Zhang Yi points out that the dynamic folio_batch allocation in
iomap_fill_dirty_folios() is problematic for the ext4 on iomap work
that is under development because it doesn't sufficiently handle the
allocation failure case (by allowing a retry, for example). We've
also seen lockdep (via syzbot) complain recently about the scope of
the allocation.
The dynamic allocation was initially added for simplicity and to
help indicate whether the batch was used or not by the calling fs.
To address these issues, put the batch on the stack of
iomap_zero_range() and use a flag to control whether the batch
should be used in the iomap folio lookup path. This keeps things
simple and eliminates allocation issues with lockdep and for ext4 on
iomap.
While here, also clean up the fill helper signature to be more
consistent with the underlying filemap helper. Pass through the
return value of the filemap helper (folio count) and update the
lookup offset via an out param.
Fixes: 395ed1ef00 ("iomap: optional zero range dirty folio processing")
Signed-off-by: Brian Foster <bfoster@redhat.com>
Link: https://patch.msgid.link/20251208140548.373411-1-bfoster@redhat.com
Acked-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Christian Brauner <brauner@kernel.org>
Commit 6f504cbf10 ("media: convert media_request_alloc() to
FD_PREPARE()") moved the call to fd_install() (now hidden in
fd_publish()) before the snprintf(), making the later write to
potentially already freed memory, as userland is free to call
close() concurrently right after the call to fd_install() which
may end up in the request_fops.release() handler freeing 'req'.
Fixes: 6f504cbf10 ("media: convert media_request_alloc() to FD_PREPARE()")
Signed-off-by: Mathias Krause <minipli@grsecurity.net>
Link: https://patch.msgid.link/20251209210903.603958-1-minipli@grsecurity.net
Signed-off-by: Christian Brauner <brauner@kernel.org>
These callbacks are necessary to synchronize ->write_thread callback
against other operations using the same device.
Signed-off-by: Marcos Paulo de Souza <mpdesouza@suse.com>
Reviewed-by: John Ogness <john.ogness@linutronix.de>
Link: https://patch.msgid.link/20251208-nbcon-device-cb-fix-v2-1-36be8d195123@suse.com
Signed-off-by: Petr Mladek <pmladek@suse.com>
IPC4 is using the NHLT blob itself and sends the SSP blob from it directly
to the firmware, there is no need for the MCLK quirk based on the SSP blob
since the SSP blob is in use.
At the same time reword the error, info and debug messages for clarity.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Reviewed-by: Balamurugan C <balamurugan.c@intel.com>
Link: https://patch.msgid.link/20251215130819.31218-1-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
While the used GML is consistent with the pattern for other Intel * Lake
SoCs, the de facto use is GLK. Update the acronym and users accordingly.
Note, a handful of the drivers for Gemini Lake in the Linux kernel use
GLK already (LPC, MEI, pin control, SDHCI, ...) and even some in ASoC.
The only ones in this patch used the inconsistent one.
Acked-by: Bjorn Helgaas <bhelgaas@google.com> # pci_ids.h
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Cezary Rojewski <cezary.rojewski@intel.com>
Link: https://patch.msgid.link/20251212181742.3944789-1-andriy.shevchenko@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Revert "ASoC: Intel: soc-acpi-intel-nvl-match: add rt722 l3 support"
NVL should be only using functional topologies for products, no monolithic
topologies are planned to be released.
In parallel a feature has been landed [1] which allows to remove the
entries from the match table for sdca codecs to rely solely on function
fragments.
This reverts commit 41566e3de4.
Link: https://lore.kernel.org/linux-sound/20251014071335.3844631-1-yung-chuan.liao@linux.intel.com/
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Link: https://patch.msgid.link/20251215101036.9370-3-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
When there is no fallback possibility available for the function topology
use it is better to try to create a profile for the card in best effort
manner, leaving out non supported links for example.
As an example: some laptops present SSPx-BT link but we don't have fragment
yet to support this. If we only have support for functional topology
without monolithic fallback then we would fail the card creation.
