Pārlūkot izejas kodu

qemu: change TLBs number to 64 in 34Kf mips cpu model

Replace OE private qemu patch with one that got upstreamed
and solves the same problem: increase qemumips CI performance
by increasing number of TLBs in CPU model and reduce need to
run software TLB refill code.

Signed-off-by: Victor Kamensky <kamensky@cisco.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
(cherry picked from commit a99dace7463d310688f4098a51316dc0743651e2)
Signed-off-by: Steve Sakoman <steve@sakoman.com>
Victor Kamensky 3 gadi atpakaļ
vecāks
revīzija
89e6fc44a3

+ 1 - 0
meta/recipes-devtools/qemu/qemu.inc

@@ -48,6 +48,7 @@ SRC_URI = "https://download.qemu.org/${BPN}-${PV}.tar.xz \
 	   file://CVE-2020-14364.patch \
 	   file://CVE-2020-14415.patch \
 	   file://CVE-2020-16092.patch \
+	   file://0001-target-mips-Increase-number-of-TLB-entries-on-the-34.patch \
 	   "
 UPSTREAM_CHECK_REGEX = "qemu-(?P<pver>\d+(\.\d+)+)\.tar"
 

+ 59 - 0
meta/recipes-devtools/qemu/qemu/0001-target-mips-Increase-number-of-TLB-entries-on-the-34.patch

@@ -0,0 +1,59 @@
+From 68fa519a6cb455005317bd61f95214b58b2f1e69 Mon Sep 17 00:00:00 2001
+From: =?UTF-8?q?Philippe=20Mathieu-Daud=C3=A9?= <f4bug@amsat.org>
+Date: Fri, 16 Oct 2020 15:20:37 +0200
+Subject: [PATCH] target/mips: Increase number of TLB entries on the 34Kf core
+ (16 -> 64)
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+Per "MIPS32 34K Processor Core Family Software User's Manual,
+Revision 01.13" page 8 in "Joint TLB (JTLB)" section:
+
+  "The JTLB is a fully associative TLB cache containing 16, 32,
+   or 64-dual-entries mapping up to 128 virtual pages to their
+   corresponding physical addresses."
+
+There is no particular reason to restrict the 34Kf core model to
+16 TLB entries, so raise its config to 64.
+
+This is helpful for other projects, in particular the Yocto Project:
+
+  Yocto Project uses qemu-system-mips 34Kf cpu model, to run 32bit
+  MIPS CI loop. It was observed that in this case CI test execution
+  time was almost twice longer than 64bit MIPS variant that runs
+  under MIPS64R2-generic model. It was investigated and concluded
+  that the difference in number of TLBs 16 in 34Kf case vs 64 in
+  MIPS64R2-generic is responsible for most of CI real time execution
+  difference. Because with 16 TLBs linux user-land trashes TLB more
+  and it needs to execute more instructions in TLB refill handler
+  calls, as result it runs much longer.
+
+(https://lists.gnu.org/archive/html/qemu-devel/2020-10/msg03428.html)
+
+Buglink: https://bugzilla.yoctoproject.org/show_bug.cgi?id=13992
+Reported-by: Victor Kamensky <kamensky@cisco.com>
+Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
+Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
+Message-Id: <20201016133317.553068-1-f4bug@amsat.org>
+
+Upstream-Status: Backport [https://github.com/qemu/qemu/commit/68fa519a6cb455005317bd61f95214b58b2f1e69]
+Signed-off-by: Victor Kamensky <kamensky@cisco.com>
+
+---
+ target/mips/translate_init.c.inc | 2 +-
+ 1 file changed, 1 insertion(+), 1 deletion(-)
+
+Index: qemu-5.1.0/target/mips/translate_init.inc.c
+===================================================================
+--- qemu-5.1.0.orig/target/mips/translate_init.inc.c
++++ qemu-5.1.0/target/mips/translate_init.inc.c
+@@ -254,7 +254,7 @@ const mips_def_t mips_defs[] =
+         .CP0_PRid = 0x00019500,
+         .CP0_Config0 = MIPS_CONFIG0 | (0x1 << CP0C0_AR) |
+                        (MMU_TYPE_R4000 << CP0C0_MT),
+-        .CP0_Config1 = MIPS_CONFIG1 | (1 << CP0C1_FP) | (15 << CP0C1_MMU) |
++        .CP0_Config1 = MIPS_CONFIG1 | (1 << CP0C1_FP) | (63 << CP0C1_MMU) |
+                        (0 << CP0C1_IS) | (3 << CP0C1_IL) | (1 << CP0C1_IA) |
+                        (0 << CP0C1_DS) | (3 << CP0C1_DL) | (1 << CP0C1_DA) |
+                        (1 << CP0C1_CA),