| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
If an incoming FUSE request can't fit on the virtqueue, the request is
placed onto a workqueue so a worker can try to resubmit it later where
there will (hopefully) be space for it next time.
This is fine for requests that aren't larger than a virtqueue's maximum
capacity. However, if a request's size exceeds the maximum capacity of the
virtqueue (even if the virtqueue is empty), it will be doomed to a life of
being placed on the workqueue, removed, discovered it won't fit, and placed
on the workqueue yet again.
Furthermore, from section 2.6.5.3.1 (Driver Requirements: Indirect
Descriptors) of the virtio spec:
"A driver MUST NOT create a descriptor chain longer than the Queue
Size of the device."
To fix this, limit the number of pages FUSE will use for an overall
request. This way, each request can realistically fit on the virtqueue
when it is decomposed into a scattergather list and avoid violating section
2.6.5.3.1 of the virtio spec.
Signed-off-by: Connor Kuehl <ckuehl@redhat.com>
Reviewed-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
|
|
When accidentally passing twice the same tag to qemu, kmemleak ended up
reporting a memory leak in virtiofs. Also, looking at the log I saw the
following error (that's when I realised the duplicated tag):
virtiofs: probe of virtio5 failed with error -17
Here's the kmemleak log for reference:
unreferenced object 0xffff888103d47800 (size 1024):
comm "systemd-udevd", pid 118, jiffies 4294893780 (age 18.340s)
hex dump (first 32 bytes):
00 00 00 00 ad 4e ad de ff ff ff ff 00 00 00 00 .....N..........
ff ff ff ff ff ff ff ff 80 90 02 a0 ff ff ff ff ................
backtrace:
[<000000000ebb87c1>] virtio_fs_probe+0x171/0x7ae [virtiofs]
[<00000000f8aca419>] virtio_dev_probe+0x15f/0x210
[<000000004d6baf3c>] really_probe+0xea/0x430
[<00000000a6ceeac8>] device_driver_attach+0xa8/0xb0
[<00000000196f47a7>] __driver_attach+0x98/0x140
[<000000000b20601d>] bus_for_each_dev+0x7b/0xc0
[<00000000399c7b7f>] bus_add_driver+0x11b/0x1f0
[<0000000032b09ba7>] driver_register+0x8f/0xe0
[<00000000cdd55998>] 0xffffffffa002c013
[<000000000ea196a2>] do_one_initcall+0x64/0x2e0
[<0000000008f727ce>] do_init_module+0x5c/0x260
[<000000003cdedab6>] __do_sys_finit_module+0xb5/0x120
[<00000000ad2f48c6>] do_syscall_64+0x33/0x40
[<00000000809526b5>] entry_SYSCALL_64_after_hwframe+0x44/0xae
Cc: stable@vger.kernel.org
Signed-off-by: Luis Henriques <lhenriques@suse.de>
Fixes: a62a8ef9d97d ("virtio-fs: add virtiofs filesystem")
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
|
|
In fuse when a direct/write-through write happens we invalidate attrs
because that might have updated mtime/ctime on server and cached
mtime/ctime will be stale.
What about page writeback path. Looks like we don't invalidate attrs
there. To be consistent, invalidate attrs in writeback path as well. Only
exception is when writeback_cache is enabled. In that case we strust local
mtime/ctime and there is no need to invalidate attrs.
Recently users started experiencing failure of xfstests generic/080,
geneirc/215 and generic/614 on virtiofs. This happened only newer "stat"
utility and not older one. This patch fixes the issue.
So what's the root cause of the issue. Here is detailed explanation.
generic/080 test does mmap write to a file, closes the file and then checks
if mtime has been updated or not. When file is closed, it leads to
flushing of dirty pages (and that should update mtime/ctime on server).
But we did not explicitly invalidate attrs after writeback finished. Still
generic/080 passed so far and reason being that we invalidated atime in
fuse_readpages_end(). This is called in fuse_readahead() path and always
seems to trigger before mmaped write.
So after mmaped write when lstat() is called, it sees that atleast one of
the fields being asked for is invalid (atime) and that results in
generating GETATTR to server and mtime/ctime also get updated and test
passes.
But newer /usr/bin/stat seems to have moved to using statx() syscall now
(instead of using lstat()). And statx() allows it to query only ctime or
mtime (and not rest of the basic stat fields). That means when querying
for mtime, fuse_update_get_attr() sees that mtime is not invalid (only
atime is invalid). So it does not generate a new GETATTR and fill stat
with cached mtime/ctime. And that means updated mtime is not seen by
xfstest and tests start failing.
Invalidating attrs after writeback completion should solve this problem in
a generic manner.
Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
|
|
When posix access ACL is set, it can have an effect on file mode and it can
also need to clear SGID if.
- None of caller's group/supplementary groups match file owner group.
