| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
When we add "atgc" to the fstab table, ATGC is not immediately enabled.
There is a 7-day time threshold, and we can use "atgc_enabled" to
show whether ATGC is enabled.
Signed-off-by: liujinbao1 <liujinbao1@xiaomi.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
This reverts commit c550e25bca660ed2554cbb48d32b82d0bb98e4b1.
Commit c550e25bca660ed2554cbb48d32b82d0bb98e4b1 ("f2fs: use flush
command instead of FUA for zoned device") used additional flush
command to keep write order.
Since Commit dd291d77cc90eb6a86e9860ba8e6e38eebd57d12 ("block:
Introduce zone write plugging") has enabled the block layer to
handle this order issue, there is no need to use flush command.
Signed-off-by: Wenjie Cheng <cwjhust@gmail.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
Convert to use folio and related functionality.
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
If lfs mode is on, buffered read may race w/ OPU dio write as below,
it may cause buffered read hits unwritten data unexpectly, and for
dio read, the race condition exists as well.
Thread A Thread B
- f2fs_file_write_iter
- f2fs_dio_write_iter
- __iomap_dio_rw
- f2fs_iomap_begin
- f2fs_map_blocks
- __allocate_data_block
- allocated blkaddr #x
- iomap_dio_submit_bio
- f2fs_file_read_iter
- filemap_read
- f2fs_read_data_folio
- f2fs_mpage_readpages
- f2fs_map_blocks
: get blkaddr #x
- f2fs_submit_read_bio
IRQ
- f2fs_read_end_io
: read IO on blkaddr #x complete
IRQ
- iomap_dio_bio_end_io
: direct write IO on blkaddr #x complete
In LFS mode, if there is inflight dio, let's wait for its completion,
this policy won't cover all race cases, however it is a tradeoff which
avoids abusing lock around IO paths.
Fixes: f847c699cff3 ("f2fs: allow out-place-update for direct IO in LFS mode")
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
It should wait all existing dio write IOs before block removal,
otherwise, previous direct write IO may overwrite data in the
block which may be reused by other inode.
Cc: stable@vger.kernel.org
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
We may trigger high frequent checkpoint for below case:
1. mkdir /mnt/dir1; set dir1 encrypted
2. touch /mnt/file1; fsync /mnt/file1
3. mkdir /mnt/dir2; set dir2 encrypted
4. touch /mnt/file2; fsync /mnt/file2
...
Although, newly created dir and file are not related, due to
commit bbf156f7afa7 ("f2fs: fix lost xattrs of directories"), we will
trigger checkpoint whenever fsync() comes after a new encrypted dir
created.
In order to avoid such performance regression issue, let's record an
entry including directory's ino in global cache whenever we update
directory's xattr data, and then triggerring checkpoint() only if
xattr metadata of target file's parent was updated.
This patch updates to cover below no encryption case as well:
1) parent is checkpointed
2) set_xattr(dir) w/ new xnid
3) create(file)
4) fsync(file)
Fixes: bbf156f7afa7 ("f2fs: fix lost xattrs of directories")
Reported-by: wangzijie <wangzijie1@honor.com>
Reported-by: Zhiguo Niu <zhiguo.niu@unisoc.com>
Tested-by: Zhiguo Niu <zhiguo.niu@unisoc.com>
Reported-by: Yunlei He <heyunlei@hihonor.com>
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
Case #1:
SQLite App GC Thread Kworker Shrinker
- f2fs_ioc_start_atomic_write
- f2fs_ioc_commit_atomic_write
- f2fs_commit_atomic_write
- filemap_write_and_wait_range
: write atomic_file's data to cow_inode
echo 3 > drop_caches
to drop atomic_file's
cache.
- f2fs_gc
- gc_data_segment
- move_data_page
- set_page_dirty
- writepages
- f2fs_do_write_data_page
: overwrite atomic_file's data
to cow_inode
- f2fs_down_write(&fi->i_gc_rwsem[WRITE])
- __f2fs_commit_atomic_write
- f2fs_up_write(&fi->i_gc_rwsem[WRITE])
Case #2:
SQLite App GC Thread Kworker
- f2fs_ioc_start_atomic_write
- __writeback_single_inode
- do_writepages
- f2fs_write_cache_pages
- f2fs_write_single_data_page
- f2fs_do_write_data_page
: write atomic_file's data to cow_inode
- f2fs_gc
- gc_data_segment
- move_data_page
- set_page_dirty
- writepages
- f2fs_do_write_data_page
: overwrite atomic_file's data to cow_inode
- f2fs_ioc_commit_atomic_write
In above cases racing in between atomic_write and GC, previous
data in atomic_file may be overwrited to cow_file, result in
data corruption.
This patch introduces PAGE_PRIVATE_ATOMIC_WRITE bit flag in page.private,
and use it to indicate that there is last dirty data in atomic file,
and the data should be writebacked into cow_file, if the flag is not
tagged in page, we should never write data across files.
Fixes: 3db1de0e582c ("f2fs: change the current atomic write way")
Cc: Daeho Jeong <daehojeong@google.com>
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
The macro stat_inc_cp_count accepts a parameter si,
but it was not used, rather the variable sbi was directly used,
which may be a local variable inside a function that calls the macros.
Signed-off-by: Julian Sun <sunjunchao2870@gmail.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
The macro on_f2fs_build_free_nids accepts a parameter nmi,
but it was not used, rather the variable nm_i was directly used,
which may be a local variable inside a function that calls the macros.
