Mark NX as being used for all non-nested shadow MMUs, as KVM will set the
NX bit for huge SPTEs if the iTLB mutli-hit mitigation is enabled.
Checking the mitigation itself is not sufficient as it can be toggled on
at any time and KVM doesn't reset MMU contexts when that happens. KVM
could reset the contexts, but that would require purging all SPTEs in all
MMUs, for no real benefit. And, KVM already forces EFER.NX=1 when TDP is
disabled (for WP=0, SMEP=1, NX=0), so technically NX is never reserved
for shadow MMUs.
LLVM does not emit optimal byteswap assembly, which results in high
stack usage in kvmhv_enter_nested_guest() due to the inlining of
byteswap_pt_regs(). With LLVM 12.0.0:
arch/powerpc/kvm/book3s_hv_nested.c:289:6: error: stack frame size of
2512 bytes in function 'kvmhv_enter_nested_guest' [-Werror,-Wframe-larger-than=]
long kvmhv_enter_nested_guest(struct kvm_vcpu *vcpu)
^
1 error generated.
While this gets fixed in LLVM, mark byteswap_pt_regs() as
noinline_for_stack so that it does not get inlined and break the build
due to -Werror by default in arch/powerpc/. Not inlining saves
approximately 800 bytes with LLVM 12.0.0:
arch/powerpc/kvm/book3s_hv_nested.c:290:6: warning: stack frame size of
1728 bytes in function 'kvmhv_enter_nested_guest' [-Wframe-larger-than=]
long kvmhv_enter_nested_guest(struct kvm_vcpu *vcpu)
^
1 warning generated.
Mark #ACs that won't be reinjected to the guest as wanted by L0 so that
KVM handles split-lock #AC from L2 instead of forwarding the exception to
L1. Split-lock #AC isn't yet virtualized, i.e. L1 will treat it like a
regular #AC and do the wrong thing, e.g. reinject it into L2.
Fixes: e6f8b6c12f03 ("KVM: VMX: Extend VMXs #AC interceptor to handle split lock #AC in guest") Cc: Xiaoyao Li <xiaoyao.li@intel.com> Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210622172244.3561540-1-seanjc@google.com> Cc: stable@vger.kernel.org Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
With global filtering, we only allow an event to be scheduled if its
filter settings exactly match those of any existing events, therefore
it is pointless to reapply the filter in that case. Much worse, though,
is that in doing that we trample the event type of counter 0 if it's
already active, and never touch the appropriate PMEVTYPERn so the new
event is likely not counting the right thing either. Don't do that.
try_grab_compound_head() is used to grab a reference to a page from
get_user_pages_fast(), which is only protected against concurrent freeing
of page tables (via local_irq_save()), but not against concurrent TLB
flushes, freeing of data pages, or splitting of compound pages.
Because no reference is held to the page when try_grab_compound_head() is
called, the page may have been freed and reallocated by the time its
refcount has been elevated; therefore, once we're holding a stable
reference to the page, the caller re-checks whether the PTE still points
to the same page (with the same access rights).
The problem is that try_grab_compound_head() has to grab a reference on
the head page; but between the time we look up what the head page is and
the time we actually grab a reference on the head page, the compound page
may have been split up (either explicitly through split_huge_page() or by
freeing the compound page to the buddy allocator and then allocating its
individual order-0 pages). If that happens, get_user_pages_fast() may end
up returning the right page but lifting the refcount on a now-unrelated
page, leading to use-after-free of pages.
To fix it: Re-check whether the pages still belong together after lifting
the refcount on the head page. Move anything else that checks
compound_head(page) below the refcount increment.
This can't actually happen on bare-metal x86 (because there, disabling
IRQs locks out remote TLB flushes), but it can happen on virtualized x86
(e.g. under KVM) and probably also on arm64. The race window is pretty
narrow, and constantly allocating and shattering hugepages isn't exactly
fast; for now I've only managed to reproduce this in an x86 KVM guest with
an artificially widened timing window (by adding a loop that repeatedly
calls `inl(0x3f8 + 5)` in `try_get_compound_head()` to force VM exits, so
that PV TLB flushes are used instead of IPIs).
As requested on the list, also replace the existing VM_BUG_ON_PAGE() with
a warning and bailout. Since the existing code only performed the BUG_ON
check on DEBUG_VM kernels, ensure that the new code also only performs the
check under that configuration - I don't want to mix two logically
separate changes together too much. The macro VM_WARN_ON_ONCE_PAGE()
doesn't return a value on !DEBUG_VM, so wrap the whole check in an #ifdef
block. An alternative would be to change the VM_WARN_ON_ONCE_PAGE()
definition for !DEBUG_VM such that it always returns false, but since that
would differ from the behavior of the normal WARN macros, it might be too
confusing for readers.
Link: https://lkml.kernel.org/r/20210615012014.1100672-1-jannh@google.com Fixes: 7aef4172c795 ("mm: handle PTE-mapped tail pages in gerneric fast gup implementaiton") Signed-off-by: Jann Horn <jannh@google.com> Reviewed-by: John Hubbard <jhubbard@nvidia.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Jan Kara <jack@suse.cz> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
During system resume, MHI host triggers M3->M0 transition and then waits
for target device to enter M0 state. Once done, the device queues a state
change event into ctrl event ring and notifies MHI host by raising an
interrupt, where a tasklet is scheduled to process this event. In most
cases, the tasklet is served timely and wait operation succeeds.
However, there are cases where CPU is busy and cannot serve this tasklet
for some time. Once delay goes long enough, the device moves itself to M1
state and also interrupts MHI host after inserting a new state change
event to ctrl ring. Later when CPU finally has time to process the ring,
there will be two events:
1. For M3->M0 event, which is the first event to be processed queued first.
The tasklet handler serves the event, updates device state to M0 and
wakes up the task.
2. For M0->M1 event, which is processed later, the tasklet handler
triggers M1->M2 transition and updates device state to M2 directly,
then wakes up the MHI host (if it is still sleeping on this wait queue).
Note that although MHI host has been woken up while processing the first
event, it may still has no chance to run before the second event is
processed. In other words, MHI host has to keep waiting till timeout
causing the M0 state to be missed.
kernel log here:
...
Apr 15 01:45:14 test-NUC8i7HVK kernel: [ 4247.911251] mhi 0000:06:00.0: Entered with PM state: M3, MHI state: M3
Apr 15 01:45:14 test-NUC8i7HVK kernel: [ 4247.917762] mhi 0000:06:00.0: State change event to state: M0
Apr 15 01:45:14 test-NUC8i7HVK kernel: [ 4247.917767] mhi 0000:06:00.0: State change event to state: M1
Apr 15 01:45:14 test-NUC8i7HVK kernel: [ 4338.788231] mhi 0000:06:00.0: Did not enter M0 state, MHI state: M2, PM state: M2
...
