public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug target/40615] New: unnecessary CSE
@ 2009-07-02 7:39 carrot at google dot com
2009-07-02 7:39 ` [Bug target/40615] " carrot at google dot com
` (9 more replies)
0 siblings, 10 replies; 11+ messages in thread
From: carrot at google dot com @ 2009-07-02 7:39 UTC (permalink / raw)
To: gcc-bugs
Compile the attached source code with options -march=armv5te -mthumb -Os
-fno-exceptions, gcc generates:
push {r4, lr}
sub sp, sp, #8
add r4, sp, #4 // redundant
mov r0, r4 // add r0, sp, 4
bl _ZN1XC1Ev
mov r0, r4 // add r0, sp, 4
bl _Z3barP1X
mov r0, r4 // add r0, sp, 4
bl _ZN1XD1Ev
add sp, sp, #8
@ sp needed for prologue
pop {r4, pc}
As mentioned in the comments, the cse is redundant. We can recompute the value
of (sp + 4) each time we want it. With this method we can save one instruction.
--
Summary: unnecessary CSE
Product: gcc
Version: 4.5.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: target
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: carrot at google dot com
GCC build triplet: i686-linux
GCC host triplet: i686-linux
GCC target triplet: arm-eabi
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40615
^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug target/40615] unnecessary CSE
2009-07-02 7:39 [Bug target/40615] New: unnecessary CSE carrot at google dot com
@ 2009-07-02 7:39 ` carrot at google dot com
2009-07-02 8:53 ` ramana at gcc dot gnu dot org
` (8 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: carrot at google dot com @ 2009-07-02 7:39 UTC (permalink / raw)
To: gcc-bugs
------- Comment #1 from carrot at google dot com 2009-07-02 07:39 -------
Created an attachment (id=18120)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=18120&action=view)
test case
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40615
^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug target/40615] unnecessary CSE
2009-07-02 7:39 [Bug target/40615] New: unnecessary CSE carrot at google dot com
2009-07-02 7:39 ` [Bug target/40615] " carrot at google dot com
@ 2009-07-02 8:53 ` ramana at gcc dot gnu dot org
2009-07-02 9:16 ` steven at gcc dot gnu dot org
` (7 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: ramana at gcc dot gnu dot org @ 2009-07-02 8:53 UTC (permalink / raw)
To: gcc-bugs
------- Comment #2 from ramana at gcc dot gnu dot org 2009-07-02 08:53 -------
This looks like one of those rematerialization problems albeit with the stack
pointer this time.
--
ramana at gcc dot gnu dot org changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Ever Confirmed|0 |1
Last reconfirmed|0000-00-00 00:00:00 |2009-07-02 08:53:25
date| |
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40615
^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug target/40615] unnecessary CSE
2009-07-02 7:39 [Bug target/40615] New: unnecessary CSE carrot at google dot com
2009-07-02 7:39 ` [Bug target/40615] " carrot at google dot com
2009-07-02 8:53 ` ramana at gcc dot gnu dot org
@ 2009-07-02 9:16 ` steven at gcc dot gnu dot org
2009-07-02 9:39 ` ramana at gcc dot gnu dot org
` (6 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: steven at gcc dot gnu dot org @ 2009-07-02 9:16 UTC (permalink / raw)
To: gcc-bugs
------- Comment #3 from steven at gcc dot gnu dot org 2009-07-02 09:15 -------
Is there a C test case? Can you add objdump of the gcc-generated asm and the
fixed asm to show the impact on code size? (/me is surprised that 3*"add
r0,sp,4" is smaller than 1**"add r0,sp,4"+3*"mov r0,r4"... Thumb is amazing :-)
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40615
^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug target/40615] unnecessary CSE
2009-07-02 7:39 [Bug target/40615] New: unnecessary CSE carrot at google dot com
` (2 preceding siblings ...)
2009-07-02 9:16 ` steven at gcc dot gnu dot org
@ 2009-07-02 9:39 ` ramana at gcc dot gnu dot org
2009-09-29 5:50 ` [Bug middle-end/40615] " pinskia at gcc dot gnu dot org
` (5 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: ramana at gcc dot gnu dot org @ 2009-07-02 9:39 UTC (permalink / raw)
To: gcc-bugs
------- Comment #4 from ramana at gcc dot gnu dot org 2009-07-02 09:39 -------
(In reply to comment #3)
> Is there a C test case? Can you add objdump of the gcc-generated asm and the
> fixed asm to show the impact on code size? (/me is surprised that 3*"add
> r0,sp,4" is smaller than 1**"add r0,sp,4"+3*"mov r0,r4"... Thumb is amazing :-)
The length of add r0,sp,4 and mov r0,r4 is the same for Thumb1 (16 bits).
I suppose the ideal code generated would be something like this modulo errors
with stack alignments in the prologue and the epilogue.
