Call drm_mode_size_dumb() to compute dumb-buffer scanline pitch and
buffer size. The hardware requires the framebuffer width to be a
multiple of 8. The scanline pitch has to be large enough to support
this. Therefore compute the byte size of 8 pixels in the given color
mode and align the pitch accordingly.
v5:
- fix typo in commit description
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Cc: Philipp Zabel <p.zabel@pengutronix.de>
Cc: Shawn Guo <shawnguo@kernel.org>
Cc: Sascha Hauer <s.hauer@pengutronix.de>
Cc: Pengutronix Kernel Team <kernel@pengutronix.de>
Cc: Fabio Estevam <festevam@gmail.com>
Link: https://lore.kernel.org/r/20250821081918.79786-11-tzimmermann@suse.de
Call drm_mode_size_dumb() to compute dumb-buffer scanline pitch and
buffer size. Align the pitch to a multiple of 8.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Link: https://lore.kernel.org/r/20250821081918.79786-5-tzimmermann@suse.de
Call drm_mode_size_dumb() to compute dumb-buffer scanline pitch and
buffer size. Align the pitch to a multiple of 8.
Push the current calculation into the only direct caller imx. Imx's
hardware requires the framebuffer width to be aligned to 8. The
driver's current approach is actually incorrect, as it only guarantees
this implicitly and requires bpp to be a multiple of 8 already. A
later commit will fix this problem by aligning the scanline pitch
such that an aligned width still fits into each scanline's memory.
A number of other drivers are build on top of gem-dma helpers and
implement their own dumb-buffer allocation. These drivers invoke
drm_gem_dma_dumb_create_internal(), which is not affected by this
commit.
v5:
- avoid reset of arguments (Tomi)
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Link: https://lore.kernel.org/r/20250821081918.79786-4-tzimmermann@suse.de
Add drm_modes_size_dumb(), a helper to calculate the dumb-buffer
scanline pitch and allocation size. Implementations of struct
drm_driver.dumb_create can call the new helper for their size
computations.
There is currently quite a bit of code duplication among DRM's
memory managers. Each calculates scanline pitch and buffer size
from the given arguments, but the implementations are inconsistent
in how they treat alignment and format support. Later patches will
unify this code on top of drm_mode_size_dumb() as much as possible.
drm_mode_size_dumb() uses existing 4CC format helpers to interpret
the given color mode. This makes the dumb-buffer interface behave
similar the kernel's video= parameter. Current per-driver implementations
again likely have subtle differences or bugs in how they support color
modes.
The dumb-buffer UAPI is only specified for known color modes. These
values describe linear, single-plane RGB color formats or legacy index
formats. Other values should not be specified. But some user space
still does. So for unknown color modes, there are a number of known
exceptions for which drm_mode_size_dumb() calculates the pitch from
the bpp value, as before. All other values work the same but print
an error.
v6:
- document additional use cases for DUMB_CREATE2 in TODO list (Tomi)
- fix typos in documentation (Tomi)
v5:
- check for overflows with check_mul_overflow() (Tomi)
v4:
- use %u conversion specifier (Geert)
- list DRM_FORMAT_Dn in UAPI docs (Geert)
- avoid dmesg spamming with drm_warn_once() (Sima)
- add more information about bpp special case (Sima)
- clarify parameters for hardware alignment
- add a TODO item for DUMB_CREATE2
v3:
- document the UAPI semantics
- compute scanline pitch from for unknown color modes (Andy, Tomi)
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Link: https://lore.kernel.org/r/20250821081918.79786-3-tzimmermann@suse.de
The ioctls MODE_CREATE_DUMB and MODE_MAP_DUMB return results into a
memory buffer supplied by user space. On errors, it is possible that
intermediate values are being returned. The exact semantics depends
on the DRM driver's implementation of these ioctls. Although this is
most-likely not a security problem in practice, avoid any uncertainty
by clearing the memory to 0 on errors.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Link: https://lore.kernel.org/r/20250821081918.79786-2-tzimmermann@suse.de
Remove superfluous newlines from inline assemblies. Compilers use the
number of lines of inline assemblies as heuristic for the complexity
and inline decisions. Therefore inline assemblies should only contain
as many lines as required.
