| Age | Commit message (Collapse) | Author |
|---|
|
Add function to enter ULLS mode for migration job and delayed worker to
exit (power saving). ULLS mode expected to entered upon page fault or
SVM prefetch. ULLS mode exit delay is currently set to 5ms.
ULLS mode only support on DGFX and USM platforms where a hardware engine
is reserved for migrations jobs. When in ULLS mode, set several flags on
migration jobs so submission backend / ring ops can properly submit in
ULLS mode.
Upon ULLS mode enter, send a job trigger waiting a semphore pipling
initial GuC / HW conetxt switch.
Upon ULLS mode exit, send a job to trigger that current ULLS
semaphore so the ring can be taken off the hardware.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Link: https://patch.msgid.link/20260925201526.1355279-20-matthew.brost@intel.com
|
|
Define memory layout for ULLS semaphores stored in LRC memory. Add
support functions to return GGTT address and set semaphore based on a
job's seqno.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Link: https://patch.msgid.link/20260925201526.1355279-19-matthew.brost@intel.com
|
|
Some multi-tile devices may want to mirror page tables across tiles for
memory-bandwidth reasons, while others may not. Add a device flag that
allows enabling or disabling page-table mirroring across tiles.
Setting the flag to true (the existing behavior) on PVC, but both modes
have been tested and are working on PVC.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Reviewed-by: Francois Dugast <francois.dugast@intel.com>
Link: https://patch.msgid.link/20260925201526.1355279-18-matthew.brost@intel.com
|
|
With CPU binds, it no longer makes sense to implement CPU bind handling
in the migrate layer, as these operations are entirely decoupled from
hardware. Introduce a dedicated CPU bind layer stored at the device
level.
Since CPU binds are tile-independent, update the PT layer to generate a
single bind job even when pages are mirrored across tiles.
This patch is large because the refactor touches multiple file / layers
and ensures functional equivalence before and after the change.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Reviewed-by: Shuicheng Lin <shuicheng.lin@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Link: https://patch.msgid.link/20260925201526.1355279-17-matthew.brost@intel.com
|
|
Since bind jobs execute on the CPU rather than the GPU, maintaining a
per-tile bind queue no longer provides value. Convert the driver to use
a single bind queue shared across tiles. The primary change is routing
all GT TLB invalidations through this unified bind queue.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Shuicheng Lin <shuicheng.lin@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Link: https://patch.msgid.link/20260925201526.1355279-16-matthew.brost@intel.com
|
|
Both populate and clear have unused void* ptr arguments, remove these.
Also consolidate ofs, num_qwords in update argument.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Shuicheng Lin <shuicheng.lin@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Link: https://patch.msgid.link/20260925201526.1355279-15-matthew.brost@intel.com
|
|
No reason to use the GPU for binds.
Benefits of CPU-based binds:
- Lower latency once dependencies are resolved, as there is no
interaction with the GuC or a hardware context switch both of which
are relatively slow.
- Large arrays of binds do not risk running out of migration PTEs,
avoiding -ENOBUFS being returned to userspace.
- Kernel binds are decoupled from the migration exec queue (which issues
copies and clears), so they cannot get stuck behind unrelated
jobs—this can be a problem with parallel GPU faults.
- Paves the for path decouping binds from tiles and individual engines
- Enables ULLS on the migration exec queue, as this queue has exclusive
access to the paging copy engine.
Update migration layer to formulate a PT job which will issue CPU bind
in the submission backend.