AND
- Caller is not priviliged (No CAP_FSETID).
As of now fuser server is responsible for changing the file mode as
well. But it does not know whether to clear SGID or not.
So add a flag FUSE_SETXATTR_ACL_KILL_SGID and send this info with SETXATTR
to let file server know that sgid needs to be cleared as well.
Reported-by: Luis Henriques <lhenriques@suse.de>
Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
|
|
Fuse client needs to send additional information to file server when it
calls SETXATTR(system.posix_acl_access), so add extra flags field to the
structure.
Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
|
|
With commit f8425c939663 ("fuse: 32-bit user space ioctl compat for fuse
device") the matching constraints for the FUSE_DEV_IOC_CLONE ioctl command
are relaxed, limited to the testing of command type and number. As Arnd
noticed, this is wrong as it wouldn't ensure the correctness of the data
size or direction for the received FUSE device ioctl.
Fix by bringing back the comparison of the ioctl received by the FUSE
device to the originally generated FUSE_DEV_IOC_CLONE.
Fixes: f8425c939663 ("fuse: 32-bit user space ioctl compat for fuse device")
Reported-by: Arnd Bergmann <arnd@kernel.org>
Signed-off-by: Alessio Balsini <balsini@android.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
|
|
s/reponsible/responsible/
Signed-off-by: Bhaskar Chowdhury <unixbhaskar@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
|
|
All callers of fuse_short_read already set the .page_zeroing flag, so no
need to do the tail zeroing again.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
|
|
'Maxmum' -> 'Maximum'
Signed-off-by: Connor Kuehl <ckuehl@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
|
|
There are two modes for write(2) and friends in fuse:
a) write through (update page cache, send sync WRITE request to userspace)
b) buffered write (update page cache, async writeout later)
The write through method kept all the page cache pages locked that were
used for the request. Keeping more than one page locked is deadlock prone
and Qian Cai demonstrated this with trinity fuzzing.
The reason for keeping the pages locked is that concurrent mapped reads
shouldn't try to pull possibly stale data into the page cache.
For full page writes, the easy way to fix this is to make the cached page
be the authoritative source by marking the page PG_uptodate immediately.
After this the page can be safely unlocked, since mapped/cached reads will
take the written data from the cache.
Concurrent mapped writes will now cause data in the original WRITE request
to be updated; this however doesn't cause any data inconsistency and this
scenario should be exceedingly rare anyway.
If the WRITE request returns with an error in the above case, currently the
page is not marked uptodate; this means that a concurrent read will always
read consistent data. After this patch the page is uptodate between
writing to the cache and receiving the error: there's window where a cached
read will read the wrong data. While theoretically this could be a
regression, it is unlikely to be one in practice, since this is normal for
buffered writes.
In case of a partial page write to an already uptodate page the locking is
also unnecessary, with the above caveats.
Partial write of a not uptodate page still needs to be handled. One way
would be to read the complete page before doing the write. This is not
possible, since it might break filesystems that don't expect any READ
requests when the file was opened O_WRONLY.
The other solution is to serialize the synchronous write with reads from
the partial pages. The easiest way to do this is to keep the partial pages
locked. The problem is that a write() may involve two such pages (one head
and one tail). This patch fixes it by only locking the partial tail page.
If there's a partial head page as well, then split that off as a separate
WRITE request.
Reported-by: Qian Cai <cai@lca.pw>
Link: https://lore.kernel.org/linux-fsdevel/4794a3fa3742a5e84fb0f934944204b55730829b.camel@lca.pw/
Fixes: ea9b9907b82a ("fuse: implement perform_write")
Cc: <stable@vger.kernel.org> # v2.6.26
Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
|
|
|
|
For each device, the nosy driver allocates a pcilynx structure.
A use-after-free might happen in the following scenario:
1. Open nosy device for the first time and call ioctl with command
NOSY_IOC_START, then a new client A will be malloced and added to
doubly linked list.
2. Open nosy device for the second time and call ioctl with command
NOSY_IOC_START, then a new client B will be malloced and added to
doubly linked list.
3. Call ioctl with command NOSY_IOC_START for client A, then client A
will be readded to the doubly linked list. Now the doubly linked
list is messed up.
4. Close the first nosy device and nosy_release will be called. In
nosy_release, client A will be unlinked and freed.
5. Close the second nosy device, and client A will be referenced,
resulting in UAF.
The root cause of this bug is that the element in the doubly linked list
is reentered into the list.
Fix this bug by adding a check before inserting a client. If a client
is already in the linked list, don't insert it.