A lot of inline assemblies contain a superfluous newline for the last
line. Remove such newlines to improve compiler inlining decisions.
Suggested-by: Juergen Christ <jchrist@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Reviewed-by: Alexander Gordeev <agordeev@linux.ibm.com>
Reviewed-by: Juergen Christ <jchrist@linux.ibm.com>
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
For the __xsch() inline assembly the conversion to flag output macros is
incomplete. Only the conditional shift of the return value was added, while
the required changes to the inline assembly itself are missing.
If compiled with GCC versions before 14.2 this leads to a double shift of
the cc output operand and therefore the returned value of __xsch() is
incorrectly always zero, instead of the expected condition code.
Fixes: e200565d43 ("s390/cio/ioasm: Convert to use flag output macros")
Cc: stable@vger.kernel.org
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Acked-by: Alexander Gordeev <agordeev@linux.ibm.com>
Reviewed-by: Juergen Christ <jchrist@linux.ibm.com>
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
If the decompressor is compiled with clang this can lead to the following
warning:
In file included from arch/s390/boot/startup.c:4:
...
In file included from ./include/linux/pgtable.h:6:
./arch/s390/include/asm/pgtable.h:2065:48: warning: passing 'unsigned long *' to parameter of type
'long *' converts between pointers to integer types with different sign [-Wpointer-sign]
2065 | value = __atomic64_or_barrier(PGSTE_PCL_BIT, ptr);
Add -Wno-pointer-sign to the decompressor compile flags, like it is also
done for the kernel. This is similar to what was done for x86 to address
the same problem [1].
[1] commit dca5203e3f ("x86/boot: Add -Wno-pointer-sign to KBUILD_CFLAGS")
Cc: stable@vger.kernel.org
Reported-by: Gerd Bayer <gbayer@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Reviewed-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
A buffer overflow arises from the usage of snprintf to write into the
buffer "buf" in target_lu_gp_members_show function located in
/drivers/target/target_core_configfs.c. This buffer is allocated with
size LU_GROUP_NAME_BUF (256 bytes).
snprintf(...) formats multiple strings into buf with the HBA name
(hba->hba_group.cg_item), a slash character, a devicename (dev->
dev_group.cg_item) and a newline character, the total formatted string
length may exceed the buffer size of 256 bytes.
Since snprintf() returns the total number of bytes that would have been
written (the length of %s/%sn ), this value may exceed the buffer length
(256 bytes) passed to memcpy(), this will ultimately cause function
memcpy reporting a buffer overflow error.
An additional check of the return value of snprintf() can avoid this
buffer overflow.
Reported-by: Wang Haoran <haoranwangsec@gmail.com>
Reported-by: ziiiro <yuanmingbuaa@gmail.com>
Signed-off-by: Wang Haoran <haoranwangsec@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The code in ast_main.c has been split into several helpers in
other source files. Delete the source file. With the generic
device init gone, fail probing on unknown hardware generations.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250922083708.45564-13-tzimmermann@suse.de
Split off device initialization for Gen7 hardware into the helpers
ast_2600_device_create() and ast_2600_detect_wide_screen(). The new
functions are duplicates of their counterparts in ast_main.c, but
stripped from most non-Gen7 support.
Simplifies maintenance as the driver's number of supported hardware
generations grows.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250922083708.45564-12-tzimmermann@suse.de
Split off device initialization for Gen6 hardware into the helpers
ast_2500_device_create() and ast_2500_detect_wide_screen(). The new
functions are duplicates of their counterparts in ast_main.c, but
stripped from most non-Gen6 support.