All code related to GPU-based binding has been removed.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Link: https://patch.msgid.link/20260925201526.1355279-14-matthew.brost@intel.com
|
|
Now that the CPU is always used for binds even in jobs, CPU bind jobs
can pass GPU jobs in the same exec queue resulting dma-fences signaling
out-of-order. Use a dedicated exec queue for binds issued from page
faults to avoid ordering issues and avoid blocking kernel binds on
unrelated copies / clears.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Link: https://patch.msgid.link/20260925201526.1355279-13-matthew.brost@intel.com
|
|
The level was previously extracted from struct xe_pt inside
xe_vm_pgtable_update during CPU binds, which always occurred during the
bind IOCTL. With CPU binds now supported in bind jobs, struct xe_pt may
no longer be valid in memory at that point. To address this, store the
level directly in struct xe_vm_pgtable_update.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Stuart Summers <stuart.summers@intel.com>
Reviewed-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Link: https://patch.msgid.link/20260925201526.1355279-12-matthew.brost@intel.com
|
|
PT jobs bypass GPU execution for the final step of a bind job, using the
CPU to program the required page tables. Teach the GuC submission backend
how to execute these jobs.
PT job submission is implemented in the GuC backend for simplicity. A
follow-up patch could introduce a dedicated backend for PT jobs.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Reviewed-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Link: https://patch.msgid.link/20260925201526.1355279-11-matthew.brost@intel.com
|
|
Add struct xe_pt_job_ops, a dynamically refcounted object that contains
the information required to issue a CPU bind via a job after the initial
bind IOCTL returns.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Link: https://patch.msgid.link/20260925201526.1355279-10-matthew.brost@intel.com
|
|
Add helpers to access PT ops, making it easier to shuffle the location of
the ops structures without requiring widespread code changes.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Link: https://patch.msgid.link/20260925201526.1355279-9-matthew.brost@intel.com
|
|
Update the scheduler job layer to support PT jobs. PT jobs are executed
entirely on the CPU and do not require LRC fences or a batch address.
Repurpose the LRC fence storage to hold PT‑job arguments and update the
scheduler job layer to distinguish between PT jobs and jobs that require
an LRC.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Reviewed-by: Francois Dugast <francois.dugast@intel.com>
Link: https://patch.msgid.link/20260925201526.1355279-8-matthew.brost@intel.com
|
|
Update the xe_bo_put_deferred arguments to include a writeback flag,
which indicates whether the BO was added to the deferred list. This is
useful when the caller needs to take additional actions after the BO has
been queued for deferred release.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Link: https://patch.msgid.link/20260925201526.1355279-7-matthew.brost@intel.com
|
|
Add the job count to the GuC exec queue snapshot, as this is useful
debug information.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Stuart Summers <stuart.summers@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Link: https://patch.msgid.link/20260925201526.1355279-6-matthew.brost@intel.com
|
|
We already maintain a job count for each exec queue, so simplify the idle
check to rely on the job count rather than the LRC state. This decouples
exec queues from LRC-based backends and avoids unnecessarily coupling idle
detection to backend-specific implementation details.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Stuart Summers <stuart.summers@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Link: https://patch.msgid.link/20260925201526.1355279-5-matthew.brost@intel.com
|
|
Add the xe_migrate_update_pgtables_cpu_execute helper, which performs
the CPU-side page-table update. This will support implementing CPU
binds, as the submission backend can call this helper once a bind job’s
dependencies are resolved. While here, add assertions to provide basic
sanity checks on tht function arguments.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Link: https://patch.msgid.link/20260925201526.1355279-4-matthew.brost@intel.com
|
|
Remove the xe_migrate_pt_update argument from the populate and clear
vfuns. This structure will not be available in run_job, where CPU binds
will be implemented. The populate path no longer needs it, and the clear
path already uses the VM field instead.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Link: https://patch.msgid.link/20260925201526.1355279-3-matthew.brost@intel.com
|
|
PT BOs share the VM's dma-resv and therefore must keep the VM alive
until all PT BOs are destroyed.
Take a VM reference from PT BOs to ensure the shared dma-resv remains
valid for the lifetime of the BOs. Scope the change to PT BOs rather
than all kernel BOs to minimize risk and avoid extending VM lifetimes
unnecessarily.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Reviewed-by: Francois Dugast <francois.dugast@intel.com>
Link: https://patch.msgid.link/20260925201526.1355279-2-matthew.brost@intel.com
|
|
Support in-band negotiation of the SerDes ports.
