| Age | Commit message (Collapse) | Author |
|---|
|
'struct __kernel_old_timespec' represents the 'native' time ABI of the
kernel. On 32-bit systems it uses 32-bit fields and on 64-bit systems
it uses 64-bit fields.
However the x86 x32 ABI uses the 64-bit time ABI natively. This is
correctly handled for the 'tv_sec' fields, through the typedefs of
'__kernel_old_time_t' -> '__kernel_long_t' -> 'long long'. The same
treatment was missed for 'tv_nsec'.
In practice this might not make much of a difference as the value of
'tv_nsec' will always fit into 32 bits and the missing bits fall
into the padding of the structure.
When introspecting the structure however, a difference can be observed.
Switch to 64-bit tv_nsec on x32. No other architectures or ABIs are
affected.
While this could be interpreted as violating the POSIX requirement of
'timespec::tv_nsec' being 'long':
* __kernel_old_timespec is not actually the POSIX timespec type
* the requirement is gone in newer versions of POSIX
* this matches glibc
Fixes: 94c467ddb273 ("y2038: add __kernel_old_timespec and __kernel_old_time_t")
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260504-timespec-x32-v2-1-0739c9047fc4@linutronix.de
|
|
If CONFIG_COMPAT_32BIT_TIME is disabled then the vDSO should not
provide any 32-bit time related functionality.
Add some build-time validations to make sure the architecture-specific
glue satisfies this requirement.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@oss.qualcomm.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Link: https://patch.msgid.link/20260702-vdso-compat_32bit_time-v3-9-db9f36d8d432@linutronix.de
|
|
If CONFIG_COMPAT_32BIT_TIME is disabled then the vDSO should not
provide any 32-bit time related functionality. This is the intended
effect of the kconfig option and also the fallback system calls would
also not be implemented.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@oss.qualcomm.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Link: https://patch.msgid.link/20260702-vdso-compat_32bit_time-v3-8-db9f36d8d432@linutronix.de
|
|
If CONFIG_COMPAT_32BIT_TIME is disabled then the vDSO should not
provide any 32-bit time related functionality. This is the intended
effect of the kconfig option and also the fallback system calls would
also not be implemented.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Link: https://patch.msgid.link/20260702-vdso-compat_32bit_time-v3-7-db9f36d8d432@linutronix.de
|
|
If CONFIG_COMPAT_32BIT_TIME is disabled then the vDSO should not
provide any 32-bit time related functionality. This is the intended
effect of the kconfig option and also the fallback system calls would
also not be implemented.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Link: https://patch.msgid.link/20260702-vdso-compat_32bit_time-v3-6-db9f36d8d432@linutronix.de
|
|
If CONFIG_COMPAT_32BIT_TIME is disabled then the vDSO should not
provide any 32-bit time related functionality. This is the intended
effect of the kconfig option and also the fallback system calls would
also not be implemented.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@oss.qualcomm.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Link: https://patch.msgid.link/20260702-vdso-compat_32bit_time-v3-5-db9f36d8d432@linutronix.de
|
|
If CONFIG_COMPAT_32BIT_TIME is disabled then the vDSO should not
provide any 32-bit time related functionality. This is the intended
effect of the kconfig option and also the fallback system calls would
also not be implemented.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@oss.qualcomm.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Link: https://patch.msgid.link/20260702-vdso-compat_32bit_time-v3-4-db9f36d8d432@linutronix.de
|
|
If CONFIG_COMPAT_32BIT_TIME is disabled then the vDSO should not
provide any 32-bit time related functionality. This is the intended
effect of the kconfig option and also the fallback system calls would
also not be implemented.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Link: https://patch.msgid.link/20260702-vdso-compat_32bit_time-v3-3-db9f36d8d432@linutronix.de
|
|
Not all architectures have the system calls for time() and
gettimeofday(). When the system call is missing, the vDSO function
should also not be present.
Validate that.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@oss.qualcomm.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Link: https://patch.msgid.link/20260702-vdso-compat_32bit_time-v3-2-db9f36d8d432@linutronix.de
|
|
The "old" time types use 32-bit seconds which are not y2038-safe.