Simplifies maintenance as the driver's number of supported hardware
generations grows.
v2:
- simplify widescreen-detection logic
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250922083708.45564-11-tzimmermann@suse.de
Split off device initialization for Gen5 hardware into the helpers
ast_2400_device_create() and ast_2400_detect_wide_screen(). The new
functions are duplicates of their counterparts in ast_main.c, but
stripped from most non-Gen5 support.
Simplifies maintenance as the driver's number of supported hardware
generations grows.
v2:
- simplify widescreen-detection logic
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250922083708.45564-10-tzimmermann@suse.de
Split off device initialization for Gen4 hardware into the helpers
ast_2300_device_create() and ast_2300_detect_wide_screen(). The new
functions are duplicates of their counterparts in ast_main.c, but
stripped from most non-Gen4 support.
Simplifies maintenance as the driver's number of supported hardware
generations grows.
v2:
- simplify widescreen-detection logic
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250922083708.45564-9-tzimmermann@suse.de
Split off device initialization for Gen3 hardware into the helpers
ast_2200_device_create() and ast_2200_detect_wide_screen(). The new
functions are duplicates of their counterparts in ast_main.c, but
stripped from most non-Gen3 support.
Simplifies maintenance as the driver's number of supported hardware
generations grows.
v2:
- simplify widescreen-detection logic
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250922083708.45564-8-tzimmermann@suse.de
Split off device initialization for Gen2 hardware into the helpers
ast_2100_device_create() and ast_2100_detect_wide_screen(). The new
functions are duplicates of their counterparts in ast_main.c, but
stripped from most non-Gen2 support.
Simplifies maintenance as the driver's number of supported hardware
generations grows.
v2:
- simplify widescreen-detection logic (Jocelyn)
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250922083708.45564-7-tzimmermann@suse.de
Split off device initialization for Gen1 hardware into the helper
ast_2000_device_create(). The new function is a duplicate of their
counterpart in ast_main.c, but stripped from most non-Gen1 support.
Simplifies maintenance as the driver's number of supported hardware
generations grows.
v2:
- remove unnecessary widescreen-detection logic
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250922083708.45564-6-tzimmermann@suse.de
Switch device creation by hardware Gen. Return the value from the
call to ast_detect_chip(). All generations are still initialized
by ast_device_create().
Also add ast_device_init() for setting some common fields in struct
ast_device.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250922083708.45564-5-tzimmermann@suse.de
Gen4 and later models detect the TX chip from VGACRD1, while earlier
models detect from VGACRA3. Split up the detection helper into
two separate helpers. Use SZ_ constants instead of plain numbers.
Then inline the call into its only caller ast_device_create(). When
ast_device_create() gets split up per Gen, either call will remain.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250922083708.45564-4-tzimmermann@suse.de
Wide-screen modes are only available on Gen2 and later. Move the
detection helpers to the appropriate source file.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250922083708.45564-3-tzimmermann@suse.de
Move display-clock tables to the appropriate per-Gen source files.
