel.org/r/d04bcc45-3471-4417-b30b-5cf9880d785d@i2se.com/
Link: https://lore.kernel.org/all/Y6OrGbqaMy2iVDWB@google.com/
Reported-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Link: https://lore.kernel.org/r/20230110172954.v2.2.I16b51f32db0c32f8a8532900bfe1c70c8572881a@changeid
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agousb: misc: onboard_hub: Invert driver registration order
Matthias Kaehlcke [Tue, 10 Jan 2023 17:32:52 +0000 (17:32 +0000)]
usb: misc: onboard_hub: Invert driver registration order

commit e5854355d76b8d768cea8e4fc3ce6dfdba25518a upstream.

The onboard_hub 'driver' consists of two drivers, a platform
driver and a USB driver. Currently when the onboard hub driver
is initialized it first registers the platform driver, then the
USB driver. This results in a race condition when the 'attach'
work is executed, which is scheduled when the platform device
is probed. The purpose of fhe 'attach' work is to bind elegible
USB hub devices to the onboard_hub USB driver. This fails if
the work runs before the USB driver has been registered.

Register the USB driver first, then the platform driver. This
increases the chances that the onboard_hub USB devices are probed
before their corresponding platform device, which the USB driver
tries to locate in _probe(). The driver already handles this
situation and defers probing if the onboard hub platform device
doesn't exist yet.

Cc: stable@vger.kernel.org
Fixes: 8bc063641ceb ("usb: misc: Add onboard_usb_hub driver")
Link: https://lore.kernel.org/lkml/Y6W00vQm3jfLflUJ@hovoldconsulting.com/T/#m0d64295f017942fd988f7c53425db302d61952b4
Reported-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Link: https://lore.kernel.org/r/20230110172954.v2.1.I75494ebee7027a50235ce4b1e930fa73a578fbe2@changeid
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoUSB: misc: iowarrior: fix up header size for USB_DEVICE_ID_CODEMERCS_IOW100
Greg Kroah-Hartman [Fri, 20 Jan 2023 13:53:30 +0000 (14:53 +0100)]
USB: misc: iowarrior: fix up header size for USB_DEVICE_ID_CODEMERCS_IOW100

commit 14ff7460bb58662d86aa50298943cc7d25532e28 upstream.

The USB_DEVICE_ID_CODEMERCS_IOW100 header size was incorrect, it should
be 12, not 13.

Cc: stable <stable@kernel.org>
Fixes: 17a82716587e ("USB: iowarrior: fix up report size handling for some devices")
Reported-by: Christoph Jung <jung@codemercs.com>
Link: https://lore.kernel.org/r/20230120135330.3842518-1-gregkh@linuxfoundation.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agostaging: vchiq_arm: fix enum vchiq_status return types
Arnd Bergmann [Tue, 17 Jan 2023 16:39:32 +0000 (17:39 +0100)]
staging: vchiq_arm: fix enum vchiq_status return types

commit 7d83299351fe7c812c529f5e39fe63b5312e4233 upstream.

gcc-13 notices a type mismatch between function declaration
and definition for a few functions that have been converted
from returning vchiq specific status values to regular error
codes:

drivers/staging/vc04_services/interface/vchiq_arm/vchiq_arm.c:662:5: error: conflicting types for 'vchiq_initialise' due to enum/integer mismatch; have 'int(struct vchiq_instance **)' [-Werror=enum-int-mismatch]
drivers/staging/vc04_services/interface/vchiq_arm/vchiq_arm.c:1411:1: error: conflicting types for 'vchiq_use_internal' due to enum/integer mismatch; have 'int(struct vchiq_state *, struct vchiq_service *, enum USE_TYPE_E)' [-Werror=enum-int-mismatch]
drivers/staging/vc04_services/interface/vchiq_arm/vchiq_arm.c:1468:1: error: conflicting types for 'vchiq_release_internal' due to enum/integer mismatch; have 'int(struct vchiq_state *, struct vchiq_service *)' [-Werror=enum-int-mismatch]

Change the declarations to match the actual function definition.

Fixes: a9fbd828be7f ("staging: vchiq_arm: drop enum vchiq_status from vchiq_*_internal")
Cc: stable <stable@kernel.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Link: https://lore.kernel.org/r/20230117163957.1109872-1-arnd@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoUSB: serial: option: add Quectel EM05CN modem
Duke Xin(辛安文) [Mon, 16 Jan 2023 02:33:28 +0000 (18:33 -0800)]
USB: serial: option: add Quectel EM05CN modem

commit 71dfd381a7c051f16a61f82fbd38a4cca563bdca upstream.

The EM05CN modem has 2 USB configurations that are configurable via the AT
command AT+QCFG="usbnet",[ 0 | 2 ] which make the modem enumerate with
the following interfaces, respectively:

"MBIM"  : AT + MBIM + DIAG + NMEA  + MODEM
"RMNET" : AT + DIAG + NMEA + Modem + QMI

The detailed description of the USB configuration for each mode as follows:

MBIM Mode
--------------
T:  Bus=01 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#=  2 Spd=480  MxCh= 0
D:  Ver= 2.00 Cls=ef(misc ) Sub=02 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=2c7c ProdID=0312 Rev= 3.18
S:  Manufacturer=Quectel
S:  Product=Quectel EM05-CN
C:* #Ifs= 6 Cfg#= 1 Atr=a0 MxPwr=500mA
A:  FirstIf#= 1 IfCount= 2 Cls=02(comm.) Sub=0e Prot=00
I:* If#= 3 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=option
E:  Ad=81(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=01(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 4 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=83(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=82(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 0 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=85(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=84(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=03(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 5 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=87(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=86(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=04(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 1 Alt= 0 #EPs= 1 Cls=02(comm.) Sub=0e Prot=00 Driver=cdc_mbim
E:  Ad=89(I) Atr=03(Int.) MxPS=  64 Ivl=32ms
I:  If#= 2 Alt= 0 #EPs= 0 Cls=0a(data ) Sub=00 Prot=02 Driver=cdc_mbim
I:* If#= 2 Alt= 1 #EPs= 2 Cls=0a(data ) Sub=00 Prot=02 Driver=cdc_mbim
E:  Ad=88(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=05(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms

RMNET Mode
--------------
T:  Bus=01 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#=  3 Spd=480  MxCh= 0
D:  Ver= 2.00 Cls=ef(misc ) Sub=02 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=2c7c ProdID=0312 Rev= 3.18
S:  Manufacturer=Quectel
S:  Product=Quectel EM05-CN
C:* #Ifs= 5 Cfg#= 1 Atr=a0 MxPwr=500mA
I:* If#= 3 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=option
E:  Ad=81(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=01(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 4 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=83(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=82(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 0 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=85(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=84(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=03(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 5 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=87(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=86(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=04(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 6 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=ff Prot=ff Driver=(none)
E:  Ad=89(I) Atr=03(Int.) MxPS=   8 Ivl=32ms
E:  Ad=88(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=05(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms

Signed-off-by: Duke Xin(辛安文) <duke_xinanwen@163.com>
Cc: stable@vger.kernel.org
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoUSB: serial: option: add Quectel EM05CN (SG) modem
Duke Xin(辛安文) [Mon, 16 Jan 2023 02:07:27 +0000 (18:07 -0800)]
USB: serial: option: add Quectel EM05CN (SG) modem

commit 1541dd0097c0f8f470e76eddf5120fc55a7e3101 upstream.

The EM05CN (SG) modem has 2 USB configurations that are configurable via the AT
command AT+QCFG="usbnet",[ 0 | 2 ] which make the modem enumerate with
the following interfaces, respectively:

"MBIM"  : AT + MBIM + DIAG + NMEA  + MODEM
"RMNET" : AT + DIAG + NMEA + Modem + QMI

The detailed description of the USB configuration for each mode as follows:

MBIM Mode
--------------
T:  Bus=01 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#=  2 Spd=480  MxCh= 0
D:  Ver= 2.00 Cls=ef(misc ) Sub=02 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=2c7c ProdID=0310 Rev= 3.18
S:  Manufacturer=Quectel
S:  Product=Quectel EM05-CN
C:* #Ifs= 6 Cfg#= 1 Atr=a0 MxPwr=500mA
A:  FirstIf#= 1 IfCount= 2 Cls=02(comm.) Sub=0e Prot=00
I:* If#= 3 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=option
E:  Ad=81(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=01(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 4 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=83(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=82(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 0 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=85(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=84(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=03(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 5 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=87(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=86(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=04(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 1 Alt= 0 #EPs= 1 Cls=02(comm.) Sub=0e Prot=00 Driver=cdc_mbim
E:  Ad=89(I) Atr=03(Int.) MxPS=  64 Ivl=32ms
I:  If#= 2 Alt= 0 #EPs= 0 Cls=0a(data ) Sub=00 Prot=02 Driver=cdc_mbim
I:* If#= 2 Alt= 1 #EPs= 2 Cls=0a(data ) Sub=00 Prot=02 Driver=cdc_mbim
E:  Ad=88(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=05(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms

RMNET Mode
--------------
T:  Bus=01 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#=  3 Spd=480  MxCh= 0
D:  Ver= 2.00 Cls=ef(misc ) Sub=02 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=2c7c ProdID=0310 Rev= 3.18
S:  Manufacturer=Quectel
S:  Product=Quectel EM05-CN
C:* #Ifs= 5 Cfg#= 1 Atr=a0 MxPwr=500mA
I:* If#= 3 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=option
E:  Ad=81(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=01(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 4 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=83(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=82(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 0 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=85(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=84(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=03(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 5 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=87(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=86(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=04(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 6 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=ff Prot=ff Driver=(none)
E:  Ad=89(I) Atr=03(Int.) MxPS=   8 Ivl=32ms
E:  Ad=88(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=05(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms

Signed-off-by: Duke Xin(辛安文) <duke_xinanwen@163.com>
Cc: stable@vger.kernel.org
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoUSB: serial: option: add Quectel EC200U modem
Ali Mirghasemi [Wed, 28 Dec 2022 11:38:47 +0000 (15:08 +0330)]
USB: serial: option: add Quectel EC200U modem

commit d9bbb15881046bd76f8710c76e26a740eee997ef upstream.

Add support for EC200U modem

0x0901: EC200U - AT + AP + CP + NMEA + DIAG + MOS

usb-device output:
T: Bus=01 Lev=02 Prnt=02 Port=02 Cnt=01 Dev#= 4 Spd=480 MxCh= 0
D: Ver= 2.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS=64 #Cfgs= 1
P: Vendor=2c7c ProdID=0901 Rev= 3.18
S: Manufacturer=Android
S: Product=Android
C:* #Ifs= 9 Cfg#= 1 Atr=e0 MxPwr=400mA
A: FirstIf#= 0 IfCount= 2 Cls=02(comm.) Sub=06 Prot=00
I:* If#= 0 Alt= 0 #EPs= 1 Cls=02(comm.) Sub=06 Prot=00 Driver=cdc_ether
E: Ad=81(I) Atr=03(Int.) MxPS= 16 Ivl=32ms
I: If#= 1 Alt= 0 #EPs= 0 Cls=0a(data ) Sub=00 Prot=00 Driver=cdc_ether
I:* If#= 1 Alt= 1 #EPs= 2 Cls=0a(data ) Sub=00 Prot=00 Driver=cdc_ether
E: Ad=82(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E: Ad=01(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 2 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E: Ad=84(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E: Ad=02(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E: Ad=83(I) Atr=03(Int.) MxPS= 512 Ivl=4096ms
I:* If#= 3 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E: Ad=85(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E: Ad=03(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 4 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E: Ad=86(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E: Ad=04(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 5 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E: Ad=87(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E: Ad=05(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 6 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E: Ad=88(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E: Ad=06(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 7 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E: Ad=8a(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E: Ad=07(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E: Ad=89(I) Atr=03(Int.) MxPS= 512 Ivl=4096ms
I:* If#= 8 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E: Ad=8b(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E: Ad=08(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms

Signed-off-by: Ali Mirghasemi <ali.mirghasemi1376@gmail.com>
Cc: stable@vger.kernel.org
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoUSB: serial: option: add Quectel EM05-G (RS) modem
Duke Xin(辛安文) [Tue, 27 Dec 2022 09:51:27 +0000 (01:51 -0800)]
USB: serial: option: add Quectel EM05-G (RS) modem

commit b72d13977689f0c717444010e108c4f20658dfee upstream.

