* [Qemu-devel] [PATCH 1.3] build: compile translate.o at -O1 optimization
@ 2012-11-27 8:36 Paolo Bonzini
2012-11-27 9:27 ` Wenchao Xia
0 siblings, 1 reply; 16+ messages in thread
From: Paolo Bonzini @ 2012-11-27 8:36 UTC (permalink / raw)
To: qemu-devel
Some versions of GCC require insane (>2GB) amounts of memory
to compile translate.o. As a countermeasure, compile it
with -O1. This should fix the buildbot failure for
default_x86_64_fedora16.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
Makefile.target | 2 ++
1 file changed, 2 insertions(+)
diff --git a/Makefile.target b/Makefile.target
index 8b658c0..3981931 100644
--- a/Makefile.target
+++ b/Makefile.target
@@ -143,6 +143,8 @@ GENERATED_HEADERS += hmp-commands.h qmp-commands-old.h
endif # CONFIG_SOFTMMU
+%/translate.o: CFLAGS := $(patsubst -O2,-O1,$(CFLAGS))
+
nested-vars += obj-y
# This resolves all nested paths, so it must come last
--
1.8.0
^ permalink raw reply related [flat|nested] 16+ messages in thread
* Re: [Qemu-devel] [PATCH 1.3] build: compile translate.o at -O1 optimization
2012-11-27 8:36 [Qemu-devel] [PATCH 1.3] build: compile translate.o at -O1 optimization Paolo Bonzini
@ 2012-11-27 9:27 ` Wenchao Xia
2012-11-27 9:37 ` Paolo Bonzini
0 siblings, 1 reply; 16+ messages in thread
From: Wenchao Xia @ 2012-11-27 9:27 UTC (permalink / raw)
To: Paolo Bonzini; +Cc: qemu-devel
> Some versions of GCC require insane (>2GB) amounts of memory
> to compile translate.o. As a countermeasure, compile it
> with -O1. This should fix the buildbot failure for
> default_x86_64_fedora16.
>
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> ---
> Makefile.target | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/Makefile.target b/Makefile.target
> index 8b658c0..3981931 100644
> --- a/Makefile.target
> +++ b/Makefile.target
> @@ -143,6 +143,8 @@ GENERATED_HEADERS += hmp-commands.h qmp-commands-old.h
>
> endif # CONFIG_SOFTMMU
>
> +%/translate.o: CFLAGS := $(patsubst -O2,-O1,$(CFLAGS))
> +
> nested-vars += obj-y
>
> # This resolves all nested paths, so it must come last
>
In tcg case I think translate.o will influent performance obviously,
how about adding an option "fast-build" to use -O1 for it by default.
If you agree I will adding that after this patch upstream which fix
build bot failure quickly.
--
Best Regards
Wenchao Xia
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [Qemu-devel] [PATCH 1.3] build: compile translate.o at -O1 optimization
2012-11-27 9:27 ` Wenchao Xia
@ 2012-11-27 9:37 ` Paolo Bonzini
2012-11-27 12:09 ` Gerd Hoffmann
0 siblings, 1 reply; 16+ messages in thread
From: Paolo Bonzini @ 2012-11-27 9:37 UTC (permalink / raw)
To: Wenchao Xia; +Cc: qemu-devel
Il 27/11/2012 10:27, Wenchao Xia ha scritto:
> In tcg case I think translate.o will influent performance obviously,
> how about adding an option "fast-build" to use -O1 for it by default.
> If you agree I will adding that after this patch upstream which fix
> build bot failure quickly.
This is not about having a fast or slow build, it's about not requiring
a ludicrous amount of memory... Besides, translate.c is usually not too
high in the profiles. Most of the time is spent _executing_
JIT-translated code, not translating it.
Note that there is probably one or two GCC options that can be
fine-tuned to avoid the explosion instead of just -O1 vs. -O2. If you
have an affected machine (F18) you can help by compiling translate.c
with -O2 -ftime-report. I planned to do this today, but I first need to
install a F18 virtual machine.
Paolo
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [Qemu-devel] [PATCH 1.3] build: compile translate.o at -O1 optimization
2012-11-27 9:37 ` Paolo Bonzini
@ 2012-11-27 12:09 ` Gerd Hoffmann
0 siblings, 0 replies; 16+ messages in thread
From: Gerd Hoffmann @ 2012-11-27 12:09 UTC (permalink / raw)
To: Paolo Bonzini; +Cc: Wenchao Xia, qemu-devel
[-- Attachment #1: Type: text/plain, Size: 367 bytes --]
Hi,
> Note that there is probably one or two GCC options that can be
> fine-tuned to avoid the explosion instead of just -O1 vs. -O2. If you
> have an affected machine (F18) you can help by compiling translate.c
> with -O2 -ftime-report. I planned to do this today, but I first need to
> install a F18 virtual machine.