Fix this issue by simply adding M2 as a valid state for resume.
When find a task from wait queue to wake up, a non-privileged task may
be found out, rather than the privileged. This maybe lead a deadlock
same as commit dfe1fe75e00e ("NFSv4: Fix deadlock between nfs4_evict_inode()
and nfs4_opendata_get_inode()"):
Privileged delegreturn task is queued to privileged list because all
the slots are assigned. If there has no enough slot to wake up the
non-privileged batch tasks(session less than 8 slot), then the privileged
delegreturn task maybe lost waked up because the found out task can't
get slot since the session is on draining.
So we should treate the privileged task as the emergency task, and
execute it as for as we can.
The 'queue->nr' will wraparound from 0 to 255 when only current
priority queue has tasks. This maybe lead a deadlock same as commit dfe1fe75e00e ("NFSv4: Fix deadlock between nfs4_evict_inode()
and nfs4_opendata_get_inode()"):
Privileged delegreturn task is queued to privileged list because all
the slots are assigned. When non-privileged task complete and release
the slot, a non-privileged maybe picked out. It maybe allocate slot
failed when the session on draining.
If the 'queue->nr' has wraparound to 255, and no enough slot to
service it, then the privileged delegreturn will lost to wake up.
Sending nulldata packets is important for sw AP link probing and detecting
4-address mode links. The checks that dropped these packets were apparently
added to work around an iwlwifi firmware bug with multi-TID aggregation.
Fixes: 41cbb0f5a295 ("mac80211: add support for HE") Cc: stable@vger.kernel.org Signed-off-by: Felix Fietkau <nbd@nbd.name> Link: https://lore.kernel.org/r/20210619101517.90806-1-nbd@nbd.name Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Rather than just indicating that transmission can start, this patch
requires the explicit flushing of the network TX queue when the driver
is informed by the device that it can transmit, next to its
configuration.
In this way, if frames have already been written by the application,
they will actually be transmitted.
Set SOCK_RCU_FREE to let RCU to call sk_destruct() on completion.
Without this patch, we will run in to j1939_can_recv() after priv was
freed by j1939_sk_release()->j1939_sk_sock_destruct()
Fixes: 25fe97cb7620 ("can: j1939: move j1939_priv_put() into sk_destruct callback") Link: https://lore.kernel.org/r/20210617130623.12705-1-o.rempel@pengutronix.de Cc: linux-stable <stable@vger.kernel.org> Reported-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Reported-by: syzbot+bdf710cfc41c186fdff3@syzkaller.appspotmail.com Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de> Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When closing the isotp socket, the potentially running hrtimers are
canceled before removing the subscription for CAN identifiers via
can_rx_unregister().
This may lead to an unintended (re)start of a hrtimer in
isotp_rcv_cf() and isotp_rcv_fc() in the case that a CAN frame is
received by isotp_rcv() while the subscription removal is processed.
However, isotp_rcv() is called under RCU protection, so after calling
can_rx_unregister, we may call synchronize_rcu in order to wait for
any RCU read-side critical sections to finish. This prevents the
reception of CAN frames after hrtimer_cancel() and therefore the
unintended (re)start of the hrtimers.
Link: https://lore.kernel.org/r/20210618173713.2296-1-socketcan@hartkopp.net Fixes: e057dd3fc20f ("can: add ISO 15765-2:2016 transport protocol") Cc: linux-stable <stable@vger.kernel.org> Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net> Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
can_can_gw_rcv() is called under RCU protection, so after calling
can_rx_unregister(), we have to call synchronize_rcu in order to wait
for any RCU read-side critical sections to finish before removing the
kmem_cache entry with the referenced gw job entry.
Link: https://lore.kernel.org/r/20210618173645.2238-1-socketcan@hartkopp.net Fixes: c1aabdf379bc ("can-gw: add netlink based CAN routing") Cc: linux-stable <stable@vger.kernel.org> Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net> Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
can_rx_register() callbacks may be called concurrently to the call to
can_rx_unregister(). The callbacks and callback data, though, are
protected by RCU and the struct sock reference count.
So the callback data is really attached to the life of sk, meaning
that it should be released on sk_destruct. However, bcm_remove_op()
calls tasklet_kill(), and RCU callbacks may be called under RCU
softirq, so that cannot be used on kernels before the introduction of
HRTIMER_MODE_SOFT.
However, bcm_rx_handler() is called under RCU protection, so after
calling can_rx_unregister(), we may call synchronize_rcu() in order to
wait for any RCU read-side critical sections to finish. That is,
bcm_rx_handler() won't be called anymore for those ops. So, we only
free them, after we do that synchronize_rcu().
Fixes: ffd980f976e7 ("[CAN]: Add broadcast manager (bcm) protocol") Link: https://lore.kernel.org/r/20210619161813.2098382-1-cascardo@canonical.com Cc: linux-stable <stable@vger.kernel.org> Reported-by: syzbot+0f7e7e5e2f4f40fa89c0@syzkaller.appspotmail.com Reported-by: Norbert Slusarek <nslusarek@gmx.net> Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Acked-by: Oliver Hartkopp <socketcan@hartkopp.net> Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 5d1b1b3f492f ("ext4: fix BUG when calling ext4_error with locked
block group") introduces ext4_grp_locked_error to handle unlocking a
group in error cases. Otherwise, there is a possibility of a sleep while
atomic. However, since 43c73221b3b1 ("ext4: replace BUG_ON with WARN_ON
in mb_find_extent()"), mb_find_extent() has contained a ext4_error()
call while a group spinlock is held. Replace this with
ext4_grp_locked_error.
Fixes: 43c73221b3b1 ("ext4: replace BUG_ON with WARN_ON in mb_find_extent()") Cc: <stable@vger.kernel.org> # 4.14+ Signed-off-by: Stephen Brennan <stephen.s.brennan@oracle.com> Reviewed-by: Lukas Czerner <lczerner@redhat.com> Reviewed-by: Junxiao Bi <junxiao.bi@oracle.com> Link: https://lore.kernel.org/r/20210623232114.34457-1-stephen.s.brennan@oracle.com Signed-off-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
After converting fs shrinkers to new scan/count API, we are no longer
pass zero nr_to_scan parameter to detect the number of objects to free,
just remove this check.
Fixes: 1ab6c4997e04 ("fs: convert fs shrinkers to new scan/count API") Cc: stable@vger.kernel.org # 3.12+ Signed-off-by: Zhang Yi <yi.zhang@huawei.com> Reviewed-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20210522103045.690103-2-yi.zhang@huawei.com Signed-off-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The cache_cnt parameter of tracepoint ext4_es_shrink_exit means the
remaining cache count after shrink, but now it is the cache count before
shrink, fix it by read sbi->s_extent_cache_cnt again.