Signed-off-by: Julian Sun <sunjunchao2870@gmail.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
Currently, we are using a mix of traditional UFS and zone UFS to support
some functionalities that cannot be achieved on zone UFS alone. However,
there are some issues with this approach. There exists a significant
performance difference between traditional UFS and zone UFS. Under normal
usage, we prioritize writes to zone UFS. However, in critical conditions
(such as when the entire UFS is almost full), we cannot determine whether
data will be written to traditional UFS or zone UFS. This can lead to
significant performance fluctuations, which is not conducive to
development and testing. To address this, we have added an option
zlu_io_enable under sys with the following three modes:
1) zlu_io_enable == 0:Normal mode, prioritize writing to zone UFS;
2) zlu_io_enable == 1:Zone UFS only mode, only allow writing to zone UFS;
3) zlu_io_enable == 2:Traditional UFS priority mode, prioritize writing to
traditional UFS.
Signed-off-by: Liao Yuanhong <liaoyuanhong@vivo.com>
Signed-off-by: Wu Bo <bo.wu@vivo.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
While calculating the end addresses of main area and segment 0, u32
may be not enough to hold the result without the danger of int
overflow.
Just in case, play it safe and cast one of the operands to a
wider type (u64).
Found by Linux Verification Center (linuxtesting.org) with static
analysis tool SVACE.
Fixes: fd694733d523 ("f2fs: cover large section in sanity check of super")
Cc: stable@vger.kernel.org
Signed-off-by: Nikita Zhandarovich <n.zhandarovich@fintech.ru>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
When dealing with large extents and calculating file offsets by
summing up according extent offsets and lengths of unsigned int type,
one may encounter possible integer overflow if the values are
big enough.
Prevent this from happening by expanding one of the addends to
(pgoff_t) type.
Found by Linux Verification Center (linuxtesting.org) with static
analysis tool SVACE.
Fixes: d323d005ac4a ("f2fs: support file defragment")
Cc: stable@vger.kernel.org
Signed-off-by: Nikita Zhandarovich <n.zhandarovich@fintech.ru>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
The result of multiplication between values derived from functions
dir_buckets() and bucket_blocks() *could* technically reach
2^30 * 2^2 = 2^32.
While unlikely to happen, it is prudent to ensure that it will not
lead to integer overflow. Thus, use mul_u32_u32() as it's more
appropriate to mitigate the issue.
Found by Linux Verification Center (linuxtesting.org) with static
analysis tool SVACE.
Fixes: 3843154598a0 ("f2fs: introduce large directory support")
Cc: stable@vger.kernel.org
Signed-off-by: Nikita Zhandarovich <n.zhandarovich@fintech.ru>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
Introudce a new help get_dnode_base() to wrap common code from
get_dnode_addr() and data_blkaddr() for cleanup.
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
|
|
The kernel sleep profile is no longer working due to a recursive locking
bug introduced by commit 42a20f86dc19 ("sched: Add wrapper for get_wchan()
to keep task blocked")
Booting with the 'profile=sleep' kernel command line option added or
executing
# echo -n sleep > /sys/kernel/profiling
after boot causes the system to lock up.
Lockdep reports
kthreadd/3 is trying to acquire lock:
ffff93ac82e08d58 (&p->pi_lock){....}-{2:2}, at: get_wchan+0x32/0x70
but task is already holding lock:
ffff93ac82e08d58 (&p->pi_lock){....}-{2:2}, at: try_to_wake_up+0x53/0x370
with the call trace being
lock_acquire+0xc8/0x2f0
get_wchan+0x32/0x70
__update_stats_enqueue_sleeper+0x151/0x430
enqueue_entity+0x4b0/0x520
enqueue_task_fair+0x92/0x6b0
ttwu_do_activate+0x73/0x140
try_to_wake_up+0x213/0x370
swake_up_locked+0x20/0x50
complete+0x2f/0x40
kthread+0xfb/0x180
However, since nobody noticed this regression for more than two years,
let's remove 'profile=sleep' support based on the assumption that nobody
needs this functionality.
Fixes: 42a20f86dc19 ("sched: Add wrapper for get_wchan() to keep task blocked")
Cc: stable@vger.kernel.org # v5.16+
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
Commit 04f08ef291d4 ("arm/arm64: dts: arm: Use generic clock and
regulator nodenames") renamed nodes and created 2 "clock-24000000" nodes
(at different paths).
The kernel can't handle these duplicate names even though they are at
different paths. Fix this by renaming one of the nodes to "clock-pclk".
This name is aligned with other Arm boards (those didn't have a known
frequency to use in the node name).
Fixes: 04f08ef291d4 ("arm/arm64: dts: arm: Use generic clock and regulator nodenames")
Reported-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Rob Herring (Arm) <robh@kernel.org>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Tested-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
The runtime constants linker script depended on documented linker
behavior [1]:
"If an output section’s name is the same as the input section’s name
and is representable as a C identifier, then the linker will
automatically PROVIDE two symbols: __start_SECNAME and __stop_SECNAME,
where SECNAME is the name of the section. These indicate the start
address and end address of the output section respectively"
to just automatically define the symbol names for the bounds of the
runtime constant arrays.
It turns out that this isn't actually something we can rely on, with old
linkers not generating these automatic symbols. It looks to have been
introduced in binutils-2.29 back in 2017, and we still support building
with versions all the way back to binutils-2.25 (from 2015).
And yes, Oleg actually seems to be using such ancient versions of
binutils.
So instead of depending on the implicit symbols from "section names
match and are representable C identifiers", just do this all manually.