The EM05-G (RS) modem has 2 USB configurations that are configurable via
the AT command AT+QCFG="usbnet",[ 0 | 2 ] which make the modem enumerate
with the following interfaces, respectively:

"RMNET" : AT + DIAG + NMEA + Modem + QMI
"MBIM"  : MBIM + AT + DIAG + NMEA + Modem

The detailed description of the USB configuration for each mode as follows:

RMNET Mode
--------------
T:  Bus=01 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#= 21 Spd=480  MxCh= 0
D:  Ver= 2.00 Cls=ef(misc ) Sub=02 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=2c7c ProdID=0314 Rev= 3.18
S:  Manufacturer=Quectel
S:  Product=Quectel EM05-G
C:* #Ifs= 5 Cfg#= 1 Atr=a0 MxPwr=500mA
I:* If#= 3 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=option
E:  Ad=81(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=01(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 4 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=83(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=82(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 2 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=85(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=84(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=03(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 5 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=87(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=86(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=04(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 6 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=ff Prot=ff Driver=(none)
E:  Ad=89(I) Atr=03(Int.) MxPS=   8 Ivl=32ms
E:  Ad=88(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=05(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms

MBIM Mode
--------------
T:  Bus=01 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#= 16 Spd=480  MxCh= 0
D:  Ver= 2.00 Cls=ef(misc ) Sub=02 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=2c7c ProdID=0314 Rev= 3.18
S:  Manufacturer=Quectel
S:  Product=Quectel EM05-G
C:* #Ifs= 6 Cfg#= 1 Atr=a0 MxPwr=500mA
A:  FirstIf#= 0 IfCount= 2 Cls=02(comm.) Sub=0e Prot=00
I:* If#= 3 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=option
E:  Ad=81(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=01(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 4 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=83(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=82(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 2 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=85(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=84(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=03(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 5 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=87(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=86(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=04(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 0 Alt= 0 #EPs= 1 Cls=02(comm.) Sub=0e Prot=00 Driver=cdc_mbim
E:  Ad=89(I) Atr=03(Int.) MxPS=  64 Ivl=32ms
I:  If#= 1 Alt= 0 #EPs= 0 Cls=0a(data ) Sub=00 Prot=02 Driver=cdc_mbim
I:* If#= 1 Alt= 1 #EPs= 2 Cls=0a(data ) Sub=00 Prot=02 Driver=cdc_mbim
E:  Ad=88(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=05(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms

Signed-off-by: Duke Xin(辛安文) <duke_xinanwen@163.com>
Cc: stable@vger.kernel.org
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoUSB: serial: option: add Quectel EM05-G (CS) modem
Duke Xin(辛安文) [Tue, 27 Dec 2022 09:28:25 +0000 (01:28 -0800)]
USB: serial: option: add Quectel EM05-G (CS) modem

commit bb78654b0b46316dac687fd4b7dc7cce636f46cd upstream.

The EM05-G (CS) modem has 2 USB configurations that are configurable via
the AT command AT+QCFG="usbnet",[ 0 | 2 ] which make the modem enumerate
with the following interfaces, respectively:

"RMNET" : AT + DIAG + NMEA + Modem + QMI
"MBIM"  : MBIM + AT + DIAG + NMEA + Modem

The detailed description of the USB configuration for each mode as follows:

RMNET Mode
--------------
T:  Bus=01 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#= 21 Spd=480  MxCh= 0
D:  Ver= 2.00 Cls=ef(misc ) Sub=02 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=2c7c ProdID=030C Rev= 3.18
S:  Manufacturer=Quectel
S:  Product=Quectel EM05-G
C:* #Ifs= 5 Cfg#= 1 Atr=a0 MxPwr=500mA
I:* If#= 3 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=option
E:  Ad=81(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=01(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 4 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=83(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=82(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 2 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=85(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=84(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=03(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 5 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=87(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=86(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=04(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 6 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=ff Prot=ff Driver=(none)
E:  Ad=89(I) Atr=03(Int.) MxPS=   8 Ivl=32ms
E:  Ad=88(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=05(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms

MBIM Mode
--------------
T:  Bus=01 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#= 16 Spd=480  MxCh= 0
D:  Ver= 2.00 Cls=ef(misc ) Sub=02 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=2c7c ProdID=030C Rev= 3.18
S:  Manufacturer=Quectel
S:  Product=Quectel EM05-G
C:* #Ifs= 6 Cfg#= 1 Atr=a0 MxPwr=500mA
A:  FirstIf#= 0 IfCount= 2 Cls=02(comm.) Sub=0e Prot=00
I:* If#= 3 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=option
E:  Ad=81(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=01(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 4 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=83(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=82(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 2 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=85(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=84(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=03(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 5 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=87(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=86(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=04(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 0 Alt= 0 #EPs= 1 Cls=02(comm.) Sub=0e Prot=00 Driver=cdc_mbim
E:  Ad=89(I) Atr=03(Int.) MxPS=  64 Ivl=32ms
I:  If#= 1 Alt= 0 #EPs= 0 Cls=0a(data ) Sub=00 Prot=02 Driver=cdc_mbim
I:* If#= 1 Alt= 1 #EPs= 2 Cls=0a(data ) Sub=00 Prot=02 Driver=cdc_mbim
E:  Ad=88(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=05(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms

Signed-off-by: Duke Xin(辛安文) <duke_xinanwen@163.com>
Cc: stable@vger.kernel.org
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoUSB: serial: option: add Quectel EM05-G (GR) modem
Duke Xin(辛安文) [Tue, 27 Dec 2022 09:44:30 +0000 (01:44 -0800)]
USB: serial: option: add Quectel EM05-G (GR) modem

commit 6c331f32e32ac71eb3e8b93fceda2802d7ecb889 upstream.

The EM05-G (GR) modem has 2 USB configurations that are configurable via
the AT command AT+QCFG="usbnet",[ 0 | 2 ] which make the modem enumerate
with the following interfaces, respectively:

"RMNET" : AT + DIAG + NMEA + Modem + QMI
"MBIM"  : MBIM + AT + DIAG + NMEA + Modem

The detailed description of the USB configuration for each mode as follows:

RMNET Mode
--------------
T:  Bus=01 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#= 21 Spd=480  MxCh= 0
D:  Ver= 2.00 Cls=ef(misc ) Sub=02 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=2c7c ProdID=0313 Rev= 3.18
S:  Manufacturer=Quectel
S:  Product=Quectel EM05-G
C:* #Ifs= 5 Cfg#= 1 Atr=a0 MxPwr=500mA
I:* If#= 3 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=option
E:  Ad=81(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=01(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 4 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=83(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=82(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 2 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=85(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=84(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=03(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 5 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=87(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=86(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=04(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 6 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=ff Prot=ff Driver=(none)
E:  Ad=89(I) Atr=03(Int.) MxPS=   8 Ivl=32ms
E:  Ad=88(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=05(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms

MBIM Mode
--------------
T:  Bus=01 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#= 16 Spd=480  MxCh= 0
D:  Ver= 2.00 Cls=ef(misc ) Sub=02 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=2c7c ProdID=0313 Rev= 3.18
S:  Manufacturer=Quectel
S:  Product=Quectel EM05-G
C:* #Ifs= 6 Cfg#= 1 Atr=a0 MxPwr=500mA
A:  FirstIf#= 0 IfCount= 2 Cls=02(comm.) Sub=0e Prot=00
I:* If#= 3 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=option
E:  Ad=81(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=01(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 4 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=83(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=82(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 2 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=85(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=84(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=03(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 5 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=87(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=86(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=04(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 0 Alt= 0 #EPs= 1 Cls=02(comm.) Sub=0e Prot=00 Driver=cdc_mbim
E:  Ad=89(I) Atr=03(Int.) MxPS=  64 Ivl=32ms
I:  If#= 1 Alt= 0 #EPs= 0 Cls=0a(data ) Sub=00 Prot=02 Driver=cdc_mbim
I:* If#= 1 Alt= 1 #EPs= 2 Cls=0a(data ) Sub=00 Prot=02 Driver=cdc_mbim
E:  Ad=88(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=05(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms

Signed-off-by: Duke Xin(辛安文) <duke_xinanwen@163.com>
Cc: stable@vger.kernel.org
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoprlimit: do_prlimit needs to have a speculation check
Greg Kroah-Hartman [Fri, 20 Jan 2023 10:03:20 +0000 (11:03 +0100)]
prlimit: do_prlimit needs to have a speculation check

commit 739790605705ddcf18f21782b9c99ad7d53a8c11 upstream.

do_prlimit() adds the user-controlled resource value to a pointer that
will subsequently be dereferenced.  In order to help prevent this
codepath from being used as a spectre "gadget" a barrier needs to be
added after checking the range.

Reported-by: Jordy Zomer <jordyzomer@google.com>
Tested-by: Jordy Zomer <jordyzomer@google.com>
Suggested-by: Linus Torvalds <torvalds@linuxfoundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoxhci: Detect lpm incapable xHC USB3 roothub ports from ACPI tables
Mathias Nyman [Mon, 16 Jan 2023 14:22:16 +0000 (16:22 +0200)]
xhci: Detect lpm incapable xHC USB3 roothub ports from ACPI tables

commit 74622f0a81d0c2bcfc39f9192b788124e8c7f0af upstream.

USB3 ports on xHC hosts may have retimers that cause too long
exit latency to work with native USB3 U1/U2 link power management states.

For now only use usb_acpi_port_lpm_incapable() to evaluate if port lpm
should be disabled while setting up the USB3 roothub.

Other ways to identify lpm incapable ports can be added here later if
ACPI _DSM does not exist.

Limit this to Intel hosts for now, this is to my knowledge only
an Intel issue.

Cc: stable@vger.kernel.org
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Link: https://lore.kernel.org/r/20230116142216.1141605-8-mathias.nyman@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agousb: acpi: add helper to check port lpm capability using acpi _DSM
Mathias Nyman [Mon, 16 Jan 2023 14:22:15 +0000 (16:22 +0200)]
usb: acpi: add helper to check port lpm capability using acpi _DSM

commit cd702d18c882d5a4ea44bbdb38edd5d5577ef640 upstream.

Add a helper to evaluate ACPI usb device specific method (_DSM) provided
in case the USB3 port shouldn't enter U1 and U2 link states.

This _DSM was added as port specific retimer configuration may lead to
exit latencies growing beyond U1/U2 exit limits, and OS needs a way to
find which ports can't support U1/U2 link power management states.

This _DSM is also used by windows:
Link: https://docs.microsoft.com/en-us/windows-hardware/drivers/bringup/usb-device-specific-method---dsm-
Some patch issues found in testing resolved by Ron Lee

Cc: stable@vger.kernel.org
Tested-by: Ron Lee <ron.lee@intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Link: https://lore.kernel.org/r/20230116142216.1141605-7-mathias.nyman@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoxhci: Add a flag to disable USB3 lpm on a xhci root port level.
Mathias Nyman [Mon, 16 Jan 2023 14:22:14 +0000 (16:22 +0200)]
xhci: Add a flag to disable USB3 lpm on a xhci root port level.

commit 0522b9a1653048440da5f21747f21e498b9220d1 upstream.

One USB3 roothub port may support link power management, while another
root port on the same xHC can't due to different retimers used for
the ports.

This is the case with Intel Alder Lake, and possible future platforms
where retimers used for USB4 ports cause too long exit latecy to
enable native USB3 lpm U1 and U2 states.

Add a flag in the xhci port structure to indicate if the port is
lpm_incapable, and check it while calculating exit latency.

Cc: stable@vger.kernel.org
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Link: https://lore.kernel.org/r/20230116142216.1141605-6-mathias.nyman@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoxhci: Add update_hub_device override for PCI xHCI hosts
Mathias Nyman [Mon, 16 Jan 2023 14:22:13 +0000 (16:22 +0200)]
xhci: Add update_hub_device override for PCI xHCI hosts

commit 23a3b8d5a2365653fd9bc5a9454d1e7f4facbf85 upstream.

Allow PCI hosts to check and tune roothub and port settings
before the hub is up and running.

This override is needed to turn off U1 and U2 LPM for some ports
based on per port ACPI _DSM, _UPC, or possibly vendor specific mmio
values for Intel xHC hosts.

Usb core calls the host update_hub_device once it creates a hub.

Entering U1 or U2 link power save state on ports with this limitation
will cause link to fail, turning the usb device unusable in that setup.

Cc: stable@vger.kernel.org
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Link: https://lore.kernel.org/r/20230116142216.1141605-5-mathias.nyman@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoxhci: Fix null pointer dereference when host dies
Mathias Nyman [Mon, 16 Jan 2023 14:22:12 +0000 (16:22 +0200)]
xhci: Fix null pointer dereference when host dies

commit a2bc47c43e70cf904b1af49f76d572326c08bca7 upstream.

Make sure xhci_free_dev() and xhci_kill_endpoint_urbs() do not race
and cause null pointer dereference when host suddenly dies.

Usb core may call xhci_free_dev() which frees the xhci->devs[slot_id]
virt device at the same time that xhci_kill_endpoint_urbs() tries to
loop through all the device's endpoints, checking if there are any
cancelled urbs left to give back.

hold the xhci spinlock while freeing the virt device

Cc: stable@vger.kernel.org
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Link: https://lore.kernel.org/r/20230116142216.1141605-4-mathias.nyman@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agousb: xhci: Check endpoint is valid before dereferencing it
Jimmy Hu [Mon, 16 Jan 2023 14:22:11 +0000 (16:22 +0200)]
usb: xhci: Check endpoint is valid before dereferencing it

commit e8fb5bc76eb86437ab87002d4a36d6da02165654 upstream.

When the host controller is not responding, all URBs queued to all
endpoints need to be killed. This can cause a kernel panic if we
dereference an invalid endpoint.

