external/binutils.git
4 years agoAdd .note.gnu.property runtime verification and merge support submit/tizen_5.5_base/20191030.000001 submit/tizen_base/20190724.060247 tizen_5.5.m2_release
Mikhail Kashkarov [Mon, 15 Apr 2019 11:15:49 +0000 (14:15 +0300)]
Add .note.gnu.property runtime verification and merge support

- Define annobin .note.gnu.property bits for sanitization/CxxABI checks
- New configure option --enable-annobin-verification=warn|strict to enable
  runtime verification support for gnu property notes.
- Add merging support for compilation flags annobin gnu notes.
- Add readelf section text for compiler properties.

Change-Id: I452029baca753f6a97ef4b7297ef9a8905a7b79c

4 years agoAdd support for elf properties section.
Mikhail Kashkarov [Mon, 17 Oct 2016 14:29:43 +0000 (15:29 +0100)]
Add support for elf properties section.

This is a set of the following backports/cherry-picks to support
.note.gnu.property section.

- Display unknown notes.  Decode NT_GNU_HWCAP notes.
- Add support for displaying and merging GNU_BUILD_NOTEs.
- Fix snafu parsing GNU_BUILD_NOTEs on ARM and AArch64 architectures.
- Fix reading numeric values from GNU BUILD NOTEs.
- Properly dump NT_GNU_PROPERTY_TYPE_0
- Try harder to find the correct symbol to associate with OPEN GNU BUILD notes.
- Add support for a GNU BUILD note type to record the enum size.
- Update support for GNU BUILD notes so that version notes can contain extra information, and stack protection notes can contain numeric values.
- Add note merging to strip and add code to merge stack size notes.
- Fix shift overflow when parsing an overlarge note value.
- Improve decoding of corrupt/unrecognised gnu build attribute notes.
- Fix off by one error when checking for empty note names.
- Add support for version 2 of the GNU Build Attribute note specification.
- Add test of V2 GNU build attribute notes.
- Always pass a valid section header offset to elf_parse_notes
- Re-apply "elf: Properly compute offsets of note descriptor and next note"
- Work around integer overflows when readelf is checking for corrupt ELF notes when run on a 32-bit host.
- Fix a seg-fault when displaying notes from a prorgam segment.  Check for a NULL string section before attempting compute the name of a separate debug info file.
- Properly handle note sections and segments
- ld: Support ELF GNU program properties
- merge_gnu_build_notes reloc deletion
- Add support for v3 binary annotation notes.
- Fix displaying the value associated a GNU BUILD note that uses an ascii name.
- Combine ,gnu.build.attributes.* sections into one .gnu.build.attributes section during linking.
- Improve readelf's selection of a file start symbol when displaying a gnu build attribute.

Change-Id: I7d828cfcc17eda6dcbfe6967e2f26b4d67238406

SHA's:

1449284bd80db7bb4cbe7fef21e1e8b753135af6
9ef920e933bf2ea228c909cf81636e6d9577e51e
c871dadee1817d4b9f3ba6ee792730c9eccf88e0
79a964dca572024447adf92e50959dc88aa4f27a
1fc87489b40b3100badf184a7c266387bae47def
c799a79d66d09c442d85467986c2e8873f09297c
a8be5506b626a57f84771c5ddfaefabf2d61c017
75d7d2986cf896fac8f0690db68ebc552e0b0339
1d15e434f43bc41a07bc7b0648fcb7e6ccbe8dcc
ddef72cdc10d82ba011a7ff81cafbbd3466acf54
3e6b644558f941d3fe482e15efbc53d8f39ef8b7
b06b2c92c06cf100f259f218337d007ee0b1c884
88305e1b9f4592a2ab129015ce409fcc16613ebb
714da62f64ef5527a22a3e897f7e21f69e442ade
08698b7b46c9c932530dcbcd7e851d27a7d2c503
276da9b31bd6e3eb8d1dd814c867266f59f29093
6ab2c4ed51f9c4243691755e1b1d2149c6a426f4
4dff97b2ce94bbb9c0c16e0e4129060b9a894f13
82ed9683ec099d8205dc499ac84febc975235af6
46bed6796d5821832e8ba373ddb2e7fdc45a109d
ee7e95efb98186c09dc2c39b32263aa15b147bb0
6f156d7a4ad1c245b357dc50f63f3564d397839e
0dd6ae21da832f351a3722d2f8bac187fc3bdfef
7d8a31665739412395f6dd370d2279acd322e78e
8fd75781892c7458702f72b3b9cb1927609716de

Change-Id: I7b201b6c0776c3a7e7e25cdbbb937170da244405

5 years agold/testing: clear compiler & linker flags when running testsuite. submit/tizen_base/20181228.105429 submit/tizen_base/20190108.221718
Andrey Drobyshev [Mon, 8 Oct 2018 12:41:37 +0000 (15:41 +0300)]
ld/testing: clear compiler & linker flags when running testsuite.

Passing CLFAGS, CXXFLAGS, LDFLAGS used previously on build stage cause failures
when running "make check". It's safer to clear them.

Change-Id: If2bba2baca0f99c3931af2a55739ff6a087f2b0f
Signed-off-by: Andrey Drobyshev <a.drobyshev@partner.samsung.com>
5 years agold/testing: fix "compilation failed" message for linker tests on x86.
Andrey Drobyshev [Thu, 4 Oct 2018 18:23:47 +0000 (21:23 +0300)]
ld/testing: fix "compilation failed" message for linker tests on x86.

 * ld-ifunc/pr18808b.c, ld-ifunc/pr19784c: change return type from int
 to void, as it should be.

Change-Id: I4b9188cf37caf24c43ec607f0a731b91b03da92a
Signed-off-by: Andrey Drobyshev <a.drobyshev@partner.samsung.com>
5 years agopackaging: fix testsuite build.
Andrey Drobyshev [Mon, 1 Oct 2018 13:35:47 +0000 (16:35 +0300)]
packaging: fix testsuite build.

 * Add list of dependent packages needed to build & run tests along with
corresponding macro run_tests;
 * Add %check section.

Signed-off-by: Andrey Drobyshev <a.drobyshev@partner.samsung.com>
Change-Id: I2cc277d6855247747025a4d3b744a5c7804f4fca

5 years agoRevert "Enable relro by default for AArch64, ARM and SCORE." submit/tizen_base/20181108.042134
Dongkyun Son [Thu, 8 Nov 2018 02:49:05 +0000 (11:49 +0900)]
Revert "Enable relro by default for AArch64, ARM and SCORE."

This reverts commit 12f75b84506d7820209fc3a919eac9d7991efd74.

IDLE memory size is increased by GNU_RELRO section.

Change-Id: Id8add9ecf47ecd23e6c8b0eccbb84fbce7ee3b2e

5 years agoEnable relro by default for AArch64, ARM and SCORE. submit/tizen_base/20181026.090949
Nick Clifton [Fri, 19 Aug 2016 08:27:52 +0000 (09:27 +0100)]
Enable relro by default for AArch64, ARM and SCORE.

ld * emultempl/aarch64elf.em (before_parse): Initialise the relro
field in the link_info structure.
* emultempl/armelf.em (before_parse): Likewise.
* emultempl/linux.em (before_parse): Likewise.
* emultempl/scoreelf.em (before_parse): Likewise.

Backported from upstream 576438f.

Signed-off-by: Andrey Drobyshev <a.drobyshev@partner.samsung.com>
Change-Id: I37738901650b0d254316bbbd8938beebd12418d1

6 years ago[TTC-15] i386/x86-64: Add unwind info for .plt.got section submit/tizen_5.0_base/20181101.000001 submit/tizen_base/20180129.030705
H.J. Lu [Fri, 19 Jan 2018 11:02:31 +0000 (14:02 +0300)]
[TTC-15] i386/x86-64: Add unwind info for .plt.got section

When there are both PLT and GOT references to the same function symbol,
linker combines GOTPLT and GOT slots into a single GOT slot and create
an entry in .plt.got section for PLT access via the GOT slot.  This
patch adds unwind info for .plt.got section.

bfd/

PR ld/20830
* elf32-i386.c (elf_i386_eh_frame_plt_got): New.
(PLT_GOT_FDE_LENGTH): Likewise.
(elf_i386_plt_layout): Add eh_frame_plt_got and
eh_frame_plt_got_size.
(elf_i386_plt): Updated.
(elf_i386_link_hash_table): Add plt_got_eh_frame.
(elf_i386_check_relocs): Create .eh_frame section for .plt.got.
(elf_i386_size_dynamic_sections): Allocate and initialize
.eh_frame section for .plt.got.
(elf_i386_finish_dynamic_sections): Adjust .eh_frame section for
.plt.got.
(elf_i386_nacl_plt): Add FIXME for eh_frame_plt_got and
eh_frame_plt_got_size.
* elf64-x86-64.c (elf_x86_64_eh_frame_plt_got): New.
(PLT_GOT_FDE_LENGTH): Likewise.
(elf_x86_64_backend_data): Add eh_frame_plt_got and
eh_frame_plt_got_size.
(elf_x86_64_arch_bed): Updated.
(elf_x86_64_bnd_arch_bed): Add FIXME for eh_frame_plt_got and
eh_frame_plt_got_size.
(elf_x86_64_nacl_arch_bed): Likewise.
(elf_x86_64_link_hash_table): Add plt_got_eh_frame.
(elf_x86_64_check_relocs): Create .eh_frame section for .plt.got.
(elf_x86_64_size_dynamic_sections): Allocate and initialize
.eh_frame section for .plt.got.
(elf_x86_64_finish_dynamic_sections): Adjust .eh_frame section
for .plt.got.

ld/

PR ld/20830
* testsuite/ld-i386/i386.exp: Run pr20830.
* testsuite/ld-x86-64/x86-64.exp: Likewise.
* testsuite/ld-i386/pr20830.d: New file.
* testsuite/ld-i386/pr20830.s: Likewise.
* testsuite/ld-x86-64/pr20830.d: Likewise.
* testsuite/ld-x86-64/pr20830.s: Likewise.