The following KASAN report reveals it:
BUG: KASAN: use-after-free in nosy_release+0x1ea/0x210
Write of size 8 at addr ffff888102ad7360 by task poc
CPU: 3 PID: 337 Comm: poc Not tainted 5.12.0-rc5+ #6
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.12.0-59-gc9ba5276e321-prebuilt.qemu.org 04/01/2014
Call Trace:
nosy_release+0x1ea/0x210
__fput+0x1e2/0x840
task_work_run+0xe8/0x180
exit_to_user_mode_prepare+0x114/0x120
syscall_exit_to_user_mode+0x1d/0x40
entry_SYSCALL_64_after_hwframe+0x44/0xae
Allocated by task 337:
nosy_open+0x154/0x4d0
misc_open+0x2ec/0x410
chrdev_open+0x20d/0x5a0
do_dentry_open+0x40f/0xe80
path_openat+0x1cf9/0x37b0
do_filp_open+0x16d/0x390
do_sys_openat2+0x11d/0x360
__x64_sys_open+0xfd/0x1a0
do_syscall_64+0x33/0x40
entry_SYSCALL_64_after_hwframe+0x44/0xae
Freed by task 337:
kfree+0x8f/0x210
nosy_release+0x158/0x210
__fput+0x1e2/0x840
task_work_run+0xe8/0x180
exit_to_user_mode_prepare+0x114/0x120
syscall_exit_to_user_mode+0x1d/0x40
entry_SYSCALL_64_after_hwframe+0x44/0xae
The buggy address belongs to the object at ffff888102ad7300 which belongs to the cache kmalloc-128 of size 128
The buggy address is located 96 bytes inside of 128-byte region [ffff888102ad7300, ffff888102ad7380)
[ Modified to use 'list_empty()' inside proper lock - Linus ]
Link: https://lore.kernel.org/lkml/1617433116-5930-1-git-send-email-zheyuma97@gmail.com/
Reported-and-tested-by: 马哲宇 (Zheyu Ma) <zheyuma97@gmail.com>
Signed-off-by: Zheyu Ma <zheyuma97@gmail.com>
Cc: Greg Kroah-Hartman <greg@kroah.com>
Cc: Stefan Richter <stefanr@s5r6.in-berlin.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
The header file <linux/errno.h> is already included above and can be
removed here.
Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com>
Signed-off-by: Mateusz Holenko <mholenko@antmicro.com>
Signed-off-by: Stafford Horne <shorne@gmail.com>
|
|
kernel test robot correctly pinpoints a compilation failure if
CONFIG_BLOCK isn't set:
fs/io_uring.c: In function '__io_complete_rw':
>> fs/io_uring.c:2509:48: error: implicit declaration of function 'io_rw_should_reissue'; did you mean 'io_rw_reissue'? [-Werror=implicit-function-declaration]
2509 | if ((res == -EAGAIN || res == -EOPNOTSUPP) && io_rw_should_reissue(req)) {
| ^~~~~~~~~~~~~~~~~~~~
| io_rw_reissue
cc1: some warnings being treated as errors
Ensure that we have a stub declaration of io_rw_should_reissue() for
!CONFIG_BLOCK.
Fixes: 230d50d448ac ("io_uring: move reissue into regular IO path")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
The big top of the file comment talk about grand plans that never
happened, so remove them to not confuse the readers. Also mark the
devname and volname fields as ignored as they were never used by the
kernel.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
It's non-obvious how retry is done for block backed files, when it happens
off the kiocb done path. It also makes it tricky to deal with the iov_iter
handling.
Just mark the req as needing a reissue, and handling it from the
submission path instead. This makes it directly obvious that we're not
re-importing the iovec from userspace past the submit point, and it means
that we can just reuse our usual -EAGAIN retry path from the read/write
handling.
At some point in the future, we'll gain the ability to always reliably
return -EAGAIN through the stack. A previous attempt on the block side
didn't pan out and got reverted, hence the need to check for this
information out-of-band right now.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
If IOCB_NOWAIT is set on submission, then that needs to get propagated to
REQ_NOWAIT on the block side. Otherwise we completely lose this
information, and any issuer of IOCB_NOWAIT IO will potentially end up
blocking on eg request allocation on the storage side.
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
NUMA is useless when NOMMU, and it leads some build error,
make it depend on MMU.
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Palmer Dabbelt <palmerdabbelt@google.com>
|
|
Eliminate the following coccicheck warning:
./arch/riscv/mm/kasan_init.c:219:2-3: Unneeded semicolon
Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Signed-off-by: Yang Li <yang.lee@linux.alibaba.com>
Signed-off-by: Palmer Dabbelt <palmerdabbelt@google.com>
|
|
In RV64, the size of each entry in excp_vect_table is 8 bytes. If the
base of the table is not 8-byte aligned, loading an entry in the table
will raise a misaligned exception. Although such exception will be
handled by opensbi/bbl, this still causes performance degradation.
