tjh.dev/kernel at v6.15-rc1

The perf events code fails to account for total_time_enabled of
inactive events.

Here is a failure case for accounting total_time_enabled for
CPU PMU events:

sudo ./perf stat -vvv -e armv8_pmuv3_0/event=0x08/ -e armv8_pmuv3_1/event=0x08/ -- stress-ng --pthread=2 -t 2s
...

armv8_pmuv3_0/event=0x08/: 1138698008 2289429840 2174835740
armv8_pmuv3_1/event=0x08/: 1826791390 1950025700 847648440
` ` `
` ` > total_time_running with child
` > total_time_enabled with child
> count with child

Performance counter stats for 'stress-ng --pthread=2 -t 2s':

1,138,698,008 armv8_pmuv3_0/event=0x08/ (94.99%)
1,826,791,390 armv8_pmuv3_1/event=0x08/ (43.47%)

The two events above are opened on two different CPU PMUs, for example,
each event is opened for a cluster in an Arm big.LITTLE system, they
will never run on the same CPU. In theory, the total enabled time should
be same for both events, as two events are opened and closed together.

As the result show, the two events' total enabled time including
child event is different (2289429840 vs 1950025700).

This is because child events are not accounted properly
if a event is INACTIVE state when the task exits:

perf_event_exit_event()
`> perf_remove_from_context()
`> __perf_remove_from_context()
`> perf_child_detach() -> Accumulate child_total_time_enabled
`> list_del_event() -> Update child event's time

The problem is the time accumulation happens prior to child event's
time updating. Thus, it misses to account the last period's time when
the event exits.

The perf core layer follows the rule that timekeeping is tied to state
change. To address the issue, make __perf_remove_from_context()
handle the task exit case by passing 'DETACH_EXIT' to it and
invoke perf_event_state() for state alongside with accounting the time.

Then, perf_child_detach() populates the time into the parent's time metrics.

After this patch, the bug is fixed:

sudo ./perf stat -vvv -e armv8_pmuv3_0/event=0x08/ -e armv8_pmuv3_1/event=0x08/ -- stress-ng --pthread=2 -t 10s
...
armv8_pmuv3_0/event=0x08/: 15396770398 32157963940 21898169000
armv8_pmuv3_1/event=0x08/: 22428964974 32157963940 10259794940

Performance counter stats for 'stress-ng --pthread=2 -t 10s':

15,396,770,398 armv8_pmuv3_0/event=0x08/ (68.10%)
22,428,964,974 armv8_pmuv3_1/event=0x08/ (31.90%)

[ mingo: Clarified the changelog. ]

Fixes: ef54c1a476aef ("perf: Rework perf_event_exit_event()")
Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Yeoreum Yun <yeoreum.yun@arm.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Tested-by: Leo Yan <leo.yan@arm.com>
Link: https://lore.kernel.org/r/20250326082003.1630986-1-yeoreum.yun@arm.com

a3c3c666

Yeoreum Yun

9 months ago

soundwire: take in count the bandwidth of a prepared stream

08ae0d61

Bard Liao

10 months ago

tools/power turbostat: re-factor sysfs code

99463389

Len Brown

9 months ago

Disable SLUB_TINY for build testing

6f110a5e

Linus Torvalds

9 months ago

rseq: Eliminate useless task_work on execve

169eae77

Mathieu Desnoyers

9 months ago

Merge tag 'rust-6.15' of git://git.kernel.org/pub/scm/linux/kernel/git/ojeda/linux

Pull Rust updates from Miguel Ojeda:
"Toolchain and infrastructure:

- Extract the 'pin-init' API from the 'kernel' crate and make it into
a standalone crate.

In order to do this, the contents are rearranged so that they can
easily be kept in sync with the version maintained out-of-tree that
other projects have started to use too (or plan to, like QEMU).

This will reduce the maintenance burden for Benno, who will now
have his own sub-tree, and will simplify future expected changes
like the move to use 'syn' to simplify the implementation.

- Add '#[test]'-like support based on KUnit.

We already had doctests support based on KUnit, which takes the
examples in our Rust documentation and runs them under KUnit.

Now, we are adding the beginning of the support for "normal" tests,
similar to those the '#[test]' tests in userspace Rust. For
instance:

#[kunit_tests(my_suite)]
mod tests {
#[test]
fn my_test() {
assert_eq!(1 + 1, 2);
}
}

Unlike with doctests, the 'assert*!'s do not map to the KUnit
assertion APIs yet.

- Check Rust signatures at compile time for functions called from C
by name.

In particular, introduce a new '#[export]' macro that can be placed
in the Rust function definition. It will ensure that the function
declaration on the C side matches the signature on the Rust
function:

#[export]
pub unsafe extern "C" fn my_function(a: u8, b: i32) -> usize {
// ...
}

The macro essentially forces the compiler to compare the types of
the actual Rust function and the 'bindgen'-processed C signature.

