* [Bug rtl-optimization/38644] [4.3/4.4/4.5/4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
@ 2010-09-30 19:16 ` sebastian.huber@embedded-brains.de
2011-01-31 21:50 ` joel at gcc dot gnu.org
` (41 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: sebastian.huber@embedded-brains.de @ 2010-09-30 19:16 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #32 from Sebastian Huber <sebastian.huber@embedded-brains.de> 2010-09-30 15:36:02 UTC ---
Which target milestone do you intend for a fix? It is still present in 4.6.0
20100925.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.3/4.4/4.5/4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
2010-09-30 19:16 ` [Bug rtl-optimization/38644] [4.3/4.4/4.5/4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code sebastian.huber@embedded-brains.de
@ 2011-01-31 21:50 ` joel at gcc dot gnu.org
2011-02-07 16:39 ` law at redhat dot com
` (40 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: joel at gcc dot gnu.org @ 2011-01-31 21:50 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
Joel Sherrill <joel at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Known to fail| |
--- Comment #33 from Joel Sherrill <joel at gcc dot gnu.org> 2011-01-31 21:26:08 UTC ---
Any chance this can get some attention before 4.6 branches? Please.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.3/4.4/4.5/4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
2010-09-30 19:16 ` [Bug rtl-optimization/38644] [4.3/4.4/4.5/4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code sebastian.huber@embedded-brains.de
2011-01-31 21:50 ` joel at gcc dot gnu.org
@ 2011-02-07 16:39 ` law at redhat dot com
2011-04-26 15:16 ` [Bug rtl-optimization/38644] [4.3/4.4/4.5/4.6/4.7 " jiangning.liu at arm dot com
` (39 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: law at redhat dot com @ 2011-02-07 16:39 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
Jeffrey A. Law <law at redhat dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |law at redhat dot com
--- Comment #34 from Jeffrey A. Law <law at redhat dot com> 2011-02-07 16:27:24 UTC ---
Typically the right thing to do is to block all memory motions across a change
in the stack pointer. It's somewhat overly pessimistic, but in reality the
few motions lost aren't going to be performance critical.
In the past each backend has emitted the blockage/barrier and it typically
happened soon after the port was converted to use RTL prologues/epilogues...
That's probably the main reason why this was never fixed in the scheduler
itself -- the first couple ports emitted a blockage and after that it became
normal practice.
I would support both emitting the suitable blockage insn in the ARM backend or
adding a dependency between the stack pointer adjustment insn and all memory
insns in the scheduler. Either is IMHO acceptable given history. The first
would be slightly preferred during this late stage of development with the
latter being more appropriate in early stage development.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.3/4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (2 preceding siblings ...)
2011-02-07 16:39 ` law at redhat dot com
@ 2011-04-26 15:16 ` jiangning.liu at arm dot com
2011-06-27 14:30 ` rguenth at gcc dot gnu.org
` (38 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: jiangning.liu at arm dot com @ 2011-04-26 15:16 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
Jiangning Liu <jiangning.liu at arm dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |jiangning.liu at arm dot
| |com
--- Comment #35 from Jiangning Liu <jiangning.liu at arm dot com> 2011-04-26 15:13:41 UTC ---
I verified that two patches in #38644 (back end) and #30282 (middle end) both
work for attached cases. Here comes my two cents,
1) The concept of red zone is to control whether instructions can write memory
below current stack frame or not, and it is only being supported by ABIs for
some particular ISAs, so it shouldn't be enabled in middle end by default for
all targets. At this point, middle end should be fixed to avoid doing things
unwanted in general for all targets.
2) Red zone is an orthogonal concept to prologue/epilogue, so it is not good to
fix this issue in prologue/epilogue back end code. At this point, we shouldn't
simply fix it in back end by adding barrier to implicitly disable red zone.
Instead, some hooks should be abstracted in middle end to support it in
scheduling dependence (middle end code). Back end like X86-64 should enable it
through hooks by itself.
The key here is red zone should be a clean feature to be supported in middle
end. Exposing this kind of stuff to back end through hooks can improve code
quality for middle end and avoid bringing the bugs to back-end.
This bug has long history, and it is being now or has ever been exposed on ARM,
POWER and X86(with some options combination). Fixing it in middle end is not
only a bug fix, but a simple infrastructure improvement. Due to the long
duration and the extensive impact for different targets, I don't see good
reason of not fixing it in mainline ASAP.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.3/4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (3 preceding siblings ...)
2011-04-26 15:16 ` [Bug rtl-optimization/38644] [4.3/4.4/4.5/4.6/4.7 " jiangning.liu at arm dot com
@ 2011-06-27 14:30 ` rguenth at gcc dot gnu.org
2011-08-04 12:35 ` [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 " sebastian.huber@embedded-brains.de
` (37 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: rguenth at gcc dot gnu.org @ 2011-06-27 14:30 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
Richard Guenther <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Target Milestone|4.3.6 |4.4.7
--- Comment #36 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-06-27 12:14:12 UTC ---
4.3 branch is being closed, moving to 4.4.7 target.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (4 preceding siblings ...)
