[Bug ld/17288] New: sh64 incremental link does not set flags in ELF header
https://sourceware.org/bugzilla/show_bug.cgi?id=17288
Bug ID: 17288
Summary: sh64 incremental link does not set flags in ELF header
Product: binutils
Version: 2.24
Status: NEW
Severity: normal
Priority: P2
Component: ld
Assignee: unassigned at sourceware dot org
Reporter: dhowells at redhat dot com
Doing an incremental link with an sh64 binutils does not set the flags field in
the ELF header. Using this script:
#!/bin/sh -x
CROSS=sh64-linux-gnu-
LDFLAGS="-EL -mshlelf64"
CFLAGS=-m5-64media-nofpu
rm -f hello.o wibble.o foo.o bar.o
echo 'int hello(int i) {return i + 42;}' |
${CROSS}gcc $CFLAGS -x c -c - -o hello.o || exit $?
${CROSS}readelf -h hello.o | grep Flags
${CROSS}ar rcsD wibble.o hello.o || exit $?
${CROSS}ld $LDFLAGS -r -o foo.o wibble.o || exit $?
${CROSS}readelf -h foo.o | grep Flags
${CROSS}ld $LDFLAGS -r -o bar.o foo.o || exit $?
${CROSS}readelf -h bar.o | grep Flags
The following is seen:
+ CROSS=sh64-linux-gnu-
+ LDFLAGS='-EL -mshlelf64'
+ CFLAGS=-m5-64media-nofpu
+ rm -f hello.o wibble.o foo.o bar.o
+ echo 'int hello(int i) {return i + 42;}'
+ sh64-linux-gnu-gcc -m5-64media-nofpu -x c -c - -o hello.o
+ sh64-linux-gnu-readelf -h hello.o
+ grep Flags
Flags: 0xa, sh5
+ sh64-linux-gnu-ar rcsD wibble.o hello.o
+ sh64-linux-gnu-ld -EL -mshlelf64 -r -o foo.o wibble.o
+ sh64-linux-gnu-readelf -h foo.o
+ grep Flags
Flags: 0x0
+ sh64-linux-gnu-ld -EL -mshlelf64 -r -o bar.o foo.o
sh64-linux-gnu-ld: unknown architecture of input file `foo.o' is incompatible
with sh5 output
+ exit 1
As can be seen, the assembler's output is flagged 0xa (sh5), but the linker's
output is flagged 0x0. Note that adding -belf64-sh64l to the ld command line
doesn't help.
--
You are receiving this mail because:
You are on the CC list for the bug.
___
bug-binutils mailing list
bug-binutils@gnu.org
https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug ld/17288] sh64 incremental link does not set flags in ELF header
https://sourceware.org/bugzilla/show_bug.cgi?id=17288 --- Comment #1 from David Howells --- The binutils was built from the 2.24 tarball plus a patch bringing it up to git release cab6c3ee9785f072a373afe31253df0451db93cf. Build configuration: + LDFLAGS='-Wl,-z,relro ' + ../binutils-2.24/configure \ --disable-dependency-tracking \ --disable-silent-rules \ --enable-checking \ --build=x86_64-redhat-linux-gnu \ --host=x86_64-redhat-linux-gnu \ --target=sh64-linux-elf \ --program-prefix=sh64-linux-gnu- \ --disable-shared \ --disable-install_libbfd \ --with-sysroot=/usr/sh64-linux-gnu/sys-root \ --enable-64-bit-bfd (I've removed the flags that set the various installation paths and the bug url) -- You are receiving this mail because: You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug ld/17288] sh64 incremental link does not set flags in ELF header
https://sourceware.org/bugzilla/show_bug.cgi?id=17288 --- Comment #3 from David Howells --- Using --whole-archive seems to work. + sh64-linux-gnu-ar rcsD wibble.o hello.o + sh64-linux-gnu-ld -EL -mshlelf64 --whole-archive -r -o foo.o wibble.o + sh64-linux-gnu-readelf -h foo.o + grep Flags Flags: 0xa, sh5 + sh64-linux-gnu-ld -EL -mshlelf64 -r -o bar.o foo.o + sh64-linux-gnu-readelf -h bar.o + grep Flags Flags: 0xa, sh5 -- You are receiving this mail because: You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug ld/17288] sh64 incremental link does not set flags in ELF header
https://sourceware.org/bugzilla/show_bug.cgi?id=17288 --- Comment #6 from David Howells --- Should the linker invent a new e_flags based on the -b and -m flags it has been given and reject any contributory object file that has different flags? -- You are receiving this mail because: You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug ld/17288] sh64 incremental link does not set flags in ELF header
https://sourceware.org/bugzilla/show_bug.cgi?id=17288 --- Comment #7 from David Howells --- Any further thoughts on this? -- You are receiving this mail because: You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug gas/18189] New: Extra-large constant handling failure in microblaze
