Otherwise, clang can effectively remove the first iteration of the for
loops where this macro is invoked, and as a result, "cmp r0, #99" fails
to assemble.
Obtained from: joerg at netbsd
MFC after: 3 days
- Dump an NT_X86_XSTATE note if XSAVE is in use. This note is designed
to match what Linux does in that 1) it dumps the entire XSAVE area
including the fxsave state, and 2) it stashes a copy of the current
xsave mask in the unused padding between the fxsave state and the
xstate header at the same location used by Linux.
- Teach readelf() to recognize NT_X86_XSTATE notes.
- Change PT_GET/SETXSTATE to take the entire XSAVE state instead of
only the extra portion. This avoids having to always make two
ptrace() calls to get or set the full XSAVE state.
- Add a PT_GET_XSTATE_INFO which returns the length of the current
XSTATE save area (so the size of the buffer needed for PT_GETXSTATE)
and the current XSAVE mask (%xcr0).
Differential Revision: https://reviews.freebsd.org/D1193
Reviewed by: kib
MFC after: 2 weeks
The powerpc support was the only supported architecture not prepending the elf format name
with "-freebsd" in base this change makes it consistent with other architectures.
On newer version of binutils the powerpc format is also prepended with "-freebsd".
Also modify the kernel ldscripts in that regards.
As a result it is now possible cross build the kernel on powerpc using newer binutils
Differential Revision: https://reviews.freebsd.org/D926
Differential Revision: https://reviews.freebsd.org/D928
names that must nnot be adjusted. This fixes a bug where code such as:
movw r2, :lower16:symbol
movt r2, :upper16:symbol
It is common for clang to generate such code when targeting armv7.
including an alignment. Previously binutils would only allow instructions
in the form "vld1.64 {d0, d1}, [r0, :128]" where the final comma should
not be there, instead the above instruction should be
"vld1.64 {d0, d1}, [r0:128]".
This change duplicates the alignment code from within the function to
handle this case.
libraries that need to be linked into an executable or library have to be
listed on the command line explicitly. This commit fixes a bug in ld(1)
where it would scan dependencies of the libraries on the command line and
link them if needed if they were also found in ld.so.cache.
The important bit of the patch is the initialisation of needed.by such that
libraries found by scanning dependencies are marked as such and not used in
the link.
The patch is a backport of binutils git commit
d5c8b1f8561426b41aa5330ed60f578178fe6be2
The author gave permission to use it under GPLv2 terms.
PR: 192062
Exp-run by: antoine
MFC after: 1 week
the same values as the -march= command line option. Add support for the
"sec" extension (security extensions).
We've been getting away without support for the sec extension because
it's bogusly enabled even on arches where its presence is optional. This
support for .arch_extension is being added mainly so that we can use the
right directives in our source code, and that helps folks using external
toolchains (and will help us when we finally update our toolchain).
instructions. Partially obtained from OpenBSD by Pedro Giffuni, while I
added the fcomip variants.
Apparently this should help with compiling certain variants of WebKit.
MFC after: 3 days
This targets the existing ARMv6 and ARMv7 SoCs that contain a VFP unit.
This is an optional coprocessors may not be present in all devices, however
it appears to be in all current SoCs we support.
armv6hf targets the VFP variant of the ARM EABI and our copy of gcc is too
old to support this. Because of this there are a number of WITH/WITHOUT
options that are unsupported and must be left as the default value. The
options and their required value are:
* WITH_ARM_EABI
* WITHOUT_GCC
* WITHOUT_GNUCXX
In addition, without an external toolchain, the following need to be left
as their default:
* WITH_CLANG
* WITH_CLANG_IS_CC
As there is a different method of passing float and double values to
functions the ABI is incompatible with existing armv6 binaries. To use
this a full rebuild of world is required. Because no floating point values
are passed into the kernel an armv6 kernel with VFP enabled will work with
an armv6hf userland and vice versa.
Add support for stac/clac instructions to manipulate the flag
that controls the behaviour of Intel's Supervisor Mode Access
Prevention (SMAP) feature.