The tables are almost identical, except for mode entries 0x17 and
0x1a. Rename to tables to match common style.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/r/20250922083708.45564-2-tzimmermann@suse.de
Merge power management utilities updates for 6.18-rc1:
- Fix and clean up the x86_energy_perf_policy utility and update its
documentation (Len Brown, Kaushlendra Kumar)
- Fix incorrect sorting of PMT telemetry in turbostat (Kaushlendra
Kumar)
- Fix incorrect size in cpuidle_state_disable() and the error return
value of cpupower_write_sysfs() in cpupower (Kaushlendra Kumar)
* pm-tools:
tools/power x86_energy_perf_policy.8: Emphasize preference for SW interfaces
tools/power x86_energy_perf_policy: Add make snapshot target
tools/power x86_energy_perf_policy: Prefer driver HWP limits
tools/power x86_energy_perf_policy: EPB access is only via sysfs
tools/power x86_energy_perf_policy: Prepare for MSR/sysfs refactoring
tools/power x86_energy_perf_policy: Enhance HWP enable
tools/power x86_energy_perf_policy: Enhance HWP enabled check
tools/power x86_energy_perf_policy: Fix incorrect fopen mode usage
tools/power turbostat: Fix incorrect sorting of PMT telemetry
tools/cpupower: Fix incorrect size in cpuidle_state_disable()
tools/cpupower: fix error return value in cpupower_write_sysfs()
Merge changes related to system sleep and runtime PM framework for
6.18-rc1:
- Annotate loops walking device links in the power management core
code as _srcu and add macros for walking device links to reduce the
likelihood of coding mistakes related to them (Rafael Wysocki)
- Document time units for *_time functions in the runtime PM API (Brian
Norris)
- Clear power.must_resume in noirq suspend error path to avoid resuming
a dependant device under a suspended parent or supplier (Rafael
Wysocki)
- Fix GFP mask handling during hybrid suspend and make the amdgpu
driver handle hybrid suspend correctly (Mario Limonciello, Rafael
Wysocki)
- Fix GFP mask handling after aborted hibernation in platform mode and
combine exit paths in power_down() to avoid code duplication (Rafael
Wysocki)
- Use vmalloc_array() and vcalloc() in the hibernation core to avoid
open-coded size computations (Qianfeng Rong)
- Fix typo in hibernation core code comment (Li Jun)
- Call pm_wakeup_clear() in the same place where other functions that do
bookkeeping prior to suspend_prepare() are called (Samuel Wu)
* pm-core:
PM: core: Add two macros for walking device links
PM: core: Annotate loops walking device links as _srcu
* pm-runtime:
PM: runtime: Documentation: ABI: Document time units for *_time
* pm-sleep:
PM: hibernate: Combine return paths in power_down()
PM: hibernate: Restrict GFP mask in power_down()
PM: hibernate: Fix pm_hibernation_mode_is_suspend() build breakage
drm/amd: Fix hybrid sleep
PM: hibernate: Add pm_hibernation_mode_is_suspend()
PM: hibernate: Fix hybrid-sleep
PM: sleep: core: Clear power.must_resume in noirq suspend error path
PM: sleep: Make pm_wakeup_clear() call more clear
PM: hibernate: Fix typo in memory bitmaps description comment
PM: hibernate: Use vmalloc_array() and vcalloc() to improve code
Merge cpuidle and power capping changes for 6.18-rc1:
- Fail cpuidle device registration if there is one already to avoid
sysfs-related issues (Rafael Wysocki)
- Use sysfs_emit()/sysfs_emit_at() instead of sprintf()/scnprintf() in
cpuidle (Vivek Yadav)
- Fix device and OF node leaks at probe in the qcom-spm cpuidle driver
and drop unnecessary initialisations from it (Johan Hovold)
- Remove unnecessary address-of operators from the intel_idle cpuidle
driver (Kaushlendra Kumar)
- Rearrange main loop in menu_select() to make the code in that funtion
easier to follow (Rafael Wysocki)
- Convert values in microseconds to ktime using us_to_ktime() where
applicable in the intel_idle power capping driver (Xichao Zhao)
* pm-cpuidle:
cpuidle: Fail cpuidle device registration if there is one already
cpuidle: sysfs: Use sysfs_emit()/sysfs_emit_at() instead of sprintf()/scnprintf()
cpuidle: qcom-spm: drop unnecessary initialisations
cpuidle: qcom-spm: fix device and OF node leaks at probe
intel_idle: Remove unnecessary address-of operators
cpuidle: governors: menu: Rearrange main loop in menu_select()
* pm-powercap:
powercap: idle_inject: use us_to_ktime() where appropriate
Merge energy model management, OPP (operating performance points) and
devfreq updates for 6.18-rc1:
