[Bug rtl-optimization/33642] unrecognizable insn for -frtl-abstract-sequences
--- Comment #18 from loki at gcc dot gnu dot org 2008-04-03 08:24 --- *** Bug 35785 has been marked as a duplicate of this bug. *** -- loki at gcc dot gnu dot org changed: What|Removed |Added CC||hjl dot tools at gmail dot ||com http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33642
[Bug rtl-optimization/35785] gcc.c-torture/compile/pr11832.c doesn't work for Linux/ia64
--- Comment #4 from loki at gcc dot gnu dot org 2008-04-03 08:24 --- The patch from http://gcc.gnu.org/ml/gcc-patches/2008-03/msg00824.html should fix this kind of problem. *** This bug has been marked as a duplicate of 33642 *** -- loki at gcc dot gnu dot org changed: What|Removed |Added Status|UNCONFIRMED |RESOLVED Resolution||DUPLICATE http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35785
[Bug rtl-optimization/33642] unrecognizable insn for -frtl-abstract-sequences
--- Comment #13 from loki at gcc dot gnu dot org 2008-03-14 08:30 --- (In reply to comment #12) > Is this the same bug? sparc-rtems4.9 on SVN trunk: Not exactly, but after I fixed a small typo in the latest patch an 'unrecognizable insn' showed up. So, this bug is still present on sparc-rtems4.9. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33642
[Bug rtl-optimization/33642] unrecognizable insn for -frtl-abstract-sequences
--- Comment #10 from loki at gcc dot gnu dot org 2008-03-11 09:45 --- Created an attachment (id=15295) --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=15295&action=view) Partially fix 'unrecognizable insn' and 'constraint' problem This patch should work on arm-eabi, i386-linux, x86_64-linux, powerpc-apple-darwin9, mips-elf, sh-elf. It fails on arm-eabi with -mthumb and/or -fPIC. Currently "-fPIC" is not well tested. -- loki at gcc dot gnu dot org changed: What|Removed |Added Attachment #15272|0 |1 is obsolete|| http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33642
[Bug rtl-optimization/33009] [4.4 Regression] -frtl-abstract-sequences causes an ICE
--- Comment #15 from loki at gcc dot gnu dot org 2008-03-07 12:33 --- (In reply to comment #14) > It is now broken on trunk on x86_64-linux with -m32 -fPIC: > > pr11832.c: In function 'foo': > pr11832.c:30: error: unrecognizable insn: > (insn 216pr11832.c:30: internal compiler error: Segmentation fault Truly, I do not understand why this bug had to be renamed and reopened! As I wrote in comment #9 the 'unrecognizable insn' case should belong to PR33642. This bug was about an infinite loop. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33009
[Bug rtl-optimization/33642] unrecognizable insn for -frtl-abstract-sequences
--- Comment #4 from loki at gcc dot gnu dot org 2008-03-06 15:01 --- Created an attachment (id=15272) --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=15272&action=view) Use own COMPUTE_RTX_LENGTH method The previous patch wasn't a good solution for some targets. This patch fixes the problem on i486-linux-gnu, arm-eabi, powerpc-elf. I am still hunting for similar bugs. -- loki at gcc dot gnu dot org changed: What|Removed |Added Attachment #15262|0 |1 is obsolete|| http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33642
[Bug rtl-optimization/33009] -frtl-abstract-sequences causes an infinite loop
--- Comment #9 from loki at gcc dot gnu dot org 2008-03-05 14:36 --- (In reply to comment #8) > With revision 132897 on i686-apple-darwin9 I get: > ... > unrecognizable insn: > (insn 118/opt/gcc/_gcc_clean/gcc/testsuite/gcc.c-torture/compile/pr33009.c:36: > internal compiler error: Segmentation fault It seems to me it is the same problem as PR33642. Would you be so kind to check it again on i686-apple-darwin9 with the attached patch from PR33642? -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33009
[Bug rtl-optimization/33642] unrecognizable insn for -frtl-abstract-sequences