Backport from upstream: fff53daefb7838b5718422c87946330e4a8288ce
Signed-off-by: Slava Barinov <v.barinov@samsung.com>
Change-Id: I15ead2a145afaf5dff5b7fbc84e3a8f676f7618e

6 years agoAdd armv7hl support. submit/tizen_base/20171024.003248
Mikhail Kashkarov [Thu, 11 May 2017 09:14:51 +0000 (12:14 +0300)]
Add armv7hl support.

Change-Id: I1abc28bf90731946353b86579e85bd391c1745d3

6 years agoAdd ChangeLog.Tizen submit/tizen_base/20170830.083645
Sangmin Seo [Wed, 9 Aug 2017 10:07:59 +0000 (19:07 +0900)]
Add ChangeLog.Tizen

ChangeLog.Tizen is to keep track of private patches for Tizen binutils.

Change-Id: I1ca9709cbf7ef92f6d4b29ca7c71782c9541f94a

6 years agoASan build fix: resolve multiple definitions. submit/tizen_4.0_base/20170811.071500 submit/tizen_4.0_base/20170828.000000 submit/tizen_4.0_base/20170828.000001 submit/tizen_base/20170804.023302
Sangmin Seo [Tue, 11 Jul 2017 10:45:45 +0000 (19:45 +0900)]
ASan build fix: resolve multiple definitions.

Put "extern" to multiply-defined variables in order to make
-fno-common work.

Change-Id: Ib5a882e3c536d868e8098e385823e50ac78f2d21

6 years agoreadelf: Update check for invalid word offsets in ARM unwind information. submit/tizen_base/20170630.042341 submit/tizen_base/20170705.023439
Nick Clifton [Mon, 3 Apr 2017 10:01:45 +0000 (11:01 +0100)]
readelf: Update check for invalid word offsets in ARM unwind information.

PR binutils/21343
* readelf.c (get_unwind_section_word): Fix snafu checking for
invalid word offsets in ARM unwind information.

Cherry-picked f32ba72991d2406b21ab17edc234a2f3fa7fb23d

Change-Id: I8e0b7cb2447d915a7d084f91d7dadd6756f9f843
Signed-off-by: Sangmin Seo <sangmin7.seo@samsung.com>
6 years agoFix --enable-new-dtags config support submit/tizen_base/20170609.081358
Alan Modra [Mon, 5 Jun 2017 05:13:54 +0000 (14:43 +0930)]
Fix --enable-new-dtags config support

and testsuite when configured with --enable-new-dtags.

* ldmain.c (main): Correct setting of link_indo.new_dtags.
* testsuite/ld-elf/now-3.d: Pass --disable-new-dtags to ld
* testsuite/ld-elf/now-4.d: Likewise.
* testsuite/ld-elf/rpath-1.d: Likewise.
* testsuite/ld-elf/rpath-2.d: Likewise.

Change-Id: I6331b0a46be98386c8248bf1e126056be84972ad

6 years agoIntroduce --enable-new-dtags configure option.
Slava Barinov [Thu, 25 May 2017 08:10:06 +0000 (11:10 +0300)]
Introduce --enable-new-dtags configure option.

ld/configure --enable-new-dtags sets ld.bfd --enable-new-dtags option by
default.

The ld option switches on generation of DT_RUNPATH tag instead of DT_RPATH tag
which is deprecated and breaks LD_LIBRARY_PATH usage.

Debian binutils bug: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=835859

        * ld/configure.ac: Add --enable-new-dtags option.
        * ld/ldmain.c: Set link_info.new_dtags to 1 if when --enable-new-dtags is
          switched on.
        * ld/configure: Regenerate.
        * ld/config.in: Regenerate.
        * packaging/binutils.spec: Switch on --enable-new-dtags.
        * packaging/binutils-armv7l.spec: Regenerate.
        * packaging/binutils-aarch64.spec: Regenerate.

Change-Id: I0ec86f61a8b97d55c29bb441a9ae4db94060ea22
Signed-off-by: Slava Barinov <v.barinov@samsung.com>
7 years agoBackport fix for PR 17704 from upstream. submit/tizen_base/20170412.005421 tizen_4.0.m1_release
Gergely Nagy [Mon, 5 Dec 2016 14:42:16 +0000 (14:42 +0000)]
Backport fix for PR 17704 from upstream.

This fix keeps the section with the highest alignment when folding sections with ICF.

PR gold/17704
* icf.cc (match_sections): Add new parameter section_addraligns.
Check section alignment and keep the section with the strictest
alignment.
(find_identical_sections): New local variable section_addraligns.
Store each section's alignment.
* Makefile.in: Regenerate.
* testsuite/pr17704a_test.s: New file.
* testsuite/Makefile.am (pr17704a_test): New test.
* testsuite/Makefile.in: Regenerate.

Change-Id: I9b8b74cabefe5dd551bfa292e1e2df7aa9ea2e85
Signed-off-by: Pavel Kopyl <p.kopyl@samsung.com>
7 years ago2.27 version
Slava Barinov [Mon, 27 Feb 2017 11:29:01 +0000 (14:29 +0300)]
2.27 version

7 years agoIgnore broken plugins
Yury Usishchev [Wed, 15 Jul 2015 17:06:43 +0000 (20:06 +0300)]
Ignore broken plugins

Allow cross-configured ld to ignore plugins that cannot be dlopened

Change-Id: I0bbca32123aeba869d843fdfb9179815af888f5c
Signed-off-by: Yury Usishchev <y.usishchev@samsung.com>
7 years agoPackaging added.
Slava Barinov [Mon, 27 Feb 2017 09:57:34 +0000 (12:57 +0300)]
Packaging added.

Signed-off-by: Slava Barinov <v.barinov@samsung.com>
7 years agoRelease 2.27 binutils-2_27 upstream/2.27
Tristan Gingold [Wed, 3 Aug 2016 09:24:16 +0000 (11:24 +0200)]
Release 2.27

bfd/
2016-08-03  Tristan Gingold  <gingold@adacore.com>

* version.m4: Bump version to 2.27
* configure: Regenerate.

binutils/
2016-08-03  Tristan Gingold  <gingold@adacore.com>

* configure: Regenerate.

gas/
2016-08-03  Tristan Gingold  <gingold@adacore.com>

* configure: Regenerate.

gprof/
2016-08-03  Tristan Gingold  <gingold@adacore.com>

* configure: Regenerate.

ld/
2016-08-03  Tristan Gingold  <gingold@adacore.com>

* configure: Regenerate.

opcodes/
2016-08-03  Tristan Gingold  <gingold@adacore.com>

* configure: Regenerate.

7 years agoRegenerate pot files.
Tristan Gingold [Wed, 3 Aug 2016 08:24:46 +0000 (10:24 +0200)]
Regenerate pot files.

7 years agoPowerPC64 ld segfault with code in non-executable sections
Alan Modra [Tue, 2 Aug 2016 13:58:47 +0000 (23:28 +0930)]
PowerPC64 ld segfault with code in non-executable sections

PR ld/20428
* elf64-ppc.c (ppc_get_stub_entry): Don't segfault on NULL group.

7 years agoAutomatic date update in version.in
GDB Administrator [Wed, 3 Aug 2016 00:01:00 +0000 (00:01 +0000)]
Automatic date update in version.in

7 years agoAutomatic date update in version.in
GDB Administrator [Tue, 2 Aug 2016 00:01:09 +0000 (00:01 +0000)]
Automatic date update in version.in

7 years agoAutomatic date update in version.in
GDB Administrator [Mon, 1 Aug 2016 00:01:04 +0000 (00:01 +0000)]
Automatic date update in version.in

7 years agoAutomatic date update in version.in
GDB Administrator [Sun, 31 Jul 2016 00:00:56 +0000 (00:00 +0000)]
Automatic date update in version.in

7 years agoAutomatic date update in version.in
GDB Administrator [Sat, 30 Jul 2016 00:01:01 +0000 (00:01 +0000)]
Automatic date update in version.in

7 years agoAutomatic date update in version.in
GDB Administrator [Fri, 29 Jul 2016 00:01:02 +0000 (00:01 +0000)]
Automatic date update in version.in

7 years agoAutomatic date update in version.in
GDB Administrator [Thu, 28 Jul 2016 00:01:07 +0000 (00:01 +0000)]
Automatic date update in version.in

7 years agoFix seg fault in linker when performing garbage collection on COFF based targets.
Nick Clifton [Wed, 27 Jul 2016 12:18:13 +0000 (13:18 +0100)]
Fix seg fault in linker when performing garbage collection on COFF based targets.

PR ld/20401
bfd * coffgen.c (fini_reloc_cookie_rels): Check for the extistence
of the coff_section_data before using it.