Tested by: dim
Obtained from: OpenBSD
MFC after: 5 days
64-bit debug data is only necessary for objects with greater than 4GB of
debug data, and is not used on other 64-bit FreeBSD targets.
Sponsored by: DARPA, AFRL
Fix the VCVT instruction. It must round towards zero when converting from
a floating-point to an integer value. This was not the case causing issues
when printing certain values.
There is a VCVTR instruction that will round depending on the current
rounding mode. We don't yet support this instruction, or setting the
rounding mode.
The header_length field is the number of bytes following the field to
the first byte of the line number program. The hard-coded constants
previously here (4 + 2 + 4) were correct only for 32-bit DWARF.
Sponsored by: DARPA, AFRL
On MIPS .dynamic is read-only and so a special section .rld_map is used
to store the pointer to the rtld information for debuggers. This
section had a hard coded size of 4 bytes which is not correct for
mips64. (Note that FreeBSD's rtld does not yet populate .rld_map.)
Sponsored by: DARPA, AFRL
Approved by: re (delphij)
linking of a shared library, leading to corrupt indexes in the dynamic
symbol table. This should fix the multimedia/ffmpegthumbnailer port.
Approved by: re (kib)
Reported by: swills
MFC after: 2 weeks
for. This is useful for software needing to know which architecture a
binary is built for as arm and armv6 have slight differences meaning only
some binaries build for one will work as expected on the other. It is
expected pkgng will be able to make use of this to simplify the logic to
determine which package ABI to use.
Approved by: re (kib)
smlal, and umlal the output registers are allowed to be the same as either
input registers, where in ARMv4 and ARMv5 they could only be the same as the
last input register.
This is the default behaviour of the newer binutils as well as most alternative linkers.
All the ports tree has been fixed to be able to link properly with this new behaviour.
Add a function to return the specific type, when the note's Name field is
'FreeBSD'.
r249558 added FreeBSD-specific ELF note types that reuse type numbers of
existing generic / Linux types. This caused 'readelf -n' to produce
incorrect output on FreeBSD core files.
Sponsored by: DARPA, AFRL
MFC after: 3 days
DW_FORM_flag_present dwarf attribute, so they do not print errors or
warnings on files that contain it. (This attribute can be emitted by
newer versions of clang and gcc.)
MFC after: 1 week
register added to the symbol table by the assembler. On further
investigation it was found the problem was with the my_get_expression
function. This is called by parse_big_immediate.
Fix this by moving the call to parse_big_immediate to the end of the if,
else if, ..., else block.
This fixes the problem on amd64 miscompiling mpboot.s causing boot
issues... We are still using gas for a few files in the kernel...
Submitted by: kib
MFC after: 1 month
Thanks to Mike Belopuhov for the pointer to the OpenBSD patch, though
OpenBSD's gcc is very different that it only helped w/ where to modify,
not how... Thanks to jhb for some early reviews...
Reviewed by: imp, kib
MFC after: 1 month
.note.ABI-tag section.
This helps on ARM EABI where the OS/ABI field is zero. It would be better
to use the NOTES program header however this would require a more invasive
change.
should never do so. This can cause global constructors and destructors
to not be executed at run-time, resulting in crashes and other strange
behaviour.
Reported by: rene
MFC after: 1 week
share/mk/sys.mk instead.
This is part of a medium term project to permit deterministic builds of
FreeBSD.
Submitted by: Erik Cederstrand <erik@cederstrand.dk>
Reviewed by: imp, toolchain@
Approved by: cperciva
MFC after: 2 weeks
encounters a DT_RUNPATH entry, the global dynamic_info[] array is
overrun, causing some other global variable to be overwritten.
In my testcase, this was the section_headers variable, leading to
segfaults or jemalloc assertions when it was freed later on.
Thanks to Koop Mast for providing samples of a few "bad" .so files.
MFC after: 1 week
r238211:
Support TARGET_ARCH=armv6 and TARGET_ARCH=armv6eb
This adds a new TARGET_ARCH for building on ARM
processors that support the ARMv6K multiprocessor
extensions. In particular, these processors have
better support for TLS and mutex operations.