2011-06-27 14:30 ` rguenth at gcc dot gnu.org
@ 2011-08-04 12:35 ` sebastian.huber@embedded-brains.de
2011-08-05 4:00 ` ramana.r at gmail dot com
` (36 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: sebastian.huber@embedded-brains.de @ 2011-08-04 12:35 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #37 from Sebastian Huber <sebastian.huber@embedded-brains.de> 2011-08-04 12:30:29 UTC ---
Is this problem also related to this bug (GCC 4.6.1 20110627) with comments
inside:
objdump -d -C /opt/rtems-4.11/lib/gcc/arm-rtems4.11/4.6.1/thumb/vfp/libstdc++.a
| grep -A 94 new_opnt.o
new_opnt.o: file format elf32-littlearm
Disassembly of section .text._ZnwmRKSt9nothrow_t:
00000000 <operator new(unsigned long, std::nothrow_t const&)>:
0: b5f0 push {r4, r5, r6, r7, lr}
2: 465f mov r7, fp
4: 4656 mov r6, sl
6: 464d mov r5, r9
8: 4644 mov r4, r8
a: b4f0 push {r4, r5, r6, r7}
c: b090 sub sp, #64 ; 0x40
e: 4b2a ldr r3, [pc, #168] ; (b8 <operator new(unsigned
long, std::nothrow_t const&)+0xb8>)
10: af00 add r7, sp, #0
12: 627b str r3, [r7, #36] ; 0x24
14: 4b29 ldr r3, [pc, #164] ; (bc <operator new(unsigned
long, std::nothrow_t const&)+0xbc>)
16: 62bb str r3, [r7, #40] ; 0x28
18: 4b29 ldr r3, [pc, #164] ; (c0 <operator new(unsigned
long, std::nothrow_t const&)+0xc0>)
1a: 6078 str r0, [r7, #4]
1c: 2240 movs r2, #64 ; 0x40
1e: 1c38 adds r0, r7, #0
20: 19d2 adds r2, r2, r7
22: 633b str r3, [r7, #48] ; 0x30
24: 300c adds r0, #12
26: 466b mov r3, sp
28: 62fa str r2, [r7, #44] ; 0x2c
2a: 637b str r3, [r7, #52] ; 0x34
2c: f7ff fffe bl 0 <_Unwind_SjLj_Register>
30: 687a ldr r2, [r7, #4]
32: 2a00 cmp r2, #0
34: d101 bne.n 3a <operator new(unsigned long, std::nothrow_t
const&)+0x3a>
36: 2301 movs r3, #1
38: 607b str r3, [r7, #4]
3a: 6878 ldr r0, [r7, #4]
3c: f7ff fffe bl 0 <malloc>
40: 6038 str r0, [r7, #0]
42: 2800 cmp r0, #0
44: d123 bne.n 8e <operator new(unsigned long, std::nothrow_t
const&)+0x8e>
46: 4a1f ldr r2, [pc, #124] ; (c4 <operator new(unsigned
long, std::nothrow_t const&)+0xc4>)
48: 6813 ldr r3, [r2, #0]
4a: 2b00 cmp r3, #0
4c: d104 bne.n 58 <operator new(unsigned long, std::nothrow_t
const&)+0x58>
4e: e021 b.n 94 <operator new(unsigned long, std::nothrow_t
const&)+0x94>
50: 4a1c ldr r2, [pc, #112] ; (c4 <operator new(unsigned
long, std::nothrow_t const&)+0xc4>)
52: 6813 ldr r3, [r2, #0]
54: 2b00 cmp r3, #0
56: d009 beq.n 6c <operator new(unsigned long, std::nothrow_t
const&)+0x6c>
58: 2201 movs r2, #1
5a: 613a str r2, [r7, #16]
5c: f000 f834 bl c8 <operator new(unsigned long, std::nothrow_t
const&)+0xc8>
60: 6878 ldr r0, [r7, #4]
62: f7ff fffe bl 0 <malloc>
66: 60b8 str r0, [r7, #8]
68: 2800 cmp r0, #0
6a: d0f1 beq.n 50 <operator new(unsigned long, std::nothrow_t
const&)+0x50>
6c: 1c38 adds r0, r7, #0
6e: 300c adds r0, #12
70: f7ff fffe bl 0 <_Unwind_SjLj_Unregister>
BAD CODE BEGIN
74: 46bd mov sp, r7
r7 is now the current stack pointer.
76: b010 add sp, #64 ; 0x40
Current stack frame is free now, r7 points to obsolete stack frame.
78: 68b8 ldr r0, [r7, #8]
Here we read from the stack frame freed previously. This is a disaster in
multi-threaded environments, because the exception code will use the stack of
an interrupted thread.
BAD CODE END
7a: bc3c pop {r2, r3, r4, r5}
7c: 4690 mov r8, r2
7e: 4699 mov r9, r3
80: 46a2 mov sl, r4
82: 46ab mov fp, r5
84: bdf0 pop {r4, r5, r6, r7, pc}
86: f7ff fffe bl 0 <__cxa_begin_catch>
8a: f7ff fffe bl 0 <__cxa_end_catch>
8e: 683b ldr r3, [r7, #0]
90: 60bb str r3, [r7, #8]
92: e7eb b.n 6c <operator new(unsigned long, std::nothrow_t
const&)+0x6c>
94: 2200 movs r2, #0
96: 60ba str r2, [r7, #8]
98: e7e8 b.n 6c <operator new(unsigned long, std::nothrow_t
const&)+0x6c>
9a: 3f40 subs r7, #64 ; 0x40
9c: 69bb ldr r3, [r7, #24]
9e: 6978 ldr r0, [r7, #20]
a0: 2b01 cmp r3, #1
a2: d0f0 beq.n 86 <operator new(unsigned long, std::nothrow_t
const&)+0x86>
a4: 1c5a adds r2, r3, #1
a6: d004 beq.n b2 <operator new(unsigned long, std::nothrow_t
const&)+0xb2>
a8: 2301 movs r3, #1
aa: 425b negs r3, r3
ac: 613b str r3, [r7, #16]
ae: f7ff fffe bl 0 <_Unwind_SjLj_Resume>
b2: 613b str r3, [r7, #16]
b4: f7ff fffe bl 0 <__cxa_call_unexpected>
...
c0: 0000009a .word 0x0000009a
c4: 00000000 .word 0x00000000
c8: 4718 bx r3
ca: 46c0 nop ; (mov r8, r8)
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (5 preceding siblings ...)
2011-08-04 12:35 ` [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 " sebastian.huber@embedded-brains.de
@ 2011-08-05 4:00 ` ramana.r at gmail dot com
2011-08-05 6:50 ` sebastian.huber@embedded-brains.de
` (35 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: ramana.r at gmail dot com @ 2011-08-05 4:00 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #39 from Ramana Radhakrishnan <ramana.r at gmail dot com> 2011-08-05 03:57:14 UTC ---
On Fri, Aug 5, 2011 at 2:33 AM, jiangning.liu at arm dot com
<gcc-bugzilla@gcc.gnu.org> wrote:
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
>
> --- Comment #38 from Jiangning Liu <jiangning.liu at arm dot com> 2011-08-05 01:33:06 UTC ---
> Hi Sebastian,
>
> I'm fixing this bug, and will send out a formal patch to community soon.