7 years agoAutomatic date update in version.in
GDB Administrator [Wed, 27 Jul 2016 00:01:03 +0000 (00:01 +0000)]
Automatic date update in version.in

7 years agoSupport -pie for aarch64*-elf targets.
Roland McGrath [Tue, 19 Jul 2016 17:19:06 +0000 (10:19 -0700)]
Support -pie for aarch64*-elf targets.

ld/
* emulparams/aarch64elf.sh (GENERATE_PIE_SCRIPT): Set to yes.
* emulparams/aarch64elf32.sh: Likewise.

(cherry picked from commit 3af09e8db3084f5e03b7f09768a2ab6415418e63)

7 years agoExpect the 'objcopy without global symbols' test to fail for AArch64 and ARM targets.
Nick Clifton [Fri, 1 Jul 2016 09:35:53 +0000 (10:35 +0100)]
Expect the 'objcopy without global symbols' test to fail for AArch64 and ARM targets.

* testsuite/binutils-all/objcopy.exp
(objcopy_test_without_global_symbol): Expect this test to fail on
the AArch64 and ARM targets, since they preserve their mapping
symbols.

7 years agoBump version to 2.26.90 (on branch 2.27)
Tristan Gingold [Fri, 1 Jul 2016 09:14:55 +0000 (11:14 +0200)]
Bump version to 2.26.90 (on branch 2.27)

bfd/
2016-07-01  Tristan Gingold  <gingold@adacore.com>

* version.m4: Bump version to 2.26.90
* configure: Regenerate.

binutils/
2016-07-01  Tristan Gingold  <gingold@adacore.com>

* configure: Regenerate.

gas/
2016-07-01  Tristan Gingold  <gingold@adacore.com>

* configure: Regenerate.

gprof/
2016-07-01  Tristan Gingold  <gingold@adacore.com>

* configure: Regenerate.

ld/
2016-07-01  Tristan Gingold  <gingold@adacore.com>

* configure: Regenerate.

opcodes/
2016-07-01  Tristan Gingold  <gingold@adacore.com>

* configure: Regenerate.

7 years agoBump version to 2.27.51
Tristan Gingold [Fri, 1 Jul 2016 09:04:44 +0000 (11:04 +0200)]
Bump version to 2.27.51

bfd/
2016-07-01  Tristan Gingold  <gingold@adacore.com>

* version.m4: Bump version to 2.27.51
* configure: Regenerate.

binutils/
2016-07-01  Tristan Gingold  <gingold@adacore.com>

* configure: Regenerate.

gas/
2016-07-01  Tristan Gingold  <gingold@adacore.com>

* configure: Regenerate.

gprof/
2016-07-01  Tristan Gingold  <gingold@adacore.com>

* configure: Regenerate.

ld/
2016-07-01  Tristan Gingold  <gingold@adacore.com>

* configure: Regenerate.

opcodes/
2016-07-01  Tristan Gingold  <gingold@adacore.com>

* configure: Regenerate.

7 years agoAdd marker for 2.27 branch.
Tristan Gingold [Fri, 1 Jul 2016 08:57:46 +0000 (10:57 +0200)]
Add marker for 2.27 branch.

binutils/
2016-07-01  Tristan Gingold  <gingold@adacore.com>

* NEWS: Add marker for 2.27.

gas/
2016-07-01  Tristan Gingold  <gingold@adacore.com>

* NEWS: Add marker for 2.27.

ld/
2016-07-01  Tristan Gingold  <gingold@adacore.com>

* NEWS: Add marker for 2.27.

7 years agoFix mis-placement in binutils.texi
Tristan Gingold [Wed, 29 Jun 2016 10:09:06 +0000 (12:09 +0200)]
Fix mis-placement in binutils.texi

binutils/
* doc/binutils.texi (objdump): Fix mis-placement.

7 years agox86-64/MPX: bndmk, bndldx, and bndstx don't allow RIP-relative addressing
Jan Beulich [Fri, 1 Jul 2016 07:07:15 +0000 (09:07 +0200)]
x86-64/MPX: bndmk, bndldx, and bndstx don't allow RIP-relative addressing

Additionally warn about scaling factors other than 1 for the latter
two, as those get ignored by the hardware.

7 years agox86/MPX: fix address size handling
Jan Beulich [Fri, 1 Jul 2016 07:06:16 +0000 (09:06 +0200)]
x86/MPX: fix address size handling

While address overrides are ignored in 64-bit mode (and hence shouldn't
really result in an error, but upon v1 converting this to a warning I
was told otherwise), trying to use 16-bit addressing is documented to
result in #UD, and hence the assembler should reject the attempt. (The
added test case at once also checks that bndc{l,n,u} won't accept
16-bit register operands.)

7 years agox86/Intel: don't accept bogus instructions
Jan Beulich [Fri, 1 Jul 2016 07:03:02 +0000 (09:03 +0200)]
x86/Intel: don't accept bogus instructions

... due to their last byte looking like a suffix, when after its
stripping a matching instruction can be found. Since memory operand
size specifiers in Intel mode get converted into suffix representation
internally, we need to keep track of the actual mnemonic suffix which
may have got trimmed off, and check its validity while looking for a
matching template. I tripper over this quite some time again after
support for AMD's SSE5 instructions got removed, as at that point some
of the SSE5 mnemonics, other than expected, didn't fail to assemble.
But the problem affects many more instructions, namely (almost) all
MMX, SSE, and AVX ones as it looks. I don't think it makes sense to
add a testcase covering all of them, nor do I think it makes sense to
pick out some random examples for a new test case.

7 years agox86: allow suffix-less movzw and 64-bit movzb
Jan Beulich [Fri, 1 Jul 2016 07:01:41 +0000 (09:01 +0200)]
x86: allow suffix-less movzw and 64-bit movzb

... just like is already the case for 16- and 32-bit movzb: I can't see
why omitting suffixes on this (and movs{b,w,l}) is not allowed, when it
is allowed for all other instructions where the suffix is redundant
with (one of) the operands.

7 years agox86: remove stray instruction attributes
Jan Beulich [Fri, 1 Jul 2016 06:57:21 +0000 (08:57 +0200)]
x86: remove stray instruction attributes

- with Cpu64 Disp16 makes no sense for memory operands
- with CpuNo64 Disp32S makes no sense
- non-64-bit lgdt doesn't allow 10-byte operands

7 years agox86/Intel: fix operand checking for MOVSD
Jan Beulich [Fri, 1 Jul 2016 06:56:13 +0000 (08:56 +0200)]
x86/Intel: fix operand checking for MOVSD

The dual purpose mnemonic (string move vs scalar double move) breaks
the assumption that the isstring flag would be set on both the first
and last entry in the current set of templates, which results in bogus
or missing diagnostics for the string move variant of the mnemonic.
Short of mostly rewriting i386_index_check() and its interaction with
the rest of the code, simply shrink the template set to just string
instructions when encountering the second memory operand, and run
i386_index_check() a second time for the first memory operand after
that reduction.

7 years agoAutomatic date update in version.in
GDB Administrator [Fri, 1 Jul 2016 00:00:22 +0000 (00:00 +0000)]
Automatic date update in version.in

7 years agoMIPS/GAS: Fix a comment typo in `get_append_method'
Maciej W. Rozycki [Thu, 30 Jun 2016 20:49:54 +0000 (21:49 +0100)]
MIPS/GAS: Fix a comment typo in `get_append_method'

gas/
* config/tc-mips.c (get_append_method): Fix a comment typo.

7 years agoChangeLog entry for the --with-cpu patch for ARC configuration.
Andrew Burgess [Thu, 30 Jun 2016 15:53:20 +0000 (16:53 +0100)]
ChangeLog entry for the --with-cpu patch for ARC configuration.

7 years agoFix typo in comment
Yao Qi [Thu, 30 Jun 2016 15:03:07 +0000 (16:03 +0100)]
Fix typo in comment

This patch fixes the typo "uf" in the comment.  I'll push it in as the
change is obvious.

2016-06-30  Yao Qi  <yao.qi@linaro.org>

* arm-dis.c (print_insn): Fix typo in comment.

7 years agoMIPS16/GAS: Fix delay slot filling across frags
Matthew Fortune [Thu, 30 Jun 2016 14:02:20 +0000 (15:02 +0100)]
MIPS16/GAS: Fix delay slot filling across frags

Fix an assertion failure like:

test.s: Assembler messages:
test.s:3: Internal error!
Assertion failure in append_insn at .../gas/config/tc-mips.c:7523.
Please report this bug.

triggered by assembling MIPS16 code like:

hello:
addiu $4, $4, 4
jr $31

with the generation of a listing file enabled, e.g.:

$ as -mips16 -O2 -aln=test.lst

The cause of the problem is the lack of support for moving instructions
across frags in MIPS16 jump swapping, which triggers more easily with
listing enabled as in that case every instruction gets placed in its own
frag.  It would trigger even with listing disabled though if the
instruction to swap a MIPS16 jump with was unfortunately enough placed
as last in a frag that became full.

This scenario is already handled correctly with branch swapping in
regular MIPS and microMIPS code, so reuse it for MIPS16 code as well,
and now that all MIPS16 handling has become the same as the regular MIPS
and microMIPS cases remove MIPS16 special casing altogether.

This effectively complements:

commit 464ab0e55ade01d2bb0b4fa45c429af7a2f85a26
Author: Maciej W. Rozycki <macro@linux-mips.org>
Date:   Mon Aug 6 20:33:00 2012 +0000

<https://sourceware.org/ml/binutils/2012-08/msg00043.html>, ("MIPS/GAS:
Correct microMIPS branch swapping assertion") for the MIPS16 case.