https://sourceware.org/bugzilla/show_bug.cgi?id=18189 Bug ID: 18189 Summary: Extra-large constant handling failure in microblaze Product: binutils Version: 2.25 Status: NEW Severity: normal Priority: P2 Component: gas Assignee: unassigned at sourceware dot org Reporter: dhowells at redhat dot com Created attachment 8220 --> https://sourceware.org/bugzilla/attachment.cgi?id=8220&action=edit libgcc derived assembly file that shows the problem. Attempting to assemble this line with a microblaze-linux-gnu target assembler: addikr23,r0,0xb746a0003f80 works when the compiler is x86_64 hosted but fails when i386 hosted with the following error: _powisf2.s: Assembler messages: _powisf2.s:71: Fatal error: operand must be a constant or a label Given that microblaze is 32-bit only, I would guess that the error should appear on x86_64 as well as on i386 - and that this is actually a symptom of a gcc bug. The error can be obtained by running the attached assembly file through the assembler thusly: microblaze-linux-gnu-as _powisf2.s -- You are receiving this mail because: You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug gas/18189] Extra-large constant handling failure in microblaze
https://sourceware.org/bugzilla/show_bug.cgi?id=18189 --- Comment #1 from David Howells --- Created attachment 8221 --> https://sourceware.org/bugzilla/attachment.cgi?id=8221&action=edit Preprocessed libgcc C file For reference, here's the preprocessed libgcc file that compiles to the previously attached assembly. The compiler command line was: /root/cross-gcc/gcc-5.0.0-20150319/microblaze-linux-gnu/gcc/xgcc -B/root/cross-gcc/gcc-5.0.0-20150319/microblaze-linux-gnu/gcc/ -B/usr/microblaze-linux-gnu/bin/ -B/usr/microblaze-linux-gnu/lib/ -O2 -fPIC -fbuilding-libgcc -fno-exceptions -fno-stack-protector -fvisibility=hidden -g -o _powisf2.o -c _powisf2.i -- You are receiving this mail because: You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug gas/18189] Extra-large constant handling failure in microblaze
https://sourceware.org/bugzilla/show_bug.cgi?id=18189 --- Comment #2 from David Howells --- binutils-2.25 was the base, plus patches as for the Fedora 21 binutils rpm. The configuration for the x86_64 binutils was: LDFLAGS='-Wl,-z,relro ' \ ../binutils-2.25/configure --disable-dependency-tracking --disable-silent-rules --enable-checking --prefix=/usr --exec-prefix=/usr --bindir=/usr/bin --sbindir=/usr/sbin --sysconfdir=/etc --datadir=/usr/share --includedir=/usr/include --libdir=/usr/lib64 --libexecdir=/usr/libexec --localstatedir=/var --sharedstatedir=/var/lib --mandir=/usr/share/man --infodir=/usr/share/info --build=x86_64-redhat-linux-gnu --host=x86_64-redhat-linux-gnu --target=microblaze-linux-gnu --program-prefix=microblaze-linux-gnu- --disable-shared --disable-install_libbfd --with-sysroot=/usr/microblaze-linux-gnu/sys-root --with-bugurl=http://bugzilla.redhat.com/bugzilla/ The i386 variant would be similar, changing x86_64 to i386. and the build line: make -C microblaze -j5 tooldir=/usr all -- You are receiving this mail because: You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug gas/18189] Extra-large constant handling failure in microblaze
https://sourceware.org/bugzilla/show_bug.cgi?id=18189 --- Comment #6 from David Howells --- Works for me, thanks! -- You are receiving this mail because: You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug binutils/19587] New: Some of the less-well used targets won't compile with gcc-6 due to static vars in opcode headers
https://sourceware.org/bugzilla/show_bug.cgi?id=19587
Bug ID: 19587
Summary: Some of the less-well used targets won't compile with
gcc-6 due to static vars in opcode headers
Product: binutils
Version: unspecified
Status: NEW
Severity: normal
Priority: P2
Component: binutils
Assignee: unassigned at sourceware dot org
Reporter: dhowells at redhat dot com
Target Milestone: ---
gcc-6 added a new warning -Wunused-const-variable that causes unused constant
variables to generate a warning. Binutils adds -Werror to the build which
turns this into an error.
