Lech writes:
After upgrading kernel on our boards from v4.19.105 to v4.19.106
we found out that syslog fails to read the messages after ones
read initially after opening /proc/kmsg just after booting.
I also found out, that output of 'dmesg --follow' also doesn't
react on new printks appearing for whatever reason - to read new
messages, reopening /proc/kmsg or /dev/kmsg was needed.
I bisected this down to commit 15341b1dd409749fa5625e4b632013b6ba81609b ("char/random: silence
a lockdep splat with printk()"), and reverting it on top of
v4.19.106 restored correct behaviour.
While people dig to find out how such an odd change causes a lockup,
let's just revert this for now as it's not all that big of a deal for
4.19.y.
In file included from ../arch/s390/purgatory/purgatory.c:10:
In file included from ../include/linux/kexec.h:18:
In file included from ../include/linux/crash_core.h:6:
In file included from ../include/linux/elfcore.h:5:
In file included from ../include/linux/user.h:1:
In file included from ../arch/s390/include/asm/user.h:11:
../arch/s390/include/asm/page.h:45:6: warning: converting the result of
'<<' to a boolean always evaluates to false
[-Wtautological-constant-compare]
if (PAGE_DEFAULT_KEY)
^
../arch/s390/include/asm/page.h:23:44: note: expanded from macro
'PAGE_DEFAULT_KEY'
#define PAGE_DEFAULT_KEY (PAGE_DEFAULT_ACC << 4)
^
1 warning generated.
Explicitly compare this against zero to silence the warning as it is
intended to be used in a boolean context.
xen_maybe_preempt_hcall() is called from the exception entry point
xen_do_hypervisor_callback with interrupts disabled.
_cond_resched() evades the might_sleep() check in cond_resched() which
would have caught that and schedule_debug() unfortunately lacks a check
for irqs_disabled().
Enable interrupts around the call and use cond_resched() to catch future
issues.
Fixes: fdfd811ddde3 ("x86/xen: allow privcmd hypercalls to be preempted") Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Link: https://lore.kernel.org/r/878skypjrh.fsf@nanos.tec.linutronix.de Reviewed-by: Juergen Gross <jgross@suse.com> Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
device_shutdown() called from reboot or power_shutdown expect
all devices to be shutdown. Same is true for even ahci pci driver.
As no ahci shutdown function is implemented, the ata subsystem
always remains alive with DMA & interrupt support. File system
related calls should not be honored after device_shutdown().
So defining ahci pci driver shutdown to freeze hardware (mask
interrupt, stop DMA engine and free DMA resources).
rxrpc_rcu_destroy_call(), which is called as an RCU callback to clean up a
put call, calls rxrpc_put_connection() which, deep in its bowels, takes a
number of spinlocks in a non-BH-safe way, including rxrpc_conn_id_lock and
local->client_conns_lock. RCU callbacks, however, are normally called from
softirq context, which can cause lockdep to notice the locking
inconsistency.
To get lockdep to detect this, it's necessary to have the connection
cleaned up on the put at the end of the last of its calls, though normally
the clean up is deferred. This can be induced, however, by starting a call
on an AF_RXRPC socket and then closing the socket without reading the
reply.
Fix this by having rxrpc_rcu_destroy_call() punt the destruction to a
workqueue if in softirq-mode and defer the destruction to process context.
Note that another way to fix this could be to add a bunch of bh-disable
annotations to the spinlocks concerned - and there might be more than just
those two - but that means spending more time with BHs disabled.
Note also that some of these places were covered by bh-disable spinlocks
belonging to the rxrpc_transport object, but these got removed without the
_bh annotation being retained on the next lock in.
Fixes: 999b69f89241 ("rxrpc: Kill the client connection bundle concept") Reported-by: syzbot+d82f3ac8d87e7ccbb2c9@syzkaller.appspotmail.com Reported-by: syzbot+3f1fd6b8cbf8702d134e@syzkaller.appspotmail.com Signed-off-by: David Howells <dhowells@redhat.com>
cc: Hillf Danton <hdanton@sina.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The user-specified hashtable size is unbound, this could
easily lead to an OOM or a hung task as we hold the global
mutex while allocating and initializing the new hashtable.
Add a max value to cap both cfg->size and cfg->max, as
suggested by Florian.
Reported-and-tested-by: syzbot+adf6c6c2be1c3a718121@syzkaller.appspotmail.com Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com> Reviewed-by: Florian Westphal <fw@strlen.de> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
snd_seq_check_queue() passes the current tick and time of the given
queue as a pointer to snd_seq_prioq_cell_out(), but those might be
updated concurrently by the seq timer update.
Fix it by retrieving the current tick and time via the proper helper
functions at first, and pass those values to snd_seq_prioq_cell_out()
later in the loops.
snd_seq_timer_get_cur_time() takes a new argument and adjusts with the
current system time only when it's requested so; this update isn't
needed for snd_seq_check_queue(), as it's called either from the
interrupt handler or right after queuing.
Also, snd_seq_timer_get_cur_tick() is changed to read the value in the
spinlock for the concurrency, too.
The queue flags are represented in bit fields and the concurrent
access may result in unexpected results. Although the current code
should be mostly OK as it's only reading a field while writing other
fields as KCSAN reported, it's safer to cover both with a proper
spinlock protection.
This patch fixes the possible concurrent read by protecting with
q->owner_lock. Also the queue owner field is protected as well since
it's the field to be protected by the lock itself.
The rawmidi state flags (opened, append, active_sensing) are stored in
bit fields that can be potentially racy when concurrently accessed
without any locks. Although the current code should be fine, there is
also no any real benefit by keeping the bitfields for this kind of
short number of members.
This patch changes those bit fields flags to the simple bool fields.
There should be no size increase of the snd_rawmidi_substream by this
change.
This if guards whether user-space wants a copy of the offload-jited
bytecode and whether this bytecode exists. By erroneously doing a bitwise
AND instead of a logical AND on user- and kernel-space buffer-size can lead
to no data being copied to user-space especially when user-space size is a
power of two and bigger then the kernel-space buffer.
Fixes: fcfb126defda ("bpf: add new jited info fields in bpf_dev_offload and bpf_prog_info") Signed-off-by: Johannes Krude <johannes@krude.de> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Jakub Kicinski <kuba@kernel.org> Link: https://lore.kernel.org/bpf/20200212193227.GA3769@phlox.h.transitiv.net Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Qian Cai reported that the WARN_ON() in the x86/msi affinity setting code,
which catches cases where the affinity setting is not done on the CPU which
is the current target of the interrupt, triggers during CPU hotplug stress
testing.
It turns out that the warning which was added with the commit addressing
the MSI affinity race unearthed yet another long standing bug.
If user space writes a bogus affinity mask, i.e. it contains no online CPUs,
then it calls irq_select_affinity_usr(). This was introduced for ALPHA in
eee45269b0f5 ("[PATCH] Alpha: convert to generic irq framework (generic part)")
and subsequently made available for all architectures in
which introduced the circumvention of the affinity setting restrictions for
interrupt which cannot be moved in process context.
The whole exercise is bogus in various aspects:
1) If the interrupt is already started up then there is absolutely
no point to honour a bogus interrupt affinity setting from user
space. The interrupt is already assigned to an online CPU and it
does not make any sense to reassign it to some other randomly
chosen online CPU.
2) If the interupt is not yet started up then there is no point
either. A subsequent startup of the interrupt will invoke
irq_setup_affinity() anyway which will chose a valid target CPU.
So the only correct solution is to just return -EINVAL in case user space
wrote an affinity mask which does not contain any online CPUs, except for
ALPHA which has it's own magic sauce for this.
