public inbox for gcc-bugs@sourceware.org help / color / mirror / Atom feed
* [Bug rtl-optimization/45051] [4.6 Regression]: gcc.c-torture/execute/builtins/abs-2.c and abs-3.c due to "track subwords of DImode allocnos" [not found] <bug-45051-4@http.gcc.gnu.org/bugzilla/> @ 2010-12-22 16:23 ` ibolton at gcc dot gnu.org 2010-12-23 1:56 ` hp at gcc dot gnu.org ` (3 subsequent siblings) 4 siblings, 0 replies; 11+ messages in thread From: ibolton at gcc dot gnu.org @ 2010-12-22 16:23 UTC (permalink / raw) To: gcc-bugs http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45051 Ian Bolton <ibolton at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED CC| |ibolton at gcc dot gnu.org Resolution|FIXED | --- Comment #7 from Ian Bolton <ibolton at gcc dot gnu.org> 2010-12-22 16:23:21 UTC --- (In reply to comment #5) > Assuming fixed and closing. Please reopen if you still have a problem. This patch has caused SpecCPU2000 Ammp to fail for ARM -O3 thumb. I assume the patch was meant to prevent deletions that shouldn't occur. This might be what happens for the original symptomatic test, but I am now seeing extra deletions that shouldn't happen for Ammp. For example, without this patch, you get these insns somewhere in the ira dump for mm_fv_update_nonbon() from rectmm.c: (insn 3163 3161 3164 107 rectmm.c:1041 (set (reg:SI 1 r1) (plus:SI (reg:SI 1 r1) (const_int 280 [0x118]))) 4 {*arm_addsi3} (nil)) (insn 3164 3163 1730 107 rectmm.c:1041 (set (reg:SI 3 r3) (reg:SI 1 r1)) 586 {*thumb2_movsi_vfp} (nil)) With the patch, you lose the add and just get this: (insn 3164 3161 1730 107 rectmm.c:1041 (set (reg:SI 3 r3) (reg:SI 1 r1)) 586 {*thumb2_movsi_vfp} (nil)) The incrementing of r1 is perfectly legitimate and useful and removing it is a bug. Other increments of r9, ip, r0 and r3 are also lost. I think the issue might be that reg_mentioned_p() considers output registers to have been "mentioned", whereas the refers_to_regno_p() does not consider an output register to have been "referred to". I can see the problem with only using reg_mentioned_p() because it doesn't handle subregs, but there is also a problem with only using refers_to_regno_p(), as we can see with this segfault in Ammp. I therefore wonder if the fix might be this: Index: gcc/reload1.c =================================================================== --- gcc/reload1.c (revision 168082) +++ gcc/reload1.c (working copy) @@ -8395,7 +8395,8 @@ delete_output_reload (rtx insn, int j, i if (NOTE_INSN_BASIC_BLOCK_P (i1)) return; if ((NONJUMP_INSN_P (i1) || CALL_P (i1)) - && refers_to_regno_p (regno, regno + nregs, PATTERN (i1), NULL)) + && (refers_to_regno_p (regno, regno + nregs, PATTERN (i1), NULL) + || reg_mentioned_p (reg, PATTERN (i1)))) { /* If this is USE in front of INSN, we only have to check that there are no more references than accounted for by inheritance. */ I am heading off for Christmas vacation shortly, so cannot look into this any further, but I wanted to record my findings so far publicly. Apologies if there is missing information. I return to work Jan 4th. ^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug rtl-optimization/45051] [4.6 Regression]: gcc.c-torture/execute/builtins/abs-2.c and abs-3.c due to "track subwords of DImode allocnos" [not found] <bug-45051-4@http.gcc.gnu.org/bugzilla/> 2010-12-22 16:23 ` [Bug rtl-optimization/45051] [4.6 Regression]: gcc.c-torture/execute/builtins/abs-2.c and abs-3.c due to "track subwords of DImode allocnos" ibolton at gcc dot gnu.org @ 2010-12-23 1:56 ` hp at gcc dot gnu.org 2010-12-23 14:29 ` hp at gcc dot gnu.org ` (2 subsequent siblings) 4 siblings, 0 replies; 11+ messages in thread From: hp at gcc dot gnu.org @ 2010-12-23 1:56 UTC (permalink / raw) To: gcc-bugs http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45051 --- Comment #8 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2010-12-23 