Currently there is a bound check missing in the dbAdjTree while
accessing the dmt_stree. To add the required check added the bool is_ctl
which is required to determine the size as suggest in the following
commit.
https://lore.kernel.org/linux-kernel-mentees/f9475918-2186-49b8-b801-6f0f9e75f4fa@oracle.com/
The issue is caused when the value of fsi becomes less than -1.
The check to break the loop when fsi value becomes -1 is present
but syzbot was able to produce value less than -1 which cause the error.
This patch simply add the change for the values less than 0.
The patch is tested via syzbot.
Reported-and-tested-by: syzbot+d4b1df2e9d4ded6488ec@syzkaller.appspotmail.com Link: https://syzkaller.appspot.com/bug?extid=d4b1df2e9d4ded6488ec Signed-off-by: Osama Muhammad <osmtendev@gmail.com> Signed-off-by: Dave Kleikamp <dave.kleikamp@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
UBSAN: array-index-out-of-bounds in fs/jfs/jfs_dmap.c:2867:6
index 196694 is out of range for type 's8[1365]' (aka 'signed char[1365]')
CPU: 1 PID: 109 Comm: jfsCommit Not tainted 6.6.0-rc3-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 08/04/2023
Call Trace:
<TASK>
__dump_stack lib/dump_stack.c:88 [inline]
dump_stack_lvl+0x1e7/0x2d0 lib/dump_stack.c:106
ubsan_epilogue lib/ubsan.c:217 [inline]
__ubsan_handle_out_of_bounds+0x11c/0x150 lib/ubsan.c:348
dbAdjTree+0x474/0x4f0 fs/jfs/jfs_dmap.c:2867
dbJoin+0x210/0x2d0 fs/jfs/jfs_dmap.c:2834
dbFreeBits+0x4eb/0xda0 fs/jfs/jfs_dmap.c:2331
dbFreeDmap fs/jfs/jfs_dmap.c:2080 [inline]
dbFree+0x343/0x650 fs/jfs/jfs_dmap.c:402
txFreeMap+0x798/0xd50 fs/jfs/jfs_txnmgr.c:2534
txUpdateMap+0x342/0x9e0
txLazyCommit fs/jfs/jfs_txnmgr.c:2664 [inline]
jfs_lazycommit+0x47a/0xb70 fs/jfs/jfs_txnmgr.c:2732
kthread+0x2d3/0x370 kernel/kthread.c:388
ret_from_fork+0x48/0x80 arch/x86/kernel/process.c:147
ret_from_fork_asm+0x11/0x20 arch/x86/entry/entry_64.S:304
</TASK>
================================================================================
Kernel panic - not syncing: UBSAN: panic_on_warn set ...
CPU: 1 PID: 109 Comm: jfsCommit Not tainted 6.6.0-rc3-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 08/04/2023
Call Trace:
<TASK>
__dump_stack lib/dump_stack.c:88 [inline]
dump_stack_lvl+0x1e7/0x2d0 lib/dump_stack.c:106
panic+0x30f/0x770 kernel/panic.c:340
check_panic_on_warn+0x82/0xa0 kernel/panic.c:236
ubsan_epilogue lib/ubsan.c:223 [inline]
__ubsan_handle_out_of_bounds+0x13c/0x150 lib/ubsan.c:348
dbAdjTree+0x474/0x4f0 fs/jfs/jfs_dmap.c:2867
dbJoin+0x210/0x2d0 fs/jfs/jfs_dmap.c:2834
dbFreeBits+0x4eb/0xda0 fs/jfs/jfs_dmap.c:2331
dbFreeDmap fs/jfs/jfs_dmap.c:2080 [inline]
dbFree+0x343/0x650 fs/jfs/jfs_dmap.c:402
txFreeMap+0x798/0xd50 fs/jfs/jfs_txnmgr.c:2534
txUpdateMap+0x342/0x9e0
txLazyCommit fs/jfs/jfs_txnmgr.c:2664 [inline]
jfs_lazycommit+0x47a/0xb70 fs/jfs/jfs_txnmgr.c:2732
kthread+0x2d3/0x370 kernel/kthread.c:388
ret_from_fork+0x48/0x80 arch/x86/kernel/process.c:147
ret_from_fork_asm+0x11/0x20 arch/x86/entry/entry_64.S:304
</TASK>
Kernel Offset: disabled
Rebooting in 86400 seconds..
The issue is caused when the value of lp becomes greater than
CTLTREESIZE which is the max size of stree. Adding a simple check
solves this issue.
Dave:
As the function returns a void, good error handling
would require a more intrusive code reorganization, so I modified
Osama's patch at use WARN_ON_ONCE for lack of a cleaner option.
The patch is tested via syzbot.
Reported-by: syzbot+39ba34a099ac2e9bd3cb@syzkaller.appspotmail.com Link: https://syzkaller.appspot.com/bug?extid=39ba34a099ac2e9bd3cb Signed-off-by: Osama Muhammad <osmtendev@gmail.com> Signed-off-by: Dave Kleikamp <dave.kleikamp@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
There are 3 synchronization issues with thermal zone suspend-resume
during system-wide transitions:
1. The resume code runs in a PM notifier which is invoked after user
space has been thawed, so it can run concurrently with user space
which can trigger a thermal zone device removal. If that happens,
the thermal zone resume code may use a stale pointer to the next
list element and crash, because it does not hold thermal_list_lock
while walking thermal_tz_list.
2. The thermal zone resume code calls thermal_zone_device_init()
outside the zone lock, so user space or an update triggered by
the platform firmware may see an inconsistent state of a
thermal zone leading to unexpected behavior.
3. Clearing the in_suspend global variable in thermal_pm_notify()
allows __thermal_zone_device_update() to continue for all thermal
zones and it may as well run before the thermal_tz_list walk (or
at any point during the list walk for that matter) and attempt to
operate on a thermal zone that has not been resumed yet. It may
also race destructively with thermal_zone_device_init().
To address these issues, add thermal_list_lock locking to
thermal_pm_notify(), especially arount the thermal_tz_list,
make it call thermal_zone_device_init() back-to-back with
__thermal_zone_device_update() under the zone lock and replace
in_suspend with per-zone bool "suspend" indicators set and unset
under the given zone's lock.
There are two major types of uncorrected recoverable (UCR) errors :
- Synchronous error: The error is detected and raised at the point of
the consumption in the execution flow, e.g. when a CPU tries to
access a poisoned cache line. The CPU will take a synchronous error
exception such as Synchronous External Abort (SEA) on Arm64 and
Machine Check Exception (MCE) on X86. OS requires to take action (for
example, offline failure page/kill failure thread) to recover this
uncorrectable error.
- Asynchronous error: The error is detected out of processor execution
context, e.g. when an error is detected by a background scrubber.
Some data in the memory are corrupted. But the data have not been
consumed. OS is optional to take action to recover this uncorrectable
error.
When APEI firmware first is enabled, a platform may describe one error
source for the handling of synchronous errors (e.g. MCE or SEA notification
), or for handling asynchronous errors (e.g. SCI or External Interrupt
notification). In other words, we can distinguish synchronous errors by
APEI notification. For synchronous errors, kernel will kill the current
process which accessing the poisoned page by sending SIGBUS with
BUS_MCEERR_AR. In addition, for asynchronous errors, kernel will notify the
process who owns the poisoned page by sending SIGBUS with BUS_MCEERR_AO in
early kill mode. However, the GHES driver always sets mf_flags to 0 so that
all synchronous errors are handled as asynchronous errors in memory failure.
To this end, set memory failure flags as MF_ACTION_REQUIRED on synchronous
events.