We also don't need r4 in that case :) . So we can save a load, a store as well
as 1 instruction over all. Smaller and faster by 1 instruction and reduced
register usage.
push {lr}
sub sp, sp, #12 (8 byte stack alignment )
add r0, sp, 4 // add r0, sp, 4
bl _ZN1XC1Ev
add r0, sp, #4 // add r0, sp, 4
bl _Z3barP1X
add r0, sp, #4 // add r0, sp, 4
bl _ZN1XD1Ev
add sp, sp, #12 (8 byte stack alignment )
@ sp needed for prologue
pop {pc}
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40615
^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug middle-end/40615] unnecessary CSE
2009-07-02 7:39 [Bug target/40615] New: unnecessary CSE carrot at google dot com
` (3 preceding siblings ...)
2009-07-02 9:39 ` ramana at gcc dot gnu dot org
@ 2009-09-29 5:50 ` pinskia at gcc dot gnu dot org
2009-09-29 5:51 ` [Bug rtl-optimization/40615] " pinskia at gcc dot gnu dot org
` (4 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: pinskia at gcc dot gnu dot org @ 2009-09-29 5:50 UTC (permalink / raw)
To: gcc-bugs
------- Comment #5 from pinskia at gcc dot gnu dot org 2009-09-29 05:50 -------
PowerPC has the same issue. Instructions on PPC are all the same size so 3
adds are better than one add plus 3 register moves.
Here is a C example:
int f(int *a);
int g(int *a);
int h(int *a);
void hh(void)
{
int t;
f(&t);
g(&t);
h(&t);
}
--- CUT ---
Most RISC will have the same issue as most will have instructions which are
fixed length.
I bet this has to do with hard registers.
--
pinskia at gcc dot gnu dot org changed:
What |Removed |Added
----------------------------------------------------------------------------
Component|target |middle-end
GCC build triplet|i686-linux |
GCC target triplet|arm-eabi |arm-eabi, powerpc*-*-*
Keywords| |missed-optimization
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40615
^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug rtl-optimization/40615] unnecessary CSE
2009-07-02 7:39 [Bug target/40615] New: unnecessary CSE carrot at google dot com
` (4 preceding siblings ...)
2009-09-29 5:50 ` [Bug middle-end/40615] " pinskia at gcc dot gnu dot org
@ 2009-09-29 5:51 ` pinskia at gcc dot gnu dot org
2010-03-20 19:30 ` ramana at gcc dot gnu dot org
` (3 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: pinskia at gcc dot gnu dot org @ 2009-09-29 5:51 UTC (permalink / raw)
To: gcc-bugs
------- Comment #6 from pinskia at gcc dot gnu dot org 2009-09-29 05:51 -------
Before CSE:
(insn 13 2 5 2 t.c:8 (set (reg:SI 119)
(plus:SI (reg/f:SI 113 sfp)
(const_int 8 [0x8]))) -1 (nil))
(insn 5 13 6 2 t.c:8 (set (reg:SI 3 3)
(reg:SI 119)) 332 {*movsi_internal1} (nil))
...
(insn 14 6 7 2 t.c:9 (set (reg:SI 120)
(plus:SI (reg/f:SI 113 sfp)
(const_int 8 [0x8]))) -1 (nil))
(insn 7 14 8 2 t.c:9 (set (reg:SI 3 3)
(reg:SI 120)) 332 {*movsi_internal1} (nil))
And then after CSE, we remove the adds but we should have moved the adds into
the move.
--
pinskia at gcc dot gnu dot org changed:
What |Removed |Added
----------------------------------------------------------------------------
Component|middle-end |rtl-optimization
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40615
^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug rtl-optimization/40615] unnecessary CSE
2009-07-02 7:39 [Bug target/40615] New: unnecessary CSE carrot at google dot com
` (5 preceding siblings ...)
2009-09-29 5:51 ` [Bug rtl-optimization/40615] " pinskia at gcc dot gnu dot org
@ 2010-03-20 19:30 ` ramana at gcc dot gnu dot org
2010-06-04 12:44 ` bernds at gcc dot gnu dot org
` (2 subsequent siblings)
9 siblings, 0 replies; 11+ messages in thread
From: ramana at gcc dot gnu dot org @ 2010-03-20 19:30 UTC (permalink / raw)
To: gcc-bugs
--
ramana at gcc dot gnu dot org changed:
What |Removed |Added
----------------------------------------------------------------------------
Severity|normal |enhancement
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40615
^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug rtl-optimization/40615] unnecessary CSE
2009-07-02 7:39 [Bug target/40615] New: unnecessary CSE carrot at google dot com
` (6 preceding siblings ...)