- Prevent CPU capacity updates after registering a perf domain from
failing on a first CPU that is not present (Christian Loehle)
- Add support for the cases in which frequency alone is not sufficient
to uniquely identify an OPP (Krishna Chaitanya Chundru)
- Use to_result() for OPP error handling in Rust (Onur Özkan)
- Add support for LPDDR5 on Rockhip RK3588 SoC to rockchip-dfi devfreq
driver (Nicolas Frattaroli)
- Fix an issue where DDR cycle counts on RK3588/RK3528 with LPDDR4(X)
are reported as half by adding a cycle multiplier to the DFI driver
in rockchip-dfi devfreq-event driver (Nicolas Frattaroli)
- Fix missing error pointer dereference check of regulator instance in
the mtk-cci devfreq driver probe and remove a redundant condition from
an if () statement in that driver (Dan Carpenter, Liao Yuanhong)
* pm-em:
PM: EM: Fix late boot with holes in CPU topology
* pm-opp:
OPP: Add support to find OPP for a set of keys
rust: opp: use to_result for error handling
* pm-devfreq:
PM / devfreq: rockchip-dfi: add support for LPDDR5
PM / devfreq: rockchip-dfi: double count on RK3588
PM / devfreq: mtk-cci: avoid redundant conditions
PM / devfreq: mtk-cci: Fix potential error pointer dereference in probe()
The display driver needs to get from the struct drm_device pointer to
the struct intel_display pointer. Currently, this depends on knowledge
of the struct drm_i915_private and struct xe_device definitions, but
we'd like to hide those definitions from display.
Require the struct drm_device and struct intel_display * members within
struct drm_i915_private and struct xe_device to be placed next to each
other, to be able to figure out the display pointer without knowledge of
the structures.
Use a generic dummy device structure to define the relative offsets of
the drm and display members, and add static assertions to ensure this
holds for both i915 and xe. Use the dummy structure to do the pointer
chase from struct drm_device * to struct intel_display *.
This requires moving the display member in struct xe_device after the
drm member.
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Ville Syrjala <ville.syrjala@linux.intel.com>
Suggested-by: Simona Vetter <simona.vetter@ffwll.ch>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20250926111032.1188876-1-jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Add i915_gem_stolen_node_alloc() and i915_gem_stolen_node_free(),
returning struct intel_stolen_node pointer. Make struct
intel_stolen_node an opaque pointer, with different implementations in
i915 and xe.
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://lore.kernel.org/r/3fe71bbb4c75ee86b4d129fafa3d4cd6526363f4.1758732183.git.jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
i915_gem_stolen.h simply defines intel_stolen_node as drm_mm_node. Make
struct intel_stolen_node an actual struct of its own right, and embed
struct drm_mm_node inside. This allow better unification between i915
and xe.
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://lore.kernel.org/r/36762f611566d81427e702369f4e8207ead5f26c.1758732183.git.jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
During initialization, the EDVD_COREx_VOLT_FREQ registers for some cores
are still at reset values and not reflecting the actual frequency. This
causes get calls to fail. Set all cores to their respective max
frequency during probe to initialize the registers to working values.
Suggested-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Aaron Kling <webgeek1234@gmail.com>
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
The original commit set all cores in a cluster to a shared policy, but
did not update set_target to apply a frequency change to all cores for
the policy. This caused most cores to remain stuck at their boot
frequency.
Fixes: be4ae8c194 ("cpufreq: tegra186: Share policy per cluster")
Signed-off-by: Aaron Kling <webgeek1234@gmail.com>
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Remove local variables from find_supply_names() and use .and_then() with
the more concise kernel::kvec![] macro, instead of KVec::with_capacity()
followed by .push() and Some().
No functional changes intended.
Signed-off-by: Thorsten Blum <thorsten.blum@linux.dev>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Make sure to drop the reference to the cci device taken by
of_find_device_by_node() on probe failure (e.g. probe deferral).