The assertion itself was introduced with:

commit 1e91584932efd70020c8c98037d0cb93a0552a20
Author: Richard Sandiford <rdsandiford@googlemail.com>
Date:   Wed Mar 9 09:17:02 2005 +0000

<https://sourceware.org/ml/binutils/2005-03/msg00217.html>, ("Rework
MIPS nop-insertion code, add -mfix-vr4130 [5/11]"), but its introduction
merely noted our existing lack of support for MIPS16 jump swapping
across frags.

gas/
* config/tc-mips.c (append_insn) <APPEND_SWAP>: Do not special
case MIPS16 handling.
* testsuite/gas/mips/branch-swap-3.d: New test.
* testsuite/gas/mips/branch-swap-4.d: New test.
* testsuite/gas/mips/mips16@branch-swap-3.d: New test.
* testsuite/gas/mips/mips16@branch-swap-4.d: New test.
* testsuite/gas/mips/micromips@branch-swap-3.d: New test.
* testsuite/gas/mips/micromips@branch-swap-4.d: New test.
* testsuite/gas/mips/branch-swap-3.s: New test source.
* testsuite/gas/mips/mips.exp: Run the new tests.

7 years agoMIPS/GAS: Simplify non-MIPS16 branch swapping sequence
Maciej W. Rozycki [Thu, 30 Jun 2016 11:46:47 +0000 (12:46 +0100)]
MIPS/GAS: Simplify non-MIPS16 branch swapping sequence

Simplify non-MIPS16 branch swapping by copying the MIPS16 variant, which
sets the new position for the current instruction first and reduces the
calculation of the new position of the previous instruction.  Also refer
to previous instruction's frag and position via `delay' for consistency.

Reintroduce an explanatory comment, updated, previously removed with:

commit 1e91584932efd70020c8c98037d0cb93a0552a20
Author: Richard Sandiford <rdsandiford@googlemail.com>
Date:   Wed Mar 9 09:17:02 2005 +0000

<https://sourceware.org/ml/binutils/2005-03/msg00217.html>, ("Rework
MIPS nop-insertion code, add -mfix-vr4130 [5/11]").

gas/
* config/tc-mips.c (append_insn): Simplify non-MIPS16 branch
swapping sequence.

7 years agoPR gas/20312: Do not pad sections to alignment on failed assembly
Maciej W. Rozycki [Wed, 29 Jun 2016 00:38:50 +0000 (01:38 +0100)]
PR gas/20312: Do not pad sections to alignment on failed assembly

Correct a regression from commit 85024cd8bcb9 ("Run write_object_file
after errors") causing unsuccessful assembly, which may be due to any
reason, such as supplying a valid source like this:

.text
.byte 0
.err

to terminate with an assertion failure like:

test.s: Assembler messages:
test.s:3: Error: .err encountered
../as-new: BFD (GNU Binutils) 2.24.51.20140628 internal error, aborting at .../gas/write.c line 608 in size_seg
../as-new: Please report this bug.

on targets whose default text section alignment is above 0, typically
RISC machines.

This is due to an attempt to set last text section's frag alignment to
0, requested from `subsegs_finish_section' where `frag_align_code
(alignment, 0)' is called with `alignment' set to 0 rather than the
section alignment if `had_errors' has returned true.  The call to
`subsegs_finish_section' is made from `subsegs_finish' from
`write_object_file' at unsuccessful completion, which previously wasn't
made.

Always set last section's frag alignment from the section alignment
then, forcing no section padding instead if completing unsuccessfully,
so that in that case alignment padding is still suppressed from any
listing generated, fixing assertion failures for these targets:

alpha-linuxecoff  -FAIL: all pr20312
arm-aout  -FAIL: all pr20312
mips-freebsd  -FAIL: all pr20312
mips-img-linux  -FAIL: all pr20312
mips-linux  -FAIL: all pr20312
mips-mti-linux  -FAIL: all pr20312
mips-netbsd  -FAIL: all pr20312
mips-sgi-irix5  -FAIL: all pr20312
mips-sgi-irix6  -FAIL: all pr20312
mips-vxworks  -FAIL: all pr20312
mips64-freebsd  -FAIL: all pr20312
mips64-img-linux  -FAIL: all pr20312
mips64-linux  -FAIL: all pr20312
mips64-mti-linux  -FAIL: all pr20312
mips64-openbsd  -FAIL: all pr20312
mips64el-freebsd  -FAIL: all pr20312
mips64el-img-linux  -FAIL: all pr20312
mips64el-linux  -FAIL: all pr20312
mips64el-mti-linux  -FAIL: all pr20312
mips64el-openbsd  -FAIL: all pr20312
mipsel-freebsd  -FAIL: all pr20312
mipsel-img-linux  -FAIL: all pr20312
mipsel-linux  -FAIL: all pr20312
mipsel-mti-linux  -FAIL: all pr20312
mipsel-netbsd  -FAIL: all pr20312
mipsel-vxworks  -FAIL: all pr20312
mipsisa32-linux  -FAIL: all pr20312
mipsisa32el-linux  -FAIL: all pr20312
mipsisa64-linux  -FAIL: all pr20312
mipsisa64el-linux  -FAIL: all pr20312
sh-pe  -FAIL: all pr20312
sparc-aout  -FAIL: all pr20312

gas/
PR gas/20312
* write.c (subsegs_finish_section): Force no section padding to
alignment on failed assembly, always set last frag's alignment
from section.
* testsuite/gas/all/pr20312.l: New list test.
* testsuite/gas/all/pr20312.s: New test source.
* testsuite/gas/all/gas.exp: Run the new test

7 years agoFix gdbserver/MI testing regression
Pedro Alves [Thu, 30 Jun 2016 10:55:22 +0000 (11:55 +0100)]
Fix gdbserver/MI testing regression

Commit 51f77c3704a6 ("Add testing infrastruture bits for running with
MI on a separate UI") broke MI testing with native-gdbserver:

 $ make check RUNTESTFLAGS="--target_board=native-gdbserver mi-var-child.exp"
 ...
 Running .../src/binutils-gdb/gdb/testsuite/gdb.mi/mi-var-child.exp ...
 can't unset "inferior_spawn_id": no such variable
     while executing
 "unset inferior_spawn_id"
     (procedure "close_gdbserver" line 20)
     invoked from within
 "close_gdbserver"
 ...

When testing with gdbserver, gdb_exit is overridden with a special
version that calls close_gdbserver, which clears inferior_spawn_id.
The problem is that the commit mentioned above made
gdb_exit/mi_gdb_exit clear inferior_spawn_id too, and clearing a
non-existing variable is a tcl error.

Since gdb_exit/mi_gdb_exit always clears inferior_spawn_id now, the
fix is simply to stop clearing it in close_gdbserver.

gdb/testsuite/
2016-06-30  Pedro Alves  <palves@redhat.com>

* lib/gdbserver-support.exp (close_gdbserver, gdb_exit): Don't
unset inferior_spawn_id.

7 years agoMake testing gdb with FORCE_SEPARATE_MI_TTY=1 actually work
Pedro Alves [Thu, 30 Jun 2016 10:55:21 +0000 (11:55 +0100)]
Make testing gdb with FORCE_SEPARATE_MI_TTY=1 actually work

Runing the whole gdb testsuite with MI on a separate tty, with:

  make check RUNTESTFLAGS="FORCE_SEPARATE_MI_TTY=1"

Doesn't actually work because commit 51f77c3704a6 ("Add testing
infrastruture bits for running with MI on a separate UI") included a
last-minute rename typo, now fixed with this commit.

gdb/testsuite/ChangeLog:
2016-06-30  Pedro Alves  <palves@redhat.com>

* lib/mi-support.exp (default_mi_gdb_start): Declare global
FORCE_SEPARATE_MI_TTY, not SEPARATE_MI_TTY.

7 years agoAllow ARC target to be configured with --with-cpu=<cpu-name>.
Andrew Burgess [Thu, 30 Jun 2016 10:14:41 +0000 (11:14 +0100)]
Allow ARC target to be configured with --with-cpu=<cpu-name>.

gas * config.in (TARGET_WITH_CPU): Undefine.
* configure.ac: Add --with-cpu support, and define in config.h.
* configure: Regenerate.
* config/tc-arc.c: Use TARGET_WITH_CPU to select default CPU.
* NEWS: Mention new configure option.

7 years ago[ARM][GAS] ARMv8.2 should enable ARMv8.1 NEON instructions.
Matthew Wahab [Thu, 30 Jun 2016 09:46:51 +0000 (10:46 +0100)]
[ARM][GAS] ARMv8.2 should enable ARMv8.1 NEON instructions.

GAS fails to recognize march=armv8.2-a as a superset of march=armv8.1-a
when assembling NEON instructions. The patch corrects this, making
-march=armv8.2-a -mfpu=neon-fp-armv8 enable the NEON intructions
introduced with ARMv8.1-A.

include/
2016-06-30  Matthew Wahab  <matthew.wahab@arm.com>

* opcode/arm.h (ARM_ARCH_V8_2a): Add FPU_NEON_EXT_RDMA to the set
of enabled FPU features.

gas/
2016-06-30  Matthew Wahab  <matthew.wahab@arm.com>

* testsuite/gas/arm/armv8_2+rdma.d: New.