>
> Is it possible you send me your source code exposing the bug first, so I can
> verify your problem can really be fixed with my local fix?
I think this is just from the libstdc++ sources - look at that file
from the build with --with-mode=thumb .
Ramana
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (6 preceding siblings ...)
2011-08-05 4:00 ` ramana.r at gmail dot com
@ 2011-08-05 6:50 ` sebastian.huber@embedded-brains.de
2011-08-09 2:08 ` jiangning.liu at arm dot com
` (34 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: sebastian.huber@embedded-brains.de @ 2011-08-05 6:50 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #40 from Sebastian Huber <sebastian.huber@embedded-brains.de> 2011-08-05 06:49:00 UTC ---
(In reply to comment #39)
> On Fri, Aug 5, 2011 at 2:33 AM, jiangning.liu at arm dot com
[...]
> > Is it possible you send me your source code exposing the bug first, so I can
> > verify your problem can really be fixed with my local fix?
>
> I think this is just from the libstdc++ sources - look at that file
> from the build with --with-mode=thumb .
Yes, this is from the libstdc++ sources (4.6.1 20110627,
libstdc++-v3/libsupc++/new_opnt.cc). You need a non-EABI ARM variant of GCC
since this bug manifestation will only show up in the SJLJ version.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (7 preceding siblings ...)
2011-08-05 6:50 ` sebastian.huber@embedded-brains.de
@ 2011-08-09 2:08 ` jiangning.liu at arm dot com
2011-08-15 8:09 ` lingyouzeng@arimacomm-hz.cn
` (33 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: jiangning.liu at arm dot com @ 2011-08-09 2:08 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #41 from Jiangning Liu <jiangning.liu at arm dot com> 2011-08-09 02:04:52 UTC ---
> Yes, this is from the libstdc++ sources (4.6.1 20110627,
> libstdc++-v3/libsupc++/new_opnt.cc). You need a non-EABI ARM variant of GCC
> since this bug manifestation will only show up in the SJLJ version.
I tried and my local patch works on this case. As you can see like below, it is
fixed!
add r0, r0, #12
bl _Unwind_SjLj_Unregister
ldr r0, [r7, #8]
mov sp, r7
add sp, sp, #68
@ sp needed for prologue
pop {r2, r3, r4, r5}
Thanks,
-Jiangning
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (8 preceding siblings ...)
2011-08-09 2:08 ` jiangning.liu at arm dot com
@ 2011-08-15 8:09 ` lingyouzeng@arimacomm-hz.cn
2011-09-06 7:46 ` sebastian.huber@embedded-brains.de
` (32 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: lingyouzeng@arimacomm-hz.cn @ 2011-08-15 8:09 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #42 from LYZ <lingyouzeng@arimacomm-hz.cn> 2011-08-15 08:08:46 UTC ---
*** Bug 50081 has been marked as a duplicate of this bug. ***
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (9 preceding siblings ...)
2011-08-15 8:09 ` lingyouzeng@arimacomm-hz.cn
@ 2011-09-06 7:46 ` sebastian.huber@embedded-brains.de
2011-09-09 13:48 ` joel at gcc dot gnu.org
` (31 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: sebastian.huber@embedded-brains.de @ 2011-09-06 7:46 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #43 from Sebastian Huber <sebastian.huber@embedded-brains.de> 2011-09-06 07:45:29 UTC ---
How long will this middle to back end ping pong last until this bug is finally
fixed? It is now open since 2008.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (10 preceding siblings ...)
2011-09-06 7:46 ` sebastian.huber@embedded-brains.de
@ 2011-09-09 13:48 ` joel at gcc dot gnu.org
2011-09-11 15:47 ` steven at gcc dot gnu.org
` (30 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: joel at gcc dot gnu.org @ 2011-09-09 13:48 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #44 from Joel Sherrill <joel at gcc dot gnu.org> 2011-09-09 13:27:54 UTC ---
Ping.. any chance of getting the proposed 4.6 fix merged? Please.
This is almost a 3 year old bug. It would be nice to get it closed.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (11 preceding siblings ...)
2011-09-09 13:48 ` joel at gcc dot gnu.org
@ 2011-09-11 15:47 ` steven at gcc dot gnu.org
2011-09-12 8:48 ` sebastian.huber@embedded-brains.de
` (29 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: steven at gcc dot gnu.org @ 2011-09-11 15:47 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
Steven Bosscher <steven at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |steven at gcc dot gnu.org
--- Comment #45 from Steven Bosscher <steven at gcc dot gnu.org> 2011-09-11 15:42:50 UTC ---
(In reply to comment #41)
> I tried and my local patch works on this case. As you can see like below, it is
> fixed!
So, where is this local patch? And why not attach it here or post it to
gcc-patches, instead of keeping it local? ;)
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (12 preceding siblings ...)
2011-09-11 15:47 ` steven at gcc dot gnu.org
@ 2011-09-12 8:48 ` sebastian.huber@embedded-brains.de
2011-09-12 15:32 ` law at redhat dot com
` (28 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: sebastian.huber@embedded-brains.de @ 2011-09-12 8:48 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #46 from Sebastian Huber <sebastian.huber@embedded-brains.de> 2011-09-12 08:42:59 UTC ---
I guess the local patch will look like this:
http://gcc.gnu.org/ml/gcc/2011-07/msg00459.html
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (13 preceding siblings ...)
2011-09-12 8:48 ` sebastian.huber@embedded-brains.de
@ 2011-09-12 15:32 ` law at redhat dot com
2011-09-12 15:37 ` rearnsha at arm dot com
` (27 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: law at redhat dot com @ 2011-09-12 15:32 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #47 from Jeffrey A. Law <law at redhat dot com> 2011-09-12 15:18:44 UTC ---
On 09/12/2011 02:42 AM, sebastian.huber@embedded-brains.de wrote:
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
>
> --- Comment #46 from Sebastian Huber<sebastian.huber@embedded-brains.de> 2011-09-12 08:42:59 UTC ---
> I guess the local patch will look like this:
>
> http://gcc.gnu.org/ml/gcc/2011-07/msg00459.html
A much simpler way to fix this is to emit a barrier just prior to
mucking around with stack pointer in the epilogue. That's how targets
have dealt with this exact issue for a couple decades.
