diff options
author | Nicolas Boichat <drinkcat@chromium.org> | 2018-12-27 13:59:13 +0800 |
---|---|---|
committer | chrome-bot <chrome-bot@chromium.org> | 2019-01-24 00:51:28 -0800 |
commit | 3cd175cd9ddec3fd88153b3edb9e1d7e4526420d (patch) | |
tree | 0053ebf42be3cce87c539a248da05408d851bd38 /Makefile.rules | |
parent | fa379e1983ec568e6ca475311af3f77dec9327ff (diff) | |
download | chrome-ec-3cd175cd9ddec3fd88153b3edb9e1d7e4526420d.tar.gz |
mt_scp: Enable I/D-cache on boot
Enable cache mapping on boot, add functions for cache support.
Invalidate the cache on boot, and add benchmarching function.
BRANCH=none
BUG=b:117804463
TEST=Boot kukui_scp with TEST-ONLY CL.
TEST=Run dcachetest command, see performance difference,
and lack of coherency if the data is updated from AP side.
> dcachetest
cached: 19745 us (val: ef915230)
cached+inval: 39402 us (val: ef915230)
direct: 94096 us (val: ef915230)
TEST=See that icachetest (cached) and perftest have similar performance:
> icachetest
run from DRAM (cached): 102779 us (total: 12a052eb9)
run from DRAM (direct): 1867168 us (total: 12a052eb9)
> perftest
run from SRAM: 102566 us (total: 12a052eb9)
TEST=cacheinfo reports sensible data:
> cacheinfo
Icache hit count: 12000127
Icache access count: 12000131
Dcache hit count: 960034
Dcache access count: 1024034
Change-Id: I3272f4ff9edc0059c4937908b1235346c582c680
Signed-off-by: Nicolas Boichat <drinkcat@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/1391529
Reviewed-by: Rong Chang <rongchang@chromium.org>
Diffstat (limited to 'Makefile.rules')
0 files changed, 0 insertions, 0 deletions