7 years agoAdd support for simulating big-endian AArch64 binaries.
Jim Wilson [Thu, 30 Jun 2016 08:10:41 +0000 (09:10 +0100)]
Add support for simulating big-endian AArch64 binaries.

* cpustate.h: Include config.h.
(union GRegisterValue): Add WORDS_BIGENDIAN check.  For big endian code
use anonymous structs to align members.
* simulator.c (aarch64_step): Use sim_core_read_buffer and
endian_le2h_4 to read instruction from pc.

7 years agoFix gold testsuite failure with GCC 6.
Cary Coutant [Thu, 30 Jun 2016 06:24:35 +0000 (23:24 -0700)]
Fix gold testsuite failure with GCC 6.

With GCC 6 when not using -static-libstdc++, the operator delete(void*)
function is defined in the shared C++ support library, rather than in
the main program. The test script is too aggressive in checking for
this symbol's presence among the exported symbols. This patch removes
the check for that symbol.

gold/
PR gold/20310
* testsuite/dynamic_list.sh: Remove check for _ZdlPv.

7 years agoUpdate "make clean" in gold/testsuite.
Cary Coutant [Thu, 30 Jun 2016 06:16:24 +0000 (23:16 -0700)]
Update "make clean" in gold/testsuite.

gold/
* testsuite/Makefile.am (MOSTLYCLEANFILES): Add eh_test_2.
* testsuite/Makefile.in: Regenerate.

7 years ago[GOLD] Pass -Wl,-z to gcc, not plain -z
Alan Modra [Wed, 29 Jun 2016 09:07:27 +0000 (18:37 +0930)]
[GOLD] Pass -Wl,-z to gcc, not plain -z

* testsuite/Makefile.am (memory_test, memory_test_2): Pass
-Wl,-z to gcc, not plain -z.
* testsuite/Makefile.in: Regenerate.

7 years agoAutomatic date update in version.in
GDB Administrator [Thu, 30 Jun 2016 00:00:19 +0000 (00:00 +0000)]
Automatic date update in version.in

7 years agoAdd copyright header in gdb.base/return.c
Yao Qi [Wed, 29 Jun 2016 16:33:19 +0000 (17:33 +0100)]
Add copyright header in gdb.base/return.c

gdb/testsuite:

2016-06-29  Yao Qi  <yao.qi@linaro.org>

* gdb.base/return.c: Add copyright header.

7 years agoDefault to --enable-compressed-debug-sections=gas for Linux/x86
H.J. Lu [Wed, 29 Jun 2016 16:29:39 +0000 (09:29 -0700)]
Default to --enable-compressed-debug-sections=gas for Linux/x86

--enable-compressed-debug-sections=gas added to binutils 2.26.  Make it
default for Linux/x86 targets in 2.27.

* NEWS: Mention --enable-compressed-debug-sections=gas is the
default for Linux/x86 targets.
* configure.tgt (ac_default_compressed_debug_sections): Default
to yes for Linux/x86 targets.

7 years agoFix PR python/20129 - use of non-existing variable
Tom Tromey [Mon, 6 Jun 2016 17:21:15 +0000 (11:21 -0600)]
Fix PR python/20129 - use of non-existing variable

PR python/20129 concerns the error message one gets from a command
like "disable frame-filter global NoSuchFilter".  Currently this
throws a second, unexpected, exception due to the use of a
non-existing variable named "name".

This patch adds regression tests and fixes a couple of spots to use
the correct variable name.

Built and regtested on x86-64 Fedora 23.

2016-06-29  Tom Tromey  <tom@tromey.com>

PR python/20129:
* python/lib/gdb/command/frame_filters.py (_do_enable_frame_filter)
(SetFrameFilterPriority._set_filter_priority): Use "frame_filter",
not "name".

2016-06-29  Tom Tromey  <tom@tromey.com>

PR python/20129:
* gdb.python/py-framefilter.exp: Add tests for setting priority
and disabling of non-existent frame filter.

7 years agoPR gdb/17210 - fix possible memory leak in read_memory_robust
Tom Tromey [Thu, 9 Jun 2016 15:46:53 +0000 (09:46 -0600)]
PR gdb/17210 - fix possible memory leak in read_memory_robust

PR gdb/17210 concerns a possible memory leak in read_memory_robust.
The bug can happen because read_memory_robust allocates memory, does
not install any cleanups, and invokes QUIT.  Similarly, target_read
calls QUIT, so it too can potentially throw.

The fix is to install cleanups to guard the allocated memory.

Built and regtested on x86-64 Fedora 23.  I couldn't think of a way to
test this, so no new test; and of course this means it should have
more careful review.

2016-06-29  Tom Tromey  <tom@tromey.com>

PR gdb/17210:
* target.c (free_memory_read_result_vector): Take a pointer to the
VEC as an argument.
(read_memory_robust): Install a cleanup for "result".
* mi/mi-main.c (mi_cmd_data_read_memory_bytes): Update.

7 years agogold: Support 386 TLS code sequences without PLT
H.J. Lu [Wed, 29 Jun 2016 15:38:43 +0000 (08:38 -0700)]
gold: Support 386 TLS code sequences without PLT

There are extensions to 386 psABI:

https://groups.google.com/forum/#!topic/ia32-abi/awsRSvJOJfs

to call tls_get_addr via GOT:

call *___tls_get_addr@GOT(%reg)

where EBX register isn't required as GOT base.

Since direct call is 4-byte long and indirect call, is 5-byte long, the
extra one byte must be handled properly.

For general dynamic model, 7-byte lea instruction before call
instruction is replaced by 6-byte one to make room for indirect call.
For local dynamic model, we simply use 5-byte indirect call.

TLS linker optimization is updated to recognize new instruction
patterns.  For local dynamic model to local exec model transition,
we generate a 6-byte lea instruction as nop, instead of a 1-byte nop
plus a 4-byte lea instruction.

PR gold/20308
* i386.cc (Target_i386::Relocate::relocate): Allow
R_386_GOT32X relocation against ___tls_get_addr.
(Target_i386::Relocate::tls_gd_to_ie): Support indirect
call to __tls_get_addr.
(Target_i386::Relocate::tls_gd_to_le): Likewise.
(Target_i386::Relocate::tls_ld_to_le): Likewise.
* testsuite/Makefile.am (check_PROGRAMS): Add pr20308a_test,
pr20308b_test, pr20308c_test, pr20308d_test, pr20308e_test.
(pr20308a_test_SOURCES): New.
(pr20308a_test_DEPENDENCIES): Likewise.
(pr20308a_test_CFLAGS): Likewise.
(pr20308a_test_LDFLAGS): Likewise.
(pr20308a_test_LDADD): Likewise.
(pr20308b_test_SOURCES): Likewise.
(pr20308b_test_DEPENDENCIES): Likewise.
(pr20308b_test_CFLAGS): Likewise.
(pr20308b_test_LDFLAGS): Likewise.
(pr20308b_test_LDADD): Likewise.
(pr20308c_test_SOURCES): Likewise.
(pr20308c_test_DEPENDENCIES): Likewise.
(pr20308c_test_CFLAGS): Likewise.
(pr20308c_test_LDFLAGS): Likewise.
(pr20308c_test_LDADD): Likewise.
(pr20308d_test_SOURCES): Likewise.
(pr20308d_test_DEPENDENCIES): Likewise.
(pr20308d_test_CFLAGS): Likewise.
(pr20308d_test_LDFLAGS): Likewise.
(pr20308d_test_LDADD): Likewise.
(pr20308e_test_SOURCES): Likewise.
(pr20308e_test_DEPENDENCIES): Likewise.
(pr20308e_test_CFLAGS): Likewise.
(pr20308e_test_LDFLAGS): Likewise.
(pr20308e_test_LDADD): Likewise.
(pr20308a.so): Likewise.
(pr20308b.so): Likewise.
(pr20308_gd.o): Likewise.
(pr20308_ld.o): Likewise.
(MOSTLYCLEANFILES): Add pr20308a.so pr20308b.so.
* testsuite/Makefile.in: Regenerated.
* testsuite/pr20308_def.c: New file.
* testsuite/pr20308_gd.S: Likewise.
* testsuite/pr20308_ld.S: Likewise.
* testsuite/pr20308_main.c: Likewise.

7 years agogold: Support x86-64 TLS code sequences without PLT
H.J. Lu [Wed, 29 Jun 2016 15:37:30 +0000 (08:37 -0700)]
gold: Support x86-64 TLS code sequences without PLT

There are extensions to x86-64 psABI:

https://groups.google.com/forum/#!topic/x86-64-abi/de5_KnLHxtI

to call tls_get_addr via GOT:

call *__tls_get_addr@GOTPCREL(%rip)

Since direct call is 4-byte long and indirect call, is 5-byte long, the
extra one byte must be handled properly.

For general dynamic model, one 0x66 prefix before call instruction is
removed to make room for indirect call.  For local dynamic model, we
simply use 5-byte indirect call.

TLS linker optimization is updated to recognize new instruction
patterns.  For local dynamic model to local exec model transition, we
generate 4 0x66 prefixes, instead of 3, before mov instruction in 64-bit
and generate a 5-byte nop, instead of 4-byte, before mov instruction in
32-bit.