These cases are rare so far. In the future, we may consider
introducing another tool, 'cbindgen', to generate C headers
automatically. Even then, having these functions explicitly marked
may be a good idea anyway.

- Enable the 'raw_ref_op' Rust feature: it is already stable, and
allows us to use the new '&raw' syntax, avoiding a couple macros.
After everyone has migrated, we will disallow the macros.

- Pass the correct target to 'bindgen' on Usermode Linux.

- Fix 'rusttest' build in macOS.

'kernel' crate:

- New 'hrtimer' module: add support for setting up intrusive timers
without allocating when starting the timer. Add support for
'Pin<Box<_>>', 'Arc<_>', 'Pin<&_>' and 'Pin<&mut _>' as pointer
types for use with timer callbacks. Add support for setting clock
source and timer mode.

- New 'dma' module: add a simple DMA coherent allocator abstraction
and a test sample driver.

- 'list' module: make the linked list 'Cursor' point between
elements, rather than at an element, which is more convenient to us
and allows for cursors to empty lists; and document it with
examples of how to perform common operations with the provided
methods.

- 'str' module: implement a few traits for 'BStr' as well as the
'strip_prefix()' method.

- 'sync' module: add 'Arc::as_ptr'.

- 'alloc' module: add 'Box::into_pin'.

- 'error' module: extend the 'Result' documentation, including a few
examples on different ways of handling errors, a warning about
using methods that may panic, and links to external documentation.

'macros' crate:

- 'module' macro: add the 'authors' key to support multiple authors.
The original key will be kept until everyone has migrated.

Documentation:

- Add error handling sections.

MAINTAINERS:

- Add Danilo Krummrich as reviewer of the Rust "subsystem".

- Add 'RUST [PIN-INIT]' entry with Benno Lossin as maintainer. It has
its own sub-tree.

- Add sub-tree for 'RUST [ALLOC]'.

- Add 'DMA MAPPING HELPERS DEVICE DRIVER API [RUST]' entry with
Abdiel Janulgue as primary maintainer. It will go through the
sub-tree of the 'RUST [ALLOC]' entry.

- Add 'HIGH-RESOLUTION TIMERS [RUST]' entry with Andreas Hindborg as
maintainer. It has its own sub-tree.

And a few other cleanups and improvements"

* tag 'rust-6.15' of git://git.kernel.org/pub/scm/linux/kernel/git/ojeda/linux: (71 commits)
rust: dma: add `Send` implementation for `CoherentAllocation`
rust: macros: fix `make rusttest` build on macOS
rust: block: refactor to use `&raw mut`
rust: enable `raw_ref_op` feature
rust: uaccess: name the correct function
rust: rbtree: fix comments referring to Box instead of KBox
rust: hrtimer: add maintainer entry
rust: hrtimer: add clocksource selection through `ClockId`
rust: hrtimer: add `HrTimerMode`
rust: hrtimer: implement `HrTimerPointer` for `Pin<Box<T>>`
rust: alloc: add `Box::into_pin`
rust: hrtimer: implement `UnsafeHrTimerPointer` for `Pin<&mut T>`
rust: hrtimer: implement `UnsafeHrTimerPointer` for `Pin<&T>`
rust: hrtimer: add `hrtimer::ScopedHrTimerPointer`
rust: hrtimer: add `UnsafeHrTimerPointer`
rust: hrtimer: allow timer restart from timer handler
rust: str: implement `strip_prefix` for `BStr`
rust: str: implement `AsRef<BStr>` for `[u8]` and `BStr`
rust: str: implement `Index` for `BStr`
rust: str: implement `PartialEq` for `BStr`
...

4e82c870

Linus Torvalds

9 months ago

ASoC: rt711-sdca: add DP0 support

9452422f

Pierre-Louis Bossart

10 months ago

tools/power turbostat: Restore GFX sysfs fflush() call

f8b136ef

Zhang Rui

9 months ago

branches 2

master 7 hours ago default

compare

for-next 11 months ago

compare

tags 914

v6.19-rc4

1 week ago latest

README

Linux kernel
============

There are several guides for kernel developers and users. These guides can
be rendered in a number of formats, like HTML and PDF. Please read
Documentation/admin-guide/README.rst first.

In order to build the documentation, use ``make htmldocs`` or
``make pdfdocs``.  The formatted documentation can also be read online at:

    https://www.kernel.org/doc/html/latest/

There are various text files in the Documentation/ subdirectory,
several of them using the reStructuredText markup notation.

Please read the Documentation/process/changes.rst file, as it contains the
requirements for building and running the kernel, and information about
the problems which may result by upgrading your kernel.

Clone this repository

Clone this repository