Fixes: 18404756765c ("genirq: Expose default irq affinity mask (take 3)") Reported-by: Qian Cai <cai@lca.pw> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Qian Cai <cai@lca.pw> Link: https://lkml.kernel.org/r/878sl8xdbm.fsf@nanos.tec.linutronix.de Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The intel_svm_is_pasid_valid() needs to be marked inline, otherwise it
causes the compile warning below:
CC [M] drivers/dma/idxd/cdev.o
In file included from drivers/dma/idxd/cdev.c:9:0:
./include/linux/intel-svm.h:125:12: warning: ‘intel_svm_is_pasid_valid’ defined but not used [-Wunused-function]
static int intel_svm_is_pasid_valid(struct device *dev, int pasid)
^~~~~~~~~~~~~~~~~~~~~~~~
Reported-by: Borislav Petkov <bp@alien8.de> Fixes: 15060aba71711 ('iommu/vt-d: Helper function to query if a pasid has any active users') Signed-off-by: Joerg Roedel <jroedel@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In crypt_scatterlist, if the crypt_stat argument is not set up
correctly, the kernel crashes. Instead, by returning an error code
upstream, the error is handled safely.
The issue is detected via a static analysis tool written by us.
When we call kobject_put() and it's the last reference to the kobject
then it calls gb_audio_module_release() and frees module. We dereference
"module" on the next line which is a use after free.
Currently the rtw_sprintf prints the contents of thread_name
onto thread_name and this can lead to a potential copy of a
string over itself. Avoid this by printing the literal string RTWHALXT
instread of the contents of thread_name.
Addresses-Coverity: ("copy of overlapping memory") Fixes: 554c0a3abf21 ("staging: Add rtl8723bs sdio wifi driver") Signed-off-by: Colin Ian King <colin.king@canonical.com> Reviewed-by: Hans de Goede <hdegoede@redhat.com> Link: https://lore.kernel.org/r/20200126220549.9849-1-colin.king@canonical.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
SuperSpeedPlus peripherals must report their bMaxPower of the
configuration descriptor in units of 8mA as per the USB 3.2
specification. The current switch statement in encode_bMaxPower()
only checks for USB_SPEED_SUPER but not USB_SPEED_SUPER_PLUS so
the latter falls back to USB 2.0 encoding which uses 2mA units.
Replace the switch with a simple if/else.
In btrfs_wait_ordered_range() once we find an ordered extent that has
finished with an error we exit the loop and don't wait for any other
ordered extents that might be still in progress.
All the users of btrfs_wait_ordered_range() expect that there are no more
ordered extents in progress after that function returns. So past fixes
such like the ones from the two following commits:
ff612ba7849964 ("btrfs: fix panic during relocation after ENOSPC before
writeback happens")
28aeeac1dd3080 ("Btrfs: fix panic when starting bg cache writeout after
IO error")
don't work when there are multiple ordered extents in the range.
Fix that by making btrfs_wait_ordered_range() wait for all ordered extents
even after it finds one that had an error.
Link: https://github.com/kdave/btrfs-progs/issues/228#issuecomment-569777554 CC: stable@vger.kernel.org # 4.4+ Reviewed-by: Qu Wenruo <wqu@suse.com> 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>
btrfs_assert_delayed_root_empty() will check if the delayed root is
completely empty, but this is a filesystem-wide check. On cleanup we
may have allowed other transactions to begin, for whatever reason, and
thus the delayed root is not empty.
So remove this check from cleanup_one_transation(). This however can
stay in btrfs_cleanup_transaction(), because it checks only after all of
the transactions have been properly cleaned up, and thus is valid.
CC: stable@vger.kernel.org # 4.4+ Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Reviewed-by: Nikolay Borisov <nborisov@suse.com> Reviewed-by: Qu Wenruo <wqu@suse.com> Signed-off-by: Josef Bacik <josef@toxicpanda.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>
While running my error injection script I hit a panic when we tried to
clean up the fs_root when freeing the fs_root. This is because
fs_info->fs_root == PTR_ERR(-EIO), which isn't great. Fix this by
setting fs_info->fs_root = NULL; if we fail to read the root.
CC: stable@vger.kernel.org # 4.4+ Reviewed-by: Nikolay Borisov <nborisov@suse.com> Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Reviewed-by: Qu Wenruo <wqu@suse.com> Signed-off-by: Josef Bacik <josef@toxicpanda.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>
This happens if we fail to insert our reserved file extent. At this
point we've already converted our reservation from ->bytes_may_use to
->bytes_reserved. However once we break we will attempt to free
everything from [cur_offset, end] from ->bytes_may_use, but our extent
reservation will overlap part of this.
Fix this problem by adding ins.offset (our extent allocation size) to
cur_offset so we remove the actual remaining part from ->bytes_may_use.
I validated this fix using my inject-error.py script
When pv_eoi_get_user() fails, 'val' may remain uninitialized and the return
value of pv_eoi_get_pending() becomes random. Fix the issue by initializing
the variable.
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com> Signed-off-by: Miaohe Lin <linmiaohe@huawei.com> Cc: stable@vger.kernel.org Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Even when APICv is disabled for L1 it can (and, actually, is) still
available for L2, this means we need to always call
vmx_deliver_nested_posted_interrupt() when attempting an interrupt
delivery.
Suggested-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com> Cc: stable@vger.kernel.org Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Consult the 'unconditional IO exiting' and 'use IO bitmaps' VM-execution
controls when checking instruction interception. If the 'use IO bitmaps'
VM-execution control is 1, check the instruction access against the IO
bitmaps to determine if the instruction causes a VM-exit.
Signed-off-by: Oliver Upton <oupton@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If EXT4_EXTENTS_FL is set on an inode while ext4_writepages() is running
on it, the following warning in ext4_add_complete_io() can be hit:
WARNING: CPU: 1 PID: 0 at fs/ext4/page-io.c:234 ext4_put_io_end_defer+0xf0/0x120
Here's a minimal reproducer (not 100% reliable) (root isn't required):
while true; do
sync
done &
while true; do
rm -f file
touch file
chattr -e file
echo X >> file
chattr +e file
done
The problem is that in ext4_writepages(), ext4_should_dioread_nolock()
(which only returns true on extent-based files) is checked once to set
the number of reserved journal credits, and also again later to select
the flags for ext4_map_blocks() and copy the reserved journal handle to
ext4_io_end::handle. But if EXT4_EXTENTS_FL is being concurrently set,
the first check can see dioread_nolock disabled while the later one can
see it enabled, causing the reserved handle to unexpectedly be NULL.
Since changing EXT4_EXTENTS_FL is uncommon, and there may be other races
related to doing so as well, fix this by synchronizing changing
EXT4_EXTENTS_FL with ext4_writepages() via the existing
s_writepages_rwsem (previously called s_journal_flag_rwsem).
This was originally reported by syzbot without a reproducer at
https://syzkaller.appspot.com/bug?extid=2202a584a00fffd19fbf,
but now that dioread_nolock is the default I also started seeing this
when running syzkaller locally.
Link: https://lore.kernel.org/r/20200219183047.47417-3-ebiggers@kernel.org Reported-by: syzbot+2202a584a00fffd19fbf@syzkaller.appspotmail.com Fixes: 6b523df4fb5a ("ext4: use transaction reservation for extent conversion in ext4_end_io") Signed-off-by: Eric Biggers <ebiggers@google.com> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Reviewed-by: Jan Kara <jack@suse.cz> Cc: stable@kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In preparation for making s_journal_flag_rwsem synchronize
ext4_writepages() with changes to both the EXTENTS and JOURNAL_DATA
flags (rather than just JOURNAL_DATA as it does currently), rename it to
s_writepages_rwsem.
