- Jan 21, 2016
-
-
Han Shen authored
Tested by build.py for linux aarch64 toolchains. The upstream patch is: commit 9a472eda40ba686e45bf4922455518ffa3c887e1 Author: Han Shen <shenhan@google.com> Date: Fri Jan 15 09:31:23 2016 -0800 [gold][aarch64] PR gold/19472 - DSOs need pc-relative stubs. The stub generated during relaxation uses absolute addressing mode for shared libraries, which is not correct. Use pc-relative addressing instead. gold/ChangeLog: 2016-01-15 Han Shen <shenhan@google.com> PR gold/19472 - DSOs need pc-relative stubs. * aarch64.cc (Reloc_stub::stub_type_for_reloc): Return PC-relative stub type for DSOs and pie executables. Change-Id: Id0022975cf93600117ff5bf300b9d736d6ad1f80
-
- Jan 15, 2016
-
-
Han Shen authored
Note, there is a small change from the origin patch - while in the upstream patch, the option --no-apply-dynamic is default off, it is on by default in this CL. This should not bring any changes to android or chromeos, because the relocation value this CL changes is not used by chromeos, and is not used by android either (after the fix in bionic months ago). The backported patch is - commit 0eccf19f96d6218dd7c2f8d13f0546c2b942cc08 Author: Cary Coutant <ccoutant@gmail.com> Date: Thu Nov 5 12:24:14 2015 -0800 Add aarch64-specific --no-apply-dynamic-relocs option. With --no-apply-dynamic-relocs on aarch64 targets, gold will not apply link-time values for absolute relocations that become dynamic relocations. This provides a workaround for broken Android dynamic linkers that use the link-time value as an extra addend to the relocation. gold/ PR gold/19163 * aarch64.cc (Target_aarch64::Relocate::relocate): Don't apply certain relocations if --no-apply-dynamic-relocs is set. * options.h (--apply-dynamic-relocs): New aarch64-specific option. Tested - build whole toolchain and built android N9 image. Change-Id: Ida524bbee2a83574cd9e78ffdad9270787327c78
-
- Jan 13, 2016
-
- Dec 18, 2015
-
-
Han Shen authored
ChromeOS reported a bug complaining about no default library search path after unification. Bug here - https://code.google.com/p/chromium/issues/detail?id=569605 Fix it for ChromeOS. This fix does not change anything for Android. Test: ChromeOS: manually test bug is fixed Android: ./build.py --host=linux --toolchain=x86_64 Change-Id: Ia900d081980d408c62a737917708e19842357e8e
-
- Nov 21, 2015
-
-
Han Shen authored
-
Han Shen authored
"%ld" is the printf specifier for "long int", which is 64-bit on 64-bit machines, and 32-bit on 32-bit hosts. While uint64_t is always 64 bits, so replace it with macro "PRIu64". Tested: ./build.py --host "linux" --toolchain aarch64-linux-android ./build.py --host "linux" --toolchain arm-linux-androideabi ./build.py --host "linux" --toolchain mipsel-linux-android ./build.py --host "windows" --toolchain arm-linux-androideabi Change-Id: I9f0b322ed14e6feb7a110bfebcff963cf94d7703
-
Han Shen authored
Bug entry here - https://b.corp.google.com/u/0/issues/25642296 Tested: ./build.py --host "linux" --toolchain aarch64-linux-android Upstream patch here - commit 4d2f5d5824be99326c2c3e459c298e338856d8c8 Author: Han Shen <shenhan@google.com> Date: Tue Oct 27 15:17:54 2015 -0700 Patch for PR gold/19042 - unsupported reloc 311/312. gold/ * aarch64.cc (Target_aarch64::Scan::local): Add support for reloc 311/312. Change-Id: If784460c1ebdfb2be1ebdf1486db5eb03c5240a5
-
- Nov 20, 2015
-
-
Han Shen authored