The reason why the monolithic topology works on the same device is that it
does not have the SSPx-BT link handled, it is ignored.
In case when there is no fallback possibility we should try to create the
card with links that we support as best effort instead of failing and
leaving the user without a card.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Link: https://patch.msgid.link/20251215101036.9370-2-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
NVL-S can be used with on-demand DSP booting, set the flag to enable it.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Link: https://patch.msgid.link/20251215132946.2155-9-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
PTL and WCL can be used with on-demand DSP booting, set the flag to
enable it.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Reviewed-by: Liam Girdwood <liam.r.girdwood@intel.com>
Link: https://patch.msgid.link/20251215132946.2155-8-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
LNL can be used with on-demand DSP booting, set the flag to enable it.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Reviewed-by: Liam Girdwood <liam.r.girdwood@intel.com>
Link: https://patch.msgid.link/20251215132946.2155-7-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
If on-demand DSP boot is used we need to make sure that the DSP is booted
up - which might not be the case - since we need ChainDMA in normal, non
DSPless mode for the BRA to work.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Reviewed-by: Liam Girdwood <liam.r.girdwood@intel.com>
Link: https://patch.msgid.link/20251215132946.2155-6-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
With the introduction of on-demand DSP boot the rpm status not necessary
tells that the DSP firmware is booted up.
Introduce the sof_client_boot_dsp() which can be used to make sure that
the DSP is booted and it can handle IPCs.
Update the client drivers to use the new function where it is expected that
the DSP is booted up.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Reviewed-by: Liam Girdwood <liam.r.girdwood@intel.com>
Link: https://patch.msgid.link/20251215132946.2155-5-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
On system suspend / resume we always power up the DSP and boot the
firmware, which is not strictly needed as right after the firmware booted
up we power the DSP down again on suspend and we also power it down after
resume after some inactivity.
Out of caution, add a new platform descriptor flag to enable on-demand
DSP boot since this might not work without changes to platform code on
certain platforms.
With the on-demand dsp boot enabled we will not boot the DSP and firmware
up on system or rpm resume, just enable audio subsystem since audio IPs,
like HDA and SoundWire might be needed (codecs suspend/resume operation).
Only boot up the DSP during the first hw_params() call when the DSP is
really going to be needed.
In this way we can handle the audio related use cases:
normal audio use (rpm suspend/resume)
system suspend/resume without active audio
system suspend/resume with active audio
system suspend/resume without active audio, and audio start before the rpm
suspend timeout
Add module option to force the on-demand DSP boot to allow it to be
disabled or enabled without kernel change for testing.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Reviewed-by: Liam Girdwood <liam.r.girdwood@intel.com>
Link: https://patch.msgid.link/20251215132946.2155-4-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Test earlier for the existence of ext_volatile_get callback and if it is
missing, skip the rpm calls to avoid needles DSP power on.
No change in functionality, we just skip the DSP power on in the unlikely
case when the ext_volatile _get is not supported and yet the topology adds
such control.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Reviewed-by: Liam Girdwood <liam.r.girdwood@intel.com>
Link: https://patch.msgid.link/20251215132946.2155-3-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The initial library loading is happening during topology loading, which is
already protected with pm_runtime_resume_and_get() via pcm.c
The redundant rpm code can be dropped from sof_ipc4_load_library()
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Reviewed-by: Liam Girdwood <liam.r.girdwood@intel.com>
Link: https://patch.msgid.link/20251215132946.2155-2-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The default topology path for IPC4 is intel/sof-ipc4-tplg with a symlink
to it as intel/sof-ace-tplg to support old kernels.
sof-bin has been released in this manner for almost two years now, it is
time to change the default path for MTL family.
Link: https://thesofproject.github.io/latest/getting_started/intel_debug/introduction.html#topology-file
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Link: https://patch.msgid.link/20251215130805.31146-1-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Currently, we send all channels to all amps and copy the channel_mask
to all ALH DMAs in playback. However, the amp may not have the
capability to run any process and SOF may need to split the channels
and send specific data channel to each amp. In that case, we need
to split the channel_mask in ALH DMA.