[x] done, see attachment
cheers,
Gerd
[-- Attachment #2: translate-times.txt --]
[-- Type: text/plain, Size: 11683 bytes --]
CC i386-softmmu/target-i386/translate.o
Execution times (seconds)
phase setup : 0.17 (100%) usr 0.06 (100%) sys 1.20 (98%) wall 2527 kB (99%) ggc
phase finalize : 0.00 ( 0%) usr 0.00 ( 0%) sys 0.02 ( 2%) wall 0 kB ( 0%) ggc
TOTAL : 0.17 0.06 1.23 2543 kB
Execution times (seconds)
phase setup : 0.00 ( 0%) usr 0.01 ( 0%) sys 0.11 ( 0%) wall 1077 kB ( 0%) ggc
phase parsing : 0.22 ( 1%) usr 0.41 ( 4%) sys 0.71 ( 1%) wall 13597 kB ( 1%) ggc
phase cgraph : 39.98 (99%) usr 9.43 (96%) sys 122.41 (99%) wall 1680634 kB (99%) ggc
phase generate : 39.98 (99%) usr 9.43 (96%) sys 122.42 (99%) wall 1680635 kB (99%) ggc
phase finalize : 0.01 ( 0%) usr 0.00 ( 0%) sys 0.09 ( 0%) wall 0 kB ( 0%) ggc
garbage collection : 0.31 ( 1%) usr 0.37 ( 4%) sys 29.45 (24%) wall 0 kB ( 0%) ggc
callgraph construction : 0.05 ( 0%) usr 0.04 ( 0%) sys 0.05 ( 0%) wall 6753 kB ( 0%) ggc
callgraph optimization : 0.00 ( 0%) usr 0.01 ( 0%) sys 0.04 ( 0%) wall 1401 kB ( 0%) ggc
varpool construction : 0.01 ( 0%) usr 0.00 ( 0%) sys 0.01 ( 0%) wall 282 kB ( 0%) ggc
ipa cp : 0.01 ( 0%) usr 0.00 ( 0%) sys 0.03 ( 0%) wall 384 kB ( 0%) ggc
ipa reference : 0.00 ( 0%) usr 0.00 ( 0%) sys 0.01 ( 0%) wall 0 kB ( 0%) ggc
ipa pure const : 0.04 ( 0%) usr 0.00 ( 0%) sys 0.03 ( 0%) wall 0 kB ( 0%) ggc
ipa SRA : 0.01 ( 0%) usr 0.01 ( 0%) sys 0.06 ( 0%) wall 3613 kB ( 0%) ggc
ipa free lang data : 0.00 ( 0%) usr 0.00 ( 0%) sys 0.02 ( 0%) wall 0 kB ( 0%) ggc
cfg construction : 0.02 ( 0%) usr 0.00 ( 0%) sys 0.02 ( 0%) wall 488 kB ( 0%) ggc
cfg cleanup : 0.36 ( 1%) usr 0.01 ( 0%) sys 0.74 ( 1%) wall 1440 kB ( 0%) ggc
trivially dead code : 0.17 ( 0%) usr 0.00 ( 0%) sys 0.17 ( 0%) wall 0 kB ( 0%) ggc
df scan insns : 0.09 ( 0%) usr 0.01 ( 0%) sys 0.08 ( 0%) wall 38 kB ( 0%) ggc
df multiple defs : 0.04 ( 0%) usr 0.00 ( 0%) sys 0.04 ( 0%) wall 0 kB ( 0%) ggc
df reaching defs : 0.10 ( 0%) usr 0.01 ( 0%) sys 0.30 ( 0%) wall 0 kB ( 0%) ggc
df live regs : 0.97 ( 2%) usr 0.06 ( 1%) sys 2.96 ( 2%) wall 0 kB ( 0%) ggc
df live&initialized regs: 0.46 ( 1%) usr 0.02 ( 0%) sys 0.69 ( 1%) wall 0 kB ( 0%) ggc
df use-def / def-use chains: 0.06 ( 0%) usr 0.00 ( 0%) sys 0.05 ( 0%) wall 0 kB ( 0%) ggc
df reg dead/unused notes: 0.31 ( 1%) usr 0.02 ( 0%) sys 0.43 ( 0%) wall 2796 kB ( 0%) ggc
register information : 0.08 ( 0%) usr 0.00 ( 0%) sys 0.10 ( 0%) wall 0 kB ( 0%) ggc
alias analysis : 0.22 ( 1%) usr 0.02 ( 0%) sys 0.34 ( 0%) wall 7380 kB ( 0%) ggc
alias stmt walking : 0.14 ( 0%) usr 0.07 ( 1%) sys 0.27 ( 0%) wall 2333 kB ( 0%) ggc
register scan : 0.03 ( 0%) usr 0.01 ( 0%) sys 0.03 ( 0%) wall 3 kB ( 0%) ggc
rebuild jump labels : 0.10 ( 0%) usr 0.00 ( 0%) sys 0.09 ( 0%) wall 0 kB ( 0%) ggc
preprocessing : 0.07 ( 0%) usr 0.15 ( 2%) sys 0.16 ( 0%) wall 1496 kB ( 0%) ggc
lexical analysis : 0.01 ( 0%) usr 0.13 ( 1%) sys 0.16 ( 0%) wall 0 kB ( 0%) ggc
parser (global) : 0.03 ( 0%) usr 0.04 ( 0%) sys 0.11 ( 0%) wall 6890 kB ( 0%) ggc
parser struct body : 0.00 ( 0%) usr 0.00 ( 0%) sys 0.03 ( 0%) wall 486 kB ( 0%) ggc
parser enumerator list : 0.00 ( 0%) usr 0.00 ( 0%) sys 0.01 ( 0%) wall 352 kB ( 0%) ggc