Signed-off-by: Shuai Xue <xueshuai@linux.alibaba.com> Tested-by: Ma Wupeng <mawupeng1@huawei.com> Reviewed-by: Kefeng Wang <wangkefeng.wang@huawei.com> Reviewed-by: Xiaofei Tan <tanxiaofei@huawei.com> Reviewed-by: Baolin Wang <baolin.wang@linux.alibaba.com> Reviewed-by: James Morse <james.morse@arm.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
There is a chance if a frequent switch of the governor
done in a loop result in timer list corruption where
timer cancel being done from two place one from
cancel_delayed_work_sync() and followed by expire_timers()
can be seen from the traces[1].
while true
do
echo "simple_ondemand" > /sys/class/devfreq/1d84000.ufshc/governor
echo "performance" > /sys/class/devfreq/1d84000.ufshc/governor
done
It looks to be issue with devfreq driver where
device_monitor_[start/stop] need to synchronized so that
delayed work should get corrupted while it is either
being queued or running or being cancelled.
Let's use polling flag and devfreq lock to synchronize the
queueing the timer instance twice and work data being
corrupted.
If we run parameterized test that uses test->priv to prepare some
custom data, then value of test->priv will leak to the next param
iteration and may be unexpected. This could be easily seen if
we promote example_priv_test to parameterized test as then only
first test iteration will be successful:
$ ./tools/testing/kunit/kunit.py run \
--kunitconfig ./lib/kunit/.kunitconfig *.example_priv*
[ ] Starting KUnit Kernel (1/1)...
[ ] ============================================================
[ ] =================== example (1 subtest) ====================
[ ] ==================== example_priv_test ====================
[ ] [PASSED] example value 3
[ ] # example_priv_test: initializing
[ ] # example_priv_test: ASSERTION FAILED at lib/kunit/kunit-example-test.c:230
[ ] Expected test->priv == ((void *)0), but
[ ] test->priv == 0000000060dfe290
[ ] ((void *)0) == 0000000000000000
[ ] # example_priv_test: cleaning up
[ ] [FAILED] example value 2
[ ] # example_priv_test: initializing
[ ] # example_priv_test: ASSERTION FAILED at lib/kunit/kunit-example-test.c:230
[ ] Expected test->priv == ((void *)0), but
[ ] test->priv == 0000000060dfe290
[ ] ((void *)0) == 0000000000000000
[ ] # example_priv_test: cleaning up
[ ] [FAILED] example value 1
[ ] # example_priv_test: initializing
[ ] # example_priv_test: ASSERTION FAILED at lib/kunit/kunit-example-test.c:230
[ ] Expected test->priv == ((void *)0), but
[ ] test->priv == 0000000060dfe290
[ ] ((void *)0) == 0000000000000000
[ ] # example_priv_test: cleaning up
[ ] [FAILED] example value 0
[ ] # example_priv_test: initializing
[ ] # example_priv_test: cleaning up
[ ] # example_priv_test: pass:1 fail:3 skip:0 total:4
[ ] ================ [FAILED] example_priv_test ================
[ ] # example: initializing suite
[ ] # module: kunit_example_test
[ ] # example: exiting suite
[ ] # Totals: pass:1 fail:3 skip:0 total:4
[ ] ===================== [FAILED] example =====================
Fix that by resetting test->priv after each param iteration, in
similar way what we did for the test->status.
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com> Cc: David Gow <davidgow@google.com> Cc: Rae Moar <rmoar@google.com> Reviewed-by: David Gow <davidgow@google.com> Signed-off-by: Shuah Khan <skhan@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Add parsing of attributes as diagnostic data. Fixes issue with test plan
being parsed incorrectly as diagnostic data when located after
suite-level attributes.
Note that if there does not exist a test plan line, the diagnostic lines
between the suite header and the first result will be saved in the suite
log rather than the first test case log.
Signed-off-by: Rae Moar <rmoar@google.com> Reviewed-by: David Gow <davidgow@google.com> Signed-off-by: Shuah Khan <skhan@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The for loop does not iterate over the last element of the node_to_pxm_map
array. This could lead to a conflict between the final fake_pxm value and
the existing pxm values. That is, the final fake_pxm value can not be
guaranteed to be an unused pxm value.
While at it, fix up white space in slit_valid().
Signed-off-by: Yuntao Wang <ytcoode@gmail.com>
[ rjw: Changelog edits ] Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Center-align all possible status reports.
Before OK and FAIL were center-aligned in relation to each other but
SKIPPED and FAILED would be left-aligned.
The gcc plugin -fanalyzer [1] tries to detect various
patterns of incorrect behaviour. The tool reports:
drivers/acpi/acpi_extlog.c: In function ‘extlog_exit’:
drivers/acpi/acpi_extlog.c:307:12: warning: check of ‘extlog_l1_addr’ for NULL after already dereferencing it [-Wanalyzer-deref-before-check]
|
| 306 | ((struct extlog_l1_head *)extlog_l1_addr)->flags &= ~FLAG_OS_OPTIN;
| | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~
| | |
| | (1) pointer ‘extlog_l1_addr’ is dereferenced here
| 307 | if (extlog_l1_addr)
| | ~
| | |
| | (2) pointer ‘extlog_l1_addr’ is checked for NULL here but it was already dereferenced at (1)
|
Fix the NULL pointer dereference check in extlog_exit().
When compiling with gcc version 14.0.0 20231126 (experimental)
and CONFIG_FORTIFY_SOURCE=y, I've noticed the following:
In file included from ./include/linux/string.h:295,
from ./include/linux/bitmap.h:12,
from ./include/linux/cpumask.h:12,
from ./arch/x86/include/asm/paravirt.h:17,
from ./arch/x86/include/asm/cpuid.h:62,
from ./arch/x86/include/asm/processor.h:19,
from ./arch/x86/include/asm/cpufeature.h:5,
from ./arch/x86/include/asm/thread_info.h:53,
from ./include/linux/thread_info.h:60,
from ./arch/x86/include/asm/preempt.h:9,
from ./include/linux/preempt.h:79,
from ./include/linux/spinlock.h:56,
from ./include/linux/mmzone.h:8,
from ./include/linux/gfp.h:7,
from ./include/linux/slab.h:16,
from ./include/linux/resource_ext.h:11,
from ./include/linux/acpi.h:13,
from drivers/pnp/pnpacpi/rsparser.c:11:
In function 'fortify_memcpy_chk',
inlined from 'pnpacpi_parse_allocated_vendor' at drivers/pnp/pnpacpi/rsparser.c:158:3,
inlined from 'pnpacpi_allocated_resource' at drivers/pnp/pnpacpi/rsparser.c:249:3:
./include/linux/fortify-string.h:588:25: warning: call to '__read_overflow2_field'
declared with attribute warning: detected read beyond size of field (2nd parameter);
maybe use struct_group()? [-Wattribute-warning]
588 | __read_overflow2_field(q_size_field, size);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
According to the comments in include/linux/fortify-string.h, 'memcpy()',
'memmove()' and 'memset()' must not be used beyond individual struct
members to ensure that the compiler can enforce protection against
buffer overflows, and, IIUC, this also applies to partial copies from
the particular member ('vendor->byte_data' in this case). So it should
be better (and safer) to do both copies at once (and 'byte_data' of
'struct acpi_resource_vendor_typed' seems to be a good candidate for
'__counted_by(byte_length)' as well).
Signed-off-by: Dmitry Antipov <dmantipov@yandex.ru> Reviewed-by: Kees Cook <keescook@chromium.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The Colorful X15 AT 23 ACPI video-bus device report spurious
ACPI_VIDEO_NOTIFY_CYCLE events resulting in spurious KEY_SWITCHVIDEOMODE
events being reported to userspace (and causing trouble there) when
an external screen plugged in.
Add a quirk setting the report_key_events mask to
REPORT_BRIGHTNESS_KEY_EVENTS so that the ACPI_VIDEO_NOTIFY_CYCLE
events will be ignored, while still reporting brightness up/down
hotkey-presses to userspace normally.
Signed-off-by: Yuluo Qiu <qyl27@outlook.com> Co-developed-by: Celeste Liu <CoelacanthusHex@gmail.com> Signed-off-by: Celeste Liu <CoelacanthusHex@gmail.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The original intention of acpi_parse_entries_array() is to return the
number of all matching entries on success. This number may be greater than
the value of the max_entries parameter. When this happens, the function
will output a warning message, indicating that `count - max_entries`
matching entries remain unprocessed and have been ignored.