This mostly touches a lot of Makefiles to extend
existing patterns for inferring CPUARCH from ARCH.
It also configures:
* GCC to default to arm1176jz-s
* GCC to predefine __FreeBSD_ARCH_armv6__
* gas to default to ARM_ARCH_V6K
* uname -p to return 'armv6'
* make so that MACHINE_ARCH defaults to 'armv6'
It also changes a number of headers to use
the compiler __ARM_ARCH_XXX__ macros to configure
processor-specific support routines.
Submitted by: Tim Kientzle <kientzle@freebsd.org>
adding appropriate table entries, the assembler had to be adjusted as
these are the first non-SSE instructions to use a 3-byte opcode (and a
mandatory prefix to boot).
MFC after: 1 month
instructions. I reimplemented this from scratch based on the Intel
manuals and the existing support for handling the fxsave and fxrstor
instructions. This will let us use these instructions natively with GCC
rather than hardcoding the opcodes in hex.
Reviewed by: kib
MFC after: 1 month
The binutils update in r218822 caused the MIPS n64 dynamic binaries to
fail because the ".interp" section is not in the initial sections.
This happens because elf64bmip-defs.sh overrides INITIAL_READONLY_SECTIONS
to add ".MIPS.options" sections instead of the ".reginfo" section used
by n32.
This used to work fine, but after r218822, INITIAL_READONLY_SECTIONS also
contains the .interp section, so the override has to be done differently.
Reported by : aduane at juniper
Obtained from: gonzo (Initial version)
* elf64-sparc.c (sparc64_elf_relocate_section): Adjust addend of
dynamic relocs against section symbols for the output section vma.
However, with the addition of TLS support in the upstream rev. 1.104
this fix was essentially reverted. After factoring out the common parts
of elf32-sparc.c and elf64-sparc.c a comment was added to elfxx-sparc.c
in the upstream rev. 1.27 as part of unrelated changes, saying that the
fix from elf64-sparc.c rev. 1.61 indeed should be implemented, but given
that some unspecified OS has a broken ld.so expecting broken relocations
deliberately is omitted.
As the current behavior actually violates the SPARC ABI, FreeBSD never
had such a broken ld.so and this is actually causing problems with at
least kernel modules linked with binutils 2.17.50 committed in r218822
without the workaround committed in r219340 in place, re-implement the
above fix in a way so that is only applied if the output format is
ELFOSABI_FREEBSD. In the upstream version it probably would make sense
to invert this check and only skip adjusting the addend for the OS with
the broken ld.so, once it's determine which one that is.
Approved by: dim
x86 CPU support, better support for powerpc64, some new directives, and
many other things. Bump __FreeBSD_version, and add a note to UPDATING.
Thanks to the many people that have helped to test this.
Obtained from: projects/binutils-2.17
'linker stubs'. Add .note.GNU-stack for the stubs objects. Without this,
final binary will have RWE mode for PT_GNU_STACK regardless of the
actual requirements.
Tested by: nwhitehorn
Reviewed by: dim, nwhitehorn
it should also be MI. The problem here arises when ld ends up linking a
link-once section with relocations against sections that point back to it
that are as yet unresolved. Instead of piecemeal finding sections we
think are potentially subject to this issue, just defer processing for
sections that have yet to be relocated instead of immediately bailing.
thread specific informations.
In order to do that, and in order to avoid KBI breakage with existing
infrastructure the following semantic is implemented:
- For live programs, a new member to the PT_LWPINFO is added (pl_tdname)
- For cores, a new ELF note is added (NT_THRMISC) that can be used for
storing thread specific, miscellaneous, informations. Right now it is
just popluated with a thread name.
GDB, then, retrieves the correct informations from the corefile via the
BFD interface, as it groks the ELF notes and create appropriate
pseudo-sections.
Sponsored by: Sandvine Incorporated
Tested by: gianni
Discussed with: dim, kan, kib
MFC after: 2 weeks
into it. Prior to this commit the .gnu_debuglink section can have up
to 3 bytes of uninitialized garbage; as a result, .ko files could
change vary between builds.