parser function body : 0.04 ( 0%) usr 0.02 ( 0%) sys 0.10 ( 0%) wall 1263 kB ( 0%) ggc
parser inl. func. body : 0.07 ( 0%) usr 0.07 ( 1%) sys 0.13 ( 0%) wall 3107 kB ( 0%) ggc
inline heuristics : 0.11 ( 0%) usr 0.02 ( 0%) sys 0.16 ( 0%) wall 3669 kB ( 0%) ggc
integration : 0.34 ( 1%) usr 0.31 ( 3%) sys 0.56 ( 0%) wall 41530 kB ( 2%) ggc
tree gimplify : 0.04 ( 0%) usr 0.02 ( 0%) sys 0.15 ( 0%) wall 3964 kB ( 0%) ggc
tree eh : 0.00 ( 0%) usr 0.00 ( 0%) sys 0.01 ( 0%) wall 2 kB ( 0%) ggc
tree CFG construction : 0.00 ( 0%) usr 0.00 ( 0%) sys 0.01 ( 0%) wall 1582 kB ( 0%) ggc
tree CFG cleanup : 0.14 ( 0%) usr 0.01 ( 0%) sys 0.05 ( 0%) wall 456 kB ( 0%) ggc
tree tail merge : 0.04 ( 0%) usr 0.01 ( 0%) sys 0.03 ( 0%) wall 6 kB ( 0%) ggc
tree VRP : 0.31 ( 1%) usr 0.02 ( 0%) sys 0.40 ( 0%) wall 6708 kB ( 0%) ggc
tree copy propagation : 0.12 ( 0%) usr 0.01 ( 0%) sys 0.18 ( 0%) wall 937 kB ( 0%) ggc
tree find ref. vars : 0.00 ( 0%) usr 0.00 ( 0%) sys 0.01 ( 0%) wall 127 kB ( 0%) ggc
tree PTA : 0.52 ( 1%) usr 0.28 ( 3%) sys 0.86 ( 1%) wall 2343 kB ( 0%) ggc
tree SSA rewrite : 0.05 ( 0%) usr 0.03 ( 0%) sys 0.14 ( 0%) wall 5755 kB ( 0%) ggc
tree SSA other : 0.01 ( 0%) usr 0.01 ( 0%) sys 0.02 ( 0%) wall 127 kB ( 0%) ggc
tree SSA incremental : 0.13 ( 0%) usr 0.02 ( 0%) sys 0.12 ( 0%) wall 680 kB ( 0%) ggc
tree operand scan : 0.19 ( 0%) usr 0.13 ( 1%) sys 0.37 ( 0%) wall 22560 kB ( 1%) ggc
dominator optimization : 0.12 ( 0%) usr 0.01 ( 0%) sys 0.14 ( 0%) wall 7169 kB ( 0%) ggc
tree CCP : 0.17 ( 0%) usr 0.00 ( 0%) sys 0.15 ( 0%) wall 670 kB ( 0%) ggc
tree PHI const/copy prop: 0.01 ( 0%) usr 0.00 ( 0%) sys 0.00 ( 0%) wall 0 kB ( 0%) ggc
tree reassociation : 0.03 ( 0%) usr 0.00 ( 0%) sys 0.02 ( 0%) wall 155 kB ( 0%) ggc
tree PRE : 0.18 ( 0%) usr 0.04 ( 0%) sys 0.26 ( 0%) wall 5335 kB ( 0%) ggc
tree FRE : 0.42 ( 1%) usr 0.10 ( 1%) sys 0.51 ( 0%) wall 12668 kB ( 1%) ggc
tree code sinking : 0.02 ( 0%) usr 0.00 ( 0%) sys 0.01 ( 0%) wall 77 kB ( 0%) ggc
tree linearize phis : 0.01 ( 0%) usr 0.00 ( 0%) sys 0.01 ( 0%) wall 2 kB ( 0%) ggc
tree forward propagate : 0.06 ( 0%) usr 0.02 ( 0%) sys 0.04 ( 0%) wall 2240 kB ( 0%) ggc
tree conservative DCE : 0.08 ( 0%) usr 0.05 ( 1%) sys 0.12 ( 0%) wall 4 kB ( 0%) ggc
tree aggressive DCE : 0.04 ( 0%) usr 0.03 ( 0%) sys 0.10 ( 0%) wall 1656 kB ( 0%) ggc
tree buildin call DCE : 0.01 ( 0%) usr 0.00 ( 0%) sys 0.00 ( 0%) wall 0 kB ( 0%) ggc
tree DSE : 0.02 ( 0%) usr 0.00 ( 0%) sys 0.05 ( 0%) wall 0 kB ( 0%) ggc
tree loop invariant motion: 0.01 ( 0%) usr 0.00 ( 0%) sys 0.00 ( 0%) wall 1 kB ( 0%) ggc
complete unrolling : 0.00 ( 0%) usr 0.00 ( 0%) sys 0.01 ( 0%) wall 182 kB ( 0%) ggc
tree iv optimization : 0.01 ( 0%) usr 0.00 ( 0%) sys 0.02 ( 0%) wall 105 kB ( 0%) ggc
tree loop init : 0.02 ( 0%) usr 0.00 ( 0%) sys 0.01 ( 0%) wall 149 kB ( 0%) ggc
tree SSA uncprop : 0.00 ( 0%) usr 0.00 ( 0%) sys 0.01 ( 0%) wall 0 kB ( 0%) ggc
tree rename SSA copies : 0.01 ( 0%) usr 0.00 ( 0%) sys 0.03 ( 0%) wall 0 kB ( 0%) ggc
tree STMT verifier : 0.00 ( 0%) usr 0.01 ( 0%) sys 0.00 ( 0%) wall 0 kB ( 0%) ggc
dominance computation : 0.11 ( 0%) usr 0.00 ( 0%) sys 0.10 ( 0%) wall 0 kB ( 0%) ggc
out of ssa : 0.06 ( 0%) usr 0.00 ( 0%) sys 0.05 ( 0%) wall 35 kB ( 0%) ggc
