[Bug gas/26263] New: [Regression 2.35] binutils breaks 32-bit mesa
https://sourceware.org/bugzilla/show_bug.cgi?id=26263 Bug ID: 26263 Summary: [Regression 2.35] binutils breaks 32-bit mesa Product: binutils Version: 2.35 Status: UNCONFIRMED Severity: normal Priority: P2 Component: gas Assignee: unassigned at sourceware dot org Reporter: nsandschn at gmx dot de CC: hjl.tools at gmail dot com Target Milestone: --- This commit broke 32-bit mesa: commit 292676c15a615b5a95bede9ee91004d3f7ee7dfd Author: H.J. Lu Date: Thu Feb 13 13:44:17 2020 -0800 x86: Resolve PLT32 reloc aganst local symbol to section To reproduce create a i386 chroot and build binutils inside that chroot. (sid-i386)niko@niko-desktop:~/build/binutils-gdb$ git checkout 292676c15a615b5a95bede9ee91004d3f7ee7dfd To fix the binutils build on i386 you need one additionial commit: (sid-i386)niko@niko-desktop:~/build/binutils-gdb$ git cherry-pick 74e10d1742f1b8312359c59a2af06c9e661252b3 This will produce a merge error in two Changelog files but they can be ignored. (sid-i386)niko@niko-desktop:~/build/binutils-gdb$ ./configure --enable-targets=x86_64-linux-gnu,x86_64-linux-gnux32 --disable-gdb --disable-gdbserver (sid-i386)niko@niko-desktop:~/build/binutils-gdb$ make && sudo make install (sid-i386)niko@niko-desktop:~/build/binutils-gdb$ ld -v GNU ld (GNU Binutils) 2.34.50.20200213 Now build mesa inside the chroot: (sid-i386)niko@niko-desktop:~/build/mesa-git$ meson build/ (sid-i386)niko@niko-desktop:~/build/mesa-git$ ninja -C build/ (sid-i386)niko@niko-desktop:~/build/mesa-git$ sudo ninja -C build/ install The build works fine without any apparent errors. However, if you try to run any program using mesa it raises a segmentation fault: (sid-i386)niko@niko-desktop:~/build/mesa-upstream/mesa$ LIBGL_DRIVERS_PATH=/usr/local/lib/i386-linux-gnu/dri glxinfo name of display: :0 Segmentation fault Reverting commit 292676c15a615b5a95bede9ee91004d3f7ee7dfd and rebuilding binutils and mesa fixes the segmentation fault: (sid-i386)niko@niko-desktop:~/build/mesa-upstream/mesa$ LIBGL_DRIVERS_PATH=/usr/local/lib/i386-linux-gnu/dri glxinfo | head -3 name of display: :0 display: :0 screen: 0 direct rendering: Yes -- You are receiving this mail because: You are on the CC list for the bug.
[Bug gas/26263] [Regression 2.35] binutils breaks 32-bit mesa
https://sourceware.org/bugzilla/show_bug.cgi?id=26263 H.J. Lu changed: What|Removed |Added Ever confirmed|0 |1 Status|UNCONFIRMED |WAITING Last reconfirmed||2020-07-19 --- Comment #1 from H.J. Lu --- I don't have i386 system to try. Which .so file under /usr/local/lib/i386-linux-gnu/dri is bad? Please provide good and bad .so files. -- You are receiving this mail because: You are on the CC list for the bug.
[Bug ld/26262] ld/plugin.c:plugin_notice may incorrectly replace an LTO bfd_link_hash_defined with bfd_link_hash_undefweak
https://sourceware.org/bugzilla/show_bug.cgi?id=26262 H.J. Lu changed: What|Removed |Added CC||hjl.tools at gmail dot com -- You are receiving this mail because: You are on the CC list for the bug.