About source code - The base version of this binutils is newer than that of aosp/binutils-2.25, it is based on the binutils that is used to build google products and ChromiumOS. And it contains *all* local Android patches as well as all patches that are cherry-picked from upstream for aosp/binutils-2.25 tree (up to Nov. 5 - 932d71b8). You may find the detailed development history for this binutils tree here - https://chromium.googlesource.com/chromiumos/third_party/binutils/+log/unification (This CL is a combination of all the CLs in it. After this CL is submitted the tree will be identical to https://chromium.googlesource.com/chromiumos/third_party/binutils/+log/unification at 2865a3615d80bd5f82d14d7e0484e84dc052596a) About testing - We tested this binutils for both ChromiumOS and Android. For android, we tested building N4, N5X, N6, N7, N9 using new binutils, we also did a full-build of toolchain (by build.py) and built a N5X image; for ChromiumOS - it passed ChromiumOS toolchain release tests on all 4 platforms (x86, x86_64, arm32 and arm64). Change-Id: I2bb2cf579f9458d0a8bc9612331dc7d5043e3d82
-
- Nov 06, 2015
-
-
Lee Campbell authored
Cherry-pick of upstream commit cdb061674c044636465e52a8c989b9011f2ec8ac gold/ChangeLog: * arm.cc (Reloc_stub::stub_type_for_reloc): Plumb to stub generation. * options.h (General_options): Add --pic-veneer option. BUG: 25512672
-
- Nov 04, 2015
-
-
Nick Clifton authored
Cherry-pick of upstream commit 31593e1b96c792abba3c5268d6423975aefa56b2 2015-02-24 Nick Clifton <nickc@redhat.com> * configure.ac (AC_CHECK_HEADERS): Add wctype.h. * configure: Regenerate. * config.in: Regenerate. * peXXigen.c: Include wctype.h if HAVE_WCTYPE_H is defined. (u16_mbtowc): Use wint_t types if HAVE_WCTYPE_H is defined. (rsrc_cmp): Use towlower instead of wcsncasecmp if HAVE_WCTYPE_H is defined. Bug: 25427405 Change-Id: I06e99d906226ce9c88ff1c568ae7a4d84df7f332
-
- Oct 16, 2015
-
-
Than McIntosh authored
Cherrypick of upstream commit 5d7908e0880030628536a0266968a15922574735 Author: Cary Coutant <ccoutant@gmail.com> Date: Thu Jun 11 13:19:38 2015 -0700 Fix gold build error at -O0. In aarch64.cc, Erratum_stub::STUB_ADDR_ALIGN and Reloc_stub::STUB_ADDR_ALIGN are declared as static const int, and initialized in the class body. These values are referenced from outside the class body, and with no optimization, the references go unresolved. gold/ * aarch64.cc (Erratum_stub::STUB_ADDR_ALIGN): Move initialization outside class body. (Reloc_stub::STUB_ADDR_ALIGN): Likewise. Change-Id: I6c0726e624be076e0d0565652a56da5ef761d88a
-
- Sep 29, 2015
-
-
Han Shen authored
gold/ChangeLog: * aarch64.cc (Target_aarch64::scan_erratum_843419_span): Use 'gold_info'. (Target_aarch64::scan_erratum_835769_span): USe 'gold_info'. (cherry picked from commit 73854cdd438dc8858290231ebcb682377d71fec2)
-
Han Shen authored
The psABI (AAELF) says this about mapping symbols: 4.6.5.1 Section-relative mapping symbols Mapping symbols defined in a section define a sequence of half-open address intervals that cover the address range of the section. Each interval starts at the address defined by the mapping symbol, and continues up to, but not including, the address defined by the next (in address order) mapping symbol or the end of the section. A section must have a mapping symbol defined at the beginning of the section; however, if the section contains only data then the mapping symbol may be omitted. That makes it pretty clear that, in the absence of a mapping symbol, the entire section can be treated as data. gold/ChangeLog: * aarch64.cc (AArch64_relobj::scan_errata): Drop missing symbol warning. * arm.cc (Arm_relobj::scan_section_for_cortex_a8_erratum): Drop missing symbol warning. (cherry picked from commit 61163dfaaedb90d78e11c1d79ad1e469daf3da47)
