]> code.ossystems Code Review - openembedded-core.git/commitdiff
qemumips: use 34Kf-64tlb CPU emulation
authorVictor Kamensky <kamensky@cisco.com>
Wed, 7 Oct 2020 20:38:38 +0000 (13:38 -0700)
committerRichard Purdie <richard.purdie@linuxfoundation.org>
Thu, 8 Oct 2020 10:28:54 +0000 (11:28 +0100)
In order to improve performance of qemumips autobuilder
test runs, let's use 34Kf-64tlb cpu type that was introduced
in OE version of qemu. 34Kf-64tlb cpu type is identical to
34Kf but it has 64 TLBs configure vs original 16 TLBs.

The change in number of TLBs in emulated CPU reduces
TLB trash and reduces number of times TLB refill kernel
code runs, and therefore siginificantly improves tests
execution time.

Note 34Kf-64tlb qemu cpu type does not exist in upstream,
so far it is added as patch in OE. See qemu 34Kf-64tlb
cpu addition commit for more details.

Fixes https://bugzilla.yoctoproject.org/show_bug.cgi?id=13992

Signed-off-by: Victor Kamensky <kamensky@cisco.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
meta/conf/machine/qemumips.conf

index 1373e4cba01f8380ef2a5090823b4329f4bdc5b0..b8c80f02ef19b064b4b163bd7fad778e5ce3328f 100644 (file)
@@ -15,4 +15,4 @@ SERIAL_CONSOLES ?= "115200;ttyS0 115200;ttyS1"
 
 QB_SYSTEM_NAME = "qemu-system-mips"
 
-QB_CPU = "-cpu 34Kf"
+QB_CPU = "-cpu 34Kf-64tlb"