expand vars : 0.07 ( 0%) usr 0.01 ( 0%) sys 0.07 ( 0%) wall 3675 kB ( 0%) ggc
expand : 0.26 ( 1%) usr 0.07 ( 1%) sys 0.32 ( 0%) wall 23890 kB ( 1%) ggc
post expand cleanups : 0.02 ( 0%) usr 0.00 ( 0%) sys 0.01 ( 0%) wall 1633 kB ( 0%) ggc
varconst : 0.00 ( 0%) usr 0.00 ( 0%) sys 0.01 ( 0%) wall 0 kB ( 0%) ggc
jump : 0.00 ( 0%) usr 0.00 ( 0%) sys 0.01 ( 0%) wall 0 kB ( 0%) ggc
forward prop : 0.13 ( 0%) usr 0.03 ( 0%) sys 0.16 ( 0%) wall 2816 kB ( 0%) ggc
CSE : 0.38 ( 1%) usr 0.03 ( 0%) sys 0.62 ( 1%) wall 2433 kB ( 0%) ggc
dead code elimination : 0.08 ( 0%) usr 0.00 ( 0%) sys 0.07 ( 0%) wall 0 kB ( 0%) ggc
dead store elim1 : 0.15 ( 0%) usr 0.01 ( 0%) sys 0.19 ( 0%) wall 2431 kB ( 0%) ggc
dead store elim2 : 0.12 ( 0%) usr 0.00 ( 0%) sys 0.15 ( 0%) wall 1876 kB ( 0%) ggc
loop analysis : 0.00 ( 0%) usr 0.00 ( 0%) sys 0.01 ( 0%) wall 80 kB ( 0%) ggc
loop invariant motion : 0.02 ( 0%) usr 0.00 ( 0%) sys 0.16 ( 0%) wall 1 kB ( 0%) ggc
CPROP : 0.38 ( 1%) usr 0.03 ( 0%) sys 0.58 ( 0%) wall 2579 kB ( 0%) ggc
PRE : 26.78 (67%) usr 7.04 (71%) sys 72.36 (59%) wall 1444332 kB (85%) ggc
CSE 2 : 0.20 ( 0%) usr 0.01 ( 0%) sys 0.18 ( 0%) wall 653 kB ( 0%) ggc
branch prediction : 0.02 ( 0%) usr 0.00 ( 0%) sys 0.02 ( 0%) wall 319 kB ( 0%) ggc
combiner : 0.35 ( 1%) usr 0.01 ( 0%) sys 0.50 ( 0%) wall 6538 kB ( 0%) ggc
if-conversion : 0.30 ( 1%) usr 0.00 ( 0%) sys 0.65 ( 1%) wall 496 kB ( 0%) ggc
regmove : 0.06 ( 0%) usr 0.00 ( 0%) sys 0.12 ( 0%) wall 224 kB ( 0%) ggc
integrated RA : 1.03 ( 3%) usr 0.02 ( 0%) sys 1.15 ( 1%) wall 26022 kB ( 2%) ggc
reload : 0.51 ( 1%) usr 0.00 ( 0%) sys 0.58 ( 0%) wall 1732 kB ( 0%) ggc
reload CSE regs : 0.40 ( 1%) usr 0.00 ( 0%) sys 0.74 ( 1%) wall 3965 kB ( 0%) ggc
ree : 0.03 ( 0%) usr 0.01 ( 0%) sys 0.05 ( 0%) wall 162 kB ( 0%) ggc
thread pro- & epilogue : 0.04 ( 0%) usr 0.00 ( 0%) sys 0.04 ( 0%) wall 822 kB ( 0%) ggc
if-conversion 2 : 0.01 ( 0%) usr 0.00 ( 0%) sys 0.01 ( 0%) wall 84 kB ( 0%) ggc
combine stack adjustments: 0.01 ( 0%) usr 0.00 ( 0%) sys 0.01 ( 0%) wall 0 kB ( 0%) ggc
peephole 2 : 0.05 ( 0%) usr 0.00 ( 0%) sys 0.05 ( 0%) wall 228 kB ( 0%) ggc
hard reg cprop : 0.05 ( 0%) usr 0.01 ( 0%) sys 0.10 ( 0%) wall 9 kB ( 0%) ggc
scheduling 2 : 0.47 ( 1%) usr 0.29 ( 3%) sys 1.15 ( 1%) wall 328 kB ( 0%) ggc
machine dep reorg : 0.05 ( 0%) usr 0.00 ( 0%) sys 0.07 ( 0%) wall 44 kB ( 0%) ggc
reorder blocks : 0.04 ( 0%) usr 0.00 ( 0%) sys 0.05 ( 0%) wall 867 kB ( 0%) ggc
final : 0.20 ( 0%) usr 0.02 ( 0%) sys 0.33 ( 0%) wall 2060 kB ( 0%) ggc
symout : 0.00 ( 0%) usr 0.00 ( 0%) sys 0.01 ( 0%) wall 0 kB ( 0%) ggc
rest of compilation : 0.26 ( 1%) usr 0.02 ( 0%) sys 0.54 ( 0%) wall 1438 kB ( 0%) ggc
remove unused locals : 0.36 ( 1%) usr 0.00 ( 0%) sys 0.41 ( 0%) wall 0 kB ( 0%) ggc
address taken : 0.04 ( 0%) usr 0.00 ( 0%) sys 0.08 ( 0%) wall 0 kB ( 0%) ggc
unaccounted todo : 0.11 ( 0%) usr 0.03 ( 0%) sys 0.12 ( 0%) wall 0 kB ( 0%) ggc
rebuild frequencies : 0.01 ( 0%) usr 0.00 ( 0%) sys 0.00 ( 0%) wall 1 kB ( 0%) ggc
repair loop structures : 0.02 ( 0%) usr 0.00 ( 0%) sys 0.03 ( 0%) wall 23 kB ( 0%) ggc
TOTAL : 40.21 9.85 123.33 1695325 kB
^ permalink raw reply [flat|nested] 16+ messages in thread
* [Qemu-devel] [PATCH 1.3] build: compile translate.o at -O1 optimization