Fixes: 0daa47325b ("cpufreq: mediatek: Link CCI device to CPU")
Cc: Jia-Wei Chang <jia-wei.chang@mediatek.com>
Cc: Rex-BC Chen <rex-bc.chen@mediatek.com>
Signed-off-by: Johan Hovold <johan@kernel.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Reviewed-by: Chen-Yu Tsai <wenst@chromium.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Fixes incorrect return vale in cpupower_write_sysfs() error path
and passing incorrect size to cpuidle_state_write_file() while
writing status to disable file in cpuidle_state_disable().
-----BEGIN PGP SIGNATURE-----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=1VZt
-----END PGP SIGNATURE-----
Merge tag 'linux-cpupower-6.18-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux
Merge cpupower utility updates for 6.18-rc1 from Shuah Khan:
"Fixes incorrect return vale in cpupower_write_sysfs() error path
and passing incorrect size to cpuidle_state_write_file() while
writing status to disable file in cpuidle_state_disable()."
* tag 'linux-cpupower-6.18-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux: (1125 commits)
tools/cpupower: Fix incorrect size in cpuidle_state_disable()
tools/cpupower: fix error return value in cpupower_write_sysfs()
Linux 6.17-rc6
MAINTAINERS: Input: Drop melfas-mip4 section
USB: core: remove the move buf action
MAINTAINERS: Update the DMA Rust entry
erofs: fix long xattr name prefix placement
Revert "net: usb: asix: ax88772: drop phylink use in PM to avoid MDIO runtime PM wakeups"
hsr: hold rcu and dev lock for hsr_get_port_ndev
hsr: use hsr_for_each_port_rtnl in hsr_port_get_hsr
hsr: use rtnl lock when iterating over ports
wifi: nl80211: completely disable per-link stats for now
net: usb: asix: ax88772: drop phylink use in PM to avoid MDIO runtime PM wakeups
net: ethtool: fix wrong type used in struct kernel_ethtool_ts_info
selftests/bpf: Skip timer cases when bpf_timer is not supported
bpf: Reject bpf_timer for PREEMPT_RT
libceph: fix invalid accesses to ceph_connection_v1_info
PM: hibernate: Restrict GFP mask in hibernation_snapshot()
MAINTAINERS: add Phil as netfilter reviewer
netfilter: nf_tables: restart set lookup on base_seq change
...
From the cover letter [1]:
This patch set introduces kmalloc_nolock() which is the next logical
step towards any context allocation necessary to remove bpf_mem_alloc
and get rid of preallocation requirement in BPF infrastructure.
In production BPF maps grew to gigabytes in size. Preallocation wastes
memory. Alloc from any context addresses this issue for BPF and other
subsystems that are forced to preallocate too.
This long task started with introduction of alloc_pages_nolock(), then
memcg and objcg were converted to operate from any context including
NMI, this set completes the task with kmalloc_nolock() that builds on
top of alloc_pages_nolock() and memcg changes.
After that BPF subsystem will gradually adopt it everywhere.
Link: https://lore.kernel.org/all/20250909010007.1660-1-alexei.starovoitov@gmail.com/ [1]
This series adds an opt-in percpu array-based caching layer to SLUB.
It has evolved to a state where kmem caches with sheaves are compatible
with all SLUB features (slub_debug, SLUB_TINY, NUMA locality
considerations). The plan is therefore that it will be later enabled for
all kmem caches and replace the complicated cpu (partial) slabs code.
Note the name "sheaf" was invented by Matthew Wilcox so we don't call
the arrays magazines like the original Bonwick paper. The per-NUMA-node
cache of sheaves is thus called "barn".