Copy the channel mask only if the widget channel count is the same
the FE channels for playback, otherwise, split the channels among the
aggregated DAIs. Like what we did in capture.
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Liam Girdwood <liam.r.girdwood@intel.com>
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Link: https://patch.msgid.link/20251215130723.31081-1-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Add some context to the error prints when sink or source widget is not
found by printing the name of the other side of the connection.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Link: https://patch.msgid.link/20251215130741.31106-1-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
try_to_free_buffers() can be called on folios with no buffers attached
when filemap_release_folio() is invoked on a folio belonging to a mapping
with AS_RELEASE_ALWAYS set but no release_folio operation defined.
In such cases, folio_needs_release() returns true because of the
AS_RELEASE_ALWAYS flag, but the folio has no private buffer data. This
causes try_to_free_buffers() to call drop_buffers() on a folio with no
buffers, leading to a null pointer dereference.
Adding a check in try_to_free_buffers() to return early if the folio has no
buffers attached, with WARN_ON_ONCE() to alert about the misconfiguration.
This provides defensive hardening.
Signed-off-by: Deepakkumar Karn <dkarn@redhat.com>
Link: https://patch.msgid.link/20251211131211.308021-1-dkarn@redhat.com
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Christian Brauner <brauner@kernel.org>
SSP/DMIC blobs have no support for FLOAT type, they are using S32 on data
bus.
Convert the format from FLOAT_LE to S32_LE to make sure that the correct
format is used within the path.
FLOAT conversion will be done on the host side (or within the path).
Fixes: f7c41911ad ("ASoC: SOF: ipc4-topology: Add support for float sample type")
Cc: stable@vger.kernel.org
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>
Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Link: https://patch.msgid.link/20251215120648.4827-3-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
With the introduction of 8-bit formats the DMIC blob lookup also needs to
be modified to prefer the 32-bit blob when 8-bit format is used on FE.
At the same time we also need to make sure that in case 8-bit format is
used, but only 16-bit blob is available for DMIC then we will not try to
look for 8-bit blob (which is invalid) as fallback, but for a 16-bit one.
Fixes: c04c2e8296 ("ASoC: SOF: ipc4-topology: Add support for 8-bit formats")
Cc: stable@vger.kernel.org
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>
Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Link: https://patch.msgid.link/20251215120648.4827-2-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Add TUXEDO Book BA15 Gen10 to the list of supported devices of the
Uniwill driver.
Signed-off-by: Werner Sembach <wse@tuxedocomputers.com>
Link: https://patch.msgid.link/20251212180319.712913-1-wse@tuxedocomputers.com
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
This reverts commit
314c828416 ("netfilter: nf_tables: can't schedule in nft_chain_validate"):
Since commit a60a5abe19 ("netfilter: nf_tables: allow iter callbacks to sleep")
the iterator callback is invoked without rcu read lock held, so this
cond_resched() is now valid.
Signed-off-by: Florian Westphal <fw@strlen.de>
Hamza Mahfooz reports cpu soft lock-ups in
nft_chain_validate():
watchdog: BUG: soft lockup - CPU#1 stuck for 27s! [iptables-nft-re:37547]
[..]
RIP: 0010:nft_chain_validate+0xcb/0x110 [nf_tables]
[..]
nft_immediate_validate+0x36/0x50 [nf_tables]
nft_chain_validate+0xc9/0x110 [nf_tables]
nft_immediate_validate+0x36/0x50 [nf_tables]
nft_chain_validate+0xc9/0x110 [nf_tables]
nft_immediate_validate+0x36/0x50 [nf_tables]
nft_chain_validate+0xc9/0x110 [nf_tables]
nft_immediate_validate+0x36/0x50 [nf_tables]
nft_chain_validate+0xc9/0x110 [nf_tables]
nft_immediate_validate+0x36/0x50 [nf_tables]
nft_chain_validate+0xc9/0x110 [nf_tables]
nft_immediate_validate+0x36/0x50 [nf_tables]
nft_chain_validate+0xc9/0x110 [nf_tables]
nft_table_validate+0x6b/0xb0 [nf_tables]
nf_tables_validate+0x8b/0xa0 [nf_tables]
nf_tables_commit+0x1df/0x1eb0 [nf_tables]
[..]