It's not like it causes us any extra pain, we already have to do that
for all the other sections that we use that often have special
characters in them.
Reported-and-tested-by: Oleg Nesterov <oleg@redhat.com>
Link: https://sourceware.org/binutils/docs/ld/Input-Section-Example.html [1]
Link: https://lore.kernel.org/all/20240802114518.GA20924@redhat.com/
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
The current "nretries > 1 || nretries >= max_retries" check in
cs_watchdog_read() will always evaluate to true, and thus pr_warn(), if
nretries is greater than 1. The intent is instead to never warn on the
first try, but otherwise warn if the successful retry was the last retry.
Therefore, change that "||" to "&&".
Fixes: db3a34e17433 ("clocksource: Retry clock read if long delays detected")
Reported-by: Borislav Petkov <bp@alien8.de>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/all/20240802154618.4149953-2-paulmck@kernel.org
|
|
To 2.50
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
NetApp server requires the file to be open with FILE_READ_EA access in
order to support FSCTL_GET_REPARSE_POINT, otherwise it will return
STATUS_INVALID_DEVICE_REQUEST. It doesn't make any sense because
there's no requirement for FILE_READ_EA bit to be set nor
STATUS_INVALID_DEVICE_REQUEST being used for something other than
"unsupported reparse points" in MS-FSA.
To fix it and improve compatibility, set FILE_READ_EA & SYNCHRONIZE
bits to match what Windows client currently does.
Tested-by: Sebastian Steinbeisser <Sebastian.Steinbeisser@lrz.de>
Acked-by: Tom Talpey <tom@talpey.com>
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
For debugging an umount failure in xfstests generic/043 generic/044 in some
configurations, we needed more information on the shutdown ioctl which
was suspected of being related to the cause, so tracepoints are added
in this patch e.g.
"trace-cmd record -e smb3_shutdown_enter -e smb3_shutdown_done -e smb3_shutdown_err"
Sample output:
godown-47084 [011] ..... 3313.756965: smb3_shutdown_enter: flags=0x1 tid=0x733b3e75
godown-47084 [011] ..... 3313.756968: smb3_shutdown_done: flags=0x1 tid=0x733b3e75
Tested-by: Anthony Nandaa (Microsoft) <profnandaa@gmail.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
Remove struct cifs_aio_ctx and its associated alloc/release functions as it
is no longer used, the functions being taken over by netfslib.
Signed-off-by: David Howells <dhowells@redhat.com>
cc: Steve French <sfrench@samba.org>
cc: linux-cifs@vger.kernel.org
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
As per MS-FSA 2.1.5.10.14, support for FSCTL_GET_REPARSE_POINT is
optional and if the server doesn't support it,
STATUS_INVALID_DEVICE_REQUEST must be returned for the operation.
If we find files with reparse points and we can't read them due to
lack of client or server support, just ignore it and then treat them
as regular files or junctions.
Fixes: 5f71ebc41294 ("smb: client: parse reparse point flag in create response")
Reported-by: Sebastian Steinbeisser <Sebastian.Steinbeisser@lrz.de>
Tested-by: Sebastian Steinbeisser <Sebastian.Steinbeisser@lrz.de>
Acked-by: Tom Talpey <tom@talpey.com>
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
Although the Arm architecture permits concurrent modification and
execution of NOP and branch instructions, it still requires some
synchronisation to ensure that other CPUs consistently execute the newly
written instruction:
> When the modified instructions are observable, each PE that is
> executing the modified instructions must execute an ISB or perform a
> context synchronizing event to ensure execution of the modified
> instructions
Prior to commit f6cc0c501649 ("arm64: Avoid calling stop_machine() when
patching jump labels"), the arm64 jump_label patching machinery
performed synchronisation using stop_machine() after each modification,
however this was problematic when flipping static keys from atomic
contexts (namely, the arm_arch_timer CPU hotplug startup notifier) and
so we switched to the _nosync() patching routines to avoid "scheduling
while atomic" BUG()s during boot.
In hindsight, the analysis of the issue in f6cc0c501649 isn't quite
right: it cites the use of IPIs in the default patching routines as the
cause of the lockup, whereas stop_machine() does not rely on IPIs and
the I-cache invalidation is performed using __flush_icache_range(),
which elides the call to kick_all_cpus_sync(). In fact, the blocking
wait for other CPUs is what triggers the BUG() and the problem remains
even after f6cc0c501649, for example because we could block on the
jump_label_mutex. Eventually, the arm_arch_timer driver was fixed to
avoid the static key entirely in commit a862fc2254bd
("clocksource/arm_arch_timer: Remove use of workaround static key").
This all leaves the jump_label patching code in a funny situation on
arm64 as we do not synchronise with other CPUs to reduce the likelihood
of a bug which no longer exists. Consequently, toggling a static key on
one CPU cannot be assumed to take effect on other CPUs, leading to
potential issues, for example with missing preempt notifiers.
Rather than revert f6cc0c501649 and go back to stop_machine() for each
patch site, implement arch_jump_label_transform_apply() and kick all
the other CPUs with an IPI at the end of patching.
Cc: Alexander Potapenko <glider@google.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Marc Zyngier <maz@kernel.org>
Fixes: f6cc0c501649 ("arm64: Avoid calling stop_machine() when patching jump labels")
Signed-off-by: Will Deacon <will@kernel.org>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Reviewed-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20240731133601.3073-1-will@kernel.org
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
The __NR_newfstat and __NR_newfstatat macros accidentally got renamed
in the conversion to the syscall.tbl format, dropping the 'new' portion
of the name.