Link: https://lore.kernel.org/r/20200219183047.47417-2-ebiggers@kernel.org Signed-off-by: Eric Biggers <ebiggers@google.com> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Reviewed-by: Jan Kara <jack@suse.cz> Cc: stable@kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When CONFIG_QFMT_V2 is configured as a module, the test in
ext4_feature_set_ok() fails and so mount of filesystems with quota or
project features fails. Fix the test to use IS_ENABLED macro which
works properly even for modules.
Link: https://lore.kernel.org/r/20200221100835.9332-1-jack@suse.cz Fixes: d65d87a07476 ("ext4: improve explanation of a mount failure caused by a misconfigured kernel") Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Cc: stable@kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
During an online resize an array of s_flex_groups structures gets replaced
so it can get enlarged. If there is a concurrent access to the array and
this memory has been reused then this can lead to an invalid memory access.
The s_flex_group array has been converted into an array of pointers rather
than an array of structures. This is to ensure that the information
contained in the structures cannot get out of sync during a resize due to
an accessor updating the value in the old structure after it has been
copied but before the array pointer is updated. Since the structures them-
selves are no longer copied but only the pointers to them this case is
mitigated.
During an online resize an array of pointers to s_group_info gets replaced
so it can get enlarged. If there is a concurrent access to the array in
ext4_get_group_info() and this memory has been reused then this can lead to
an invalid memory access.
During an online resize an array of pointers to buffer heads gets
replaced so it can get enlarged. If there is a racing block
allocation or deallocation which uses the old array, and the old array
has gotten reused this can lead to a GPF or some other random kernel
memory getting modified.
We tested a soft lockup problem in linux 4.19 which could also
be found in linux 5.x.
When dir inode takes up a large number of blocks, and if the
directory is growing when we are searching, it's possible the
restart branch could be called many times, and the do while loop
could hold cpu a long time.
EXT4_I(inode)->i_disksize could be accessed concurrently as noticed by
KCSAN,
BUG: KCSAN: data-race in ext4_write_end [ext4] / ext4_writepages [ext4]
write to 0xffff91c6713b00f8 of 8 bytes by task 49268 on cpu 127:
ext4_write_end+0x4e3/0x750 [ext4]
ext4_update_i_disksize at fs/ext4/ext4.h:3032
(inlined by) ext4_update_inode_size at fs/ext4/ext4.h:3046
(inlined by) ext4_write_end at fs/ext4/inode.c:1287
generic_perform_write+0x208/0x2a0
ext4_buffered_write_iter+0x11f/0x210 [ext4]
ext4_file_write_iter+0xce/0x9e0 [ext4]
new_sync_write+0x29c/0x3b0
__vfs_write+0x92/0xa0
vfs_write+0x103/0x260
ksys_write+0x9d/0x130
__x64_sys_write+0x4c/0x60
do_syscall_64+0x91/0xb47
entry_SYSCALL_64_after_hwframe+0x49/0xbe
read to 0xffff91c6713b00f8 of 8 bytes by task 24872 on cpu 37:
ext4_writepages+0x10ac/0x1d00 [ext4]
mpage_map_and_submit_extent at fs/ext4/inode.c:2468
(inlined by) ext4_writepages at fs/ext4/inode.c:2772
do_writepages+0x5e/0x130
__writeback_single_inode+0xeb/0xb20
writeback_sb_inodes+0x429/0x900
__writeback_inodes_wb+0xc4/0x150
wb_writeback+0x4bd/0x870
wb_workfn+0x6b4/0x960
process_one_work+0x54c/0xbe0
worker_thread+0x80/0x650
kthread+0x1e0/0x200
ret_from_fork+0x27/0x50
Reported by Kernel Concurrency Sanitizer on:
CPU: 37 PID: 24872 Comm: kworker/u261:2 Tainted: G W O L 5.5.0-next-20200204+ #5
Hardware name: HPE ProLiant DL385 Gen10/ProLiant DL385 Gen10, BIOS A40 07/10/2019
Workqueue: writeback wb_workfn (flush-7:0)
Since only the read is operating as lockless (outside of the
"i_data_sem"), load tearing could introduce a logic bug. Fix it by
adding READ_ONCE() for the read and WRITE_ONCE() for the write.
While certain modeset operations on gv100+ need us to temporarily
disable the LUT, we make the mistake of sometimes neglecting to
reprogram the LUT after such modesets. In particular, moving a head from
one encoder to another seems to trigger this quite often. GV100+ is very
picky about having a LUT in most scenarios, so this causes the display
engine to hang with the following error code:
Walter Wu has reported a potential case in which init_stack_slab() is
called after stack_slabs[STACK_ALLOC_MAX_SLABS - 1] has already been
initialized. In that case init_stack_slab() will overwrite
stack_slabs[STACK_ALLOC_MAX_SLABS], which may result in a memory
corruption.
Link: http://lkml.kernel.org/r/20200218102950.260263-1-glider@google.com Fixes: cd11016e5f521 ("mm, kasan: stackdepot implementation. Enable stackdepot for SLAB") Signed-off-by: Alexander Potapenko <glider@google.com> Reported-by: Walter Wu <walter-zh.wu@mediatek.com> Cc: Dmitry Vyukov <dvyukov@google.com> Cc: Matthias Brugger <matthias.bgg@gmail.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Josh Poimboeuf <jpoimboe@redhat.com> Cc: Kate Stewart <kstewart@linuxfoundation.org> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> 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: Sasha Levin <sashal@kernel.org>
The driver only supports FIFO mode so setting and checking this variable
is unnecessary. If DMA support is ever added then such checks can be
introduced.
Signed-off-by: Ryan Case <ryandcase@chromium.org> Reviewed-by: Evan Green <evgreen@chromium.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The variables of tx_wm and rx_wm were set to the same define value in
all cases, never updated, and the define was sometimes used
interchangably. Remove the variables/function and use the fixed value.
Signed-off-by: Ryan Case <ryandcase@chromium.org> Reviewed-by: Evan Green <evgreen@chromium.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
A frequent side comment has been to remove the use of writel_relaxed,
readl_relaxed, and mb. This reduces driver complexity and the _relaxed
variants were not known to provide any noticeable performance benefit.
Signed-off-by: Ryan Case <ryandcase@chromium.org> Reviewed-by: Evan Green <evgreen@chromium.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Disable M_TX_FIFO_WATERMARK_EN after we've sent all data for a given
transaction so we don't continue to receive a flurry of free space
interrupts while waiting for the M_CMD_DONE notification. Re-enable the
watermark when establishing the next transaction.
Also clear the watermark interrupt after filling the FIFO so we do not
receive notification again prior to actually having free space.
Signed-off-by: Ryan Case <ryandcase@chromium.org> Reviewed-by: Douglas Anderson <dianders@chromium.org> Tested-by: Douglas Anderson <dianders@chromium.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
If a serial console write occured while a UART transmit command was
waiting for a done signal then no further data would be sent until
something new kicked the system into gear. If there is already data
waiting in the circular buffer we must re-enable the tx watermark so we
receive the expected interrupts.
Signed-off-by: Ryan Case <ryandcase@chromium.org> Reviewed-by: Evan Green <evgreen@chromium.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Commit 13db77347db1 ("KVM: x86: don't notify userspace IOAPIC on edge
EOI") said, edge-triggered interrupts don't set a bit in TMR, which means
that IOAPIC isn't notified on EOI. And var level indicates level-triggered
interrupt.
But commit 3159d36ad799 ("KVM: x86: use generic function for MSI parsing")
replace var level with irq.level by mistake. Fix it by changing irq.level
to irq.trig_mode.