--- Comment #3 from loki at gcc dot gnu dot org 2008-03-05 11:09 --- Created an attachment (id=15262) --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=15262&action=view) Avoid GET_ATTR_LENGTH when RECOG_MEMOIZED fails It seems this patch fixes the described problem on arm-eabi. I am also going to test it on other targets. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33642
[Bug rtl-optimization/33009] -frtl-abstract-sequences causes an infinite loop
--- Comment #7 from loki at gcc dot gnu dot org 2008-03-05 10:27 --- Fixed. -- loki at gcc dot gnu dot org changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33009
[Bug rtl-optimization/33009] -frtl-abstract-sequences causes an infinite loop
--- Comment #6 from loki at gcc dot gnu dot org 2008-03-05 10:16 --- Subject: Bug 33009 Author: loki Date: Wed Mar 5 10:15:45 2008 New Revision: 132893 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=132893 Log: 2008-03-05 Gabor Loki <[EMAIL PROTECTED]> PR gcc/33009 * rtl-factoring.c (clear_regs_live_in_seq): Fix backward steps. (split_block_and_df_analyze): New. Split basic block and rebuild dataflow. (block_label_after): Use SPLIT_BLOCK_AND_DF_ANALYZE instead of SPLIT_BLOCK. (split_pattern_seq): Likewise. (erase_matching_seqs): Likewise. (split_pattern_seq): Skip return insn in case of REG_NORETURN note. PR testsuite/33009 * gcc.c-torture/compile/pr11832.c: Check -frtl-abstract-sequences. * gcc.c-torture/compile/pr33009.c: Likewise. Added: trunk/gcc/testsuite/gcc.c-torture/compile/pr11832.c trunk/gcc/testsuite/gcc.c-torture/compile/pr33009.c Modified: trunk/gcc/ChangeLog trunk/gcc/rtl-factoring.c trunk/gcc/testsuite/ChangeLog -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33009
[Bug rtl-optimization/33642] unrecognizable insn for -frtl-abstract-sequences
--- Comment #2 from loki at gcc dot gnu dot org 2008-03-03 13:51 --- I can confirm the unrecognizable insn too (r132833). I am going to look into it. -- loki at gcc dot gnu dot org changed: What|Removed |Added AssignedTo|unassigned at gcc dot gnu |loki at gcc dot gnu dot org |dot org | Status|UNCONFIRMED |ASSIGNED Ever Confirmed|0 |1 Last reconfirmed|-00-00 00:00:00 |2008-03-03 13:51:21 date|| http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33642
[Bug rtl-optimization/33009] -frtl-abstract-sequences causes an infinite loop
--- Comment #5 from loki at gcc dot gnu dot org 2008-02-29 13:06 --- Created an attachment (id=15245) --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=15245&action=view) infinite loop patch v1 This proposed patch fixes the reported bug on i386. I am going to test it for some other targets, and create some testcases. After that I will send it to the gcc-patches list. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33009
[Bug rtl-optimization/33009] -frtl-abstract-sequences causes an infinite loop
--- Comment #3 from loki at gcc dot gnu dot org 2008-02-21 15:48 --- I can confirm the infinite loop too (r132521). I am going to look into it. -- loki at gcc dot gnu dot org changed: What|Removed |Added CC||loki at gcc dot gnu dot org AssignedTo|unassigned at gcc dot gnu |loki at gcc dot gnu dot org |dot org | Status|UNCONFIRMED |ASSIGNED Ever Confirmed|0 |1 Last reconfirmed|-00-00 00:00:00 |2008-02-21 15:48:53 date|| http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33009
[Bug bootstrap/25670] New: [4.2 Regression] build fail with 'make all-gcc'
I've configured with: ./configure --prefix=.../install --disable-nls --disable-libgcj --enable-languages=c Executed: make all-gcc Output: ... make[2]: Leaving directory `.../intl' cat: stage_last: No such file or directory make: invalid option -- a Usage: make [options] [target] ... ... It appeared after: Revision 108512 I think the problem is the missing check for 'stage_last' file in the Makefile. The following modification solved the problem, but I didn't test it very thoroughly. Makefile.tpl: -unstage = [ -f stage_current ] || $(MAKE) `cat stage_last`-start +unstage = [ -f stage_current ] || if [ -f stage_last ]; then $(MAKE) `cat stage_last`-start; fi -- Summary: [4.2 Regression] build fail with 'make all-gcc' Product: gcc Version: 4.2.0 Status: UNCONFIRMED Keywords: build Severity: major Priority: P3 Component: bootstrap AssignedTo: unassigned at gcc dot gnu dot org ReportedBy: loki at gcc dot gnu dot org GCC host triplet: i686-pc-linux-gnu http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25670