Currently nf_tables will traverse the entire table (chain graph), starting
from the entry points (base chains), exploring all possible paths
(chain jumps). But there are cases where we could avoid revalidation.
Consider:
1 input -> j2 -> j3
2 input -> j2 -> j3
3 input -> j1 -> j2 -> j3
Then the second rule does not need to revalidate j2, and, by extension j3,
because this was already checked during validation of the first rule.
We need to validate it only for rule 3.
This is needed because chain loop detection also ensures we do not exceed
the jump stack: Just because we know that j2 is cycle free, its last jump
might now exceed the allowed stack size. We also need to update all
reachable chains with the new largest observed call depth.
Care has to be taken to revalidate even if the chain depth won't be an
issue: chain validation also ensures that expressions are not called from
invalid base chains. For example, the masquerade expression can only be
called from NAT postrouting base chains.
Therefore we also need to keep record of the base chain context (type,
hooknum) and revalidate if the chain becomes reachable from a different
hook location.
Reported-by: Hamza Mahfooz <hamzamahfooz@linux.microsoft.com>
Closes: https://lore.kernel.org/netfilter-devel/20251118221735.GA5477@linuxonhyperv3.guj3yctzbm1etfxqx2vob5hsef.xx.internal.cloudapp.net/
Tested-by: Hamza Mahfooz <hamzamahfooz@linux.microsoft.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
The RK3368 has only one VOP, so there is no source selection happening
and the controller uses an internal PHY for the HDMI output.
Signed-off-by: Heiko Stuebner <heiko.stuebner@cherry.de>
Reviewed-by: Andy Yan <andyshrk@163.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patch.msgid.link/20251021074254.87065-3-heiko@sntech.de
Define a new compatible for RK3368 HDMI.
The RK3368 HDMI also uses a PHY internal to the controller, so works
similar to other controllers, with the exception that the RK3368
only has one VOP, so there is no source selection needed.
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Signed-off-by: Heiko Stuebner <heiko.stuebner@cherry.de>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patch.msgid.link/20251021074254.87065-2-heiko@sntech.de
Now that the last in-tree filesystem has been converted to the new mount
API, remove all legacy mount API code designed to handle un-converted
filesystems, and remove associated documentation as well.
(The code to handle the legacy mount(2) syscall from userspace is still
in place, of course.)
Tested with an allmodconfig build on x86_64, and a sanity check of an
old mount(2) syscall mount.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Link: https://patch.msgid.link/20251212174403.2882183-1-sandeen@redhat.com
Signed-off-by: Christian Brauner <brauner@kernel.org>
Mateusz Guzik <mjguzik@gmail.com> says:
When spawning and killing threads in separate processes in parallel the
primary bottleneck on the stock kernel is pidmap_lock, largely because
of a back-to-back acquire in the common case.
Benchmark code at the end.
With this patchset alloc_pid() only takes the lock once and consequently
alleviates the problem. While scalability improves, the lock remains the
primary bottleneck by a large margin.
I believe idr is a poor choice for the task at hand to begin with, but
sorting out that out beyond the scope of this patchset. At the same time
any replacement would be best evaluated against a state where the
above relock problem is fixed.