The opcode header files define a bunch of static constant variables, mostly
arrays, that now fall foul of this. Errors like the following are generated:
../../binutils-2.26/opcodes/../include/opcode/metag.h:686:30: error:
'metag_scondtab' defined but not used [-Werror=unused-const-variable]
This error can be suppressed by adding -Wno-unused-const-variable to the build.
Note that whilst -Wno-error=unused-const-variable could be used with gcc-6, it
cannot be used with anything older.
The following are all potential problems:
warthog>git grep "static const.*=" master -- include/opcode/*.h | sed
s/master://
include/opcode/dlx.h:static const struct dlx_opcode dlx_opcodes[] =
include/opcode/hppa.h:static const char *const completer_chars =
",CcY<>?!@+&U~FfGHINnOoZMadu|/=0123%e$m}";
include/opcode/hppa.h:static const struct pa_opcode pa_opcodes[] =
include/opcode/i860.h:static const struct i860_opcode i860_opcodes[] =
include/opcode/metag.h:static const metag_reg metag_regtab[] =
include/opcode/metag.h:static const metag_reg metag_dsp_regtab[] =
include/opcode/metag.h:static const metag_reg metag_dsp_tmpl_regtab[2][56] =
include/opcode/metag.h:static const metag_acf metag_acftab[] =
include/opcode/metag.h:static const split_condition metag_scondtab[] =
include/opcode/metag.h:static const split_condition metag_dsp_scondtab[] =
include/opcode/metag.h:static const split_condition metag_fpu_scondtab[] =
include/opcode/metag.h:static const insn_template metag_optab[] =
include/opcode/mips.h:static const unsigned int mips_isa_table[] = {
include/opcode/nds32.h:static const int nds32_r45map[] =
include/opcode/nds32.h:static const int nds32_r54map[] =
include/opcode/ns32k.h:static const struct ns32k_opcode ns32k_opcodes[]=
include/opcode/score-datadep.h:static const struct insn_to_dependency
insn_to_dependency_table[] =
include/opcode/score-datadep.h:static const struct data_dependency
data_dependency_table[] =
include/opcode/tic30.h:static const reg tic30_regtab[] = {
include/opcode/tic30.h:static const ind_addr_type tic30_indaddr_tab[] = {
include/opcode/tic30.h:static const insn_template tic30_optab[] = {
include/opcode/tic30.h:static const insn_template *const tic30_optab_end =
include/opcode/tic30.h:static const partemplate tic30_paroptab[] = {
include/opcode/tic30.h:static const partemplate *const tic30_paroptab_end =
include/opcode/tic4x.h:static const tic4x_register_t tic3x_registers[] =
include/opcode/tic4x.h:static const tic4x_register_t tic4x_registers[] =
include/opcode/tic4x.h:static const tic4x_cond_t tic4x_conds[] =
include/opcode/tic4x.h:static const tic4x_indirect_t tic4x_indirects[] =
include/opcode/tic4x.h:static const tic4x_inst_t tic4x_insts[] =
include/opcode/visium.h:static const struct reg_entry gen_reg_table[] =
include/opcode/visium.h:static const struct reg_entry fp_reg_table[] =
--
You are receiving this mail because:
You are on the CC list for the bug.
___
bug-binutils mailing list
bug-binutils@gnu.org
https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug gas/19614] New: gas for c6x prints "Error: inconsistent uses of .cfi_sections" when gcc compiles things with it
https://sourceware.org/bugzilla/show_bug.cgi?id=19614
Bug ID: 19614
Summary: gas for c6x prints "Error: inconsistent uses of
.cfi_sections" when gcc compiles things with it
Product: binutils
Version: 2.26
Status: NEW
Severity: normal
Priority: P2
Component: gas
Assignee: unassigned at sourceware dot org
Reporter: dhowells at redhat dot com
Target Milestone: ---
When gcc is building for c6x with a command like the following:
echo 'int main() { return 0; }' | ./gcc/xgcc -x c -c -g - -B ./gcc
-fexceptions
the following occurs with the 2.26 gas:
/tmp/cc6wIVaX.s: Assembler messages:
/tmp/cc6wIVaX.s:12: Error: inconsistent uses of .cfi_sections
The problem is the following in the assembly source output by gcc:
...
.cfi_sections .debug_frame
...
.cfi_sections .debug_frame, .c6xabi.exidx
...
If the first directive is modified to have the same parameter as the second
then the error goes away.