Fixes: 1ab6c4997e04 ("fs: convert fs shrinkers to new scan/count API") Cc: stable@vger.kernel.org # 3.12+ Signed-off-by: Zhang Yi <yi.zhang@huawei.com> Reviewed-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20210522103045.690103-3-yi.zhang@huawei.com Signed-off-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
After commit c89128a00838 ("ext4: handle errors on
ext4_commit_super"), 'ret' may be set to 0 before calling
ext4_fill_flex_info(), if ext4_fill_flex_info() fails ext4_mount()
doesn't return error code, it makes 'root' is null which causes crash
in legacy_get_tree().
Fixes: c89128a00838 ("ext4: handle errors on ext4_commit_super") Reported-by: Hulk Robot <hulkci@huawei.com> Cc: <stable@vger.kernel.org> # v4.18+ Signed-off-by: Yang Yingliang <yangyingliang@huawei.com> Link: https://lore.kernel.org/r/20210510111051.55650-1-yangyingliang@huawei.com Signed-off-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
A code in iomap alloc may overflow block number when converting it to
byte offset. Luckily this is mostly harmless as we will just use more
expensive method of writing using unwritten extents even though we are
writing beyond i_size.
Cc: stable@kernel.org Fixes: 378f32bab371 ("ext4: introduce direct I/O write using iomap infrastructure") Signed-off-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20210412102333.2676-4-jack@suse.cz Signed-off-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Initialize eh_generation of struct ext4_extent_header to prevent leaking
info to userspace. Fixes KMSAN kernel-infoleak bug reported by syzbot at:
http://syzkaller.appspot.com/bug?id=78e9ad0e6952a3ca16e8234724b2fa92d041b9b8
In ext4_orphan_cleanup(), if ext4_truncate() failed to get a transaction
handle, it didn't remove the inode from the in-core orphan list, which
may probably trigger below error dump in ext4_destroy_inode() during the
final iput() and could lead to memory corruption on the later orphan
list changes.
The defrag loop processes leaves in batches and starting transaction for
each. The whole defragmentation on a given root is protected by a bit
but in case the transaction fails, the bit is not cleared
In case the transaction fails the bit would prevent starting
defragmentation again, so make sure it's cleared.
The early check if we should attempt compression does not take into
account the number of input pages. It can happen that there's only one
page, eg. a tail page after some ranges of the BTRFS_MAX_UNCOMPRESSED
have been processed, or an isolated page that won't be converted to an
inline extent.
The single page would be compressed but a later check would drop it
again because the result size must be at least one block shorter than
the input. That can never work with just one page.
During an incremental send operation, when processing the new references
for the current inode, we might send an unlink operation for another inode
that has a conflicting path and has more than one hard link. However this
path was computed and cached before we processed previous new references
for the current inode. We may have orphanized a directory of that path
while processing a previous new reference, in which case the path will
be invalid and cause the receiver process to fail.
The following reproducer triggers the problem and explains how/why it
happens in its comments:
$ cat test-send-unlink.sh
#!/bin/bash
DEV=/dev/sdi
MNT=/mnt/sdi
mkfs.btrfs -f $DEV >/dev/null
mount $DEV $MNT
# Create our test files and directory. Inode 259 (file3) has two hard
# links.
touch $MNT/file1
touch $MNT/file2
touch $MNT/file3
# Now create the base snapshot, which is going to be the parent snapshot
# for a later incremental send.
btrfs subvolume snapshot -r $MNT $MNT/snap1
btrfs send -f /tmp/snap1.send $MNT/snap1
# Move inode 257 into directory inode 260. This results in computing the
# path for inode 260 as "/A" and caching it.
mv $MNT/file1 $MNT/A/file1
# Move inode 258 (file2) into directory inode 260, with a name of
# "hard_link", moving first inode 259 away since it currently has that
# location and name.
mv $MNT/A/hard_link $MNT/tmp
mv $MNT/file2 $MNT/A/hard_link
# Now rename inode 260 to something else (B for example) and then create
# a hard link for inode 258 that has the old name and location of inode
# 260 ("/A").
mv $MNT/A $MNT/B
ln $MNT/B/hard_link $MNT/A
# Create another snapshot of our subvolume and use it for an incremental
# send.
btrfs subvolume snapshot -r $MNT $MNT/snap2
btrfs send -f /tmp/snap2.send -p $MNT/snap1 $MNT/snap2
# Now unmount the filesystem, create a new one, mount it and try to
# apply both send streams to recreate both snapshots.
umount $DEV
mkfs.btrfs -f $DEV >/dev/null
mount $DEV $MNT
# First add the first snapshot to the new filesystem by applying the
# first send stream.
btrfs receive -f /tmp/snap1.send $MNT
# The incremental receive operation below used to fail with the
# following error:
#
# ERROR: unlink A/hard_link failed: No such file or directory
#
# This is because when send is processing inode 257, it generates the
# path for inode 260 as "/A", since that inode is its parent in the send
# snapshot, and caches that path.
#
# Later when processing inode 258, it first processes its new reference
# that has the path of "/A", which results in orphanizing inode 260
# because there is a a path collision. This results in issuing a rename
# operation from "/A" to "/o260-6-0".
#
# Finally when processing the new reference "B/hard_link" for inode 258,
# it notices that it collides with inode 259 (not yet processed, because
# it has a higher inode number), since that inode has the name
# "hard_link" under the directory inode 260. It also checks that inode
# 259 has two hardlinks, so it decides to issue a unlink operation for
# the name "hard_link" for inode 259. However the path passed to the
# unlink operation is "/A/hard_link", which is incorrect since currently
# "/A" does not exists, due to the orphanization of inode 260 mentioned
# before. The path is incorrect because it was computed and cached
# before the orphanization. This results in the receiver to fail with
# the above error.
btrfs receive -f /tmp/snap2.send $MNT
umount $MNT
When running the test, it fails like this:
$ ./test-send-unlink.sh
Create a readonly snapshot of '/mnt/sdi' in '/mnt/sdi/snap1'
At subvol /mnt/sdi/snap1
Create a readonly snapshot of '/mnt/sdi' in '/mnt/sdi/snap2'
At subvol /mnt/sdi/snap2
At subvol snap1
At snapshot snap2
ERROR: unlink A/hard_link failed: No such file or directory
Fix this by recomputing a path before issuing an unlink operation when
processing the new references for the current inode if we previously
have orphanized a directory.
A test case for fstests will follow soon.
CC: stable@vger.kernel.org # 4.4+ 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>
Commit 3a95200d3f89 ("arm_pmu: Change API to support 64bit counter values")
changes the input "value" type from 32-bit to 64-bit, which introduces the
following problem: ARMv7 PMU counters is 32-bit width, in big-endian mode,
write counter uses high 32-bit, which writes an incorrect value.
