Summary:
The Freescale e500v2 PowerPC core does not use a standard FPU.
Instead, it uses a Signal Processing Engine (SPE)--a DSP-style vector processor
unit, which doubles as a FPU. The PowerPC SPE ABI is incompatible with the
stock powerpc ABI, so a new MACHINE_ARCH was created to deal with this.
Additionaly, the SPE opcodes overlap with Altivec, so these are mutually
exclusive. Taking advantage of this fact, a new file, powerpc/booke/spe.c, was
created with the same function set as in powerpc/powerpc/altivec.c, so it
becomes effectively a drop-in replacement. setjmp/longjmp were modified to save
the upper 32-bits of the now-64-bit GPRs (upper 32-bits are only accessible by
the SPE).
Note: This does _not_ support the SPE in the e500v1, as the e500v1 SPE does not
support double-precision floating point.
Also, without a new MACHINE_ARCH it would be impossible to provide binary
packages which utilize the SPE.
Additionally, no work has been done to support ports, work is needed for this.
This also means no newer gcc can yet be used. However, gcc's powerpc support
has been refactored which would make adding a powerpcspe-freebsd target very
easy.
Test Plan:
This was lightly tested on a RouterBoard RB800 and an AmigaOne A1222
(P1022-based) board, compiled against the new ABI. Base system utilities
(/bin/sh, /bin/ls, etc) still function appropriately, the system is able to boot
multiuser.
Reviewed By: bdrewery, imp
Relnotes: yes
Differential Revision: https://reviews.freebsd.org/D5683
Bring older verbatim version of cpuid.h
This file is used regularly in FreeBSD builds but we usually use the
similar file provided by clang.
By providing the older file introduced in GCC 4.3, we hope to mimic
better what is provided by an external toolchain.
Obtained from: GCC-4_3-branch (SVN rev. 129548, pre GPLv3)
We brought an original __COUNTER__ implementation in r228474, however, it
was missing documentation and it had a different behaviour for precompiled
headers with respect to the upstream version. Since the upstream version
is under the same license as GCC4.2, bring the missing pieces to reduce
differences against upstream.
Optained from: GCC pre-4.3 (rev. 125041 ; GPLv2)
llvm libunwind includes a libunwind.cpp, but on ARM libunwind.S is found
first in .PATH. Rename the latter one, since it is not going to be
updated again.
Reviewed by: andrew
MFC after: 3 days
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D7162
DR #289[0] came down and gcc4.2.1 was on the wrong side of history.
Partially revert GCC r42574 (just remove the error) to rectify the parse
bug to match Clang and other compliant C99 compilers.
An example declaration gcc tripped on before this fix:
void foobar(int [static 1]);
An example declaration gcc did not trip on before this fix:
void foobar(int name[static 1]);
Bump __FreeBSD_cc_version.
[0]: http://www.open-std.org/JTC1/SC22/WG14/www/docs/dr_289.htm
Reported by: allanjude
Sponsored by: EMC / Isilon Storage Division
This is a C11 feature that is starting to get used in places such as Mesa.
This implementation takes a different approach to upstream and is
therefore not covered by GPLv3.
Obtained from: OpenBSD (CVS rev. 1.2)
MFC after: 3 weeks
This is required to build csu.
Reviewed by: andrew
Sponsored by: DARPA, AFRL
Sponsored by: HEIF5
Differential Revision: https://reviews.freebsd.org/D5039
From OpenBSD's commit log:
This was responsible for memory corruption with recent versions
of Mesa where c and c++ code share a header with a packed enum type.
Reference:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=39219
Obtained from: OpenBSD (CVS rev. 1.2)
MFC after: 1 week
This includes additional functions to be protected: those that
have local array definitions, or have references to local frame
addresses. This is a new option in GCC-4.9 that was relicensed
by Han Shen from Google under GPLv2 for OpenBSD.
Obtained from: OpenBSD (2014-01-14)
MFC after: 2 weeks
Some point after gcc-4.2 the MIPS inline assembly restrictions changed -
=h (hi register) disappeared from the list of restrictions and can no
longer be used.
So, until someone requires an assembly version of this function,
just use a non-assembly version and let the compiler sort it out.
Suggested by: kan
The __builtin_init_dwarf_reg_size_table function is unimplemented in
clang 3.6 for AArch64. Comment it out for now and replace it with
a message and abort.
Tracked in upstream LLVM PR 22997
https://llvm.org/bugs/show_bug.cgi?id=22997
Submitted by: andrew
This will result in __ARM_ARCH_7A__ being defined during the compile.
When compiling with gcc, it will still only generate armv6 opcodes itself,
but should pass the arch to gas so that inline asm can use v7 opcodes.
PR c++/48211
* name-lookup.h (cp_class_binding): Make base a pointer.
* name-lookup.c (new_class_binding): Adjust.
(poplevel_class): Adjust.