PR gold/20216
* configure.ac (DEFAULT_TARGET_X86_64_OR_X32): New
AM_CONDITIONAL.
* configure: Regenerated.
* x86_64.cc (Target_x86_64<size>::Relocate::relocate): Allow
R_X86_64_GOTPCRELX relocation against __tls_get_addr.
(Target_x86_64<size>::Relocate::tls_gd_to_ie): Support indirect
call to __tls_get_addr.
(Target_x86_64<size>::Relocate::tls_gd_to_le): Likewise.
(Target_x86_64<size>::Relocate::tls_ld_to_le): Likewise.
* testsuite/Makefile.am (check_PROGRAMS): Add pr20216a_test,
pr20216b_test, pr20216c_test, pr20216d_test, pr20216e_test.
(pr20216a_test_SOURCES): New.
(pr20216a_test_DEPENDENCIES): Likewise.
(pr20216a_test_CFLAGS): Likewise.
(pr20216a_test_LDFLAGS): Likewise.
(pr20216a_test_LDADD): Likewise.
(pr20216b_test_SOURCES): Likewise.
(pr20216b_test_DEPENDENCIES): Likewise.
(pr20216b_test_CFLAGS): Likewise.
(pr20216b_test_LDFLAGS): Likewise.
(pr20216b_test_LDADD): Likewise.
(pr20216c_test_SOURCES): Likewise.
(pr20216c_test_DEPENDENCIES): Likewise.
(pr20216c_test_CFLAGS): Likewise.
(pr20216c_test_LDFLAGS): Likewise.
(pr20216c_test_LDADD): Likewise.
(pr20216d_test_SOURCES): Likewise.
(pr20216d_test_DEPENDENCIES): Likewise.
(pr20216d_test_CFLAGS): Likewise.
(pr20216d_test_LDFLAGS): Likewise.
(pr20216d_test_LDADD): Likewise.
(pr20216e_test_SOURCES): Likewise.
(pr20216e_test_DEPENDENCIES): Likewise.
(pr20216e_test_CFLAGS): Likewise.
(pr20216e_test_LDFLAGS): Likewise.
(pr20216e_test_LDADD): Likewise.
(pr20216a.so): Likewise.
(pr20216b.so): Likewise.
(pr20216_gd.o): Likewise.
(pr20216_ld.o): Likewise.
(MOSTLYCLEANFILES): Add pr20216a.so pr20216b.so.
* testsuite/Makefile.in: Regenerated.
* testsuite/pr20216_def.c: New file.
* testsuite/pr20216_gd.S: Likewise.
* testsuite/pr20216_ld.S: Likewise.
* testsuite/pr20216_main.c: Likewise.

7 years agoInitialize strtok_r's saveptr to NULL
Manish Goregaokar [Wed, 29 Jun 2016 11:39:43 +0000 (17:09 +0530)]
Initialize strtok_r's saveptr to NULL

Building gdb with --enable-build-with-cxx=no trips on a warning:

 ../../binutils-gdb/gdb/rust-lang.c:173:15: error: saveptr may be used
 uninitialized in this function [-Werror=maybe-uninitialized]
     ret.name = concat (TYPE_NAME (type), "::", token, (char *) NULL);

The problem is that gcc doesn't understand that "tail" can never be
NULL in the call to strtok_r:

      name = xstrdup (TYPE_FIELD_NAME (type, 0));
      cleanup = make_cleanup (xfree, name);
      tail = name + strlen (RUST_ENUM_PREFIX);
...
      for (token = strtok_r (tail, "$", &saveptr);

Fix this by always initializing saveptr.

2016-06-29  Manish Goregaokar  <manish@mozilla.com>

gdb/ChangeLog:
    * rust-lang.c (rust_get_disr_info): Initialize saveptr to NULL.

7 years agoSet unknown_syscall differently on arm linux
Yao Qi [Wed, 29 Jun 2016 13:51:41 +0000 (14:51 +0100)]
Set unknown_syscall differently on arm linux

Currently, we use 123456789 as unknown or illegal syscall number, and
expect program return ENOSYS.  Although 123456789 is an illegal syscall
number on arm linux, kernel sends SIGILL rather than returns -ENOSYS.
However, arm linux kernel returns -ENOSYS if syscall number is within
0xf0001..0xf07ff, so we can use 0xf07ff for unknown_syscall in test.

gdb/testsuite:

2016-06-29  Yao Qi  <yao.qi@linaro.org>

* gdb.base/catch-syscall.c [__arm__]: Set unknown_syscall to
0x0f07ff.

7 years agosparc: make SPARC_OPCODE_ARCH_MAX part of its enum
Trevor Saunders [Fri, 27 May 2016 08:52:45 +0000 (04:52 -0400)]
sparc: make SPARC_OPCODE_ARCH_MAX part of its enum

include/ChangeLog:

2016-06-29  Trevor Saunders  <tbsaunde+binutils@tbsaunde.org>

* opcode/sparc.h (enum sparc_opcode_arch_val): Move
SPARC_OPCODE_ARCH_MAX into the enum.

7 years agoUse strtok_r instead of strsep in rust_get_disr_info
Manish Goregaokar [Wed, 29 Jun 2016 10:12:28 +0000 (15:42 +0530)]
Use strtok_r instead of strsep in rust_get_disr_info

strsep doesn't exist on Windows.

2016-06-29  Manish Goregaokar  <manish@mozilla.com>

gdb/ChangeLog:
    * rust-lang.c (rust_get_disr_info): Use strtok_r instead of strsep.

7 years agoPreserve all mapping symbols in ARM and AArch64 object files.
Nick Clifton [Wed, 29 Jun 2016 10:17:40 +0000 (11:17 +0100)]
Preserve all mapping symbols in ARM and AArch64 object files.

bfd * elfnn-aarch64.c (is_aarch64_mapping_symbol): New function.
Returns TRUE for AArch64 mapping symbols.
(elfNN_aarch64_backend_symbol_processing): New function.  Marks
mapping symbols as precious in object files so that they will not
be stripped.
(elf_backend_symbol_processing): Define.

* elf32-arm.c (is_arm_mapping_symbol): New function.  Returns TRUE
for ARM mapping symbols.
(elf32_arm_backend_symbol_processing): Make use of the new function.

7 years agoCorrect fix for typo
Nick Clifton [Wed, 29 Jun 2016 08:09:03 +0000 (09:09 +0100)]
Correct fix for typo

7 years agoFix typo
Nick Clifton [Wed, 29 Jun 2016 08:06:55 +0000 (09:06 +0100)]
Fix typo

7 years agoGAS: Fix `abort' expansion in write.c
Maciej W. Rozycki [Tue, 28 Jun 2016 16:33:38 +0000 (17:33 +0100)]
GAS: Fix `abort' expansion in write.c

Remove an internal diagnostic regression introduced with the inclusion of
"libbfd.h" from write.c, added with:

commit e7ff5c732e7b95aafccd0910ea1a5cb8251a1033
Author: Alan Modra <amodra@gmail.com>
Date:   Fri Feb 16 03:40:17 2007 +0000

That change made "libbfd.h" override the `abort' definition provided by
"as.h" earlier on, making the message produced by any calls reached from
write.c, which is a part of the GAS proper, look like they came from
BFD, e.g.:

.../gas/testsuite/gas/elf/type.s: Assembler messages:
.../gas/testsuite/gas/elf/type.s:30: Error: symbol type "gnu_unique_object" is supported only by GNU targets
../as-new: BFD (GNU Binutils) 2.26.51.20160628 internal error, aborting at .../gas/write.c:608 in size_seg

../as-new: Please report this bug.

vs:

.../gas/testsuite/gas/elf/type.s: Assembler messages:
.../gas/testsuite/gas/elf/type.s:30: Error: symbol type "gnu_unique_object" is supported only by GNU targets
.../gas/testsuite/gas/elf/type.s: Internal error, aborting at .../gas/write.c:602 in size_seg
Please report this bug.

With the removal of "libbfd.h" restore the latter message format.

gas/
* write.c: Remove "libbfd.h" inclusion.

7 years agoAutomatic date update in version.in
GDB Administrator [Wed, 29 Jun 2016 00:00:26 +0000 (00:00 +0000)]
Automatic date update in version.in

7 years ago[GOLD] Modify script_test_12
Alan Modra [Tue, 28 Jun 2016 22:30:58 +0000 (08:00 +0930)]
[GOLD] Modify script_test_12

Changes necessary to make this test pass on powerpc64.

* script_test_12.t: Delete .plt, specify 64k page size.
* script_test_12i.t: Likewise.

7 years ago[GOLD] alignment test
Alan Modra [Tue, 28 Jun 2016 22:30:15 +0000 (08:00 +0930)]
[GOLD] alignment test

PowerPC gcc aligns char arrays to a word which results in the test
failing since it expects a char alignment.  As the test already uses
gcc attributes to assign variables to sections, we may as well use an
attribute to align too.

* testsuite/plugin_layout_with_alignment.c: Explicitly align all
variables.

7 years ago[GOLD] Disable copy_test_protected on powerpc
Alan Modra [Tue, 28 Jun 2016 22:29:34 +0000 (07:59 +0930)]
[GOLD] Disable copy_test_protected on powerpc

A target like powerpc64 that is PIC by default doesn't need copy relocs.

* testsuite/Makefile.am (copy_test_protected): Disable for powerpc.
* testsuite/Makefile.in: Regenerate.

7 years agoFix execute permission on gold/testsuite/script_test_14.sh.
Cary Coutant [Tue, 28 Jun 2016 22:44:22 +0000 (15:44 -0700)]
Fix execute permission on gold/testsuite/script_test_14.sh.