Performance improvement varies between reboots. When benchmarking with
20 processes creating and killing threads in a loop, the unpatched
baseline hovers around 465k ops/s, while patched is anything between
~510k ops/s and ~560k depending on false-sharing (which I only minimally
sanitized). So this is at least 10% if you are unlucky.
bench from will-it-scale:
char *testcase_description = "Thread creation and teardown";
static void *worker(void *arg)
{
return (NULL);
}
void testcase(unsigned long long *iterations, unsigned long nr)
{
pthread_t thread[1];
int error;
while (1) {
for (int i = 0; i < 1; i++) {
error = pthread_create(&thread[i], NULL, worker, NULL);
assert(error == 0);
}
for (int i = 0; i < 1; i++) {
error = pthread_join(thread[i], NULL);
assert(error == 0);
}
(*iterations)++;
}
}
v2:
- cosmetic fixes from Oleg
- drop idr_preload_many, relock pidmap + call idr_preload again instead
- write a commit message for the alloc pid patch
* patches from https://patch.msgid.link/20251203092851.287617-1-mjguzik@gmail.com:
pid: only take pidmap_lock once on alloc
ns: pad refcount
Link: https://patch.msgid.link/20251203092851.287617-1-mjguzik@gmail.com
Signed-off-by: Christian Brauner <brauner@kernel.org>
Opening and closing an inode dirties the ->i_readcount field.
Depending on the alignment of the inode, it may happen to false-share
with other fields loaded both for both operations to various extent.
This notably concerns the ->i_flctx field.
Since most inodes don't have the field populated, this bit can be managed
with a flag in ->i_opflags instead which bypasses the problem.
Here are results I obtained while opening a file read-only in a loop
with 24 cores doing the work on Sapphire Rapids. Utilizing the flag as
opposed to reading ->i_flctx field was toggled at runtime as the benchmark
was running, to make sure both results come from the same alignment.
before: 3233740
after: 3373346 (+4%)
before: 3284313
after: 3518711 (+7%)
before: 3505545
after: 4092806 (+16%)
Or to put it differently, this varies wildly depending on how (un)lucky
you get.
The primary bottleneck before and after is the avoidable lockref trip in
do_dentry_open().
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Mateusz Guzik <mjguzik@gmail.com>
Link: https://patch.msgid.link/20251203094837.290654-2-mjguzik@gmail.com
Signed-off-by: Christian Brauner <brauner@kernel.org>
When spawning and killing threads in separate processes in parallel the
primary bottleneck on the stock kernel is pidmap_lock, largely because
of a back-to-back acquire in the common case. This aspect is fixed with
the patch.
Performance improvement varies between reboots. When benchmarking with
20 processes creating and killing threads in a loop, the unpatched
baseline hovers around 465k ops/s, while patched is anything between
~510k ops/s and ~560k depending on false-sharing (which I only minimally
sanitized). So this is at least 10% if you are unlucky.
The change also facilitated some cosmetic fixes.
It has an unintentional side effect of no longer issuing spurious
idr_preload() around idr_replace().
Signed-off-by: Mateusz Guzik <mjguzik@gmail.com>
Link: https://patch.msgid.link/20251203092851.287617-3-mjguzik@gmail.com
Reviewed-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Christian Brauner <brauner@kernel.org>
Matches the idiom of storing a pointer with a release fence and safely
getting the content with a consume fence after.
Eliminates an actual fence on some archs.
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Mateusz Guzik <mjguzik@gmail.com>
Link: https://patch.msgid.link/20251203094837.290654-1-mjguzik@gmail.com
Signed-off-by: Christian Brauner <brauner@kernel.org>
Note no effort is made to make sure structs embedding the namespace are
themselves aligned, so this is not guaranteed to eliminate cacheline
bouncing due to refcount management.
Signed-off-by: Mateusz Guzik <mjguzik@gmail.com>
Link: https://patch.msgid.link/20251203092851.287617-2-mjguzik@gmail.com
Signed-off-by: Christian Brauner <brauner@kernel.org>
No need for the crapper to be susceptible to false-sharing.
Signed-off-by: Mateusz Guzik <mjguzik@gmail.com>
Link: https://patch.msgid.link/20251203095508.291073-1-mjguzik@gmail.com
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Christian Brauner <brauner@kernel.org>
min_t(unsigned int, a, b) casts an 'unsigned long' to 'unsigned int'.
Use min(a, b) instead as it promotes any 'unsigned int' to 'unsigned long'
and so cannot discard significant bits.