Signed-off-by: Zihao Yu <yuzihao@ict.ac.cn>
Reviewed-by: Anup Patel <anup@brainfault.org>
Signed-off-by: Palmer Dabbelt <palmerdabbelt@google.com>
|
|
The <asm/uaccess.h> header has a problem with put_user(a, ptr) if
the 'a' is not a simple variable, such as a function. This can lead
to the compiler producing code as so:
1: enable_user_access()
2: evaluate 'a' into register 'r'
3: put 'r' to 'ptr'
4: disable_user_acess()
The issue is that 'a' is now being evaluated with the user memory
protections disabled. So we try and force the evaulation by assigning
'x' to __val at the start, and hoping the compiler barriers in
enable_user_access() do the job of ordering step 2 before step 1.
This has shown up in a bug where 'a' sleeps and thus schedules out
and loses the SR_SUM flag. This isn't sufficient to fully fix, but
should reduce the window of opportunity. The first instance of this
we found is in scheudle_tail() where the code does:
$ less -N kernel/sched/core.c
4263 if (current->set_child_tid)
4264 put_user(task_pid_vnr(current), current->set_child_tid);
Here, the task_pid_vnr(current) is called within the block that has
enabled the user memory access. This can be made worse with KASAN
which makes task_pid_vnr() a rather large call with plenty of
opportunity to sleep.
Signed-off-by: Ben Dooks <ben.dooks@codethink.co.uk>
Reported-by: syzbot+e74b94fe601ab9552d69@syzkaller.appspotmail.com
Suggested-by: Arnd Bergman <arnd@arndb.de>
--
Changes since v1:
- fixed formatting and updated the patch description with more info
Changes since v2:
- fixed commenting on __put_user() (schwab@linux-m68k.org)
Change since v3:
- fixed RFC in patch title. Should be ready to merge.
Signed-off-by: Palmer Dabbelt <palmerdabbelt@google.com>
|
|
The const annotation should not be used for 'sp', or it will
become read only and lead to bad stack output.
Fixes: dec822771b01 ("riscv: stacktrace: Move register keyword to beginning of declaration")
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Palmer Dabbelt <palmerdabbelt@google.com>
|
|
Merge module sections only when using Clang LTO. With ld.bfd, merging
sections does not appear to update the symbol tables for the module,
e.g. 'readelf -s' shows the value that a symbol would have had, if
sections were not merged. ld.lld does not show this problem.
The stale symbol table breaks gdb's function disassembler, and presumably
other things, e.g.
gdb -batch -ex "file arch/x86/kvm/kvm.ko" -ex "disassemble kvm_init"
reads the wrong bytes and dumps garbage.
Fixes: dd2776222abb ("kbuild: lto: merge module sections")
Cc: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Reviewed-by: Sami Tolvanen <samitolvanen@google.com>
Tested-by: Sami Tolvanen <samitolvanen@google.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Link: https://lore.kernel.org/r/20210322234438.502582-1-seanjc@google.com
|
|
Commit cbc3b92ce037 fixed an issue to modify the macros of the stack trace
event so that user space could parse it properly. Originally the stack
trace format to user space showed that the called stack was a dynamic
array. But it is not actually a dynamic array, in the way that other
dynamic event arrays worked, and this broke user space parsing for it. The
update was to make the array look to have 8 entries in it. Helper
functions were added to make it parse it correctly, as the stack was
dynamic, but was determined by the size of the event stored.
Although this fixed user space on how it read the event, it changed the
internal structure used for the stack trace event. It changed the array
size from [0] to [8] (added 8 entries). This increased the size of the
stack trace event by 8 words. The size reserved on the ring buffer was the
size of the stack trace event plus the number of stack entries found in
the stack trace. That commit caused the amount to be 8 more than what was
needed because it did not expect the caller field to have any size. This
produced 8 entries of garbage (and reading random data) from the stack
trace event:
<idle>-0 [002] d... 1976396.837549: <stack trace>
=> trace_event_raw_event_sched_switch
=> __traceiter_sched_switch
=> __schedule
=> schedule_idle
=> do_idle
=> cpu_startup_entry
=> secondary_startup_64_no_verify
=> 0xc8c5e150ffff93de
=> 0xffff93de
=> 0
=> 0
=> 0xc8c5e17800000000
=> 0x1f30affff93de
=> 0x00000004
=> 0x200000000
Instead, subtract the size of the caller field from the size of the event
to make sure that only the amount needed to store the stack trace is
reserved.
Link: https://lore.kernel.org/lkml/your-ad-here.call-01617191565-ext-9692@work.hours/
Cc: stable@vger.kernel.org
Fixes: cbc3b92ce037 ("tracing: Set kernel_stack's caller size properly")
Reported-by: Vasily Gorbik <gor@linux.ibm.com>
Tested-by: Vasily Gorbik <gor@linux.ibm.com>
Acked-by: Vasily Gorbik <gor@linux.ibm.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
|
|
iov_iter_revert() is done in completion handlers that happensf before
read/write returns -EIOCBQUEUED, no need to repeat reverting afterwards.
