Adjust the conditions for entering the aligned copy loop to
improve performance on mutually misaligned buffer copies.
2013-07-01 Will Newton <will.newton@linaro.org>
* libc/machine/arm/memcpy-armv7a.S: Adjust entry to
aligned loop to improve misaligned copy performance.
path without long path prefixing.
* wide_path.h (wide_path::wide_path): Allow extra bool parameter to
specify whether or not performing Windows long path prefixing.
* mips.h: Fix comment for "1": it is now STYPE rather than SHAMT.
Use "source" rather than "destination" for microMIPS "G".
gas/
* config/tc-mips.c (validate_mips_insn): Use STYPE rather than SHAMT.
(PTR_REG): Ditto.
(PTR_SIZE): Ditto.
(PTR_LOG_SIZE): Ditto.
(start): Use GEN_DWORD to replace the .dword of HeapBase,
__bss_start__, __bss_end__, FUNCTION(_fini), env and
CommandLine; when __ILP32__ is defined, set the stack base to
the top end of the 32-bit address space if the returned value
from the Angel API call is larger than or equal to 4 GiB.
Also carry out sanity check on the heap base; abort if the
base is larger than or equal to 4 GiB. Use other new
macros in the instructions that processes the argv arrays.
(StackBase): New lable; replace __stack_base__.
(__stack_base__): Set with StackBase or StackBase + 4.
* elfxx-mips.h (_bfd_mips_elf_get_synthetic_symtab): New
prototype.
* elf32-mips.c (elf_backend_plt_sym_val): Remove macro.
(bfd_elf32_get_synthetic_symtab): New macro.
* elfxx-mips.c (plt_entry): New structure.
(mips_elf_link_hash_entry): Add use_plt_entry member.
(mips_elf_link_hash_table): Rename plt_entry_size member to
plt_mips_entry_size. Add plt_comp_entry_size, plt_mips_offset,
plt_comp_offset, plt_got_index entries and plt_header_is_comp
members.
(STUB_LW_MICROMIPS, STUB_MOVE_MICROMIPS): New macros.
(STUB_LUI_MICROMIPS, STUB_JALR_MICROMIPS): Likewise.
(STUB_ORI_MICROMIPS, STUB_LI16U_MICROMIPS): Likewise.
(STUB_LI16S_MICROMIPS): Likewise.
(MICROMIPS_FUNCTION_STUB_NORMAL_SIZE): Likewise.
(MICROMIPS_FUNCTION_STUB_BIG_SIZE): Likewise.
(micromips_o32_exec_plt0_entry): New variable.
(mips16_o32_exec_plt_entry): Likewise.
(micromips_o32_exec_plt_entry): Likewise.
(mips_elf_link_hash_newfunc): Initialize use_plt_entry.
(mips_elf_output_extsym): Update to use gotplt_union's plist
member rather than offset.
(mips_elf_gotplt_index): Likewise. Remove the VxWorks
restriction. Use MIPS_ELF_GOT_SIZE to calculate GOT address.
(mips_elf_count_got_symbols): Update to use gotplt_union's plist
member rather than offset.
(mips_elf_calculate_relocation): Handle MIPS16/microMIPS PLT
entries.
(_bfd_mips_elf_create_dynamic_sections): Don't set PLT sizes
here.
(mips_elf_make_plt_record): New function.
(_bfd_mips_elf_check_relocs): Update comment. Record occurences
of JAL relocations that might need a PLT entry.
(_bfd_mips_elf_adjust_dynamic_symbol): Update to use
gotplt_union's plist member rather than offset. Set individual
PLT entry sizes here. Handle MIPS16/microMIPS PLT entries.
Don't set the symbol's value in the symbol table for PLT
references here. Don't set the PLT or PLT GOT section sizes
here.
(mips_elf_estimate_stub_size): Handle microMIPS stubs.
(mips_elf_allocate_lazy_stub): Likewise.
(mips_elf_lay_out_lazy_stubs): Likewise. Define a _MIPS_STUBS_
magic symbol.
(mips_elf_set_plt_sym_value): New function.
(_bfd_mips_elf_size_dynamic_sections): Set PLT header size and
PLT and PLT GOT section sizes here. Set the symbol values in
the symbol table for PLT references here. Handle microMIPS
annotation of the _PROCEDURE_LINKAGE_TABLE_ magic symbol.
(_bfd_mips_elf_finish_dynamic_symbol): Update to use
gotplt_union's plist member rather than offset. Handle
MIPS16/microMIPS PLT entries. Handle microMIPS stubs.
(_bfd_mips_vxworks_finish_dynamic_symbol): Update to use
gotplt_union's plist member rather than offset. Use
MIPS_ELF_GOT_SIZE to calculate GOT address.
(mips_finish_exec_plt): Handle microMIPS PLT. Return status.
(_bfd_mips_elf_finish_dynamic_sections): Handle result from
mips_finish_exec_plt.
(_bfd_mips_elf_link_hash_table_create): Update to use
gotplt_union's plist member rather than offset.
(_bfd_mips_elf_get_synthetic_symtab): New function.
include/elf/
* mips.h (ELF_ST_IS_MIPS_PLT): Respect STO_MIPS16 setting.
(ELF_ST_SET_MIPS_PLT): Likewise.
gdb/
* mips-tdep.c (mips_elf_make_msymbol_special): Handle MIPS16 and
microMIPS synthetic symbols.
ld/
* emulparams/elf32btsmip.sh: Arrange for .got.plt to be placed
as close to .plt as possible.