Fixes: 3a95200d3f89 ("arm_pmu: Change API to support 64bit counter values") Cc: <stable@vger.kernel.org> Signed-off-by: Yang Jihong <yangjihong1@huawei.com> Acked-by: Mark Rutland <mark.rutland@arm.com> Link: https://lore.kernel.org/r/20210430012659.232110-1-yangjihong1@huawei.com Signed-off-by: Will Deacon <will@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Annotate the firmware files CCP might need using MODULE_FIRMWARE().
This will get them included into an initrd when CCP is also included
there. Otherwise the CCP module will not find its firmware when loaded
before the root-fs is mounted.
This can cause problems when the pre-loaded SEV firmware is too old to
support current SEV and SEV-ES virtualization features.
Fixes: e93720606efd ("crypto: ccp - Allow SEV firmware to be chosen based on Family and Model") Cc: stable@vger.kernel.org # v4.20+ Acked-by: Tom Lendacky <thomas.lendacky@amd.com> Signed-off-by: Joerg Roedel <jroedel@suse.de> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Even though we validate user-provided inputs we then traverse past
validated data when applying the new map. The issue was originally
discovered by Murray McAllister with this simple POC (if the following
is executed by an unprivileged user it will instantly panic the system):
int main(void) {
int fd, ret;
unsigned int buffer[10000];
Some environments do not set $SHELL when running tests. There's no
need to use $SHELL here anyway, since "cat" can be used to receive any
delivered signals from the kernel. Additionally avoid using bash-isms
in the command, and record stderr for posterity.
When checking the file name attribute, we want to ensure that it fits
within the bounds of ATTR_RECORD. To do this, we should check that (attr
record + file name offset + file name length) < (attr record + attr record
length).
However, the original check did not include the file name offset in the
calculation. This means that corrupted on-disk metadata might not caught
by the incorrect file name check, and lead to an invalid memory access.
An example can be seen in the crash report of a memory corruption error
found by Syzbot:
https://syzkaller.appspot.com/bug?id=a1a1e379b225812688566745c3e2f7242bffc246
Adding the file name offset to the validity check fixes this error and
passes the Syzbot reproducer test.
On filesystems with a block size smaller than PAGE_SIZE and non-empty
files smaller then PAGE_SIZE, gfs2_page_mkwrite could end up allocating
excess blocks beyond the end of the file, similar to fallocate. This
doesn't make sense; fix it.
Reported-by: Bob Peterson <rpeterso@redhat.com> Fixes: 184b4e60853d ("gfs2: Fix end-of-file handling in gfs2_page_mkwrite") Cc: stable@vger.kernel.org # v5.5+ Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
when system is doing s4, the process of xhci_resume may be as below:
1、xhci_mem_cleanup
2、xhci_init->xhci_mem_init->xhci_mem_cleanup(when memory is not enough).
xhci_mem_cleanup will be executed twice when system is out of memory.
xhci->port_caps is freed in xhci_mem_cleanup,but it isn't set to NULL.
It will be freed twice when xhci_mem_cleanup is called the second time.
We got following bug when system resumes from s4:
kernel BUG at mm/slub.c:309!
Internal error: Oops - BUG: 0 [#1] PREEMPT SMP
CPU: 0 PID: 5929 Tainted: G S W 5.4.96-arm64-desktop #1
pc : __slab_free+0x5c/0x424
lr : kfree+0x30c/0x32c
Creation EP's debugfs called earlier than debugfs folder for dwc3
device created. As result EP's debugfs are created in '/sys/kernel/debug'
instead of '/sys/kernel/debug/usb/dwc3.1.auto'.
Moved dwc3_debugfs_init() function call before calling
dwc3_core_init_mode() to allow create dwc3 debugfs parent before
creating EP's debugfs's.
The device (32a7:0000 Heimann Sensor GmbH USB appset demo) claims to be
a CDC-ACM device in its descriptors but in fact is not. If it is run
with echo disabled it returns garbled data, probably due to something
that happens in the TTY layer. And when run with echo enabled (the
default), it will mess up the calibration data of the sensor the first
time any data is sent to the device.
In short, I had a bad time after connecting the sensor and trying to get
it to work. I hope blacklisting it in the cdc-acm driver will save
someone else a bit of trouble.
when receive eem echo command, it will send a response,
but queue this response to the usb request which allocate
from gadget device endpoint zero,
and transmit the request to IN endpoint of eem interface.
on dwc3 gadget, it will trigger following warning in function
__dwc3_gadget_ep_queue(),
if (WARN(req->dep != dep, "request %pK belongs to '%s'\n",
&req->request, req->dep->name))
return -EINVAL;
fix it by allocating a usb request from IN endpoint of eem interface,
and transmit the usb request to same IN endpoint of eem interface.
In ems_usb_disconnect() dev pointer, which is netdev private data, is
used after free_candev() call:
| if (dev) {
| unregister_netdev(dev->netdev);
| free_candev(dev->netdev);
|
| unlink_all_urbs(dev);
|
| usb_free_urb(dev->intr_urb);
|
| kfree(dev->intr_in_buffer);
| kfree(dev->tx_msg_buffer);
| }
Fix it by simply moving free_candev() at the end of the block.
Fail log:
| BUG: KASAN: use-after-free in ems_usb_disconnect
| Read of size 8 at addr ffff88804e041008 by task kworker/1:2/2895
|
| CPU: 1 PID: 2895 Comm: kworker/1:2 Not tainted 5.13.0-rc5+ #164
| Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.14.0-0-g155821a-rebuilt.opensuse.4
| Workqueue: usb_hub_wq hub_event
| Call Trace:
| dump_stack (lib/dump_stack.c:122)
| print_address_description.constprop.0.cold (mm/kasan/report.c:234)
| kasan_report.cold (mm/kasan/report.c:420 mm/kasan/report.c:436)
| ems_usb_disconnect (drivers/net/can/usb/ems_usb.c:683 drivers/net/can/usb/ems_usb.c:1058)
Fixes: 702171adeed3 ("ems_usb: Added support for EMS CPC-USB/ARM7 CAN/USB interface") Link: https://lore.kernel.org/r/20210617185130.5834-1-paskripkin@gmail.com Cc: linux-stable <stable@vger.kernel.org> Signed-off-by: Pavel Skripkin <paskripkin@gmail.com> Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The direction of the pipe argument must match the request-type direction
bit or control requests may fail depending on the host-controller-driver
implementation.
Fix the four control requests which erroneously used usb_rcvctrlpipe().