Signed-off-by: David Yang <mmyangfl@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://patch.msgid.link/20260923021945.2203368-9-mmyangfl@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Split it in order to be used in other modules.
Signed-off-by: David Yang <mmyangfl@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://patch.msgid.link/20260923021945.2203368-8-mmyangfl@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
yt921x_poll_mib() open-codes the same container_of_const() expression
that leds.c defines as to_yt921x_priv(). Move the macro to chip.h as
yt921x_port_to_priv() so both files use one definition.
Signed-off-by: David Yang <mmyangfl@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://patch.msgid.link/20260923021945.2203368-7-mmyangfl@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
FIELD_PREP values cannot be used as case labels. Introduce enum for
future use.
Signed-off-by: David Yang <mmyangfl@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://patch.msgid.link/20260923021945.2203368-6-mmyangfl@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Review and fix the incorrectness in the bit field names. No functional
change.
Signed-off-by: David Yang <mmyangfl@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://patch.msgid.link/20260923021945.2203368-5-mmyangfl@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
The yt921x_port_is_internal/external() macros hardcoded the port layout:
they treat ports 8 and 9 as external on every chip and count the dummy
ports as internal.
Identify the actual port capabilities using chip info and track
configuration of external ports (although only SerDes is supported now).
This is a hardening and cleanup step ahead of the SerDes PCS support, no
behavior changes intended.
Signed-off-by: David Yang <mmyangfl@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://patch.msgid.link/20260923021945.2203368-4-mmyangfl@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Replace external_mask with xmii_mask and serdes_mask, and advertise the
interface modes it actually supports.
Also move struct yt921x_info into chip.h for future use.
Signed-off-by: David Yang <mmyangfl@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://patch.msgid.link/20260923021945.2203368-3-mmyangfl@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Remove YT921X_PORT_MASK_* macros since they are only thin wrappers
around BIT() and GENMASK() and are not used in other places (plain BIT()
and GENMASK() are used there). No functional change.
Signed-off-by: David Yang <mmyangfl@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://patch.msgid.link/20260923021945.2203368-2-mmyangfl@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
The current PCS code unconditionally uses SGMII autonegotiation, though
the hardware supports both SGMII and 1000BASE-X modes. Decouple the
choice of PCS enablement from use of the SGMII mode when running at
gigabit rates and announce to phylink that 1000BASE-X is a supported
operating mode. This enables direct attachment of the PCS to e.g. an
SFP.
The 1000BASE-X code in phylink also sometimes calls the autonegotiation
restart method, so add an implementation of autonegotiation restart and
make sure that pcs_config() signals to phylink when the AN advertisement
has changed.
Signed-off-by: Nathan Whitehorn <nwhitehorn@pa.msu.edu>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://patch.msgid.link/20260922151429.23272-3-nwhitehorn@pa.msu.edu
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
When used with an offboard transceiver (e.g. an SFP), link state needs
to be measured at the PCS rather than an onboard PHY. Poll the link
state when the PCS is active to allow the kernel to detect link state
changes in this case.
Signed-off-by: Nathan Whitehorn <nwhitehorn@pa.msu.edu>
Reviewed-by: Théo Lebrun <theo.lebrun@bootlin.com>
Link: https://patch.msgid.link/20260922151429.23272-2-nwhitehorn@pa.msu.edu
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
idt82p33_ptp_clock_unregister_all() cancels each channel's adjtime work
before unregistering its PTP clock. An adjustment callback already in
progress can schedule that work after cancellation returns. Device
removal then frees the channel while its delayed work remains queued,
causing a use-after-free.
Mark the device as stopping under its lock before draining work. Reject
new adjustment and EXTS requests, clear the EXTS polling mask, and
synchronously drain the EXTS and channel adjustment work before the
device storage is released.