Cc: stable@vger.kernel.org Fixes: 3159d36ad799 ("KVM: x86: use generic function for MSI parsing") Signed-off-by: Miaohe Lin <linmiaohe@huawei.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
vmx_check_intercept is not yet fully implemented. To avoid emulating
instructions disallowed by the L1 hypervisor, refuse to emulate
instructions by default.
Commit 68600f623d69 ("mm: don't miss the last page because of round-off
error") makes the scan size round up to @denominator regardless of the
memory cgroup's state, online or offline. This affects the overall
reclaiming behavior: the corresponding LRU list is eligible for
reclaiming only when its size logically right shifted by @sc->priority
is bigger than zero in the former formula.
For example, the inactive anonymous LRU list should have at least 0x4000
pages to be eligible for reclaiming when we have 60/12 for
swappiness/priority and without taking scan/rotation ratio into account.
After the roundup is applied, the inactive anonymous LRU list becomes
eligible for reclaiming when its size is bigger than or equal to 0x1000
in the same condition.
aarch64 has 512MB huge page size when the base page size is 64KB. The
memory cgroup that has a huge page is always eligible for reclaiming in
that case.
The reclaiming is likely to stop after the huge page is reclaimed,
meaing the further iteration on @sc->priority and the silbing and child
memory cgroups will be skipped. The overall behaviour has been changed.
This fixes the issue by applying the roundup to offlined memory cgroups
only, to give more preference to reclaim memory from offlined memory
cgroup. It sounds reasonable as those memory is unlikedly to be used by
anyone.
The issue was found by starting up 8 VMs on a Ampere Mustang machine,
which has 8 CPUs and 16 GB memory. Each VM is given with 2 vCPUs and
2GB memory. It took 264 seconds for all VMs to be completely up and
784MB swap is consumed after that. With this patch applied, it took 236
seconds and 60MB swap to do same thing. So there is 10% performance
improvement for my case. Note that KSM is disable while THP is enabled
in the testing.
total used free shared buff/cache available
Mem: 16196 10065 2049 16 4081 3749
Swap: 8175 784 7391
total used free shared buff/cache available
Mem: 16196 11324 3656 24 1215 2936
Swap: 8175 60 8115
Link: http://lkml.kernel.org/r/20200211024514.8730-1-gshan@redhat.com Fixes: 68600f623d69 ("mm: don't miss the last page because of round-off error") Signed-off-by: Gavin Shan <gshan@redhat.com> Acked-by: Roman Gushchin <guro@fb.com> Cc: <stable@vger.kernel.org> [4.20+] 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>
This was noticed when printing debugfs for MSIs on my ARM64 server. The
new dstate IRQD_MSI_NOMASK_QUIRK came out surprisingly while it should only
be the x86 stuff for the time being...
The new MSI quirk flag uses the same bit as IRQ_DOMAIN_NAME_ALLOCATED which
is oddly defined as bit 6 for no good reason.
nvme_mpath_init() is called by nvme_init_identify() which is called in
multiple places (nvme_reset_work(), nvme_passthru_end(), etc). This
means nvme_mpath_init() may be called multiple times before
nvme_mpath_uninit() (which is only called on nvme_free_ctrl()).
When nvme_mpath_init() is called multiple times, it overwrites the
ana_log_buf pointer with a new allocation, thus leaking the previous
allocation.
To fix this, free ana_log_buf before allocating a new one.
Fixes: 0d0b660f214dc490 ("nvme: add ANA support") Cc: <stable@vger.kernel.org> Reviewed-by: Sagi Grimberg <sagi@grimberg.me> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Logan Gunthorpe <logang@deltatee.com> Signed-off-by: Keith Busch <kbusch@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit a97955844807 ("ipc,sem: remove uneeded sem_undo_list lock usage
in exit_sem()") removes a lock that is needed. This leads to a process
looping infinitely in exit_sem() and can also lead to a crash. There is
a reproducer available in [1] and with the commit reverted the issue
does not reproduce anymore.
Using the reproducer found in [1] is fairly easy to reach a point where
one of the child processes is looping infinitely in exit_sem between
for(;;) and if (semid == -1) block, while it's trying to free its last
sem_undo structure which has already been freed by freeary().
Each sem_undo struct is on two lists: one per semaphore set (list_id)
and one per process (list_proc). The list_id list tracks undos by
semaphore set, and the list_proc by process.
Undo structures are removed either by freeary() or by exit_sem(). The
freeary function is invoked when the user invokes a syscall to remove a
semaphore set. During this operation freeary() traverses the list_id
associated with the semaphore set and removes the undo structures from
both the list_id and list_proc lists.
For this case, exit_sem() is called at process exit. Each process
contains a struct sem_undo_list (referred to as "ulp") which contains
the head for the list_proc list. When the process exits, exit_sem()
traverses this list to remove each sem_undo struct. As in freeary(),
whenever a sem_undo struct is removed from list_proc, it is also removed
from the list_id list.
Removing elements from list_id is safe for both exit_sem() and freeary()
due to sem_lock(). Removing elements from list_proc is not safe;
freeary() locks &un->ulp->lock when it performs
list_del_rcu(&un->list_proc) but exit_sem() does not (locking was
removed by commit a97955844807 ("ipc,sem: remove uneeded sem_undo_list
lock usage in exit_sem()").
This can result in the following situation while executing the
reproducer [1] : Consider a child process in exit_sem() and the parent
in freeary() (because of semctl(sid[i], NSEM, IPC_RMID)).
- The list_proc for the child contains the last two undo structs A and
B (the rest have been removed either by exit_sem() or freeary()).
- The semid for A is 1 and semid for B is 2.
- exit_sem() removes A and at the same time freeary() removes B.
- Since A and B have different semid sem_lock() will acquire different
locks for each process and both can proceed.
The bug is that they remove A and B from the same list_proc at the same
time because only freeary() acquires the ulp lock. When exit_sem()
removes A it makes ulp->list_proc.next to point at B and at the same
time freeary() removes B setting B->semid=-1.
At the next iteration of for(;;) loop exit_sem() will try to remove B.
The only way to break from for(;;) is for (&un->list_proc ==
&ulp->list_proc) to be true which is not. Then exit_sem() will check if
B->semid=-1 which is and will continue looping in for(;;) until the
memory for B is reallocated and the value at B->semid is changed.
At that point, exit_sem() will crash attempting to unlink B from the
lists (this can be easily triggered by running the reproducer [1] a
second time).
To prove this scenario instrumentation was added to keep information
about each sem_undo (un) struct that is removed per process and per
semaphore set (sma).
CPU0 CPU1
[caller holds sem_lock(sma for A)] ...
freeary() exit_sem()
... ...
... sem_lock(sma for B)
spin_lock(A->ulp->lock) ...
list_del_rcu(un_A->list_proc) list_del_rcu(un_B->list_proc)
Undo structures A and B have different semid and sem_lock() operations
proceed. However they belong to the same list_proc list and they are
removed at the same time. This results into ulp->list_proc.next
pointing to the address of B which is already removed.
After reverting commit a97955844807 ("ipc,sem: remove uneeded
sem_undo_list lock usage in exit_sem()") the issue was no longer
reproducible.
The serdev tty-port controller driver should reset the tty-port client
operations also on deregistration to avoid a NULL-pointer dereference in
case the port is later re-registered as a normal tty device.
Note that this can only happen with tty drivers such as 8250 which have
statically allocated port structures that can end up being reused and
where a later registration would not register a serdev controller (e.g.
due to registration errors or if the devicetree has been changed in
between).
Specifically, this can be an issue for any statically defined ports that
would be registered by 8250 core when an 8250 driver is being unbound.