Fixes: 1d3e20236d7a ("[PATCH] USB: usbtouchscreen: unified USB touchscreen driver") Fixes: 24ced062a296 ("usbtouchscreen: add support for DMC TSC-10/25 devices") Fixes: 9e3b25837a20 ("Input: usbtouchscreen - add support for e2i touchscreen controller") Signed-off-by: Johan Hovold <johan@kernel.org> Cc: stable@vger.kernel.org # 2.6.17 Link: https://lore.kernel.org/r/20210524092048.4443-1-johan@kernel.org Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The HP EliteBook 830 G8 Notebook PC using ALC285 codec which using 0x04 to
control mute LED and 0x01 to control micmute LED.
Therefore, add a quirk to make it works.
Asus Zenbook 14 UM431D has two speaker pins and a headphone pin, and
the auto-parser ends up assigning the bass to the third DAC 0x06.
Although the tone comes out, it's inconvenient because this DAC has no
volume control unlike two other DACs.
For obtaining the volume control for the bass speaker, this patch
enforces the mapping to let both front and bass speaker pins sharing
the same DAC. It's not ideal but a little bit of improvement.
Since we've already applied the same workaround for another ASUS
machine, we just need to hook the chain to the existing quirk.
On HP Spectre x360 15-df0xxx, after system boot with plugged headset, the
headset mic are not detected.
Moving pincfg and DAC's config to single fixup function fix this.
[ The actual bug in the original code was that it used a chain to
ALC286_FIXUP_SPEAKER2_TO_DAC1, and it contains not only the DAC1
route fix but also another chain to ALC269_FIXUP_THINKPAD_ACPI.
I thought the latter one is harmless for non-Thinkpad, but it
doesn't seem so; it contains again yet another chain to
ALC269_FIXUP_SKI_IGNORE, and this might be bad for some machines,
including this HP machine. -- tiwai ]
The HP EliteBook x360 830 G8 using ALC285 codec which using 0x04 to
control mute LED and 0x01 to control micmute LED.
Therefore, add a quirk to make it works.
The HP ProBook 630 G8 using ALC236 codec which using 0x02 to
control mute LED and 0x01 to control micmute LED.
Therefore, add a quirk to make it works.
The HP ProBook 445 G8 using ALC236 codec.
COEF index 0x34 bit 5 is used to control the playback mute LED, but the
microphone mute LED is controlled using pin VREF instead of a COEF index.
Therefore, add a quirk to make it works.
The HP ProBook 450 G8 using ALC236 codec which using 0x02 to
control mute LED and 0x01 to control micmute LED.
Therefore, add a quirk to make it works.
The recent workaround for the wild interrupts in commit c1f0616124c4
("ALSA: intel8x0: Don't update period unless prepared") leaded to a
regression, causing the interrupt storm during ac97 clock measurement
at the driver probe. We need to handle the interrupt while the clock
measurement as well as the proper PCM streams.
The current way of the scarlett2 mixer code managing the
usb_mixer_elem_info object is wrong in two ways: it passes its
internal index to the head.id field, and the val_type field is
uninitialized. This ended up with the wrong execution at the resume
because a bogus unit id is passed wrongly. Also, in the later code
extensions, we'll have more mixer elements, and passing the index will
overflow the unit id size (of 256).
This patch corrects those issues. It introduces a new value type,
USB_MIXER_BESPOKEN, which indicates a non-standard mixer element, and
use this type for all scarlett2 mixer elements, as well as
initializing the fixed unit id 0 for avoiding the overflow.
At extending the available mixer values for 32bit types, we forgot to
add the corresponding entries for the format dump in the proc output.
This may result in OOB access. Here adds the missing entries.
Even with rate limited reporting this is very spammy and since
it is remote device that is providing bogus data there is no
need to report this as error.
Since real_len variable was used only to allow conditional error
message it is now also removed.
[72454.143336] bt_err_ratelimited: 10 callbacks suppressed
[72454.143337] Bluetooth: hci0: advertising data len corrected
[72454.296314] Bluetooth: hci0: advertising data len corrected
[72454.892329] Bluetooth: hci0: advertising data len corrected
[72455.051319] Bluetooth: hci0: advertising data len corrected
[72455.357326] Bluetooth: hci0: advertising data len corrected
[72455.663295] Bluetooth: hci0: advertising data len corrected
[72455.787278] Bluetooth: hci0: advertising data len corrected
[72455.942278] Bluetooth: hci0: advertising data len corrected
[72456.094276] Bluetooth: hci0: advertising data len corrected
[72456.249137] Bluetooth: hci0: advertising data len corrected
[72459.416333] bt_err_ratelimited: 13 callbacks suppressed
[72459.416334] Bluetooth: hci0: advertising data len corrected
[72459.721334] Bluetooth: hci0: advertising data len corrected
[72460.011317] Bluetooth: hci0: advertising data len corrected
[72460.327171] Bluetooth: hci0: advertising data len corrected
[72460.638294] Bluetooth: hci0: advertising data len corrected
[72460.946350] Bluetooth: hci0: advertising data len corrected
[72461.225320] Bluetooth: hci0: advertising data len corrected
[72461.690322] Bluetooth: hci0: advertising data len corrected
[72462.118318] Bluetooth: hci0: advertising data len corrected
[72462.427319] Bluetooth: hci0: advertising data len corrected
[72464.546319] bt_err_ratelimited: 7 callbacks suppressed
[72464.546319] Bluetooth: hci0: advertising data len corrected
[72464.857318] Bluetooth: hci0: advertising data len corrected
[72465.163332] Bluetooth: hci0: advertising data len corrected
[72465.278331] Bluetooth: hci0: advertising data len corrected
[72465.432323] Bluetooth: hci0: advertising data len corrected
[72465.891334] Bluetooth: hci0: advertising data len corrected
[72466.045334] Bluetooth: hci0: advertising data len corrected
[72466.197321] Bluetooth: hci0: advertising data len corrected
[72466.340318] Bluetooth: hci0: advertising data len corrected
[72466.498335] Bluetooth: hci0: advertising data len corrected
[72469.803299] bt_err_ratelimited: 10 callbacks suppressed
struct firmware::data is marked const, and when the firmware is
compressed with xz (default at least with Fedora) it's mapped read-only
which results in a crash:
This also seems to fix a failure to suspend due to the firmware
download on bootup getting interrupted by the crash:
Bluetooth: hci0: SSR or FW download time out
PM: dpm_run_callback(): acpi_subsys_suspend+0x0/0x60 returns -110
PM: Device serial0-0 failed to suspend: error -110
PM: Some devices failed to suspend, or early wake event detected
In order to avoid a race condition for user events when changing
cpu affinity reset the active flag only when EOI-ing the event.
This is working fine as all user events are lateeoi events. Note that
lateeoi_ack_mask_dynirq() is not modified as there is no explicit call
to xen_irq_lateeoi() expected later.