A couple of places need umin() because of loops like:
nfolios = DIV_ROUND_UP(ret + start, PAGE_SIZE);
for (i = 0; i < nfolios; i++) {
struct folio *folio = page_folio(pages[i]);
...
unsigned int len = umin(ret, PAGE_SIZE - start);
...
ret -= len;
...
}
where the compiler doesn't track things well enough to know that
'ret' is never negative.
The alternate loop:
for (i = 0; ret > 0; i++) {
struct folio *folio = page_folio(pages[i]);
...
unsigned int len = min(ret, PAGE_SIZE - start);
...
ret -= len;
...
}
would be equivalent and doesn't need 'nfolios'.
Most of the 'unsigned long' actually come from PAGE_SIZE.
Detected by an extra check added to min_t().
Signed-off-by: David Laight <david.laight.linux@gmail.com>
Link: https://patch.msgid.link/20251119224140.8616-31-david.laight.linux@gmail.com
Signed-off-by: Christian Brauner <brauner@kernel.org>
...which do nothing. They were deprecated (in documentation) in
6b99e6e6aa ("Documentation/admin-guide: blockdev/ramdisk: remove use of
"rdev"") in 2020 and in kernel messages in c8376994c8 ("initrd: remove
support for multiple floppies") in 2020.
Signed-off-by: Askar Safin <safinaskar@gmail.com>
Link: https://patch.msgid.link/20251119222407.3333257-2-safinaskar@gmail.com
Acked-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Christian Brauner <brauner@kernel.org>
During GT reset recovery in do_gt_restart(), xe_uc_start() was called
before xe_reg_sr_apply_mmio() restored engine-specific registers. This
created a race window where the scheduler could run jobs before hardware
state was fully restored.
This caused failures in eudebug tests (xe_exec_sip_eudebug@breakpoint-
waitsip-*) where TD_CTL register (containing TD_CTL_GLOBAL_DEBUG_ENABLE)
wasn't restored before jobs started executing. Breakpoints would fail to
trigger SIP entry because the debug enable bit wasn't set yet.
Fix by moving xe_uc_start() after all MMIO register restoration,
including engine registers and CCS mode configuration, ensuring all
hardware state is fully restored before any jobs can be scheduled.
Fixes: dd08ebf6c3 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Signed-off-by: Jan Maslak <jan.maslak@intel.com>
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patch.msgid.link/20251210145618.169625-2-jan.maslak@intel.com
(cherry picked from commit 825aed0328)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
There are some corner cases where flushing transient
data may take slightly longer than the 150us timeout
we currently allow. Update the driver to use a 300us
timeout instead based on the latest guidance from
the hardware team. An update to the bspec to formally
document this is expected to arrive soon.
Fixes: c01c6066e6 ("drm/xe/device: implement transient flush")
Signed-off-by: Jagmeet Randhawa <jagmeet.randhawa@intel.com>
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patch.msgid.link/0201b1d6ec64d3651fcbff1ea21026efa915126a.1765487866.git.jagmeet.randhawa@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
(cherry picked from commit d69d3636f5)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Modify kernel-doc comments in ti_sci.h to eliminate all kernel-doc
warnings:
* use correct/matching struct names in kdoc comments
* use correct struct member names in kdoc comments
* add a ':' after struct member names where needed
* change a blank kdoc line to " *"
* convert 3 structs to kernel-doc comments:
struct ti_sci_msg_rm_udmap_tx_ch_cfg_req
struct ti_sci_msg_rm_udmap_rx_ch_cfg_req
struct ti_sci_msg_rm_udmap_flow_cfg_req
Fixes 13 kernel-doc warnings:
Warning: drivers/firmware/ti_sci.h:609 expecting prototype for struct
tisci_msg_req_prepare_sleep. Prototype was for struct
ti_sci_msg_req_prepare_sleep instead
Warning: drivers/firmware/ti_sci.h:609 struct member 'hdr' not described
in 'ti_sci_msg_req_prepare_sleep'