This fixes a potential segfault when compiling gold, a part of the
devel/binutils port, with gcc. See also the upstream bug report:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=48211
Thanks to Jason Merrill, Tom Callaway and Red Hat legal for approving
the use of this patch under the GNU GPL, version 2 or later.
MFC after: 1 week
the first argument of the following builtin function:
* __builtin_ia32_psrlqi128() takes __v2di instead of __v4si
This should fix the following errors when building the graphics/webp
port with base gcc:
lossless_sse2.c:403: error: incompatible type for argument 1 of '__builtin_ia32_psrlqi128'
lossless_sse2.c:404: error: incompatible type for argument 1 of '__builtin_ia32_psrlqi128'
Reported by: Jos Chrispijn <ports@webrz.net>
MFC after: 3 days
From the OpenBSD log:
x86-64 ABI requires arrays greater than 16 bytes to be aligned to
16byte boundary. However, GCC 16-byte aligns arrays of >=16 BITS,
not BYTES.
This diff improves bug detectability for code which has local arrays
of [16 .. 127] bits: in those cases SSP will now detect even 1-byte
overflows.
Obtained from: OpenBSD (CVS rev 1.4)
MFC after: 1 week
This mimics the behaviour in clang and lets us build cleanly
the libdispatch port on platforms where the base gcc is still
the default compiler.
Bump __FreeBSD_version for ports.
Tested by: theraven
MFC after: 3 days
all of the features in the current working draft of the upcoming C++
standard, provisionally named C++1y.
The code generator's performance is greatly increased, and the loop
auto-vectorizer is now enabled at -Os and -O2 in addition to -O3. The
PowerPC backend has made several major improvements to code generation
quality and compile time, and the X86, SPARC, ARM32, Aarch64 and SystemZ
backends have all seen major feature work.
Release notes for llvm and clang can be found here:
<http://llvm.org/releases/3.4/docs/ReleaseNotes.html>
<http://llvm.org/releases/3.4/tools/clang/docs/ReleaseNotes.html>
MFC after: 1 month
Support for warnings about missing prototypes in C++ was added by Apple
GCC (Radar 6261539). Most of the code crept into r260311 so it felt
natural to make use of it.
Obtained from: Apple GCC - 5646
MFC after: 5 days
Among some of the objc changes from Apple that crept into r260311,
Radar 5355344 is incomplete and is not used since we don't carry
ObjC in the base system.
The dead code seems to have caused issues in some Tinderboxes so
get rid of it altogether.
Reported by: luigi
MFC after: 9 days
GCC-PR rtl-optimization/34628
* combine.c (try_combine): Stop and undo after the first combination
if an autoincrement side-effect on the first insn has effectively
been lost.
The issue was detected in OpenBSD but their fix was not very good. Huge
thanks to the upstream author, Eric Botcazou, for permitting the use of
this patch under GPLv2.
MFC after: 5 days
Block objects [1] are a C-level syntactic and runtime feature. They
are similar to standard C functions, but in addition to executable
code they may also contain variable bindings to automatic (stack)
or managed (heap) memory. A block can therefore maintain a set of
state (data) that it can use to impact behavior when executed.
This port is based on Apple's GCC 5646 with some bugfixes from
Apple GCC 5666.3. It has some small differences with the support
in clang, which remains the recommended compiler.
Perhaps the most notable difference is that in GCC that __block
is not actually a keyword, but a macro. There will be workaround
for this issue in a near future. Other issues can be consulted in
the clang documentation [2]
For better compatiblity with Apple's GCC and llvm-gcc some related
fixes and features from Apple have been included. Support for the
non-standard nested functions in GCC is now off by default.
No effort was made to update the ObjC support since FreeBSD doesn't
carry ObjC in the base system, but some of the code crept in and
was more difficult to remove than to adjust.
Reference:
[1]
https://developer.apple.com/library/mac/documentation/Cocoa/Conceptual/Blocks/Articles/00_Introduction.html
[2]
http://clang.llvm.org/compatibility.html#block-variable-initialization
Obtained from: Apple GCC 4.2
MFC after: 3 weeks
Unfortunately this causes ICE on powerpc and sparc64.
Reducing these differences against upstream is not important
anymore so hopefully I have finished breaking the compiler
occasionally.
Apple GCC has extensions to support for both label attributes and
an "unavailable" attribute. These are critical for objc but are
also useful in regular C/C++.
Apparently at least the label attributes might have found their way to
upstream GCC but the code doesn't seem available on the GPLv2 tree so
we are taking the code directly from Apple. To make this clearer we
are preserving the annoying "APPLE LOCAL" tags and the ChangeLogs
when they are available.
Obtained from: Apple GCC 4.2 - 5531
MFC after: 3 weeks
Very small updates: fixes GCC-PR target/31152
Tested by building the cross-compiler.
Obtained from: gcc 4.3 (rev. r118461, 125973: GPLv2)
MFC after: 2 weeks
Fixes GCC libstdc++/29286
The fix seems to work for amd64 but causes segfaults on powerpc.
At this time gcc is much more important on powerpc than on amd64.
Reported by: andreast