Moreover, even though it may appear being just a no-op, it's actually
races with 1) user forging a new iovec of a different size 2) reissue,
that is done via io-wq continues completely asynchronously.
Fixes: 3e6a0d3c7571c ("io_uring: fix -EAGAIN retry with IOPOLL")
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
task_pid may be large enough to not fit into the left space of
TASK_COMM_LEN-sized buffers and overflow in sprintf. We not so care
about uniqueness, so replace it with safer snprintf().
Reported-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/1702c6145d7e1c46fbc382f28334c02e1a3d3994.1617267273.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
S_ISBLK is marked as unbounded work for async preparation, because it
doesn't match S_ISREG. That is incorrect, as any read/write to a block
device is also a bounded operation. Fix it up and ensure that S_ISBLK
isn't marked unbounded.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
Memory backed or zoned null block devices may generate actual request
timeout errors due to the submission path being blocked on memory
allocation or zone locking. Unlike fake timeouts or injected timeouts,
the request submission path will call blk_mq_complete_request() or
blk_mq_end_request() for these real timeout errors, causing a double
completion and use after free situation as the block layer timeout
handler executes blk_mq_rq_timed_out() and __blk_mq_free_request() in
blk_mq_check_expired(). This problem often triggers a NULL pointer
dereference such as:
BUG: kernel NULL pointer dereference, address: 0000000000000050
RIP: 0010:blk_mq_sched_mark_restart_hctx+0x5/0x20
...
Call Trace:
dd_finish_request+0x56/0x80
blk_mq_free_request+0x37/0x130
null_handle_cmd+0xbf/0x250 [null_blk]
? null_queue_rq+0x67/0xd0 [null_blk]
blk_mq_dispatch_rq_list+0x122/0x850
__blk_mq_do_dispatch_sched+0xbb/0x2c0
__blk_mq_sched_dispatch_requests+0x13d/0x190
blk_mq_sched_dispatch_requests+0x30/0x60
__blk_mq_run_hw_queue+0x49/0x90
process_one_work+0x26c/0x580
worker_thread+0x55/0x3c0
? process_one_work+0x580/0x580
kthread+0x134/0x150
? kthread_create_worker_on_cpu+0x70/0x70
ret_from_fork+0x1f/0x30
This problem very often triggers when running the full btrfs xfstests
on a memory-backed zoned null block device in a VM with limited amount
of memory.
Avoid this by executing blk_mq_complete_request() in null_timeout_rq()
only for commands that are marked for a fake timeout completion using
the fake_timeout boolean in struct null_cmd. For timeout errors injected
through debugfs, the timeout handler will execute
blk_mq_complete_request()i as before. This is safe as the submission
path does not execute complete requests in this case.
In null_timeout_rq(), also make sure to set the command error field to
BLK_STS_TIMEOUT and to propagate this error through to the request
completion.
Reported-by: Johannes Thumshirn <Johannes.Thumshirn@wdc.com>
Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>
Tested-by: Johannes Thumshirn <Johannes.Thumshirn@wdc.com>
Reviewed-by: Johannes Thumshirn <Johannes.Thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20210331225244.126426-1-damien.lemoal@wdc.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
Instead of just reporting an assertion failure, report enough information
that we can start diagnosing exactly went wrong.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
|
|
The throbber could race with creation of the anchor entry and cause the
IDR to have zero entries in it, which would cause the test to fail.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
|
|
When run on a single CPU, this test would frequently access already-freed
memory. Due to timing, this bug never showed up on multi-CPU tests.
Reported-by: Chris von Recklinghausen <crecklin@redhat.com>
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
|
|
Several test runners register individual worker threads with the
RCU library, but neglect to register the main thread, which can lead
to objects being freed while the main thread is in what appears to be
an RCU critical section.
Reported-by: Chris von Recklinghausen <crecklin@redhat.com>
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
|
|
Commit 496121c02127 ("ACPI: processor: idle: Allow probing on platforms
with one ACPI C-state") broke CPU0 hotplug on certain systems, e.g.
I'm observing the following on AWS Nitro (e.g r5b.xlarge but other
instance types are affected as well):
# echo 0 > /sys/devices/system/cpu/cpu0/online
# echo 1 > /sys/devices/system/cpu/cpu0/online
<10 seconds delay>
-bash: echo: write error: Input/output error
In fact, the above mentioned commit only revealed the problem and did
not introduce it. On x86, to wakeup CPU an NMI is being used and
hlt_play_dead()/mwait_play_dead() loops are prepared to handle it:
/*
* If NMI wants to wake up CPU0, start CPU0.