7 years agoImplement the R_AARCH64_NONE relocation.
Igor Kudrin [Tue, 28 Jun 2016 22:43:43 +0000 (15:43 -0700)]
Implement the R_AARCH64_NONE relocation.

According to "ELF for the ARM(r) 64-bit Architecture (AArch64)",
this relocation can be used "to prevent removal of sections that
might otherwise appear to be unused."

gold/
* aarch64-reloc.def (NONE): New relocation.
* aarch64.cc (Target_aarch64::Scan::local): Handle R_AARCH64_NONE.
(Target_aarch64::Scan::global): Likewise.
* testsuite/Makefile.am (aarch64_reloc_none): New test.
* testsuite/Makefile.in: Regenerate.
* testsuite/aarch64_reloc_none.s: New test source file.
* testsuite/aarch64_reloc_none.sh: New test script.

7 years agoConvert indirect calls to direct when possible.
Sriraman Tallam [Tue, 28 Jun 2016 22:42:33 +0000 (15:42 -0700)]
Convert indirect calls to direct when possible.

Please see patch discussion:
https://www.sourceware.org/ml/binutils/2016-05/msg00322.html

2016-06-28  Sriraman Tallam  <tmsriram@google.com>

* x86_64.cc (Lazy_view): New class.
(can_convert_mov_to_lea): Templatize function.  Make the function
check for appropriate relocation types and use the view parameter
to get section contents.
(can_convert_callq_to_direct): New function.
(Target_x86_64<size>::Scan::global): Refactor.
(Target_x86_64<size>::Relocate::relocate): Refactor. Change any indirect
call via GOT that can be converted.
* testsuite/Makefile.am (x86_64_indirect_call_to_direct.sh): New test.
* testsuite/Makefile.in: Regenerate.
* testsuite/x86_64_indirect_call_to_direct1.s: New file.
* testsuite/x86_64_indirect_jump_to_direct1.s: New file.

7 years agoImplement SORT_BY_INIT_PRIORITY.
Igor Kudrin [Tue, 28 Jun 2016 22:34:11 +0000 (15:34 -0700)]
Implement SORT_BY_INIT_PRIORITY.

2016-06-28  Igor Kudrin  <ikudrin@accesssoftek.com>

gold/
PR gold/18098
* script-c.h (Sort_wildcard): Add SORT_WILDCARD_BY_INIT_PRIORITY.
* script-sections.cc (Input_section_sorter::get_init_priority): New method.
(Input_section_sorter::operator()): Handle SORT_WILDCARD_BY_INIT_PRIORITY.
(Output_section_element_input::print): Likewise.
* script.cc (script_keyword_parsecodes): Add entry SORT_BY_INIT_PRIORITY.
* yyscript.y (SORT_BY_INIT_PRIORITY): New token.
(wildcard_section): Handle SORT_BY_INIT_PRIORITY.

* testsuite/Makefile.am (script_test_14): New test.
* testsuite/Makefile.in: Regenerate.
* testsuite/script_test_14.s: New test source file.
* testsuite/script_test_14.sh: New test script.
* testsuite/script_test_14.t: New test linker script.

7 years agoImplement SORT_BY_INIT_PRIORITY.
Igor Kudrin [Tue, 28 Jun 2016 20:56:05 +0000 (13:56 -0700)]
Implement SORT_BY_INIT_PRIORITY.

2016-06-28  Igor Kudrin  <ikudrin@accesssoftek.com>

gold/
PR gold/18098
* script-c.h (Sort_wildcard): Add SORT_WILDCARD_BY_INIT_PRIORITY.
* script-sections.cc (Input_section_sorter::get_init_priority): New method.
(Input_section_sorter::operator()): Handle SORT_WILDCARD_BY_INIT_PRIORITY.
(Output_section_element_input::print): Likewise.
* script.cc (script_keyword_parsecodes): Add entry SORT_BY_INIT_PRIORITY.
* yyscript.y (SORT_BY_INIT_PRIORITY): New token.
(wildcard_section): Handle SORT_BY_INIT_PRIORITY.

* testsuite/Makefile.am (script_test_14): New test.
* testsuite/Makefile.in: Regenerate.
* testsuite/script_test_14.s: New test source file.
* testsuite/script_test_14.sh: New test script.
* testsuite/script_test_14.t: New test linker script.

7 years ago[TILEPro] Don't build gdb
Walter Lee [Tue, 28 Jun 2016 18:15:32 +0000 (14:15 -0400)]
[TILEPro] Don't build gdb

In configure.ac, add gdb to noconfigdirs for TILEPro.

       * configure.ac (tilepro-*-*): Add gdb to noconfigdirs.
       * configure: Regenerate.

7 years ago[AArch64] Use int64_t for address offset
Yao Qi [Tue, 28 Jun 2016 16:24:25 +0000 (17:24 +0100)]
[AArch64] Use int64_t for address offset

In AArch64 displaced stepping and fast tracepoint, GDB/GDBserver needs
to check whether the offset can fit in the range.  We are using int32_t
for offset, it is sufficient to get an offset from an instruction, but
it is not enough to get an offset from two addresses.  For example,
we have a BL in shared lib which is at 0x0000002000040774, and the
scratch pad for displaced stepping is at 0x400698.  The offset can't
fit in 28 bit imm.  However, since we are using int32_t for offset, GDB
thinks the offset can fit it, and generate the B instruction with wrong
offset.

It fixes the following fail,

-FAIL: gdb.base/dso2dso.exp: next over call to sub2

gdb:

2016-06-28  Yao Qi  <yao.qi@linaro.org>

* aarch64-tdep.c (aarch64_displaced_step_b): Use int64_t for
variable new_offset.

gdb/gdbserver:

2016-06-28  Yao Qi  <yao.qi@linaro.org>

* linux-aarch64-low.c (aarch64_ftrace_insn_reloc_b): Use int64_t
for variable new_offset.
(aarch64_ftrace_insn_reloc_b_cond): Likewise.
(aarch64_ftrace_insn_reloc_cb): Likewise.
(aarch64_ftrace_insn_reloc_tb): Likewise.
(aarch64_install_fast_tracepoint_jump_pad): Likewise.  Use
PRIx64 instead of PRIx32.

7 years agoFix typo in previous commit
Nick Clifton [Tue, 28 Jun 2016 16:03:11 +0000 (17:03 +0100)]
Fix typo in previous commit

7 years agoSkip version check for unreferenced and undefined symbol
H.J. Lu [Tue, 28 Jun 2016 15:08:04 +0000 (08:08 -0700)]
Skip version check for unreferenced and undefined symbol

No need to check version if symbol is unreferenced and undefined.

bfd/

PR ld/20306
* elflink.c (elf_link_check_versioned_symbol): Return false
for unreferenced undefined symbol.

ld/testsuite/

* testsuite/ld-gc/gc.exp: Run pr20306 test.
* ld-gc/pr20306.c: New file.
* ld-gc/pr20306.d: Likewise.

7 years agoFix more linker testsuite failures.
Nick Clifton [Tue, 28 Jun 2016 14:55:22 +0000 (15:55 +0100)]
Fix more linker testsuite failures.

bfin * elf32-bfin.c (bfin_adjust_dynamic_symbol): Fail if a COPY reloc
is needed.

ld * testsuite/ld-elf/comm-data.exp: Expect comm-data2 test to fail
for bfin.
* testsuite/ld-elf/elf.exp: Expect pr14170 and symbolic function
tests to fail for bfin.
* testsuite/ld-elf/endsym.d: Expect to fail with cr16, crx, dlx,
nds32 and visium.
* testsuite/ld-elf/var1.d: Expect to fail with d30v, dlx, ft32 and
microblaze.
* testsuite/ld-pe/pe.exp: Expect foreign symbol test to fail for
mcore-pe.

7 years agoRelax previous restriction on running binutils ar tests for Alpha targets, allowing...
Nick Clifton [Tue, 28 Jun 2016 13:05:34 +0000 (14:05 +0100)]
Relax previous restriction on running binutils ar tests for Alpha targets, allowing the tests to be run for ELF variants.

* testsuite/binutils-all/ar.exp: Relax previous restriction on
Alpha targets.  Allow ELF based Alpha targets.

7 years agoUse `supports_gnu_unique' with the `unique_symbol' and `type' tests
Maciej W. Rozycki [Tue, 28 Jun 2016 12:21:36 +0000 (13:21 +0100)]
Use `supports_gnu_unique' with the `unique_symbol' and `type' tests

Complement commit a43942db49b0 ("LD/ELF: Unify STB_GNU_UNIQUE handling")
and use `supports_gnu_unique' with the `unique_symbol' and `type' tests,
fixing failures like:

.../binutils/testsuite/binutils-all/unique.s: Assembler messages:
.../binutils/testsuite/binutils-all/unique.s:2: Error: symbol type "gnu_unique_object" is supported only by GNU targets

ERROR: .../binutils/testsuite/binutils-all/unique.s: assembly failed
UNRESOLVED: ar unique symbol in archive
.../binutils/ar -s -r -c tmpdir/artest.a tmpdir/unique.o
Executing on host: .../binutils/ar -s -r -c tmpdir/artest.a tmpdir/unique.o   (timeout = 300)
.../binutils/ar: tmpdir/unique.o: No such file or directory

FAIL: ar unique symbol in archive

and:

.../gas/testsuite/gas/elf/type.s: Assembler messages:
.../gas/testsuite/gas/elf/type.s:30: Error: symbol type "gnu_unique_object" is supported only by GNU targets
../as-new: BFD (GNU Binutils) 2.26.51.20160628 internal error, aborting at .../gas/write.c:608 in size_seg

../as-new: Please report this bug.