[Bug ld/26262] ld/plugin.c:plugin_notice may incorrectly replace an LTO bfd_link_hash_defined with bfd_link_hash_undefweak
https://sourceware.org/bugzilla/show_bug.cgi?id=26262 H.J. Lu changed: What|Removed |Added Status|UNCONFIRMED |WAITING Ever confirmed|0 |1 Last reconfirmed||2020-07-19 --- Comment #1 from H.J. Lu --- clang has different behavior depending on -fuse-ld=XXX: [hjl@gnu-cfl-2 pr26262]$ clang -fuse-ld=gold -Wl,-y,__llvm_profile_raw_version -Wl,-plugin-opt=save-temps -fprofile-generate -flto -o a a.c -save-temps a.o: definition of __llvm_profile_raw_version /usr/lib64/clang/10.0.0/lib/linux/libclang_rt.profile-x86_64.a(InstrProfiling.c.o): definition of __llvm_profile_raw_version a.o: definition of __llvm_profile_raw_version [hjl@gnu-cfl-2 pr26262]$ clang -fuse-ld=bfd -Wl,-y,__llvm_profile_raw_version -Wl,-plugin-opt=save-temps -fprofile-generate -flto -o a a.o /usr/bin/ld.bfd: a.o: definition of __llvm_profile_raw_version [hjl@gnu-cfl-2 pr26262]$ clang -fuse-ld=bfd -Wl,-y,__llvm_profile_raw_version -Wl,-plugin-opt=save-temps -fprofile-generate -flto -o a a.c -save-temps /usr/bin/ld.bfd: a.o (symbol from plugin): definition of __llvm_profile_raw_version /usr/bin/ld.bfd: /usr/lib64/clang/10.0.0/lib/linux/libclang_rt.profile-x86_64.a(InstrProfiling.c.o): definition of __llvm_profile_raw_version [hjl@gnu-cfl-2 pr26262]$ clang -fuse-ld=gold -Wl,-y,__llvm_profile_raw_version -Wl,-plugin-opt=save-temps -fprofile-generate -flto -o a a.o /usr/lib64/clang/10.0.0/lib/linux/libclang_rt.profile-x86_64.a(InstrProfiling.c.o): definition of __llvm_profile_raw_version [hjl@gnu-cfl-2 pr26262]$ rm a.o [hjl@gnu-cfl-2 pr26262]$ clang -fprofile-generate -flto -c a.c [hjl@gnu-cfl-2 pr26262]$ clang -fuse-ld=bfd -Wl,-y,__llvm_profile_raw_version -Wl,-plugin-opt=save-temps -fprofile-generate -flto -o a a.o /usr/bin/ld.bfd: a.o (symbol from plugin): definition of __llvm_profile_raw_version /usr/bin/ld.bfd: /usr/lib64/clang/10.0.0/lib/linux/libclang_rt.profile-x86_64.a(InstrProfiling.c.o): definition of __llvm_profile_raw_version [hjl@gnu-cfl-2 pr26262]$ clang -fuse-ld=gold -Wl,-y,__llvm_profile_raw_version -Wl,-plugin-opt=save-temps -fprofile-generate -flto -o a a.o /usr/lib64/clang/10.0.0/lib/linux/libclang_rt.profile-x86_64.a(InstrProfiling.c.o): definition of __llvm_profile_raw_version [hjl@gnu-cfl-2 pr26262]$ It looks like a clang bug. -- You are receiving this mail because: You are on the CC list for the bug.
[Bug gas/26263] [Regression 2.35] binutils breaks 32-bit mesa
https://sourceware.org/bugzilla/show_bug.cgi?id=26263 --- Comment #3 from Niko Sandschneider --- Created attachment 12709 --> https://sourceware.org/bugzilla/attachment.cgi?id=12709&action=edit Good version of libglapi.so -- You are receiving this mail because: You are on the CC list for the bug.
[Bug gas/26263] [Regression 2.35] binutils breaks 32-bit mesa
https://sourceware.org/bugzilla/show_bug.cgi?id=26263 --- Comment #2 from Niko Sandschneider --- Created attachment 12708 --> https://sourceware.org/bugzilla/attachment.cgi?id=12708&action=edit Bad version of libglapi.so I tried to get a back trace of the segmentation fault and it points to libglapi.so. I attached the good and bad versions of that file. -- You are receiving this mail because: You are on the CC list for the bug.