-
- Aug 16, 2015
-
-
Andrew Hsieh authored
-
- Jul 31, 2015
-
-
Andrew Hsieh authored
This was added before ld.gold scans for erratum 835769, but GCC has been hard-wired to pass "--fix-cortex-a53-835769" to linker (ld.bfd is default for AArch64 for now) unconditionally in Android platform. Change-Id: I39bacdc2a52d09afdc0993c17b46ac438d1fa246
-
- Jul 30, 2015
-
-
Han Shen authored
2015-07-07 Han Shen <shenhan@google.com> gold/ChangeLog: 2015-07-06 Han Shen <shenhan@google.com> * aarch64.cc (AArch64_relobj::do_count_local_symbols): Make legal of mapping symbols. Change-Id: If62f577eeb704d8c7b0b2f525e91b3cb311001ca (cherry picked from commit b91deca94e36b9f18156909e4c615fe7af8e3ae8)
-
Han Shen authored
The crash reason is that the insn to be moved to stub may be a relocation spot, so instead of placing the origin insn (that is insn before-relocation) to the stub, I have to place the relocated one. Note the relocation involved is non-pc-relative, so it is safe to move the relocated insn. gold/ChangeLog: 2015-06-29 Han Shen <shenhan@google.com> * AArch64.cc (Erratum_stub::Insn_utilities): New typedef. (Erratum_stub::update_erratum_insn): New method. (Stub_table::relocate_stubs): Modified to place relocated insn. (AArch64_relobj::fix_errata): Modified gold_assert. Change-Id: I9d90ac6e569bbbab086f877c7446dc6f9ea96247 (cherry picked from commit 56b06706ebe6bb3decc674b788b6f5b243b8610d)
-
Han Shen authored
gold/ChangeLog: * aarch64.cc (AArch64_insn_utilities::BYTES_PER_INSN): Move defintion outside class definition. (AArch64_insn_utilities::AARCH64_ZR): New static constant. (AArch64_insn_utilities::aarch64_op31): New member. (AArch64_insn_utilities::aarch64_ra): New member. (AArch64_insn_utilities::aarch64_mac): New member. (AArch64_insn_utilities::aarch64_mlxl): New member. (ST_E_835769): New global enum member. (Stub_table::relocate_stubs): Add 835769 handler. (Stub_template_repertoire::Stub_template_repertoire): Install new stub type. (AArch64_relobj::scan_errata): This func is renamed from scan_erratum_843419. (AArch64_relobj::do_count_local_symbols): Add 835769 handler. (AArch64_relobj::do_relocate_sections): Add 835769 handler. (AArch64_relobj::scan_sections_for_stubs): Add 835769 handler. (Target_aarch64::scan_erratum_835769_span): New method. (Target_aarch64::create_erratum_stub): New method. (Target_aarch64::is_erratum_835769_sequence): New method. (Target_aarch64::scan_erratum_843419_sequence): Move part of the code into create_erratum_stub. * options.h (fix_cortex_a53_835769): New option. Change-Id: Ia355fad38de61f82b9a42c601129fbbe2730ba27 (cherry picked from commit 2f0c79aa61e82cbc5da4b9ab399362b2cdd8ea2e)
-
Han Shen authored