This isn't a problem in 2.25 and is due to the following commit:
commit 2f0c68f23bb3132cd5ac466ca8775c0d9e4960cd
Author: Catherine Moore
Date: Thu May 28 14:50:36 2015 -0700
Compact EH Support
adding a check:
if (cfi_sections_set && cfi_sections != sections)
as_bad (_("inconsistent uses of .cfi_sections"));
Can the check be relaxed so that if the parameter hasn't been specified yet,
it's made retroactive, but only incurs an error if changed once specified?
gcc bug: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69747
--
You are receiving this mail because:
You are on the CC list for the bug.
___
bug-binutils mailing list
bug-binutils@gnu.org
https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug gas/20047] New: The arc assembler outputs an "unrecognized option" error with a weird character in it
https://sourceware.org/bugzilla/show_bug.cgi?id=20047 Bug ID: 20047 Summary: The arc assembler outputs an "unrecognized option" error with a weird character in it Product: binutils Version: 2.26 Status: NEW Severity: normal Priority: P2 Component: gas Assignee: unassigned at sourceware dot org Reporter: dhowells at redhat dot com Target Milestone: --- Running the arc assembler produces an error with a weird character in it: echo | /data/fedora/cross-gcc/gcc-6.1.1-20160427/arc-linux-gnu/./gcc/as -v -mARC700 -mEA -o conftest.o GNU assembler version 2.26 (arc-linux-uclibc) using BFD version version 2.26.20160125 Assembler messages: Error: unrecognized option -Î It appears that it doesn't like the -mEA argument supplied by gcc, but gets the error message wrong. The version was binutils-2.26. The configuration was: LDFLAGS='-Wl,-z,relro ' \ ../binutils-2.26/configure --disable-dependency-tracking --disable-silent-rules --enable-checking --prefix=/usr --exec-prefix=/usr --bindir=/usr/bin --sbindir=/usr/sbin --sysconfdir=/etc --datadir=/usr/share --includedir=/usr/include --libdir=/usr/lib64 --libexecdir=/usr/libexec --localstatedir=/var --sharedstatedir=/var/lib --mandir=/usr/share/man --infodir=/usr/share/info --build=x86_64-redhat-linux-gnu --host=x86_64-redhat-linux-gnu --target=arc-linux-uclibc --program-prefix=arc-linux-gnu- --disable-shared --disable-install_libbfd --with-sysroot=/usr/arc-linux-gnu/sys-root --enable-deterministic-archives=no --with-bugurl=http://bugzilla.redhat.com/bugzilla/ -- You are receiving this mail because: You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug gas/21021] New: Assertion failure in ignore_rest_of_line at gas/read.c:3758
https://sourceware.org/bugzilla/show_bug.cgi?id=21021 Bug ID: 21021 Summary: Assertion failure in ignore_rest_of_line at gas/read.c:3758 Product: binutils Version: 2.27 Status: NEW Severity: normal Priority: P2 Component: gas Assignee: unassigned at sourceware dot org Reporter: dhowells at redhat dot com Target Milestone: --- Created attachment 9736 --> https://sourceware.org/bugzilla/attachment.cgi?id=9736&action=edit Test code Compiling the attached test case with: xtensa-linux-gnu-as /tmp/xtensa.s results in the following: /tmp/xtensa.s: Assembler messages: /tmp/xtensa.s:39: Error: bad expression /tmp/xtensa.s:39: Error: found '�', expected: ')' /tmp/xtensa.s:39: Error: junk at end of line, first unrecognized character valued 0xfff9 /tmp/xtensa.s:39: Internal error! Assertion failure in ignore_rest_of_line at ../../binutils-2.27/gas/read.c:3758. Please report this bug. Line 39 is: bbci.l a10, 0, 1f but attempting to remove any of the other lines affects the error produced. The unexpected character changes, depending on the run. Valgrind doesn't find anything. The binutils-2.27 package was built on x86_64 using: LDFLAGS='-Wl,-z,relro ' ./configure --disable-dependency-tracking --disable-silent-rules --enable-checking --prefix=/usr --exec-prefix=/usr --bindir=/usr/bin --sbindir=/usr/sbin --sysconfdir=/etc --datadir=/usr/share --includedir=/usr/include --libdir=/usr/lib64 --libexecdir=/usr/libexec --localstatedir=/var --sharedstatedir=/var/lib --mandir=/usr/share/man --infodir=/usr/share/info --build=x86_64-redhat-linux-gnu --host=x86_64-redhat-linux-gnu --target=xtensa-linux-gnu --program-prefix=xtensa-linux-gnu- --disable-shared --disable-install_libbfd --with-sysroot=/usr/xtensa-linux-gnu/sys-root --enable-deterministic-archives=no --enable-compressed-debug-sections=none --enable-lto --with-bugurl=http://bugzilla.redhat.com/bugzilla/ -- You are receiving this mail because: You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug ld/21914] [x86_64] SEGV during link