@ 2012-11-27 8:34 Paolo Bonzini
2012-11-27 13:24 ` Peter Maydell
2012-11-27 15:53 ` Avi Kivity
0 siblings, 2 replies; 16+ messages in thread
From: Paolo Bonzini @ 2012-11-27 8:34 UTC (permalink / raw)
To: qemu-devel
Some versions of GCC require insane (>2GB) amounts of memory
to compile translate.o. As a countermeasure, compile it
with -O1. This should fix the buildbot failure for
default_x86_64_fedora16.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
Makefile.target | 2 ++
1 file changed, 2 insertions(+)
diff --git a/Makefile.target b/Makefile.target
index 8b658c0..3981931 100644
--- a/Makefile.target
+++ b/Makefile.target
@@ -143,6 +143,8 @@ GENERATED_HEADERS += hmp-commands.h qmp-commands-old.h
endif # CONFIG_SOFTMMU
+%/translate.o: CFLAGS := $(patsubst -O2,-O1,$(CFLAGS))
+
nested-vars += obj-y
# This resolves all nested paths, so it must come last
--
1.8.0
^ permalink raw reply related [flat|nested] 16+ messages in thread
* Re: [Qemu-devel] [PATCH 1.3] build: compile translate.o at -O1 optimization
2012-11-27 8:34 Paolo Bonzini
@ 2012-11-27 13:24 ` Peter Maydell
2012-11-27 13:45 ` Gerd Hoffmann
2012-11-27 15:53 ` Avi Kivity
1 sibling, 1 reply; 16+ messages in thread
From: Peter Maydell @ 2012-11-27 13:24 UTC (permalink / raw)
To: Paolo Bonzini; +Cc: qemu-devel
On 27 November 2012 08:34, Paolo Bonzini <pbonzini@redhat.com> wrote:
> Some versions of GCC require insane (>2GB) amounts of memory
> to compile translate.o. As a countermeasure, compile it
> with -O1. This should fix the buildbot failure for
> default_x86_64_fedora16.
This is a well known bug in old gcc (ie fixed in 4.5, 4.6 and
trunk a year ago). Use a newer gcc, or a 64 bit build system
with a reasonable amount of RAM, or as a workaround apply some
suitable compiler flags by passing configure
'--extra-cflags=-fno-var-tracking'. This patch definitely
shouldn't be applied as we shouldn't be hampering the majority
for the benefit of old broken systems.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48190
https://bugs.launchpad.net/gcc-linaro/+bug/714921
-- PMM
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [Qemu-devel] [PATCH 1.3] build: compile translate.o at -O1 optimization
2012-11-27 13:24 ` Peter Maydell
@ 2012-11-27 13:45 ` Gerd Hoffmann
2012-11-27 13:46 ` Peter Maydell
2012-11-27 13:49 ` 陳韋任 (Wei-Ren Chen)
0 siblings, 2 replies; 16+ messages in thread
From: Gerd Hoffmann @ 2012-11-27 13:45 UTC (permalink / raw)
To: Peter Maydell; +Cc: Paolo Bonzini, qemu-devel
On 11/27/12 14:24, Peter Maydell wrote:
> On 27 November 2012 08:34, Paolo Bonzini <pbonzini@redhat.com> wrote:
>> Some versions of GCC require insane (>2GB) amounts of memory
>> to compile translate.o. As a countermeasure, compile it
>> with -O1. This should fix the buildbot failure for
>> default_x86_64_fedora16.
>
> This is a well known bug in old gcc (ie fixed in 4.5, 4.6 and
> trunk a year ago). Use a newer gcc, or a 64 bit build system
> with a reasonable amount of RAM, or as a workaround apply some
> suitable compiler flags by passing configure
> '--extra-cflags=-fno-var-tracking'. This patch definitely
> shouldn't be applied as we shouldn't be hampering the majority
> for the benefit of old broken systems.