However, commit 4ceacd02f5a1 ("ACPI / table: Always count matched and
successfully parsed entries") changed this logic to return the number of
entries successfully processed by the handler. In this case, when the
max_entries parameter is not zero, the number of entries successfully
processed can never be greater than the value of max_entries. In other
words, the expression `count > max_entries` will always evaluate to false.
This means that the logic in the final if statement will never be executed.
Commit 99b0efd7c886 ("ACPI / tables: do not report the number of entries
ignored by acpi_parse_entries()") mentioned this issue, but it tried to fix
it by removing part of the warning message. This is meaningless because the
pr_warn statement will never be executed in the first place.
Commit 8726d4f44150 ("ACPI / tables: fix acpi_parse_entries_array() so it
traverses all subtables") introduced an errs variable, which is intended to
make acpi_parse_entries_array() always traverse all of the subtables,
calling as many of the callbacks as possible. However, it seems that the
commit does not achieve this goal. For example, when a handler returns an
error, none of the handlers will be called again in the subsequent
iterations. This result appears to be no different from before the change.
This patch corrects and cleans up the logic of acpi_parse_entries_array(),
making it return the number of all matching entries, rather than the number
of entries successfully processed by handlers. Additionally, if an error
occurs when executing a handler, the function will return -EINVAL immediately.
This patch should not affect existing users of acpi_parse_entries_array().
Signed-off-by: Yuntao Wang <ytcoode@gmail.com> Reviewed-by: Dave Jiang <dave.jiang@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
When auditd_set sets the auditd_conn pointer, audit messages can
immediately be put on the socket by other kernel threads. If the backlog
is large or the rate is high, this can immediately fill the socket
buffer. If the audit daemon requested an ACK for this operation, a full
socket buffer causes the ACK to get dropped, also setting ENOBUFS on the
socket.
To avoid this race and ensure ACKs get through, fast-track the ACK in
this specific case to ensure it is sent before auditd_conn is set.
Signed-off-by: Chris Riches <chris.riches@nutanix.com>
[PM: fix some tab vs space damage] Signed-off-by: Paul Moore <paul@paul-moore.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The use_count of a regulator should only be incremented when the
enable_count changes from 0 to 1. Similarly, the use_count should
only be decremented when the enable_count changes from 1 to 0.
In the previous implementation, use_count was sometimes decremented
to 0 when some consumer called unbalanced disable,
leading to unexpected disable even the regulator is enabled by
other consumers. With this change, the use_count accurately reflects
the number of users which the regulator is enabled.
This should make things more robust in the case where a consumer does
leak references.
When a CPU is taken offline, the contribution of its cfs_rqs to task_groups'
load may remain and will negatively impact the calculation of the share of
the online CPUs.
To fix this bug, clear the contribution of an offlining CPU to task groups'
load and skip its contribution while it is inactive.
Here's the reproducer of the anomaly, by Imran Khan:
"So far I have encountered only one rather lengthy way of reproducing this issue,
which is as follows:
1. Take a KVM guest (booted with 4 CPUs and can be scaled up to 124 CPUs) and
create 2 custom cgroups: /sys/fs/cgroup/cpu/test_group_1 and /sys/fs/cgroup/
cpu/test_group_2
2. Assign a CPU intensive workload to each of these cgroups and start the
workload.
For my tests I am using following app:
int main(int argc, char *argv[])
{
unsigned long count, i, val;
if (argc != 2) {
printf("usage: ./a.out <number of random nums to generate> \n");
return 0;
}
count = strtoul(argv[1], NULL, 10);
printf("Generating %lu random numbers \n", count);
for (i = 0; i < count; i++) {
val = rand();
val = val % 2;
//usleep(1);
}
printf("Generated %lu random numbers \n", count);
return 0;
}
Also since the system is booted with 4 CPUs, in order to completely load the
system I am also launching 4 instances of same test app under:
/sys/fs/cgroup/cpu/
3. We can see that both of the cgroups get similar CPU time:
4. Now move systemd-udevd to one of these test groups, say test_group_1, and
perform scale up to 124 CPUs followed by scale down back to 4 CPUs from the
host side.
5. Run the same workload i.e 4 instances of CPU hogger under /sys/fs/cgroup/cpu
and one instance of CPU hogger each in /sys/fs/cgroup/cpu/test_group_1 and
/sys/fs/cgroup/test_group_2.
It can be seen that test_group_1 (the one where systemd-udevd was moved) is getting
much less CPU time than the test_group_2, even though at this point of time both of
these groups have only CPU hogger running:
From sched_debug data it can be seen that in bad case the load.weight of per-CPU
sched entities corresponding to test_group_1 has reduced significantly and
also load_avg of test_group_1 remains much higher than that of test_group_2,
even though systemd-udevd stopped running long time back and at this point of
time both cgroups just have the CPU hogger app as running entity."
[ mingo: Added details from the original discussion, plus minor edits to the patch. ]
After release of the hashbucket lock the tracking object can be modified or
freed by a concurrent thread. Using it in such a case is error prone, even
for printing the object state:
1. T1 tries to deactivate destroyed object, debugobjects detects it,
hash bucket lock is released.
2. T2 preempts T1 and frees the tracking object.
3. The freed tracking object is allocated and initialized for a
different to be tracked kernel object.
4. T1 resumes and reports error for wrong kernel object.
Create a local copy of the tracking object before releasing the hash bucket
lock and use the local copy for reporting and fixups to prevent this.
Since commit fc137c0ddab2 ("sched/numa: enhance vma scanning logic")
NUMA Balancing allows updating PTEs to trap NUMA hinting faults if the
task had previously accessed VMA. However unconditional scan of VMAs are
allowed during initial phase of VMA creation until process's
mm numa_scan_seq reaches 2 even though current task had not accessed VMA.
Rationale:
- Without initial scan subsequent PTE update may never happen.
- Give fair opportunity to all the VMAs to be scanned and subsequently
understand the access pattern of all the VMAs.
But it has a corner case where, if a VMA is created after some time,
process's mm numa_scan_seq could be already greater than 2.
For e.g., values of mm numa_scan_seq when VMAs are created by running
mmtest autonuma benchmark briefly looks like:
start_seq=0 : 459
start_seq=2 : 138
start_seq=3 : 144
start_seq=4 : 8
start_seq=8 : 1
start_seq=9 : 1
This results in no unconditional PTE updates for those VMAs created after
some time.
Fix:
- Note down the initial value of mm numa_scan_seq in per VMA start_seq.
- Allow unconditional scan till start_seq + 2.
Result:
SUT: AMD EPYC Milan with 2 NUMA nodes 256 cpus.
base kernel: upstream 6.6-rc6 with Mels patches [1] applied.
Memory errors don't happen very often, especially fatal ones. However,
in large-scale scenarios such as data centers, that probability
increases with the amount of machines present.
When a fatal machine check happens, mce_panic() is called based on the
severity grading of that error. The page containing the error is not
marked as poison.
However, when kexec is enabled, tools like makedumpfile understand when
pages are marked as poison and do not touch them so as not to cause
a fatal machine check exception again while dumping the previous
kernel's memory.
Therefore, mark the page containing the error as poisoned so that the
kexec'ed kernel can avoid accessing the page.
[ bp: Rewrite commit message and comment. ]
Co-developed-by: Youquan Song <youquan.song@intel.com> Signed-off-by: Youquan Song <youquan.song@intel.com> Signed-off-by: Zhiquan Li <zhiquan1.li@intel.com> Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de> Reviewed-by: Naoya Horiguchi <naoya.horiguchi@nec.com> Link: https://lore.kernel.org/r/20231014051754.3759099-1-zhiquan1.li@intel.com Signed-off-by: Sasha Levin <sashal@kernel.org>
Some of the fp/vmx code in sstep.c assume a certain maximum size for the
instructions being emulated. The size of those operations however is
determined separately in analyse_instr().
Add a check to validate the assumption on the maximum size of the
operations, so as to prevent any unintended kernel stack corruption.