Fix this by using xhci_get_virt_ep() helper to find the endpoint and
checking if the endpoint is valid before dereferencing it.

[233311.853271] xhci-hcd xhci-hcd.1.auto: xHCI host controller not responding, assume dead
[233311.853393] Unable to handle kernel NULL pointer dereference at virtual address 00000000000000e8

[233311.853964] pc : xhci_hc_died+0x10c/0x270
[233311.853971] lr : xhci_hc_died+0x1ac/0x270

[233311.854077] Call trace:
[233311.854085]  xhci_hc_died+0x10c/0x270
[233311.854093]  xhci_stop_endpoint_command_watchdog+0x100/0x1a4
[233311.854105]  call_timer_fn+0x50/0x2d4
[233311.854112]  expire_timers+0xac/0x2e4
[233311.854118]  run_timer_softirq+0x300/0xabc
[233311.854127]  __do_softirq+0x148/0x528
[233311.854135]  irq_exit+0x194/0x1a8
[233311.854143]  __handle_domain_irq+0x164/0x1d0
[233311.854149]  gic_handle_irq.22273+0x10c/0x188
[233311.854156]  el1_irq+0xfc/0x1a8
[233311.854175]  lpm_cpuidle_enter+0x25c/0x418 [msm_pm]
[233311.854185]  cpuidle_enter_state+0x1f0/0x764
[233311.854194]  do_idle+0x594/0x6ac
[233311.854201]  cpu_startup_entry+0x7c/0x80
[233311.854209]  secondary_start_kernel+0x170/0x198

Fixes: 50e8725e7c42 ("xhci: Refactor command watchdog and fix split string.")
Cc: stable@vger.kernel.org
Signed-off-by: Jimmy Hu <hhhuuu@google.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Message-ID: <0fe978ed-8269-9774-1c40-f8a98c17e838@linux.intel.com>
Link: https://lore.kernel.org/r/20230116142216.1141605-3-mathias.nyman@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoxhci-pci: set the dma max_seg_size
Ricardo Ribalda [Mon, 16 Jan 2023 14:22:10 +0000 (16:22 +0200)]
xhci-pci: set the dma max_seg_size

commit 93915a4170e9defd56a767a18e6c4076f3d18609 upstream.

Allow devices to have dma operations beyond 64K, and avoid warnings such
as:

xhci_hcd 0000:00:14.0: mapping sg segment longer than device claims to support [len=98304] [max=65536]

Cc: stable@vger.kernel.org
Cc: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Ricardo Ribalda <ribalda@chromium.org>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Link: https://lore.kernel.org/r/20230116142216.1141605-2-mathias.nyman@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoRevert "serial: stm32: Merge hard IRQ and threaded IRQ handling into single IRQ handler"
Greg Kroah-Hartman [Fri, 20 Jan 2023 10:16:59 +0000 (11:16 +0100)]
Revert "serial: stm32: Merge hard IRQ and threaded IRQ handling into single IRQ handler"

commit 2cbafffbf69addd7509072f4be5917f81d238cf6 upstream.

This reverts commit f24771b62a83239f0dce816bddf0f6807f436235 as it is
reported to break the build.

Reported-by: kernel test robot <lkp@intel.com>
Link: https://lore.kernel.org/r/202301200130.ttBiTzfO-lkp@intel.com
Fixes: f24771b62a83 ("serial: stm32: Merge hard IRQ and threaded IRQ handling into single IRQ handler")
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Valentin Caron <valentin.caron@foss.st.com> # V3
Cc: Marek Vasut <marex@denx.de>
Cc: Johan Hovold <johan@kernel.org>
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoserial: stm32: Merge hard IRQ and threaded IRQ handling into single IRQ handler
Marek Vasut [Thu, 12 Jan 2023 18:04:17 +0000 (19:04 +0100)]
serial: stm32: Merge hard IRQ and threaded IRQ handling into single IRQ handler

commit f24771b62a83239f0dce816bddf0f6807f436235 upstream.

Requesting an interrupt with IRQF_ONESHOT will run the primary handler
in the hard-IRQ context even in the force-threaded mode. The
force-threaded mode is used by PREEMPT_RT in order to avoid acquiring
sleeping locks (spinlock_t) in hard-IRQ context. This combination
makes it impossible and leads to "sleeping while atomic" warnings.

Use one interrupt handler for both handlers (primary and secondary)
and drop the IRQF_ONESHOT flag which is not needed.

Fixes: e359b4411c283 ("serial: stm32: fix threaded interrupt handling")
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Tested-by: Valentin Caron <valentin.caron@foss.st.com> # V3
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20230112180417.25595-1-marex@denx.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agomm/khugepaged: fix collapse_pte_mapped_thp() to allow anon_vma
Hugh Dickins [Thu, 22 Dec 2022 20:41:50 +0000 (12:41 -0800)]
mm/khugepaged: fix collapse_pte_mapped_thp() to allow anon_vma

commit ab0c3f1251b4670978fde0bd54161795a139b060 upstream.

uprobe_write_opcode() uses collapse_pte_mapped_thp() to restore huge pmd,
when removing a breakpoint from hugepage text: vma->anon_vma is always set
in that case, so undo the prohibition.  And MADV_COLLAPSE ought to be able
to collapse some page tables in a vma which happens to have anon_vma set
from CoWing elsewhere.

Is anon_vma lock required?  Almost not: if any page other than expected
subpage of the non-anon huge page is found in the page table, collapse is
aborted without making any change.  However, it is possible that an anon
page was CoWed from this extent in another mm or vma, in which case a
concurrent lookup might look here: so keep it away while clearing pmd (but
perhaps we shall go back to using pmd_lock() there in future).

Note that collapse_pte_mapped_thp() is exceptional in freeing a page table
without having cleared its ptes: I'm uneasy about that, and had thought
pte_clear()ing appropriate; but exclusive i_mmap lock does fix the
problem, and we would have to move the mmu_notification if clearing those
ptes.

What this fixes is not a dangerous instability.  But I suggest Cc stable
because uprobes "healing" has regressed in that way, so this should follow
8d3c106e19e8 into those stable releases where it was backported (and may
want adjustment there - I'll supply backports as needed).

Link: https://lkml.kernel.org/r/b740c9fb-edba-92ba-59fb-7a5592e5dfc@google.com
Fixes: 8d3c106e19e8 ("mm/khugepaged: take the right locks for page table retraction")
Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Jann Horn <jannh@google.com>
Cc: Yang Shi <shy828301@gmail.com>
Cc: Zach O'Keefe <zokeefe@google.com>
Cc: Song Liu <songliubraving@fb.com>
Cc: <stable@vger.kernel.org> [5.4+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agohugetlb: unshare some PMDs when splitting VMAs
James Houghton [Wed, 4 Jan 2023 23:19:10 +0000 (23:19 +0000)]
hugetlb: unshare some PMDs when splitting VMAs

commit b30c14cd61025eeea2f2e8569606cd167ba9ad2d upstream.

PMD sharing can only be done in PUD_SIZE-aligned pieces of VMAs; however,
it is possible that HugeTLB VMAs are split without unsharing the PMDs
first.

Without this fix, it is possible to hit the uffd-wp-related WARN_ON_ONCE
in hugetlb_change_protection [1].  The key there is that
hugetlb_unshare_all_pmds will not attempt to unshare PMDs in
non-PUD_SIZE-aligned sections of the VMA.

It might seem ideal to unshare in hugetlb_vm_op_open, but we need to
unshare in both the new and old VMAs, so unsharing in hugetlb_vm_op_split
seems natural.

[1]: https://lore.kernel.org/linux-mm/CADrL8HVeOkj0QH5VZZbRzybNE8CG-tEGFshnA+bG9nMgcWtBSg@mail.gmail.com/

Link: https://lkml.kernel.org/r/20230104231910.1464197-1-jthoughton@google.com
Fixes: 6dfeaff93be1 ("hugetlb/userfaultfd: unshare all pmds for hugetlbfs when register wp")
Signed-off-by: James Houghton <jthoughton@google.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Acked-by: Peter Xu <peterx@redhat.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: Muchun Song <songmuchun@bytedance.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agomm/shmem: restore SHMEM_HUGE_DENY precedence over MADV_COLLAPSE
Zach O'Keefe [Sat, 24 Dec 2022 08:20:35 +0000 (00:20 -0800)]
mm/shmem: restore SHMEM_HUGE_DENY precedence over MADV_COLLAPSE

commit 3de0c269adc6c2fac0bb1fb11965f0de699dc32b upstream.

SHMEM_HUGE_DENY is for emergency use by the admin, to disable allocation
of shmem huge pages if, for example, a dangerous bug is found in their
usage: see "deny" in Documentation/mm/transhuge.rst.  An app using
madvise(,,MADV_COLLAPSE) should not be allowed to override it: restore its
precedence over shmem_huge_force.

Restore SHMEM_HUGE_DENY precedence over MADV_COLLAPSE.

Link: https://lkml.kernel.org/r/20221224082035.3197140-2-zokeefe@google.com
Fixes: 7c6c6cc4d3a2 ("mm/shmem: add flag to enforce shmem THP in hugepage_vma_check()")
Signed-off-by: Zach O'Keefe <zokeefe@google.com>
Suggested-by: Hugh Dickins <hughd@google.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Yang Shi <shy828301@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agonilfs2: fix general protection fault in nilfs_btree_insert()
Ryusuke Konishi [Thu, 5 Jan 2023 05:53:56 +0000 (14:53 +0900)]
nilfs2: fix general protection fault in nilfs_btree_insert()

commit 7633355e5c7f29c049a9048e461427d1d8ed3051 upstream.

If nilfs2 reads a corrupted disk image and tries to reads a b-tree node
block by calling __nilfs_btree_get_block() against an invalid virtual
block address, it returns -ENOENT because conversion of the virtual block
address to a disk block address fails.  However, this return value is the
same as the internal code that b-tree lookup routines return to indicate
that the block being searched does not exist, so functions that operate on
that b-tree may misbehave.

When nilfs_btree_insert() receives this spurious 'not found' code from
nilfs_btree_do_lookup(), it misunderstands that the 'not found' check was
successful and continues the insert operation using incomplete lookup path
data, causing the following crash:

 general protection fault, probably for non-canonical address
 0xdffffc0000000005: 0000 [#1] PREEMPT SMP KASAN
 KASAN: null-ptr-deref in range [0x0000000000000028-0x000000000000002f]
 ...
 RIP: 0010:nilfs_btree_get_nonroot_node fs/nilfs2/btree.c:418 [inline]
 RIP: 0010:nilfs_btree_prepare_insert fs/nilfs2/btree.c:1077 [inline]
 RIP: 0010:nilfs_btree_insert+0x6d3/0x1c10 fs/nilfs2/btree.c:1238
 Code: bc 24 80 00 00 00 4c 89 f8 48 c1 e8 03 42 80 3c 28 00 74 08 4c 89
 ff e8 4b 02 92 fe 4d 8b 3f 49 83 c7 28 4c 89 f8 48 c1 e8 03 <42> 80 3c
 28 00 74 08 4c 89 ff e8 2e 02 92 fe 4d 8b 3f 49 83 c7 02
 ...
 Call Trace:
 <TASK>
  nilfs_bmap_do_insert fs/nilfs2/bmap.c:121 [inline]
  nilfs_bmap_insert+0x20d/0x360 fs/nilfs2/bmap.c:147
  nilfs_get_block+0x414/0x8d0 fs/nilfs2/inode.c:101
  __block_write_begin_int+0x54c/0x1a80 fs/buffer.c:1991
  __block_write_begin fs/buffer.c:2041 [inline]
  block_write_begin+0x93/0x1e0 fs/buffer.c:2102
  nilfs_write_begin+0x9c/0x110 fs/nilfs2/inode.c:261
  generic_perform_write+0x2e4/0x5e0 mm/filemap.c:3772
  __generic_file_write_iter+0x176/0x400 mm/filemap.c:3900
  generic_file_write_iter+0xab/0x310 mm/filemap.c:3932
  call_write_iter include/linux/fs.h:2186 [inline]
  new_sync_write fs/read_write.c:491 [inline]
  vfs_write+0x7dc/0xc50 fs/read_write.c:584
  ksys_write+0x177/0x2a0 fs/read_write.c:637
  do_syscall_x64 arch/x86/entry/common.c:50 [inline]
  do_syscall_64+0x3d/0xb0 arch/x86/entry/common.c:80
  entry_SYSCALL_64_after_hwframe+0x63/0xcd
 ...
 </TASK>

This patch fixes the root cause of this problem by replacing the error
code that __nilfs_btree_get_block() returns on block address conversion
failure from -ENOENT to another internal code -EINVAL which means that the
b-tree metadata is corrupted.

By returning -EINVAL, it propagates without glitches, and for all relevant
b-tree operations, functions in the upper bmap layer output an error
message indicating corrupted b-tree metadata via
nilfs_bmap_convert_error(), and code -EIO will be eventually returned as
it should be.