[Bug gas/26263] [Regression 2.35] binutils breaks 32-bit mesa
https://sourceware.org/bugzilla/show_bug.cgi?id=26263 H.J. Lu changed: What|Removed |Added Status|WAITING |NEW --- Comment #4 from H.J. Lu --- -libglapi_good.so: file format elf32-i386 +libglapi_bad.so: file format elf32-i386 Disassembly of section .init: @@ -209,7 +209,7 @@ Disassembly of section .text: 5231: c7 46 e4 00 00 ff a0movl $0xa0ff,-0x1c(%esi) 5238: c7 46 e8 34 12 00 00movl $0x1234,-0x18(%esi) 523f: c7 46 ec 90 90 90 90movl $0x90909090,-0x14(%esi) -5246: e8 95 ff ff ff call 51e0 +5246: e8 2e ff ff ff call 5179 524b: 89 6e e8mov%ebp,-0x18(%esi) 524e: 83 c5 04add$0x4,%ebp 5251: 89 46 e2mov%eax,-0x1e(%esi) @@ -264,7 +264,7 @@ Disassembly of section .text: 52e2: 89 46 08mov%eax,0x8(%esi) 52e5: 8b 44 24 0c mov0xc(%esp),%eax 52e9: 89 46 0cmov%eax,0xc(%esi) -52ec: e8 ef fe ff ff call 51e0 +52ec: e8 e2 fd ff ff call 50d3 <__cxa_finalize@plt+0x3> 52f1: 8b 4c 24 20 mov0x20(%esp),%ecx 52f5: 89 46 02mov%eax,0x2(%esi) 52f8: 89 f0 mov%esi,%eax -- You are receiving this mail because: You are on the CC list for the bug.
[Bug gas/26263] [Regression 2.35] binutils breaks 32-bit mesa
https://sourceware.org/bugzilla/show_bug.cgi?id=26263 H.J. Lu changed: What|Removed |Added Status|NEW |WAITING --- Comment #5 from H.J. Lu --- Which version of mesa are you using? -- You are receiving this mail because: You are on the CC list for the bug.
[Bug gas/26263] [Regression 2.35] binutils breaks 32-bit mesa
https://sourceware.org/bugzilla/show_bug.cgi?id=26263 --- Comment #6 from Niko Sandschneider --- I used mesa-git from yesterday (b85ef043245e318422dbf99701267734a336ddf5), but the problem also appears when building mesa-20.1.2. I didn't try any older versions yet. -- You are receiving this mail because: You are on the CC list for the bug.
[Bug gas/26263] [Regression 2.35] binutils breaks 32-bit mesa
https://sourceware.org/bugzilla/show_bug.cgi?id=26263 H.J. Lu changed: What|Removed |Added Target Milestone|--- |2.35 Status|WAITING |NEW --- Comment #7 from H.J. Lu --- [hjl@gnu-cfl-2 pr25551]$ cat x.s .section .init.text,"ax",@progbits .global foo foo: callprintk@PLT callprintk .text printk: ret [hjl@gnu-cfl-2 pr25551]$ make as --32 -o x.o x.s ld -shared -melf_i386 -o x.so x.o objdump -dw x.so x.so: file format elf32-i386 Disassembly of section .text: 1000 : 1000: c3 ret Disassembly of section .init.text: 1001 : 1001: e8 f9 ff ff ff call fff 1006: e8 f5 ff ff ff call 1000 [hjl@gnu-cfl-2 pr25551]$ -- You are receiving this mail because: You are on the CC list for the bug.
[Bug binutils/26246] invalid access in bfd/elf32-xtensa.c
https://sourceware.org/bugzilla/show_bug.cgi?id=26246 Alan Modra changed: What|Removed |Added CC||jcmvbkbc at gcc dot gnu.org -- You are receiving this mail because: You are on the CC list for the bug.
[Bug gas/26263] [Regression 2.35] binutils breaks 32-bit mesa
https://sourceware.org/bugzilla/show_bug.cgi?id=26263 H.J. Lu changed: What|Removed |Added Assignee|unassigned at sourceware dot org |hjl.tools at gmail dot com --- Comment #8 from H.J. Lu --- Created attachment 12710 --> https://sourceware.org/bugzilla/attachment.cgi?id=12710&action=edit A patch Please try this. -- You are receiving this mail because: You are on the CC list for the bug.
[Bug gas/26263] [Regression 2.35] binutils breaks 32-bit mesa
https://sourceware.org/bugzilla/show_bug.cgi?id=26263 --- Comment #9 from cvs-commit at gcc dot gnu.org --- The master branch has been updated by H.J. Lu : https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=2585b7a5ce5830e60a089aa2316a329558902f0c commit 2585b7a5ce5830e60a089aa2316a329558902f0c Author: H.J. Lu Date: Sun Jul 19 06:51:19 2020 -0700 x86: Change PLT32 reloc against section to PC32 Commit 292676c1 resolved PLT32 reloc aganst local symbol to section. Since PLT32 relocation must be against symbols, turn such PLT32 relocation into PC32 relocation. gas/ PR gas/26263 * config/tc-i386.c (i386_validate_fix): Change PLT32 reloc against section to PC32 reloc. * testsuite/gas/i386/relax-5.d: Updated. * testsuite/gas/i386/x86-64-relax-4.d: Likewise. ld/ PR gas/26263 * testsuite/ld-i386/i386.exp: Run PR gas/26263 test. * testsuite/ld-x86-64/x86-64.exp: Likewise. * testsuite/ld-i386/pr26263.d: New file. * testsuite/ld-x86-64/pr26263.d: Likewise. * testsuite/ld-x86-64/pr26263.s: Likewise. -- You are receiving this mail because: You are on the CC list for the bug.