This caching may seem similar to the arrays we had in SLAB, but there
are some important differences:
- deals differently with NUMA locality of freed objects, thus there are
no per-node "shared" arrays (with possible lock contention) and no
"alien" arrays that would need periodical flushing
- instead, freeing remote objects (which is rare) bypasses the sheaves
- percpu sheaves thus contain only local objects (modulo rare races
and local node exhaustion)
- NUMA restricted allocations and strict_numa mode is still honoured
- improves kfree_rcu() handling by reusing whole sheaves
- there is an API for obtaining a preallocated sheaf that can be used
for guaranteed and efficient allocations in a restricted context, when
the upper bound for needed objects is known but rarely reached
- opt-in, not used for every cache (for now)
The motivation comes mainly from the ongoing work related to VMA locking
scalability and the related maple tree operations. This is why VMA and
maple nodes caches are sheaf-enabled in the patchset.
A sheaf-enabled cache has the following expected advantages:
- Cheaper fast paths. For allocations, instead of local double cmpxchg,
thanks to local_trylock() it becomes a preempt_disable() and no atomic
operations. Same for freeing, which is otherwise a local double cmpxchg
only for short term allocations (so the same slab is still active on the
same cpu when freeing the object) and a more costly locked double
cmpxchg otherwise.
- kfree_rcu() batching and recycling. kfree_rcu() will put objects to a
separate percpu sheaf and only submit the whole sheaf to call_rcu()
when full. After the grace period, the sheaf can be used for
allocations, which is more efficient than freeing and reallocating
individual slab objects (even with the batching done by kfree_rcu()
implementation itself). In case only some cpus are allowed to handle rcu
callbacks, the sheaf can still be made available to other cpus on the
same node via the shared barn. The maple_node cache uses kfree_rcu() and
thus can benefit from this.
Note: this path is currently limited to !PREEMPT_RT
- Preallocation support. A prefilled sheaf can be privately borrowed to
perform a short term operation that is not allowed to block in the
middle and may need to allocate some objects. If an upper bound (worst
case) for the number of allocations is known, but only much fewer
allocations actually needed on average, borrowing and returning a sheaf
is much more efficient then a bulk allocation for the worst case
followed by a bulk free of the many unused objects. Maple tree write
operations should benefit from this.
- Compatibility with slub_debug. When slub_debug is enabled for a cache,
we simply don't create the percpu sheaves so that the debugging hooks
(at the node partial list slowpaths) are reached as before. The same
thing is done for CONFIG_SLUB_TINY. Sheaf preallocation still works by
reusing the (ineffective) paths for requests exceeding the cache's
sheaf_capacity. This is in line with the existing approach where
debugging bypasses the fast paths and SLUB_TINY preferes memory
savings over performance.
The above is adapted from the cover letter [1], which contains also
in-kernel microbenchmark results showing the lower overhead of sheaves.
Results from Suren Baghdasaryan [2] using a mmap/munmap microbenchmark
also show improvements.
Results from Sudarsan Mahendran [3] using will-it-scale show both
benefits and regressions, probably due to overall noisiness of those
tests.
Link: https://lore.kernel.org/all/20250910-slub-percpu-caches-v8-0-ca3099d8352c@suse.cz/ [1]
Link: https://lore.kernel.org/all/CAJuCfpEQ%3DRUgcAvRzE5jRrhhFpkm8E2PpBK9e9GhK26ZaJQt%3DQ@mail.gmail.com/ [2]
Link: https://lore.kernel.org/all/20250913000935.1021068-1-sudarsanm@google.com/ [3]
kmalloc_nolock() relies on ability of local_trylock_t to detect
the situation when per-cpu kmem_cache is locked.
In !PREEMPT_RT local_(try)lock_irqsave(&s->cpu_slab->lock, flags)
disables IRQs and marks s->cpu_slab->lock as acquired.
local_lock_is_locked(&s->cpu_slab->lock) returns true when
slab is in the middle of manipulating per-cpu cache
of that specific kmem_cache.
kmalloc_nolock() can be called from any context and can re-enter
into ___slab_alloc():
kmalloc() -> ___slab_alloc(cache_A) -> irqsave -> NMI -> bpf ->
kmalloc_nolock() -> ___slab_alloc(cache_B)
or
kmalloc() -> ___slab_alloc(cache_A) -> irqsave -> tracepoint/kprobe -> bpf ->
kmalloc_nolock() -> ___slab_alloc(cache_B)
Hence the caller of ___slab_alloc() checks if &s->cpu_slab->lock
can be acquired without a deadlock before invoking the function.