[Bug target/23436] Insn does not satisfy its constraints (reload_cse_simplify_operands)
--- Additional Comments From loki at gcc dot gnu dot org 2005-08-17 08:27 --- Created an attachment (id=9513) --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=9513&action=view) Test case -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23436
[Bug target/23436] New: Insn does not satisfy its constraints (reload_cse_simplify_operands)
I have the following internal compiler error at -O3. Command line: arm-elf-gcc -c -O3 -mthumb pr.c Output: pr.c: In function 'nrrdAxisInfoGet': pr.c:1934: error: insn does not satisfy its constraints: (insn:HI 59 275 240 7 (set (mem:SI (plus:SI (reg:SI 2 r2) (reg/f:SI 13 sp)) [2 S4 A32]) (reg:SI 3 r3 [orig:130 .size ] [130])) 143 {*thumb_movsi_insn} (insn_list:REG_DEP_TRUE 58 (nil)) (nil)) pr.c:1934: internal compiler error: in reload_cse_simplify_operands, at postreload.c:393 Last working version: 2005-08-04 12:00:00 UTC -- Summary: Insn does not satisfy its constraints (reload_cse_simplify_operands) Product: gcc Version: 4.1.0 Status: UNCONFIRMED Severity: normal Priority: P2 Component: target AssignedTo: unassigned at gcc dot gnu dot org ReportedBy: loki at gcc dot gnu dot org CC: gcc-bugs at gcc dot gnu dot org GCC target triplet: thumb-elf http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23436
[Bug target/23435] Unrecognizable insn (in extract_insn, at recog.c)
--- Additional Comments From loki at gcc dot gnu dot org 2005-08-17 08:12 --- Created an attachment (id=9512) --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=9512&action=view) Test case -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23435
[Bug target/23435] New: Unrecognizable insn (in extract_insn, at recog.c)
I have the following error when compiling a file from linux kernel (attached). Command line: m68k-elf-gcc -w -O1 -c pr.c Output: pr.c: In function 'nfs_statfs': pr.c:8757: error: unrecognizable insn: (insn 42 40 43 3 (set (mem/s/j:DI (plus:SI (reg/f:SI 26 virtual-stack-vars) (const_int -64 [0xffc0])) [0 res.bsize+0 S8 A16]) (zero_extend:DI (mem/s/j:SI (plus:SI (reg/v/f:SI 45 [ sb ]) (const_int 10 [0xa])) [0 .s_blocksize+0 S4 A16]))) -1 (nil) (nil)) pr.c:8757: internal compiler error: in extract_insn, at recog.c:2084 Last working version: "2005-07-29 12:00:00 UTC" -- Summary: Unrecognizable insn (in extract_insn, at recog.c) Product: gcc Version: 4.1.0 Status: UNCONFIRMED Severity: normal Priority: P2 Component: target AssignedTo: unassigned at gcc dot gnu dot org ReportedBy: loki at gcc dot gnu dot org CC: gcc-bugs at gcc dot gnu dot org GCC target triplet: m68k-elf http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23435
[Bug tree-optimization/22564] [4.1 Regression] Compilation time increased about 11%
--- Additional Comments From loki at gcc dot gnu dot org 2005-07-20 12:31 --- (In reply to comment #1) You are probably right, but it is hard to find such a test case, because there is only one significant daily regression (I guess from merging), but this one doesn't explain the overall change. IMHO the graphs suggest that the daily bugfixes increased the compilation time day after day. So, it's maybe difficult to clearly show which algorithms are responsible for this regression. Soever, I will search for some tests which will demonstarte this compile-time-hog. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22564
[Bug tree-optimization/22564] New: [4.1 Regression] Compilation time increased about 11%