JEff
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (14 preceding siblings ...)
2011-09-12 15:32 ` law at redhat dot com
@ 2011-09-12 15:37 ` rearnsha at arm dot com
2011-09-12 15:50 ` law at redhat dot com
` (26 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: rearnsha at arm dot com @ 2011-09-12 15:37 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #48 from Richard Earnshaw <rearnsha at arm dot com> 2011-09-12 15:31:51 UTC ---
On 12/09/11 16:18, law at redhat dot com wrote:
> A much simpler way to fix this is to emit a barrier just prior to
> mucking around with stack pointer in the epilogue. That's how targets
> have dealt with this exact issue for a couple decades.
Simpler, but wrong. The compiler should not be generating unsafe code
by default. The problem is in the mid-end and expecting every port to
get this right in order to work-around a mid-end bug is just stupid
stupid stupid.
The mid end should not be scheduling around stack moves unless it has
been explicitly told it is safe to do this. I don't understand why
there is so much resistance to fixing the problem properly.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (15 preceding siblings ...)
2011-09-12 15:37 ` rearnsha at arm dot com
@ 2011-09-12 15:50 ` law at redhat dot com
2011-09-12 18:40 ` steven at gcc dot gnu.org
` (25 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: law at redhat dot com @ 2011-09-12 15:50 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #49 from Jeffrey A. Law <law at redhat dot com> 2011-09-12 15:33:56 UTC ---
On 09/12/2011 09:31 AM, rearnsha at arm dot com wrote:
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
>
> --- Comment #48 from Richard Earnshaw<rearnsha at arm dot com> 2011-09-12 15:31:51 UTC ---
> On 12/09/11 16:18, law at redhat dot com wrote:
>
>> A much simpler way to fix this is to emit a barrier just prior to
>> mucking around with stack pointer in the epilogue. That's how targets
>> have dealt with this exact issue for a couple decades.
> Simpler, but wrong. The compiler should not be generating unsafe code
> by default. The problem is in the mid-end and expecting every port to
> get this right in order to work-around a mid-end bug is just stupid
> stupid stupid.
>
> The mid end should not be scheduling around stack moves unless it has
> been explicitly told it is safe to do this. I don't understand why
> there is so much resistance to fixing the problem properly.
I don't disagree with you Richard, but we're at, what, 3 years on this
bug... That's absurd, particularly when there's a trivial, correct fix.
jeff
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (16 preceding siblings ...)
2011-09-12 15:50 ` law at redhat dot com
@ 2011-09-12 18:40 ` steven at gcc dot gnu.org
2011-09-26 8:11 ` rguenther at suse dot de
` (24 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: steven at gcc dot gnu.org @ 2011-09-12 18:40 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #50 from Steven Bosscher <steven at gcc dot gnu.org> 2011-09-12 18:24:58 UTC ---
Perhaps someone can comment on, and test, Joern's patch here:
http://gcc.gnu.org/ml/gcc/2011-07/msg00461.html
@Richard E: That would be a middle-end fix, if it's correct, so perhaps you
could do the honors?
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (17 preceding siblings ...)
2011-09-12 18:40 ` steven at gcc dot gnu.org
@ 2011-09-26 8:11 ` rguenther at suse dot de
2011-10-15 8:49 ` sebastian.huber@embedded-brains.de
` (23 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: rguenther at suse dot de @ 2011-09-26 8:11 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #51 from rguenther at suse dot de <rguenther at suse dot de> 2011-09-26 08:04:37 UTC ---
On Mon, 12 Sep 2011, rearnsha at arm dot com wrote:
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
>
> --- Comment #48 from Richard Earnshaw <rearnsha at arm dot com> 2011-09-12 15:31:51 UTC ---
> On 12/09/11 16:18, law at redhat dot com wrote:
>
> > A much simpler way to fix this is to emit a barrier just prior to
> > mucking around with stack pointer in the epilogue. That's how targets
> > have dealt with this exact issue for a couple decades.
>
> Simpler, but wrong. The compiler should not be generating unsafe code
> by default. The problem is in the mid-end and expecting every port to
> get this right in order to work-around a mid-end bug is just stupid
> stupid stupid.
>
> The mid end should not be scheduling around stack moves unless it has
> been explicitly told it is safe to do this. I don't understand why
> there is so much resistance to fixing the problem properly.
The middle-end does not treat stack moves specially, they are just
memory accesses. Extra dependences have to be modeled accordingly.
It's a hack to treat stack moves specially, not a proper fix.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (18 preceding siblings ...)
2011-09-26 8:11 ` rguenther at suse dot de
@ 2011-10-15 8:49 ` sebastian.huber@embedded-brains.de
2011-10-24 13:09 ` sebastian.huber@embedded-brains.de
` (22 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: sebastian.huber@embedded-brains.de @ 2011-10-15 8:49 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #52 from Sebastian Huber <sebastian.huber@embedded-brains.de> 2011-10-15 08:48:10 UTC ---
In GCC 4.6.2 20111014 (prerelease) the problem is still not fixed and
"arm-eabi-gcc -march=armv5t -mthumb -O2" produces wrong code. Please don't let
it slip through the next release.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (19 preceding siblings ...)
2011-10-15 8:49 ` sebastian.huber@embedded-brains.de
@ 2011-10-24 13:09 ` sebastian.huber@embedded-brains.de
2011-10-28 7:34 ` sebastian.huber@embedded-brains.de
` (21 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: sebastian.huber@embedded-brains.de @ 2011-10-24 13:09 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #53 from Sebastian Huber <sebastian.huber@embedded-brains.de> 2011-10-24 13:06:03 UTC ---
I tested the above patch proposed by Mikael Pettersson (from 2010-05-26, more
than one year ago) with GCC 4.6 20111021. It still fixes the test case
provided by Dave Murphy (from 2008-12-27, nearly three years ago). I run the
GCC C and C++ testsuite with the arm-rtemseabi4.11 target and the results are
identical with and without the patch. Even if we have no perfect solution yet,
it would be very kind if someone can check in an intermediate fix.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (20 preceding siblings ...)