*/
if (wakeup_cpu0())
start_cpu0();
cpuidle_play_dead() -> acpi_idle_play_dead() (which is now being called on
systems where it wasn't called before the above mentioned commit) serves
the same purpose but it doesn't have a path for CPU0. What happens now on
wakeup is:
- NMI is sent to CPU0
- wakeup_cpu0_nmi() works as expected
- we get back to while (1) loop in acpi_idle_play_dead()
- safe_halt() puts CPU0 to sleep again.
The straightforward/minimal fix is add the special handling for CPU0 on x86
and that's what the patch is doing.
Fixes: 496121c02127 ("ACPI: processor: idle: Allow probing on platforms with one ACPI C-state")
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: 5.10+ <stable@vger.kernel.org> # 5.10+
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
|
|
Add a test for the issue when KVM_SET_CLOCK(0) call could cause
TSC page value to go very big because of a signedness issue around
hv_clock->system_time.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Message-Id: <20210326155551.17446-3-vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
When guest time is reset with KVM_SET_CLOCK(0), it is possible for
'hv_clock->system_time' to become a small negative number. This happens
because in KVM_SET_CLOCK handling we set 'kvm->arch.kvmclock_offset' based
on get_kvmclock_ns(kvm) but when KVM_REQ_CLOCK_UPDATE is handled,
kvm_guest_time_update() does (masterclock in use case):
hv_clock.system_time = ka->master_kernel_ns + v->kvm->arch.kvmclock_offset;
And 'master_kernel_ns' represents the last time when masterclock
got updated, it can precede KVM_SET_CLOCK() call. Normally, this is not a
problem, the difference is very small, e.g. I'm observing
hv_clock.system_time = -70 ns. The issue comes from the fact that
'hv_clock.system_time' is stored as unsigned and 'system_time / 100' in
compute_tsc_page_parameters() becomes a very big number.
Use 'master_kernel_ns' instead of get_kvmclock_ns() when masterclock is in
use and get_kvmclock_base_ns() when it's not to prevent 'system_time' from
going negative.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Message-Id: <20210331124130.337992-2-vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
pvclock_gtod_sync_lock can be taken with interrupts disabled if the
preempt notifier calls get_kvmclock_ns to update the Xen
runstate information:
spin_lock include/linux/spinlock.h:354 [inline]
get_kvmclock_ns+0x25/0x390 arch/x86/kvm/x86.c:2587
kvm_xen_update_runstate+0x3d/0x2c0 arch/x86/kvm/xen.c:69
kvm_xen_update_runstate_guest+0x74/0x320 arch/x86/kvm/xen.c:100
kvm_xen_runstate_set_preempted arch/x86/kvm/xen.h:96 [inline]
kvm_arch_vcpu_put+0x2d8/0x5a0 arch/x86/kvm/x86.c:4062
So change the users of the spinlock to spin_lock_irqsave and
spin_unlock_irqrestore.
Reported-by: syzbot+b282b65c2c68492df769@syzkaller.appspotmail.com
Fixes: 30b5c851af79 ("KVM: x86/xen: Add support for vCPU runstate information")
Cc: David Woodhouse <dwmw@amazon.co.uk>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
There is no need to include changes to vcpu->requests into
the pvclock_gtod_sync_lock critical section. The changes to
the shared data structures (in pvclock_update_vm_gtod_copy)
already occur under the lock.
Cc: David Woodhouse <dwmw@amazon.co.uk>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
Fixing nested_vmcb_check_save to avoid all TOC/TOU races
is a bit harder in released kernels, so do the bare minimum
by avoiding that EFER.SVME is cleared. This is problematic
because svm_set_efer frees the data structures for nested
virtualization if EFER.SVME is cleared.
Also check that EFER.SVME remains set after a nested vmexit;
clearing it could happen if the bit is zero in the save area
that is passed to KVM_SET_NESTED_STATE (the save area of the
nested state corresponds to the nested hypervisor's state
and is restored on the next nested vmexit).
Cc: stable@vger.kernel.org
Fixes: 2fcf4876ada ("KVM: nSVM: implement on demand allocation of the nested state")
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
Avoid races between check and use of the nested VMCB controls. This
for example ensures that the VMRUN intercept is always reflected to the
nested hypervisor, instead of being processed by the host. Without this
patch, it is possible to end up with svm->nested.hsave pointing to
the MSR permission bitmap for nested guests.
This bug is CVE-2021-29657.
Reported-by: Felix Wilhelm <fwilhelm@google.com>
Cc: stable@vger.kernel.org
Fixes: 2fcf4876ada ("KVM: nSVM: implement on demand allocation of the nested state")
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
The page table of AMDGPU requires an alignment to CPU page so we should
check ioctl parameters for it. Return -EINVAL if some parameter is
unaligned to CPU page, instead of corrupt the page table sliently.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Xi Ruoyao <xry111@mengyan1223.wang>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
|
|
In Mesa, dev_info.gart_page_size is used for alignment and it was
set to AMDGPU_GPU_PAGE_SIZE(4KB). However, the page table of AMDGPU
driver requires an alignment on CPU pages. So, for non-4KB page system,
gart_page_size should be max_t(u32, PAGE_SIZE, AMDGPU_GPU_PAGE_SIZE).
