[PATCH -v2] sched,x86: optimize switch_mm for multi-threadedworkloads

From: Rik van Riel
Date: Wed Jul 31 2013 - 22:14:36 EST

On Wed, 31 Jul 2013 17:41:39 -0700
Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:

> However, as Rik points out, activate_mm() is different in that we
> shouldn't have any preexisting MMU state anyway. And besides, that
> should never trigger the "prev == next" case.
> But it does look a bit messy, and even your comment is a bit
> misleading (it might make somebody think that all of switch_mm() is
> protected from interrupts)

Is this better? Not that I really care which version gets applied :)


Subject: sched,x86: optimize switch_mm for multi-threaded workloads

Dick Fowles, Don Zickus and Joe Mario have been working on improvements
to perf, and noticed heavy cache line contention on the mm_cpumask,
running linpack on a 60 core / 120 thread system.

The cause turned out to be unnecessary atomic accesses to the
mm_cpumask. When in lazy TLB mode, the CPU is only removed from
the mm_cpumask if there is a TLB flush event.

Most of the time, no such TLB flush happens, and the kernel
skips the TLB reload. It can also skip the atomic memory
set & test.

Here is a summary of Joe's test results:

* The __schedule function dropped from 24% of all program cycles down
to 5.5%.
* The cacheline contention/hotness for accesses to that bitmask went
from being the 1st/2nd hottest - down to the 84th hottest (0.3% of
all shared misses which is now quite cold)
* The average load latency for the bit-test-n-set instruction in
__schedule dropped from 10k-15k cycles down to an average of 600 cycles.
* The linpack program results improved from 133 GFlops to 144 GFlops.
Peak GFlops rose from 133 to 153.

Reported-by: Don Zickus <dzickus@xxxxxxxxxx>
Reported-by: Joe Mario <jmario@xxxxxxxxxx>
Tested-by: Joe Mario <jmario@xxxxxxxxxx>
Signed-off-by: Rik van Riel <riel@xxxxxxxxxx>

Signed-off-by: Rik van Riel <riel@xxxxxxxxxx>
arch/x86/include/asm/mmu_context.h | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/arch/x86/include/asm/mmu_context.h b/arch/x86/include/asm/mmu_context.h
index cdbf367..3ac6089 100644
--- a/arch/x86/include/asm/mmu_context.h
+++ b/arch/x86/include/asm/mmu_context.h
@@ -59,7 +59,13 @@ static inline void switch_mm(struct mm_struct *prev, struct mm_struct *next,
this_cpu_write(cpu_tlbstate.state, TLBSTATE_OK);
BUG_ON(this_cpu_read(cpu_tlbstate.active_mm) != next);

- if (!cpumask_test_and_set_cpu(cpu, mm_cpumask(next))) {
+ if (!cpumask_test_cpu(cpu, mm_cpumask(next))) {
+ /* On established mms, the mm_cpumask is only changed
+ * from irq context, from ptep_clear_flush while in
+ * lazy tlb mode, and here. Irqs are blocked during
+ * schedule, protecting us from simultaneous changes.
+ */
+ cpumask_set_cpu(cpu, mm_cpumask(next));
/* We were in lazy tlb mode and leave_mm disabled
* tlb flush IPI delivery. We must reload CR3
* to make sure to use no freed page tables.
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/