2011-10-24 13:09 ` sebastian.huber@embedded-brains.de
@ 2011-10-28 7:34 ` sebastian.huber@embedded-brains.de
2011-10-29 23:29 ` davem at devkitpro dot org
` (20 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: sebastian.huber@embedded-brains.de @ 2011-10-28 7:34 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #54 from Sebastian Huber <sebastian.huber@embedded-brains.de> 2011-10-28 07:31:19 UTC ---
I tested with GCC 4.6.2 and the patch provided by Mikael Pettersson. It works
for -march=armv4t and -march=armv5t, but not for -march=armv5te:
--- test-armv5te.s 2011-10-28 09:22:24.627388063 +0200
+++ test-armv5t.s 2011-10-28 09:22:19.923643155 +0200
@@ -1,4 +1,4 @@
- .arch armv5te
+ .arch armv5t
.fpu softvfp
.eabi_attribute 20, 1
.eabi_attribute 21, 1
@@ -27,8 +27,8 @@
mov r1, r4
mov r2, #1
bl doStreamReadBlock
- add sp, sp, #8
ldrb r0, [r4]
+ add sp, sp, #8
@ sp needed for prologue
pop {r4, pc}
.size readStream, .-readStream
Command line:
arm-eabi-gcc -O2 -march=armv5t -mthumb -S test.c -o test-armv5t.s
arm-eabi-gcc -O2 -march=armv5te -mthumb -S test.c -o test-armv5te.s
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (21 preceding siblings ...)
2011-10-28 7:34 ` sebastian.huber@embedded-brains.de
@ 2011-10-29 23:29 ` davem at devkitpro dot org
2011-10-31 7:52 ` jiangning.liu at arm dot com
` (19 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: davem at devkitpro dot org @ 2011-10-29 23:29 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #55 from Dave Murphy <davem at devkitpro dot org> 2011-10-29 23:27:02 UTC ---
(In reply to comment #54)
> I tested with GCC 4.6.2 and the patch provided by Mikael Pettersson. It works
> for -march=armv4t and -march=armv5t, but not for -march=armv5te:
For what it's worth I've been using Richard Earnshaw's patch from
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30282#c8 with my own gcc builds and
it's working fine for all -march values.
There's also Joern's patch at http://gcc.gnu.org/ml/gcc/2011-07/msg00461.html
which I haven't tested but looks like it should work.
I still don't understand why there seems to be so much resistance to Richard's
suggestion that targets with redzones should explicitly enable this behaviour.
How can it be a hack to treat stack moves specially? Isn't the stack generally
a special register?
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (22 preceding siblings ...)
2011-10-29 23:29 ` davem at devkitpro dot org
@ 2011-10-31 7:52 ` jiangning.liu at arm dot com
2011-10-31 8:34 ` mikpe at it dot uu.se
` (18 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: jiangning.liu at arm dot com @ 2011-10-31 7:52 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #56 from Jiangning Liu <jiangning.liu at arm dot com> 2011-10-31 07:48:25 UTC ---
(In reply to comment #54)
> I tested with GCC 4.6.2 and the patch provided by Mikael Pettersson. It works
> for -march=armv4t and -march=armv5t, but not for -march=armv5te:
>
Sebastian,
Actually you may try this,
diff --git a/gcc/config/arm/arm.c b/gcc/config/arm/arm.c
index aed748c..8269c1a 100755
--- a/gcc/config/arm/arm.c
+++ b/gcc/config/arm/arm.c
@@ -22273,6 +22273,8 @@ thumb1_expand_epilogue (void)
gcc_assert (amount >= 0);
if (amount)
{
+ emit_insn (gen_blockage ());
+
if (amount < 512)
emit_insn (gen_addsi3 (stack_pointer_rtx, stack_pointer_rtx,
GEN_INT (amount)));
Thanks,
-Jiangning
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (23 preceding siblings ...)
2011-10-31 7:52 ` jiangning.liu at arm dot com
@ 2011-10-31 8:34 ` mikpe at it dot uu.se
2011-10-31 10:46 ` sebastian.huber@embedded-brains.de
` (17 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: mikpe at it dot uu.se @ 2011-10-31 8:34 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #57 from Mikael Pettersson <mikpe at it dot uu.se> 2011-10-31 08:32:09 UTC ---
(In reply to comment #56)
> (In reply to comment #54)
> > I tested with GCC 4.6.2 and the patch provided by Mikael Pettersson. It works
> > for -march=armv4t and -march=armv5t, but not for -march=armv5te:
> >
>
> Sebastian,
>
> Actually you may try this,
>
> diff --git a/gcc/config/arm/arm.c b/gcc/config/arm/arm.c
> index aed748c..8269c1a 100755
> --- a/gcc/config/arm/arm.c
> +++ b/gcc/config/arm/arm.c
> @@ -22273,6 +22273,8 @@ thumb1_expand_epilogue (void)
> gcc_assert (amount >= 0);
> if (amount)
> {
> + emit_insn (gen_blockage ());
> +
> if (amount < 512)
> emit_insn (gen_addsi3 (stack_pointer_rtx, stack_pointer_rtx,
> GEN_INT (amount)));
>
> Thanks,
> -Jiangning
Yeah, that should be even better than my patch. Thanks.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (24 preceding siblings ...)
2011-10-31 8:34 ` mikpe at it dot uu.se
@ 2011-10-31 10:46 ` sebastian.huber@embedded-brains.de
2011-11-04 16:53 ` jye2 at gcc dot gnu.org
` (16 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: sebastian.huber@embedded-brains.de @ 2011-10-31 10:46 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #58 from Sebastian Huber <sebastian.huber@embedded-brains.de> 2011-10-31 10:45:43 UTC ---
I tested Jiangning Liu's latest patch. With it GCC 4.6.2 produces valid code
for -march=armv4t, -march=armv5t, -march=armv5te, -march=armv6, and
-march=armv7-m. GCC 4.6.2 produces valid code for -march-armv7-m also without
the patch, for all other listed options the code is wrong.
