Messages in this thread | | | From | John Stultz <> | Subject | [PATCH 0/5] Arch-generic sched_clock NMI safety and optimizations for -tip | Date | Tue, 17 Mar 2015 10:38:21 -0700 |
| |
Ingo, Peter, Thomas,
The following patches from Daniel extend the arch-generic sched_clock implementation so that it can be safely called from NMI (or FIQ on ARM) context. They also optimize the sched_clock logic to improve cache performance. I wanted to send them along so they could be queued in -tip for 4.1.
From Daniel's last post: "The data cache profile of sched_clock() in both the original code and my previous patch was somewhere between 2 and 3 (64-byte) cache lines, depending on alignment of struct clock_data. After patching, the cache profile for the normal case should be a single cacheline.
NMI safety was tested on i.MX6 with perf drowning the system in FIQs and using the perf handler to check that sched_clock() returned monotonic values. At the same time I forcefully reduced kt_wrap so that update_sched_clock() is being called at >1000Hz.
Without the patches the above system is grossly unstable, surviving [9K,115K,25K] perf event cycles during three separate runs. With the patch I ran for over 9M perf event cycles before getting bored."
I'm relaying these along because in the past I've queued and submitted arch-generic sched_clock.c changes from Stephen via Thomas, but I wanted to make sure this got a look from Ingo and Peter.
In the future, I think Stephen is probably the right person to act as patch-catcher for the kernel/time/sched_clock.c file. But I think those changes should continue to be merged in via the -tip tree w/ other scheduler code.
Let me know if you have any objections or concerns.
thanks -john
Cc: Daniel Thompson <daniel.thompson@linaro.org> Cc: Russell King <linux@arm.linux.org.uk> Cc: Will Deacon <will.deacon@arm.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Stephen Boyd <sboyd@codeaurora.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org>
These patches can also be pulled from the pull request below.
The following changes since commit c517d838eb7d07bbe9507871fab3931deccff539:
Linux 4.0-rc1 (2015-02-22 18:21:14 -0800)
are available in the git repository at:
https://git.linaro.org/people/john.stultz/linux.git fortglx/4.1/schedclock
for you to fetch changes up to 4bb74b85d2e74cfec9e21a6de44b232a2983bcc7:
sched_clock: Avoid deadlock during read from NMI (2015-03-10 20:28:58 -0700)
----------------------------------------------------------------
Daniel Thompson (5): sched_clock: Match scope of read and write seqcounts sched_clock: Optimize cache line usage sched_clock: Remove suspend from clock_read_data sched_clock: Remove redundant notrace from update function sched_clock: Avoid deadlock during read from NMI
kernel/time/sched_clock.c | 195 ++++++++++++++++++++++++++++++++-------------- 1 file changed, 138 insertions(+), 57 deletions(-)
-- 1.9.1
| |