https://sourceware.org/bugzilla/show_bug.cgi?id=21914 --- Comment #1 from David Howells --- binutils was configured thusly: + LDFLAGS='-Wl,-z,relro ' + ../binutils-2.29/configure --disable-dependency-tracking --disable-silent-rules --enable-checking --prefix=/usr --exec-prefix=/usr --bindir=/usr/bin --sbindir=/usr/sbin --sysconfdir=/etc --datadir=/usr/share --includedir=/usr/include --libdir=/usr/lib64 --libexecdir=/usr/libexec --localstatedir=/var --sharedstatedir=/var/lib --mandir=/usr/share/man --infodir=/usr/share/info --build=x86_64-redhat-linux-gnu --host=x86_64-redhat-linux-gnu --target=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --disable-shared --disable-install_libbfd --with-sysroot=/usr/x86_64-linux-gnu/sys-root --enable-deterministic-archives=no --enable-compressed-debug-sections=none --enable-lto --with-bugurl=http://bugzilla.redhat.com/bugzilla/ -- You are receiving this mail because: You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug ld/21914] New: [x86_64] SEGV during link
https://sourceware.org/bugzilla/show_bug.cgi?id=21914 Bug ID: 21914 Summary: [x86_64] SEGV during link Product: binutils Version: 2.29 Status: NEW Severity: normal Priority: P2 Component: ld Assignee: unassigned at sourceware dot org Reporter: dhowells at redhat dot com Target Milestone: --- Created attachment 10322 --> https://sourceware.org/bugzilla/attachment.cgi?id=10322&action=edit Test data ld from binutils-2.29 can be made to crash when attempting to link the attached test data. The test data needs to be decompressed and then ld commanded as follows: x86_64-linux-gnu-ld -m elf_i386 -N --no-check-sections --section-start=.prefix=0 --gc-sections -static -o usbdisk.bin --oformat binary -e 0 usbdisk.o Note that the ld being used here is an x86_64-on-x86_64 cross. Poking around in the crashed ld process with gdb shows: Program received signal SIGSEGV, Segmentation fault. 0x0042e3a4 in gldelf_i386_place_orphan (s=0x82cea0, secname=0x82c692 ".prefix", constraint=382) at eelf_i386.c:1978 1978&& (elf_section_data (os->bfd_section)->this_hdr.sh_info (gdb) bt #0 0x0042e3a4 in gldelf_i386_place_orphan (s=0x82cea0, secname=0x82c692 ".prefix", constraint=382) at eelf_i386.c:1978 #1 0x00412bfa in lang_place_orphans () at ../../binutils-2.29/ld/ldlang.c:6396 #2 0x00416f21 in lang_process () at ../../binutils-2.29/ld/ldlang.c:7198 #3 0x00403ee0 in main (argc=, argv=) at ../../binutils-2.29/ld/ldmain.c:413 The fault apparently occurs on line 1978 in the following: 1977if (os->bfd_section != NULL 1978&& (elf_section_data (os->bfd_section)->this_hdr.sh_info 1979== elf_section_data (s)->this_hdr.sh_info) 1980&& (os->bfd_section->flags == 0 1981|| ((!bfd_link_relocatable (&link_info) 1982 || (iself && (((elf_section_flags (s) 1983 ^ elf_section_flags (os->bfd_section)) 1984& SHF_EXCLUDE) == 0))) (gdb) p os $1 = (lang_output_section_statement_type *) 0x81e168 (gdb) p os->bfd_section $2 = (asection *) 0x820720 (gdb) p s $3 = (asection *) 0x82cea0 Looking at the disassembly, line 1977 starts here: 0x0042e386 <+822>: mov0x38(%rbx),%rsi 0x0042e38a <+826>: test %rsi,%rsi 0x0042e38d <+829>: je 0x42e448 0x0042e393 <+835>: mov0xe0(%rsi),%rax 0x0042e39a <+842>: mov0xe0(%rbp),%rdx 0x0042e3a1 <+849>: mov0x2c(%rdx),%ecx => 0x0042e3a4 <+852>: cmp%ecx,0x2c(%rax) So %rbx holds 'os', %rsi holds 'os->bfd_section' and %rbp holds 's'. (gdb) p/x *(long*)($rsi+0xe0) $13 = 0x0 (gdb) p/x *(long*)($rbp+0xe0) $14 = 0x82f170 So elf_section_data (os->bfd_section) returns NULL and a SEGV occurs when we try and dereference it. -- You are receiving this mail because: You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