Approved by: dim
MFC after: 7 days
us up to version 2.17.50.20070703, at the last GPLv2 commit.
Amongst others, this added upstream support for some FreeBSD-specific
things that we previously had to manually hack in, such as the OSABI
label support, and so on.
There are also quite a number of new files, some for cpu's (e.g. SPU)
that we may or may not be interested in, but those can be cleaned up
later on, if needed.
64-bit PowerPC when linking multiple C++ files referencing the same
method, defined in a common header, when that method had a switch
statement with more than 4 cases. This change fixes compilation of LLVM
tblgen on 64-bit PPC with binutils 2.17.
Lots of help from: dim
Upstream after: more testing
(still under GPLv2 at that time):
Author: H.J. Lu <hjl@lucon.org>
Date: Wed Sep 27 04:18:16 2006 +0000
PR ld/3223
PR ld/3267
* bfd/elf.c (assign_file_positions_for_non_load_sections): Don't warn
zero size allocated sections.
* ld/ldlang.h (lang_output_section_statement_type): Add
section_relative_symbol.
* ld/ldlang.c (strip_excluded_output_sections): Don't strip a section
with a symbol relative to it.
(lang_size_sections_1): Mark if an output section has a symbol symbol
relative to it.
This prevents warnings like the following during stripping of debug info
from kernel modules on i386:
===> zlib (all)
...
objcopy --only-keep-debug zlib.ko.debug zlib.ko.symbols
objcopy --strip-debug --add-gnu-debuglink=zlib.ko.symbols zlib.ko.debug zlib.ko
BFD: zlib.ko: warning: allocated section `.plt' not in segment
BFD: zlib.ko: warning: allocated section `.got' not in segment
(still under GPLv2 at that time):
Author: Nick Clifton <nickc@redhat.com>
Date: Thu Aug 24 14:59:24 2006 +0000
* ldlang.c (lang_size_sections_1, lang_assignment_statement_enum):
Adjust the current address of DEFAULT_MEMORY_REGION even when dot
hasn't changed.
This is a prerequisite for the fix coming just after this.
moved from ld/emultempl/elf32.em to ld/ldlang.c, so apply approximately
the same change as upstream, which has description:
* ldlang.c (lang_insert_orphan): Add __start_<section> symbol assignment
inside output section statement. Ensure only one set of symbols per
output section.
* emultempl/pe.em (gld_${EMULATION_NAME}_place_orphan): Add non-dollar
sections before dollar sections. Correct add_child list insertion.
Taken from upstream git commit 7e01d69a19a8fd079887f26853c8565da15ff340,
with permission to use it under GPLv2 from the author.
The change made to bfd/elf.c in upstream revision 1.217.4.3 (which was a
revert of an earlier change), caused objcopy on powerpc to fail to copy
debug info from kernel modules. This had to be fixed by applying the
diff from upstream revision 1.243 on top of it.
__start_SECNAME and __stop_SECNAME symbols are automatically generated
by ld for orphan sections, i.e. those not explicitely referenced by a
linker script. The symbols are supposed to be placed correspondingly
at the start and the end of the section in output file. In some cases
__start_SECNAME may be placed at the address after the end of the
previous section (if any) and before the start the section. This
happens when following conditions are met:
1. the orphan section is found in more than one input file
2. the orphan section has different alignment requirements across input
files
3. the first instance of the section encountered doesn't have the
greatest alignment requirement
In these conditions resulting output section will be placed at address
after the end of the previous section aligned to the greatest alignment
requirement in the inputs, but __start_SECNAME will be placed at address
after the end of the previous section aligned to the alignment
requirement of the first input in which the section is encountered.
See commit message of r196118 for a concrete example of problems caused
by this bug.
The fix is to place __start_SECNAME inside the section and use ABSOLUTE
directive, rather than placing __start_SECNAME outside the section and
trying to guess address alignment.
This fix is in line with upstream binutils change/fix made between
versions 2.19 and 2.20 in revision of 1.307 ldlang.c.
MFC after: 3 weeks