Link: https://lkml.kernel.org/r/000000000000bd89e205f0e38355@google.com
Link: https://lkml.kernel.org/r/20230105055356.8811-1-konishi.ryusuke@gmail.com
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Reported-by: syzbot+ede796cecd5296353515@syzkaller.appspotmail.com
Tested-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agozonefs: Detect append writes at invalid locations
Damien Le Moal [Fri, 6 Jan 2023 08:43:06 +0000 (17:43 +0900)]
zonefs: Detect append writes at invalid locations

commit a608da3bd730d718f2d3ebec1c26f9865f8f17ce upstream.

Using REQ_OP_ZONE_APPEND operations for synchronous writes to sequential
files succeeds regardless of the zone write pointer position, as long as
the target zone is not full. This means that if an external (buggy)
application writes to the zone of a sequential file underneath the file
system, subsequent file write() operation will succeed but the file size
will not be correct and the file will contain invalid data written by
another application.

Modify zonefs_file_dio_append() to check the written sector of an append
write (returned in bio->bi_iter.bi_sector) and return -EIO if there is a
mismatch with the file zone wp offset field. This change triggers a call
to zonefs_io_error() and a zone check. Modify zonefs_io_error_cb() to
not expose the unexpected data after the current inode size when the
errors=remount-ro mode is used. Other error modes are correctly handled
already.

Fixes: 02ef12a663c7 ("zonefs: use REQ_OP_ZONE_APPEND for sync DIO")
Cc: stable@vger.kernel.org
Signed-off-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoLoongArch: Add HWCAP_LOONGARCH_CPUCFG to elf_hwcap
Huacai Chen [Tue, 17 Jan 2023 03:42:16 +0000 (11:42 +0800)]
LoongArch: Add HWCAP_LOONGARCH_CPUCFG to elf_hwcap

commit d52fec86a465355b379e839fa372ead0334d62e6 upstream.

HWCAP_LOONGARCH_CPUCFG is missing in elf_hwcap, so add it for glibc's
later use.

Cc: stable@vger.kernel.org
Reported-by: Yinyu Cai <caiyinyu@loongson.cn>
Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoAdd exception protection processing for vd in axi_chan_handle_err function
Shawn.Shao [Thu, 12 Jan 2023 05:58:02 +0000 (13:58 +0800)]
Add exception protection processing for vd in axi_chan_handle_err function

commit 57054fe516d59d03a7bcf1888e82479ccc244f87 upstream.

Since there is no protection for vd, a kernel panic will be
triggered here in exceptional cases.

You can refer to the processing of axi_chan_block_xfer_complete function

The triggered kernel panic is as follows:

[   67.848444] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000060
[   67.848447] Mem abort info:
[   67.848449]   ESR = 0x96000004
[   67.848451]   EC = 0x25: DABT (current EL), IL = 32 bits
[   67.848454]   SET = 0, FnV = 0
[   67.848456]   EA = 0, S1PTW = 0
[   67.848458] Data abort info:
[   67.848460]   ISV = 0, ISS = 0x00000004
[   67.848462]   CM = 0, WnR = 0
[   67.848465] user pgtable: 4k pages, 48-bit VAs, pgdp=00000800c4c0b000
[   67.848468] [0000000000000060] pgd=0000000000000000, p4d=0000000000000000
[   67.848472] Internal error: Oops: 96000004 [#1] SMP
[   67.848475] Modules linked in: dmatest
[   67.848479] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 5.10.100-emu_x2rc+ #11
[   67.848483] pstate: 62000085 (nZCv daIf -PAN -UAO +TCO BTYPE=--)
[   67.848487] pc : axi_chan_handle_err+0xc4/0x230
[   67.848491] lr : axi_chan_handle_err+0x30/0x230
[   67.848493] sp : ffff0803fe55ae50
[   67.848495] x29: ffff0803fe55ae50 x28: ffff800011212200
[   67.848500] x27: ffff0800c42c0080 x26: ffff0800c097c080
[   67.848504] x25: ffff800010d33880 x24: ffff80001139d850
[   67.848508] x23: ffff0800c097c168 x22: 0000000000000000
[   67.848512] x21: 0000000000000080 x20: 0000000000002000
[   67.848517] x19: ffff0800c097c080 x18: 0000000000000000
[   67.848521] x17: 0000000000000000 x16: 0000000000000000
[   67.848525] x15: 0000000000000000 x14: 0000000000000000
[   67.848529] x13: 0000000000000000 x12: 0000000000000040
[   67.848533] x11: ffff0800c0400248 x10: ffff0800c040024a
[   67.848538] x9 : ffff800010576cd4 x8 : ffff0800c0400270
[   67.848542] x7 : 0000000000000000 x6 : ffff0800c04003e0
[   67.848546] x5 : ffff0800c0400248 x4 : ffff0800c4294480
[   67.848550] x3 : dead000000000100 x2 : dead000000000122
[   67.848555] x1 : 0000000000000100 x0 : ffff0800c097c168
[   67.848559] Call trace:
[   67.848562]  axi_chan_handle_err+0xc4/0x230
[   67.848566]  dw_axi_dma_interrupt+0xf4/0x590
[   67.848569]  __handle_irq_event_percpu+0x60/0x220
[   67.848573]  handle_irq_event+0x64/0x120
[   67.848576]  handle_fasteoi_irq+0xc4/0x220
[   67.848580]  __handle_domain_irq+0x80/0xe0
[   67.848583]  gic_handle_irq+0xc0/0x138
[   67.848585]  el1_irq+0xc8/0x180
[   67.848588]  arch_cpu_idle+0x14/0x2c
[   67.848591]  default_idle_call+0x40/0x16c
[   67.848594]  do_idle+0x1f0/0x250
[   67.848597]  cpu_startup_entry+0x2c/0x60
[   67.848600]  rest_init+0xc0/0xcc
[   67.848603]  arch_call_rest_init+0x14/0x1c
[   67.848606]  start_kernel+0x4cc/0x500
[   67.848610] Code: eb0002ff 9a9f12d6 f2fbd5a2 f2fbd5a3 (a94602c1)
[   67.848613] ---[ end trace 585a97036f88203a ]---

Signed-off-by: Shawn.Shao <shawn.shao@jaguarmicro.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20230112055802.1764-1-shawn.shao@jaguarmicro.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoproc: fix PIE proc-empty-vm, proc-pid-vm tests
Alexey Dobriyan [Fri, 6 Jan 2023 19:30:14 +0000 (22:30 +0300)]
proc: fix PIE proc-empty-vm, proc-pid-vm tests

commit 5316a017d093f644675a56523bcf5787ba8f4fef upstream.

vsyscall detection code uses direct call to the beginning of
the vsyscall page:

asm ("call %P0" :: "i" (0xffffffffff600000))

It generates "call rel32" instruction but it is not relocated if binary
is PIE, so binary segfaults into random userspace address and vsyscall
page status is detected incorrectly.

Do more direct:

asm ("call *%rax")

which doesn't do need any relocaltions.

Mark g_vsyscall as volatile for a good measure, I didn't find instruction
setting it to 0. Now the code is obviously correct:

xor eax, eax
mov rdi, rbp
mov rsi, rbp
mov DWORD PTR [rip+0x2d15], eax      # g_vsyscall = 0
mov rax, 0xffffffffff600000
call rax
mov DWORD PTR [rip+0x2d02], 1        # g_vsyscall = 1
mov eax, DWORD PTR ds:0xffffffffff600000
mov DWORD PTR [rip+0x2cf1], 2        # g_vsyscall = 2
mov edi, [rip+0x2ceb]                # exit(g_vsyscall)
call exit

Note: fixed proc-empty-vm test oopses 5.19.0-28-generic kernel
but this is separate story.

Link: https://lkml.kernel.org/r/Y7h2xvzKLg36DSq8@p183
Fixes: 5bc73bb3451b9 ("proc: test how it holds up with mapping'less process")
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Reported-by: Mirsad Goran Todorovac <mirsad.todorovac@alu.unizg.hr>
Tested-by: Mirsad Goran Todorovac <mirsad.todorovac@alu.unizg.hr>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agonommu: fix split_vma() map_count error
Liam Howlett [Mon, 9 Jan 2023 20:58:20 +0000 (20:58 +0000)]
nommu: fix split_vma() map_count error

commit fd9edbdbdcde6b489ce59f326755ef16a2ffadd7 upstream.

During the maple tree conversion of nommu, an error in counting the VMAs
was introduced by counting the existing VMA again.  The counting used to
be decremented by one and incremented by two, but now it only increments
by two.  Fix the counting error by moving the increment outside the
setup_vma_to_mm() function to the callers.

Link: https://lkml.kernel.org/r/20230109205809.956325-1-Liam.Howlett@oracle.com
Fixes: 8220543df148 ("nommu: remove uses of VMA linked list")
Signed-off-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yu Zhao <yuzhao@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agonommu: fix do_munmap() error path
Liam Howlett [Mon, 9 Jan 2023 20:57:21 +0000 (20:57 +0000)]
nommu: fix do_munmap() error path

commit 80be727ec87225797771a39f3e6801baf291faaf upstream.

When removing a VMA from the tree fails due to no memory, do not free the
VMA since a reference still exists.

Link: https://lkml.kernel.org/r/20230109205708.956103-1-Liam.Howlett@oracle.com
Fixes: 8220543df148 ("nommu: remove uses of VMA linked list")
Signed-off-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yu Zhao <yuzhao@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agonommu: fix memory leak in do_mmap() error path
Liam Howlett [Mon, 9 Jan 2023 20:55:21 +0000 (20:55 +0000)]
nommu: fix memory leak in do_mmap() error path

commit 7f31cced5724e6d414fe750aa1cd7e7b578ec22f upstream.

The preallocation of the maple tree nodes may leak if the error path to
"error_just_free" is taken.  Fix this by moving the freeing of the maple
tree nodes to a shared location for all error paths.

Link: https://lkml.kernel.org/r/20230109205507.955577-1-Liam.Howlett@oracle.com
Fixes: 8220543df148 ("nommu: remove uses of VMA linked list")
Signed-off-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yu Zhao <yuzhao@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agowifi: mac80211: fix initialization of rx->link and rx->link_sta
Felix Fietkau [Fri, 30 Dec 2022 20:07:47 +0000 (21:07 +0100)]
wifi: mac80211: fix initialization of rx->link and rx->link_sta

commit e66b7920aa5ac5b1a1997a454004ba9246a3c005 upstream.

There are some codepaths that do not initialize rx->link_sta properly. This
causes a crash in places which assume that rx->link_sta is valid if rx->sta
is valid.
One known instance is triggered by __ieee80211_rx_h_amsdu being called from
fast-rx. It results in a crash like this one:

 BUG: kernel NULL pointer dereference, address: 00000000000000a8
 #PF: supervisor write access in kernel mode
 #PF: error_code(0x0002) - not-present page PGD 0 P4D 0
 Oops: 0002 [#1] PREEMPT SMP PTI
 CPU: 1 PID: 506 Comm: mt76-usb-rx phy Tainted: G            E      6.1.0-debian64x+1.7 #3
 Hardware name: ZOTAC ZBOX-ID92/ZBOX-IQ01/ZBOX-ID92/ZBOX-IQ01, BIOS B220P007 05/21/2014
 RIP: 0010:ieee80211_deliver_skb+0x62/0x1f0 [mac80211]
 Code: 00 48 89 04 24 e8 9e a7 c3 df 89 c0 48 03 1c c5 a0 ea 39 a1 4c 01 6b 08 48 ff 03 48
       83 7d 28 00 74 11 48 8b 45 30 48 63 55 44 <48> 83 84 d0 a8 00 00 00 01 41 8b 86 c0
       11 00 00 8d 50 fd 83 fa 01
 RSP: 0018:ffff999040803b10 EFLAGS: 00010286
 RAX: 0000000000000000 RBX: ffffb9903f496480 RCX: 0000000000000000
 RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000
 RBP: ffff999040803ce0 R08: 0000000000000000 R09: 0000000000000000
 R10: 0000000000000000 R11: 0000000000000000 R12: ffff8d21828ac900
 R13: 000000000000004a R14: ffff8d2198ed89c0 R15: ffff8d2198ed8000
 FS:  0000000000000000(0000) GS:ffff8d24afe80000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: 00000000000000a8 CR3: 0000000429810002 CR4: 00000000001706e0
 Call Trace:
  <TASK>
  __ieee80211_rx_h_amsdu+0x1b5/0x240 [mac80211]
  ? ieee80211_prepare_and_rx_handle+0xcdd/0x1320 [mac80211]
  ? __local_bh_enable_ip+0x3b/0xa0
  ieee80211_prepare_and_rx_handle+0xcdd/0x1320 [mac80211]
  ? prepare_transfer+0x109/0x1a0 [xhci_hcd]
  ieee80211_rx_list+0xa80/0xda0 [mac80211]
  mt76_rx_complete+0x207/0x2e0 [mt76]
  mt76_rx_poll_complete+0x357/0x5a0 [mt76]
  mt76u_rx_worker+0x4f5/0x600 [mt76_usb]
  ? mt76_get_min_avg_rssi+0x140/0x140 [mt76]
  __mt76_worker_fn+0x50/0x80 [mt76]
  kthread+0xed/0x120
  ? kthread_complete_and_exit+0x20/0x20
  ret_from_fork+0x22/0x30

Since the initialization of rx->link and rx->link_sta is rather convoluted
and duplicated in many places, clean it up by using a helper function to
set it.