Cc: stable@vger.kernel.org Reported-by: Julien Grall <julien@xen.org> Fixes: b6622798bc50b62 ("xen/events: avoid handling the same event on two cpus at the same time") Tested-by: Julien Grall <julien@xen.org> Signed-off-by: Juergen Gross <jgross@suse.com> Reviewed-by: Boris Ostrovsky <boris.ostrvsky@oracle.com> Link: https://lore.kernel.org/r/20210623130913.9405-1-jgross@suse.com Signed-off-by: Juergen Gross <jgross@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Some devices reference an output terminal as the source of extension
units. This is incorrect, as output terminals only have an input pin,
and thus can't be connected to any entity in the forward direction. The
resulting topology would cause issues when registering the media
controller graph. To avoid this problem, connect the extension unit to
the source of the output terminal instead.
While at it, and while no device has been reported to be affected by
this issue, also handle forward scans where two output terminals would
be connected together, and skip the terminals found through such an
invalid connection.
Reported-and-tested-by: John Nealy <jnealy3@yahoo.com> Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Signed-off-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The Facility Status and Control Register is a privileged SPR that
defines the availability of some features in problem state. Since it
can be written by the guest, we must restore it to the previous host
value after guest exit.
This restoration is currently done by taking the value from
current->thread.fscr, which in the P9 path is not enough anymore
because the guest could context switch the QEMU thread, causing the
guest-current value to be saved into the thread struct.
The above situation manifested when running a QEMU linked against a
libc with System Call Vectored support, which causes scv
instructions to be run by QEMU early during the guest boot (during
SLOF), at which point the FSCR is 0 due to guest entry. After a few
scv calls (1 to a couple hundred), the context switching happens and
the QEMU thread runs with the guest value, resulting in a Facility
Unavailable interrupt.
This patch saves and restores the host value of FSCR in the inner
guest entry loop in a way independent of current->thread.fscr. The old
way of doing it is still kept in place because it works for the old
entry path.
Restore CR4.LA57 to the mmu_role to fix an amusing edge case with nested
virtualization. When KVM (L0) is using TDP, CR4.LA57 is not reflected in
mmu_role.base.level because that tracks the shadow root level, i.e. TDP
level. Normally, this is not an issue because LA57 can't be toggled
while long mode is active, i.e. the guest has to first disable paging,
then toggle LA57, then re-enable paging, thus ensuring an MMU
reinitialization.
But if L1 is crafty, it can load a new CR4 on VM-Exit and toggle LA57
without having to bounce through an unpaged section. L1 can also load a
new CR3 on exit, i.e. it doesn't even need to play crazy paging games, a
single entry PML5 is sufficient. Such shenanigans are only problematic
if L0 and L1 use TDP, otherwise L1 and L2 share an MMU that gets
reinitialized on nested VM-Enter/VM-Exit due to mmu_role.base.guest_mode.
Note, in the L2 case with nested TDP, even though L1 can switch between
L2s with different LA57 settings, thus bypassing the paging requirement,
in that case KVM's nested_mmu will track LA57 in base.level.
Allow creating FDB steering rules only when in switchdev mode.
The only software model where a userspace application can manipulate
FDB entries is when it manages the eswitch. This is only possible in
switchdev mode where we expose a single RDMA device with representors
for all the vports that are connected to the eswitch.
Fixes: 52438be44112 ("RDMA/mlx5: Allow inserting a steering rule to the FDB") Link: https://lore.kernel.org/r/e928ae7c58d07f104716a2a8d730963d1bd01204.1623052923.git.leonro@nvidia.com Reviewed-by: Maor Gottlieb <maorg@nvidia.com> Signed-off-by: Mark Bloch <mbloch@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
[sudip: use old mlx5_eswitch_mode] Signed-off-by: Sudip Mukherjee <sudipm.mukherjee@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Both of these drivers use ioport_map(), so they need to
depend on HAS_IOPORT_MAP. Otherwise, they cannot be built
even with COMPILE_TEST on architectures without an ioport
implementation, such as ARCH=um.
Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
A disabled/masked interrupt marked as wakeup source must be re-enable
and unmasked in order to be able to wake-up the host. That can be done
by flaging the irqchip with IRQCHIP_ENABLE_WAKEUP_ON_SUSPEND.
Note: It 'sometimes' works without that change, but only thanks to the
lazy generic interrupt disabling (keeping interrupt unmasked).
Handle a reported media event code of 3. This indicates that the media has
been removed from the drive and user intervention is required to proceed.
Return DISK_EVENT_EJECT_REQUEST in that case.
During boot the Secure Boot Forbidden Signature Database, dbx,
is loaded into the blacklist keyring. Systems booted with shim
have an equivalent Forbidden Signature Database called mokx.
Currently mokx is only used by shim and grub, the contents are
ignored by the kernel.
Add the ability to load mokx into the blacklist keyring during boot.
Add a new Kconfig option called SYSTEM_REVOCATION_KEYS. If set,
this option should be the filename of a PEM-formated file containing
X.509 certificates to be included in the default blacklist keyring.
DH Changes:
- Make the new Kconfig option depend on SYSTEM_REVOCATION_LIST.
- Fix SYSTEM_REVOCATION_KEYS=n, but CONFIG_SYSTEM_REVOCATION_LIST=y[1][2].
- Use CONFIG_SYSTEM_REVOCATION_LIST for extract-cert[3].
- Use CONFIG_SYSTEM_REVOCATION_LIST for revocation_certificates.o[3].
The Secure Boot Forbidden Signature Database, dbx, contains a list of now
revoked signatures and keys previously approved to boot with UEFI Secure
Boot enabled. The dbx is capable of containing any number of
EFI_CERT_X509_SHA256_GUID, EFI_CERT_SHA256_GUID, and EFI_CERT_X509_GUID
entries.
Currently when EFI_CERT_X509_GUID are contained in the dbx, the entries are
skipped.
Add support for EFI_CERT_X509_GUID dbx entries. When a EFI_CERT_X509_GUID
is found, it is added as an asymmetrical key to the .blacklist keyring.
Anytime the .platform keyring is used, the keys in the .blacklist keyring
are referenced, if a matching key is found, the key will be rejected.
[DH: Made the following changes:
- Added to have a config option to enable the facility. This allows a
Kconfig solution to make sure that pkcs7_validate_trust() is
enabled.[1][2]
- Moved the functions out from the middle of the blacklist functions.
- Added kerneldoc comments.]
It's not sufficient to skip reading when the pos is beyond the EOF.
There may be data at the head of the page that we need to fill in
before the write.
Add a new helper function that corrects and clarifies the logic of
when we can skip reads, and have it only zero out the part of the page
that won't have data copied in for the write.
Finally, don't set the page Uptodate after zeroing. It's not up to date
since the write data won't have been copied in yet.
[DH made the following changes:
- Prefixed the new function with "netfs_".