[Bug gas/26263] [Regression 2.35] binutils breaks 32-bit mesa
https://sourceware.org/bugzilla/show_bug.cgi?id=26263 Niko Sandschneider changed: What|Removed |Added Resolution|--- |FIXED Status|NEW |RESOLVED --- Comment #10 from Niko Sandschneider --- The patch indeed fixes the segmentation fault. Thank you very much for the quick fix! -- You are receiving this mail because: You are on the CC list for the bug.
[Bug gas/26263] [Regression 2.35] binutils breaks 32-bit mesa
https://sourceware.org/bugzilla/show_bug.cgi?id=26263 --- Comment #11 from cvs-commit at gcc dot gnu.org --- The binutils-2_35-branch branch has been updated by H.J. Lu : https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=52da8d36c60fc2ad7bef1256a9bbc3e3b1602787 commit 52da8d36c60fc2ad7bef1256a9bbc3e3b1602787 Author: H.J. Lu Date: Sun Jul 19 06:51:19 2020 -0700 x86: Change PLT32 reloc against section to PC32 Commit 292676c1 resolved PLT32 reloc aganst local symbol to section. Since PLT32 relocation must be against symbols, turn such PLT32 relocation into PC32 relocation. gas/ PR gas/26263 * config/tc-i386.c (i386_validate_fix): Change PLT32 reloc against section to PC32 reloc. * testsuite/gas/i386/relax-5.d: Updated. * testsuite/gas/i386/x86-64-relax-4.d: Likewise. ld/ PR gas/26263 * testsuite/ld-i386/i386.exp: Run PR gas/26263 test. * testsuite/ld-x86-64/x86-64.exp: Likewise. * testsuite/ld-i386/pr26263.d: New file. * testsuite/ld-x86-64/pr26263.d: Likewise. * testsuite/ld-x86-64/pr26263.s: Likewise. (cherry picked from commit 2585b7a5ce5830e60a089aa2316a329558902f0c) -- You are receiving this mail because: You are on the CC list for the bug.
[Bug gas/26263] [Regression 2.35] binutils breaks 32-bit mesa
https://sourceware.org/bugzilla/show_bug.cgi?id=26263 --- Comment #12 from H.J. Lu --- Fixed for master and 2.35. -- You are receiving this mail because: You are on the CC list for the bug.
[Bug ld/26265] New: Spurious 'gc-sections requires either an entry or an undefined symbol' when --init or --fini is specified
https://sourceware.org/bugzilla/show_bug.cgi?id=26265 Bug ID: 26265 Summary: Spurious 'gc-sections requires either an entry or an undefined symbol' when --init or --fini is specified Product: binutils Version: unspecified Status: UNCONFIRMED Severity: normal Priority: P2 Component: ld Assignee: unassigned at sourceware dot org Reporter: i at maskray dot me Target Milestone: --- link_info.init_function & link_info.fini_function are GC roots. When they are explicitly specified, there is no need for the error: % ld.bfd -r --gc-sections -init _init a.o ld.bfd: gc-sections requires either an entry or an undefined symbol -- You are receiving this mail because: You are on the CC list for the bug.