Fixes: ccdde7c74ffd ("wifi: mac80211: properly implement MLO key handling")
Fixes: b320d6c456ff ("wifi: mac80211: use correct rx link_sta instead of default")
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20221230200747.19040-1-nbd@nbd.name
[remove unnecessary rx->sta->sta.mlo check]
Cc: stable@vger.kernel.org
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agowifi: mac80211: sdata can be NULL during AMPDU start
Alexander Wetzel [Fri, 30 Dec 2022 12:18:50 +0000 (13:18 +0100)]
wifi: mac80211: sdata can be NULL during AMPDU start

commit 69403bad97aa0162e3d7911b27e25abe774093df upstream.

ieee80211_tx_ba_session_handle_start() may get NULL for sdata when a
deauthentication is ongoing.

Here a trace triggering the race with the hostapd test
multi_ap_fronthaul_on_ap:

(gdb) list *drv_ampdu_action+0x46
0x8b16 is in drv_ampdu_action (net/mac80211/driver-ops.c:396).
391             int ret = -EOPNOTSUPP;
392
393             might_sleep();
394
395             sdata = get_bss_sdata(sdata);
396             if (!check_sdata_in_driver(sdata))
397                     return -EIO;
398
399             trace_drv_ampdu_action(local, sdata, params);
400

wlan0: moving STA 02:00:00:00:03:00 to state 3
wlan0: associated
wlan0: deauthenticating from 02:00:00:00:03:00 by local choice (Reason: 3=DEAUTH_LEAVING)
wlan3.sta1: Open BA session requested for 02:00:00:00:00:00 tid 0
wlan3.sta1: dropped frame to 02:00:00:00:00:00 (unauthorized port)
wlan0: moving STA 02:00:00:00:03:00 to state 2
wlan0: moving STA 02:00:00:00:03:00 to state 1
wlan0: Removed STA 02:00:00:00:03:00
wlan0: Destroyed STA 02:00:00:00:03:00
BUG: unable to handle page fault for address: fffffffffffffb48
PGD 11814067 P4D 11814067 PUD 11816067 PMD 0
Oops: 0000 [#1] PREEMPT SMP PTI
CPU: 2 PID: 133397 Comm: kworker/u16:1 Tainted: G        W          6.1.0-rc8-wt+ #59
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.0-20220807_005459-localhost 04/01/2014
Workqueue: phy3 ieee80211_ba_session_work [mac80211]
RIP: 0010:drv_ampdu_action+0x46/0x280 [mac80211]
Code: 53 48 89 f3 be 89 01 00 00 e8 d6 43 bf ef e8 21 46 81 f0 83 bb a0 1b 00 00 04 75 0e 48 8b 9b 28 0d 00 00 48 81 eb 10 0e 00 00 <8b> 93 58 09 00 00 f6 c2 20 0f 84 3b 01 00 00 8b 05 dd 1c 0f 00 85
RSP: 0018:ffffc900025ebd20 EFLAGS: 00010287
RAX: 0000000000000000 RBX: fffffffffffff1f0 RCX: ffff888102228240
RDX: 0000000080000000 RSI: ffffffff918c5de0 RDI: ffff888102228b40
RBP: ffffc900025ebd40 R08: 0000000000000001 R09: 0000000000000001
R10: 0000000000000001 R11: 0000000000000000 R12: ffff888118c18ec0
R13: 0000000000000000 R14: ffffc900025ebd60 R15: ffff888018b7efb8
FS:  0000000000000000(0000) GS:ffff88817a600000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: fffffffffffffb48 CR3: 0000000105228006 CR4: 0000000000170ee0
Call Trace:
 <TASK>
 ieee80211_tx_ba_session_handle_start+0xd0/0x190 [mac80211]
 ieee80211_ba_session_work+0xff/0x2e0 [mac80211]
 process_one_work+0x29f/0x620
 worker_thread+0x4d/0x3d0
 ? process_one_work+0x620/0x620
 kthread+0xfb/0x120
 ? kthread_complete_and_exit+0x20/0x20
 ret_from_fork+0x22/0x30
 </TASK>

Signed-off-by: Alexander Wetzel <alexander@wetzel-home.de>
Link: https://lore.kernel.org/r/20221230121850.218810-2-alexander@wetzel-home.de
Cc: stable@vger.kernel.org
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agowifi: mac80211: reset multiple BSSID options in stop_ap()
Aloka Dixit [Wed, 21 Dec 2022 18:56:16 +0000 (10:56 -0800)]
wifi: mac80211: reset multiple BSSID options in stop_ap()

commit 0eb38842ada035d71bb06fb9116f26f24ee0f998 upstream.

Reset multiple BSSID options when all AP related configurations are
reset in ieee80211_stop_ap().

Stale values result in HWSIM test failures (e.g. p2p_group_cli_invalid),
if run after 'he_ap_ema'.

Reported-by: Jouni Malinen <j@w1.fi>
Signed-off-by: Aloka Dixit <quic_alokad@quicinc.com>
Link: https://lore.kernel.org/r/20221221185616.11514-1-quic_alokad@quicinc.com
Cc: stable@vger.kernel.org
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agowifi: mac80211: fix MLO + AP_VLAN check
Felix Fietkau [Wed, 14 Dec 2022 13:03:26 +0000 (14:03 +0100)]
wifi: mac80211: fix MLO + AP_VLAN check

commit f216033d770f7ca0eda491fe01a9f02e7af59576 upstream.

Instead of preventing adding AP_VLAN to MLO enabled APs, this check was
preventing adding more than one 4-addr AP_VLAN regardless of the MLO status.
Fix this by adding missing extra checks.

Fixes: ae960ee90bb1 ("wifi: mac80211: prevent VLANs on MLDs")
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20221214130326.37756-1-nbd@nbd.name
Cc: stable@vger.kernel.org
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agowifi: brcmfmac: fix regression for Broadcom PCIe wifi devices
Arend van Spriel [Wed, 11 Jan 2023 11:24:19 +0000 (12:24 +0100)]
wifi: brcmfmac: fix regression for Broadcom PCIe wifi devices

commit ed05cb177ae5cd7f02f1d6e7706ba627d30f1696 upstream.

A sanity check was introduced considering maximum flowrings above
256 as insane and effectively aborting the device probe. This
resulted in regression for number of users as the value turns out
to be sane after all.

Fixes: 2aca4f3734bd ("brcmfmac: return error when getting invalid max_flowrings from dongle")
Reported-by: chainofflowers <chainofflowers@posteo.net>
Link: https://lore.kernel.org/all/4781984.GXAFRqVoOG@luna/
Reported-by: Christian Marillat <marillat@debian.org>
Link: https://bugzilla.kernel.org/show_bug.cgi?id=216894
Cc: stable@vger.kernel.org
Signed-off-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230111112419.24185-1-arend.vanspriel@broadcom.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoBluetooth: hci_qca: Fix driver shutdown on closed serdev
Krzysztof Kozlowski [Thu, 29 Dec 2022 10:28:29 +0000 (11:28 +0100)]
Bluetooth: hci_qca: Fix driver shutdown on closed serdev

commit 272970be3dabd24cbe50e393ffee8f04aec3b9a8 upstream.

The driver shutdown callback (which sends EDL_SOC_RESET to the device
over serdev) should not be invoked when HCI device is not open (e.g. if
hci_dev_open_sync() failed), because the serdev and its TTY are not open
either.  Also skip this step if device is powered off
(qca_power_shutdown()).

The shutdown callback causes use-after-free during system reboot with
Qualcomm Atheros Bluetooth:

  Unable to handle kernel paging request at virtual address
  0072662f67726fd7
  ...
  CPU: 6 PID: 1 Comm: systemd-shutdow Tainted: G        W
  6.1.0-rt5-00325-g8a5f56bcfcca #8
  Hardware name: Qualcomm Technologies, Inc. Robotics RB5 (DT)
  Call trace:
   tty_driver_flush_buffer+0x4/0x30
   serdev_device_write_flush+0x24/0x34
   qca_serdev_shutdown+0x80/0x130 [hci_uart]
   device_shutdown+0x15c/0x260
   kernel_restart+0x48/0xac

KASAN report:

  BUG: KASAN: use-after-free in tty_driver_flush_buffer+0x1c/0x50
  Read of size 8 at addr ffff16270c2e0018 by task systemd-shutdow/1

  CPU: 7 PID: 1 Comm: systemd-shutdow Not tainted
  6.1.0-next-20221220-00014-gb85aaf97fb01-dirty #28
  Hardware name: Qualcomm Technologies, Inc. Robotics RB5 (DT)
  Call trace:
   dump_backtrace.part.0+0xdc/0xf0
   show_stack+0x18/0x30
   dump_stack_lvl+0x68/0x84
   print_report+0x188/0x488
   kasan_report+0xa4/0xf0
   __asan_load8+0x80/0xac
   tty_driver_flush_buffer+0x1c/0x50
   ttyport_write_flush+0x34/0x44
   serdev_device_write_flush+0x48/0x60
   qca_serdev_shutdown+0x124/0x274
   device_shutdown+0x1e8/0x350
   kernel_restart+0x48/0xb0
   __do_sys_reboot+0x244/0x2d0
   __arm64_sys_reboot+0x54/0x70
   invoke_syscall+0x60/0x190
   el0_svc_common.constprop.0+0x7c/0x160
   do_el0_svc+0x44/0xf0
   el0_svc+0x2c/0x6c
   el0t_64_sync_handler+0xbc/0x140
   el0t_64_sync+0x190/0x194

Fixes: 7e7bbddd029b ("Bluetooth: hci_qca: Fix qca6390 enable failure after warm reboot")
Cc: <stable@vger.kernel.org>
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoBluetooth: hci_sync: Fix use HCI_OP_LE_READ_BUFFER_SIZE_V2
Luiz Augusto von Dentz [Mon, 19 Dec 2022 21:32:51 +0000 (13:32 -0800)]
Bluetooth: hci_sync: Fix use HCI_OP_LE_READ_BUFFER_SIZE_V2

commit 3a4d29b6d631bb00236a98887e1039bbfc1b6ab5 upstream.

Don't try to use HCI_OP_LE_READ_BUFFER_SIZE_V2 if controller don't
support ISO channels, but in order to check if ISO channels are
supported HCI_OP_LE_READ_LOCAL_FEATURES needs to be done earlier so the
features bits can be checked on hci_le_read_buffer_size_sync.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=216817
Fixes: c1631dbc00c1 ("Bluetooth: hci_sync: Fix hci_read_buffer_size_sync")
Cc: stable@vger.kernel.org # 6.1
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agofbdev: omapfb: avoid stack overflow warning
Arnd Bergmann [Thu, 15 Dec 2022 17:02:28 +0000 (18:02 +0100)]
fbdev: omapfb: avoid stack overflow warning

[ Upstream commit 634cf6ead93988b0da9ac054521ab63a3ba189db ]

The dsi_irq_stats structure is a little too big to fit on the
stack of a 32-bit task, depending on the specific gcc options:

fbdev/omap2/omapfb/dss/dsi.c: In function 'dsi_dump_dsidev_irqs':
fbdev/omap2/omapfb/dss/dsi.c:1621:1: error: the frame size of 1064 bytes is larger than 1024 bytes [-Werror=frame-larger-than=]

Since this is only a debugfs file, performance is not critical,
so just dynamically allocate it, and print an error message
in there in place of a failure code when the allocation fails.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Helge Deller <deller@gmx.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agoof: fdt: Honor CONFIG_CMDLINE* even without /chosen node, take 2
Rob Herring [Tue, 3 Jan 2023 18:00:32 +0000 (12:00 -0600)]
of: fdt: Honor CONFIG_CMDLINE* even without /chosen node, take 2

[ Upstream commit 064e32dc5b03114d0767893fecdaf7b5dfd8c286 ]

I do not read a strict requirement on /chosen node in either ePAPR or in
Documentation/devicetree. Help text for CONFIG_CMDLINE and
CONFIG_CMDLINE_EXTEND doesn't make their behavior explicitly dependent on
the presence of /chosen or the presense of /chosen/bootargs.

However the early check for /chosen and bailing out in
early_init_dt_scan_chosen() skips CONFIG_CMDLINE handling which is not
really related to /chosen node or the particular method of passing cmdline
from bootloader.

This leads to counterintuitive combinations (assuming
CONFIG_CMDLINE_EXTEND=y):

a) bootargs="foo", CONFIG_CMDLINE="bar" => cmdline=="foo bar"
b) /chosen missing, CONFIG_CMDLINE="bar" => cmdline==""
c) bootargs="", CONFIG_CMDLINE="bar" => cmdline==" bar"

Rework early_init_dt_scan_chosen() so that the cmdline config options are
always handled.