Signed-off-by: Naveen N Rao <naveen@kernel.org> Reviewed-by: Gustavo A. R. Silva <gustavoars@kernel.org> Build-tested-by: Gustavo A. R. Silva <gustavoars@kernel.org> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://msgid.link/20231123071705.397625-1-naveen@kernel.org Signed-off-by: Sasha Levin <sashal@kernel.org>
The linux-next build of powerpc64 allnoconfig fails with:
arch/powerpc/mm/book3s64/pgtable.c:557:5: error: no previous prototype for 'pmd_move_must_withdraw'
557 | int pmd_move_must_withdraw(struct spinlock *new_pmd_ptl,
| ^~~~~~~~~~~~~~~~~~~~~~
Caused by commit:
c6345dfa6e3e ("Makefile.extrawarn: turn on missing-prototypes globally")
Fix it by moving the function definition under
CONFIG_TRANSPARENT_HUGEPAGE like the prototype. The function is only
called when CONFIG_TRANSPARENT_HUGEPAGE=y.
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
[mpe: Flesh out change log from linux-next patch] Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://msgid.link/20231127132809.45c2b398@canb.auug.org.au Signed-off-by: Sasha Levin <sashal@kernel.org>
When there are two racing NMIs on x86, the first NMI invokes NMI handler and
the 2nd NMI is latched until IRET is executed.
If panic on NMI and panic kexec are enabled, the first NMI triggers
panic and starts booting the next kernel via kexec. Note that the 2nd
NMI is still latched. During the early boot of the next kernel, once
an IRET is executed as a result of a page fault, then the 2nd NMI is
unlatched and invokes the NMI handler.
However, NMI handler is not set up at the early stage of boot, which
results in a boot failure.
Avoid such problems by setting up a NOP handler for early NMIs.
[ mingo: Refined the changelog. ]
Signed-off-by: Jun'ichi Nomura <junichi.nomura@nec.com> Signed-off-by: Derek Barbosa <debarbos@redhat.com> Signed-off-by: Ingo Molnar <mingo@kernel.org> Cc: Kees Cook <keescook@chromium.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@kernel.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
arch/powerpc/mm/book3s64/pgtable.c:275:15: error: no previous prototype for ‘create_section_mapping’ [-Werror=missing-prototypes]
275 | int __meminit create_section_mapping(unsigned long start, unsigned long end,
| ^~~~~~~~~~~~~~~~~~~~~~
That happens because the prototype for create_section_mapping() is in
asm/mmzone.h, but asm/mmzone.h is only included by linux/mmzone.h
when CONFIG_NUMA=y.
In fact the prototype is only needed by arch/powerpc/mm code, so move
the prototype into arch/powerpc/mm/mmu_decl.h, which also fixes the
build error.
With NUMA=n and FA_DUMP=y or PRESERVE_FA_DUMP=y the build fails with:
arch/powerpc/kernel/fadump.c:1739:22: error: no previous prototype for ‘arch_reserved_kernel_pages’ [-Werror=missing-prototypes]
1739 | unsigned long __init arch_reserved_kernel_pages(void)
| ^~~~~~~~~~~~~~~~~~~~~~~~~~
The prototype for arch_reserved_kernel_pages() is in include/linux/mm.h,
but it's guarded by __HAVE_ARCH_RESERVED_KERNEL_PAGES. The powerpc
headers define __HAVE_ARCH_RESERVED_KERNEL_PAGES in asm/mmzone.h, which
is not included into the generic headers when NUMA=n.
Move the definition of __HAVE_ARCH_RESERVED_KERNEL_PAGES into asm/mmu.h
which is included regardless of NUMA=n.
Additionally the ifdef around __HAVE_ARCH_RESERVED_KERNEL_PAGES needs to
also check for CONFIG_PRESERVE_FA_DUMP.
arch/powerpc/kernel/traps.c:1442:5: error: no previous prototype for ‘is_valid_bugaddr’ [-Werror=missing-prototypes]
1442 | int is_valid_bugaddr(unsigned long addr)
| ^~~~~~~~~~~~~~~~
The prototype is only defined, and the function is only needed, when
CONFIG_GENERIC_BUG=y, so move the implementation under that.
The SW_INCR event is somewhat unusual, and depends on the specific HW
counter that it is programmed into. When programmed into PMEVCNTR<n>,
SW_INCR will count any writes to PMSWINC_EL0 with bit n set, ignoring
writes to SW_INCR with bit n clear.
Event rotation means that there's no fixed relationship between
perf_events and HW counters, so this isn't all that useful.
Further, we program PMUSERENR.{SW,EN}=={0,0}, which causes EL0 writes to
PMSWINC_EL0 to be trapped and handled as UNDEFINED, resulting in a
SIGILL to userspace.
Given that, it's not a good idea to expose SW_INCR in sysfs. Hide it as
we did for CHAIN back in commit:
4ba2578fa7b55701 ("arm64: perf: don't expose CHAIN event in sysfs")
In current code, init_irq_stacks() will call cpu_to_node().
The cpu_to_node() depends on percpu "numa_node" which is initialized in:
arch_call_rest_init() --> rest_init() -- kernel_init()
--> kernel_init_freeable() --> smp_prepare_cpus()
But init_irq_stacks() is called in init_IRQ() which is before
arch_call_rest_init().
So in init_irq_stacks(), the cpu_to_node() does not work, it
always return 0. In NUMA, it makes the node 1 cpu accesses the IRQ stack which
is in the node 0.
This patch fixes it by:
1.) export the early_cpu_to_node(), and use it in the init_irq_stacks().
2.) change init_irq_stacks() to __init function.
kasprintf() returns a pointer to dynamically allocated memory
which can be NULL upon failure. Ensure the allocation was successful
by checking the pointer validity.
__put_unaligned_be24() and friends use implicit casts to convert
larger-sized data to bytes, which trips sparse truncation warnings when
the argument is a constant:
CC [M] drivers/input/touchscreen/hynitron_cstxxx.o
CHECK drivers/input/touchscreen/hynitron_cstxxx.c
drivers/input/touchscreen/hynitron_cstxxx.c: note: in included file (through arch/x86/include/generated/asm/unaligned.h):
include/asm-generic/unaligned.h:119:16: warning: cast truncates bits from constant value (aa01a0 becomes a0)
include/asm-generic/unaligned.h:120:20: warning: cast truncates bits from constant value (aa01 becomes 1)
include/asm-generic/unaligned.h:119:16: warning: cast truncates bits from constant value (ab00d0 becomes d0)
include/asm-generic/unaligned.h:120:20: warning: cast truncates bits from constant value (ab00 becomes 0)
To avoid this let's mask off upper bits explicitly, the resulting code
should be exactly the same, but it will keep sparse happy.
Presently ia32 registers stored in ptregs are unconditionally cast to
unsigned int by the ia32 stub. They are then cast to long when passed to
__se_sys*, but will not be sign extended.
This takes the sign of the syscall argument into account in the ia32
stub. It still casts to unsigned int to avoid implementation specific
behavior. However then casts to int or unsigned int as necessary. So that
the following cast to long sign extends the value.
This fixes the io_pgetevents02 LTP test when compiled with -m32. Presently
the systemcall io_pgetevents_time64() unexpectedly accepts -1 for the
maximum number of events.
It doesn't appear other systemcalls with signed arguments are effected
because they all have compat variants defined and wired up.
Fixes: ebeb8c82ffaf ("syscalls/x86: Use 'struct pt_regs' based syscall calling for IA32_EMULATION and x32") Suggested-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Richard Palethorpe <rpalethorpe@suse.com> Signed-off-by: Nikolay Borisov <nik.borisov@suse.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Reviewed-by: Arnd Bergmann <arnd@arndb.de> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20240110130122.3836513-1-nik.borisov@suse.com Link: https://lore.kernel.org/ltp/20210921130127.24131-1-rpalethorpe@suse.com/ Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 71fee48f ("tick-sched: Fix idle and iowait sleeptime accounting vs
CPU hotplug") preserved total idle sleep time and iowait sleeptime across
CPU hotplug events.