- Don't call zero_user_segments() for a full-page write.
- Altered the beyond-last-page check to avoid a DIV instruction and got
rid of then-redundant zero-length file check.
]
[ Note: this fix is commit 827a746f405d in mainline kernels. The
original bug was in ceph, but got lifted into the fs/netfs
library for v5.13. This backport should apply to stable
kernels v5.10 though v5.12. ]
in case of driver wants to sync part of ranges with offset,
swiotlb_tbl_sync_single() copies from orig_addr base to tlb_addr with
offset and ends up with data mismatch.
It was removed from
"swiotlb: don't modify orig_addr in swiotlb_tbl_sync_single",
but said logic has to be added back in.
From Linus's email:
"That commit which the removed the offset calculation entirely, because the old
(unsigned long)tlb_addr & (IO_TLB_SIZE - 1)
was wrong, but instead of removing it, I think it should have just
fixed it to be
(tlb_addr - mem->start) & (IO_TLB_SIZE - 1);
instead. That way the slot offset always matches the slot index calculation."
(Unfortunatly that broke NVMe).
The use-case that drivers are hitting is as follow:
[One fix which was Linus's that made more sense to as it created a
symmetry would break NVMe. The reason for that is the:
unsigned int offset = (tlb_addr - mem->start) & (IO_TLB_SIZE - 1);
would come up with the proper offset, but it would lose the
alignment (which this patch contains).]
Fixes: 16fc3cef33a0 ("swiotlb: don't modify orig_addr in swiotlb_tbl_sync_single") Signed-off-by: Bumyong Lee <bumyong.lee@samsung.com> Signed-off-by: Chanho Park <chanho61.park@samsung.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reported-by: Dominique MARTINET <dominique.martinet@atmark-techno.com> Reported-by: Horia Geantă <horia.geanta@nxp.com> Tested-by: Horia Geantă <horia.geanta@nxp.com> CC: stable@vger.kernel.org Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Send SEV_CMD_DECOMMISSION command to PSP firmware if ASID binding
fails. If a failure happens after a successful LAUNCH_START command,
a decommission command should be executed. Otherwise, guest context
will be unfreed inside the AMD SP. After the firmware will not have
memory to allocate more SEV guest context, LAUNCH_START command will
begin to fail with SEV_RET_RESOURCE_LIMIT error.
The existing code calls decommission inside sev_unbind_asid, but it is
not called if a failure happens before guest activation succeeds. If
sev_bind_asid fails, decommission is never called. PSP firmware has a
limit for the number of guests. If sev_asid_binding fails many times,
PSP firmware will not have resources to create another guest context.
Cc: stable@vger.kernel.org Fixes: 59414c989220 ("KVM: SVM: Add support for KVM_SEV_LAUNCH_START command") Reported-by: Peter Gonda <pgonda@google.com> Signed-off-by: Alper Gun <alpergun@google.com> Reviewed-by: Marc Orr <marcorr@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20210610174604.2554090-1-alpergun@google.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If more than one futex is placed on a shmem huge page, it can happen
that waking the second wakes the first instead, and leaves the second
waiting: the key's shared.pgoff is wrong.
When 3.11 commit 13d60f4b6ab5 ("futex: Take hugepages into account when
generating futex_key"), the only shared huge pages came from hugetlbfs,
and the code added to deal with its exceptional page->index was put into
hugetlb source. Then that was missed when 4.8 added shmem huge pages.
page_to_pgoff() is what others use for this nowadays: except that, as
currently written, it gives the right answer on hugetlbfs head, but
nonsense on hugetlbfs tails. Fix that by calling hugetlbfs-specific
hugetlb_basepage_index() on PageHuge tails as well as on head.
Yes, it's unconventional to declare hugetlb_basepage_index() there in
pagemap.h, rather than in hugetlb.h; but I do not expect anything but
page_to_pgoff() ever to need it.
[akpm@linux-foundation.org: give hugetlb_basepage_index() prototype the correct scope]
Link: https://lkml.kernel.org/r/b17d946b-d09-326e-b42a-52884c36df32@google.com Fixes: 800d8c63b2e9 ("shmem: add huge pages support") Reported-by: Neel Natu <neelnatu@google.com> Signed-off-by: Hugh Dickins <hughd@google.com> Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org> Acked-by: Thomas Gleixner <tglx@linutronix.de> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Zhang Yi <wetpzy@gmail.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Darren Hart <dvhart@infradead.org> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Aha! Shouldn't that quick scan over pte_none()s make sure that it holds
ptlock in the PVMW_SYNC case? That too might have been responsible for
BUGs or WARNs in split_huge_page_to_list() or its unmap_page(), though
I've never seen any.
Link: https://lkml.kernel.org/r/1bdf384c-8137-a149-2a1e-475a4791c3c@google.com Link: https://lore.kernel.org/linux-mm/20210412180659.B9E3.409509F4@e16-tech.com/ Fixes: ace71a19cec5 ("mm: introduce page_vma_mapped_walk()") Signed-off-by: Hugh Dickins <hughd@google.com> Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Tested-by: Wang Yugui <wangyugui@e16-tech.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Peter Xu <peterx@redhat.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Zi Yan <ziy@nvidia.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Running certain tests with a DEBUG_VM kernel would crash within hours,
on the total_mapcount BUG() in split_huge_page_to_list(), while trying
to free up some memory by punching a hole in a shmem huge page: split's
try_to_unmap() was unable to find all the mappings of the page (which,
on a !DEBUG_VM kernel, would then keep the huge page pinned in memory).
Crash dumps showed two tail pages of a shmem huge page remained mapped
by pte: ptes in a non-huge-aligned vma of a gVisor process, at the end
of a long unmapped range; and no page table had yet been allocated for
the head of the huge page to be mapped into.
Although designed to handle these odd misaligned huge-page-mapped-by-pte
cases, page_vma_mapped_walk() falls short by returning false prematurely
when !pmd_present or !pud_present or !p4d_present or !pgd_present: there
are cases when a huge page may span the boundary, with ptes present in
the next.
Restructure page_vma_mapped_walk() as a loop to continue in these cases,
while keeping its layout much as before. Add a step_forward() helper to
advance pvmw->address across those boundaries: originally I tried to use
mm's standard p?d_addr_end() macros, but hit the same crash 512 times
less often: because of the way redundant levels are folded together, but
folded differently in different configurations, it was just too
difficult to use them correctly; and step_forward() is simpler anyway.
Link: https://lkml.kernel.org/r/fedb8632-1798-de42-f39e-873551d5bc81@google.com Fixes: ace71a19cec5 ("mm: introduce page_vma_mapped_walk()") Signed-off-by: Hugh Dickins <hughd@google.com> Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Peter Xu <peterx@redhat.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Wang Yugui <wangyugui@e16-tech.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Zi Yan <ziy@nvidia.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
page_vma_mapped_walk() cleanup: add a level of indentation to much of
the body, making no functional change in this commit, but reducing the
later diff when this is all converted to a loop.