With an i2c-stub device and kprobe-controlled callback timing, three
runs of the unmodified driver reported KASAN use-after-free after
unbind. Three runs with this change completed without a kernel
diagnostic. No physical IDT82P33 device was tested.
Cc: stable+noautosel@kernel.org # untested fix to a driver init path race
Signed-off-by: Yibo Tan <lhfff@tju.edu.cn>
Link: https://patch.msgid.link/20260922113601.855416-1-lhfff@tju.edu.cn
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
While driver advertises gPTP support on Gen4 platforms it is in fact
completely broken. On R-Car Gen4 devices the RAVB module have no
internal gPTP clock as generations before it. Instead it utilizes a
system-wide gPTP clock.
This change utilizes the refactoring of the RAVB gPTP code to add
support for a system-wide clock and stops the Gen4 devices trying to use
the non-existing internal gPTP clock.
To remain backward compatible the ptp-timer property is optional.If not
present, or not enabled, the RAVB driver will no longer advertise gPTP
support to user-space.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Reviewed-by: Sergey Shtylyov <sergei.shtylyov@gmail.com>
Link: https://patch.msgid.link/20260921175037.4084310-11-niklas.soderlund+renesas@ragnatech.se
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Prepare for adding Gen4 support which have an optional external gPTP
clock. Add a callback to get the clock index and use it to determine if
the device shall report gPTP support.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Reviewed-by: Sergey Shtylyov <sergei.shtylyov@gmail.com>
Link: https://patch.msgid.link/20260921175037.4084310-9-niklas.soderlund+renesas@ragnatech.se
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Different generations of the RAVB IP have different needs when it probes
the gPTP timer clock. Add a callback in the PTP information to allow
each generation to probe its own way.
With this the last gPTP specific flag (gptp_ref_clk) can be removed.
However the primary motivation for the change is to prepare for Gen4
support, which compared to other generations with gPTP support does not
have the clock as part of the IP itself.
Gen4 will not need to compute GTI value as it have nowhere to write it,
as the gPTP clock is external. For this reason move the computation of
it into the newly gPTP probe specific callbacks for the RAVB IP's that
support it.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Reviewed-by: Sergey Shtylyov <sergei.shtylyov@gmail.com>
Link: https://patch.msgid.link/20260921175037.4084310-8-niklas.soderlund+renesas@ragnatech.se
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Prepare for adding Gen4 support which will add a third and new way to
interact with the gPTP clock by replacing the flags for Gen2 behavior
(info->gptp) and Gen3 behavior (info->ccc_gac) with callbacks.
This will make adding Gen4 support cleaner as the code will not have "if
else if else" sprinkled all over to handle each generations special
cases.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Reviewed-by: Sergey Shtylyov <sergei.shtylyov@gmail.com>
Link: https://patch.msgid.link/20260921175037.4084310-7-niklas.soderlund+renesas@ragnatech.se
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
If the PTP clock registration failed the error is silently ignored.
Before reworking all callers of ravb_ptp_init() handle and propagate the
error so it can be used.
Link: https://sashiko.dev/#/patchset/20260610102432.3538432-2-niklas.soderlund+renesas@ragnatech.se?part=1
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Reviewed-by: Sergey Shtylyov <sergei.shtylyov@gmail.com>
Link: https://patch.msgid.link/20260921175037.4084310-6-niklas.soderlund+renesas@ragnatech.se
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
There is no need to explicitly pass the struct platform_device pointer
to ravb_ptp_init(), it can retrieve it directly from the private data
structure.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Reviewed-by: Sergey Shtylyov <sergei.shtylyov@gmail.com>
Link: https://patch.msgid.link/20260921175037.4084310-5-niklas.soderlund+renesas@ragnatech.se
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
For devices that do not support the gPTP clock in config mode the
somewhat oddly named flag gptp is set, compared to devices that do
support the gPTP clock in config and operation mode where the flag
ccc_gac is set instead. The two flags are mutually exclusive.