Testsuite results:
http://gcc.gnu.org/ml/gcc-testresults/2011-10/msg03523.html
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6/4.7 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (25 preceding siblings ...)
2011-10-31 10:46 ` sebastian.huber@embedded-brains.de
@ 2011-11-04 16:53 ` jye2 at gcc dot gnu.org
2011-11-09 9:17 ` [Bug rtl-optimization/38644] [4.4/4.5/4.6 " sebastian.huber@embedded-brains.de
` (15 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: jye2 at gcc dot gnu.org @ 2011-11-04 16:53 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #59 from jye2 at gcc dot gnu.org 2011-11-04 16:50:11 UTC ---
Author: jye2
Date: Fri Nov 4 16:50:04 2011
New Revision: 180964
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=180964
Log:
2011-11-04 Jiangning Liu <jiangning.liu@arm.com>
PR rtl-optimization/38644
* config/arm/arm.c (thumb1_expand_epilogue): Add memory barrier
for epilogue having stack adjustment.
testcase:
* gcc.target/arm/stack-red-zone.c: New.
Added:
trunk/gcc/testsuite/gcc.target/arm/stack-red-zone.c
Modified:
trunk/gcc/ChangeLog
trunk/gcc/config/arm/arm.c
trunk/gcc/testsuite/ChangeLog
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (26 preceding siblings ...)
2011-11-04 16:53 ` jye2 at gcc dot gnu.org
@ 2011-11-09 9:17 ` sebastian.huber@embedded-brains.de
2011-11-16 10:33 ` liujiangning at gcc dot gnu.org
` (14 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: sebastian.huber@embedded-brains.de @ 2011-11-09 9:17 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #60 from Sebastian Huber <sebastian.huber@embedded-brains.de> 2011-11-09 09:00:26 UTC ---
Jiangning Liu thank you very much for your update.
The target milestone is currently 4.4.7. Are there plans to commit this fix
the the 4.4, 4.5, and 4.6 branches?
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (27 preceding siblings ...)
2011-11-09 9:17 ` [Bug rtl-optimization/38644] [4.4/4.5/4.6 " sebastian.huber@embedded-brains.de
@ 2011-11-16 10:33 ` liujiangning at gcc dot gnu.org
2012-01-09 16:58 ` ramana at gcc dot gnu.org
` (13 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: liujiangning at gcc dot gnu.org @ 2011-11-16 10:33 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #61 from Jiangning Liu <liujiangning at gcc dot gnu.org> 2011-11-16 09:47:01 UTC ---
Author: liujiangning
Date: Wed Nov 16 09:46:58 2011
New Revision: 181406
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=181406
Log:
2011-11-16 Jiangning Liu <jiangning.liu@arm.com>
Backport r180964 from mainline
2011-11-04 Jiangning Liu <jiangning.liu@arm.com>
PR rtl-optimization/38644
* config/arm/arm.c (thumb1_expand_epilogue): Add memory barrier
for epilogue having stack adjustment.
testsuite:
2011-11-16 Jiangning Liu <jiangning.liu@arm.com>
Backport r180964 from mainline
2011-11-04 Jiangning Liu <jiangning.liu@arm.com>
PR rtl-optimization/38644
* gcc.target/arm/stack-red-zone.c: New.
Added:
branches/ARM/embedded-4_6-branch/gcc/testsuite/gcc.target/arm/stack-red-zone.c
Modified:
branches/ARM/embedded-4_6-branch/gcc/ChangeLog.arm
branches/ARM/embedded-4_6-branch/gcc/config/arm/arm.c
branches/ARM/embedded-4_6-branch/gcc/testsuite/ChangeLog.arm
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.4/4.5/4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (28 preceding siblings ...)
2011-11-16 10:33 ` liujiangning at gcc dot gnu.org
@ 2012-01-09 16:58 ` ramana at gcc dot gnu.org
2012-03-13 14:55 ` [Bug rtl-optimization/38644] [4.5/4.6 " jakub at gcc dot gnu.org
` (12 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: ramana at gcc dot gnu.org @ 2012-01-09 16:58 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #62 from Ramana Radhakrishnan <ramana at gcc dot gnu.org> 2012-01-09 16:55:24 UTC ---
Author: ramana
Date: Mon Jan 9 16:55:16 2012
New Revision: 183019
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=183019
Log:
2012-01-09 Ramana Radhakrishnan <ramana.radhakrishnan@linaro.org>
Backport from mainline
2011-11-04 Jiangning Liu <jiangning.liu@arm.com>
PR rtl-optimization/38644
* config/arm/arm.c (thumb1_expand_epilogue): Add memory barrier
for epilogue having stack adjustment.
2012-01-09 Ramana Radhakrishnan <ramana.radhakrishnan@linaro.org>
Backport from mainline:
2011-11-04 Jiangning Liu <jiangning.liu@arm.com>
PR rtl-optimization/38644
* gcc.target/arm/stack-red-zone.c: New.
Added:
branches/gcc-4_6-branch/gcc/testsuite/gcc.target/arm/stack-red-zone.c
Modified:
branches/gcc-4_6-branch/gcc/ChangeLog
branches/gcc-4_6-branch/gcc/config/arm/arm.c
branches/gcc-4_6-branch/gcc/testsuite/ChangeLog
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.5/4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (29 preceding siblings ...)
2012-01-09 16:58 ` ramana at gcc dot gnu.org
@ 2012-03-13 14:55 ` jakub at gcc dot gnu.org
2012-07-02 11:35 ` rguenth at gcc dot gnu.org
` (11 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: jakub at gcc dot gnu.org @ 2012-03-13 14:55 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Target Milestone|4.4.7 |4.5.4
--- Comment #63 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-03-13 12:47:29 UTC ---
4.4 branch is being closed, moving to 4.5.4 target.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.5/4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (30 preceding siblings ...)