.../gas/testsuite/../../binutils/readelf -s dump.o | grep "1 *\[FIONTCU\]" > dump.out
Executing on host: sh -c {.../gas/testsuite/../../binutils/readelf -s dump.o >readelf.out 2>gas.stderr}  /dev/null  (timeout = 300)
readelf: Error: dump.o: Failed to read file's magic number
FAIL: elf type list

on MIPS/FreeBSD targets:

mips-freebsd  -FAIL: ar unique symbol in archive
mips-freebsd  -FAIL: elf type list
mips64-freebsd  -FAIL: ar unique symbol in archive
mips64-freebsd  -FAIL: elf type list
mips64el-freebsd  -FAIL: ar unique symbol in archive
mips64el-freebsd  -FAIL: elf type list
mipsel-freebsd  -FAIL: ar unique symbol in archive
mipsel-freebsd  -FAIL: elf type list

binutils/
* testsuite/binutils-all/ar.exp: Use `supports_gnu_unique' with
the `unique_symbol' test.

gas/
* testsuite/gas/elf/elf.exp: Use `supports_gnu_unique' with the
`type' test.

7 years agoMark ARM mapping symbols in object files are precious, so that strip will not remove...
Nick Clifton [Tue, 28 Jun 2016 12:22:49 +0000 (13:22 +0100)]
Mark ARM mapping symbols in object files are precious, so that strip will not remove them.

* elf32-arm.c (elf32_arm_backend_symbol_processing): New
function.  Marks mapping symbols in object files as precious, so
that strip will not remove them.
(elf_backend_symbol_processing): Define.

7 years agoFix RX and M68HC11 linker testsuite failures.
Nick Clifton [Tue, 28 Jun 2016 11:43:14 +0000 (12:43 +0100)]
Fix RX and M68HC11 linker testsuite failures.

ld * testsuite/ld-elf/merge.d: Add m68hc11 to list of targets that
expect to fail this test.
* testsuite/ld-scripts/overlay-size.d: Skip the entire test for
RX.
* testsuite/ld-scripts/rgn-at10.d: No longer expect this test to
fail for the RX.
* testsuite/ld-scripts/rgn-at11.d: Likewise.
* testsuite/ld-scripts/rgn-at2.d: Likewise.
* testsuite/ld-scripts/rgn-at6.d: Likewise.
* testsuite/ld-scripts/rgn-at7.d: Likewise.
* testsuite/ld-scripts/rgn-at8.d: Likewise.

7 years agoImplement get_syscall_trapinfo for arm-linux
Yao Qi [Tue, 28 Jun 2016 11:02:36 +0000 (12:02 +0100)]
Implement get_syscall_trapinfo for arm-linux

gdb/gdbserver:

2016-06-28  Yao Qi  <yao.qi@linaro.org>

* linux-arm-low.c (arm_get_syscall_trapinfo): New function.
(the_low_target): Install arm_get_syscall_trapinfo.

7 years agoImplement get_syscall_trapinfo for aarch64-linux
Yao Qi [Tue, 28 Jun 2016 11:02:35 +0000 (12:02 +0100)]
Implement get_syscall_trapinfo for aarch64-linux

gdb/gdbserver:

2016-06-28  Yao Qi  <yao.qi@linaro.org>

* linux-aarch64-low.c (aarch64_get_syscall_trapinfo): New
function.
(the_low_target): Install aarch64_get_syscall_trapinfo.

7 years agoRemove parameter sysret from linux_target_ops.get_syscall_trapinfo
Yao Qi [Tue, 28 Jun 2016 11:02:35 +0000 (12:02 +0100)]
Remove parameter sysret from linux_target_ops.get_syscall_trapinfo

When I implement linux_target_ops.get_syscall_trapinfo for aarch64 and arm,
I find the second parameter sysret isn't used at all.  In RSP, we don't
need syscall return value either, because GDB can figure out the return
value from registers content got by 'g' packet.

This patch is to remove them.

gdb/gdbserver:

2016-06-28  Yao Qi  <yao.qi@linaro.org>

* linux-low.c (get_syscall_trapinfo): Remove parameter sysret.
Callers updated.
* linux-low.h (struct linux_target_ops) <get_syscall_trapinfo>:
Remove parameter sysno.
* linux-x86-low.c (x86_get_syscall_trapinfo): Remove parameter
sysret.

7 years agoProbe catch syscall support
Yao Qi [Tue, 28 Jun 2016 11:02:35 +0000 (12:02 +0100)]
Probe catch syscall support

In 82075af2c14b1f8a54fa5796fb63f7ef23f98d9d (Implement 'catch syscall'
for gdbserver), only x86 is supported, but the test can still be run
on other linux targets, like aarch64 and ppc, with native-gdbserver.
This causes many new fails.

This patch removes the check on isnative and on target triplets.
Instead, we can insert catch point, and resume the program to see whether
catch syscall is supported or not.

gdb/testsuite:

2016-06-28  Yao Qi  <yao.qi@linaro.org>

* gdb.base/catch-syscall.exp: Remove check on isnative and target
triplets.  Start gdb, execute catch syscall, and continue.  Check
gdb's output to determine catch syscall is supported.

7 years agoDon't convert R_SPARC_32 to R_SPARC_RELATIVE if class is ELFCLASS64.
James Clarke [Tue, 28 Jun 2016 11:00:56 +0000 (12:00 +0100)]
Don't convert R_SPARC_32 to R_SPARC_RELATIVE if class is ELFCLASS64.

bfd * elfxx-sparc.c (_bfd_sparc_elf_relocate_section): Don't convert
R_SPARC_32 to R_SPARC_RELATIVE if class is ELFCLASS64.

gold * sparc.cc (Target_sparc::Scan::local): Don't convert R_SPARC_32
to R_SPARC_RELATIVE if class is ELFCLASS64.
(Target_sparc::Scan::global): Likewise.

ld * testsuite/ld-elf/symbolic-func.r: Allow non-zero offsets from
.text.

7 years agoMIPS/LD/testsuite: Accept any UNIX OS/ABI with GNU attribute tests
Maciej W. Rozycki [Tue, 28 Jun 2016 10:08:58 +0000 (11:08 +0100)]
MIPS/LD/testsuite: Accept any UNIX OS/ABI with GNU attribute tests

Remove failures with FreeBSD targets, e.g:

regexp_diff match failure
regexp "^  OS/ABI:                            UNIX - System V$"
line   "  OS/ABI:                            UNIX - FreeBSD"
FAIL: ld-mips-elf/attr-gnu-4-50

introduced with commit 351cdf24d223 ("[MIPS] Implement O32 FPXX, FP64
and FP64A ABI extensions").

ld/
* testsuite/ld-mips-elf/attr-gnu-4-10.d: Match any UNIX OS/ABI.
* testsuite/ld-mips-elf/attr-gnu-4-50.d: Likewise.
* testsuite/ld-mips-elf/attr-gnu-4-60.d: Likewise.
* testsuite/ld-mips-elf/attr-gnu-4-70.d: Likewise.

7 years agoFix new testcase for hppa64
Alan Modra [Tue, 28 Jun 2016 09:33:59 +0000 (19:03 +0930)]
Fix new testcase for hppa64

Anything in first column is a label on hppa64.

PR gas/20247
* testsuite/gas/elf/section11.s: Don't start directives in first column.

7 years agoDon't run ld-scripts/pr20302 on linuxaout
Alan Modra [Tue, 28 Jun 2016 09:32:08 +0000 (19:02 +0930)]
Don't run ld-scripts/pr20302 on linuxaout

PR ld/20302
* testsuite/ld-scripts/pr20302.d: Exclude *-*-*aout.

7 years agoInvalid read in _bfd_elf_get_symbol_version_string
Alan Modra [Tue, 28 Jun 2016 09:29:33 +0000 (18:59 +0930)]
Invalid read in _bfd_elf_get_symbol_version_string

PR 20304
* objdump.c (objdump_print_symname): Don't attempt to retrieve
version info from synthetic symbols.

7 years ago[AArch64] Make register indices be full 64-bit values
Richard Sandiford [Tue, 28 Jun 2016 08:21:04 +0000 (09:21 +0100)]
[AArch64] Make register indices be full 64-bit values

aarch64_opnd_info used bitfields to hold vector element indices,
but values were stored into those bitfields before their ranges had
been checked.  This meant large invalid indices could be silently
truncated to smaller valid indices.

The two obvious fixes were to do the range checking earlier or use
a full 64-bit field for the index.  I went for the latter for two
reasons:

      - Doing the range checking in operand_general_constraint_met_p
        seems structurally cleaner than doing it while parsing.

      - The bitfields didn't really buy us anything.  The imm field
        of the union is already 128 bits, so we can use a full int64_t
        index without growing the structure.

The patch also adds missing range checks for the elements in a register
list index.

include/
* opcode/aarch64.h (aarch64_opnd_info): Change index fields to int64_t.

opcodes/
* aarch64-opc.c (operand_general_constraint_met_p): Check the
range of ldst_elemlist operands.
(print_register_list): Use PRIi64 to print the index.
(aarch64_print_operand): Likewise.

gas/
* testsuite/gas/aarch64/diagnostic.s,
testsuite/gas/aarch64/diagnostic.l: Add tests for out-of-range indices.