01:56:08 UTC --- (In reply to comment #7) > (insn 3163 3161 3164 107 rectmm.c:1041 (set (reg:SI 1 r1) > (plus:SI (reg:SI 1 r1) > (const_int 280 [0x118]))) 4 {*arm_addsi3} (nil)) > I think the issue might be that reg_mentioned_p() considers output registers to > have been "mentioned", whereas the refers_to_regno_p() does not consider an > output register to have been "referred to". But r1 is an input as well as an output , i.e. "referred to", so insn 3163 should have matched without your patch. I'm missing analysis on why that didn't happen. Nevertheless, I'm testing r162418 with your patch. Also, you should open a separate PR for the regression you see, not just reuse this one. Use the depends-on features for the dependency. ^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug rtl-optimization/45051] [4.6 Regression]: gcc.c-torture/execute/builtins/abs-2.c and abs-3.c due to "track subwords of DImode allocnos" [not found] <bug-45051-4@http.gcc.gnu.org/bugzilla/> 2010-12-22 16:23 ` [Bug rtl-optimization/45051] [4.6 Regression]: gcc.c-torture/execute/builtins/abs-2.c and abs-3.c due to "track subwords of DImode allocnos" ibolton at gcc dot gnu.org 2010-12-23 1:56 ` hp at gcc dot gnu.org @ 2010-12-23 14:29 ` hp at gcc dot gnu.org 2011-01-03 19:58 ` rguenth at gcc dot gnu.org 2011-01-03 20:01 ` rguenth at gcc dot gnu.org 4 siblings, 0 replies; 11+ messages in thread From: hp at gcc dot gnu.org @ 2010-12-23 14:29 UTC (permalink / raw) To: gcc-bugs http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45051 --- Comment #9 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2010-12-23 14:29:13 UTC --- (In reply to comment #8) > Nevertheless, I'm testing r162418 with your patch. FWIW, no regressions for cris-elf, with your patch (manually) applied on top of Bernds's patch applied to r162418. ^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug rtl-optimization/45051] [4.6 Regression]: gcc.c-torture/execute/builtins/abs-2.c and abs-3.c due to "track subwords of DImode allocnos" [not found] <bug-45051-4@http.gcc.gnu.org/bugzilla/> ` (2 preceding siblings ...) 2010-12-23 14:29 ` hp at gcc dot gnu.org @ 2011-01-03 19:58 ` rguenth at gcc dot gnu.org 2011-01-03 20:01 ` rguenth at gcc dot gnu.org 4 siblings, 0 replies; 11+ messages in thread From: rguenth at gcc dot gnu.org @ 2011-01-03 19:58 UTC (permalink / raw) To: gcc-bugs http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45051 Richard Guenther <rguenth at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- Target Milestone|--- |4.6.0 ^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug rtl-optimization/45051] [4.6 Regression]: gcc.c-torture/execute/builtins/abs-2.c and abs-3.c due to "track subwords of DImode allocnos" [not found] <bug-45051-4@http.gcc.gnu.org/bugzilla/> ` (3 preceding siblings ...) 2011-01-03 19:58 ` rguenth at gcc dot gnu.org @ 2011-01-03 20:01 ` rguenth at gcc dot gnu.org 4 siblings, 0 replies; 11+ messages in thread From: rguenth at gcc dot gnu.org @ 2011-01-03 20:01 UTC (permalink / raw) To: gcc-bugs http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45051 Richard Guenther <rguenth at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED --- Comment #10 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-01-03 20:01:41 UTC --- Original problem was fixed. Please open a new bug for any other issues. ^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug rtl-optimization/45051] New: [4.6 Regression]: gcc.c-torture/execute/builtins/abs-2.c and abs-3.c due to "track subwords of DImode allocnos"
@ 2010-07-24 2:43 hp at gcc dot gnu dot org
2010-07-24 2:46 ` [Bug rtl-optimization/45051] " hp at gcc dot gnu dot org
` (5 more replies)
0 siblings, 6 replies; 11+ messages in thread
From: hp at gcc dot gnu dot org @ 2010-07-24 2:43 UTC (permalink / raw)
To: gcc-bugs
With revision 162417 this test passed.