Now fixing for 843419 is fully functional. The first part of the erratum fix CL is here - https://sourceware.org/ml/binutils/2015-04/msg00229.html gold/ChangeLog 2015-06-10 Han Shen <shenhan@google.com> * aarch64.cc(global enum): New constants representing stub types. (Stub_template): New POD struct. (Stub_template_repertoire): New class. (Stub_base): New class. (Erratum_stub): New class. (Reloc_stub): Refactored to be a subclass of Stub_base. (Reloc_stub::Stub_type): Removed. (Reloc_stub::offset): Moved to Stub_base. (Reloc_stub::set_offset): Moved to Stub_base. (Reloc_stub::destination_address): Moved to Stub_base. (Reloc_stub::set_destination_address): Moved to Stub_base. (Reloc_stub::reset_destination_address): Moved to Stub_base. (Reloc_stub::stub_type): Renamed and moved to Stub_base. (Reloc_stub::stub_size): Renamed and moved to Stub_base. (Reloc_stub::stub_insns): Renamed and moved to Stub_base. (Reloc_stub::write): Moved to Stub_base. (Reloc_stub::invalid_offset): Moved to Stub_base. (Reloc_stub::invalid_address): Moved to Stub_base. (Reloc_stub::stub_type_): Renamed and moved to Stub_base. (Reloc_stub::stub_insns_): Moved to Stub_base. (Reloc_stub::offset_): Moved to Stub_base. (Reloc_stub::destination_address_): Moved to Stub_base. (Stub_table::The_aarch64_relobj): New typedef. (Stub_table::The_erratum_stub): New typedef. (Stub_table::The_erratum_stub_less): New typedef. (Stub_table::The_erratum_stub_set): New typedef. (Stub_table::The_erratum_stub_set_iter): New typedef. (Stub_table::empty): Added emptiness testing for erratum stubs. (Stub_table::add_erratum_stub): New method to add an erratum stub. (Stub_table::find_erratum_stub): New method. (Stub_table::find_erratum_stubs_for_input_section): New method. (Stub_table::erratum_stub_address): New method. (Stub_table::update_date_size_changed_p): Modified to handle erratum stubs. (Stub_table::do_addralign): Modified to handle erratum stubs. (Stub_table::erratum_stubs_): New member. (Stub_table::erratum_stub_size_): New member. (Stub_table::relocate_stubs): Modified to handle erratum stubs. (Stub_table::do_write): Modified to handle erratum stubs. (AArch64_relobj::The_erratum_stub): New typedef. (AArch64_relobj::Erratum_stub_set_iter): New typedef. (AArch64_relobj::fix_errata): New method. (Target_aarch64::The_reloc_stub_type): Removed. (Target_aarch64::The_erratum_stub): New typede. (AArch64_relocate_functions::construct_b): New method. Change-Id: I37ccc9fbefd5559dae700637e119a2a885cf3eb9 (cherry picked from commit a48d0c12f893a922692492829f04189d9f216172)
-
- Jul 27, 2015
-
-
Joseph Myers authored
Cherry-picked from upstream: d5cff5df74b18e1e5ed94de8f4c9adee3ffd95c6 For the case of MIPS n64 target and 32-bit host, the computation of the DT_MIPS_RLD_MAP_REL tag involves sdyn->output_section->vma + sdyn->output_offset (64-bit) being added to b (32-bit host pointer), so losing the high part and resulting in an incorrect DT_MIPS_RLD_MAP_REL tag, and all dynamically linked glibc tests failing for n64. This patch fixes this (spot-tested with glibc tests; however, I don't have a self-contained testcase for this bug). * elfxx-mips.c (_bfd_mips_elf_finish_dynamic_sections) <DT_MIPS_RLD_MAP_REL>: Add target address to host address difference, not to host pointer. Change-Id: If4984d632723a36a3d6c739a96706b5636f975bc
-
- Jul 15, 2015
-
-
Cary Coutant authored
2015-02-02 Khem Raj <raj.khem@gmail.com> gold/ * attributes.h (class Output_attributes_section_data ): Add do_print_to_mapfile function. BUG=21451238 b.android.com/177690 Change-Id: I56d853784a48c247578d08bca848869b3a4c2c78
-
Cary Coutant authored
2015-02-02 Khem Raj <raj.khem@gmail.com> gold/ * attributes.h (class Output_attributes_section_data ): Add do_print_to_mapfile function.