In an unrelated change, the two syscalls are no longer architecture
specific but are once more defined on all 64-bit architectures, so the
'newstat' ABI keyword can be dropped from the table as a simplification.
Fixes: Fixes: 4fe53bf2ba0a ("syscalls: add generic scripts/syscall.tbl")
Closes: https://lore.kernel.org/lkml/838053e0-b186-4e9f-9668-9a3384a71f23@app.fastmail.com/T/#t
Reported-by: Florian Weimer <fweimer@redhat.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
|
|
Despite multiple attempts to get the syscall number assignment right
for the newly added uretprobe syscall, we ended up with a bit of a mess:
- The number is defined as 467 based on the assumption that the
xattrat family of syscalls would use 463 through 466, but those
did not make it into 6.11.
- The include/uapi/asm-generic/unistd.h file still lists the number
463, but the new scripts/syscall.tbl that was supposed to have the
same data lists 467 instead as the number for arc, arm64, csky,
hexagon, loongarch, nios2, openrisc and riscv. None of these
architectures actually provide a uretprobe syscall.
- All the other architectures (powerpc, arm, mips, ...) don't list
this syscall at all.
There are two ways to make it consistent again: either list it with
the same syscall number on all architectures, or only list it on x86
but not in scripts/syscall.tbl and asm-generic/unistd.h.
Based on the most recent discussion, it seems like we won't need it
anywhere else, so just remove the inconsistent assignment and instead
move the x86 number to the next available one in the architecture
specific range, which is 335.
Fixes: 5c28424e9a34 ("syscalls: Fix to add sys_uretprobe to syscall.tbl")
Fixes: 190fec72df4a ("uprobe: Wire up uretprobe system call")
Fixes: 63ded110979b ("uprobe: Change uretprobe syscall scope and number")
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Reviewed-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
|
|
The thermal sysfs API document is outdated. One of the problems with
it is that is still documents thermal_zone_device_register() which
does not exit any more and it does not reflect the current thermal
zone operations definition.
Replace the thermal_zone_device_register() description in it with
a thermal_zone_device_register_with_trips() description, including
an update of the thermal zone operations list.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Link: https://patch.msgid.link/2767845.mvXUDI8C0e@rjwysocki.net
|
|
This reverts commit d45bb9c5f7a6f7b6e47939856b28cb1da0cdc119.
Just got a report that this causes some suspend/resume issues,
so back it out and I'll investigate it later.
Reported-by: Mike Galbraith <efault@gmx.de>
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
both callers have verified that fd is not greater than ->max_fds;
however, misprediction might end up with
tofree = fdt->fd[fd];
being speculatively executed. That's wrong for the same reasons
why it's wrong in close_fd()/file_close_fd_locked(); the same
solution applies - array_index_nospec(fd, fdt->max_fds) could differ
from fd only in case of speculative execution on mispredicted path.
Cc: stable@vger.kernel.org
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
|
|
While zeroing the upper 32 bits of an 8-byte getuser on 32-bit x86 was
fixed by commit 8c860ed825cb ("x86/uaccess: Fix missed zeroing of ia32 u64
get_user() range checking") it was broken again in commit 8a2462df1547
("x86/uaccess: Improve the 8-byte getuser() case").
This is because the register which holds the upper 32 bits (%ecx) is being
cleared _after_ the check_range, so if the range check fails, %ecx is never
cleared.
This can be reproduced with:
./tools/testing/kunit/kunit.py run --arch i386 usercopy
Instead, clear %ecx _before_ check_range in the 8-byte case. This
reintroduces a bit of the ugliness we were trying to avoid by adding
another #ifndef CONFIG_X86_64, but at least keeps check_range from needing
a separate bad_get_user_8 jump.
Fixes: 8a2462df1547 ("x86/uaccess: Improve the 8-byte getuser() case")
Signed-off-by: David Gow <davidgow@google.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Link: https://lore.kernel.org/all/20240731073031.4045579-1-davidgow@google.com
|
|
The RISC-V kernel already has checks to ensure that memory which would
lie outside of the linear mapping is not used. However those checks
use memory_limit, which is used to implement the mem= kernel command
line option (to limit the total amount of memory, not its address
range). When memory is made up of two or more non-contiguous memory
banks this check is incorrect.
Two changes are made here:
- add a call in setup_bootmem() to memblock_cap_memory_range() which
will cause any memory which falls outside the linear mapping to be
removed from the memory regions.
- remove the check in create_linear_mapping_page_table() which was
intended to remove memory which is outside the liner mapping based
on memory_limit, as it is no longer needed. Note a check for
mapping more memory than memory_limit (to implement mem=) is
unnecessary because of the existing call to
memblock_enforce_memory_limit().
This issue was seen when booting on a SV39 platform with two memory
banks:
0x00,80000000 1GiB
0x20,00000000 32GiB
This memory range is 158GiB from top to bottom, but the linear mapping
is limited to 128GiB, so the lower block of RAM will be mapped at
PAGE_OFFSET, and the upper block straddles the top of the linear
mapping.
This causes the following Oops:
[ 0.000000] Linux version 6.10.0-rc2-gd3b8dd5b51dd-dirty (stuart.menefy@codasip.com) (riscv64-codasip-linux-gcc (GCC) 13.2.0, GNU ld (GNU Binutils) 2.41.0.20231213) #20 SMP Sat Jun 22 11:34:22 BST 2024
[ 0.000000] memblock_add: [0x0000000080000000-0x00000000bfffffff] early_init_dt_add_memory_arch+0x4a/0x52
[ 0.000000] memblock_add: [0x0000002000000000-0x00000027ffffffff] early_init_dt_add_memory_arch+0x4a/0x52
...