For the gptp-flag devices (Gen2) the clock is tied to the AVB-DMAC, when
it is stopped so is the gPTP clock. For ccc_gac-flag devices (Gen3) the
gPTP clock is available whenever the ndev is open.
Prepare to add Gen4 support which will add a third way by cleaning the
Gen2 and Gen3 cases up a bit.
Fold the gptp-flag start and stop calls into ravb_dmac_init() and
ravb_stop_dma(), which start and stop the AVB-DMAC. There are no
functional change as all call sites to the construct:
if (info->gptp)
ravb_ptp_init(ndev, priv->pdev);
are always just after a call to into ravb_dmac_init() and all call sites
to the construct:
if (info->gptp)
ravb_ptp_stop(ndev);
are always directly followed by a call to ravb_stop_dma().
There are two special cases where the calling construct covers both the
gptp-flag and info->ccc_gac devices, one for start and one for stop. The
condition that it is preceded by a call to ravb_dmac_init(), or followed
by a call to ravb_stop_dma() are however true for them too. Reworked the
two special cases to drop the check of info->gptp.
The end result is that the gPTP clock will be started or stopped for the
gptp-flag devices in tandem with the AVB-DMAC, while the info->ccc_gac
devices will be controlled, as before, when the ndev is opened or
closed.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Reviewed-by: Sergey Shtylyov <sergei.shtylyov@gmail.com>
Link: https://patch.msgid.link/20260921175037.4084310-4-niklas.soderlund+renesas@ragnatech.se
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Commit f384ab481cab ("net: ravb: Split GTI computation and set
operations") broke apart the operations of computing the timer interval
and programming of it. However it kept the programming of the interval
in the RAVB main logic.
Having split the two apart this can be improved further by moving the
programming to the gPTP initialization function, as the first action of
the gPTP init function is to wait for the timer interval programming to
be acknowledge by the hardware.
As an added bonus the interaction with the gPTP registers for the
programming can then also be done while holding the gPTP registers lock.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Reviewed-by: Sergey Shtylyov <sergei.shtylyov@gmail.com>
Link: https://patch.msgid.link/20260921175037.4084310-3-niklas.soderlund+renesas@ragnatech.se
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Since commit a6a85ba36fd0 ("net: ravb: Move PTP initialization in the
driver's ndo_open API for ccc_gac platorms") the gPTP clock (if
supported) is stopped and started by opening and closing the ndev.
This makes the special case to stop and start it when resuming from WoL
redundant. As the ndev will always be closed and re-opened when
resuming the system.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Reviewed-by: Sergey Shtylyov <sergei.shtylyov@gmail.com>
Link: https://patch.msgid.link/20260921175037.4084310-2-niklas.soderlund+renesas@ragnatech.se
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Initialize a per-device RSS key in nsim_ethtool_init() using
netdev_rss_key_fill() and wire up get_rxfh_key_size, get_rxfh_indir_size,
get_rxfh, and get_rxfh_fields on nsim_ethtool_ops.
This allows RSS selftests under tools/testing/selftests/drivers/net/ to
exercise both /proc/sys/net/core/netdev_rss_key and ethtool netlink RSS
reporting against netdevsim without requiring physical hardware.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20260922163458.3900996-4-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Each kernel make invocation begins with ~30 compiler and linker runs each
of which performs duplicate probe for a number of compiler and linker
options.
This is useless work - the compiler and its version is known, so use these
to determine which options are available, once.
Using the newly introduced cc-option-str, track the result of each probe in
CC_OPT_*, LD_OPT_* variables and have Makefile's simply append the symbol.
Some care has to be taken around negative options, i.e. -Wno-xxx, as gcc
accepts an unknown -Wno- without complaint and only complains later, so
instead probe the positive form for these cases.
Use CC_HAS_* For those probes that gate a Makefile conditional rather than
a simple option, as per the established pattern for this.