Fixes: bed35c6dfa6a ("serdev: add a tty port controller driver") Cc: stable <stable@vger.kernel.org> # 4.11 Reported-by: Loic Poulain <loic.poulain@linaro.org> Signed-off-by: Johan Hovold <johan@kernel.org> Link: https://lore.kernel.org/r/20200210145730.22762-1-johan@kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There has oops as below happen on i.MX8MP EVK platform that has
6G bytes DDR memory.
when (xmit->tail < xmit->head) && (xmit->head == 0),
it setups one sg entry with sg->length is zero:
sg_set_buf(sgl + 1, xmit->buf, xmit->head);
if xmit->buf is allocated from >4G address space, and SDMA only
support <4G address space, then dma_map_sg() will call swiotlb_map()
to do bounce buffer copying and mapping.
But swiotlb_map() don't allow sg entry's length is zero, otherwise
report BUG_ON().
So the patch is to correct the tx DMA scatter list.
In atmel_shutdown() we call atmel_stop_rx() and atmel_stop_tx() functions.
Prevent the rx restart that is implemented in RS485 or ISO7816 modes when
calling atmel_stop_tx() by using the atomic information tasklet_shutdown
that is already in place for this purpose.
The commit 54e53b2e8081
("tty: serial: 8250: pass IRQ shared flag to UART ports")
nicely explained the problem:
---8<---8<---
On some systems IRQ lines between multiple UARTs might be shared. If so, the
irqflags have to be configured accordingly. The reason is: The 8250 port startup
code performs IRQ tests *before* the IRQ handler for that particular port is
registered. This is performed in serial8250_do_startup(). This function checks
whether IRQF_SHARED is configured and only then disables the IRQ line while
testing.
This test is performed upon each open() of the UART device. Imagine two UARTs
share the same IRQ line: On is already opened and the IRQ is active. When the
second UART is opened, the IRQ line has to be disabled while performing IRQ
tests. Otherwise an IRQ might handler might be invoked, but the IRQ itself
cannot be handled, because the corresponding handler isn't registered,
yet. That's because the 8250 code uses a chain-handler and invokes the
corresponding port's IRQ handling routines himself.
Unfortunately this IRQF_SHARED flag isn't configured for UARTs probed via device
tree even if the IRQs are shared. This way, the actual and shared IRQ line isn't
disabled while performing tests and the kernel correctly detects a spurious
IRQ. So, adding this flag to the DT probe solves the issue.
Note: The UPF_SHARE_IRQ flag is configured unconditionally. Therefore, the
IRQF_SHARED flag can be set unconditionally as well.
Example stack trace by performing `echo 1 > /dev/ttyS2` on a non-patched system:
But unfortunately didn't fix the root cause. Let's try again here by moving
IRQ flag assignment from serial_link_irq_chain() to serial8250_do_startup().
This should fix the similar issue reported for 8250_pnp case.
Since this change we don't need to have custom solutions in 8250_aspeed_vuart
and 8250_of drivers, thus, drop them.
Fixes: 1c2f04937b3e ("serial: 8250: add IRQ trigger support") Reported-by: Li RongQing <lirongqing@baidu.com> Cc: Kurt Kanzenbach <kurt@linutronix.de> Cc: Vikram Pandita <vikram.pandita@ti.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Cc: stable <stable@vger.kernel.org> Acked-by: Kurt Kanzenbach <kurt@linutronix.de> Link: https://lore.kernel.org/r/20200211135559.85960-1-andriy.shevchenko@linux.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
added support for access to the free-running counter via 'perf -e
msr/irperf/', but when exercised, it always returns a 0 count:
BEFORE:
$ perf stat -e instructions,msr/irperf/ true
Performance counter stats for 'true':
624,833 instructions
0 msr/irperf/
Simply set its enable bit - HWCR bit 30 - to make it start counting.
Enablement is restricted to all machines advertising IRPERF capability,
except those susceptible to an erratum that makes the IRPERF return
bad values.
That erratum occurs in Family 17h models 00-1fh [1], but not in F17h
models 20h and above [2].
AFTER (on a family 17h model 31h machine):
$ perf stat -e instructions,msr/irperf/ true
Performance counter stats for 'true':
621,690 instructions
622,490 msr/irperf/
[1] Revision Guide for AMD Family 17h Models 00h-0Fh Processors
[2] Revision Guide for AMD Family 17h Models 30h-3Fh Processors
The revision guides are available from the bugzilla Link below.
Accessing the MCA thresholding controls in sysfs concurrently with CPU
hotplug can lead to a couple of KASAN-reported issues:
BUG: KASAN: use-after-free in sysfs_file_ops+0x155/0x180
Read of size 8 at addr ffff888367578940 by task grep/4019
and
BUG: KASAN: use-after-free in show_error_count+0x15c/0x180
Read of size 2 at addr ffff888368a05514 by task grep/4454
for example. Both result from the fact that the threshold block
creation/teardown code frees the descriptor memory itself instead of
defining proper ->release function and leaving it to the driver core to
take care of that, after all sysfs accesses have completed.
Do that and get rid of the custom freeing code, fixing the above UAFs in
the process.
[ bp: write commit message. ]
Fixes: 95268664390b ("[PATCH] x86_64: mce_amd support for family 0x10 processors") Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Borislav Petkov <bp@suse.de> Cc: <stable@vger.kernel.org> Link: https://lkml.kernel.org/r/20200214082801.13836-1-bp@alien8.de Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
threshold_create_bank() creates a bank descriptor per MCA error
thresholding counter which can be controlled over sysfs. It publishes
the pointer to that bank in a per-CPU variable and then goes on to
create additional thresholding blocks if the bank has such.
However, that creation of additional blocks in
allocate_threshold_blocks() can fail, leading to a use-after-free
through the per-CPU pointer.
Therefore, publish that pointer only after all blocks have been setup
successfully.
Fixes: 019f34fccfd5 ("x86, MCE, AMD: Move shared bank to node descriptor") Reported-by: Saar Amar <Saar.Amar@microsoft.com> Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Borislav Petkov <bp@suse.de> Cc: <stable@vger.kernel.org> Link: http://lkml.kernel.org/r/20200128140846.phctkvx5btiexvbx@kili.mountain Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
I found a NULL pointer dereference in ocfs2_block_group_clear_bits().
The running environment:
kernel version: 4.19
A cluster with two nodes, 5 luns mounted on two nodes, and do some
file operations like dd/fallocate/truncate/rm on every lun with storage
network disconnection.
The fallocate operation on dm-23-45 caused an null pointer dereference.
My analysis process as follows:
ocfs2_fallocate
__ocfs2_change_file_space
ocfs2_allocate_extents
ocfs2_extend_allocation
ocfs2_add_inode_data
ocfs2_add_clusters_in_btree
ocfs2_insert_extent
ocfs2_do_insert_extent
ocfs2_rotate_tree_right
ocfs2_extend_rotate_transaction
ocfs2_extend_trans
jbd2_journal_restart
jbd2__journal_restart
/* handle->h_transaction is NULL,
* is_handle_aborted(handle) is true
*/
handle->h_transaction = NULL;
start_this_handle
return -EROFS;
ocfs2_free_clusters
_ocfs2_free_clusters
_ocfs2_free_suballoc_bits
ocfs2_block_group_clear_bits
ocfs2_journal_access_gd
__ocfs2_journal_access
jbd2_journal_get_undo_access
/* I think jbd2_write_access_granted() will
* return true, because do_get_write_access()
* will return -EROFS.
*/
if (jbd2_write_access_granted(...)) return 0;
do_get_write_access
/* handle->h_transaction is NULL, it will
* return -EROFS here, so do_get_write_access()
* was not called.