>From revision 162418 and on, these tests (or "this test" as abs-3.c is
effectively the same as abs-2.c) has failed as follows:
Running
/tmp/badabs/gcc/gcc/testsuite/gcc.c-torture/execute/builtins/builtins.exp ...
FAIL: gcc.c-torture/execute/builtins/abs-2.c execution, -O2
FAIL: gcc.c-torture/execute/builtins/abs-2.c execution, -O3
-fomit-frame-pointer
FAIL: gcc.c-torture/execute/builtins/abs-2.c execution, -O3 -g
FAIL: gcc.c-torture/execute/builtins/abs-2.c execution, -Os
FAIL: gcc.c-torture/execute/builtins/abs-2.c execution, -O2 -flto
FAIL: gcc.c-torture/execute/builtins/abs-2.c execution, -O2 -fwhopr
FAIL: gcc.c-torture/execute/builtins/abs-3.c execution, -O2
FAIL: gcc.c-torture/execute/builtins/abs-3.c execution, -O3
-fomit-frame-pointer
FAIL: gcc.c-torture/execute/builtins/abs-3.c execution, -O3 -g
FAIL: gcc.c-torture/execute/builtins/abs-3.c execution, -Os
FAIL: gcc.c-torture/execute/builtins/abs-3.c execution, -O2 -flto
FAIL: gcc.c-torture/execute/builtins/abs-3.c execution, -O2 -fwhopr
With the message in the logfile being:
...
PASS: gcc.c-torture/execute/builtins/abs-2.c compilation, -O2
program stopped with signal 6.
FAIL: gcc.c-torture/execute/builtins/abs-2.c execution, -O2
...
(i.e. abort being called)
Author of patch in suspect revision range CC:ed.
At a glance, the code change causes a register containing 0 to be used as
containing 0xffffffff for the low-part of 9223372036854775807LL ((1<<63)-1),
i.e. a mis-tracking of a DImode subreg.
I'll attach a smaller test-case and an assembly diff pointing out the wrong
code.
--
Summary: [4.6 Regression]: gcc.c-torture/execute/builtins/abs-2.c
and abs-3.c due to "track subwords of DImode allocnos"
Product: gcc
Version: 4.6.0
Status: UNCONFIRMED
Keywords: wrong-code
Severity: normal
Priority: P3
Component: rtl-optimization
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: hp at gcc dot gnu dot org
GCC host triplet: x86_64-unknown-linux-gnu
GCC target triplet: cris-axis-elf
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45051
^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug rtl-optimization/45051] [4.6 Regression]: gcc.c-torture/execute/builtins/abs-2.c and abs-3.c due to "track subwords of DImode allocnos" 2010-07-24 2:43 [Bug rtl-optimization/45051] New: " hp at gcc dot gnu dot org @ 2010-07-24 2:46 ` hp at gcc dot gnu dot org 2010-07-24 2:51 ` hp at gcc dot gnu dot org ` (4 subsequent siblings) 5 siblings, 0 replies; 11+ messages in thread From: hp at gcc dot gnu dot org @ 2010-07-24 2:46 UTC (permalink / raw) To: gcc-bugs ------- Comment #1 from hp at gcc dot gnu dot org 2010-07-24 02:46 ------- Created an attachment (id=21295) --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=21295&action=view) Shortened gcc.c-torture/execute/builtins/abs-2.c Compile with e.g. -O2 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45051 ^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug rtl-optimization/45051] [4.6 Regression]: gcc.c-torture/execute/builtins/abs-2.c and abs-3.c due to "track subwords of DImode allocnos" 2010-07-24 2:43 [Bug rtl-optimization/45051] New: " hp at gcc dot gnu dot org 2010-07-24 2:46 ` [Bug rtl-optimization/45051] " hp at gcc dot gnu dot org @ 2010-07-24 2:51 ` hp at gcc dot gnu dot org 2010-07-24 3:03 ` hp at gcc dot gnu dot org ` (3 subsequent