Respect COMPAT_32BIT_TIME for functions using those types.
time(), stime() and gettimeofday() are disabled completely.
settimeofday() is kept as it is required to do the initial timewarping
after boot. However the 'tv' argument will be rejected.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Link: https://lore.kernel.org/lkml/e9487ebe-3730-438a-9c23-e45f75986ecc@app.fastmail.com/
Link: https://patch.msgid.link/20260702-vdso-compat_32bit_time-v3-1-db9f36d8d432@linutronix.de
|
|
The logic for CONFIG_GENERIC_GETTIMEOFDAY=n and !timens_page is
identical now.
Use this to simplify the logic a bit.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260630-vdso-mlockall-v4-6-6c93708ce723@linutronix.de
|
|
While mlockall() is meant to lock page *memory*, effectively it will
also create and lock the corresponding page table entries.
Latency-sensitive applications expect not to experience any pagefaults
after calling mlockall(). However mlockall() ignores VM_IO mappings,
which is used by the generic vDSO datastore.
While the fault handler itself is very fast, going through the full
pagefault exception handling is much slower, on the order of 20us in a
test machine.
Since the memory behind the datastore mappings is always present and
accessible it is not necessary to use VM_IO for them.
The data page mapping is now also aligned with the architecture-specific
code pages. Some architecture-specific data pages, like the x86 VCLOCK
pages, continue to use VM_IO as they are not always mappable. They will
require their own special handling later when the general approach has
been agreed upon.
As a side-effect this will allow GUP on these pages and allow more ways
to access the data in them. This is fine, as all data in this mapping
is globally visible anyways. Either because it is mapped into all tasks,
or in the case of the time namespace pages, can be read from procfs.
Regular mlock() would also work, but userspace does not know the boundaries
of the vDSO.
Reported-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Tested-by: Nam Cao <namcao@linutronix.de>
Link: https://patch.msgid.link/20260630-vdso-mlockall-v4-5-6c93708ce723@linutronix.de
|
|
The fault handler for the timens page does not have access to the target
task and therefore can not be invoked remotely.
Currently the handler relies on the fact that the vvar mapping is marked as
VM_IO and VM_PFNMAP for which the mm core always prevents remote access.
However the VM_IO and VM_PFNMAP flags are going to be removed.
Add an explicit check to prevent remote access to the mapping.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Tested-by: Nam Cao <namcao@linutronix.de>
Link: https://patch.msgid.link/20260630-vdso-mlockall-v4-4-6c93708ce723@linutronix.de
|
|
mlockall() stops if a page in a VMA is unmappable. As the datastore VMA
can contain holes, mlockall() would not process all data pages.
Replace the mapping error VM_FAULT_SIGBUS by just mapping the underlying
unused and zeroed-out data page. The vDSO will not access these pages in
any case and for other userspace these pages have undefined contents.
This will allow mlockall() to process all pages within the VMA as soon
as VM_IO is removed from the VMA.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Tested-by: Nam Cao <namcao@linutronix.de>
Link: https://patch.msgid.link/20260630-vdso-mlockall-v4-3-6c93708ce723@linutronix.de
|
|
To support mlockall() on the datapages the VMA can have no holes where
inner pages return VM_FAULT_SIGBUS. An upcoming change will avoid these
holes by mapping a zeroed pages into these holes. That logic will be
simpler when the mapping logic is based on vmf->pgoff instead of the
vdso_k_ symbols.
Switch to the equivalent vmf->pgoff logic.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Tested-by: Nam Cao <namcao@linutronix.de>
Link: https://patch.msgid.link/20260630-vdso-mlockall-v4-2-6c93708ce723@linutronix.de
|
|
An upcoming change will make this a file-scoped variable, for which it
should have a clearer name.
Rename the variable to prepare for that.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Tested-by: Nam Cao <namcao@linutronix.de>
Link: https://patch.msgid.link/20260630-vdso-mlockall-v4-1-6c93708ce723@linutronix.de
|
|
v3d_get_cpu_indirect_csd_params() looks up the indirect buffer object
from a userspace-supplied handle but never checks the result. A bogus
or stale handle makes drm_gem_object_lookup() return NULL, which is
then stored in info->indirect and only dereferenced later when the
indirect CSD job runs, turning a userspace mistake into a NULL pointer
dereference in the kernel.
Bail out with -ENOENT as soon as the lookup fails, so the bad handle is
rejected at submission time.