Warning: drivers/firmware/ti_sci.h:609 struct member 'mode' not described
in 'ti_sci_msg_req_prepare_sleep'
Warning: drivers/firmware/ti_sci.h:609 struct member 'ctx_lo' not described
in 'ti_sci_msg_req_prepare_sleep'
Warning: drivers/firmware/ti_sci.h:609 struct member 'ctx_hi' not described
in 'ti_sci_msg_req_prepare_sleep'
Warning: drivers/firmware/ti_sci.h:609 struct member 'debug_flags' not
described in 'ti_sci_msg_req_prepare_sleep'
Warning: drivers/firmware/ti_sci.h:623 expecting prototype for struct
tisci_msg_set_io_isolation_req. Prototype was for struct
ti_sci_msg_req_set_io_isolation instead
Warning: drivers/firmware/ti_sci.h:696 struct member 'latency' not
described in 'ti_sci_msg_req_lpm_set_latency_constraint'
Warning: drivers/firmware/ti_sci.h:857 bad line:
Warning: drivers/firmware/ti_sci.h:1002 This comment starts with '/**', but
isn't a kernel-doc comment. Refer to Documentation/doc-guide/kernel-doc.rst
* Configures a Navigator Subsystem UDMAP transmit channel
Warning: drivers/firmware/ti_sci.h:1130 This comment starts with '/**', but
isn't a kernel-doc comment. Refer to Documentation/doc-guide/kernel-doc.rst
* Configures a Navigator Subsystem UDMAP receive channel
Warning: drivers/firmware/ti_sci.h:1249 This comment starts with '/**', but
isn't a kernel-doc comment. Refer to Documentation/doc-guide/kernel-doc.rst
* Configures a Navigator Subsystem UDMAP receive flow
Warning: drivers/firmware/ti_sci.h:1421 struct member 'valid_params'
not described in 'ti_sci_msg_rm_udmap_flow_cfg_req'
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Link: https://patch.msgid.link/20251128055731.812460-1-rdunlap@infradead.org
Signed-off-by: Nishanth Menon <nm@ti.com>
Ensure VF migration recovery work is only queued when no recovery is
already queued and teardown is not in progress.
Fixes: b47c0c07c3 ("drm/xe/vf: Teardown VF post migration worker on driver unload")
Signed-off-by: Satyanarayana K V P <satyanarayana.k.v.p@intel.com>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Tomasz Lis <tomasz.lis@intel.com>
Reviewed-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Link: https://patch.msgid.link/20251210052546.622809-5-satyanarayana.k.v.p@intel.com
(cherry picked from commit 8d8cf42b03)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Some Xe bos are allocated with extra backing-store for the CCS
metadata. It's never been the intention to share the CCS metadata
when exporting such bos as dma-buf. Don't include it in the
dma-buf sg-table.
Fixes: dd08ebf6c3 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: <stable@vger.kernel.org> # v6.8+
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Karol Wachowski <karol.wachowski@linux.intel.com>
Link: https://patch.msgid.link/20251209204920.224374-1-thomas.hellstrom@linux.intel.com
(cherry picked from commit a4ebfb9d95)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
MEI GSC interrupt comes from i915 or xe driver. It has top half and
bottom half. Top half is called from i915/xe interrupt handler. It
should be in irq disabled context.
With RT kernel(PREEMPT_RT enabled), by default IRQ handler is in
threaded IRQ. MEI GSC top half might be in threaded IRQ context.
generic_handle_irq_safe API could be called from either IRQ or
process context, it disables local IRQ then calls MEI GSC interrupt
top half.
This change fixes B580 GPU boot issue with RT enabled.
Fixes: e02cea83d3 ("drm/xe/gsc: add Battlemage support")
Tested-by: Baoli Zhang <baoli.zhang@intel.com>
Signed-off-by: Junxiao Chang <junxiao.chang@intel.com>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patch.msgid.link/20251107033152.834960-1-junxiao.chang@intel.com
Signed-off-by: Maarten Lankhorst <dev@lankhorst.se>
(cherry picked from commit 3efadf0287)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