siblings) 5 siblings, 0 replies; 11+ messages in thread From: hp at gcc dot gnu dot org @ 2010-07-24 2:51 UTC (permalink / raw) To: gcc-bugs ------- Comment #2 from hp at gcc dot gnu dot org 2010-07-24 02:51 ------- (In reply to comment #0) Oops; replace: > containing 0xffffffff for the low-part of 9223372036854775807LL ((1<<63)-1), with: > containing 0x7fffffff for the high-part of 9223372036854775807LL ((1<<63)-1), -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45051 ^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug rtl-optimization/45051] [4.6 Regression]: gcc.c-torture/execute/builtins/abs-2.c and abs-3.c due to "track subwords of DImode allocnos" 2010-07-24 2:43 [Bug rtl-optimization/45051] New: " hp at gcc dot gnu dot org 2010-07-24 2:46 ` [Bug rtl-optimization/45051] " hp at gcc dot gnu dot org 2010-07-24 2:51 ` hp at gcc dot gnu dot org @ 2010-07-24 3:03 ` hp at gcc dot gnu dot org 2010-07-27 9:35 ` bernds at gcc dot gnu dot org ` (2 subsequent siblings) 5 siblings, 0 replies; 11+ messages in thread From: hp at gcc dot gnu dot org @ 2010-07-24 3:03 UTC (permalink / raw) To: gcc-bugs ------- Comment #3 from hp at gcc dot gnu dot org 2010-07-24 03:02 ------- --- good:r162417/abs-2.s Sat Jul 24 03:13:36 2010 +++ bad:r162418/abs-2.s Sat Jul 24 03:15:53 2010 @@ -9,27 +9,27 @@ _main_test: move $srp,[$sp] subu.b 84,$sp movem $r8,[$sp] - clear.d $r4 - clear.d $r5 - move.b 1,$r4 - move.d $r4,[$sp+76] - move.d $r5,[$sp+80] + clear.d $r0 ; (set (reg:SI r0) (const_int 0)) + clear.d $r1 + move.b 1,$r0 ; (set (strict_low_part:QI (reg:SI r0)) (const_int 1)) + move.d $r0,[$sp+76] + move.d $r1,[$sp+80] moveq -1,$r12 move.d $r12,[$sp+68] move.d $r12,[$sp+72] - clear.d $r10 - move.d $r4,$r11 - move.d $r11,$r8 - move.d -2147483648,$r0 - move.d $r11,[$sp+60] - move.d $r10,[$sp+64] - move.d 2147483647,$r11 + move.d $r0,$r10 + move.d -2147483648,$r11 + move.d $r10,[$sp+60] + move.d $r11,[$sp+64] + clear.b $r0 ; (set (strict_low_part:QI (reg:SI r9)) (const_int 0)) + move.d $r0,$r9 ; (set (reg:SI r9) (reg:SI r0)) + move.d 2147483647,$r0 move.d $r12,[$sp+52] - move.d $r11,[$sp+56] - move.d $r8,[$sp+44] - move.d $r0,[$sp+48] + move.d $r9,[$sp+56] ; stored here + move.d $r10,[$sp+44] + move.d $r11,[$sp+48] move.d $r12,[$sp+36] - move.d $r11,[$sp+40] + move.d $r0,[$sp+40] move.d [$sp+52],$r12 move.d [$sp+56],$r13 ; loaded here test.d $r12 ; fatally tested below; containing 0, expected 0x7fffffff -- hp 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 |2010-07-24 03:02:58 date| | http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45051 ^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug rtl-optimization/45051] [4.6 Regression]: gcc.c-torture/execute/builtins/abs-2.c and abs-3.c due to "track subwords of DImode allocnos" 2010-07-24 2:43 [Bug rtl-optimization/45051] New: " hp at gcc dot gnu dot org ` (2 preceding siblings ...) 