kumquat-buildroot/package/gcc/6.2.0/831-ARM-PR-target-70473-Reduce-size-of-Cortex-A8-automat.patch
Thomas Petazzoni 8c97033a01 gcc: add patches to fix PR70473
gcc 5.x introduced a regression in the ARM build, which causes the
s-automata program to consume a very significant amount of RAM during
the gcc build. This causes numerous failures with our Travis-CI based
testing of defconfigs.

In order to address this, this commit backports a commit from the gcc
master branch, to both our gcc 5.x and gcc 6.x support.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
2016-10-15 16:38:58 +02:00

70 lines
2.5 KiB
Diff

From 07453188b18a2cbe815917dbb893b4ebfdd793b6 Mon Sep 17 00:00:00 2001
From: ktkachov <ktkachov@138bc75d-0d04-0410-961f-82ee72b054a4>
Date: Fri, 26 Aug 2016 10:23:20 +0000
Subject: [PATCH] [ARM] PR target/70473: Reduce size of Cortex-A8 automaton
PR target/70473
* config/arm/cortex-a8-neon.md (cortex_a8_vfp_muld): Reduce
reservation duration to 15 cycles.
(cortex_a8_vfp_macs): Likewise.
(cortex_a8_vfp_macd): Likewise.
(cortex_a8_vfp_divs): Likewise.
(cortex_a8_vfp_divd): Likewise.
git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/trunk@239772
138bc75d-0d04-0410-961f-82ee72b054a4
[Backport from gcc upstream commit 79f5d5e3cd5a668d5ecb2b6b0cce80ef5706ac53.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
---
gcc/config/arm/cortex-a8-neon.md | 14 +++++++++-----
1 file changed, 9 insertions(+), 5 deletions(-)
diff --git a/gcc/config/arm/cortex-a8-neon.md b/gcc/config/arm/cortex-a8-neon.md
index 45f861f..b16c299 100644
--- a/gcc/config/arm/cortex-a8-neon.md
+++ b/gcc/config/arm/cortex-a8-neon.md
@@ -357,30 +357,34 @@
(eq_attr "type" "fmuls"))
"cortex_a8_vfp,cortex_a8_vfplite*11")
+;; Don't model a reservation for more than 15 cycles as this explodes the
+;; state space of the automaton for little gain. It is unlikely that the
+;; scheduler will find enough instructions to hide the full latency of the
+;; instructions.
(define_insn_reservation "cortex_a8_vfp_muld" 17
(and (eq_attr "tune" "cortexa8")
(eq_attr "type" "fmuld"))
- "cortex_a8_vfp,cortex_a8_vfplite*16")
+ "cortex_a8_vfp,cortex_a8_vfplite*15")
(define_insn_reservation "cortex_a8_vfp_macs" 21
(and (eq_attr "tune" "cortexa8")
(eq_attr "type" "fmacs,ffmas"))
- "cortex_a8_vfp,cortex_a8_vfplite*20")
+ "cortex_a8_vfp,cortex_a8_vfplite*15")
(define_insn_reservation "cortex_a8_vfp_macd" 26
(and (eq_attr "tune" "cortexa8")
(eq_attr "type" "fmacd,ffmad"))
- "cortex_a8_vfp,cortex_a8_vfplite*25")
+ "cortex_a8_vfp,cortex_a8_vfplite*15")
(define_insn_reservation "cortex_a8_vfp_divs" 37
(and (eq_attr "tune" "cortexa8")
(eq_attr "type" "fdivs, fsqrts"))
- "cortex_a8_vfp,cortex_a8_vfplite*36")
+ "cortex_a8_vfp,cortex_a8_vfplite*15")
(define_insn_reservation "cortex_a8_vfp_divd" 65
(and (eq_attr "tune" "cortexa8")
(eq_attr "type" "fdivd, fsqrtd"))
- "cortex_a8_vfp,cortex_a8_vfplite*64")
+ "cortex_a8_vfp,cortex_a8_vfplite*15")
;; Comparisons can actually take 7 cycles sometimes instead of four,
;; but given all the other instructions lumped into type=ffarith that
--
2.7.4