*/
if (is_handle_aborted(handle)) return -EROFS;
/* bh2jh(group_bh) is NULL, caused NULL
pointer dereference */
undo_bg = (struct ocfs2_group_desc *)
bh2jh(group_bh)->b_committed_data;
If handle->h_transaction == NULL, then jbd2_write_access_granted()
does not really guarantee that journal_head will stay around,
not even speaking of its b_committed_data. The bh2jh(group_bh)
can be removed after ocfs2_journal_access_gd() and before call
"bh2jh(group_bh)->b_committed_data". So, we should move
is_handle_aborted() check from do_get_write_access() into
jbd2_journal_get_undo_access() and jbd2_journal_get_write_access()
before the call to jbd2_write_access_granted().
Link: https://lore.kernel.org/r/f72a623f-b3f1-381a-d91d-d22a1c83a336@huawei.com Signed-off-by: Yan Wang <wangyan122@huawei.com> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Reviewed-by: Jun Piao <piaojun@huawei.com> Reviewed-by: Jan Kara <jack@suse.cz> Cc: stable@kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
After a treclaim, we expect to be in non-transactional state. If we
don't clear the current thread's MSR[TS] before we get preempted, then
tm_recheckpoint_new_task() will recheckpoint and we get rescheduled in
suspended transaction state.
When handling a signal caught in transactional state,
handle_rt_signal64() calls get_tm_stackpointer() that treclaims the
transaction using tm_reclaim_current() but without clearing the
thread's MSR[TS]. This can cause the TM Bad Thing exception below if
later we pagefault and get preempted trying to access the user's
sigframe, using __put_user(). Afterwards, when we are rescheduled back
into do_page_fault() (but now in suspended state since the thread's
MSR[TS] was not cleared), upon executing 'rfid' after completion of
the page fault handling, the exception is raised because a transition
from suspended to non-transactional state is invalid.
The simplified sequence of events that triggers the above exception is:
... # userspace in NON-TRANSACTIONAL state
tbegin # userspace in TRANSACTIONAL state
signal delivery # kernelspace in SUSPENDED state
handle_rt_signal64()
get_tm_stackpointer()
treclaim # kernelspace in NON-TRANSACTIONAL state
__put_user()
page fault happens. We will never get back here because of the TM Bad Thing exception.
page fault handling kicks in and we voluntarily preempt ourselves
do_page_fault()
__schedule()
__switch_to(other_task)
our task is rescheduled and we recheckpoint because the thread's MSR[TS] was not cleared
__switch_to(our_task)
switch_to_tm()
tm_recheckpoint_new_task()
trechkpt # kernelspace in SUSPENDED state
The page fault handling resumes, but now we are in suspended transaction state
do_page_fault() completes
rfid <----- trying to get back where the page fault happened (we were non-transactional back then)
TM Bad Thing # illegal transition from suspended to non-transactional
This patch fixes that issue by clearing the current thread's MSR[TS]
just after treclaim in get_tm_stackpointer() so that we stay in
non-transactional state in case we are preempted. In order to make
treclaim and clearing the thread's MSR[TS] atomic from a preemption
perspective when CONFIG_PREEMPT is set, preempt_disable/enable() is
used. It's also necessary to save the previous value of the thread's
MSR before get_tm_stackpointer() is called so that it can be exposed
to the signal handler later in setup_tm_sigcontexts() to inform the
userspace MSR at the moment of the signal delivery.
Found with tm-signal-context-force-tm kernel selftest.
Fixes: 2b0a576d15e0 ("powerpc: Add new transactional memory state to the signal context") Cc: stable@vger.kernel.org # v3.9 Signed-off-by: Gustavo Luiz Duarte <gustavold@linux.ibm.com> Acked-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/20200211033831.11165-1-gustavold@linux.ibm.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In routine wpa_supplicant_ioctl(), the user-controlled p->length is
checked to be at least the size of struct ieee_param size, but the code
does not detect the case where p->length is greater than the size
of the struct, thus a malicious user could be wasting kernel memory.
Fixes commit 554c0a3abf216 ("staging: Add rtl8723bs sdio wifi driver").
Reported by: Pietro Oliva <pietroliva@gmail.com> Cc: Pietro Oliva <pietroliva@gmail.com> Cc: Stable <stable@vger.kernel.org> Fixes: 554c0a3abf216 ("staging: Add rtl8723bs sdio wifi driver"). Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net> Link: https://lore.kernel.org/r/20200210180235.21691-5-Larry.Finger@lwfinger.net Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In routine rtw_hostapd_ioctl(), the user-controlled p->length is assumed
to be at least the size of struct ieee_param size, but this assumption is
never checked. This could result in out-of-bounds read/write on kernel
heap in case a p->length less than the size of struct ieee_param is
specified by the user. If p->length is allowed to be greater than the size
of the struct, then a malicious user could be wasting kernel memory.
Fixes commit 554c0a3abf216 ("0taging: Add rtl8723bs sdio wifi driver").
Reported by: Pietro Oliva <pietroliva@gmail.com> Cc: Pietro Oliva <pietroliva@gmail.com> Cc: Stable <stable@vger.kernel.org>
Fixes 554c0a3abf216 ("0taging: Add rtl8723bs sdio wifi driver"). Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net> Link: https://lore.kernel.org/r/20200210180235.21691-3-Larry.Finger@lwfinger.net Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In routine wpa_supplicant_ioctl(), the user-controlled p->length is
checked to be at least the size of struct ieee_param size, but the code
does not detect the case where p->length is greater than the size
of the struct, thus a malicious user could be wasting kernel memory.
Fixes commit a2c60d42d97c ("Add files for new driver - part 16").
Reported by: Pietro Oliva <pietroliva@gmail.com> Cc: Pietro Oliva <pietroliva@gmail.com> Cc: Stable <stable@vger.kernel.org>
Fixes commit a2c60d42d97c ("Add files for new driver - part 16"). Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net> Link: https://lore.kernel.org/r/20200210180235.21691-4-Larry.Finger@lwfinger.net Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In routine rtw_hostapd_ioctl(), the user-controlled p->length is assumed
to be at least the size of struct ieee_param size, but this assumption is
never checked. This could result in out-of-bounds read/write on kernel
heap in case a p->length less than the size of struct ieee_param is
specified by the user. If p->length is allowed to be greater than the size
of the struct, then a malicious user could be wasting kernel memory.
Fixes commit a2c60d42d97c ("Add files for new driver - part 16").
Reported by: Pietro Oliva <pietroliva@gmail.com> Cc: Pietro Oliva <pietroliva@gmail.com> Cc: Stable <stable@vger.kernel.org> Fixes: a2c60d42d97c ("staging: r8188eu: Add files for new driver - part 16") Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net> Link: https://lore.kernel.org/r/20200210180235.21691-2-Larry.Finger@lwfinger.net Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The current code in dwc3_gadget_ep_reclaim_completed_trb() will
check for IOC/LST bit in the event->status and returns if
IOC/LST bit is set. This logic doesn't work if multiple TRBs
are queued per request and the IOC/LST bit is set on the last
TRB of that request.
Consider an example where a queued request has multiple queued
TRBs and IOC/LST bit is set only for the last TRB. In this case,
the core generates XferComplete/XferInProgress events only for
the last TRB (since IOC/LST are set only for the last TRB). As
per the logic in dwc3_gadget_ep_reclaim_completed_trb()
event->status is checked for IOC/LST bit and returns on the
first TRB. This leaves the remaining TRBs left unhandled.
Similarly, if the gadget function enqueues an unaligned request
with sglist already in it, it should fail the same way, since we
will append another TRB to something that already uses more than
one TRB.