This change avoids a subtle issue with ppc32 (big endian) - clang doesn't
accept -fno-stack-clash-protection on this sub-target, so keep this option
in the Makefile.
With gcc and clang, defconfig and allmodconfig, the recorded command
lines are unchanged, a build with nothing to do rebuilds nothing and W=1
continues to function correctly.
Doing this improves all builds, but has a particularly positive impact on
no-op builds (builds where nothing has changed).
Whole build, 128-thread Threadripper 9980X, best of N runs:
before after delta
-------------------------------
x86 defconfig, no-op make, gcc 0.78s 0.60s -0.18s (-23%)
x86 defconfig, no-op make, clang 0.94s 0.66s -0.28s (-30%)
x86 defconfig, touch mm/vma.c, gcc 8.6s 8.5s -0.16s (-2%)
x86 defconfig, touch mm/vma.c, clang 8.0s 7.7s -0.24s (-3%)
x86 defconfig, clean, clang 29.8s 29.4s -0.37s (-1%)
x86 allmodconfig, no-op make, gcc 12.9s 12.7s -0.23s (-2%)
x86 allmodconfig, no-op make, clang 14.2s 13.9s -0.29s (-2%)
x86 allmodconfig, touch mm/vma.c, clang 32.2s 31.9s -0.30s (-1%)
Assisted-by: LLM
Signed-off-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>
Link: https://patch.msgid.link/20260923-build-speedup-v4-12-73128809a4a4@kernel.org
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
|
|
Pull drm fixes from Dave Airlie:
"While most of this is AI inspired fixes for error handling paths,
leaks and use after frees, there are some normal things.
nouveau has probably the biggest changes with some fixes to stabilise
runtime suspend/resume on 570 firmware which regressed after we moved
from 535, there are some fixes to stackframe issues seen with amdgpu,
and otherwise the usual bunch of i915/xe/amdgpu fixes, and some
virtio-gpu fixes.
Hopefully it will start to quiten down a bit from here.
client:
- fix restore of partially initialized client
i915:
- Fix incorrect RCU teardown order leading to endless loop
- Fix DP MST TU and FEC handling for disconnected streams
- Fix selective fetch disable, again
- Fix export namespace for kunit helpers
- Workaround eDP flicker on a specific laptop model
xe:
- CRI throttle reasons report
- TLB invalidation at wedge
- SVM eviction and VM close
- Display corruption on LNL on Xen PV
- W/a fix and addition
amdgpu:
- Display ref count fix
- Userq fixes
- VCN 4, 5 reset fixes
- Fixes for various error paths
- Stack frame size fixes for various combinations of compilers and
configs
amdkfd:
- Possible UAF fix
nouveau:
- runtime suspend/resume fixes for newer firmware
- rcu free the scheduler
- fix VRAM pinning
- fix double free
- fix reference leaks
- fix runtime PM leak
- fix cursor list usage problems
- fix HDMI config rejection without SCDC
virtio:
- fix a bunch of object/memory leaks in failure paths
- add pixel blend mode property to cursor plane
- revert prime buffers import
- sync shmem backing on guest transfers
imagination:
- propogate map failures properly
- fix page count in map interface
- clamp freelist reconstruction requests
ivpu:
- use separate flag for job timeout
bridge:
- samsung-dsim: fix TE GPIO lifetime for host attach"
* tag 'drm-fixes-2026-09-26' of https://gitlab.freedesktop.org/drm/kernel: (60 commits)
drm/amd/display: Bump frame warning limit for all builds of dml
drm/imagination: clamp freelist reconstruction requests
drm/imagination: Fix page count for page table for map() interface
drm/imagination: Propagate map failures correctly from pvr_mmu_map_sgl()
drm/amd/display: Bump frame warning limit for clang builds of dml
drm/amd/display: Relax DML frame limit with UBSAN
drm/amdgpu: Fix runtime PM leak in amdgpu_debugfs_test_ib_show()
drm/amdgpu: Fix last_update fence leak in amdgpu_vm_init()
drm/amdgpu: Fix acpi device leak in amdgpu_acpi_enumerate_xcc()
drm/amdgpu: Fix vmid_wait fence leak in amdgpu_ring_init()
drm/amdkfd: fix use-after-free and multi-container gap in kfd_dev_mapping
drm/amdgpu/vcn4.0.3: fix video_timeout unit mismatch in jpeg reset wait
drm/amdgpu/vcn5.0.1: fix video_timeout unit mismatch in jpeg reset wait
drm/amdgpu/userq: fix double jiffies conversion in hang detect timeout
drm/amdgpu: move userq fence wait out of signalling section
drm/amd/display: Fix dc stream excess put in dm_update_crtc_state()
drm/xe: Add wa_14025941587 to xe2, xe3 and xe3p platforms
drm/xe: harden adjust_idledly() against divide-by-zero and overflow
drm/xe: Limit sg segment size to PAGE_SIZE on Xen PV
drm/i915: fix incorrect RCU teardown order
...