Similar reasoning applies to the number of idle calls and idle sleeps to
get the proper average of sleep time per idle invocation.
Preserve those fields too.
Fixes: 71fee48f ("tick-sched: Fix idle and iowait sleeptime accounting vs CPU hotplug") Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20240122233534.3094238-1-tim.c.chen@linux.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There have been reports of the watchdog marking clocksources unstable on
machines with 8 NUMA nodes:
clocksource: timekeeping watchdog on CPU373:
Marking clocksource 'tsc' as unstable because the skew is too large:
clocksource: 'hpet' wd_nsec: 14523447520
clocksource: 'tsc' cs_nsec: 14524115132
The measured clocksource skew - the absolute difference between cs_nsec
and wd_nsec - was 668 microseconds:
The kernel used 200 microseconds for the uncertainty_margin of both the
clocksource and watchdog, resulting in a threshold of 400 microseconds (the
md variable). Both the cs_nsec and the wd_nsec value indicate that the
readout interval was circa 14.5 seconds. The observed behaviour is that
watchdog checks failed for large readout intervals on 8 NUMA node
machines. This indicates that the size of the skew was directly proportinal
to the length of the readout interval on those machines. The measured
clocksource skew, 668 microseconds, was evaluated against a threshold (the
md variable) that is suited for readout intervals of roughly
WATCHDOG_INTERVAL, i.e. HZ >> 1, which is 0.5 second.
The intention of 2e27e793e280 ("clocksource: Reduce clocksource-skew
threshold") was to tighten the threshold for evaluating skew and set the
lower bound for the uncertainty_margin of clocksources to twice
WATCHDOG_MAX_SKEW. Later in c37e85c135ce ("clocksource: Loosen clocksource
watchdog constraints"), the WATCHDOG_MAX_SKEW constant was increased to
125 microseconds to fit the limit of NTP, which is able to use a
clocksource that suffers from up to 500 microseconds of skew per second.
Both the TSC and the HPET use default uncertainty_margin. When the
readout interval gets stretched the default uncertainty_margin is no
longer a suitable lower bound for evaluating skew - it imposes a limit
that is far stricter than the skew with which NTP can deal.
The root causes of the skew being directly proportinal to the length of
the readout interval are:
* the inaccuracy of the shift/mult pairs of clocksources and the watchdog
* the conversion to nanoseconds is imprecise for large readout intervals
Prevent this by skipping the current watchdog check if the readout
interval exceeds 2 * WATCHDOG_INTERVAL. Considering the maximum readout
interval of 2 * WATCHDOG_INTERVAL, the current default uncertainty margin
(of the TSC and HPET) corresponds to a limit on clocksource skew of 250
ppm (microseconds of skew per second). To keep the limit imposed by NTP
(500 microseconds of skew per second) for all possible readout intervals,
the margins would have to be scaled so that the threshold value is
proportional to the length of the actual readout interval.
As for why the readout interval may get stretched: Since the watchdog is
executed in softirq context the expiration of the watchdog timer can get
severely delayed on account of a ksoftirqd thread not getting to run in a
timely manner. Surely, a system with such belated softirq execution is not
working well and the scheduling issue should be looked into but the
clocksource watchdog should be able to deal with it accordingly.
If we still own the FPU after initializing fcr31, when we are preempted
the dirty value in the FPU will be read out and stored into fcr31,
clobbering our setting. This can cause an improper floating-point
environment after execve(). For example:
zsh% cat measure.c
#include <fenv.h>
int main() { return fetestexcept(FE_INEXACT); }
zsh% cc measure.c -o measure -lm
zsh% echo $((1.0/3)) # raising FE_INEXACT
0.33333333333333331
zsh% while ./measure; do ; done
(stopped in seconds)
Call lose_fpu(0) before setting fcr31 to prevent this.
Closes: https://lore.kernel.org/linux-mips/7a6aa1bbdbbe2e63ae96ff163fab0349f58f1b9e.camel@xry111.site/ Fixes: 9b26616c8d9d ("MIPS: Respect the ISA level in FCSR handling") Cc: stable@vger.kernel.org Signed-off-by: Xi Ruoyao <xry111@xry111.site> Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Creating a region with 16 memory devices caused a problem. The div_u64_rem
function, used for dividing an unsigned 64-bit number by a 32-bit one,
faced an issue when SZ_256M * p->interleave_ways. The result surpassed
the maximum limit of the 32-bit divisor (4G), leading to an overflow
and a remainder of 0.
note: At this point, p->interleave_ways is 16, meaning 16 * 256M = 4G
To fix this issue, I replaced the div_u64_rem function with div64_u64_rem
and adjusted the type of the remainder.
Signed-off-by: Quanquan Cao <caoqq@fujitsu.com> Reviewed-by: Dave Jiang <dave.jiang@intel.com> Fixes: 23a22cd1c98b ("cxl/region: Allocate HPA capacity to regions") Cc: <stable@vger.kernel.org> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The FORCE_STOP_STATE bit is unsuitable to force the DSI link into LP-11
mode. It seems the bridge internally queues DSI packets and when the
FORCE_STOP_STATE bit is cleared, they are sent in close succession
without any useful timing (this also means that the DSI lanes won't go
into LP-11 mode). The length of this gibberish varies between 1ms and
5ms. This sometimes breaks an attached bridge (TI SN65DSI84 in this
case). In our case, the bridge will fail in about 1 per 500 reboots.
The FORCE_STOP_STATE handling was introduced to have the DSI lanes in
LP-11 state during the .pre_enable phase. But as it turns out, none of
this is needed at all. Between samsung_dsim_init() and
samsung_dsim_set_display_enable() the lanes are already in LP-11 mode.
The code as it was before commit 20c827683de0 ("drm: bridge:
samsung-dsim: Fix init during host transfer") and 0c14d3130654 ("drm:
bridge: samsung-dsim: Fix i.MX8M enable flow to meet spec") was correct
in this regard.
This patch basically reverts both commits. It was tested on an i.MX8M
SoC with an SN65DSI84 bridge. The signals were probed and the DSI
packets were decoded during initialization and link start-up. After this
patch the first DSI packet on the link is a VSYNC packet and the timing
is correct.
Command mode between .pre_enable and .enable was also briefly tested by
a quick hack. There was no DSI link partner which would have responded,
but it was made sure the DSI packet was send on the link. As a side
note, the command mode seems to just work in HS mode. I couldn't find
that the bridge will handle commands in LP mode.
Lantiq uses a common kernel config for devices with 24Kc and 34Kc cores.
The changes made previously to add support for interrupts on all cores
work on 24Kc platforms with SMP disabled and 34Kc platforms with SMP
enabled. This patch fixes boot issues on Danube (single core 24Kc) with
SMP enabled.
Fixes: 730320fd770d ("MIPS: lantiq: enable all hardware interrupts on second VPE") Signed-off-by: Aleksander Jan Bajkowski <olek2@wp.pl> Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
Machines which have more than 8 nodes fail to boot SMP after commit a2ccf46333d7b2cf96 ("LoongArch/smp: Call rcutree_report_cpu_starting()
earlier"). Because such machines use tlb-based per-cpu base address
rather than dmw-based per-cpu base address, resulting per-cpu variables
can only be accessed after tlb_init(). But rcutree_report_cpu_starting()
is now called before tlb_init() and accesses per-cpu variables indeed.
Since the original patch want to avoid the lockdep warning caused by
page allocation in tlb_init(), we can move rcutree_report_cpu_starting()
to tlb_init() where after tlb exception configuration but before page
allocation.
In __spi_pump_transfer_message(), the message was not finalized in the
first error return as it is in the other error return paths. Not
finalizing the message could cause anything waiting on the message to
complete to hang forever.
This adds the missing call to spi_finalize_current_message().
Fixes: ae7d2346dc89 ("spi: Don't use the message queue if possible in spi_sync") Signed-off-by: David Lechner <dlechner@baylibre.com> Link: https://msgid.link/r/20240125205312.3458541-2-dlechner@baylibre.com Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
A recent change moved the code that decides to skip
a channel or disable multichannel entirely, into a
helper function.