Fixes: 18b8413b25b7 ("drm/v3d: Create a CPU job extension for a indirect CSD job")
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Signed-off-by: Maíra Canal <mcanal@igalia.com>
Link: https://patch.msgid.link/20260703-v3d-cpu-job-fixes-v3-2-bc51b1f3eeb5@igalia.com
|
|
All architectures are now fully using the generic vDSO infrastructure.
They don't need these declarations anymore to implement the functions
in architecture-specific code.
Move them to the private header.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260707-timekeeping-header-cleanup-v1-2-e85ad96409a9@linutronix.de
|
|
The only caller of vdso_time_update_aux() is already gated behind
CONFIG_POSIX_AUX. The additional check in the header files is not
necessary.
Remove it and then fold the declarations into the existing
CONFIG_GENERIC_GETTIMEOFDAY ifdeffery.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260707-timekeeping-header-cleanup-v1-1-e85ad96409a9@linutronix.de
|
|
hrtimer.h is used all over the kernel. Any change to hrtimer_bases.h
effectively triggers a full rebuild.
As all logical dependencies from hrtimer.h to hrtimer_bases.h have been
removed, the inclusion is now unncessary. Remove it.
Signed-off-by: Thomas Weißschuh (Schneider Electric) <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260702-hrtimer-header-dependencies-v1-8-c50b19bda473@linutronix.de
|
|
The usage of 'bool' requires linux/types.h, which is currently only
included through a transitive dependency chain.
Include linux/types.h as that chain is going to go away.
Signed-off-by: Thomas Weißschuh (Schneider Electric) <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260702-hrtimer-header-dependencies-v1-7-c50b19bda473@linutronix.de
|
|
Multiple used types and symbols are only visible through transitive
dependency chains.
Include the headers explicitly as those chains are going to go away.
Signed-off-by: Thomas Weißschuh (Schneider Electric) <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260702-hrtimer-header-dependencies-v1-6-c50b19bda473@linutronix.de
|
|
This header uses some definitions from linux/hrtimer_bases.h.
Currently this header is included transitively, which will change.
Include the header explicitly.
Signed-off-by: Thomas Weißschuh (Schneider Electric) <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260702-hrtimer-header-dependencies-v1-5-c50b19bda473@linutronix.de
|
|
This header uses some definitions from linux/hrtimer_bases.h.
Currently this header is included transitively, which will change.
Include the header explicitly.
Signed-off-by: Thomas Weißschuh (Schneider Electric) <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260702-hrtimer-header-dependencies-v1-4-c50b19bda473@linutronix.de
|
|
The usage of the hrtimer base forces hrtimer.h to also expose the base
structure definitions.
Move the function to hrtimer.c to avoid this.
Signed-off-by: Thomas Weißschuh (Schneider Electric) <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260702-hrtimer-header-dependencies-v1-3-c50b19bda473@linutronix.de
|
|
The usage of the hrtimer base introduces a dependency on the timer base
structure definitions from the widely-used hrtimer.h.
Move the helper to hrtimer_bases.h to trim this dependency.
Also adapt the two only callers to now include hrtimer_bases.h.
Signed-off-by: Thomas Weißschuh (Schneider Electric) <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260702-hrtimer-header-dependencies-v1-2-c50b19bda473@linutronix.de
|
|
This header was originally added for #defines and was later extended
with the hrtimer base structures. All the #defines have been removed
in the meantime, so the naming is off now.
Rename the header to fit its contents more.
This will also make the upcoming addition of some functions nicer.
Signed-off-by: Thomas Weißschuh (Schneider Electric) <thomas.weissschuh@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260702-hrtimer-header-dependencies-v1-1-c50b19bda473@linutronix.de
|
|
hrtimer_get_next_event() runs on every tick-stop decision via
get_next_timer_interrupt() -> cmp_next_hrtimer_event(). When high
resolution timers are active it must return KTIME_MAX -- the caller
documents and depends on this.
The function takes cpu_base->lock, checks hres_active, and returns
KTIME_MAX.
Taking the lock is not required because cpu_base->hres_active is only
written by the local CPU in hrtimer_switch_to_hres() from hard interrupt
context and in hrtimers_cpu_starting() during bring-up.
