| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
| |
Complement the addition of the required kernel support, present upstream
as from commit 2b5e869ecfcb3112f7e1267cb0328f3ff6d49b18 ("MIPS: ELF:
Interpret the NAN2008 file header flag") and released with Linux 4.5-rc1
on Jan 24th, 2016.
* sysdeps/unix/sysv/linux/mips/configure.ac: Set
`arch_minimum_kernel' to 4.5.0 if 2008 NaN encoding is used.
* sysdeps/unix/sysv/linux/mips/configure: Regenerate.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
On MIPS when the toolchain is using the O32 FPXX ABI, the testsuite
fails to build for pre-R2 CPU.
It assumes that it is possible to use the -mfp64 option to build
tst-abi-fp64amod and tst-abi-fp64mod, while this requires a CPU which
supports the mfhc1 and mthc1 instructions, ie at least a R2 CPU:
error: '-mgp32' and '-mfp64' can only be combined if the target
supports the mfhc1 and mthc1 instructions
The same way it assumes that it is possible to use the -modd-spreg option
to build tst-abi-fpxxomod and tst-abi-fp64mod, while this requires at
least a R1 CPU:
warning: the 'mips2' architecture does not support odd
single-precision registers
This patches changes that by checking the usability of -mfp64 and
-modd-spreg options in configure, and disable those tests when they can
not be used.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* elf/elf.h (PT_MIPS_ABIFLAGS): Define.
(Elf_MIPS_ABIFlags_v0): New structure.
(EF_MIPS_FP64): Define.
(MIPS_AFL_REG_NONE, MIPS_AFL_REG_32, MIPS_AFL_REG_64): Likewise.
(MIPS_AFL_REG_128, MIPS_AFL_ASE_DSP, MIPS_AFL_ASE_DSP64): Likewise.
(MIPS_AFL_ASE_DSPR2, MIPS_AFL_ASE_EVA, MIPS_AFL_ASE_MCU): Likewise.
(MIPS_AFL_ASE_MDMX, MIPS_AFL_ASE_MIPS3D, MIPS_AFL_ASE_MT): Likewise.
(MIPS_AFL_ASE_SMARTMIPS, MIPS_AFL_ASE_VIRT): Likewise.
(MIPS_AFL_ASE_VIRT64, MIPS_AFL_ASE_MSA, MIPS_AFL_ASE_MSA64): Likewise.
(MIPS_AFL_ASE_MIPS16, MIPS_AFL_ASE_MICROMIPS): Likewise.
(MIPS_AFL_ASE_XPA, MIPS_AFL_EXT_XLR, MIPS_AFL_EXT_OCTEON2): Likewise.
(MIPS_AFL_EXT_OCTEONP, MIPS_AFL_EXT_LOONGSON_3A): Likewise.
(MIPS_AFL_EXT_OCTEON, MIPS_AFL_EXT_5900, MIPS_AFL_EXT_4010): Likewise.
(MIPS_AFL_EXT_4100, MIPS_AFL_EXT_3900, MIPS_AFL_EXT_10000): Likewise.
(MIPS_AFL_EXT_SB1, MIPS_AFL_EXT_4111, MIPS_AFL_EXT_4120): Likewise.
(MIPS_AFL_EXT_5400, MIPS_AFL_EXT_5500): Likewise.
(MIPS_AFL_EXT_LOONGSON_2E, MIPS_AFL_EXT_LOONGSON_2F): Likewise.
(Val_GNU_MIPS_ABI_FP_ANY, Val_GNU_MIPS_ABI_FP_DOUBLE): New enum values.
(Val_GNU_MIPS_ABI_FP_SINGLE, Val_GNU_MIPS_ABI_FP_SOFT): Likewise.
(Val_GNU_MIPS_ABI_FP_OLD_64, Val_GNU_MIPS_ABI_FP_XX): Likewise.
(Val_GNU_MIPS_ABI_FP_64, Val_GNU_MIPS_ABI_FP_64A): Likewise.
(Val_GNU_MIPS_ABI_FP_MAX): Likewise.
* sysdeps/mips/Makefile [subdir=elf]: Add tst-abi-interlink,
tst-mode-switch-1, tst-mode-switch-2, tst-mode-switch-3 tests.
* sysdeps/mips/bits/linkmap.h (struct link_map_machine): Add fpmode
field.
* sysdeps/mips/dl-machine.h (elf_machine_matches_host): Reject
EF_MIPS_FP64.
* sysdeps/mips/dl-machine-reject-phdr.h: New file.
* sysdeps/mips/tst-abi-fp32mod.c: Likewise.
* sysdeps/mips/tst-abi-fpxxmod.c: Likewise.
* sysdeps/mips/tst-abi-fpxxomod.c: Likewise.
* sysdeps/mips/tst-abi-fp64mod.c: Likewise.
* sysdeps/mips/tst-abi-fp64amod.c: Likewise.