[commit msg written by Alexander Sverdlin]

Cc: Alexander Sverdlin <alexander.sverdlin@gmail.com>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Tested-by: Geoff Levand <geoff@infradead.org>
Reviewed-by: Alexander Sverdlin <alexander.sverdlin@gmail.com>
Link: https://lore.kernel.org/r/20230103-dt-cmdline-fix-v1-2-7038e88b18b6@kernel.org
Signed-off-by: Rob Herring <robh@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agoperf/x86/rapl: Add support for Intel Emerald Rapids
Zhang Rui [Wed, 4 Jan 2023 14:58:31 +0000 (22:58 +0800)]
perf/x86/rapl: Add support for Intel Emerald Rapids

[ Upstream commit 57512b57dcfaf63c52d8ad2fb35321328cde31b0 ]

Emerald Rapids RAPL support is the same as previous Sapphire Rapids.
Add Emerald Rapids model for RAPL.

Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Link: https://lore.kernel.org/r/20230104145831.25498-2-rui.zhang@intel.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agoperf/x86/rapl: Add support for Intel Meteor Lake
Zhang Rui [Wed, 4 Jan 2023 14:58:30 +0000 (22:58 +0800)]
perf/x86/rapl: Add support for Intel Meteor Lake

[ Upstream commit f52853a668bfeddd79f319d536a506f68cc2b478 ]

Meteor Lake RAPL support is the same as previous Sky Lake.
Add Meteor Lake model for RAPL.

Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Link: https://lore.kernel.org/r/20230104145831.25498-1-rui.zhang@intel.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agomemblock tests: Fix compilation error.
Aaron Thompson [Wed, 4 Jan 2023 10:07:37 +0000 (10:07 +0000)]
memblock tests: Fix compilation error.

[ Upstream commit 340726747336716350eb5a928b860a29db955f05 ]

Commit cf4694be2b2cf ("tools: Add atomic_test_and_set_bit()") changed
tools/arch/x86/include/asm/atomic.h to include <asm/asm.h>, which causes
'make -C tools/testing/memblock' to fail with:

In file included from ../../include/asm/atomic.h:6,
                 from ../../include/linux/atomic.h:5,
                 from ./linux/mmzone.h:5,
                 from ../../include/linux/mm.h:5,
                 from ../../include/linux/pfn.h:5,
                 from ./linux/memory_hotplug.h:6,
                 from ./linux/init.h:7,
                 from ./linux/memblock.h:11,
                 from tests/common.h:8,
                 from tests/basic_api.h:5,
                 from main.c:2:
../../include/asm/../../arch/x86/include/asm/atomic.h:11:10: fatal error: asm/asm.h: No such file or directory
   11 | #include <asm/asm.h>
      |          ^~~~~~~~~~~

Create a symlink to asm/asm.h in the same manner as the existing one to
asm/cmpxchg.h.

Signed-off-by: Aaron Thompson <dev@aaront.org>
Link: https://lore.kernel.org/r/010101857c402765-96e2dbc6-b82b-47e2-a437-4834dbe0b96b-000000@us-west-2.amazonses.com
Signed-off-by: Mike Rapoport (IBM) <rppt@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agocifs: fix race in assemble_neg_contexts()
Paulo Alcantara [Thu, 29 Dec 2022 15:33:56 +0000 (12:33 -0300)]
cifs: fix race in assemble_neg_contexts()

[ Upstream commit 775e44d6d86dca400d614cbda5dab4def4951fe7 ]

Serialise access of TCP_Server_Info::hostname in
assemble_neg_contexts() by holding the server's mutex otherwise it
might end up accessing an already-freed hostname pointer from
cifs_reconnect() or cifs_resolve_server().

Signed-off-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Reviewed-by: Enzo Matsumiya <ematsumiya@suse.de>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agoperf/x86/rapl: Treat Tigerlake like Icelake
Chris Wilson [Wed, 28 Dec 2022 11:34:54 +0000 (06:34 -0500)]
perf/x86/rapl: Treat Tigerlake like Icelake

[ Upstream commit c07311b5509f6035f1dd828db3e90ff4859cf3b9 ]

Since Tigerlake seems to have inherited its cstates and other RAPL power
caps from Icelake, assume it also follows Icelake for its RAPL events.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Zhang Rui <rui.zhang@intel.com>
Link: https://lore.kernel.org/r/20221228113454.1199118-1-rodrigo.vivi@intel.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agof2fs: let's avoid panic if extent_tree is not created
Jaegeuk Kim [Thu, 22 Dec 2022 00:14:10 +0000 (16:14 -0800)]
f2fs: let's avoid panic if extent_tree is not created

[ Upstream commit df9d44b645b83fffccfb4e28c1f93376585fdec8 ]

This patch avoids the below panic.

pc : __lookup_extent_tree+0xd8/0x760
lr : f2fs_do_write_data_page+0x104/0x87c
sp : ffffffc010cbb3c0
x29: ffffffc010cbb3e0 x28: 0000000000000000
x27: ffffff8803e7f020 x26: ffffff8803e7ed40
x25: ffffff8803e7f020 x24: ffffffc010cbb460
x23: ffffffc010cbb480 x22: 0000000000000000
x21: 0000000000000000 x20: ffffffff22e90900
x19: 0000000000000000 x18: ffffffc010c5d080
x17: 0000000000000000 x16: 0000000000000020
x15: ffffffdb1acdbb88 x14: ffffff888759e2b0
x13: 0000000000000000 x12: ffffff802da49000
x11: 000000000a001200 x10: ffffff8803e7ed40
x9 : ffffff8023195800 x8 : ffffff802da49078
x7 : 0000000000000001 x6 : 0000000000000000
x5 : 0000000000000006 x4 : ffffffc010cbba28
x3 : 0000000000000000 x2 : ffffffc010cbb480
x1 : 0000000000000000 x0 : ffffff8803e7ed40
Call trace:
 __lookup_extent_tree+0xd8/0x760
 f2fs_do_write_data_page+0x104/0x87c
 f2fs_write_single_data_page+0x420/0xb60
 f2fs_write_cache_pages+0x418/0xb1c
 __f2fs_write_data_pages+0x428/0x58c
 f2fs_write_data_pages+0x30/0x40
 do_writepages+0x88/0x190
 __writeback_single_inode+0x48/0x448
 writeback_sb_inodes+0x468/0x9e8
 __writeback_inodes_wb+0xb8/0x2a4
 wb_writeback+0x33c/0x740
 wb_do_writeback+0x2b4/0x400
 wb_workfn+0xe4/0x34c
 process_one_work+0x24c/0x5bc
 worker_thread+0x3e8/0xa50
 kthread+0x150/0x1b4

Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agox86/asm: Fix an assembler warning with current binutils
Mikulas Patocka [Tue, 3 Jan 2023 15:24:11 +0000 (10:24 -0500)]
x86/asm: Fix an assembler warning with current binutils

[ Upstream commit 55d235361fccef573990dfa5724ab453866e7816 ]

Fix a warning: "found `movsd'; assuming `movsl' was meant"

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agobtrfs: always report error in run_one_delayed_ref()
Qu Wenruo [Mon, 26 Dec 2022 01:00:40 +0000 (09:00 +0800)]
btrfs: always report error in run_one_delayed_ref()

[ Upstream commit 39f501d68ec1ed5cd5c66ac6ec2a7131c517bb92 ]

Currently we have a btrfs_debug() for run_one_delayed_ref() failure, but
if end users hit such problem, there will be no chance that
btrfs_debug() is enabled.  This can lead to very little useful info for
debugging.

This patch will:

- Add extra info for error reporting
  Including:
  * logical bytenr
  * num_bytes
  * type
  * action
  * ref_mod

- Replace the btrfs_debug() with btrfs_err()

- Move the error reporting into run_one_delayed_ref()
  This is to avoid use-after-free, the @node can be freed in the caller.

This error should only be triggered at most once.

As if run_one_delayed_ref() failed, we trigger the error message, then
causing the call chain to error out:

btrfs_run_delayed_refs()
`- btrfs_run_delayed_refs()
   `- btrfs_run_delayed_refs_for_head()
      `- run_one_delayed_ref()

And we will abort the current transaction in btrfs_run_delayed_refs().
If we have to run delayed refs for the abort transaction,
run_one_delayed_ref() will just cleanup the refs and do nothing, thus no
new error messages would be output.

Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agoselftests: net: fix cmsg_so_mark.sh test hang
Po-Hsu Lin [Thu, 29 Dec 2022 05:41:06 +0000 (13:41 +0800)]
selftests: net: fix cmsg_so_mark.sh test hang

[ Upstream commit 1573c6882018f69991aead951d09423ce978adac ]

This cmsg_so_mark.sh test will hang on non-amd64 systems because of the
infinity loop for argument parsing in cmsg_sender.

Variable "o" in cs_parse_args() for taking getopt() should be an int,
otherwise it will be 255 when getopt() returns -1 on non-amd64 system
and thus causing infinity loop.

Link: https://lore.kernel.org/lkml/CA+G9fYsM2k7mrF7W4V_TrZ-qDauWM394=8yEJ=-t1oUg8_40YA@mail.gmail.com/t/
Signed-off-by: Po-Hsu Lin <po-hsu.lin@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agoRDMA/srp: Move large values to a new enum for gcc13
Jiri Slaby (SUSE) [Mon, 12 Dec 2022 12:04:11 +0000 (13:04 +0100)]
RDMA/srp: Move large values to a new enum for gcc13

[ Upstream commit 56c5dab20a6391604df9521f812c01d1e3fe1bd0 ]

Since gcc13, each member of an enum has the same type as the enum [1]. And
that is inherited from its members. Provided these two:
  SRP_TAG_NO_REQ        = ~0U,
  SRP_TAG_TSK_MGMT = 1U << 31
all other members are unsigned ints.

Esp. with SRP_MAX_SGE and SRP_TSK_MGMT_SQ_SIZE and their use in min(),
this results in the following warnings:
  include/linux/minmax.h:20:35: error: comparison of distinct pointer types lacks a cast
  drivers/infiniband/ulp/srp/ib_srp.c:563:42: note: in expansion of macro 'min'

  include/linux/minmax.h:20:35: error: comparison of distinct pointer types lacks a cast
  drivers/infiniband/ulp/srp/ib_srp.c:2369:27: note: in expansion of macro 'min'

So move the large values away to a separate enum, so that they don't
affect other members.

[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=36113

Link: https://lore.kernel.org/r/20221212120411.13750-1-jirislaby@kernel.org
Signed-off-by: Jiri Slaby (SUSE) <jirislaby@kernel.org>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agobpf: keep a reference to the mm, in case the task is dead.
Kui-Feng Lee [Fri, 16 Dec 2022 22:18:54 +0000 (14:18 -0800)]
bpf: keep a reference to the mm, in case the task is dead.

[ Upstream commit 7ff94f276f8ea05df82eb115225e9b26f47a3347 ]

Fix the system crash that happens when a task iterator travel through
vma of tasks.

In task iterators, we used to access mm by following the pointer on
the task_struct; however, the death of a task will clear the pointer,
even though we still hold the task_struct.  That can cause an
unexpected crash for a null pointer when an iterator is visiting a
task that dies during the visit.  Keeping a reference of mm on the
iterator ensures we always have a valid pointer to mm.

Co-developed-by: Song Liu <song@kernel.org>
Signed-off-by: Song Liu <song@kernel.org>
Signed-off-by: Kui-Feng Lee <kuifeng@meta.com>
Reported-by: Nathan Slingerland <slinger@meta.com>
Acked-by: Yonghong Song <yhs@fb.com>
Link: https://lore.kernel.org/r/20221216221855.4122288-2-kuifeng@meta.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agor8169: fix dmar pte write access is not set error
Chunhao Lin [Mon, 26 Dec 2022 12:31:53 +0000 (20:31 +0800)]
r8169: fix dmar pte write access is not set error

[ Upstream commit bb41c13c05c23d9bc46b4e37d8914078c6a40e3a ]

When close device, if wol is enabled, rx will be enabled. When open
device it will cause rx packet to be dma to the wrong memory address
after pci_set_master() and system log will show blow messages.

DMAR: DRHD: handling fault status reg 3
DMAR: [DMA Write] Request device [02:00.0] PASID ffffffff fault addr
ffdd4000 [fault reason 05] PTE Write access is not set

In this patch, driver disable tx/rx when close device. If wol is
enabled, only enable rx filter and disable rxdv_gate(if support) to
let hardware only receive packet to fifo but not to dma it.

Signed-off-by: Chunhao Lin <hau@realtek.com>
Reviewed-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agor8169: move rtl_wol_enable_rx() and rtl_prepare_power_down()
Chunhao Lin [Mon, 26 Dec 2022 12:31:52 +0000 (20:31 +0800)]
r8169: move rtl_wol_enable_rx() and rtl_prepare_power_down()

[ Upstream commit ad425666a1f05d9b215a84cf010c3789b2ea8206 ]

There is no functional change. Moving these two functions for following
patch "r8169: fix dmar pte write access is not set error".

Signed-off-by: Chunhao Lin <hau@realtek.com>
Reviewed-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agonet/ethtool/ioctl: return -EOPNOTSUPP if we have no phy stats
Daniil Tatianin [Mon, 26 Dec 2022 11:48:23 +0000 (14:48 +0300)]
net/ethtool/ioctl: return -EOPNOTSUPP if we have no phy stats

[ Upstream commit 9deb1e9fb88b1120a908676fa33bdf9e2eeaefce ]

It's not very useful to copy back an empty ethtool_stats struct and
return 0 if we didn't actually have any stats. This also allows for
further simplification of this function in the future commits.