All callers of hrtimer_get_next_event() reach it from the tick-stop /
cpuidle paths with interrupts disabled on that CPU. No writer can therefore
race with the read, so an unlocked hres_active check is stable and the lock
can be skipped in this case.
On a 176-thread AMD EPYC 9D64 running a production workload, bucketing
callers of native_queued_spin_lock_slowpath(), the slowpath had 199 samples
in total, of which 54 are attributed to hrtimer_get_next_event(), i.e. this
accounts for ~27% of slowpath hits on this specific workload.
Move the hres_active check before the lock guard region to address this.
[ tglx: Massaged change log and comments ]
Signed-off-by: Usama Arif <usama.arif@linux.dev>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260707153800.542394-1-usama.arif@linux.dev
|
|
print_cpu() prints hrtimer_cpu_base and tick_sched state without
synchronizing with concurrent updates. The output is diagnostic only, so
use data_race(READ_ONCE()) for these scalar reads to document the
intentional races and avoid KCSAN reports.
Reported-by: syzbot+8f0e958900a14d08a51d@syzkaller.appspotmail.com
Signed-off-by: Yu Peng <pengyu@kylinos.cn>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260707125813.2182532-1-pengyu@kylinos.cn
Closes: https://syzkaller.appspot.com/bug?extid=8f0e958900a14d08a51d
|
|
The function validates tv_sec to be within [-KTIME_SEC_MAX, KTIME_SEC_MAX]
but never validates that tv_nsec is within the valid range of
[0, NSEC_PER_SEC-1] before using it in timespec64_add().
timespec64_add() expects both timespec64 structures to have normalized
values with tv_nsec in the range [0, 999999999]. If off->val.tv_nsec
contains invalid values (negative or >= NSEC_PER_SEC), it could lead to
incorrect calculations or unexpected behavior.
Add validation to ensure tv_nsec is within the valid range before
performing the addition.
Fixes: 04a8682a71be ("fs/proc: Introduce /proc/pid/timens_offsets")
Signed-off-by: Malaya Kumar Rout <malayarout91@gmail.com>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260704093429.89350-1-malayarout91@gmail.com
|
|
When the WARN_ON_ONCE(i >= tmigr_hierarchy_levels) assertion triggers,
the function returns -EINVAL without freeing the 'stack' memory allocated
via kzalloc_objs() at the beginning of the function.
Add kfree(stack) before returning to prevent the memory leak.
Fixes: 6c181b5667ee ("timers/migration: Convert "while" loops to use "for"")
Signed-off-by: Malaya Kumar Rout <malayarout91@gmail.com>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260704085533.87098-1-malayarout91@gmail.com
|
|
tk_aux_sysfs_init() creates one child kobject per auxiliary clock. If a
later child or sysfs group creation fails, the current error path only
puts the parent kobjects and leaves earlier children and groups behind.
Store the child kobjects during init and remove the successfully created
groups and kobjects on failure.
Fixes: 7b5ab04f035f ("timekeeping: Fix resource leak in tk_aux_sysfs_init() error paths")
Signed-off-by: Yuho Choi <dbgh9129@gmail.com>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260703165337.168445-1-dbgh9129@gmail.com
|
|
init_clocksource_sysfs() registers the clocksource subsystem before
registering the clocksource device. If device_register() fails, the
function returns the error while leaving the subsystem registered.
Unregister the clocksource subsystem on that failure path so the
successful subsystem registration is unwound before returning.
Fixes: d369a5d8fc70 ("clocksource: convert sysdev_class to a regular subsystem")
Signed-off-by: Yuho Choi <dbgh9129@gmail.com>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260702215733.84588-1-dbgh9129@gmail.com
|
|
Commit 98b74e1f3104 ("kselftests: timers: leap-a-day: Change default
arguments to help test runs") replaced the -s option with -w and made
"wait for the leap second" the non-default behaviour, but it only
updated the switch/case handling. Two things were left inconsistent:
- The getopt() option string still lists 's' instead of 'w', so
passing -w is rejected as an invalid option and the new behaviour
cannot be selected at all.
- The file header comment still documents the removed -s option and
an outdated default for -i.
Fix the getopt() string to accept 'w' (matching the existing case 'w':
handler) and update the header comment to describe -w, -t and the
current -i default.