During this, a mutex_unlock of the session_mutex
should have been removed. Doing that here.
Fixes: f591062bdbf4 ("cifs: handle servers that still advertise multichannel after disabling") Signed-off-by: Shyam Prasad N <sprasad@microsoft.com> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
In the existing implementation, when executing interleaved write and read
operations in the ISR for a transfer length greater than the FIFO size,
the TXFIFO write precedes the RXFIFO read. Consequently, the initially
received data in the RXFIFO is pushed out and lost, leading to a failure
in data integrity. To address this issue, reverse the order of interleaved
operations and conduct the RXFIFO read followed by the TXFIFO write.
Fixes: 6afe2ae8dc48 ("spi: spi-cadence: Interleave write of TX and read of RX FIFO") Signed-off-by: Amit Kumar Mahapatra <amit.kumar-mahapatra@amd.com> Link: https://msgid.link/r/20231218090652.18403-1-amit.kumar-mahapatra@amd.com Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
SFDP read shall use the mspi reads when using the bcm_qspi_exec_mem_op()
call. This fixes SFDP parameter page read failures seen with parts that
now use SFDP protocol to read the basic flash parameter table.
Fixes: 5f195ee7d830 ("spi: bcm-qspi: Implement the spi_mem interface") Signed-off-by: Kamal Dasu <kamal.dasu@broadcom.com> Tested-by: Florian Fainelli <florian.fainelli@broadcom.com> Reviewed-by: Florian Fainelli <florian.fainelli@broadcom.com> Link: https://msgid.link/r/20240109210033.43249-1-kamal.dasu@broadcom.com Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Similar to commit 26db46bc9c67 ("drm/bridge: parade-ps8640: Ensure bridge
is suspended in .post_disable()"). Add a mutex to ensure that aux transfer
won't race with atomic_disable by holding the PM reference and prevent
the bridge from suspend.
Also we need to use pm_runtime_put_sync_suspend() to suspend the bridge
instead of idle with pm_runtime_put_sync().
Fixes: 3203e497eb76 ("drm/bridge: anx7625: Synchronously run runtime suspend.") Fixes: adca62ec370c ("drm/bridge: anx7625: Support reading edid through aux channel") Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org> Tested-by: Xuxin Xiong <xuxinxiong@huaqin.corp-partner.google.com> Reviewed-by: Pin-yen Lin <treapking@chromium.org> Reviewed-by: Douglas Anderson <dianders@chromium.org> Signed-off-by: Douglas Anderson <dianders@chromium.org> Link: https://patchwork.freedesktop.org/patch/msgid/20240118015916.2296741-1-hsinyi@chromium.org Signed-off-by: Sasha Levin <sashal@kernel.org>
Commit 1a721de8489f ("block: don't add or resize partition on the disk
with GENHD_FL_NO_PART") prevented all operations about partitions on disks
with GENHD_FL_NO_PART in blkpg_do_ioctl() since they are meaningless.
However, it changed error code in some scenarios. So move checking
GENHD_FL_NO_PART to bdev_add_partition() to eliminate impact.
Fixes: 1a721de8489f ("block: don't add or resize partition on the disk with GENHD_FL_NO_PART") Reported-by: Allison Karlitskaya <allison.karlitskaya@redhat.com> Closes: https://lore.kernel.org/all/CAOYeF9VsmqKMcQjo1k6YkGNujwN-nzfxY17N3F-CMikE1tYp+w@mail.gmail.com/ Signed-off-by: Li Lingfeng <lilingfeng3@huawei.com> Reviewed-by: Yu Kuai <yukuai3@huawei.com> Link: https://lore.kernel.org/r/20240118130401.792757-1-lilingfeng@huaweicloud.com Signed-off-by: Jens Axboe <axboe@kernel.dk> Signed-off-by: Sasha Levin <sashal@kernel.org>
Turns out this "SoC" side controller does not support certain commands,
such as reading chip JEDEC ID, so the controller is pretty much unusable
in Linux. We should be using the "PCH" side controller instead. For this
reason remove this PCI ID from the list.
Fixes: c2912d42e86e ("spi: intel-pci: Add support for Meteor Lake-S SPI serial flash") Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com> Link: https://msgid.link/r/20240122120034.2664812-2-mika.westerberg@linux.intel.com Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The raw interrupt status of eic maybe set before the interrupt is enabled,
since the eic interrupt has a latch function, which would trigger the
interrupt event once enabled it from user side. To solve this problem,
interrupts generated before setting the interrupt trigger type are ignored.
When wmi_install_notify_handler()/wmi_remove_notify_handler() are
unable to enable/disable the WMI device, they unconditionally return
an error to the caller.
When registering legacy WMI notify handlers, this means that the
callback remains registered despite wmi_install_notify_handler()
having returned an error.
When removing legacy WMI notify handlers, this means that the
callback is removed despite wmi_remove_notify_handler() having
returned an error.
Fix this by only warning when the WMI device could not be enabled.
This behaviour matches the bus-based WMI interface.
Tested on a Dell Inspiron 3505 and a Acer Aspire E1-731.
Fixes: 58f6425eb92f ("WMI: Cater for multiple events with same GUID") Signed-off-by: Armin Wolf <W_Armin@gmx.de> Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Reviewed-by: Hans de Goede <hdegoede@redhat.com> Link: https://lore.kernel.org/r/20240103192707.115512-2-W_Armin@gmx.de Signed-off-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Add a check to verify the result of xa_load() during the partition
lookups done while registering/unregistering the scheduler receiver
interrupt callbacks and while executing the main scheduler receiver
interrupt callback handler.
Add the missing rwlock initialization for the FF-A partition associated
the driver in ffa_setup_partitions(). It will the primary scheduler
partition in the host or the VM partition in the virtualised environment.
IOW, it corresponds to the partition with VM ID == drv_info->vm_id.
The clock protocol version as per the SCMI v3.2 specification is 0x30000.
Enable the v3.0 clock protocol features only when clock protocol version
equals 0x30000.
The previous beta version of the spec had this value set to 0x20001 and
th same value trickled down from the initial development. The version
update were missed in the driver.
When storing opps by level or index use xa_insert() instead of xa_store()
and add error-checking to spot bad duplicates indexes possibly wrongly
provided by the platform firmware.
gcc rightfully complains about excessive stack usage in the fimd_win_set_pixfmt()
function:
drivers/gpu/drm/exynos/exynos_drm_fimd.c: In function 'fimd_win_set_pixfmt':
drivers/gpu/drm/exynos/exynos_drm_fimd.c:750:1: error: the frame size of 1032 bytes is larger than 1024 byte
drivers/gpu/drm/exynos/exynos5433_drm_decon.c: In function 'decon_win_set_pixfmt':
drivers/gpu/drm/exynos/exynos5433_drm_decon.c:381:1: error: the frame size of 1032 bytes is larger than 1024 bytes
There is really no reason to copy the large exynos_drm_plane
structure to the stack before using one of its members, so just
use a pointer instead.
Jiri Slaby reported a futex state inconsistency resulting in -EINVAL during
a lock operation for a PI futex. It requires that the a lock process is
interrupted by a timeout or signal:
T1 Owns the futex in user space.
T2 Tries to acquire the futex in kernel (futex_lock_pi()). Allocates a
pi_state and attaches itself to it.
T2 Times out and removes its rt_waiter from the rt_mutex. Drops the
rtmutex lock and tries to acquire the hash bucket lock to remove
the futex_q. The lock is contended and T2 schedules out.
T1 Unlocks the futex (futex_unlock_pi()). Finds a futex_q but no
rt_waiter. Unlocks the futex (do_uncontended) and makes it available
to user space.
T3 Acquires the futex in user space.
T4 Tries to acquire the futex in kernel (futex_lock_pi()). Finds the
existing futex_q of T2 and tries to attach itself to the existing
pi_state. This (attach_to_pi_state()) fails with -EINVAL because uval
contains the TID of T3 but pi_state points to T1.
It's incorrect to unlock the futex and make it available for user space to
acquire as long as there is still an existing state attached to it in the
kernel.