-
- Jul 08, 2015
-
-
Elliott Hughes authored
Change-Id: I26a47a29e4d16e75e21e9295504079a815c334c6
-
- Jun 23, 2015
-
-
Dan Albert authored
AFAICT this was an Android hack. I don't see it anywhere in the history for upstream. Just remove it. Bug: http://b/21534147 Change-Id: I7c69f543a80a0e4805fbea2934eaec052544d6c7
-
- Jun 18, 2015
-
-
Dan Albert authored
-
Dan Albert authored
Change-Id: If634d8d30cfa8c88ca3ba59691a0df4ee5596f85
-
- Jun 17, 2015
-
-
Yiran Wang authored
resolved. It looks like it is WAI, but in some scenario there could a lot of them and the build script just die on any warning. Please refer to bug 21502954 for some details. Change-Id: I6cc90adccc63d26f15821bed6e4c6ff336722eff
-
- May 06, 2015
-
-
Andrew Hsieh authored
1. In Drawin PTHREAD_ONCE_INIT is {0x30B1BCBA, {0}} and the GCC < 4.4 doesn't support ended initializer list 2. wcsncasecmp doesn't exist in MacSDK10.6.x Change-Id: I69204a72f853f5263dffedc448379d75ed4eca2e
-
- May 05, 2015
-
-
Andrew Hsieh authored
-
- Apr 28, 2015
-
-
Han Shen authored
Keep gold consistent with bfd erratum-fixing option names, so as to ease life in Makefile/scripts. gold/ * options.h (--fix-cortex-a53-843419): Rename option. * aarch64.cc (AArch64_relobj::do_count_local_symbols): Use renamed option. (AArch64_relobj::scan_sections_for_stubs): Use renamed option. Change-Id: I8941fb38ef34fa5eb297e83725d87fdee994fc3a
-
Han Shen authored
It implemented scanning the binary and reporting occurrences to users when '--fix-cortex-a53' is turned on. With this, gold users will be able to see if or not there are such erratum occurrences in the output binary. Also included in the CL is reading/recording mapping symbols, which is needed during scan. gold/ChangeLog: * aarch64.cc (AArch64_insn_utilities): New utility class. (AArch64_relobj::Mapping_symbol_position): New struct. (AArch64_relobj::Mapping_symbol_info): New typedef. (AArch64_relobj::do_count_local_symbols): New function overriding parent's implementation. (AArch64_relobj::mapping_symbol_info_): New member (AArch64_relobj::scan_erratum_843419): New method. (Target_aarch64::scan_erratum_843419_span): New method. (Target_aarch64::is_erratum_843419_sequence): New method. * options.h (fix_cortex_a53): New option. Change-Id: I4fe7a2b8858cda358ac28eede42cd5a2f49c2238
-
- Apr 22, 2015
-
-
Andrew Bennett authored
Change-Id: I7ea5748faea472beb79e34aee8d19c4e0f37981e
-
- Apr 21, 2015
-
-
Alexander Ivchenko authored
2015-04-01 Ilya Tocar <ilya.tocar@intel.com> PR gold/17640 * i386.cc (Target_i386::can_convert_mov_to_lea): New. (Target_i386::Scan::local): Don't create GOT entry, when we can convert GOT to GOTOFF. (Target_i386::Scan::global): Ditto. (Target_i386::Relocate::relocate): Convert mov foo@GOT(%reg), %reg to lea foo@GOTOFF(%reg), %reg if possible. * testsuite/Makefile.am (i386_mov_to_lea): New test. * testsuite/i386_mov_to_lea1.s: New. * testsuite/i386_mov_to_lea2.s: Ditto. * testsuite/i386_mov_to_lea3.s: Ditto. * testsuite/i386_mov_to_lea4.s: Ditto. * testsuite/i386_mov_to_lea5.s: Ditto. * testsuite/i386_mov_to_lea.sh: Ditto. Change-Id: I7916a1da20873600a910ce358990b1fe2c1f9897 Signed-off-by:
Alexander Ivchenko <alexander.ivchenko@intel.com>
-
- Apr 06, 2015
-
-
Lai Wei-Chih authored
Change-Id: I7664f43e31327f2f9bb97b0eaad584f80948d4fe
-
- Mar 31, 2015
-
-
Marcus Shawcroft authored
Some early revisions of the Cortex-A53 have an erratum (843419). The details of the erratum are quite complex and involve dynamic conditions. For the purposes of the workaround we have simplified the static conditions to an ADRP in the last two instructions of a 4KByte page, followed within four instructions by a load/store dependent on the ADRP. This patch adds support to conservatively scan for and workaround the erratum. There are two different workaround strategies used. The first is to rewrite ADRP instructions which form part of an erratum sequence with an ADR instruction. In situations where the ADR provides insufficient offset the dependent load or store instruction from the sequence is moved to a stub section and branches are inserted from the original sequence to the relocated instruction and back again. Stub section sizes are rounded up to a multiple of 4096 in order to ensure that the act of inserting work around stubs does not create more errata sequences. Workaround stubs are always inserted into the stub section associated with the input section containing the erratum sequence. This ensures that the fully relocated form of the veneered load store instruction is available at the point in time when the stub section is written. 