Home > Internal Compiler > Internal Compiler Error In Reload_cse_simplify_operands

Internal Compiler Error In Reload_cse_simplify_operands

I boiled down a very small example of the problem: Let's say we have a file 'test_pc.c' as follow -------------------------------------------- void func_2(const unsigned long aPC) { } int func_1() { register See for instructions. Comment 3 Bernd Schmidt 2010-07-06 23:45:13 UTC Subject: Bug 44787 Author: bernds Date: Tue Jul 6 23:44:55 2010 New Revision: 161893 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=161893 Log: PR rtl-optimization/44787 * config/arm/arm.md (arith_shiftsi): Allow stack Modified: trunk/gcc/ChangeLog trunk/gcc/config/arm/arm.c trunk/gcc/config/arm/arm.md Comment 3 Ramana Radhakrishnan 2011-06-03 19:34:06 UTC Fixed on trunk . http://mttags.com/internal-compiler/internal-compiler-error-in-reload-cse-simplify-operands-at-postreload-c.php

When I compile with gcc-4.8 I get the following: -------------------------------------------- $ $ANDROID_NDK_HOME/toolchains/arm-linux-androideabi-4.8/prebuilt/linux-x86_64/bin/arm-linux-androideabi-gcc -O0 -g -Wall -o test_pc.o -c test_pc.c test_pc.c: In function 'func_1': test_pc.c:10:1: error: insn does not satisfy its constraints: Backport r172297 from mainline 2011-04-11 Chung-Lin Tang Richard Earnshaw PR target/48250 * config/arm/arm.c (arm_legitimize_reload_address): Update cases to use sign-magnitude offsets. Added: trunk/gcc/testsuite/gcc.c-torture/compile/pr44788.c trunk/gcc/testsuite/gcc.target/arm/pr44788.c Modified: trunk/gcc/ChangeLog trunk/gcc/config/arm/arm.md trunk/gcc/config/arm/thumb2.md trunk/gcc/testsuite/ChangeLog Comment 4 Bernd Schmidt 2010-07-06 23:46:24 UTC Fixed. The failing memory store started out as insn 1485, and gcc was supposed to reload the reg to fr0, which is a dumb thing to do but a correct choice none https://gcc.gnu.org/bugzilla/show_bug.cgi?id=44787

Status: FutureRelease Owner: [email protected] Dec 21, 2012 Project Member #4 [email protected] (No comment was entered for this change.) Status: Released ► Sign in to add a comment Terms - Privacy Should I file this as a bug? -------------------------------------------------------------- The output of `arm-none-eabi-g++ -v` is : Using built-in specs. See for instructions. Thank you.) More information and reason for this action is here: https://fedoraproject.org/wiki/BugZappers/HouseKeeping/Fedora19 Comment 15 Peter Bergner 2013-07-26 12:34:22 EDT I'm trying to clean up bugs on the IBM side and came

  1. Add detailed description in comments. * config/arm/arm.md (reload_outdf): Disable for ARM mode; change condition from TARGET_32BIT to TARGET_ARM.
  2. It seems that reading PC through _Unwind_GetIP is the portable way across different platforms and I won't be surprised that GCC deliberately added the ICE to dissuade reading PC directly.
  3. Description Jing Yu 2010-07-02 17:38:29 UTC Revision 160947 breaks the following test case.
  4. Still, something is very fishy with reg_equiv_init to have it point at the insn where the pseudo dies.
  5. Please Cc: me on replies!
  6. Bootstrap and regression test with that change plus the fix in (4) show no regressions, and cure the testcase.

to the correct ones to reproduce): /cygdrive/d/sdk/android-ndk/toolchains/arm-linux-androideabi-4.4.3/prebuilt/windows/bin/arm-linux-androideabi-g++ -fpic -ffunction-sections -funwind-tables -fstack-protector -D__ARM_ARCH_5__ -D__ARM_ARCH_5T__ -D__ARM_ARCH_5E__ -D__ARM_ARCH_5TE__ -Wno-psabi -march=armv7-a -mfloat-abi=softfp -mfpu=vfp -fno-exceptions -fno-rtti -O2 -fomit-frame-pointer -fstrict-aliasing -funswitch-loops -finline-limit=300 -mfpu=neon -DANDROID -gstabs+ -Wa,--noexecstack -frtti Obviously tightening the predicate would work, but as I said in comment #2 the insn is good. Output from gcc -save-temps that triggers the bug Download all attachments as: .zip Oldest first Newest first Threaded Comments only Change History (3) Changed 9 years ago by paulw2 Attachment target.i​ Back to list Status: Released Owner: [email protected] Closed: Nov 2012 Cc: [email protected], [email protected], [email protected] Type-Defect Priority-Medium ReportedBy-Developer Sign in to add a comment Reported by

Since the patch was regarding long double support on Darwin, I can understand poking at ti and tf patterns, but I wonder why Geoff did anything with a di pattern? It does fail with a ToT NDK, albeit with a slightly different message (it is now 4.9): [armeabi-v7a] Compile thumb : issue65705-asm-pc <= issue65705-asm-pc.c jni/issue65705-asm-pc.c: In function 'func_1': jni/issue65705-asm-pc.c:26:1: error: insn Backport r171251 from mainline 2011-03-21 Daniel Jacobowitz * config/arm/unwind-arm.c (__gnu_unwind_pr_common): Correct test for barrier handlers. Format For Printing -XML -Clone This Bug -Top of page First Last Prev Next This bug is not in your last search results.