T1 cannot hand over the futex to T2 because T2 already gave up and started
to clean up and is blocked on the hash bucket lock, so T2's futex_q with
the pi_state pointing to T1 is still queued.
T2 observes the futex_q, but ignores it as there is no waiter on the
corresponding rt_mutex and takes the uncontended path which allows the
subsequent caller of futex_lock_pi() (T4) to observe that stale state.
To prevent this the unlock path must dequeue all futex_q entries which
point to the same pi_state when there is no waiter on the rt mutex. This
requires obviously to make the dequeue conditional in the locking path to
prevent a double dequeue. With that it's guaranteed that user space cannot
observe an uncontended futex which has kernel state attached.
After commit 61167ad5fecd ("mm: pass nid to reserve_bootmem_region()")
nid of a reserved region is used by init_reserved_page() (with
CONFIG_DEFERRED_STRUCT_PAGE_INIT=y) to access node strucure.
In many cases the nid of the reserved memory is not set and this causes
a crash.
When the nid of a reserved region is not set, fall back to
early_pfn_to_nid(), so that nid of the first_online_node will be passed
to init_reserved_page().
Fixes: 61167ad5fecd ("mm: pass nid to reserve_bootmem_region()") Signed-off-by: Yajun Deng <yajun.deng@linux.dev> Link: https://lore.kernel.org/r/20240118061853.2652295-1-yajun.deng@linux.dev
[rppt: massaged the commit message] Signed-off-by: Mike Rapoport (IBM) <rppt@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
After commit 26db46bc9c67 ("drm/bridge: parade-ps8640: Ensure bridge
is suspended in .post_disable()"), if we hit the error case in
ps8640_aux_transfer() then we return without dropping the mutex. Fix
this oversight.
The ps8640 bridge seems to expect everything to be power cycled at the
disable process, but sometimes ps8640_aux_transfer() holds the runtime
PM reference and prevents the bridge from suspend.
Prevent that by introducing a mutex lock between ps8640_aux_transfer()
and .post_disable() to make sure the bridge is really powered off.
Fixes: 826cff3f7ebb ("drm/bridge: parade-ps8640: Enable runtime power management") Signed-off-by: Pin-yen Lin <treapking@chromium.org> Reviewed-by: Douglas Anderson <dianders@chromium.org> Signed-off-by: Douglas Anderson <dianders@chromium.org> Link: https://patchwork.freedesktop.org/patch/msgid/20240109120528.1292601-1-treapking@chromium.org Signed-off-by: Sasha Levin <sashal@kernel.org>
The driver never unregisters the audio codec platform device, which can
lead to a crash on module reloading, nor does it handle the return value
from sii902x_audio_codec_init().
- tidss probes, but is deferred as sii902x is still missing.
- sii902x starts probing and enters sii902x_init().
- sii902x calls drm_bridge_add(). Now the sii902x bridge is ready from
DRM's perspective.
- sii902x calls sii902x_audio_codec_init() and
platform_device_register_data()
- The registration of the audio platform device causes probing of the
deferred devices.
- tidss probes, which eventually causes sii902x_bridge_get_edid() to be
called.
- sii902x_bridge_get_edid() tries to use the i2c to read the edid.
However, the sii902x driver has not set up the i2c part yet, leading
to the crash.
Fix this by moving the drm_bridge_add() to the end of the
sii902x_init(), which is also at the very end of sii902x_probe().
As with several other panel drivers, this fails to link without the DP
helper library:
ld: drivers/gpu/drm/panel/panel-raydium-rm692e5.o: in function `rm692e5_prepare':
panel-raydium-rm692e5.c:(.text+0x11f4): undefined reference to `drm_dsc_pps_payload_pack'
Select the same symbols that the others already use.
It turns out that I had misconfigured the device I was using the panel
with; the bus data polarity is not high for this panel, I had to change
the config on the display controller's side.
Fix the panel config to properly reflect its accurate settings.
Unlike what is claimed in commit f5aa7d46b0ee ("drm/bridge:
parade-ps8640: Provide wait_hpd_asserted() in struct drm_dp_aux"), if
someone manually tries to do an AUX transfer (like via `i2cdump ${bus}
0x50 i`) while the panel is off we don't just get a simple transfer
error. Instead, the whole ps8640 gets thrown for a loop and goes into
a bad state.
Let's put the function to wait for the HPD (and the magical 50 ms
after first reset) back in when we're doing an AUX transfer. This
shouldn't actually make things much slower (assuming the panel is on)
because we should immediately poll and see the HPD high. Mostly this
is just an extra i2c transfer to the bridge.
This needs to be set to 1 to avoid a potential deadlock in
the GC 10.x and newer. On GC 9.x and older, this needs
to be set to 0. This can lead to hangs in some mixed
graphics and compute workloads. Updated firmware is also
required for AQL.
This needs to be set to 1 to avoid a potential deadlock in
the GC 10.x and newer. On GC 9.x and older, this needs
to be set to 0. This can lead to hangs in some mixed
graphics and compute workloads. Updated firmware is also
required for AQL.
This improves latency if the GPU is already busy with other work.
This is useful for VR compositors that submit highly latency-sensitive
compositing work on high-priority compute queues while the GPU is busy
rendering the next frame.
Reviewed-by: Marek Olšák <marek.olsak@amd.com> Signed-off-by: Friedrich Vock <friedrich.vock@gmx.de> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Stable-dep-of: 03ff6d7238b7 ("drm/amdgpu/gfx10: set UNORD_DISPATCH in compute MQDs") Signed-off-by: Sasha Levin <sashal@kernel.org>
[Why & How]
res_pool is not initialized in all situations such as virtual
environments, and therefore link encoder resources should not be
initialized if res_pool is NULL.
Cc: Mario Limonciello <mario.limonciello@amd.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: stable@vger.kernel.org Reviewed-by: Martin Leung <martin.leung@amd.com> Acked-by: Alex Hung <alex.hung@amd.com> Signed-off-by: Dillon Varone <dillon.varone@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
[why]
When ODM slice count is changed, otg master pipe's pixel clock params is
no longer valid as the value is dependent on ODM slice count.
Reviewed-by: Chaitanya Dhere <chaitanya.dhere@amd.com> Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com> Signed-off-by: Wenjing Liu <wenjing.liu@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Stable-dep-of: aa36d8971fcc ("drm/amd/display: Init link enc resources in dc_state only if res_pool presents") Signed-off-by: Sasha Levin <sashal@kernel.org>
[Why]
When mapping resources, resources could be unavailable.
Cc: Mario Limonciello <mario.limonciello@amd.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: stable@vger.kernel.org Reviewed-by: Sung joon Kim <sungjoon.kim@amd.com> Acked-by: Alex Hung <alex.hung@amd.com> Signed-off-by: Charlene Liu <charlene.liu@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
[Why]
Not clearing the memory select bits prior to OPTC disable can cause DSC
corruption issues when attempting to reuse a memory instance for another
OPTC that enables ODM.
[How]
Clear the memory select bits prior to disabling an OPTC.
Cc: Mario Limonciello <mario.limonciello@amd.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: stable@vger.kernel.org Reviewed-by: Charlene Liu <charlene.liu@amd.com> Acked-by: Alex Hung <alex.hung@amd.com> Signed-off-by: Ilya Bakoulin <ilya.bakoulin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
[Why]
If an OPP is used for a different OPTC without first being disconnected
from the previous OPTC, unexpected behaviour can occur. This also
applies to phantom pipes, which is what the current logic missed.
[How]
Disconnect OPPs from OPTC for phantom pipes before disabling OTG master.
Also move the disconnection to before the OTG master disable, since the
register is double buffered.
Reviewed-by: Dillon Varone <dillon.varone@amd.com> Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com> Signed-off-by: George Shen <george.shen@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Stable-dep-of: 3ba2a0bfd8cf ("drm/amd/display: Clear OPTC mem select on disable") Signed-off-by: Sasha Levin <sashal@kernel.org>
[Why]
Under some circumstances, disabling an OPTC and attempting to reclaim
its OPP(s) for a different OPTC could cause a hang/underflow due to OPPs
not being properly disconnected from the disabled OPTC.