[Bug ld/26262] ld/plugin.c:plugin_notice may incorrectly replace an LTO bfd_link_hash_defined with bfd_link_hash_undefweak
https://sourceware.org/bugzilla/show_bug.cgi?id=26262 --- Comment #2 from Fangrui Song --- (In reply to H.J. Lu from comment #1) > clang has different behavior depending on -fuse-ld=XXX: > > ... > > It looks like a clang bug. Thanks for the report. It is an LLVMgold.so bug. LLVMgold.so is used by GNU ld and gold for clang LTO (full LTO or ThinLTO). The problem is that clang -fuse-ld=bfd -flto a.o -o a -Wl,-plugin-opt=save-temps # override a.o I will fix it in https://reviews.llvm.org/D84132 and cherry pick it into llvm-project's release/11.x branch (newly created for the 11.0.0 release). Let's move back to the original topic. Use a different output filename (than the basename of the source file) to avoid the -plugin-opt=save-temps issue. - % clang -fuse-ld=bfd -fprofile-generate=out -flto a.c -Wl,-y,__llvm_profile_raw_version -Wl,-plugin-opt=save-temps -o aa /usr/bin/ld.bfd: /tmp/a-1cd442.o (symbol from plugin): definition of __llvm_profile_raw_version /usr/bin/ld.bfd: /tmp/RelA/lib/clang/12.0.0/lib/linux/libclang_rt.profile-x86_64.a(InstrProfiling.c.o): definition of __llvm_profile_raw_version % ./aa % llvm-profdata merge out -o a.profdata % llvm-profdata show a.profdata Instrumentation level: Front-end### incorrect, it should be IR Total functions: 1 Maximum function count: 1 Maximum internal block count: 0 % rm -r out % clang -fuse-ld=gold -fprofile-generate=out -flto a.c -Wl,-y,__llvm_profile_raw_version -Wl,-plugin-opt=save-temps -o aa /tmp/a-3b2969.o: definition of __llvm_profile_raw_version /tmp/RelA/lib/clang/12.0.0/lib/linux/libclang_rt.profile-x86_64.a(InstrProfiling.c.o): definition of __llvm_profile_raw_version aa.o: definition of __llvm_profile_raw_version % ./aa % llvm-profdata merge out -o a.profdata % llvm-profdata show a.profdata Instrumentation level: IR Total functions: 1 Maximum function count: 1 Maximum internal block count: 0 As my first comment says, you can inspect aa.resolution.txt to see that GNU ld incorrectly thinks that __llvm_profile_raw_versio is non-prevailing. -- You are receiving this mail because: You are on the CC list for the bug.
[Bug ld/26265] Spurious 'gc-sections requires either an entry or an undefined symbol' when --init or --fini is specified
https://sourceware.org/bugzilla/show_bug.cgi?id=26265 Alan Modra changed: What|Removed |Added Last reconfirmed||2020-07-20 Assignee|unassigned at sourceware dot org |amodra at gmail dot com Status|UNCONFIRMED |ASSIGNED Ever confirmed|0 |1 -- You are receiving this mail because: You are on the CC list for the bug.
[Bug gas/4572] division doesn't work on solaris/x86
https://sourceware.org/bugzilla/show_bug.cgi?id=4572
--- Comment #13 from cvs-commit at gcc dot gnu.org ---
The master branch has been updated by Jan Beulich :
https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=750e4bf70f2caab39dc5d0a1b2d26c9ca1fbd909
commit 750e4bf70f2caab39dc5d0a1b2d26c9ca1fbd909
Author: Jan Beulich
Date: Mon Jul 20 08:56:23 2020 +0200
gas: generalize comment character escaping recognition
PR gas/4572
Generalize what ab1fadc6b2f0 ("PR22714, Assembler preprocessor loses
track of \@") did to always honor escaped comment chars. Use this then
to support escaped /, %, and * operators on x86, when / is a comment
char (to match the Sun assembler's behavior).
--
You are receiving this mail because:
You are on the CC list for the bug.
[Bug gas/4572] division doesn't work on solaris/x86
https://sourceware.org/bugzilla/show_bug.cgi?id=4572 --- Comment #14 from cvs-commit at gcc dot gnu.org --- The master branch has been updated by Jan Beulich : https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=b3983e5f53caad175563f8c842f2ab2a1277c2bc commit b3983e5f53caad175563f8c842f2ab2a1277c2bc Author: Jan Beulich Date: Mon Jul 20 08:57:18 2020 +0200 x86: handle SVR4 escaped binary operators PR gas/4572 When / is a comment character, its use as binary "divide" operator needs escaping by a backslash. Besides the scrubber needing to support this (addressed in an earlier change), there are also a few provisions needed in target specific operator handling. As the spec calls for % and * to also be escaped because of being "overloaded", also recognize these, despite the overloading there not really preventing their use as operators in most (%) or all (*) cases, given the way how the rest of the assembler works. To bring source and testsuite in line, also drop the TE_I386AIX part of the respective conditional, as i?86-*-aix* support had been removed a while ago. -- You are receiving this mail because: You are on the CC list for the bug.