Signed-off-by: Daniil Tatianin <d-tatianin@yandex-team.ru>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agovdpa_sim_net: should not drop the multicast/broadcast packet
Cindy Lu [Wed, 14 Dec 2022 05:43:06 +0000 (13:43 +0800)]
vdpa_sim_net: should not drop the multicast/broadcast packet

[ Upstream commit 72455a1142527e607e1d69439f3ffa2ef6d09e26 ]

In the receive_filter(), should not drop the packet with the
broadcast/multicast address. Add the check for this

Signed-off-by: Cindy Lu <lulu@redhat.com>
Message-Id: <20221214054306.24145-1-lulu@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agovduse: Validate vq_num in vduse_validate_config()
Harshit Mogalapalli [Mon, 28 Nov 2022 15:57:15 +0000 (07:57 -0800)]
vduse: Validate vq_num in vduse_validate_config()

[ Upstream commit 937c783aa3d8d77963ec91918d3298edb45b9161 ]

Add a limit to 'config->vq_num' which is user controlled data which
comes from an vduse_ioctl to prevent large memory allocations.

Micheal says  - This limit is somewhat arbitrary.
However, currently virtio pci and ccw are limited to a 16 bit vq number.
While MMIO isn't it is also isn't used with lots of VQs due to
current lack of support for per-vq interrupts.
Thus, the 0xffff limit on number of VQs corresponding
to a 16-bit VQ number seems sufficient for now.

This is found using static analysis with smatch.

Suggested-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
Message-Id: <20221128155717.2579992-1-harshit.m.mogalapalli@oracle.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agovirtio_pci: modify ENOENT to EINVAL
Angus Chen [Tue, 1 Nov 2022 11:16:54 +0000 (19:16 +0800)]
virtio_pci: modify ENOENT to EINVAL

[ Upstream commit b66ead2d0ecac00c3a06a6218af5411cb5fcb5d5 ]

Virtio_crypto use max_data_queues+1 to setup vqs,
we use vp_modern_get_num_queues to protect the vq range in setup_vq.
We could enter index >= vp_modern_get_num_queues(mdev) in setup_vq
if common->num_queues is not set well,and it return -ENOENT.
It is better to use -EINVAL instead.

Signed-off-by: Angus Chen <angus.chen@jaguarmicro.com>
Message-Id: <20221101111655.1947-1-angus.chen@jaguarmicro.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agovdpa/mlx5: Avoid overwriting CVQ iotlb
Eli Cohen [Mon, 14 Nov 2022 13:17:56 +0000 (15:17 +0200)]
vdpa/mlx5: Avoid overwriting CVQ iotlb

[ Upstream commit 38fc462f57ef4e5dc722bab6824854b105de8aa2 ]

When qemu uses different address spaces for data and control virtqueues,
the current code would overwrite the control virtqueue iotlb through the
dup_iotlb call. Fix this by referring to the address space identifier
and the group to asid mapping to determine which mapping needs to be
updated. We also move the address space logic from mlx5 net to core
directory.

Reported-by: Eugenio Pérez <eperezma@redhat.com>
Signed-off-by: Eli Cohen <elic@nvidia.com>
Message-Id: <20221114131759.57883-6-elic@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agovdpa/mlx5: Avoid using reslock in event_handler
Eli Cohen [Mon, 14 Nov 2022 13:17:55 +0000 (15:17 +0200)]
vdpa/mlx5: Avoid using reslock in event_handler

[ Upstream commit 0dbc1b4ae07d003b2e88ba9d4142846320f8e349 ]

event_handler runs under atomic context and may not acquire reslock. We
can still guarantee that the handler won't be called after suspend by
clearing nb_registered, unregistering the handler and flushing the
workqueue.

Signed-off-by: Eli Cohen <elic@nvidia.com>
Message-Id: <20221114131759.57883-5-elic@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agovdpa/mlx5: Return error on vlan ctrl commands if not supported
Eli Cohen [Mon, 14 Nov 2022 13:17:53 +0000 (15:17 +0200)]
vdpa/mlx5: Return error on vlan ctrl commands if not supported

[ Upstream commit 5aec804936bbff182081f1cdc271fcb76af1a4ff ]

Check if VIRTIO_NET_F_CTRL_VLAN is negotiated and return error if
control VQ command is received.

Signed-off-by: Eli Cohen <elic@nvidia.com>
Message-Id: <20221114131759.57883-3-elic@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agotools/virtio: initialize spinlocks in vring_test.c
Ricardo Cañuelo [Wed, 12 Oct 2022 06:29:49 +0000 (08:29 +0200)]
tools/virtio: initialize spinlocks in vring_test.c

[ Upstream commit c262f75cb6bb5a63828e72ce3b8fe808e5029479 ]

The virtio_device vqs_list spinlocks must be initialized before use to
prevent functions that manipulate the device virtualqueues, such as
vring_new_virtqueue(), from blocking indefinitely.

Signed-off-by: Ricardo Cañuelo <ricardo.canuelo@collabora.com>
Message-Id: <20221012062949.1526176-1-ricardo.canuelo@collabora.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agonet: ethernet: marvell: octeontx2: Fix uninitialized variable warning
Anuradha Weeraman [Sun, 25 Dec 2022 17:42:22 +0000 (23:12 +0530)]
net: ethernet: marvell: octeontx2: Fix uninitialized variable warning

[ Upstream commit d3805695fe1e7383517903715cefc9bbdcffdc90 ]

Fix for uninitialized variable warning.

Addresses-Coverity: ("Uninitialized scalar variable")
Signed-off-by: Anuradha Weeraman <anuradha@debian.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agoselftests/bpf: check null propagation only neither reg is PTR_TO_BTF_ID
Hao Sun [Thu, 22 Dec 2022 02:44:14 +0000 (10:44 +0800)]
selftests/bpf: check null propagation only neither reg is PTR_TO_BTF_ID

[ Upstream commit cedebd74cf3883f0384af9ec26b4e6f8f1964dd4 ]

Verify that nullness information is not porpagated in the branches
of register to register JEQ and JNE operations if one of them is
PTR_TO_BTF_ID. Implement this in C level so we can use CO-RE.

Signed-off-by: Hao Sun <sunhao.th@gmail.com>
Suggested-by: Martin KaFai Lau <martin.lau@kernel.org>
Link: https://lore.kernel.org/r/20221222024414.29539-2-sunhao.th@gmail.com
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agopNFS/filelayout: Fix coalescing test for single DS
Olga Kornievskaia [Tue, 20 Dec 2022 17:31:29 +0000 (12:31 -0500)]
pNFS/filelayout: Fix coalescing test for single DS

[ Upstream commit a6b9d2fa0024e7e399c26facd0fb466b7396e2b9 ]

When there is a single DS no striping constraints need to be placed on
the IO. When such constraint is applied then buffered reads don't
coalesce to the DS's rsize.

Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agowifi: iwlwifi: fw: skip PPAG for JF
Johannes Berg [Tue, 13 Dec 2022 21:15:04 +0000 (23:15 +0200)]
wifi: iwlwifi: fw: skip PPAG for JF

[ Upstream commit 1c4c0b28b517d778d37900deedfe91088839f07a ]

For JF RFs we don't support PPAG, but many firmware
images lie about it. Always skip support for JF to
avoid firmware errors when sending the command.

Reported-and-tested-by: Íñigo Huguet <ihuguet@redhat.com>
Link: https://lore.kernel.org/linux-wireless/CACT4oufQsqHGp6bah2c4+jPn2wG1oZqY=UKa_TmPx=F6Lxng8Q@mail.gmail.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Gregory Greenman <gregory.greenman@intel.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20221213225723.2a43415d8990.I9ac210740a45b41f1b2e15274e1daf4284f2808a@changeid
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agobtrfs: fix trace event name typo for FLUSH_DELAYED_REFS
Naohiro Aota [Wed, 14 Dec 2022 02:06:07 +0000 (11:06 +0900)]
btrfs: fix trace event name typo for FLUSH_DELAYED_REFS

[ Upstream commit 0a3212de8ab3e2ce5808c6265855e528d4a6767b ]

Fix a typo of printing FLUSH_DELAYED_REFS event in flush_space() as
FLUSH_ELAYED_REFS.

Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agodma-buf: fix dma_buf_export init order v2
Christian König [Tue, 6 Dec 2022 13:07:49 +0000 (14:07 +0100)]
dma-buf: fix dma_buf_export init order v2

[ Upstream commit f728a5ea27c92133893590e731ce10f6561ced87 ]

The init order and resulting error handling in dma_buf_export
was pretty messy.

Subordinate objects like the file and the sysfs kernel objects
were initializing and wiring itself up with the object in the
wrong order resulting not only in complicating and partially
incorrect error handling, but also in publishing only halve
initialized DMA-buf objects.

Clean this up thoughtfully by allocating the file independent
of the DMA-buf object. Then allocate and initialize the DMA-buf
object itself, before publishing it through sysfs. If everything
works as expected the file is then connected with the DMA-buf
object and publish it through debugfs.

Also adds the missing dma_resv_fini() into the error handling.

v2: add some missing changes to dma_bug_getfile() and a missing NULL
    check in dma_buf_file_release()

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Reviewed-by: T.J. Mercier <tjmercier@google.com>
Acked-by: Sumit Semwal <sumit.semwal@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20221209071535.933698-1-christian.koenig@amd.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agoLinux 6.1.7
Greg Kroah-Hartman [Wed, 18 Jan 2023 10:58:34 +0000 (11:58 +0100)]
Linux 6.1.7

Link: https://lore.kernel.org/r/20230116154803.321528435@linuxfoundation.org
Tested-by: Salvatore Bonaccorso <carnil@debian.org>
Tested-by: Ronald Warsow <rwarsow@gmx.de>
Tested-by: Conor Dooley <conor.dooley@microchip.com>
Tested-by: Shuah Khan <skhan@linuxfoundation.org>
Tested-by: Justin M. Forbes <jforbes@fedoraproject.org>
Tested-by: Takeshi Ogasawara <takeshi.ogasawara@futuring-girl.com>
Tested-by: Rudi Heitbaum <rudi@heitbaum.com>
Tested-by: Bagas Sanjaya <bagasdotme@gmail.com>
Tested-by: Ron Economos <re@w6rz.net>
Tested-by: Linux Kernel Functional Testing <lkft@linaro.org>
Tested-by: Jon Hunter <jonathanh@nvidia.com>
Tested-by: Sudip Mukherjee <sudip.mukherjee@codethink.co.uk>
Tested-by: Allen Pais <apais@linux.microsoft.com>
Link: https://lore.kernel.org/r/20230117124546.116438951@linuxfoundation.org
Tested-by: Allen Pais <apais@linux.microsoft.com>
Tested-by: Ronald Warsow <rwarsow@gmx.de>
Tested-by: Ron Economos <re@w6rz.net>
Tested-by: Takeshi Ogasawara <takeshi.ogasawara@futuring-girl.com>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Tested-by: Kelsey Steele <kelseysteele@linux.microsoft.com>
Tested-by: Bagas Sanjaya <bagasdotme@gmail.com>
Tested-by: Fenil Jain <fkjainco@gmail.com>
Tested-by: Linux Kernel Functional Testing <lkft@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agopinctrl: amd: Add dynamic debugging for active GPIOs
Mario Limonciello [Thu, 13 Oct 2022 13:47:29 +0000 (08:47 -0500)]
pinctrl: amd: Add dynamic debugging for active GPIOs

commit 1d66e379731f79ae5039a869c0fde22a4f6a6a91 upstream.

Some laptops have been reported to wake up from s2idle when plugging
in the AC adapter or by closing the lid.  This is a surprising
behavior that is further clarified by commit cb3e7d624c3ff ("PM:
wakeup: Add extra debugging statement for multiple active IRQs").

With that commit in place the following interaction can be seen
when the lid is closed:

[   28.946038] PM: suspend-to-idle
[   28.946083] ACPI: EC: ACPI EC GPE status set
[   28.946101] ACPI: PM: Rearming ACPI SCI for wakeup
[   28.950152] Timekeeping suspended for 3.320 seconds
[   28.950152] PM: Triggering wakeup from IRQ 9
[   28.950152] ACPI: EC: ACPI EC GPE status set
[   28.950152] ACPI: EC: ACPI EC GPE dispatched
[   28.995057] ACPI: EC: ACPI EC work flushed
[   28.995075] ACPI: PM: Rearming ACPI SCI for wakeup
[   28.995131] PM: Triggering wakeup from IRQ 9
[   28.995271] ACPI: EC: ACPI EC GPE status set
[   28.995291] ACPI: EC: ACPI EC GPE dispatched
[   29.098556] ACPI: EC: ACPI EC work flushed
[   29.207020] ACPI: EC: ACPI EC work flushed
[   29.207037] ACPI: PM: Rearming ACPI SCI for wakeup
[   29.211095] Timekeeping suspended for 0.739 seconds
[   29.211095] PM: Triggering wakeup from IRQ 9
[   29.211079] PM: Triggering wakeup from IRQ 7
[   29.211095] ACPI: PM: ACPI non-EC GPE wakeup
[   29.211095] PM: resume from suspend-to-idle

* IRQ9 on this laptop is used for the ACPI SCI.
* IRQ7 on this laptop is used for the GPIO controller.