2010-07-24 3:03 ` hp at gcc dot gnu dot org @ 2010-07-27 9:35 ` bernds at gcc dot gnu dot org 2010-07-27 21:49 ` bernds at gcc dot gnu dot org 2010-09-21 21:26 ` hp at gcc dot gnu dot org 5 siblings, 0 replies; 11+ messages in thread From: bernds at gcc dot gnu dot org @ 2010-07-27 9:35 UTC (permalink / raw) To: gcc-bugs ------- Comment #4 from bernds at gcc dot gnu dot org 2010-07-27 09:35 ------- Subject: Bug 45051 Author: bernds Date: Tue Jul 27 09:34:51 2010 New Revision: 162558 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=162558 Log: PR rtl-optimization/45051 * reload1.c (delete_output_reload): Use refers_to_regno_p rather than reg_mentioned_p. Modified: trunk/gcc/ChangeLog trunk/gcc/reload1.c -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45051 ^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug rtl-optimization/45051] [4.6 Regression]: gcc.c-torture/execute/builtins/abs-2.c and abs-3.c due to "track subwords of DImode allocnos" 2010-07-24 2:43 [Bug rtl-optimization/45051] New: " hp at gcc dot gnu dot org ` (3 preceding siblings ...) 2010-07-27 9:35 ` bernds at gcc dot gnu dot org @ 2010-07-27 21:49 ` bernds at gcc dot gnu dot org 2010-09-21 21:26 ` hp at gcc dot gnu dot org 5 siblings, 0 replies; 11+ messages in thread From: bernds at gcc dot gnu dot org @ 2010-07-27 21:49 UTC (permalink / raw) To: gcc-bugs ------- Comment #5 from bernds at gcc dot gnu dot org 2010-07-27 21:49 ------- Assuming fixed and closing. Please reopen if you still have a problem. -- 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=45051 ^ permalink raw reply [flat|nested] 11+ messages in thread
* [Bug rtl-optimization/45051] [4.6 Regression]: gcc.c-torture/execute/builtins/abs-2.c and abs-3.c due to "track subwords of DImode allocnos" 2010-07-24 2:43 [Bug rtl-optimization/45051] New: " hp at gcc dot gnu dot org ` (4 preceding siblings ...) 2010-07-27 21:49 ` bernds at gcc dot gnu dot org @ 2010-09-21 21:26 ` hp at gcc dot gnu dot org 5 siblings, 0 replies; 11+ messages in thread From: hp at gcc dot gnu dot org @ 2010-09-21 21:26 UTC (permalink / raw) To: gcc-bugs ------- Comment #6 from hp at gcc dot gnu dot org 2010-09-21 21:26 ------- Subject: Bug 45051 Author: hp Date: Tue Sep 21 21:25:57 2010 New Revision: 164498 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=164498 Log: PR rtl-optimization/41085 Backport from mainline 2010-07-27 Bernd Schmidt <bernds@codesourcery.com> PR rtl-optimization/45051 * reload1.c (delete_output_reload): Use refers_to_regno_p rather than reg_mentioned_p. Modified: branches/gcc-4_5-branch/gcc/ChangeLog branches/gcc-4_5-branch/gcc/reload1.c -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45051 ^ permalink raw reply [flat|nested] 11+ messages in thread
end of thread, other threads:[~2011-01-03 20:01 UTC | newest] Thread overview: 11+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- [not found] <bug-45051-4@http.gcc.gnu.org/bugzilla/> 2010-12-22 16:23 ` [Bug rtl-optimization/45051] [4.6 Regression]: gcc.c-torture/execute/builtins/abs-2.c and abs-3.c due to "track subwords of DImode allocnos" ibolton at gcc dot gnu.org 2010-12-23 1:56 ` hp at gcc dot gnu.org 2010-12-23 14:29 ` hp at gcc dot gnu.org 2011-01-03 19:58 ` rguenth at gcc dot gnu.org 2011-01-03 20:01 ` rguenth at gcc dot gnu.org 2010-07-24 2:43 [Bug rtl-optimization/45051] New: " hp at gcc dot gnu dot org 2010-07-24 2:46 ` [Bug rtl-optimization/45051] " hp at gcc dot gnu dot org 2010-07-24 2:51 ` hp at gcc dot gnu dot org 2010-07-24 3:03 ` hp at gcc dot gnu dot org 2010-07-27 9:35 ` bernds at gcc dot gnu dot org 2010-07-27 21:49 ` bernds at gcc dot gnu dot org 2010-09-21 21:26 ` hp at gcc dot gnu dot org
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).