From: Robert Foley <robert.foley@linaro.org>
To: qemu-devel@nongnu.org
Cc: robert.foley@linaro.org, cota@braap.org,
Paolo Bonzini <pbonzini@redhat.com>,
peter.puhov@linaro.org, alex.bennee@linaro.org,
Richard Henderson <rth@twiddle.net>
Subject: [PATCH v10 73/73] cputlb: queue async flush jobs without the BQL
Date: Wed, 17 Jun 2020 17:02:31 -0400 [thread overview]
Message-ID: <20200617210231.4393-74-robert.foley@linaro.org> (raw)
In-Reply-To: <20200617210231.4393-1-robert.foley@linaro.org>
From: "Emilio G. Cota" <cota@braap.org>
This yields sizable scalability improvements, as the below results show.
Host: Two Intel Xeon Silver 4114 20-core CPUs at 2.20 GHz
VM: Ubuntu 18.04 ppc64
Speedup vs a single thread for kernel build
7 +-----------------------------------------------------------------------+
| + + + + + + |
| ########### baseline ******* |
| ##### #### cpu lock ####### |
| ## #### |
6 |-+ ## ## +-|
| ## #### |
| ## ### |
| ## ***** # |
| ## **** *** # |
| ## *** * |
5 |-+ ## *** **** +-|
| # **** ** |
| # ** ** |
| #* ** |
| #* ** |
| #* * |
| # ****** |
| # ** |
| # * |
3 |-+ # +-|
| # |
| # |
| # |
| # |
2 |-+ # +-|
| # |
| # |
| # |
| # |
| # + + + + + + |
1 +-----------------------------------------------------------------------+
0 5 10 15 20 25 30 35
Guest vCPUs
Pictures are also here:
https://drive.google.com/file/d/1ASg5XyP9hNfN9VysXC3qe5s9QSJlwFAt/view?usp=sharing
Some notes:
- baseline corresponds to the commit before this series
- cpu-lock is this series
Single-threaded performance is affected very lightly. Results
below for debian aarch64 bootup+test for the entire series
on an Intel(R) Core(TM) i7-6700K CPU @ 4.00GHz host:
- Before:
Performance counter stats for 'taskset -c 0 ../img/aarch64/die.sh' (10 runs):
7269.033478 task-clock (msec) # 0.998 CPUs utilized
( +- 0.06% )
30,659,870,302 cycles # 4.218 GHz
( +- 0.06% )
54,790,540,051 instructions # 1.79 insns per cycle
( +- 0.05% )
9,796,441,380 branches # 1347.695 M/sec
( +- 0.05% )
165,132,201 branch-misses # 1.69% of all branches
( +- 0.12% )
7.287011656 seconds time elapsed
( +- 0.10% )
- After:
7375.924053 task-clock (msec) # 0.998 CPUs utilized
( +- 0.13% )
31,107,548,846 cycles # 4.217 GHz
( +- 0.12% )
55,355,668,947 instructions # 1.78 insns per cycle
( +- 0.05% )
9,929,917,664 branches # 1346.261 M/sec
( +- 0.04% )
166,547,442 branch-misses # 1.68% of all branches
( +- 0.09% )
7.389068145 seconds time elapsed
( +- 0.13% )
That is, a 1.37% slowdown.
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Emilio G. Cota <cota@braap.org>
[Updated the speedup chart results for re-based series.]
Signed-off-by: Robert Foley <robert.foley@linaro.org>
---
accel/tcg/cputlb.c | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c
index 1e815357c7..7f75054643 100644
--- a/accel/tcg/cputlb.c
+++ b/accel/tcg/cputlb.c
@@ -299,7 +299,7 @@ static void flush_all_helper(CPUState *src, run_on_cpu_func fn,
CPU_FOREACH(cpu) {
if (cpu != src) {
- async_run_on_cpu(cpu, fn, d);
+ async_run_on_cpu_no_bql(cpu, fn, d);
}
}
}
@@ -367,8 +367,8 @@ void tlb_flush_by_mmuidx(CPUState *cpu, uint16_t idxmap)
tlb_debug("mmu_idx: 0x%" PRIx16 "\n", idxmap);
if (cpu->created && !qemu_cpu_is_self(cpu)) {
- async_run_on_cpu(cpu, tlb_flush_by_mmuidx_async_work,
- RUN_ON_CPU_HOST_INT(idxmap));
+ async_run_on_cpu_no_bql(cpu, tlb_flush_by_mmuidx_async_work,
+ RUN_ON_CPU_HOST_INT(idxmap));
} else {
tlb_flush_by_mmuidx_async_work(cpu, RUN_ON_CPU_HOST_INT(idxmap));
}
@@ -562,7 +562,7 @@ void tlb_flush_page_by_mmuidx(CPUState *cpu, target_ulong addr, uint16_t idxmap)
* we can stuff idxmap into the low TARGET_PAGE_BITS, avoid
* allocating memory for this operation.