[How]
Ensure that all OPPs are unassigned from an OPTC when it gets disabled.
Reviewed-by: Alvin Lee <alvin.lee2@amd.com> Acked-by: Wayne Lin <wayne.lin@amd.com> Signed-off-by: Ilya Bakoulin <ilya.bakoulin@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Stable-dep-of: 3ba2a0bfd8cf ("drm/amd/display: Clear OPTC mem select on disable") Signed-off-by: Sasha Levin <sashal@kernel.org>
Rename AUO 0x405c B116XAK01 to B116XAK01.0 and adjust the timing of
auo_b116xak01: T3=200, T12=500, T7_max = 50 according to decoding edid
and datasheet.
Fixes: da458286a5e2 ("drm/panel: Add support for AUO B116XAK01 panel") Cc: stable@vger.kernel.org Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org> Reviewed-by: Douglas Anderson <dianders@chromium.org> Acked-by: Maxime Ripard <mripard@kernel.org> Signed-off-by: Douglas Anderson <dianders@chromium.org> Link: https://patchwork.freedesktop.org/patch/msgid/20231107204611.3082200-2-hsinyi@chromium.org Signed-off-by: Sasha Levin <sashal@kernel.org>
[Why]
There are a number of instances where we convert HostVMMinPageSize or
GPUVMMinPageSize from bytes to KB by dividing (rather than multiplying) and
vice versa.
Additionally, in some cases, a parameter is passed through DML in KB but
later checked as if it were in bytes.
Cc: stable@vger.kernel.org Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com> Signed-off-by: Taimur Hassan <syed.hassan@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
[Why]
DMCUB can be in idle when we attempt to interface with the HW through
the GPINT mailbox resulting in a system hang.
[How]
Add dc_wake_and_execute_gpint() to wrap the wake, execute, sleep
sequence.
If the GPINT executes successfully then DMCUB will be put back into
sleep after the optional response is returned.
It functions similar to the inbox command interface.
Cc: Mario Limonciello <mario.limonciello@amd.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: stable@vger.kernel.org Reviewed-by: Hansen Dsouza <hansen.dsouza@amd.com> Acked-by: Wayne Lin <wayne.lin@amd.com> Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
[Why]
We can hang in place trying to send commands when the DMCUB isn't
powered on.
[How]
For functions that execute within a DC context or DC lock we can
wrap the direct calls to dm_execute_dmub_cmd/list with code that
exits idle power optimizations and reallows once we're done with
the command submission on success.
For DM direct submissions the DM will need to manage the enter/exit
sequencing manually.
We cannot invoke a DMCUB command directly within the DM execution
helper or we can deadlock.
Cc: Mario Limonciello <mario.limonciello@amd.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: stable@vger.kernel.org Reviewed-by: Hansen Dsouza <hansen.dsouza@amd.com> Acked-by: Wayne Lin <wayne.lin@amd.com> Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
[Why]
We can hang in place trying to send commands when the DMCUB isn't
powered on.
[How]
We need to exit out of the idle state prior to sending a command,
but the process that performs the exit also invokes a command itself.
Fixing this issue involves the following:
1. Using a software state to track whether or not we need to start
the process to exit idle or notify idle.
It's possible for the hardware to have exited an idle state without
driver knowledge, but entering one is always restricted to a driver
allow - which makes the SW state vs HW state mismatch issue purely one
of optimization, which should seldomly be hit, if at all.
2. Refactor any instances of exit/notify idle to use a single wrapper
that maintains this SW state.
This works simialr to dc_allow_idle_optimizations, but works at the
DMCUB level and makes sure the state is marked prior to any notify/exit
idle so we don't enter an infinite loop.
3. Make sure we exit out of idle prior to sending any commands or
waiting for DMCUB idle.
This patch takes care of 1/2. A future patch will take care of wrapping
DMCUB command submission with calls to this new interface.
Cc: Mario Limonciello <mario.limonciello@amd.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: stable@vger.kernel.org Reviewed-by: Hansen Dsouza <hansen.dsouza@amd.com> Acked-by: Wayne Lin <wayne.lin@amd.com> Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Stable-dep-of: 8892780834ae ("drm/amd/display: Wake DMCUB before sending a command") Signed-off-by: Sasha Levin <sashal@kernel.org>
[Why]
On resume from S3, may get apply_idle_optimizations call while DMUB
is inactive which will just time out.
[How]
Set and track power state in dmub_srv and check power state before
sending commands to DMUB. Add interface in both dmub_srv and
dc_dmub_srv
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Acked-by: Wayne Lin <wayne.lin@amd.com> Signed-off-by: Samson Tam <samson.tam@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Stable-dep-of: 8892780834ae ("drm/amd/display: Wake DMCUB before sending a command") Signed-off-by: Sasha Levin <sashal@kernel.org>
Writing sequentially to a huge file on btrfs on a SMR HDD revealed a
decline of the performance (220 MiB/s to 30 MiB/s after 500 minutes).
The performance goes down because of increased latency of the extent
allocation, which is induced by a traversing of a lot of full block groups.
So, this patch optimizes the ffe_ctl->hint_byte by choosing a block group
with sufficient size from the active block group list, which does not
contain full block groups.
After applying the patch, the performance is maintained well.
The conversion to CCI also converted the multi-byte register access to
big-endian. Correct the register definition by using the correct
little-endian ones.
Fixes: af73323b9770 ("media: imx290: Convert to new CCI register access helpers") Cc: stable@vger.kernel.org Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com> Reviewed-by: Hans de Goede <hdegoede@redhat.com> Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
[Sakari Ailus: Fixed the Fixes: tag.] Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Sasha Levin <sashal@kernel.org>
Some sensors, e.g. Sony IMX290, are using little-endian registers. Add
support for those by encoding the endianness into Bit 20 of the register
address.
Fixes: af73323b9770 ("media: imx290: Convert to new CCI register access helpers") Cc: stable@vger.kernel.org Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com> Reviewed-by: Hans de Goede <hdegoede@redhat.com> Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
[Sakari Ailus: Fixed commit message.] Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Sasha Levin <sashal@kernel.org>
The kernel allocates a memory buffer and provides its location to the
hardware, which uses it to update the HFI table. This allocation occurs
during boot and remains constant throughout runtime.
When resuming from hibernation, the restore kernel allocates a second
memory buffer and reprograms the HFI hardware with the new location as
part of a normal boot. The location of the second memory buffer may
differ from the one allocated by the image kernel.
When the restore kernel transfers control to the image kernel, its HFI
buffer becomes invalid, potentially leading to memory corruption if the
hardware writes to it (the hardware continues to use the buffer from the
restore kernel).
It is also possible that the hardware "forgets" the address of the memory
buffer when resuming from "deep" suspend. Memory corruption may also occur
in such a scenario.
To prevent the described memory corruption, disable HFI when preparing to
suspend or hibernate. Enable it when resuming.
Add syscore callbacks to handle the package of the boot CPU (packages of
non-boot CPUs are handled via CPU offline). Syscore ops always run on the
boot CPU. Additionally, HFI only needs to be disabled during "deep" suspend
and hibernation. Syscore ops only run in these cases.
Cc: 6.1+ <stable@vger.kernel.org> # 6.1+ Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
[ rjw: Comment adjustment, subject and changelog edits ] Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
In preparation to support hibernation, add functionality to disable an HFI
instance during CPU offline. The last CPU of an instance that goes offline
will disable such instance.
The Intel Software Development Manual states that the operating system must
wait for the hardware to set MSR_IA32_PACKAGE_THERM_STATUS[26] after
disabling an HFI instance to ensure that it will no longer write on the HFI
memory. Some processors, however, do not ever set such bit. Wait a minimum
of 2ms to give time hardware to complete any pending memory writes.
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Stable-dep-of: 97566d09fd02 ("thermal: intel: hfi: Add syscore callbacks for system-wide PM") Signed-off-by: Sasha Levin <sashal@kernel.org>
projects / thirdparty / kernel / stable.git / log