2015-03-30 Tejas Belagod <tejas.belagod@arm.com> Marcus Shawcroft <marcus.shawcroft@arm.com> Jiong Wang <jiong.wang@arm.com> * bfd-in.h (bfd_elf64_aarch64_set_options) (bfd_elf32_aarch64_set_options): Add parameter. * bfd-in2.h: Regenerated. * elfnn-aarch64.c (aarch64_erratum_843419_stub) (_bfd_aarch64_adrp_p, _bfd_aarch64_erratum_843419_sequence_p) (_bfd_aarch64_erratum_843419_stub_name) (_bfd_aarch64_erratum_843419_fixup) (_bfd_aarch64_erratum_843419_scan) (_bfd_aarch64_erratum_843419_branch_to_stub) (_bfd_aarch64_erratum_843419_p): Define. (enum elf_aarch64_stub_type): Define aarch64_stub_erratum_843419_veneer. (struct elf_aarch64_stub_hash_entry): Define adrp_offset. (struct elf_aarch64_link_hash_table): Define fix_erratum_843419 and fix_erratum_843419_adr. (stub_hash_newfunc): Initialize adrp_offset; (_bfd_aarch64_add_stub_entry_after): Define. (aarch64_map_one_stub, aarch64_build_one_stub) (aarch64_size_one_stub): Handle aarch64_stub_erratum_843419_veneer. (_bfd_aarch64_resize_stubs): Round stub section size. (elfNN_aarch64_size_stubs): Add scan for 843419. (bfd_elfNN_aarch64_set_options): Add parameter. Initialize fix_erratum_843419 and fix_erratum_843419_adr. (struct erratum_835769_branch_to_stub_data): Add info. (elfNN_aarch64_write_section): Initialise info. Handle 843419. (elfNN_aarch64_size_dynamic_sections): Handle 843419. * elfxx-aarch64.c (_bfd_aarch64_decode_adrp_imm) (_bfd_aarch64_sign_extend): Define. (reencode_adr_imm): Remove static. Rename to: (_bfd_aarch64_reencode_adr_imm): Define. (_bfd_aarch64_elf_put_addend): Call _bfd_aarch64_reencode_adr_imm. * elfxx-aarch64.h (AARCH64_ADR_OP, AARCH64_ADRP_OP) (AARCH64_ADRP_OP_MASK, _bfd_aarch64_sign_extend) (_bfd_aarch64_decode_adrp_imm, _bfd_aarch64_reencode_adr_imm): Define. ld/testsuite: 2015-03-24 Tejas Belagod <tejas.belagod@arm.com> * emultempl/aarch64elf.em (aarch64_elf_create_output_section_statements): Add parameter in bfd_elf${ELFSIZE}_aarch64_set_options call. (OPTION_FIX_ERRATUM_843419): Define. (PARSE_AND_LIST_LONGOPTS): Add fix-cortex-a53-843419. (PARSE_AND_LIST_ARGS_CASES): Add OPTION_FIX_ERRATUM_843419. Change-Id: I758b04c90d431f4e72a978871d5a66676c413d92
-
Marcus Shawcroft authored
Ensure that injection of a stub section does not break a link where there is an xpectation that flow of control can pass from one input section to another simply by linking the input sections in series. The solution here is to allow stub sections to be inserted after any input section (existing behaviour), but inject an additional branch at the start of each stub section such that control flow falling into the stub section will branch over the stub section. Change-Id: I8ccadcfb2f43e6409cb5a649087a47c0c8826b15
-
Marcus Shawcroft authored
Lower stub alignment from 8 to 4 bytes thus removing padding between input section content and stub section content. Change-Id: Ic3afff9bc2f013b4b47e181b269b40231385080a
-
Marcus Shawcroft authored
Change-Id: Idc292f60196ed4bf417eb77acc239761666adb8b
-
Marcus Shawcroft authored
Change-Id: Ib6171d8646bc0e825a5038c7087acc534a8612a9
-
Marcus Shawcroft authored
Change the behaviour of section_group[] such that .stub_sec points to the stub section attached to the indexed section rather than the stub section attached to the link_section pointed to be the index section. This provides a mechanism to get to the stub section following any input section. While still allowing the section grouping mechanism to find the section group stub section associated with an input section by first following the link_sec pointer. Change-Id: I19bce3feea77bbc71c04f2d8b3a204c527a21d9a
-