Fixes: 98b74e1f3104 ("kselftests: timers: leap-a-day: Change default arguments to help test runs")
Signed-off-by: Jiangshan Yi <yijiangshan@kylinos.cn>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Acked-by: John Stultz <jstultz@google.com>
Link: https://patch.msgid.link/20260702093915.2652638-1-yijiangshan@kylinos.cn
|
|
The copy of performance query results to the output buffer open-codes
the 32-bit/64-bit selection with two nearly identical loops. As the
write_to_buffer() helper already encapsulates the do_64bit decision,
use it instead of open-coding it.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Link: https://patch.msgid.link/20260703-v3d-cpu-job-fixes-v3-3-bc51b1f3eeb5@igalia.com
Signed-off-by: Maíra Canal <mcanal@igalia.com>
|
|
Fix its node leak when its_probe_one() failed in
gic_acpi_parse_madt_its().
Fixes: 9585a495ac936 ("irqchip/gic-v3-its: Split allocation from initialisation of its_node")
Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Reviewed-by: Radu Rendec <radu@rendec.net>
Link: https://patch.msgid.link/20260702033050.1583-3-shikemeng@huaweicloud.com
|
|
Fix collection leak when its_init_domain() failed in its_probe_one().
Fixes: 4c21f3c26ecc2 ("irqchip: GICv3: ITS: DT probing and initialization")
Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Reviewed-by: Radu Rendec <radu@rendec.net>
Link: https://patch.msgid.link/20260702033050.1583-2-shikemeng@huaweicloud.com
|
|
Pull the changes in which were provided for the GPIO tree.
|
|
All PDC hardware versions support pass-through mode in which both Direct
SPIs and GPIO interrupts (as SPIs) are sent to the GIC without latching at
the PDC level.
Newer PDCs (v3.0 onwards) also support an additional secondary controller
mode where PDC latches the GPIO interrupts and sends them to the GIC as
level type interrupts. Direct SPIs works the same as pass-through mode
without latching at PDC even in secondary controller mode.
All the SoCs use pass-through mode with the exception of x1e. The x1e PDC
may be set to secondary controller mode for builds on CRD boards whereas it
may be set to pass through mode for IoT-EVK boards. The mode configuration
is done in firmware and initially shipped windows firmware did not have a
SCM interface to read or modify the PDC mode. Only later write access was
opened up for the non-secure world. The availability of write access can be
probed.
Therefore try to modify the to modify the PDC mode to pass-through mode via
SCM write. When the write fails on older firmware assume it to work in
secondary mode.
In secondary mode set the separate irqchip for the GPIOs to perform
additional operations only for the GPIO interrupts.
[ tglx: Massage change log ]
Co-developed-by: Sneh Mankad <sneh.mankad@oss.qualcomm.com>
Signed-off-by: Sneh Mankad <sneh.mankad@oss.qualcomm.com>
Signed-off-by: Maulik Shah <maulik.shah@oss.qualcomm.com>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260707-hamoa_pdc_v3-v4-4-dfd1f4a3ae89@oss.qualcomm.com
|
|
Before commit 4dc70713dc24 ("irqchip/qcom-pdc: Kill non-wakeup irqdomain")
there were separate domains for direct SPIs and GPIOs used as SPIs.
Separate domains can be useful to have separate interrupt chips for
different functionalities. Since the commit unified both domains there
is no way to differentiate.
In preparation to add the second level interrupt controller support where
GPIO interrupts get latched at PDC (but not direct SPIs) there is a need to
differentiate between SPIs and GPIOs as SPIs. Reverting above commit does
is not a good option as it leads to waste of resources.
PDC hardware provides the IRQ_PARAM register to enumerate the number of
direct SPIs and the number of GPIOs as SPIs. Further PDC allocates direct
SPIs at the beginning and all GPIOs as SPIs are allocated at the end. This
information can be used in the driver to differentiate them.
Add the support to read this register and keep this information in struct
pdc_desc. Later change utilizes it.
[ tglx: Massage change log ]
Signed-off-by: Maulik Shah <maulik.shah@oss.qualcomm.com>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260707-hamoa_pdc_v3-v4-3-dfd1f4a3ae89@oss.qualcomm.com
|
|
There are multiple static variables used in the driver. Move all to struct
pdc_desc to better align with versioning support. Document them.