[ 0.000000] memblock_alloc_try_nid: 23724 bytes align=0x8 nid=-1 from=0x0000000000000000 max_addr=0x0000000000000000 early_init_dt_alloc_memory_arch+0x1e/0x48
[ 0.000000] memblock_reserve: [0x00000027ffff5350-0x00000027ffffaffb] memblock_alloc_range_nid+0xb8/0x132
[ 0.000000] Unable to handle kernel paging request at virtual address fffffffe7fff5350
[ 0.000000] Oops [#1]
[ 0.000000] Modules linked in:
[ 0.000000] CPU: 0 PID: 0 Comm: swapper Not tainted 6.10.0-rc2-gd3b8dd5b51dd-dirty #20
[ 0.000000] Hardware name: codasip,a70x (DT)
[ 0.000000] epc : __memset+0x8c/0x104
[ 0.000000] ra : memblock_alloc_try_nid+0x74/0x84
[ 0.000000] epc : ffffffff805e88c8 ra : ffffffff806148f6 sp : ffffffff80e03d50
[ 0.000000] gp : ffffffff80ec4158 tp : ffffffff80e0bec0 t0 : fffffffe7fff52f8
[ 0.000000] t1 : 00000027ffffb000 t2 : 5f6b636f6c626d65 s0 : ffffffff80e03d90
[ 0.000000] s1 : 0000000000005cac a0 : fffffffe7fff5350 a1 : 0000000000000000
[ 0.000000] a2 : 0000000000005cac a3 : fffffffe7fffaff8 a4 : 000000000000002c
[ 0.000000] a5 : ffffffff805e88c8 a6 : 0000000000005cac a7 : 0000000000000030
[ 0.000000] s2 : fffffffe7fff5350 s3 : ffffffffffffffff s4 : 0000000000000000
[ 0.000000] s5 : ffffffff8062347e s6 : 0000000000000000 s7 : 0000000000000001
[ 0.000000] s8 : 0000000000002000 s9 : 00000000800226d0 s10: 0000000000000000
[ 0.000000] s11: 0000000000000000 t3 : ffffffff8080a928 t4 : ffffffff8080a928
[ 0.000000] t5 : ffffffff8080a928 t6 : ffffffff8080a940
[ 0.000000] status: 0000000200000100 badaddr: fffffffe7fff5350 cause: 000000000000000f
[ 0.000000] [<ffffffff805e88c8>] __memset+0x8c/0x104
[ 0.000000] [<ffffffff8062349c>] early_init_dt_alloc_memory_arch+0x1e/0x48
[ 0.000000] [<ffffffff8043e892>] __unflatten_device_tree+0x52/0x114
[ 0.000000] [<ffffffff8062441e>] unflatten_device_tree+0x9e/0xb8
[ 0.000000] [<ffffffff806046fe>] setup_arch+0xd4/0x5bc
[ 0.000000] [<ffffffff806007aa>] start_kernel+0x76/0x81a
[ 0.000000] Code: b823 02b2 bc23 02b2 b023 04b2 b423 04b2 b823 04b2 (bc23) 04b2
[ 0.000000] ---[ end trace 0000000000000000 ]---
[ 0.000000] Kernel panic - not syncing: Attempted to kill the idle task!
[ 0.000000] ---[ end Kernel panic - not syncing: Attempted to kill the idle task! ]---
The problem is that memblock (unaware that some physical memory cannot
be used) has allocated memory from the top of memory but which is
outside the linear mapping region.
Signed-off-by: Stuart Menefy <stuart.menefy@codasip.com>
Fixes: c99127c45248 ("riscv: Make sure the linear mapping does not use the kernel mapping")
Reviewed-by: David McKay <david.mckay@codasip.com>
Reviewed-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Link: https://lore.kernel.org/r/20240622114217.2158495-1-stuart.menefy@codasip.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
|
|
The `if (req_max_level)` test was meant ignore req_max_level if
PG_LEVEL_NONE was returned. Hence, this function should return
max_level instead of the ignored req_max_level.
This is only a latent issue for now, since guest_memfd does not
support large pages.
Signed-off-by: Ackerley Tng <ackerleytng@google.com>
Message-ID: <20240801173955.1975034-1-ackerleytng@google.com>
Fixes: f32fb32820b1 ("KVM: x86: Add hook for determining max NPT mapping level")
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
The sysfs "attention" file normally controls the Slot Control Attention
Indicator with 0 (off), 1 (on), 2 (blink) settings.
576243b3f9ea ("PCI: pciehp: Allow exclusive userspace control of
indicators") added pciehp_set_raw_indicator_status() to allow userspace to
directly control all four bits in both the Attention Indicator and the
Power Indicator fields via the "attention" file.
This is used on Intel VMD bridges so utilities like "ledmon" can use sysfs
"attention" to control up to 16 indicators for NVMe device RAID status.
abaaac4845a0 ("PCI: hotplug: Use FIELD_GET/PREP()") broke this by masking
the sysfs data with PCI_EXP_SLTCTL_AIC, which discards the upper two bits
intended for the Power Indicator Control field (PCI_EXP_SLTCTL_PIC).