To aviod this, this patch changes the code to check for IOC/LST
bits in TRB->ctrl instead.
At a practical level, this patch resolves USB transfer stalls seen
with adb on dwc3 based HiKey960 after functionfs gadget added
scatter-gather support around v4.20.
Cc: Felipe Balbi <balbi@kernel.org> Cc: Yang Fei <fei.yang@intel.com> Cc: Thinh Nguyen <thinhn@synopsys.com> Cc: Tejas Joglekar <tejas.joglekar@synopsys.com> Cc: Andrzej Pietrasiewicz <andrzej.p@collabora.com> Cc: Jack Pham <jackp@codeaurora.org> Cc: Todd Kjos <tkjos@google.com> Cc: Greg KH <gregkh@linuxfoundation.org> Cc: Linux USB List <linux-usb@vger.kernel.org> Cc: stable <stable@vger.kernel.org> Tested-by: Tejas Joglekar <tejas.joglekar@synopsys.com> Reviewed-by: Thinh Nguyen <thinhn@synopsys.com> Signed-off-by: Anurag Kumar Vulisha <anurag.kumar.vulisha@xilinx.com>
[jstultz: forward ported to mainline, reworded commit log, reworked
to only check trb->ctrl as suggested by Felipe] Signed-off-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Felipe Balbi <balbi@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
SET/CLEAR_FEATURE for Remote Wakeup allowance not handled correctly.
GET_STATUS handling provided not correct data on DATA Stage.
Issue seen when gadget's dr_mode set to "otg" mode and connected
to MacOS.
Both are fixed and tested using USBCV Ch.9 tests.
Renesas R-Car H3ULCB + Kingfisher Infotainment Board is either not able
to detect the USB3.0 mass storage devices or is detecting those as
USB2.0 high speed devices.
The explanation given by Renesas is that, due to a HW issue, the XHCI
driver does not wake up after going to sleep on connecting a USB3.0
device.
In order to mitigate that, disable the auto-suspend feature
specifically for SMSC hubs from hub_probe() function, as a quirk.
Renesas Kingfisher Infotainment Board has two USB3.0 ports (CN2) which
are connected via USB5534B 4-port SuperSpeed/Hi-Speed, low-power,
configurable hub controller.
[1] SanDisk USB 3.0 device detected as USB-2.0 before the patch
[ 74.036390] usb 5-1.1: new high-speed USB device number 4 using xhci-hcd
[ 74.061598] usb 5-1.1: New USB device found, idVendor=0781, idProduct=5581, bcdDevice= 1.00
[ 74.069976] usb 5-1.1: New USB device strings: Mfr=1, Product=2, SerialNumber=3
[ 74.077303] usb 5-1.1: Product: Ultra
[ 74.080980] usb 5-1.1: Manufacturer: SanDisk
[ 74.085263] usb 5-1.1: SerialNumber: 4C530001110208116550
[2] SanDisk USB 3.0 device detected as USB-3.0 after the patch
[ 34.565078] usb 6-1.1: new SuperSpeed Gen 1 USB device number 3 using xhci-hcd
[ 34.588719] usb 6-1.1: New USB device found, idVendor=0781, idProduct=5581, bcdDevice= 1.00
[ 34.597098] usb 6-1.1: New USB device strings: Mfr=1, Product=2, SerialNumber=3
[ 34.604430] usb 6-1.1: Product: Ultra
[ 34.608110] usb 6-1.1: Manufacturer: SanDisk
[ 34.612397] usb 6-1.1: SerialNumber: 4C530001110208116550
Paul Zimmerman reports that his USB Bluetooth adapter sometimes
crashes following system resume, when it receives a
Get-Device-Descriptor request while it is busy doing something else.
Such a request was added by commit a4f55d8b8c14 ("usb: hub: Check
device descriptor before resusciation"). It gets sent when the hub
driver's work thread checks whether a connect-change event on an
enabled port really indicates a new device has been connected, as
opposed to an old device momentarily disconnecting and then
reconnecting (which can happen with xHCI host controllers, since they
automatically enable connected ports).
The same kind of thing occurs when a port's power session is lost
during system suspend. When the system wakes up it sees a
connect-change event on the port, and if the child device's
persist_enabled flag was set then hub_activate() sets the device's
reset_resume flag as well as the port's bit in hub->change_bits. The
reset-resume code then takes responsibility for checking that the same
device is still attached to the port, and it does this as part of the
device's resume pathway. By the time the hub driver's work thread
starts up again, the device has already been fully reinitialized and
is busy doing its own thing. There's no need for the work thread to
do the same check a second time, and in fact this unnecessary check is
what caused the problem that Paul observed.
Note that performing the unnecessary check is not actually a bug.
Devices are supposed to be able to send descriptors back to the host
even when they are busy doing something else. The underlying cause of
Paul's problem lies in his Bluetooth adapter. Nevertheless, we
shouldn't perform the same check twice in a row -- and as a nice side
benefit, removing the extra check allows the Bluetooth adapter to work
more reliably.
The work thread performs its check when it sees that the port's bit is
set in hub->change_bits. In this situation that bit is interpreted as
though a connect-change event had occurred on the port _after_ the
reset-resume, which is not what actually happened.
One possible fix would be to make the reset-resume code clear the
port's bit in hub->change_bits. But it seems simpler to just avoid
setting the bit during hub_activate() in the first place. That's what
this patch does.
(Proving that the patch is correct when CONFIG_PM is disabled requires
a little thought. In that setting hub_activate() will be called only
for initialization and resets, since there won't be any resumes or
reset-resumes. During initialization and hub resets the hub doesn't
have any child devices, and so this code path never gets executed.)
When a uas disk is plugged into an external hub, uas_probe()
will be called by the hub thread to do the probe. It will
first create a SCSI host and then do the scan for this host.
During the scan, it will probe the LUN using SCSI INQUERY command
which will be packed in the URB and submitted to uas disk.
There might be a chance that this external hub with uas disk
attached is unplugged during the scan. In this case, uas driver
will fail to submit the URB (due to the NOTATTACHED state of uas
device) and try to put this SCSI command back to request queue
waiting for next chance to run.
In normal case, this cycle will terminate when hub thread gets
disconnection event and calls into uas_disconnect() accordingly.
But in this case, uas_disconnect() will not be called because
hub thread of external hub gets stuck waiting for the completion
of this SCSI command. A deadlock happened.
In this fix, uas will call scsi_scan_host() asynchronously to
avoid the blocking of hub thread.
This device has a broken vendor-specific altsetting for interface 1,
where endpoint 0x85 is declared as an isochronous endpoint despite being
used by interface 2 for audio capture.
Since commit 3e4f8e21c4f2 ("USB: core: fix check for duplicate
endpoints") USB core ignores any duplicate endpoints found during
descriptor parsing, but in this case we need to ignore the first
instance in order to avoid breaking the audio capture interface.
Add a new device quirk that can be used to blacklist endpoints.
Since commit 3e4f8e21c4f2 ("USB: core: fix check for duplicate
endpoints") USB core ignores any duplicate endpoints found during
descriptor parsing.
In order to handle devices where the first interfaces with duplicate
endpoints are the ones that should have their endpoints ignored, we need
to add a blacklist.
On some situations, the software handles TRB events slower
than adding TRBs, then xhci_handle_event can't return zero
long time, the xHC will consider the event ring is full,
and trigger "Event Ring Full" error, but in fact, the software
has already finished lots of events, just no chance to
update ERDP (event ring dequeue pointer).