The compilation time graphs for all target in CSiBE have been drastically increased from 2005-03-01. For example: Os / O2 / O3 i686-linux: 11% 11% 17% arm-elf: 10% 11% 18% m68k-elf:12% 12% 18% (See CSiBE for more detailed results) -- Summary: [4.1 Regression] Compilation time increased about 11% Product: gcc Version: 4.1.0 Status: UNCONFIRMED Keywords: compile-time-hog Severity: normal Priority: P2 Component: tree-optimization AssignedTo: unassigned at gcc dot gnu dot org ReportedBy: loki at gcc dot gnu dot org CC: gcc-bugs at gcc dot gnu dot org http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22564
[Bug c/17913] [4.0/4.1 Regression] ICE jumping into statement expression
--- Additional Comments From loki at gcc dot gnu dot org 2005-05-10 13:50 --- CVSROOT:/cvs/gcc Module name:gcc Changes by: [EMAIL PROTECTED] 2005-05-10 13:47:05 Modified files: gcc: ChangeLog c-typeck.c fold-const.c gcc/testsuite : ChangeLog Added files: gcc/testsuite/gcc.c-torture/compile: pr17913.c Log message: 2005-05-10 Gabor Loki <[EMAIL PROTECTED]> PR c/17913 * c-typeck.c (build_conditional_expr): Remove reducing cond_expr. * fold-const.c (fold): Expand the condition of reducing cond_expr. (contains_label_1, contains_label_p): New functions for checking labels in a sub-tree. testsuite: 2005-05-10 Gabor Loki <[EMAIL PROTECTED]> PR c/17913 * gcc.c-torture/compile/pr17913.c: Computed jump test for PR17913 Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.8704&r2=2.8705 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/c-typeck.c.diff?cvsroot=gcc&r1=1.440&r2=1.441 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/fold-const.c.diff?cvsroot=gcc&r1=1.574&r2=1.575 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&r1=1.5455&r2=1.5456 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.c-torture/compile/pr17913.c.diff?cvsroot=gcc&r1=NONE&r2=1.1 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17913
[Bug middle-end/21356] New: Dominance error after aggressive dead code elimination (cd_dce)
I have internal compiler error for the following c code: int a; void* p; void foo (void) { switch (a) { a0: case 0: p = &&a1; a1: case 1: p = &&a2; a2: default: p = &&a1; } goto *p; } Command line: gcc -c -O2 1.c Output: 1.c: In function 'foo': 1.c:5: error: dominator of 1 should be 2, not 0 1.c:5: internal compiler error: in verify_dominators, at dominance.c:875 ... -- Summary: Dominance error after aggressive dead code elimination (cd_dce) Product: gcc Version: 4.1.0 Status: UNCONFIRMED Severity: normal Priority: P2 Component: middle-end AssignedTo: unassigned at gcc dot gnu dot org ReportedBy: loki at gcc dot gnu dot org CC: gcc-bugs at gcc dot gnu dot org GCC target triplet: i686-pc-linux-gnu-gcc http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21356
[Bug middle-end/21282] New: Converting floor into lfloor built-in function
I have segfault for the following c code: extern double floor (double); long foo (float f) { return (long) floor (f); } Command line: arm-elf-gcc -c -O1 1.c Output message: 1.c: In function 'foo': 1.c:5: internal compiler error: in expand_builtin_int_roundingfn, at builtins.c:2225 ... The environment was configured with: binutils-2.14/configure --target=arm-elf gcc/configure --target=arm-elf --enable-language=c newlib-1.12.0/configure --target/arm-elf I have found the same problem for {i386, m68k, mips, ppc}-elf. The bug appeared after the following patch: http://gcc.gnu.org/ml/gcc-cvs/2005-04/msg00503.html -- Summary: Converting floor into lfloor built-in function Product: gcc Version: 4.1.0 Status: UNCONFIRMED Severity: normal Priority: P2 Component: middle-end AssignedTo: unassigned at gcc dot gnu dot org ReportedBy: loki at gcc dot gnu dot org CC: gcc-bugs at gcc dot gnu dot org,uros at gcc dot gnu dot org GCC target triplet: arm-elf i386-elf m68k-elf mips-elf ppc-elf http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21282