If that specific per-cpu kmem_cache is busy the kmalloc_nolock()
retries in a different kmalloc bucket. The second attempt will
likely succeed, since this cpu locked different kmem_cache.
Similarly, in PREEMPT_RT local_lock_is_locked() returns true when
per-cpu rt_spin_lock is locked by current _task_. In this case
re-entrance into the same kmalloc bucket is unsafe, and
kmalloc_nolock() tries a different bucket that is most likely is
not locked by the current task. Though it may be locked by a
different task it's safe to rt_spin_lock() and sleep on it.
Similar to alloc_pages_nolock() the kmalloc_nolock() returns NULL
immediately if called from hard irq or NMI in PREEMPT_RT.
kfree_nolock() defers freeing to irq_work when local_lock_is_locked()
and (in_nmi() or in PREEMPT_RT).
SLUB_TINY config doesn't use local_lock_is_locked() and relies on
spin_trylock_irqsave(&n->list_lock) to allocate,
while kfree_nolock() always defers to irq_work.
Note, kfree_nolock() must be called _only_ for objects allocated
with kmalloc_nolock(). Debug checks (like kmemleak and kfence)
were skipped on allocation, hence obj = kmalloc(); kfree_nolock(obj);
will miss kmemleak/kfence book keeping and will cause false positives.
large_kmalloc is not supported by either kmalloc_nolock()
or kfree_nolock().
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Reviewed-by: Harry Yoo <harry.yoo@oracle.com>
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Since the combination of valid upper bits in slab->obj_exts with
OBJEXTS_ALLOC_FAIL bit can never happen,
use OBJEXTS_ALLOC_FAIL == (1ull << 0) as a magic sentinel
instead of (1ull << 2) to free up bit 2.
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Shakeel Butt <shakeel.butt@linux.dev>
Reviewed-by: Harry Yoo <harry.yoo@oracle.com>
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
kmalloc_nolock() can be called from any context
the ___slab_alloc() can acquire local_trylock_t (which is rt_spin_lock
in PREEMPT_RT) and attempt to acquire a different local_trylock_t
while in the same task context.
The calling sequence might look like:
kmalloc() -> tracepoint -> bpf -> kmalloc_nolock()
or more precisely:
__lock_acquire+0x12ad/0x2590
lock_acquire+0x133/0x2d0
rt_spin_lock+0x6f/0x250
___slab_alloc+0xb7/0xec0
kmalloc_nolock_noprof+0x15a/0x430
my_debug_callback+0x20e/0x390 [testmod]
___slab_alloc+0x256/0xec0
__kmalloc_cache_noprof+0xd6/0x3b0
Make LOCKDEP understand that local_trylock_t-s protect
different kmem_caches. In order to do that add lock_class_key
for each kmem_cache and use that key in local_trylock_t.
This stack trace is possible on both PREEMPT_RT and !PREEMPT_RT,
but teach lockdep about it only for PREEMPT_RT, since
in !PREEMPT_RT the ___slab_alloc() code is using
local_trylock_irqsave() when lockdep is on.
Note, this patch applies this logic to local_lock_t
while the next one converts it to local_trylock_t.
Both are mapped to rt_spin_lock in PREEMPT_RT.
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Split alloc_pages_nolock() and introduce alloc_frozen_pages_nolock()
to be used by alloc_slab_page().
Reviewed-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Reviewed-by: Shakeel Butt <shakeel.butt@linux.dev>
Reviewed-by: Harry Yoo <harry.yoo@oracle.com>
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