It isn't that simple. It's Fedora 17 with gcc 4.7.2 which runs oom
while compiling translate.c
cheers,
Gerd
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [Qemu-devel] [PATCH 1.3] build: compile translate.o at -O1 optimization
2012-11-27 13:45 ` Gerd Hoffmann
@ 2012-11-27 13:46 ` Peter Maydell
2012-11-27 14:10 ` Paolo Bonzini
2012-11-27 13:49 ` 陳韋任 (Wei-Ren Chen)
1 sibling, 1 reply; 16+ messages in thread
From: Peter Maydell @ 2012-11-27 13:46 UTC (permalink / raw)
To: Gerd Hoffmann; +Cc: Paolo Bonzini, qemu-devel
On 27 November 2012 13:45, Gerd Hoffmann <kraxel@redhat.com> wrote:
> On 11/27/12 14:24, Peter Maydell wrote:
>> This is a well known bug in old gcc (ie fixed in 4.5, 4.6 and
>> trunk a year ago). Use a newer gcc, or a 64 bit build system
>> with a reasonable amount of RAM, or as a workaround apply some
>> suitable compiler flags by passing configure
>> '--extra-cflags=-fno-var-tracking'. This patch definitely
>> shouldn't be applied as we shouldn't be hampering the majority
>> for the benefit of old broken systems.
>
> It isn't that simple. It's Fedora 17 with gcc 4.7.2 which runs oom
> while compiling translate.c
In that case it is a new (or regressed) gcc bug and we should be
pursuing it with the upstream gcc folk. We still shouldn't be
putting random workarounds in our configure script.
-- PMM
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [Qemu-devel] [PATCH 1.3] build: compile translate.o at -O1 optimization
2012-11-27 13:46 ` Peter Maydell
@ 2012-11-27 14:10 ` Paolo Bonzini
0 siblings, 0 replies; 16+ messages in thread
From: Paolo Bonzini @ 2012-11-27 14:10 UTC (permalink / raw)
To: Peter Maydell; +Cc: Gerd Hoffmann, qemu-devel
Il 27/11/2012 14:46, Peter Maydell ha scritto:
> On 27 November 2012 13:45, Gerd Hoffmann <kraxel@redhat.com> wrote:
>> On 11/27/12 14:24, Peter Maydell wrote:
>>> This is a well known bug in old gcc (ie fixed in 4.5, 4.6 and
>>> trunk a year ago). Use a newer gcc, or a 64 bit build system
>>> with a reasonable amount of RAM, or as a workaround apply some
>>> suitable compiler flags by passing configure
>>> '--extra-cflags=-fno-var-tracking'. This patch definitely
>>> shouldn't be applied as we shouldn't be hampering the majority
>>> for the benefit of old broken systems.
>>
>> It isn't that simple. It's Fedora 17 with gcc 4.7.2 which runs oom
>> while compiling translate.c
>
> In that case it is a new (or regressed) gcc bug and we should be
> pursuing it with the upstream gcc folk. We still shouldn't be
> putting random workarounds in our configure script.
Sure we should, but we're also one week from release. Pretty sure a lot
of people will compile QEMU on a machine with <2GB and fail; the choice
is between documenting the problem and workaround in the release
information, or doing it for everyone in the Makefile.
Unfortunately, this one is not fixed by settling for worse debug
information, and it happens with GCC 4.7.2 (latest stable). It is fixed
in GCC trunk, but it is also possible that it is just latent because I
found that several GCC switches fix it:
* -fPIE is what causes it. Removing it obviously fixes it, but I'm not
sure it is a good idea (does it break linux-user self-virtualization?)
* -fno-gcse turns off the pass that explodes, and fixes it
* -fno-tree-pre turns off a similar but unrelated pass, but also fixes
it (found by chance because I confused it with -fno-gcse :)).
I think -fno-gcse is a valid workaround, and better than this patch.
Paolo
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [Qemu-devel] [PATCH 1.3] build: compile translate.o at -O1 optimization
2012-11-27 13:45 ` Gerd Hoffmann
2012-11-27 13:46 ` Peter Maydell
@ 2012-11-27 13:49 ` 陳韋任 (Wei-Ren Chen)
2012-11-27 15:05 ` Paolo Bonzini
2012-11-27 16:24 ` Andreas Färber
1 sibling, 2 replies; 16+ messages in thread
From: 陳韋任 (Wei-Ren Chen) @ 2012-11-27 13:49 UTC (permalink / raw)
To: Gerd Hoffmann; +Cc: Peter Maydell, qemu-devel, Paolo Bonzini
On Tue, Nov 27, 2012 at 02:45:07PM +0100, Gerd Hoffmann wrote:
> On 11/27/12 14:24, Peter Maydell wrote:
> > On 27 November 2012 08:34, Paolo Bonzini <pbonzini@redhat.com> wrote:
> >> Some versions of GCC require insane (>2GB) amounts of memory
> >> to compile translate.o. As a countermeasure, compile it
> >> with -O1. This should fix the buildbot failure for
> >> default_x86_64_fedora16.