* sysdeps/mips/tst-abi-interlink.c: Likewise.
* sysdeps/mips/tst-mode-switch-1.c: Likewise.
* sysdeps/mips/tst-mode-switch-2.c: Likewise.
* sysdeps/mips/tst-mode-switch-3.c: Likewise.
* sysdeps/unix/sysv/linux/mips/configure.ac (o32-fpabi): Define to
record the current FP ABI extension.
(mips-mode-switch): Define to show if kernel headers support mode
switching.
* sysdeps/unix/sysv/linux/mips/configure: Regenerate.
* sysdeps/unix/sysv/linux/mips/ldsodefs.h: Increase maximum
supported SYSV ABI version to 3.
* sysdeps/unix/sysv/linux/mips/libc-abis: Add new MIPS_O32_FP64
feature.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch continues removing architecture-specific cases from
non-architecture-specific files by moving the logic to use directories
such as /lib64 out of sysdeps/gnu/configure.ac.
A new macro LIBC_SLIBDIR_RTLDDIR is created that sysdeps configure
scripts can use to declare the library directories to be used; the
logic was previously duplicated in configure fragments for aarch64,
mips and x32 as well as in sysdeps/gnu/configure.ac. This macro is
used directly in sysdeps/gnu/configure.ac only to provide the /lib
default (the logic saying that with --prefix=/usr shared libraries go
in /lib not /usr/lib); the architecture cases formerly there are moved
into various new or existing configure.ac files. The new macro is
also used in the various architecture fragments that already had such
logic. In the x32 there was previously a configure fragment, but it
was a directly written one without a .ac file; now a .ac file is used
there instead to generate configure.
Tested x86_64 that the installed shared libraries, and the directory
structure of the installation, are unchanged by this patch.
There is an old bug report - bug 6441 - about library directories
changing after reconfiguring. If this is still applicable - and I
haven't attempted to confirm it or review the old patch pointed to in
that bug - then this patch should reduce the number of places needing
changing in any fix.
* aclocal.m4 (LIBC_SLIBDIR_RTLDDIR): New macro.
* sysdeps/gnu/configure.ac: Use LIBC_SLIBDIR_RTLDDIR. Remove
cases for individual architectures.
* sysdeps/gnu/configure: Regenerated.
* sysdeps/unix/sysv/linux/aarch64/configure.ac: Use
LIBC_SLIBDIR_RTLDDIR.
* sysdeps/unix/sysv/linux/aarch64/configure: Regenerated.
* sysdeps/unix/sysv/linux/mips/configure.ac: Use
LIBC_SLIBDIR_RTLDDIR.
* sysdeps/unix/sysv/linux/mips/configure: Regenerated.
* sysdeps/unix/sysv/linux/powerpc/powerpc64/configure.ac: Use
LIBC_SLIBDIR_RTLDDIR.
* sysdeps/unix/sysv/linux/powerpc/powerpc64/configure:
Regenerated.
* sysdeps/unix/sysv/linux/s390/s390-64/configure.ac: New file.
* sysdeps/unix/sysv/linux/s390/s390-64/configure: New generated
file.
* sysdeps/unix/sysv/linux/sparc/sparc64/configure.ac: New file.
* sysdeps/unix/sysv/linux/sparc/sparc64/configure: New generated
file.
* sysdeps/unix/sysv/linux/x86_64/64/configure.ac: New file.
* sysdeps/unix/sysv/linux/x86_64/64/configure: New generated file.
* sysdeps/unix/sysv/linux/x86_64/x32/configure.ac: New file.
* sysdeps/unix/sysv/linux/x86_64/x32/configure: Generate.
|
|
I've moved the MIPS port from ports to the main sysdeps hierarchy.
Beyond the README update, the move of the files was simply
git mv ports/sysdeps/mips sysdeps/mips
git mv ports/sysdeps/unix/mips sysdeps/unix/mips
git mv ports/sysdeps/unix/sysv/linux/mips sysdeps/unix/sysv/linux/mips
and in addition to the ChangeLog entries here, I put a note at the top
of ports/ChangeLog.mips similar to those in other files.
Tested that disassembly of installed shared libraries for mips is the
same before and after this patch (except for ld.so where paths in
assertions are involved, as for arm).
* sysdeps/mips: Move directory from ports/sysdeps/mips.
* sysdeps/unix/mips: Move directory from ports/sysdeps/unix/mips.
* sysdeps/unix/sysv/linux/mips: Move directory from
ports/sysdeps/unix/sysv/linux/mips.
* README: Update listing for mips-*-linux-gnu and
mips64-*-linux-gnu.
* sysdeps/mips: Move directory to ../sysdeps/mips.
* sysdeps/unix/mips: Move directory to ../sysdeps/unix/mips.
* sysdeps/unix/sysv/linux/mips: Move directory to
../sysdeps/unix/sysv/linux/mips.
|