|
|
https://gitlab.freedesktop.org/drm/amdgpu/kernel into drm-next
amd-drm-next-7.4-2026-09-24:
amdgpu:
- Display ref count fix
- Userq fixes
- VCN 4, 5 reset fixes
- Fixes for various error paths
- Stack frame size fixes for various combinations of compilers and configs
- GC 12.1 fixes
- RAS updates
- SR-IOV udpates
- MES fixes
- UALink updates
- ABM updates
- DCN 6 updates
- Rework display instance handling
- KUnit test updates
- Documentation fixes
- DCN 4.2 updates
- Freesync updates
- Cursor refactor
- amdgpu_vm_eviction_(try)lock cleanup
amdkfd:
- Possible UAF fix
- GC 12.1 trap fixes
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patch.msgid.link/20260924174339.635232-1-alexander.deucher@amd.com
|
|
https://gitlab.freedesktop.org/drm/misc/kernel into drm-fixes
A number of fixes:
- bridge:
- samsung-dsim: fix GPIO lifetime
- client: Null pointer dereference fix
- imagination: error handling fix, page handling fix
- nouveau: fix reference leaks, double-frees, out-of-bounds accesses,
use-after-frees, don't reject config without SCDC, a number of
workarounds
- virtio: fix memory leak, reference leaks, null pointer dereference,
add pixel blend mode, cache coherency fix
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Maxime Ripard <self@mripard.dev>
Link: https://patch.msgid.link/arU22zzqUGDEco1y@houat
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Pull SCSI fixes from James Bottomley:
"Mostly small driver fixes. The biggest fix is the one to the block
zone handling which might trip for real or virtual hardware if the
number of zones is > 2^32"
* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
scsi: megaraid_sas: Protect megasas_get_ctrl_info() in megasas_resume()
scsi: sd_zbc: Reject disks with too many zones
scsi: block: Fix zones_cond out-of-bounds write on zone report
scsi: leapraid: Avoid -Wformat-security warning
scsi: devinfo: Add BLIST_SKIP_IO_HINTS for EMC Symmetrix
scsi: libiscsi_tcp: Check the data direction of a Data-In PDU
scsi: ufs: pltfrm: Add quirk for R-Car S4 lacking lanes-per-direction
scsi: ufs: core: Keep internal commands dispatchable during error handling
|
|
The GCC and Clang attribute '__counted_by_ptr' associates a pointer
field with an integer field holding its element count. This enables
runtime bounds checking by KASAN and UBSAN to prevent out-of-bounds
accesses to the pointer.
In 'struct fw_device' (defined in 'include/linux/firewire.h'), the
field 'config_rom' is a pointer to the device's Configuration ROM data,
and its associated element count is stored in 'config_rom_length'.