What has occurred is when the lid was closed the EC woke up the
SoC from it's deepest sleep state and the kernel's s2idle loop
processed all EC events.  When it was finished processing EC events,
it checked for any other reasons to wake (break the s2idle loop).

The IRQ for the GPIO controller was active so the loop broke, and
then this IRQ was processed.  This is not a kernel bug but it is
certainly a surprising behavior, and to better debug it we should
have a dynamic debugging message that we can enact to catch it.

Acked-by: Basavaraj Natikar <Basavaraj.Natikar@amd.com>
Acked-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
Acked-by: Mark Pearson <markpearson@lenovo.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Link: https://lore.kernel.org/r/20221013134729.5592-2-mario.limonciello@amd.com
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoRevert "usb: ulpi: defer ulpi_register on ulpi_read_id timeout"
Ferry Toth [Thu, 22 Dec 2022 20:53:02 +0000 (21:53 +0100)]
Revert "usb: ulpi: defer ulpi_register on ulpi_read_id timeout"

commit b659b613cea2ae39746ca8bd2b69d1985dd9d770 upstream.

This reverts commit 8a7b31d545d3a15f0e6f5984ae16f0ca4fd76aac.

This patch results in some qemu test failures, specifically xilinx-zynq-a9
machine and zynq-zc702 as well as zynq-zed devicetree files, when trying
to boot from USB drive.

Link: https://lore.kernel.org/lkml/20221220194334.GA942039@roeck-us.net/
Fixes: 8a7b31d545d3 ("usb: ulpi: defer ulpi_register on ulpi_read_id timeout")
Cc: stable@vger.kernel.org
Reported-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Ferry Toth <ftoth@exalondelft.nl>
Link: https://lore.kernel.org/r/20221222205302.45761-1-ftoth@exalondelft.nl
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoblock: handle bio_split_to_limits() NULL return
Jens Axboe [Wed, 4 Jan 2023 15:51:19 +0000 (08:51 -0700)]
block: handle bio_split_to_limits() NULL return

commit 613b14884b8595e20b9fac4126bf627313827fbe upstream.

This can't happen right now, but in preparation for allowing
bio_split_to_limits() returning NULL if it ended the bio, check for it
in all the callers.

Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoio_uring/io-wq: only free worker if it was allocated for creation
Jens Axboe [Sun, 8 Jan 2023 17:39:17 +0000 (10:39 -0700)]
io_uring/io-wq: only free worker if it was allocated for creation

commit e6db6f9398dadcbc06318a133d4c44a2d3844e61 upstream.

We have two types of task_work based creation, one is using an existing
worker to setup a new one (eg when going to sleep and we have no free
workers), and the other is allocating a new worker. Only the latter
should be freed when we cancel task_work creation for a new worker.

Fixes: af82425c6a2d ("io_uring/io-wq: free worker if task_work creation is canceled")
Reported-by: syzbot+d56ec896af3637bdb7e4@syzkaller.appspotmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoio_uring/io-wq: free worker if task_work creation is canceled
Jens Axboe [Mon, 2 Jan 2023 23:49:46 +0000 (16:49 -0700)]
io_uring/io-wq: free worker if task_work creation is canceled

commit af82425c6a2d2f347c79b63ce74fca6dc6be157f upstream.

If we cancel the task_work, the worker will never come into existance.
As this is the last reference to it, ensure that we get it freed
appropriately.

Cc: stable@vger.kernel.org
Reported-by: 진호 <wnwlsgh98@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agodrm/i915: Fix CFI violations in gt_sysfs
Nathan Chancellor [Tue, 25 Oct 2022 19:50:15 +0000 (21:50 +0200)]
drm/i915: Fix CFI violations in gt_sysfs

commit a8a4f0467d706fc22d286dfa973946e5944b793c upstream.

When booting with CONFIG_CFI_CLANG, there are numerous violations when
accessing the files under
/sys/devices/pci0000:00/0000:00:02.0/drm/card0/gt/gt0:

  $ cd /sys/devices/pci0000:00/0000:00:02.0/drm/card0/gt/gt0

  $ grep . *
  id:0
  punit_req_freq_mhz:350
  rc6_enable:1
  rc6_residency_ms:214934
  rps_act_freq_mhz:1300
  rps_boost_freq_mhz:1300
  rps_cur_freq_mhz:350
  rps_max_freq_mhz:1300
  rps_min_freq_mhz:350
  rps_RP0_freq_mhz:1300
  rps_RP1_freq_mhz:350
  rps_RPn_freq_mhz:350
  throttle_reason_pl1:0
  throttle_reason_pl2:0
  throttle_reason_pl4:0
  throttle_reason_prochot:0
  throttle_reason_ratl:0
  throttle_reason_status:0
  throttle_reason_thermal:0
  throttle_reason_vr_tdc:0
  throttle_reason_vr_thermalert:0

  $ sudo dmesg &| grep "CFI failure at"
  [  214.595903] CFI failure at kobj_attr_show+0x19/0x30 (target: id_show+0x0/0x70 [i915]; expected type: 0xc527b809)
  [  214.596064] CFI failure at kobj_attr_show+0x19/0x30 (target: punit_req_freq_mhz_show+0x0/0x40 [i915]; expected type: 0xc527b809)
  [  214.596407] CFI failure at kobj_attr_show+0x19/0x30 (target: rc6_enable_show+0x0/0x40 [i915]; expected type: 0xc527b809)
  [  214.596528] CFI failure at kobj_attr_show+0x19/0x30 (target: rc6_residency_ms_show+0x0/0x270 [i915]; expected type: 0xc527b809)
  [  214.596682] CFI failure at kobj_attr_show+0x19/0x30 (target: act_freq_mhz_show+0x0/0xe0 [i915]; expected type: 0xc527b809)
  [  214.596792] CFI failure at kobj_attr_show+0x19/0x30 (target: boost_freq_mhz_show+0x0/0xe0 [i915]; expected type: 0xc527b809)
  [  214.596893] CFI failure at kobj_attr_show+0x19/0x30 (target: cur_freq_mhz_show+0x0/0xe0 [i915]; expected type: 0xc527b809)
  [  214.596996] CFI failure at kobj_attr_show+0x19/0x30 (target: max_freq_mhz_show+0x0/0xe0 [i915]; expected type: 0xc527b809)
  [  214.597099] CFI failure at kobj_attr_show+0x19/0x30 (target: min_freq_mhz_show+0x0/0xe0 [i915]; expected type: 0xc527b809)
  [  214.597198] CFI failure at kobj_attr_show+0x19/0x30 (target: RP0_freq_mhz_show+0x0/0xe0 [i915]; expected type: 0xc527b809)
  [  214.597301] CFI failure at kobj_attr_show+0x19/0x30 (target: RP1_freq_mhz_show+0x0/0xe0 [i915]; expected type: 0xc527b809)
  [  214.597405] CFI failure at kobj_attr_show+0x19/0x30 (target: RPn_freq_mhz_show+0x0/0xe0 [i915]; expected type: 0xc527b809)
  [  214.597538] CFI failure at kobj_attr_show+0x19/0x30 (target: throttle_reason_bool_show+0x0/0x50 [i915]; expected type: 0xc527b809)
  [  214.597701] CFI failure at kobj_attr_show+0x19/0x30 (target: throttle_reason_bool_show+0x0/0x50 [i915]; expected type: 0xc527b809)
  [  214.597836] CFI failure at kobj_attr_show+0x19/0x30 (target: throttle_reason_bool_show+0x0/0x50 [i915]; expected type: 0xc527b809)
  [  214.597952] CFI failure at kobj_attr_show+0x19/0x30 (target: throttle_reason_bool_show+0x0/0x50 [i915]; expected type: 0xc527b809)
  [  214.598071] CFI failure at kobj_attr_show+0x19/0x30 (target: throttle_reason_bool_show+0x0/0x50 [i915]; expected type: 0xc527b809)
  [  214.598177] CFI failure at kobj_attr_show+0x19/0x30 (target: throttle_reason_bool_show+0x0/0x50 [i915]; expected type: 0xc527b809)
  [  214.598307] CFI failure at kobj_attr_show+0x19/0x30 (target: throttle_reason_bool_show+0x0/0x50 [i915]; expected type: 0xc527b809)
  [  214.598439] CFI failure at kobj_attr_show+0x19/0x30 (target: throttle_reason_bool_show+0x0/0x50 [i915]; expected type: 0xc527b809)
  [  214.598542] CFI failure at kobj_attr_show+0x19/0x30 (target: throttle_reason_bool_show+0x0/0x50 [i915]; expected type: 0xc527b809)

With kCFI, indirect calls are validated against their expected type
versus actual type and failures occur when the two types do not match.
The ultimate issue is that these sysfs functions are expecting to be
called via dev_attr_show() but they may also be called via
kobj_attr_show(), as certain files are created under two different
kobjects that have two different sysfs_ops in intel_gt_sysfs_register(),
hence the warnings above. When accessing the gt_ files under
/sys/devices/pci0000:00/0000:00:02.0/drm/card0, which are using the same
sysfs functions, there are no violations, meaning the functions are
being called with the proper type.

To make everything work properly, adjust certain functions to match the
type of the ->show() and ->store() members in 'struct kobj_attribute'.
Add a macro to generate functions for that can be called via both
dev_attr_{show,store}() or kobj_attr_{show,store}() so that they can be
called through both kobject locations without violating kCFI and adjust
the attribute groups to account for this.

Link: https://github.com/ClangBuiltLinux/linux/issues/1716
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Reviewed-by: Andrzej Hajda <andrzej.hajda@intel.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20221013205909.1282545-1-nathan@kernel.org
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agoio_uring/poll: attempt request issue after racy poll wakeup
Jens Axboe [Sat, 14 Jan 2023 15:46:14 +0000 (08:46 -0700)]
io_uring/poll: attempt request issue after racy poll wakeup

commit 6e5aedb9324aab1c14a23fae3d8eeb64a679c20e upstream.

If we have multiple requests waiting on the same target poll waitqueue,
then it's quite possible to get a request triggered and get disappointed
in not being able to make any progress with it. If we race in doing so,
we'll potentially leave the poll request on the internal tables, but
removed from the waitqueue. That means that any subsequent trigger of
the poll waitqueue will not kick that request into action, causing an
application to potentially wait for completion of a request that will
never happen.

Fix this by adding a new poll return state, IOU_POLL_REISSUE. Rather
than have complicated logic for how to re-arm a given type of request,
just punt it for a reissue.

While in there, move the 'ret' variable to the only section where it
gets used. This avoids confusion the scope of it.

Cc: stable@vger.kernel.org
Fixes: eb0089d629ba ("io_uring: single shot poll removal optimisation")
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>
17 months agoio_uring: lock overflowing for IOPOLL
Pavel Begunkov [Sat, 14 Jan 2023 15:41:33 +0000 (08:41 -0700)]
io_uring: lock overflowing for IOPOLL

commit 544d163d659d45a206d8929370d5a2984e546cb7 upstream.

syzbot reports an issue with overflow filling for IOPOLL:

WARNING: CPU: 0 PID: 28 at io_uring/io_uring.c:734 io_cqring_event_overflow+0x1c0/0x230 io_uring/io_uring.c:734
CPU: 0 PID: 28 Comm: kworker/u4:1 Not tainted 6.2.0-rc3-syzkaller-16369-g358a161a6a9e #0
Workqueue: events_unbound io_ring_exit_work
Call trace:
 io_cqring_event_overflow+0x1c0/0x230 io_uring/io_uring.c:734
 io_req_cqe_overflow+0x5c/0x70 io_uring/io_uring.c:773
 io_fill_cqe_req io_uring/io_uring.h:168 [inline]
 io_do_iopoll+0x474/0x62c io_uring/rw.c:1065
 io_iopoll_try_reap_events+0x6c/0x108 io_uring/io_uring.c:1513
 io_uring_try_cancel_requests+0x13c/0x258 io_uring/io_uring.c:3056
 io_ring_exit_work+0xec/0x390 io_uring/io_uring.c:2869
 process_one_work+0x2d8/0x504 kernel/workqueue.c:2289
 worker_thread+0x340/0x610 kernel/workqueue.c:2436
 kthread+0x12c/0x158 kernel/kthread.c:376
 ret_from_fork+0x10/0x20 arch/arm64/kernel/entry.S:863

There is no real problem for normal IOPOLL as flush is also called with
uring_lock taken, but it's getting more complicated for IOPOLL|SQPOLL,
for which __io_cqring_overflow_flush() happens from the CQ waiting path.

Reported-and-tested-by: syzbot+6805087452d72929404e@syzkaller.appspotmail.com
Cc: stable@vger.kernel.org # 5.10+
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>