> >
> > This is a well known bug in old gcc (ie fixed in 4.5, 4.6 and
> > trunk a year ago). Use a newer gcc, or a 64 bit build system
> > with a reasonable amount of RAM, or as a workaround apply some
> > suitable compiler flags by passing configure
> > '--extra-cflags=-fno-var-tracking'. This patch definitely
> > shouldn't be applied as we shouldn't be hampering the majority
> > for the benefit of old broken systems.
>
> It isn't that simple. It's Fedora 17 with gcc 4.7.2 which runs oom
> while compiling translate.c
Even apply Peter's suggestion? Do all gcc 4.7.2 on various platform
have the same problem, or it only happen on Fedora 17.
Regards,
chenwj
--
Wei-Ren Chen (陳韋任)
Computer Systems Lab, Institute of Information Science,
Academia Sinica, Taiwan (R.O.C.)
Tel:886-2-2788-3799 #1667
Homepage: http://people.cs.nctu.edu.tw/~chenwj
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [Qemu-devel] [PATCH 1.3] build: compile translate.o at -O1 optimization
2012-11-27 13:49 ` 陳韋任 (Wei-Ren Chen)
@ 2012-11-27 15:05 ` Paolo Bonzini
2012-11-27 16:24 ` Andreas Färber
1 sibling, 0 replies; 16+ messages in thread
From: Paolo Bonzini @ 2012-11-27 15:05 UTC (permalink / raw)
To: "陳韋任 (Wei-Ren Chen)"
Cc: Peter Maydell, Gerd Hoffmann, qemu-devel
Il 27/11/2012 14:49, 陳韋任 (Wei-Ren Chen) ha scritto:
>> > It isn't that simple. It's Fedora 17 with gcc 4.7.2 which runs oom
>> > while compiling translate.c
> Even apply Peter's suggestion? Do all gcc 4.7.2 on various platform
> have the same problem, or it only happen on Fedora 17.
All. It is reproducible with GCC's stable branch from the FSF
repository. I'm testing a patch, but it will take more than 1 week to
get it to the distros.
Paolo
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [Qemu-devel] [PATCH 1.3] build: compile translate.o at -O1 optimization
2012-11-27 13:49 ` 陳韋任 (Wei-Ren Chen)
2012-11-27 15:05 ` Paolo Bonzini
@ 2012-11-27 16:24 ` Andreas Färber
1 sibling, 0 replies; 16+ messages in thread
From: Andreas Färber @ 2012-11-27 16:24 UTC (permalink / raw)
To: "陳韋任 (Wei-Ren Chen)"
Cc: Peter Maydell, Paolo Bonzini, Gerd Hoffmann, qemu-devel
Am 27.11.2012 14:49, schrieb 陳韋任 (Wei-Ren Chen):
> On Tue, Nov 27, 2012 at 02:45:07PM +0100, Gerd Hoffmann wrote:
>> On 11/27/12 14:24, Peter Maydell wrote:
>>> On 27 November 2012 08:34, Paolo Bonzini <pbonzini@redhat.com> wrote:
>>>> Some versions of GCC require insane (>2GB) amounts of memory
>>>> to compile translate.o. As a countermeasure, compile it
>>>> with -O1. This should fix the buildbot failure for
>>>> default_x86_64_fedora16.
>>>
>>> This is a well known bug in old gcc (ie fixed in 4.5, 4.6 and
>>> trunk a year ago). Use a newer gcc, or a 64 bit build system
>>> with a reasonable amount of RAM, or as a workaround apply some
>>> suitable compiler flags by passing configure
>>> '--extra-cflags=-fno-var-tracking'. This patch definitely
>>> shouldn't be applied as we shouldn't be hampering the majority
>>> for the benefit of old broken systems.
>>
>> It isn't that simple. It's Fedora 17 with gcc 4.7.2 which runs oom
>> while compiling translate.c
>
> Even apply Peter's suggestion? Do all gcc 4.7.2 on various platform
> have the same problem, or it only happen on Fedora 17.
I ran into the same problem with various versions of openSUSE in the
openSUSE Build Service where v1.2 built fine.
openSUSE Factory is using gcc 4.7.2, 12.2 4.7.1.
Andreas
--
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [Qemu-devel] [PATCH 1.3] build: compile translate.o at -O1 optimization
2012-11-27 8:34 Paolo Bonzini
2012-11-27 13:24 ` Peter Maydell
@ 2012-11-27 15:53 ` Avi Kivity
2012-11-27 16:02 ` Paolo Bonzini
2012-11-27 16:49 ` Markus Armbruster
1 sibling, 2 replies; 16+ messages in thread
From: Avi Kivity @ 2012-11-27 15:53 UTC (permalink / raw)
To: Paolo Bonzini; +Cc: qemu-devel
On 11/27/2012 10:34 AM, Paolo Bonzini wrote:
> Some versions of GCC require insane (>2GB) amounts of memory
> to compile translate.o. As a countermeasure, compile it
> with -O1. This should fix the buildbot failure for
> default_x86_64_fedora16.
>
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> ---
> Makefile.target | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/Makefile.target b/Makefile.target
> index 8b658c0..3981931 100644
> --- a/Makefile.target
> +++ b/Makefile.target
> @@ -143,6 +143,8 @@ GENERATED_HEADERS += hmp-commands.h qmp-commands-old.h
>
> endif # CONFIG_SOFTMMU
>
> +%/translate.o: CFLAGS := $(patsubst -O2,-O1,$(CFLAGS))
> +
This may change some string argument in CFLAGS, for example an argument
to -I.