2012-03-13 14:55 ` [Bug rtl-optimization/38644] [4.5/4.6 " jakub at gcc dot gnu.org
@ 2012-07-02 11:35 ` rguenth at gcc dot gnu.org
2012-07-31 16:29 ` [Bug rtl-optimization/38644] [4.6 " hagayg at broadcom dot com
` (10 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: rguenth at gcc dot gnu.org @ 2012-07-02 11:35 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
Richard Guenther <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Target Milestone|4.5.4 |4.6.4
--- Comment #64 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-07-02 11:32:12 UTC ---
The 4.5 branch is being closed, adjusting target milestone.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (31 preceding siblings ...)
2012-07-02 11:35 ` rguenth at gcc dot gnu.org
@ 2012-07-31 16:29 ` hagayg at broadcom dot com
2012-07-31 16:53 ` pinskia at gcc dot gnu.org
` (9 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: hagayg at broadcom dot com @ 2012-07-31 16:29 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
Hagay <hagayg at broadcom dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |hagayg at broadcom dot com
--- Comment #65 from Hagay <hagayg at broadcom dot com> 2012-07-31 16:26:15 UTC ---
Also saw the same problem on MIPS
Compiling using 'gcc version 4.5.2 (Sourcery CodeBench Lite 2011.09-86)'
Produces the below assembly (mips16)
4c13 addiu a0,19
6478 restore 64,ra,s0-s1
f3a6 dd50 sw v0,13232(a1)
e820 jr ra
8c40 lh v0,0(a0)
6500 nop
You can see the 'lh' comes after 'restore' , so the problem exist on MIPS
Sourcery as well.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (32 preceding siblings ...)
2012-07-31 16:29 ` [Bug rtl-optimization/38644] [4.6 " hagayg at broadcom dot com
@ 2012-07-31 16:53 ` pinskia at gcc dot gnu.org
2012-07-31 17:37 ` hagayg at broadcom dot com
` (8 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: pinskia at gcc dot gnu.org @ 2012-07-31 16:53 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #66 from Andrew Pinski <pinskia at gcc dot gnu.org> 2012-07-31 16:49:54 UTC ---
(In reply to comment #65)
> Also saw the same problem on MIPS
> Compiling using 'gcc version 4.5.2 (Sourcery CodeBench Lite 2011.09-86)'
You should report that issue to Mentor because it was fixed with:
2011-10-02 Richard Sandiford <rdsandiford@googlemail.com>
* config/mips/mips.c (mips_frame_barrier): New function.
(mips_expand_prologue): Call it after allocating stack space.
(mips_deallocate_stack): New function.
(mips_expand_epilogue): Call mips_frame_barrier and
mips_deallocate_stack.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (33 preceding siblings ...)
2012-07-31 16:53 ` pinskia at gcc dot gnu.org
@ 2012-07-31 17:37 ` hagayg at broadcom dot com
2013-04-05 3:51 ` peter at axium dot co.nz
` (7 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: hagayg at broadcom dot com @ 2012-07-31 17:37 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #67 from Hagay <hagayg at broadcom dot com> 2012-07-31 17:34:02 UTC ---
Just reported to Mentor. Thanks
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (34 preceding siblings ...)
2012-07-31 17:37 ` hagayg at broadcom dot com
@ 2013-04-05 3:51 ` peter at axium dot co.nz
2013-04-05 4:23 ` pinskia at gcc dot gnu.org
` (6 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: peter at axium dot co.nz @ 2013-04-05 3:51 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
incrediball <peter at axium dot co.nz> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |peter at axium dot co.nz
--- Comment #68 from incrediball <peter at axium dot co.nz> 2013-04-05 03:51:14 UTC ---
Is this problem resolved? The status is still set to "NEW" but "known to work"
shows that it either has been resolved in v4.7.0 or that version is somehow not
affected.
I spent the best part of this week trying to track this problem down. It was
VERY hard to reproduce but I eventually found that the -fschedule-insns2
compile option caused the problem. The code that was affected is almost exactly
the same as the test case in comment #1.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (35 preceding siblings ...)
2013-04-05 3:51 ` peter at axium dot co.nz
@ 2013-04-05 4:23 ` pinskia at gcc dot gnu.org
2013-04-05 7:15 ` sebastian.huber@embedded-brains.de
` (5 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: pinskia at gcc dot gnu.org @ 2013-04-05 4:23 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |RESOLVED
Resolution| |FIXED
Target Milestone|4.6.4 |4.6.0
--- Comment #69 from Andrew Pinski <pinskia at gcc dot gnu.org> 2013-04-05 04:23:43 UTC ---
(In reply to comment #68)
> Is this problem resolved? The status is still set to "NEW" but "known to work"
> shows that it either has been resolved in v4.7.0 or that version is somehow not
> affected.
It was resolved with the patch in comment #59 and comment #62. In GCC 4.6.0.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (36 preceding siblings ...)
2013-04-05 4:23 ` pinskia at gcc dot gnu.org
@ 2013-04-05 7:15 ` sebastian.huber@embedded-brains.de
2014-02-16 10:01 ` jackie.rosen at hushmail dot com
` (4 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: sebastian.huber@embedded-brains.de @ 2013-04-05 7:15 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #70 from Sebastian Huber <sebastian.huber@embedded-brains.de> 2013-04-05 07:15:34 UTC ---
(In reply to comment #69)
> (In reply to comment #68)
> > Is this problem resolved? The status is still set to "NEW" but "known to work"
> > shows that it either has been resolved in v4.7.0 or that version is somehow not
> > affected.
>
> It was resolved with the patch in comment #59 and comment #62. In GCC 4.6.0.
It was fixed in GCC 4.6.3 and above.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (37 preceding siblings ...)
2013-04-05 7:15 ` sebastian.huber@embedded-brains.de
@ 2014-02-16 10:01 ` jackie.rosen at hushmail dot com
2022-10-14 9:56 ` cvs-commit at gcc dot gnu.org
` (3 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: jackie.rosen at hushmail dot com @ 2014-02-16 10:01 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
Jackie Rosen <jackie.rosen at hushmail dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |jackie.rosen at hushmail dot com
--- Comment #71 from Jackie Rosen <jackie.rosen at hushmail dot com> ---
*** Bug 260998 has been marked as a duplicate of this bug. ***
Seen from the domain http://volichat.com
Marked for reference. Resolved as fixed @bugzilla.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (38 preceding siblings ...)