For NVMe devices behind an Intel VMD, ledmon settings that use the
PCI_EXP_SLTCTL_PIC bits, i.e., ATTENTION_REBUILD (0x5), ATTENTION_LOCATE
(0x7), ATTENTION_FAILURE (0xD), ATTENTION_OFF (0xF), no longer worked
correctly.
Mask with PCI_EXP_SLTCTL_AIC | PCI_EXP_SLTCTL_PIC to retain both the
Attention Indicator and the Power Indicator bits.
Fixes: abaaac4845a0 ("PCI: hotplug: Use FIELD_GET/PREP()")
Link: https://lore.kernel.org/r/20240722141440.7210-1-blazej.kucman@intel.com
Signed-off-by: Blazej Kucman <blazej.kucman@intel.com>
[bhelgaas: commit log]
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: stable@vger.kernel.org # v6.7+
|
|
pci_intx() becomes managed if pcim_enable_device() has been called in
advance. Commit 25216afc9db5 ("PCI: Add managed pcim_intx()") changed this
behavior so that pci_intx() always leads to creation of a separate device
resource for itself, whereas earlier, a shared resource was used for all
PCI devres operations.
Unfortunately, pci_intx() seems to be used in some drivers' remove() paths;
in the managed case this causes a device resource to be created on driver
detach, which causes .probe() to fail if the driver is reloaded:
pci 0000:00:1f.2: Resources present before probing
Fix the regression by only redirecting pci_intx() to its managed twin
pcim_intx() if the pci_command changes.
Link: https://lore.kernel.org/r/20240725120729.59788-2-pstanner@redhat.com
Fixes: 25216afc9db5 ("PCI: Add managed pcim_intx()")
Reported-by: Damien Le Moal <dlemoal@kernel.org>
Closes: https://lore.kernel.org/all/b8f4ba97-84fc-4b7e-ba1a-99de2d9f0118@kernel.org/
Signed-off-by: Philipp Stanner <pstanner@redhat.com>
[bhelgaas: add error message to commit log]
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Tested-by: Damien Le Moal <dlemoal@kernel.org>
|
|
When using the shadow call stack sanitizer, all code must be compiled
with the -ffixed-x18 flag, but this flag is not currently being passed
to Rust. This results in crashes that are extremely difficult to debug.
To ensure that nobody else has to go through the same debugging session
that I had to, prevent configurations that enable both SHADOW_CALL_STACK
and RUST.
It is rather common for people to backport 724a75ac9542 ("arm64: rust:
Enable Rust support for AArch64"), so I recommend applying this fix all
the way back to 6.1.
Cc: stable@vger.kernel.org # 6.1 and later
Fixes: 724a75ac9542 ("arm64: rust: Enable Rust support for AArch64")
Signed-off-by: Alice Ryhl <aliceryhl@google.com>
Acked-by: Miguel Ojeda <ojeda@kernel.org>
Link: https://lore.kernel.org/r/20240729-shadow-call-stack-v4-1-2a664b082ea4@google.com
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
A number of Arm Ltd CPUs suffer from errata whereby an MSR to the SSBS
special-purpose register does not affect subsequent speculative
instructions, permitting speculative store bypassing for a window of
time.
We worked around this for a number of CPUs in commits:
* 7187bb7d0b5c7dfa ("arm64: errata: Add workaround for Arm errata 3194386 and 3312417")
* 75b3c43eab594bfb ("arm64: errata: Expand speculative SSBS workaround")
Since then, similar errata have been published for a number of other Arm
Ltd CPUs, for which the same mitigation is sufficient. This is described
in their respective Software Developer Errata Notice (SDEN) documents:
* Cortex-A76 (MP052) SDEN v31.0, erratum 3324349
https://developer.arm.com/documentation/SDEN-885749/3100/
* Cortex-A77 (MP074) SDEN v19.0, erratum 3324348
https://developer.arm.com/documentation/SDEN-1152370/1900/
* Cortex-A78 (MP102) SDEN v21.0, erratum 3324344
https://developer.arm.com/documentation/SDEN-1401784/2100/
* Cortex-A78C (MP138) SDEN v16.0, erratum 3324346
https://developer.arm.com/documentation/SDEN-1707916/1600/
* Cortex-A78C (MP154) SDEN v10.0, erratum 3324347
https://developer.arm.com/documentation/SDEN-2004089/1000/
* Cortex-A725 (MP190) SDEN v5.0, erratum 3456106
https://developer.arm.com/documentation/SDEN-2832921/0500/
* Cortex-X1 (MP077) SDEN v21.0, erratum 3324344
https://developer.arm.com/documentation/SDEN-1401782/2100/
* Cortex-X1C (MP136) SDEN v16.0, erratum 3324346
https://developer.arm.com/documentation/SDEN-1707914/1600/
* Neoverse-N1 (MP050) SDEN v32.0, erratum 3324349
https://developer.arm.com/documentation/SDEN-885747/3200/
* Neoverse-V1 (MP076) SDEN v19.0, erratum 3324341
https://developer.arm.com/documentation/SDEN-1401781/1900/
Note that due to the manner in which Arm develops IP and tracks errata,
some CPUs share a common erratum number and some CPUs have multiple
erratum numbers for the same HW issue.
On parts without SB, it is necessary to use ISB for the workaround. The
spec_bar() macro used in the mitigation will expand to a "DSB SY; ISB"
sequence in this case, which is sufficient on all affected parts.
Enable the existing mitigation by adding the relevant MIDRs to
erratum_spec_ssbs_list. The list is sorted alphanumerically (involving
moving Neoverse-V3 after Neoverse-V2) so that this is easy to audit and
potentially extend again in future. The Kconfig text is also updated to
clarify the set of affected parts and the mitigation.