[hughd@google.com: : page_vma_mapped_walk(): add a level of indentation fix] Link: https://lkml.kernel.org/r/7f817555-3ce1-c785-e438-87d8efdcaf26@google.com Link: https://lkml.kernel.org/r/efde211-f3e2-fe54-977-ef481419e7f3@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Peter Xu <peterx@redhat.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Wang Yugui <wangyugui@e16-tech.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Zi Yan <ziy@nvidia.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
page_vma_mapped_walk() cleanup: adjust the test for crossing page table
boundary - I believe pvmw->address is always page-aligned, but nothing
else here assumed that; and remember to reset pvmw->pte to NULL after
unmapping the page table, though I never saw any bug from that.
Link: https://lkml.kernel.org/r/799b3f9c-2a9e-dfef-5d89-26e9f76fd97@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Peter Xu <peterx@redhat.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Wang Yugui <wangyugui@e16-tech.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Zi Yan <ziy@nvidia.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
page_vma_mapped_walk() cleanup: rearrange the !pmd_present() block to
follow the same "return not_found, return not_found, return true"
pattern as the block above it (note: returning not_found there is never
premature, since existence or prior existence of huge pmd guarantees
good alignment).
Link: https://lkml.kernel.org/r/378c8650-1488-2edf-9647-32a53cf2e21@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Reviewed-by: Peter Xu <peterx@redhat.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Wang Yugui <wangyugui@e16-tech.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Zi Yan <ziy@nvidia.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When debugging the bug reported by Wang Yugui [1], try_to_unmap() may
fail, but the first VM_BUG_ON_PAGE() just checks page_mapcount() however
it may miss the failure when head page is unmapped but other subpage is
mapped. Then the second DEBUG_VM BUG() that check total mapcount would
catch it. This may incur some confusion.
As this is not a fatal issue, so consolidate the two DEBUG_VM checks
into one VM_WARN_ON_ONCE_PAGE().
Link: https://lkml.kernel.org/r/d0f0db68-98b8-ebfb-16dc-f29df24cf012@google.com Signed-off-by: Yang Shi <shy828301@gmail.com> Reviewed-by: Zi Yan <ziy@nvidia.com> Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Jan Kara <jack@suse.cz> Cc: Jue Wang <juew@google.com> Cc: "Matthew Wilcox (Oracle)" <willy@infradead.org> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Peter Xu <peterx@redhat.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Wang Yugui <wangyugui@e16-tech.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Note on stable backport: fixed up variables in split_huge_page_to_list().
There is a race between THP unmapping and truncation, when truncate sees
pmd_none() and skips the entry, after munmap's zap_huge_pmd() cleared
it, but before its page_remove_rmap() gets to decrement
compound_mapcount: generating false "BUG: Bad page cache" reports that
the page is still mapped when deleted. This commit fixes that, but not
in the way I hoped.
The first attempt used try_to_unmap(page, TTU_SYNC|TTU_IGNORE_MLOCK)
instead of unmap_mapping_range() in truncate_cleanup_page(): it has
often been an annoyance that we usually call unmap_mapping_range() with
no pages locked, but there apply it to a single locked page.
try_to_unmap() looks more suitable for a single locked page.
However, try_to_unmap_one() contains a VM_BUG_ON_PAGE(!pvmw.pte,page):
it is used to insert THP migration entries, but not used to unmap THPs.
Copy zap_huge_pmd() and add THP handling now? Perhaps, but their TLB
needs are different, I'm too ignorant of the DAX cases, and couldn't
decide how far to go for anon+swap. Set that aside.
The second attempt took a different tack: make no change in truncate.c,
but modify zap_huge_pmd() to insert an invalidated huge pmd instead of
clearing it initially, then pmd_clear() between page_remove_rmap() and
unlocking at the end. Nice. But powerpc blows that approach out of the
water, with its serialize_against_pte_lookup(), and interesting pgtable
usage. It would need serious help to get working on powerpc (with a
minor optimization issue on s390 too). Set that aside.
Just add an "if (page_mapped(page)) synchronize_rcu();" or other such
delay, after unmapping in truncate_cleanup_page()? Perhaps, but though
that's likely to reduce or eliminate the number of incidents, it would
give less assurance of whether we had identified the problem correctly.
This successful iteration introduces "unmap_mapping_page(page)" instead
of try_to_unmap(), and goes the usual unmap_mapping_range_tree() route,
with an addition to details. Then zap_pmd_range() watches for this
case, and does spin_unlock(pmd_lock) if so - just like
page_vma_mapped_walk() now does in the PVMW_SYNC case. Not pretty, but
safe.
Note that unmap_mapping_page() is doing a VM_BUG_ON(!PageLocked) to
assert its interface; but currently that's only used to make sure that
page->mapping is stable, and zap_pmd_range() doesn't care if the page is
locked or not. Along these lines, in invalidate_inode_pages2_range()
move the initial unmap_mapping_range() out from under page lock, before
then calling unmap_mapping_page() under page lock if still mapped.
Link: https://lkml.kernel.org/r/a2a4a148-cdd8-942c-4ef8-51b77f643dbe@google.com Fixes: fc127da085c2 ("truncate: handle file thp") Signed-off-by: Hugh Dickins <hughd@google.com> Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Reviewed-by: Yang Shi <shy828301@gmail.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Jan Kara <jack@suse.cz> Cc: Jue Wang <juew@google.com> Cc: "Matthew Wilcox (Oracle)" <willy@infradead.org> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Peter Xu <peterx@redhat.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Wang Yugui <wangyugui@e16-tech.com> Cc: Zi Yan <ziy@nvidia.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Note on stable backport: fixed up call to truncate_cleanup_page()
in truncate_inode_pages_range().
Anon THP tails were already supported, but memory-failure may need to
use page_address_in_vma() on file THP tails, which its page->mapping
check did not permit: fix it.
hughd adds: no current usage is known to hit the issue, but this does
fix a subtle trap in a general helper: best fixed in stable sooner than
later.
Link: https://lkml.kernel.org/r/a0d9b53-bf5d-8bab-ac5-759dc61819c1@google.com Fixes: 800d8c63b2e9 ("shmem: add huge pages support") Signed-off-by: Jue Wang <juew@google.com> Signed-off-by: Hugh Dickins <hughd@google.com> Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org> Reviewed-by: Yang Shi <shy828301@gmail.com> Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Jan Kara <jack@suse.cz> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Peter Xu <peterx@redhat.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Wang Yugui <wangyugui@e16-tech.com> Cc: Zi Yan <ziy@nvidia.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
projects / thirdparty / kernel / stable.git / log
