This is the mail archive of the binutils@sourceware.org mailing list for the binutils project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[Patch, AArch64, BFD, LD][2.25] Cortex-A53 erratum 835769 linker workaround.


Hi,

Some early revisions of the Cortex-A53 have an erratum (835769) whereby
it is possible for a 64-bit multiply-accumulate instruction in AArch64 state to
generate an incorrect result.  The details are quite complex and hard to
determine statically, since branches in the code may exist in some
circumstances, but all cases end with a memory (load, store, or
prefetch) instruction followed immediately by the multiply-accumulate
operation.

We employ a linker patching technique, whereby we move the potentially
affected multiply-accumulate instruction into a patch region and
replace the original instruction with a branch to the patch.

This patch achieves the linker patch. Also, a patch to turn this on from the compiler driver is being submitted independently to the gcc lists.

The fix is disabled by default and can be turned on by the
--fix-cortex-a53-835769 linker option.

Tested on aarch64-none-linux-gnu (and bootstrap with the option) and built and run various large
benchmarks with it.

Ok for 2.25?

Thanks,
Tejas.

Changelog:

2014-10-24  Tejas Belagod  <tejas.belagod@arm.com>

bfd/
	* bfd-in.h (bfd_elf64_aarch64_set_options): Add a parameter.
	* bfd-in2.h (bfd_elf64_aarch64_set_options): Likewise.
	* elfnn-aarch64.c (aarch64_erratum_835769_stub): New.
	(elf_aarch64_stub_type): Add new type 		
	aarch64_stub_erratum_835769_veneer.
	(elf_aarch64_stub_hash_entry): New fields for erratum 835769.
	(aarch64_erratum_835769_fix): New data struct to record erratum
	835769.
	(elf_aarch64_link_hash_table: Global flags for 835769.
	(aarch64_build_one_stub): Add case for 835769.
	(aarch64_size_one_stub): Likewise.
	(aarch64_mem_op_p, aarch64_mlxl_p, 	
	aarch64_erratum_sequence,erratum_835769_scan):
	New. Decode and scan functions for erratum 835769.
	(elf_aarch64_create_or_find_stub_sec): New.
	(elfNN_aarch64_size_stubs): Look for erratum 835769 and record
	them.
	(bfd_elfNN_aarch64_set_options: Set global flag for 835769.
	(erratum_835769_branch_to_stub_data, 	
	make_branch_to_erratum_835769_stub):New. Connect up all the
	erratum stubs to occurances by branches.
	(elfNN_aarch64_write_section): New hook.
	(aarch64_map_one_stub): Output erratum stub symbol.
	(elfNN_aarch64_size_dynamic_sections): Init mapping symbol
	information for erratum 835769.
	(elf_backend_write_section): Define.

ld/
	* emultempl/aarch64elf.em: Add command-line option for erratum
	835769.

ld/testsuite/

	* ld-aarch64/aarch64-elf.exp (aarch64elftests): Drive erratum
	835769 tests.
	* ld-aarch64/erratum835769.d: New.
	* ld-aarch64/erratum835769.s: New.

Attachment: a_53-erratum-225-1.txt
Description: Text document


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]