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: James Morse <james.morse@arm.com>
Cc: Will Deacon <will@kernel.org>
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Acked-by: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20240801101803.1982459-4-mark.rutland@arm.com
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
Add cputype definitions for Cortex-A725. These will be used for errata
detection in subsequent patches.
These values can be found in the Cortex-A725 TRM:
https://developer.arm.com/documentation/107652/0001/
... in table A-247 ("MIDR_EL1 bit descriptions").
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: James Morse <james.morse@arm.com>
Cc: Will Deacon <will@kernel.org>
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Link: https://lore.kernel.org/r/20240801101803.1982459-3-mark.rutland@arm.com
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
Add cputype definitions for Cortex-X1C. These will be used for errata
detection in subsequent patches.
These values can be found in the Cortex-X1C TRM:
https://developer.arm.com/documentation/101968/0002/
... in section B2.107 ("MIDR_EL1, Main ID Register, EL1").
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: James Morse <james.morse@arm.com>
Cc: Will Deacon <will@kernel.org>
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Link: https://lore.kernel.org/r/20240801101803.1982459-2-mark.rutland@arm.com
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
Sometimes the hotplug cpu stalls at the arch_cpu_idle() for a while after
workqueue_online_cpu(). When cpu stalls at the idle loop, the reschedule
IPI is pending. However the enable bit is not enabled yet so the cpu stalls
at WFI until watchdog timeout. Therefore enable the IPI before the
workqueue_online_cpu() to fix the issue.
Fixes: 63c5484e7495 ("workqueue: Add multiple affinity scopes and interface to select them")
Signed-off-by: Nick Hu <nick.hu@sifive.com>
Reviewed-by: Anup Patel <anup@brainfault.org>
Link: https://lore.kernel.org/r/20240717031714.1946036-1-nick.hu@sifive.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
|
|
Handle VM_FAULT_SIGSEGV in the page fault path so that we correctly
kill the process and we don't BUG() the kernel.
Fixes: 07037db5d479 ("RISC-V: Paging and MMU")
Signed-off-by: Zhe Qiao <qiaozhe@iscas.ac.cn>
Reviewed-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Link: https://lore.kernel.org/r/20240731084547.85380-1-qiaozhe@iscas.ac.cn
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
|
|
It is required to check event type before checking event config.
Events with the different types can have the same config.
This check is missed for legacy mode code
For such perf usage:
sysctl -w kernel.perf_user_access=2
perf stat -e cycles,L1-dcache-loads --
driver will try to force both events to CYCLE counter.
This commit implements event type check before forcing
events on the special counters.
Signed-off-by: Shifrin Dmitry <dmitry.shifrin@syntacore.com>
Reviewed-by: Atish Patra <atishp@rivosinc.com>
Fixes: cc4c07c89aad ("drivers: perf: Implement perf event mmap support in the SBI backend")
Link: https://lore.kernel.org/r/20240729125858.630653-1-dmitry.shifrin@syntacore.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
|
|
This has a bunch of {read,write}q() calls, so it won't work on 32-bit
systems. I don't think there's any 32-bit StarFive systems, so for now
just require 64-bit.
Fixes: cabff60ca77d ("cache: Add StarFive StarLink cache management")
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Reviewed-by: Emil Renner Berthing <emil.renner.berthing@canonical.com>
Link: https://lore.kernel.org/r/20240722154519.25375-2-palmer@rivosinc.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
|
|
Currently, the RISC-V firmware JSON file has duplicate event name
"FW_SFENCE_VMA_RECEIVED". According to the RISC-V SBI PMU extension[1],
the event name should be "FW_SFENCE_VMA_ASID_SENT".
Before this patch:
$ perf list
firmware:
fw_access_load
[Load access trap event. Unit: cpu]
fw_access_store
[Store access trap event. Unit: cpu]
....
fw_set_timer
[Set timer event. Unit: cpu]
fw_sfence_vma_asid_received
[Received SFENCE.VMA with ASID request from other HART event. Unit: cpu]
fw_sfence_vma_received
[Sent SFENCE.VMA with ASID request to other HART event. Unit: cpu]
After this patch:
$ perf list
firmware:
fw_access_load
[Load access trap event. Unit: cpu]
fw_access_store
[Store access trap event. Unit: cpu]
.....
fw_set_timer
[Set timer event. Unit: cpu]
fw_sfence_vma_asid_received
[Received SFENCE.VMA with ASID request from other HART event. Unit: cpu]
fw_sfence_vma_asid_sent
[Sent SFENCE.VMA with ASID request to other HART event. Unit: cpu]
fw_sfence_vma_received
[Received SFENCE.VMA request from other HART event. Unit: cpu]
Link: https://github.com/riscv-non-isa/riscv-sbi-doc/blob/master/src/ext-pmu.adoc#event-firmware-events-type-15 [1]
Fixes: 8f0dcb4e7364 ("perf arch events: riscv sbi firmware std event files")
Fixes: c4f769d4093d ("perf vendor events riscv: add Sifive U74 JSON file")
Fixes: acbf6de674ef ("perf vendor events riscv: Add StarFive Dubhe-80 JSON file")
Fixes: 7340c6df49df ("perf vendor events riscv: add T-HEAD C9xx JSON file")
Fixes: f5102e31c209 ("riscv: andes: Support specifying symbolic firmware and hardware raw event")
Signed-off-by: Eric Lin <eric.lin@sifive.com>
Reviewed-by: Samuel Holland <samuel.holland@sifive.com>
Reviewed-by: Nikita Shubin <n.shubin@yadro.com>
Reviewed-by: Inochi Amaoto <inochiama@outlook.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Reviewed-by: Atish Patra <atishp@rivosinc.com>
Link: https://lore.kernel.org/r/20240719115018.27356-1-eric.lin@sifive.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
|
|
When alignment handling is delegated to the kernel, everything must be
word-aligned in purgatory, since the trap handler is then set to the
kexec one. Without the alignment, hitting the exception would
ultimately crash. On other occasions, the kernel's handler would take
care of exceptions.