In this commit, we force update ERDP if half of TRBS_PER_SEGMENT
events have handled to avoid "Event Ring Full" error.
xhci driver assumed that xHC controllers have at most one custom
supported speed table (PSI) for all usb 3.x ports.
Memory was allocated for one PSI table under the xhci hub structure.
Turns out this is not the case, some controllers have a separate
"supported protocol capability" entry with a PSI table for each port.
This means each usb3 roothub port can in theory support different custom
speeds.
To solve this, cache all supported protocol capabilities with their PSI
tables in an array, and add pointers to the xhci port structure so that
every port points to its capability entry in the array.
When creating the SuperSpeedPlus USB Device Capability BOS descriptor
for the xhci USB 3.1 roothub we for now will use only data from the
first USB 3.1 capable protocol capability entry in the array.
This could be improved later, this patch focuses resolving
the memory leak.
Reported-by: Paul Menzel <pmenzel@molgen.mpg.de> Reported-by: Sajja Venkateswara Rao <VenkateswaraRao.Sajja@amd.com> Fixes: 47189098f8be ("xhci: parse xhci protocol speed ID list for usb 3.1 usage") Cc: stable <stable@vger.kernel.org> # v4.4+ Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Link: https://lore.kernel.org/r/20200211150158.14475-1-mathias.nyman@linux.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Intel hosts that need the XHCI_PME_STUCK_QUIRK flag should enable
runtime pm by calling xhci_pme_acpi_rtd3_enable() before
usb_hcd_pci_probe() calls pci_dev_run_wake().
Otherwise usage count for the device won't be decreased, and runtime
suspend is prevented.
usb_hcd_pci_probe() only decreases the usage count if device can
generate run-time wake-up events, i.e. when pci_dev_run_wake()
returns true.
This issue was exposed by pci_dev_run_wake() change in
commit 8feaec33b986 ("PCI / PM: Always check PME wakeup capability for
runtime wakeup support")
and should be backported to kernels with that change
A Full-speed bulk USB audio device (DJ-Tech CTRL) with a invalid Maximum
Packet Size of 4 causes a xHC "Parameter Error" at enumeration.
This is because valid Maximum packet sizes for Full-speed bulk endpoints
are 8, 16, 32 and 64 bytes. Hosts are not required to support other values
than these. See usb 2 specs section 5.8.3 for details.
The device starts working after forcing the maximum packet size to 8.
This is most likely the case with other devices as well, so force the
maximum packet size to a valid range.
When ashmem file is mmapped, the resulting vma->vm_file points to the
backing shmem file with the generic fops that do not check ashmem
permissions like fops of ashmem do. If an mremap is done on the ashmem
region, then the permission checks will be skipped. Fix that by disallowing
mapping operation on the backing shmem file.
When pasting a selection to a vt, the task is set as INTERRUPTIBLE while
waiting for a tty to unthrottle. But signals are not handled at all.
Normally, this is not a problem as tty_ldisc_receive_buf receives all
the goods and a user has no reason to interrupt the task.
There are two scenarios where this matters:
1) when the tty is throttled and a signal is sent to the process, it
spins on a CPU until the tty is unthrottled. schedule() does not
really echedule, but returns immediately, of course.
2) when the sel_buffer becomes invalid, KASAN prevents any reads from it
and the loop simply does not proceed and spins forever (causing the
tty to throttle, but the code never sleeps, the same as above). This
sometimes happens as there is a race in the sel_buffer handling code.
So add signal handling to this ioctl (TIOCL_PASTESEL) and return -EINTR
in case a signal is pending.
Commit a6dbe4427559 ("vt: perform safe console erase in the right
order") provided fixes to an earlier commit by gathering all console
scrollback flushing operations in a function of its own. This includes
the invocation of vc_sw->con_switch() as previously done through a
update_screen() call. That commit failed to carry over the
con_is_visible() conditional though, as well as cursor handling, which
caused problems when "\e[3J" was written to a background console.
One could argue for preserving the call to update_screen(). However
this does far more than we need, and it is best to remove scrollback
assumptions from it. Instead let's gather the minimum needed to actually
perform scrollback flushing properly in that one place.
While at it, let's document the vc_sw->con_switch() side effect being
relied upon.
Jordy Zomer reported a KASAN out-of-bounds read in the floppy driver in
wait_til_ready().
Which on the face of it can't happen, since as Willy Tarreau points out,
the function does no particular memory access. Except through the FDCS
macro, which just indexes a static allocation through teh current fdc,
which is always checked against N_FDC.
Except the checking happens after we've already assigned the value.
The floppy driver is a disgrace (a lot of it going back to my original
horrd "design"), and has no real maintainer. Nobody has the hardware,
and nobody really cares. But it still gets used in virtual environment
because it's one of those things that everybody supports.
The whole thing should be re-written, or at least parts of it should be
seriously cleaned up. The 'current fdc' index, which is used by the
FDCS macro, and which is often shadowed by a local 'fdc' variable, is a
prime example of how not to write code.
But because nobody has the hardware or the motivation, let's just fix up
the immediate problem with a nasty band-aid: test the fdc index before
actually assigning it to the static 'fdc' variable.
The driver does not populate .reg_read callback for the non-active NVMem
because the file is supposed to be write-only. However, it turns out
NVMem subsystem does not yet support this and expects that the .reg_read
callback is provided. If user reads the binary attribute it triggers
NULL pointer dereference like this one:
In ecryptfs_init_messaging(), if the allocation for 'ecryptfs_msg_ctx_arr'
fails, the previously allocated 'ecryptfs_daemon_hash' is not deallocated,
leading to a memory leak bug. To fix this issue, free
'ecryptfs_daemon_hash' before returning the error.
Cc: stable@vger.kernel.org Fixes: 88b4a07e6610 ("[PATCH] eCryptfs: Public key transport mechanism") Signed-off-by: Wenwen Wang <wenwen@cs.uga.edu> Signed-off-by: Tyler Hicks <tyhicks@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In parse_tag_1_packet(), if tag 1 packet contains a key larger than
ECRYPTFS_MAX_ENCRYPTED_KEY_BYTES, no cleanup is executed, leading to a
memory leak on the allocated 'auth_tok_list_item'. To fix this issue, go to
the label 'out_free' to perform the cleanup work.
Some code in HD-audio driver calls snprintf() in a loop and still
expects that the return value were actually written size, while
snprintf() returns the expected would-be length instead. When the
given buffer limit were small, this leads to a buffer overflow.
Use scnprintf() for addressing those issues. It returns the actually
written size unlike snprintf().
Currently, the implementation of qcom_iommu_domain_free() is guaranteed
to do one of two things: WARN() and leak everything, or dereference NULL
and crash. That alone is terrible, but in fact the whole idea of trying
to track the liveness of a domain via the qcom_domain->iommu pointer as
a sanity check is full of fundamentally flawed assumptions. Make things
robust and actually functional by not trying to be quite so clever.
Reported-by: Brian Masney <masneyb@onstation.org> Tested-by: Brian Masney <masneyb@onstation.org> Reported-by: Naresh Kamboju <naresh.kamboju@linaro.org> Fixes: 0ae349a0f33f ("iommu/qcom: Add qcom_iommu") Signed-off-by: Robin Murphy <robin.murphy@arm.com> Tested-by: Stephan Gerhold <stephan@gerhold.net> Cc: stable@vger.kernel.org # v4.14+ Signed-off-by: Joerg Roedel <jroedel@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In case devlink_dpipe_entry_ctx_prepare() failed, release RTNL that was
previously taken and free the memory allocated by
mlxsw_sp_erif_entry_prepare().
Fixes: 2ba5999f009d ("mlxsw: spectrum: Add Support for erif table entries access") Signed-off-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>