Signed-off-by: Rui Wang <wangr@lemote.com>
Signed-off-by: Huacai Chen <chenhc@lemote.com>
Link: https://github.com/loongson-community/linux-stable/commit/caa9c0a1
[Xi: rebased for drm-next, use max_t for checkpatch,
and reworded commit message.]
Signed-off-by: Xi Ruoyao <xry111@mengyan1223.wang>
BugLink: https://gitlab.freedesktop.org/drm/amd/-/issues/1549
Tested-by: Dan Horák <dan@danny.cz>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
|
|
Do the same thing we do for Renoir. We can check, but since
the sbios has started DPM, it will always return true which
causes the driver to skip some of the SMU init when it shouldn't.
Reviewed-by: Zhan Liu <zhan.liu@amd.com>
Acked-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
|
|
Amdgpu driver uses 4-byte data type as DQM fence memory,
and transmits GPU address of fence memory to microcode
through query status PM4 message. However, query status
PM4 message definition and microcode processing are all
processed according to 8 bytes. Fence memory only allocates
4 bytes of memory, but microcode does write 8 bytes of memory,
so there is a memory corruption.
Changes since v1:
* Change dqm->fence_addr as a u64 pointer to fix this issue,
also fix up query_status and amdkfd_fence_wait_timeout function
uses 64 bit fence value to make them consistent.
Signed-off-by: Qu Huang <jinsdb@126.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
|
|
For multiple split bios, if one of the bio is fail, the whole
should return error to application. But we found there is a race
between bio_integrity_verify_fn and bio complete, which return
io success to application after one of the bio fail. The race as
following:
split bio(READ) kworker
nvme_complete_rq
blk_update_request //split error=0
bio_endio
bio_integrity_endio
queue_work(kintegrityd_wq, &bip->bip_work);
bio_integrity_verify_fn
bio_endio //split bio
__bio_chain_endio
if (!parent->bi_status)
<interrupt entry>
nvme_irq
blk_update_request //parent error=7
req_bio_endio
bio->bi_status = 7 //parent bio
<interrupt exit>
parent->bi_status = 0
parent->bi_end_io() // return bi_status=0
The bio has been split as two: split and parent. When split
bio completed, it depends on kworker to do endio, while
bio_integrity_verify_fn have been interrupted by parent bio
complete irq handler. Then, parent bio->bi_status which have
been set in irq handler will overwrite by kworker.
In fact, even without the above race, we also need to conside
the concurrency beteen mulitple split bio complete and update
the same parent bi_status. Normally, multiple split bios will
be issued to the same hctx and complete from the same irq
vector. But if we have updated queue map between multiple split
bios, these bios may complete on different hw queue and different
irq vector. Then the concurrency update parent bi_status may
cause the final status error.
Suggested-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Yufen Yu <yuyufen@huawei.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20210331115359.1125679-1-yuyufen@huawei.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
syzbot is reporting NULL pointer dereference at reiserfs_security_init()
[1], for commit ab17c4f02156c4f7 ("reiserfs: fixup xattr_root caching")
is assuming that REISERFS_SB(s)->xattr_root != NULL in
reiserfs_xattr_jcreate_nblocks() despite that commit made
REISERFS_SB(sb)->priv_root != NULL && REISERFS_SB(s)->xattr_root == NULL
case possible.
I guess that commit 6cb4aff0a77cc0e6 ("reiserfs: fix oops while creating
privroot with selinux enabled") wanted to check xattr_root != NULL
before reiserfs_xattr_jcreate_nblocks(), for the changelog is talking
about the xattr root.
The issue is that while creating the privroot during mount
reiserfs_security_init calls reiserfs_xattr_jcreate_nblocks which
dereferences the xattr root. The xattr root doesn't exist, so we get
an oops.
Therefore, update reiserfs_xattrs_initialized() to check both the
privroot and the xattr root.
Link: https://syzkaller.appspot.com/bug?id=8abaedbdeb32c861dc5340544284167dd0e46cde # [1]
Reported-and-tested-by: syzbot <syzbot+690cb1e51970435f9775@syzkaller.appspotmail.com>
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Fixes: 6cb4aff0a77c ("reiserfs: fix oops while creating privroot with selinux enabled")
Acked-by: Jeff Mahoney <jeffm@suse.com>
Acked-by: Jan Kara <jack@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
Don't call into get_signal() with the sqd mutex held, it'll fail if we're
freezing the task and we'll get complaints on locks still being held:
====================================
WARNING: iou-sqp-8386/8387 still has locks held!