Add a new pdc->enable_intr() callback to point to the respective versions
specific enable function. Remove pdc_enable_intr() and __pdc_enable_intr()
and invoke pdc->enable_intr() from the call sites.
Locking in pdc_enable_intr() applies lock to all version specific
pdc->enable_intr() however lock is needed only for pdc_enable_intr_bank()
which uses a shared bank across on PDC v2.7 and PDC v3.0.
pdc_enable_intr_cfg() do not require locking as IRQ_CFG registers are one
per interrupt. Move the locking to pdc_enable_intr_bank().
No functional impact intended
[ tglx: Massage change log. ]
Signed-off-by: Maulik Shah <maulik.shah@oss.qualcomm.com>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260707-hamoa_pdc_v3-v4-2-dfd1f4a3ae89@oss.qualcomm.com
|
|
The PDC irqchip driver updates IRQ_ENABLE and IRQ_CFG for three different
versions v2.7, v3.0 and v3.2. These registers are organized in hardware as
below on various SoCs:
+---------------------------------------------------------------+
| SM8350, SM8450 | SM8550, Hamoa | SM8650, SM8750 |
|---------------------------------------------------------------|
| v2.7 | v3.0 | v3.2 |
|---------------------------------------------------------------|
| IRQ_ENABLE_BANK | IRQ_ENABLE_BANK | NA |
|---------------------------------------------------------------|
| IRQ_CFG | IRQ_CFG | IRQ_CFG |
| | | |
| | | [31:6] Unused |
| | [31:5] Unused | [5] GPIO_STATUS |
| | [4] GPIO_STATUS| [4] GPIO_MASK |
| [31:3] Unused | [3] GPIO_MASK | [3] IRQ_ENABLE |
| [0:2] Type | [0:2] Type | [0:2] Type |
+---------------------------------------------------------------|
All SoCs PDC chips support "pass through mode" in which all interrupts are
forwarded to the GIC without any latching in the PDC hardware.
So far the driver did not utilize GPIO_STATUS and GPIO_MASK from IRQ_CFG
register for v3.0 and v3.2 since they are only needed to be configured when
PDC runs in a specific mode named "second level interrupt controller". In
that mode it can latch the GPIO interrupts in GPIO_STATUS and forward GPIO
interrupts to GIC as LEVEL_HIGH type SPI interrupt.
All the SoCs defaulted to pass through mode with the exception of some x1e
types. x1e PDC may be set to secondary controller mode for builds on CRD
boards whereas it may be set to pass through mode for IoT-EVK boards.
Restructure in preparation to add the second level interrupt controller
mode utilizing GPIO_STATUS and GPIO_MASK bits which changed the bit
positions between v3.0 and v3.2.
No functional change intended.
[ tglx: Massaged change log ]
Signed-off-by: Maulik Shah <maulik.shah@oss.qualcomm.com>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
Link: https://patch.msgid.link/20260707-hamoa_pdc_v3-v4-1-dfd1f4a3ae89@oss.qualcomm.com
|
|
A non-global perfmon is meant to count events generated by a specific
submission, but the scheduler can run jobs from different queues
concurrently on the same V3D core. Without explicit serialization, an
unrelated job running in parallel with a perfmon-carrying job pollutes
the counters and generates unusable results.
To address such issue, we must enforce cross-queue serialization when we
detect a perfmon-carrying submission. It's possible to implement
serialization by enforcing two rules:
1. A job that carries a non-global perfmon must wait for every job
currently in-flight across all HW queues to finish.
2. While a perfmon-carrying job is still in-flight, all subsequently
submitted jobs must wait for it.
Note that serialization is not needed in the global perfmon case, as the
global perfmon tracks activity from all jobs, so concurrency is desirable.
Therefore, check if serialization is needed during job submission and if
so, attach fence dependences to enforce cross-queue serialization.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Link: https://patch.msgid.link/20260706-v3d-perfmon-lifetime-v4-2-d7b312ff2c83@igalia.com
Signed-off-by: Maíra Canal <mcanal@igalia.com>
|
|
v3d exposes a single set of performance counters per core, so at any
moment at most one performance monitor can be programmed in HW. In
software, this singleton is represented by v3d_dev->active_perfmon, but
until now nothing actually serialized access to it: scheduler callbacks,
the GPU-reset path, and perfmon ioctls all read and wrote that field
lock-free.