*/
- async_run_on_cpu(cpu, tlb_flush_page_by_mmuidx_async_1,
+ async_run_on_cpu_no_bql(cpu, tlb_flush_page_by_mmuidx_async_1,
RUN_ON_CPU_TARGET_PTR(addr | idxmap));
} else {
TLBFlushPageByMMUIdxData *d = g_new(TLBFlushPageByMMUIdxData, 1);
@@ -570,7 +570,7 @@ void tlb_flush_page_by_mmuidx(CPUState *cpu, target_ulong addr, uint16_t idxmap)
/* Otherwise allocate a structure, freed by the worker. */
d->addr = addr;
d->idxmap = idxmap;
- async_run_on_cpu(cpu, tlb_flush_page_by_mmuidx_async_2,
+ async_run_on_cpu_no_bql(cpu, tlb_flush_page_by_mmuidx_async_2,
RUN_ON_CPU_HOST_PTR(d));
}
}
--
2.17.1
next prev parent reply other threads:[~2020-06-17 21:37 UTC|newest]
Thread overview: 75+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-06-17 21:01 [PATCH v10 00/73] per-CPU locks Robert Foley
2020-06-17 21:01 ` [PATCH v10 01/73] cpu: rename cpu->work_mutex to cpu->lock Robert Foley
2020-06-17 21:01 ` [PATCH v10 02/73] cpu: introduce cpu_mutex_lock/unlock Robert Foley
2020-06-17 21:01 ` [PATCH v10 03/73] cpu: make qemu_work_cond per-cpu Robert Foley
2020-06-17 21:01 ` [PATCH v10 04/73] cpu: move run_on_cpu to cpus-common Robert Foley
2020-06-17 21:01 ` [PATCH v10 05/73] cpu: introduce process_queued_cpu_work_locked Robert Foley
2020-06-17 21:01 ` [PATCH v10 06/73] cpu: make per-CPU locks an alias of the BQL in TCG rr mode Robert Foley
2020-06-17 21:01 ` [PATCH v10 07/73] tcg-runtime: define helper_cpu_halted_set Robert Foley
2020-06-17 21:01 ` [PATCH v10 08/73] ppc: convert to helper_cpu_halted_set Robert Foley
2020-06-17 21:01 ` [PATCH v10 09/73] cris: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 10/73] hppa: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 11/73] m68k: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 12/73] alpha: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 13/73] microblaze: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 14/73] cpu: define cpu_halted helpers Robert Foley
2020-06-17 21:01 ` [PATCH v10 15/73] tcg-runtime: convert to cpu_halted_set Robert Foley
2020-06-17 21:01 ` [PATCH v10 16/73] hw/semihosting: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 17/73] arm: convert to cpu_halted Robert Foley
2020-06-17 21:01 ` [PATCH v10 18/73] ppc: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 19/73] sh4: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 20/73] i386: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 21/73] lm32: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 22/73] m68k: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 23/73] mips: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 24/73] riscv: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 25/73] s390x: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 26/73] sparc: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 27/73] xtensa: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 28/73] gdbstub: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 29/73] openrisc: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 30/73] cpu-exec: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 31/73] cpu: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 32/73] cpu: define cpu_interrupt_request helpers Robert Foley
2020-06-17 21:01 ` [PATCH v10 33/73] ppc: use cpu_reset_interrupt Robert Foley
2020-06-17 21:01 ` [PATCH v10 34/73] exec: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 35/73] i386: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 36/73] s390x: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 37/73] openrisc: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 38/73] arm: convert to cpu_interrupt_request Robert Foley
2020-06-17 21:01 ` [PATCH v10 39/73] i386: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 40/73] i386/kvm: " Robert Foley
2020-06-17 21:01 ` [PATCH v10 41/73] i386/hax-all: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 42/73] i386/whpx-all: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 43/73] i386/hvf: convert to cpu_request_interrupt Robert Foley
2020-06-17 21:02 ` [PATCH v10 44/73] ppc: convert to cpu_interrupt_request Robert Foley
2020-06-17 21:02 ` [PATCH v10 45/73] sh4: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 46/73] cris: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 47/73] hppa: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 48/73] lm32: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 49/73] m68k: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 50/73] mips: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 51/73] nios: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 52/73] s390x: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 53/73] alpha: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 54/73] moxie: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 55/73] sparc: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 56/73] openrisc: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 57/73] unicore32: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 58/73] microblaze: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 59/73] accel/tcg: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 60/73] cpu: convert to interrupt_request Robert Foley
2020-06-17 21:02 ` [PATCH v10 61/73] cpu: call .cpu_has_work with the CPU lock held Robert Foley
2020-06-17 21:02 ` [PATCH v10 62/73] cpu: introduce cpu_has_work_with_iothread_lock Robert Foley
2020-06-17 21:02 ` [PATCH v10 63/73] ppc: convert to cpu_has_work_with_iothread_lock Robert Foley
2020-06-17 21:02 ` [PATCH v10 64/73] mips: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 65/73] s390x: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 66/73] riscv: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 67/73] sparc: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 68/73] xtensa: " Robert Foley
2020-06-17 21:02 ` [PATCH v10 69/73] cpu: rename all_cpu_threads_idle to qemu_tcg_rr_all_cpu_threads_idle Robert Foley
2020-06-17 21:02 ` [PATCH v10 70/73] cpu: protect CPU state with cpu->lock instead of the BQL Robert Foley
2020-06-17 21:02 ` [PATCH v10 71/73] cpus-common: release BQL earlier in run_on_cpu Robert Foley
2020-06-17 21:02 ` [PATCH v10 72/73] cpu: add async_run_on_cpu_no_bql Robert Foley
2020-06-17 21:02 ` Robert Foley [this message]
2020-06-17 22:20 ` [PATCH v10 00/73] per-CPU locks no-reply
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200617210231.4393-74-robert.foley@linaro.org \
--to=robert.foley@linaro.org \
--cc=alex.bennee@linaro.org \
--cc=cota@braap.org \
--cc=pbonzini@redhat.com \
--cc=peter.puhov@linaro.org \
--cc=qemu-devel@nongnu.org \
--cc=rth@twiddle.net \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).