It's encouraged to use GCC 4.6 available and default since r8b, though. Comment 13 IBM Bug Proxy 2012-11-09 12:32:01 EST Alan's patched have been committed for a while now. The output is " error: insn does not satisfy its constraints:" and "internal compiler error: in reload_cse_simplify_operands, at postreload.c:411". Or perhaps http://gcc.gnu.org/ml/gcc-patches/2012-04/msg01688.html and its followup?

On 4.7 is doesn't have a reg at all, ie. ppc-koji build --scratch f18 gcc-4.7.1-1.fc18.src.rpm 2. 3. Needs backporting to 4.6 branch. It can and should generate stw 6,32760(19); stw 7,32764(19). 2) The failing constraint is "o<>", specifically "o" so offsettable_memref_p is involved.

The error output is m68k-rtems4.8-gcc -m5200 -O target.i ../../../../iprcf/voiplib/net-snmp/src/agent/mibgroup/target/target.c: In functi on 'get_target_sessions': ../../../../iprcf/voiplib/net-snmp/src/agent/mibgroup/target/target.c:156: error : insn does not satisfy its constraints: (insn 328 326 329 34 (set (mem/s/j:SI (plus:SI (reg:SI More about the author More correct is "val + 0x8000 >= 0x10000 - (TARGET_POWERPC64 ? 8 : 12)" Comment 7 Jakub Jelinek 2012-07-10 04:09:23 EDT As long as something like void foo (void *x, long Current mainline does just that. I think this is more or less just luck of the draw, so the same thing could happen on mainline with a slightly different test case.

Seems this works because + 32764 is not rs6000_legitimate_offset_address_p. I tried revision 160946 and did not see such error. Comment 10 Alan Modra 2012-07-16 23:49:13 EDT What's happening here is that the pseudo for the reg in insn 1485 doesn't get a hard reg. check my blog Comment 5 Alan Modra 2012-07-09 07:43:01 EDT Incidentally, the "o<>" constraint in movdi_internal32 came in with revision 85419.

Bye, [1] http://buildd.debian.org/fetch.php?&pkg=gstreamer0.10-ffmpeg&ver=0.10.0-1&arch=hppa&stamp=1138998904&file=log&as=raw -- Loïc Minier Current Earth status: NOT DESTROYED Reply to: debian-hppa@lists.debian.org Loïc Minier (on-list) Loïc Minier (off-list) Follow-Ups: Re: [hppa] internal compiler error: in reload_cse_simplify_operands, at postreload.c:391 We have a mass rebuild of all our packages coming up, does this issue need to be fixed before that ? Backport r171632 from mainline 2011-03-28 Richard Sandiford * builtins.c (expand_builtin_memset_args): Use gen_int_mode instead of GEN_INT.

See ycode==PLUS checks in offsettable_address_addr_space_p. (2) and (3) together is why we have a problem, since the insn constraints don't cover all possibilities allowed by the insn predicate.

lda : 1), _colFactor(layout == ROW_MAJOR ? 1 : lda) {} inline int operator()(int i, int j) { return i * _rowFactor + j * _colFactor; } private: int _rowFactor; int Status: RESOLVED FIXED Alias: None Product: gcc Classification: Unclassified Component: rtl-optimization (show other bugs) Version: 4.6.0 Importance: P3 normal Target Milestone: 4.6.0 Assignee: Not yet assigned to anyone URL: Keywords: ice-on-valid-code Jun 26, 2015 Project Member #7 [email protected] (No comment was entered for this change.) Labels: Subcomponent-Tools-ndk Jun 26, 2015 Project Member #8 [email protected] (No comment was entered for this rs6000.md doesn't, due to the "o" constraint problem mentioned in comment #3.

Changing version to '19'. (As we did not run this process for some time, it could affect also pre-Fedora 19 development cycle bugs. See for instructions. -------------------------------------------- The expected behaviour is that such code should compile fine. Actual results: Expected results: Additional info: Comment 1 Karsten Hopp 2012-07-05 08:23:17 EDT this issue exists on PPC only, ppc64 builds fine. news Mitt kontoSökMapsYouTubePlayGmailDriveKalenderGoogle+ÖversättFotonMerDokumentBloggerKontakterHangoutsÄnnu mer från GoogleLogga inDolda fältSök efter grupper eller meddelanden My favorites ▼ | Sign in android Android Open Source Project - Issue Tracker ProjectHome Issues New issue Search

Are we going to live with Jakub's "=m,..." workaround or has Alan's patches been merged into the F18 gcc tree yet? We are very sorry. Comment 6 Alan Modra 2012-07-10 00:28:51 EDT Re comment #2 point 4. Format For Printing -XML -Clone This Bug -Top of page Home | New | Browse | Search | [?] | Reports | Help | NewAccount | Log In Remember [x] |

I'm putting [email protected] in Cc:, perhaps they have a clue. Comment 11 Karsten Hopp 2012-07-17 05:06:48 EDT As I have noe clue what all this is about, I need your advice. I can't see any later constraint pairs that would match the non-offsettable parts of "m". Ah, no, that patch was for find_moveable_pseudos which we don't have on 4.7.