How about:
CFLAGS_opt = -O2
CFLAGS += $(CFLAGS_opt)
...
%/translate.o: CFLAGS_opt = -O1
--
error compiling committee.c: too many arguments to function
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [Qemu-devel] [PATCH 1.3] build: compile translate.o at -O1 optimization
2012-11-27 15:53 ` Avi Kivity
@ 2012-11-27 16:02 ` Paolo Bonzini
2012-11-27 16:49 ` Markus Armbruster
1 sibling, 0 replies; 16+ messages in thread
From: Paolo Bonzini @ 2012-11-27 16:02 UTC (permalink / raw)
To: Avi Kivity; +Cc: qemu-devel
Il 27/11/2012 16:53, Avi Kivity ha scritto:
> This may change some string argument in CFLAGS, for example an argument
> to -I.
>
> How about:
>
> CFLAGS_opt = -O2
> CFLAGS += $(CFLAGS_opt)
> ...
> %/translate.o: CFLAGS_opt = -O1
Not possible because CFLAGS comes from configure, but anyway what I'm
going to do is
%/translate.o: QEMU_CFLAGS += -fno-gcse
now that Gerd found the culprit patch.
Paolo
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [Qemu-devel] [PATCH 1.3] build: compile translate.o at -O1 optimization
2012-11-27 15:53 ` Avi Kivity
2012-11-27 16:02 ` Paolo Bonzini
@ 2012-11-27 16:49 ` Markus Armbruster
2012-11-27 16:55 ` Paolo Bonzini
1 sibling, 1 reply; 16+ messages in thread
From: Markus Armbruster @ 2012-11-27 16:49 UTC (permalink / raw)
To: Avi Kivity; +Cc: Paolo Bonzini, qemu-devel
Avi Kivity <avi@redhat.com> writes:
> On 11/27/2012 10:34 AM, Paolo Bonzini wrote:
>> Some versions of GCC require insane (>2GB) amounts of memory
>> to compile translate.o. As a countermeasure, compile it
>> with -O1. This should fix the buildbot failure for
>> default_x86_64_fedora16.
>>
>> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
>> ---
>> Makefile.target | 2 ++
>> 1 file changed, 2 insertions(+)
>>
>> diff --git a/Makefile.target b/Makefile.target
>> index 8b658c0..3981931 100644
>> --- a/Makefile.target
>> +++ b/Makefile.target
>> @@ -143,6 +143,8 @@ GENERATED_HEADERS += hmp-commands.h qmp-commands-old.h
>>
>> endif # CONFIG_SOFTMMU
>>
>> +%/translate.o: CFLAGS := $(patsubst -O2,-O1,$(CFLAGS))
>> +
>
> This may change some string argument in CFLAGS, for example an argument
> to -I.
>
> How about:
>
> CFLAGS_opt = -O2
> CFLAGS += $(CFLAGS_opt)
> ...
> %/translate.o: CFLAGS_opt = -O1
Just append -O1 to CFLAGS; the last -O wins.
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [Qemu-devel] [PATCH 1.3] build: compile translate.o at -O1 optimization
2012-11-27 16:49 ` Markus Armbruster
@ 2012-11-27 16:55 ` Paolo Bonzini
0 siblings, 0 replies; 16+ messages in thread
From: Paolo Bonzini @ 2012-11-27 16:55 UTC (permalink / raw)
To: Markus Armbruster; +Cc: Avi Kivity, qemu-devel
Il 27/11/2012 17:49, Markus Armbruster ha scritto:
>>> +%/translate.o: CFLAGS := $(patsubst -O2,-O1,$(CFLAGS))
>>> +
>>
>> This may change some string argument in CFLAGS, for example an argument
>> to -I.
>>
>> How about:
>>
>> CFLAGS_opt = -O2
>> CFLAGS += $(CFLAGS_opt)
>> ...
>> %/translate.o: CFLAGS_opt = -O1
>
> Just append -O1 to CFLAGS; the last -O wins.
But you don't want to override -O0...
Anyhow, v2 was posted and does it in a different, more fine-grained manner.
Paolo
^ permalink raw reply [flat|nested] 16+ messages in thread
end of thread, other threads:[~2012-11-27 17:32 UTC | newest]
Thread overview: 16+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2012-11-27 8:36 [Qemu-devel] [PATCH 1.3] build: compile translate.o at -O1 optimization Paolo Bonzini
2012-11-27 9:27 ` Wenchao Xia
2012-11-27 9:37 ` Paolo Bonzini
2012-11-27 12:09 ` Gerd Hoffmann
-- strict thread matches above, loose matches on Subject: below --
2012-11-27 8:34 Paolo Bonzini
2012-11-27 13:24 ` Peter Maydell
2012-11-27 13:45 ` Gerd Hoffmann
2012-11-27 13:46 ` Peter Maydell
2012-11-27 14:10 ` Paolo Bonzini
2012-11-27 13:49 ` 陳韋任 (Wei-Ren Chen)
2012-11-27 15:05 ` Paolo Bonzini
2012-11-27 16:24 ` Andreas Färber
2012-11-27 15:53 ` Avi Kivity
2012-11-27 16:02 ` Paolo Bonzini
2012-11-27 16:49 ` Markus Armbruster
2012-11-27 16:55 ` Paolo Bonzini
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).