2014-02-16 10:01 ` jackie.rosen at hushmail dot com
@ 2022-10-14 9:56 ` cvs-commit at gcc dot gnu.org
2022-10-14 9:57 ` cvs-commit at gcc dot gnu.org
` (2 subsequent siblings)
42 siblings, 0 replies; 43+ messages in thread
From: cvs-commit at gcc dot gnu.org @ 2022-10-14 9:56 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #74 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Eric Botcazou <ebotcazou@gcc.gnu.org>:
https://gcc.gnu.org/g:e39b170695a161feba7401b7d21d824db9ee1f8f
commit r13-3296-ge39b170695a161feba7401b7d21d824db9ee1f8f
Author: Eric Botcazou <ebotcazou@adacore.com>
Date: Fri Oct 14 11:52:04 2022 +0200
Fix PR target/107248
This is the infamous PR rtl-optimization/38644 rearing its ugly head for
leaf functions on SPARC more than a decade later... Richard E.'s generic
solution has never been implemented so let's do as other RISC back-ends
did.
gcc/
PR target/107248
* config/sparc/sparc.cc (sparc_expand_prologue): Emit a frame
blockage for leaf functions.
(sparc_flat_expand_prologue): Emit frame instead of full blockage.
(sparc_expand_epilogue): Emit a frame blockage for leaf functions.
(sparc_flat_expand_epilogue): Emit frame instead of full blockage.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (39 preceding siblings ...)
2022-10-14 9:56 ` cvs-commit at gcc dot gnu.org
@ 2022-10-14 9:57 ` cvs-commit at gcc dot gnu.org
2022-10-14 9:58 ` cvs-commit at gcc dot gnu.org
2022-10-14 10:00 ` cvs-commit at gcc dot gnu.org
42 siblings, 0 replies; 43+ messages in thread
From: cvs-commit at gcc dot gnu.org @ 2022-10-14 9:57 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #75 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-12 branch has been updated by Eric Botcazou
<ebotcazou@gcc.gnu.org>:
https://gcc.gnu.org/g:a5a6598d5b1d29741993371310c0bb8ca57e190c
commit r12-8831-ga5a6598d5b1d29741993371310c0bb8ca57e190c
Author: Eric Botcazou <ebotcazou@adacore.com>
Date: Fri Oct 14 11:52:04 2022 +0200
Fix PR target/107248
This is the infamous PR rtl-optimization/38644 rearing its ugly head for
leaf functions on SPARC more than a decade later... Richard E.'s generic
solution has never been implemented so let's do as other RISC back-ends
did.
gcc/
PR target/107248
* config/sparc/sparc.cc (sparc_expand_prologue): Emit a frame
blockage for leaf functions.
(sparc_flat_expand_prologue): Emit frame instead of full blockage.
(sparc_expand_epilogue): Emit a frame blockage for leaf functions.
(sparc_flat_expand_epilogue): Emit frame instead of full blockage.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (40 preceding siblings ...)
2022-10-14 9:57 ` cvs-commit at gcc dot gnu.org
@ 2022-10-14 9:58 ` cvs-commit at gcc dot gnu.org
2022-10-14 10:00 ` cvs-commit at gcc dot gnu.org
42 siblings, 0 replies; 43+ messages in thread
From: cvs-commit at gcc dot gnu.org @ 2022-10-14 9:58 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #76 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-11 branch has been updated by Eric Botcazou
<ebotcazou@gcc.gnu.org>:
https://gcc.gnu.org/g:3f4b65df625edae3e8829718af721ad2330b3f22
commit r11-10311-g3f4b65df625edae3e8829718af721ad2330b3f22
Author: Eric Botcazou <ebotcazou@adacore.com>
Date: Fri Oct 14 11:52:04 2022 +0200
Fix PR target/107248
This is the infamous PR rtl-optimization/38644 rearing its ugly head for
leaf functions on SPARC more than a decade later... Richard E.'s generic
solution has never been implemented so let's do as other RISC back-ends
did.
gcc/
PR target/107248
* config/sparc/sparc.c (sparc_expand_prologue): Emit a frame
blockage for leaf functions.
(sparc_flat_expand_prologue): Emit frame instead of full blockage.
(sparc_expand_epilogue): Emit a frame blockage for leaf functions.
(sparc_flat_expand_epilogue): Emit frame instead of full blockage.
^ permalink raw reply [flat|nested] 43+ messages in thread* [Bug rtl-optimization/38644] [4.6 Regression] Optimization flag -O1 -fschedule-insns2 causes wrong code
[not found] <bug-38644-4@http.gcc.gnu.org/bugzilla/>
` (41 preceding siblings ...)
2022-10-14 9:58 ` cvs-commit at gcc dot gnu.org
@ 2022-10-14 10:00 ` cvs-commit at gcc dot gnu.org
42 siblings, 0 replies; 43+ messages in thread
From: cvs-commit at gcc dot gnu.org @ 2022-10-14 10:00 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=38644
--- Comment #77 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-10 branch has been updated by Eric Botcazou
<ebotcazou@gcc.gnu.org>:
https://gcc.gnu.org/g:d0ef37c35b7ff7324b4567652380f32079d46088
commit r10-11034-gd0ef37c35b7ff7324b4567652380f32079d46088
Author: Eric Botcazou <ebotcazou@adacore.com>
Date: Fri Oct 14 11:52:04 2022 +0200
Fix PR target/107248
This is the infamous PR rtl-optimization/38644 rearing its ugly head for
leaf functions on SPARC more than a decade later... Richard E.'s generic
solution has never been implemented so let's do as other RISC back-ends
did.
gcc/
PR target/107248
* config/sparc/sparc.c (sparc_expand_prologue): Emit a frame
blockage for leaf functions.
(sparc_flat_expand_prologue): Emit frame instead of full blockage.
(sparc_expand_epilogue): Emit a frame blockage for leaf functions.
(sparc_flat_expand_epilogue): Emit frame instead of full blockage.
^ permalink raw reply [flat|nested] 43+ messages in thread