* scripttempl/elf.sc: Handle $INITIAL_READWRITE_SECTIONS and
$PLT_NEXT_DATA variables.
ld/testsuite/
* ld-mips-elf/jalx-2.dd: Update for microMIPS PLT support.
* ld-mips-elf/pic-and-nonpic-3a.dd: Update for the _MIPS_STUBS_
magic symbol.
* ld-mips-elf/pic-and-nonpic-3b.dd: Likewise.
* ld-mips-elf/pic-and-nonpic-6-n32.dd: Likewise.
* ld-mips-elf/pic-and-nonpic-6-n64.dd: Likewise.
* ld-mips-elf/pic-and-nonpic-6-o32.dd: Likewise.
* ld-mips-elf/stub-dynsym-1-10000.d: Likewise.
* ld-mips-elf/stub-dynsym-1-2fe80.d: Likewise.
* ld-mips-elf/stub-dynsym-1-7fff.d: Likewise.
* ld-mips-elf/stub-dynsym-1-8000.d: Likewise.
* ld-mips-elf/stub-dynsym-1-fff0.d: Likewise.
* ld-mips-elf/tlslib-o32.d: Likewise.
opcodes/
* mips-dis.c (is_mips16_plt_tail): New function.
(print_insn_mips16): Handle MIPS16 PLT entry's GOT slot address
word.
(is_compressed_mode_p): Handle MIPS16/microMIPS PLT entries.
Import the latest version of strlen from the Linaro cortex-strings
package. This version is faster across a variety of block size and
alignments on ARMv7.
newlib/ChangeLog:
2013-06-21 Will Newton <will.newton@linaro.org>
* libc/machine/arm/strlen-armv7.S: Import latest strlen
code from Linaro cortex-strings.
of conglomerated newargv.setup(). Let newargv.setup() decide when to call
dup_all(). Only set argc and argv for cygwin processes.
(av::setup): Rename from av::fixup. Accept argc and argv parameters. Fill out
argv and argc here. Duplicate whole argv structure when this is a Cygwin
executable.
* winf.cc (linebuf::fromargv): Don't bother duplicating argv elements since
they will never be used.
* winf.h (av::set): Delete.
(av::setup): Rename from av::fixup. Add two parameters.
(av::replace0_maybe): Assign calloced to 1 rather than 'true' for clarity.
(av::dup_maybe): Delete.
(av::dup_all): Set calloced to show that we have duplicated all of the
arguments in the list.
* times.cc (GetSystemTimePreciseAsFileTime): Temporarily declare here
to workaround missing definition in 32 bit w32api headers.
(get_system_time): New always inline function to call either
GetSystemTimePreciseAsFileTime or GetSystemTimeAsFileTime on a per OS
basis. Call throughout instead of GetSystemTimeAsFileTime.
* wincap.h (wincaps::has_precise_system_time): New element.
* wincap.cc: Implement above element throughout.
Windows 9x compatibility.
* fhandler_disk_file.cc (fhandler_base::fstat_helper): Drop now unneeded
casts in calls to_timestruc_t.
(fhandler_base::utimens_fs): Ditto for timespec_to_filetime.
* fhandler_proc.cc (format_proc_stat): Ditto for to_time_t.
* hires.h (class hires_ms): Remove unused member initime_ns.
Remove declarations for timeGetTime_ns and prime.
(hires_ms::uptime): Remove.
* posix_ipc.cc (ipc_cond_timedwait): Ditto for timespec_to_filetime.
* fhandler_registry.cc (fhandler_registry::fstat): Add cast.
* resource.cc (fill_rusage): Call NtQueryInformationProcess rather than
GetProcessTimes to deal with LARGE_INTEGER rather than FILETIME.
* times.cc: Simplify time handling. Throughout, use LARGE_INTEGER
rather than FILETIME to simplify computations. Throughout use
{u}int64_t rather than {unsigned} long long. Drop unneeded casts since
NSPERSEC is 64 bit anyway.
(systime_ns): Remove.
(times): Call NtQuerySystemInformation to fetch boot time. Call
NtQueryInformationProcess rather than GetProcessTimes to deal with
LARGE_INTEGER rather than FILETIME. Call GetSystemTimeAsFileTime.
(totimeval): Use constant 1000000 as in other functions.
(time_t_to_filetime): Remove.
(to_time_t): Change return type to time_t.
(time_as_timestruc_t): Rename filetime to systime.
(time): Ditto. Add cast.
(hires_ns::nsecs): Fix return type cast.
(hires_ms::timeGetTime_ns): Remove.
(hires_ns::prime): Remove.
(hires_ms::nsecs): Drop call to prime. Call GetSystemTimeAsFileTime
directly. Subtract FACTOR here since it's the only function needing
to do so.
(minperiod): Cosmetically change to ULONG.
(hires_ns::resolution): Fix return type cast.
(hires_ms::resolution): Simplify, rely on NtQueryTimerResolution.
* winsup.h: Align time related prototypes to above changes.
Will Newton
Corinna Vinschen
Jeff Johnston
Yufeng Zhang
Richard Sandiford
Maciej W. Rozycki
Yaakov Selkowitz
Nick Clifton
Christopher Faylor
Catherine Moore
Sandra Loosemore