This has been tested on a JH7110 SoC with oreboot and its SBI delegating
unaligned access exceptions and the kernel configured to handle them.
Fixes: 736e30af583fb ("RISC-V: Add purgatory")
Signed-off-by: Daniel Maslowski <cyrevolt@gmail.com>
Reviewed-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Link: https://lore.kernel.org/r/20240719170437.247457-1-cyrevolt@gmail.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
|
|
Compile-testing the crypto/caam driver on alpha showed a pre-existing
problem on alpha with iowrite64be() missing:
ERROR: modpost: "iowrite64be" [drivers/crypto/caam/caam_jr.ko] undefined!
The prototypes were added a while ago when we started using asm-generic/io.h,
but the implementation was still missing. At some point the ioread64/iowrite64
helpers were added, but the big-endian versions are still missing, and
the generic version (using readq/writeq) is would not work here.
Change it to wrap ioread64()/iowrite64() instead.
Fixes: beba3771d9e0 ("crypto: caam: Make CRYPTO_DEV_FSL_CAAM dependent of COMPILE_TEST")
Fixes: e19d4ebc536d ("alpha: add full ioread64/iowrite64 implementation")
Fixes: 7e772dad9913 ("alpha: Use generic <asm-generic/io.h>")
Closes: https://lore.kernel.org/all/CAHk-=wgEyzSxTs467NDOVfBSzWvUS6ztcwhiy=M3xog==KBmTw@mail.gmail.com/
Tested-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
|
|
While x86_64 has PMD aligned text sections, i386 does not have this
luxery. Notably ALIGN_ENTRY_TEXT_END is empty and _etext has PAGE
alignment.
This means that text on i386 can be page granular at the tail end,
which in turn means that the PTI text clones should consistently
account for this.
Make pti_clone_entry_text() consistent with pti_clone_kernel_text().
Fixes: 16a3fe634f6a ("x86/mm/pti: Clone kernel-image on PTE level for 32 bit")
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
|
|
Guenter reported dodgy crashes on an i386-nosmp build using GCC-11
that had the form of endless traps until entry stack exhaust and then
#DF from the stack guard.
It turned out that pti_clone_pgtable() had alignment assumptions on
the start address, notably it hard assumes start is PMD aligned. This
is true on x86_64, but very much not true on i386.
These assumptions can cause the end condition to malfunction, leading
to a 'short' clone. Guess what happens when the user mapping has a
short copy of the entry text?
Use the correct increment form for addr to avoid alignment
assumptions.
Fixes: 16a3fe634f6a ("x86/mm/pti: Clone kernel-image on PTE level for 32 bit")
Reported-by: Guenter Roeck <linux@roeck-us.net>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Suggested-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20240731163105.GG33588@noisy.programming.kicks-ass.net
|
|
If a client sends out a cap update dropping caps with the prior 'seq'
just before an incoming cap revoke request, then the client may drop
the revoke because it believes it's already released the requested
capabilities.
This causes the MDS to wait indefinitely for the client to respond
to the revoke. It's therefore always a good idea to ack the cap
revoke request with the bumped up 'seq'.
Currently if the cap->issued equals to the newcaps the check_caps()
will do nothing, we should force flush the caps.
Cc: stable@vger.kernel.org
Link: https://tracker.ceph.com/issues/61782
Signed-off-by: Xiubo Li <xiubli@redhat.com>
Reviewed-by: Venky Shankar <vshankar@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
|
|
When a subflow receives and discards duplicate data, the mptcp
stack assumes that the consumed offset inside the current skb is
zero.
With multiple subflows receiving data simultaneously such assertion
does not held true. As a result the subflow-level copied_seq will
be incorrectly increased and later on the same subflow will observe
a bad mapping, leading to subflow reset.
Address the issue taking into account the skb consumed offset in
mptcp_subflow_discard_data().
Fixes: 04e4cd4f7ca4 ("mptcp: cleanup mptcp_subflow_discard_data()")
Cc: stable@vger.kernel.org
Link: https://github.com/multipath-tcp/mptcp_net-next/issues/501
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Mat Martineau <martineau@kernel.org>
Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
liujinbao1
Wenjie Cheng
Chao Yu
Julian Sun
Liao Yuanhong
Nikita Zhandarovich
Linus Torvalds
Tetsuo Handa
Rob Herring (Arm)
Paul E. McKenney
Steve French
Paulo Alcantara
David Howells
Will Deacon
Arnd Bergmann
Rafael J. Wysocki
Dave Airlie
Al Viro
David Gow
Stuart Menefy
Ackerley Tng
Blazej Kucman
Philipp Stanner
Alice Ryhl
Mark Rutland
Nick Hu
Zhe Qiao
Shifrin Dmitry
Palmer Dabbelt
Eric Lin
Daniel Maslowski
Peter Zijlstra
Xiubo Li
Paolo Abeni