2010-03-20 19:30 ` ramana at gcc dot gnu dot org
@ 2010-06-04 12:44 ` bernds at gcc dot gnu dot org
2010-06-07 22:50 ` bernds at gcc dot gnu dot org
2010-06-08 15:31 ` hjl dot tools at gmail dot com
9 siblings, 0 replies; 11+ messages in thread
From: bernds at gcc dot gnu dot org @ 2010-06-04 12:44 UTC (permalink / raw)
To: gcc-bugs
------- Comment #7 from bernds at gcc dot gnu dot org 2010-06-04 12:44 -------
Subject: Bug 40615
Author: bernds
Date: Fri Jun 4 12:44:01 2010
New Revision: 160260
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=160260
Log:
PR rtl-optimization/39871
PR rtl-optimization/40615
PR rtl-optimization/42500
PR rtl-optimization/42502
* ira.c (init_reg_equiv_memory_loc: New function.
(ira): Call it twice.
* reload.h (calculate_elim_costs_all_insns): Declare.
* ira-costs.c: Include "reload.h".
(regno_equiv_gains): New static variable.
(init_costs): Allocate it.
(finish_costs): Free it.
(ira_costs): Call calculate_elim_costs_all_insns.
(find_costs_and_classes): Take estimated elimination costs
into account.
(ira_adjust_equiv_reg_cost): New function.
* ira.h (ira_adjust_equiv_reg_cost): Declare it.
* reload1.c (init_eliminable_invariants, free_reg_equiv,
elimination_costs_in_insn, note_reg_elim_costly): New static
functions.
(elim_bb): New static variable.
(reload): Move code out of here into init_eliminable_invariants and
free_reg_equiv. Call them.
(calculate_elim_costs_all_insns): New function.
(eliminate_regs_1): Declare. Add extra arg FOR_COSTS;
all callers changed. If FOR_COSTS is true, don't call alter_reg,
but call note_reg_elim_costly if we turned a valid memory address
into an invalid one.
* Makefile.in (ira-costs.o): Depend on reload.h.
testsuite/
PR rtl-optimization/39871
PR rtl-optimization/40615
PR rtl-optimization/42500
PR rtl-optimization/42502
* gcc.target/arm/eliminate.c: New test.
Added:
trunk/gcc/testsuite/gcc.target/arm/eliminate.c
Modified:
trunk/gcc/ChangeLog
trunk/gcc/Makefile.in
trunk/gcc/ira-costs.c
trunk/gcc/ira.c
trunk/gcc/ira.h
trunk/gcc/reload.h
trunk/gcc/reload1.c
trunk/gcc/testsuite/ChangeLog
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40615
^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug rtl-optimization/40615] unnecessary CSE
2009-07-02 7:39 [Bug target/40615] New: unnecessary CSE carrot at google dot com
` (7 preceding siblings ...)
2010-06-04 12:44 ` bernds at gcc dot gnu dot org
@ 2010-06-07 22:50 ` bernds at gcc dot gnu dot org
2010-06-08 15:31 ` hjl dot tools at gmail dot com
9 siblings, 0 replies; 11+ messages in thread
From: bernds at gcc dot gnu dot org @ 2010-06-07 22:50 UTC (permalink / raw)
To: gcc-bugs
------- Comment #8 from bernds at gcc dot gnu dot org 2010-06-07 22:49 -------
Fixed.
--
bernds at gcc dot gnu dot org changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |RESOLVED
Resolution| |FIXED
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40615
^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug rtl-optimization/40615] unnecessary CSE
2009-07-02 7:39 [Bug target/40615] New: unnecessary CSE carrot at google dot com
` (8 preceding siblings ...)
2010-06-07 22:50 ` bernds at gcc dot gnu dot org
@ 2010-06-08 15:31 ` hjl dot tools at gmail dot com
9 siblings, 0 replies; 11+ messages in thread
From: hjl dot tools at gmail dot com @ 2010-06-08 15:31 UTC (permalink / raw)
To: gcc-bugs
--
hjl dot tools at gmail dot com changed:
What |Removed |Added
----------------------------------------------------------------------------
Target Milestone|--- |4.6.0
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40615
^ permalink raw reply [flat|nested] 11+ messages in thread
end of thread, other threads:[~2010-06-08 15:31 UTC | newest]
Thread overview: 11+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2009-07-02 7:39 [Bug target/40615] New: unnecessary CSE carrot at google dot com
2009-07-02 7:39 ` [Bug target/40615] " carrot at google dot com
2009-07-02 8:53 ` ramana at gcc dot gnu dot org
2009-07-02 9:16 ` steven at gcc dot gnu dot org
2009-07-02 9:39 ` ramana at gcc dot gnu dot org
2009-09-29 5:50 ` [Bug middle-end/40615] " pinskia at gcc dot gnu dot org
2009-09-29 5:51 ` [Bug rtl-optimization/40615] " pinskia at gcc dot gnu dot org
2010-03-20 19:30 ` ramana at gcc dot gnu dot org
2010-06-04 12:44 ` bernds at gcc dot gnu dot org
2010-06-07 22:50 ` bernds at gcc dot gnu dot org
2010-06-08 15:31 ` hjl dot tools at gmail dot com
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 read-only IMAP folder(s) and NNTP newsgroup(s).