The existence of v3d_perfmon->lock mutex did not close the gap. It
serialized start/stop of *one* perfmon object against itself, but the
invariant that needs protection is device-wide: there can be exactly one
active perfmon at any moment in HW. Two threads acting on different
perfmon objects could race through v3d_dev->active_perfmon and the
counter registers, leaving software and HW out of sync.
This commit moves the locking to where the invariant actually lives. Group
the active perfmon pointer with a device-wide spinlock and route every
state transition (job start, job completion, set global, reset,
suspend/resume, destruction) through a small set of locked entry points
that are the only mutators of the HW counters.
Some design improvements needed to be made for the refactor:
1. Stop the perfmon from the IRQ handler at job-completion time (the
natural boundary for "active perfmon follows the active job"). This
required a change from a mutex to a spinlock. This solves another
issue of the existing design: perfmon start/stop was exclusively
attached to run_job() callbacks, which means that if nothing was
further queued up, a perfmon would never actually be stopped.
2. Pause/resume the HW counters across runtime-PM transitions without
dropping the software reference. This preserves the perfmon state
while the device is idle.
3. Move the global perfmon lifecycle management to the set_global
IOCTL. This simplifies the logic in v3d_perfmon_start() and
v3d_perfmon_stop(), as there is no need to always check if the
global perfmon is enabled.
4. v3d_perfmon_get_values_ioctl() doesn't stop the perfmon when
capturing the values. All lifecycle management is handled by the
job (for per-job perfmons) or the set_global IOCTL (for global
perfmons).
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Link: https://patch.msgid.link/20260706-v3d-perfmon-lifetime-v4-1-d7b312ff2c83@igalia.com
Signed-off-by: Maíra Canal <mcanal@igalia.com>
|
|
syscon-reboot-mode schema allows arbitrary "mode-.* properties but only
a subset actually makes sense and is valid. Provide negative look-ahead
pattern to disallow any modes not supported by the device, which
tightens the binding.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@oss.qualcomm.com>
Acked-by: Rob Herring (Arm) <robh@kernel.org>
Link: https://patch.msgid.link/20260701105849.197086-4-krzysztof.kozlowski@oss.qualcomm.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
|
|
The 'reboot-mode' child of PMU node must be a syscon-reboot-mode one, so
add missing $ref to enforce the validation and disallow other
reboot-mode handlers.
Defining 'type: object' is on the other hand not necessary when other
schema is referenced.
This can be further restricted to match hardware/firmware:
syscon-reboot-mode schema allows arbitrary "mode-.* properties but only
a subset actually makes sense and is valid. Provide negative look-ahead
pattern to disallow any modes not supported by the device, which
tightens the binding.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@oss.qualcomm.com>
Acked-by: Rob Herring (Arm) <robh@kernel.org>
Link: https://patch.msgid.link/20260701105849.197086-3-krzysztof.kozlowski@oss.qualcomm.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
|
|
The kstat_irqs field of struct irq_desc is used to store a per-cpu count
of interrupt events. It is initialized in init_desc(), along with all
the other fields in struct irq_desc that need explicit initialization,
and therefore it's always available (non-NULL) for any valid interrupt
descriptor.
On parisc, CONFIG_SPARSE_IRQ is disabled, interrupt descriptors are
allocated statically in the irq_desc[] array, and kstat_irqs is
initialized implicitly to NULL. The kstat_irqs field is initialized
later, for all descriptors, via start_kernel() -> early_irq_init() ->
init_desc().
smp_boot_one_cpu() is used only for CPU hotplugging. On a SMP system,
the boot CPU initializes the interrupt descriptors as described above,
and smp_boot_one_cpu() is called later, for the secondary CPU(s).
A previous change made similar changes across the generic interrupt code in
kernel/irq/irqdesc.c, and provides a more detailed explanation of why
kstat_irqs is guaranteed to be non-NULL.
Signed-off-by: Radu Rendec <radu@rendec.net>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260705210951.2717741-3-radu@rendec.net
|
|
The kstat_irqs field of struct irq_desc is used to store a per-cpu count
of interrupt events. It is initialized in init_desc(), along with all
the other fields in struct irq_desc that need explicit initialization,
and therefore it's always available (non-NULL) for any valid interrupt
descriptor (with a caveat - see below).