Additionally, update the existing KUnit test in
'drivers/firewire/device-attribute-test.c' where 'config_rom_length'
was incorrectly initialized with the byte size
('sizeof(simple_avc_config_rom)') instead of the element count
('ARRAY_SIZE(simple_avc_config_rom)'). This ensures the
'__counted_by_ptr' bounds-checking annotation does not trigger any
false-positive panics or compile/runtime checks.
Cc: codemender-patching+linux@google.com
Assisted-by: LLM
Signed-off-by: Bill Wendling <morbo@google.com>
Link: https://lore.kernel.org/r/20260925212552.125652-1-morbo@google.com
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
|
|
The R-Car Gen4 PCIe controller does contain core reset, which puts the
controller back into well-defined state, but it does also tear down the
PCIe link. This is called Application/Local register reset in the
documentation.
Deassert the core reset after the core clock has been enabled, and assert
the core reset in case of error or when shutting down the controller.
This also fixes reinitialization failure of the controller, where if the
controller driver is unbound and bound again via sysfs attributes, the
controller driver will fail to probe the second time because it fails to
access the controller DBI, which returns all zeroes unless the core reset
is toggled. The following commands trigger this fault on R-Car V4H:
$ echo e65d0000.pcie > /sys/bus/platform/drivers/pcie-rcar-gen4/unbind
$ echo e65d0000.pcie > /sys/bus/platform/drivers/pcie-rcar-gen4/bind
Fixes: faf5a975ee3b ("PCI: rcar-gen4: Add support for R-Car V4H")
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
[mani: added fixes tag and cced noautosel]
Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@oss.qualcomm.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: <stable+noautosel@kernel.org> # DTS dependency
Link: https://patch.msgid.link/20260903205153.283553-3-marek.vasut+renesas@mailbox.org
|
|
The PECI interrupt handler uses the lock and completion object as soon as
the interrupt is delivered. Initialize both before registering the
handler so an early interrupt cannot access uninitialized state.
Fixes: 3e16184a1bd8 ("peci: Add peci-npcm controller driver")
Cc: stable@vger.kernel.org
Assisted-by: Codex:GPT-5
Signed-off-by: Runyu Xiao <runyu.xiao@seu.edu.cn>
Reviewed-by: Iwona Winiarska <iwona.winiarska@intel.com>
Link: https://lore.kernel.org/r/20260902152256.3591658-1-runyu.xiao@seu.edu.cn
Signed-off-by: Iwona Winiarska <iwona.winiarska@intel.com>
|
|
The PECI interrupt handler uses the lock and completion object as soon as
the interrupt is delivered. Initialize both before registering the
handler so an early interrupt cannot access uninitialized state.
Fixes: a85e4c52086c ("peci: Add peci-aspeed controller driver")
Cc: stable@vger.kernel.org
Assisted-by: Codex:GPT-5
Signed-off-by: Runyu Xiao <runyu.xiao@seu.edu.cn>
Reviewed-by: Iwona Winiarska <iwona.winiarska@intel.com>
Link: https://lore.kernel.org/r/20260902142600.3580382-1-runyu.xiao@seu.edu.cn
Signed-off-by: Iwona Winiarska <iwona.winiarska@intel.com>
|
|
The clk_init_data structure contains several mutually-exclusive members
for different methods to specify the possible parents of a clock,
prompting drivers to initialize only the members they need. However,
not initializing all members may cause subtle issues, which are only
exposed when CONFIG_INIT_STACK_ALL_PATTERN or CONFIG_INIT_STACK_NONE is
enabled.
Make sure all members are fully initialized, to avoid such bugs, and to
prevent future breakage when converting drivers to a different method
for specifying the parents.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Iwona Winiarska <iwona.winiarska@intel.com>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Link: https://lore.kernel.org/r/48aa77c45e109f78ccd806bc3b7bf380c5bc5e71.1787240145.git.geert+renesas@glider.be
Signed-off-by: Iwona Winiarska <iwona.winiarska@intel.com>
|