5.12.0-rc4-syzkaller #0 Not tainted
------------------------------------
1 lock held by iou-sqp-8386/8387:
#0: ffff88801e1d2470 (&sqd->lock){+.+.}-{3:3}, at: io_sq_thread+0x24c/0x13a0 fs/io_uring.c:6731
stack backtrace:
CPU: 1 PID: 8387 Comm: iou-sqp-8386 Not tainted 5.12.0-rc4-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
__dump_stack lib/dump_stack.c:79 [inline]
dump_stack+0x141/0x1d7 lib/dump_stack.c:120
try_to_freeze include/linux/freezer.h:66 [inline]
get_signal+0x171a/0x2150 kernel/signal.c:2576
io_sq_thread+0x8d2/0x13a0 fs/io_uring.c:6748
Fold the get_signal() case in with the parking checks, as we need to drop
the lock in both cases, and since we need to be checking for parking when
juggling the lock anyway.
Reported-by: syzbot+796d767eb376810256f5@syzkaller.appspotmail.com
Fixes: dbe1bdbb39db ("io_uring: handle signals for IO threads like a normal thread")
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
Commit 71da201f38df ("ACPI: scan: Defer enumeration of devices with
_DEP lists") dropped the following 2 lines from acpi_init_device_object():
/* Assume there are unmet deps until acpi_device_dep_initialize() runs */
device->dep_unmet = 1;
Leaving the initial value of dep_unmet at the 0 from the kzalloc(). This
causes the acpi_bus_get_status() call in acpi_add_single_object() to
actually call _STA, even though there maybe unmet deps, leading to errors
like these:
[ 0.123579] ACPI Error: No handler for Region [ECRM] (00000000ba9edc4c)
[GenericSerialBus] (20170831/evregion-166)
[ 0.123601] ACPI Error: Region GenericSerialBus (ID=9) has no handler
(20170831/exfldio-299)
[ 0.123618] ACPI Error: Method parse/execution failed
\_SB.I2C1.BAT1._STA, AE_NOT_EXIST (20170831/psparse-550)
Fix this by re-adding the dep_unmet = 1 initialization to
acpi_init_device_object() and modifying acpi_bus_check_add() to make sure
that dep_unmet always gets setup there, overriding the initial 1 value.
This re-fixes the issue initially fixed by
commit 63347db0affa ("ACPI / scan: Use acpi_bus_get_status() to initialize
ACPI_TYPE_DEVICE devs"), which introduced the removed
"device->dep_unmet = 1;" statement.
This issue was noticed; and the fix tested on a Dell Venue 10 Pro 5055.
Fixes: 71da201f38df ("ACPI: scan: Defer enumeration of devices with _DEP lists")
Suggested-by: Rafael J. Wysocki <rafael@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Cc: 5.11+ <stable@vger.kernel.org> # 5.11+
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
|
|
The SOR resets are exclusively shared with the SOR power domain. This
means that exclusive access can only be granted temporarily and in order
for that to work, a rigorous sequence must be observed. To ensure that a
single consumer gets exclusive access to a reset, each consumer must
implement a rigorous protocol using the reset_control_acquire() and
reset_control_release() functions.
However, these functions alone don't provide any guarantees at the
system level. Drivers need to ensure that the only a single consumer has
access to the reset at the same time. In order for the SOR to be able to
exclusively access its reset, it must therefore ensure that the SOR
power domain is not powered off by holding on to a runtime PM reference
to that power domain across the reset assert/deassert operation.
This used to work fine by accident, but was revealed when recently more
devices started to rely on the SOR power domain.
Fixes: 11c632e1cfd3 ("drm/tegra: sor: Implement acquire/release for reset")
Reported-by: Jonathan Hunter <jonathanh@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
|
|
Commit 4bba4c4bb09a added tools/include/linux/compiler_types.h which
includes linux/compiler-gcc.h. Unfortunately, we had our own (empty)
compiler_types.h which overrode the one added by that commit, and
so we lost the definition of __must_be_array(). Removing our empty
compiler_types.h fixes the problem and reduces our divergence from the
rest of the tools.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
|
Connor Kuehl
Luis Henriques
Vivek Goyal
Alessio Balsini
Bhaskar Chowdhury
Miklos Szeredi
Linus Torvalds
Zheyu Ma
Zhen Lei
Jens Axboe
Christoph Hellwig
Pavel Begunkov
Kefeng Wang
Yang Li
Zihao Yu
Ben Dooks
Sean Christopherson
Steven Rostedt (VMware)
Damien Le Moal
Matthew Wilcox (Oracle)
Vitaly Kuznetsov
Paolo Bonzini
Xℹ Ruoyao
Huacai Chen
Alex Deucher
Qu Huang
Yufen Yu
Tetsuo Handa
Hans de Goede
Thierry Reding