When CONFIG_SPARSE_IRQ is enabled, all interrupt descriptors are always
allocated dynamically via alloc_desc(), which calls init_desc(), so in
that case kstat_irqs is guaranteed to be non-NULL before a valid struct
irq_desc pointer is even returned.
By contrast, when CONFIG_SPARSE_IRQ is disabled, interrupt descriptors
are allocated statically in the irq_desc[] array, and kstat_irqs is
initialized implicitly to NULL. The per-cpu pointer is initialized only
later, for all descriptors, via start_kernel() -> early_irq_init() ->
init_desc().
The kstat_irqs field is used mostly for printing interrupt statistics
(i.e. reading /proc/interrupts), and that cannot happen until much
later, when user-space is fully initialized. So, there is no concern
with that use case.
The list below includes all functions where the NULL check is removed,
along with a list of all possible call chains and/or a brief explanation
of why it's safe to remove the NULL check in that case.
* irq_desc_kstat_cpu() [include/linux/irqdesc.h]
- All direct call sites use it for printing IRQ statistics.
- Indirect call site: per_cpu_count_show() - also used for printing
interruptstatistics.
* kstat_irqs_cpu() [kernel/irq/irqdesc.c]
- Called by sun3_int7() and sun3_int5() [arch/m68k/sun3/sun3ints.c]
These are interrupt handlers and cannot be called until their
corresponding interrupts are initialized in sun3_init_IRQ(). The
call chain leading to that is:
start_kernel() -> init_IRQ() [arch/m68k/kernel/ints.c] ->
mach_init_IRQ = sun3_init_IRQ()
The init_IRQ() call happens right *after* the early_irq_init() call,
which means the descriptors are already fully initialized by the
time the interrupt handlers are even registered.
- Called by show_interrupts() [arch/s390/kernel/irq.c] - used for
printing interrupt statistics.
* kstat_irqs() [kernel/irq/irqdesc.c]
The only possible call chain is via fs/proc/stat.c:
stat_open() -> show_stat() -> show_all_irqs() -> kstat_irqs_usr() ->
kstat_irqs()
It is used for printing interrupt statistics.
* kstat_snapshot_irqs()
The only possible call chain is via kernel/watchdog.c:
watchdog_timer_fn() -> is_softlockup() -> start_counting_irqs() ->
kstat_snapshot_irqs()
The watchdog timer cannot fire early, before early_irq_init().
* kstat_get_irq_since_snapshot()
The only possible call chain is via kernel/watchdog.c:
watchdog_timer_fn() -> report_cpu_status() -> print_irq_counts() ->
kstat_get_irq_since_snapshot()
The watchdog timer cannot fire early, before early_irq_init().
Signed-off-by: Radu Rendec <radu@rendec.net>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Link: https://patch.msgid.link/20260705210951.2717741-2-radu@rendec.net
|
|
The calloc-allocated buffer in attr_lsm_count() is never released on
any exit path, including both the normal return path and the early
return when read_sysfs_lsms fails, resulting in a heap memory leak.
Add free() for the buffer on all return branches to fix the leak.
Fixes: d3d929a8b0cd ("LSM: selftests for Linux Security Module syscalls")
Signed-off-by: Wang Yan <wangyan01@kylinos.cn>
Reviewed-by: William Roberts <bill.c.roberts@gmail.com>
Tested-by: William Roberts <bill.c.roberts@gmail.com>
Signed-off-by: Paul Moore <paul@paul-moore.com>
|
|
task_euid() is a very weird operation. You can see how weird it is by
grepping for task_euid() - binder is its only user. task_euid() obtains
the objective effective UID - it looks at the credentials of the task
for purposes of acting on it as an object, but then accesses the
effective UID (which the credentials.7 man page describes as "[...] used
by the kernel to determine the permissions that the process will have
when accessing shared resources [...]").
Since usage in Binder has now been removed, get rid of the resulting
dead code.
Changes to the zh_CN translation was carried out with the help of
Gemini and Google Translate, and since adjusted as per Alex Shi's
feedback.
Suggested-by: Jann Horn <jannh@google.com>
Reviewed-by: Gary Guo <gary@garyguo.net>
Signed-off-by: Alice Ryhl <aliceryhl@google.com>
Signed-off-by: Paul Moore <paul@paul-moore.com>
|
