Incorrect codegen from rdseed intrinsic use (CVE-2017-11671)
We should not expand call arguments in between flags reg setting and
flags reg using instructions, as it may expand with flags reg
clobbering insn (ADD in this case).
Attached patch moves expansion out of the link. Also, change
zero-extension to non-flags reg clobbering sequence in case we perform
zero-extension with and.
2017-03-25 Uros Bizjak
(NetBSD/mips64 progress further with gsed)
Adding gsed as a tool creates circular dependencies if pkgsrc gcc is used
as the bootstrap compiler.
Pointed out by jperkin, sorry.
This is a somewhat blind commit. I've long had issues on various platforms
with libgcc getting misconfigured (on netbsd/mips,arm,powerpc), for example
build failures see:
http://mail-index.netbsd.org/port-mips/2017/06/21/msg000832.html
testing GCC 7.1 with netbsd/mips64el I got a lot further using it, but
still didn't complete the build. It took hours to reach this failure,
so I'd rather blindly commit the same change in the hopes it might help
other architectures.
applied to GCC 8 and above - thanks Krister Walfridsson)
libstdc++: if it thinks vswprintf is broken, don't block the
entire definition of to_string, just that of to_wstring. backport
of GCC r223830 already applied to GCC 4.9. this applies to netbsd-6-0.
Doesn't fix any build I wanted to fix, but it does get the build
of LLVM4.x further on netbsd-6-0. maybe there's an odd package that
could benefit from it.
bump PKGREVISION.
configs for gcc on netbsd don't build -march=native support, but this
stray prototype was left, causing build errors:
gcc.o:(.rodata+0x58c4): undefined reference to
`host_detect_local_cpu(int, char const**)'
pointed out by jperkin, thanks
fix netbsd gcc-java option build
netbsd has gethostbyname_r in libc, but not in any headers.
configure test succeeds, but compilation fails.
workaround with passing ac_cv_func_gethostbyname_r=no to autoconf
thanks joerg for the suggestion.
Reset PKGREVISION, set gcc48-libs PKGREVISION to 1 (+1 of gcc48)
The following list of 74 PRs have been resolved in the 4.8.5 release
(non-exhaustive list):
64557 get_addr in true_dependence_1 cannot handle VALUE inside an expr
66215 [4.8/4.9/5/6 Regression] Wrong after label NOP emission for -mhotpatch
66233 [4.8/4.9 Regression] internal compiler error: in expand_fix
65550 [4.8/4.9 Regression] ICE (segfault) with pch
62642 [4.8/4.9 Regression] x86 rdtsc is moved through barrier
63608 [4.8/4.9 Regression] error: type mismatch in binary expression
59626 [4.8 Regression] /usr/include/bits/unistd.h:173:1: error: inlining failed in call to always_inline 'readlinkat': recursive inlining
65549 [4.9 Regression] crash in htab_hash_string with -flto -g
59990 [4.8 regression] incorrect memcpy optimization
60656 [4.8 regression] x86 vectorization produces wrong code
66470 [4.8/4.9/5/6 Regression] TLS ICE due to ix86_split_long_move
66481 Implicit lambda capture of variable named ""closure"" causes crash
66275 __attribute__((sysv_abi)) with x86_64-w64-mingw32-gcc generates incorrect code
65680 [4.8/4.9 Regression] ICE at -O1 and above on x86_64-linux-gnu in expand_assignment
65368 [4.8/4.9 Regression]_bzhi_u32 intrinsic generates incorrect code when -O1 or above is specified and index is an immediate
64634 [4.8/4.9 Regression] gcov reports catch(...) as not executed
66123 [4.8 Regression] Array of labels as values + ternary operator + pointer arithmetic = internal compiler error
65518 [4.8 Regression] gcc consumes all memory with -O3
66140 ICE at extract_insn
65721 [4.8/4.9/5/6 Regression] Internal compiler error segmentation fault
65727 [4.8/4.9/5/6 Regression] Segfault With Decltype In Lambda Expression Used To Initialize Static Class Member
59016 f951: internal compiler error: Segmentation fault
65693 [4.8/4.9 Regression] ICE in assign_by_spills
61977 [4.8/4.9/5 Regression] powerpc preprocessor breaks on lines that end with ""vector
59513 [4.8/4.9/5 Regression] Fortran runtime error: Sequential READ or WRITE not allowed after EOF marker
65279 [4.8/4.9 Regression] std::scoped_allocator_adaptor is not assignable
65543 rvalue stream insertion and extraction operators incorrectly implemented
61138 [4.8/4.9/5 Regression] Wrong code with pointer-bounds remapping
65072 Segfault when parsing dectlype in trailing return type
65327 GCC rejects ""constexpr volatile int i = 5;
65409 [4.8/4.9/5 Regression] ICE in store_field
52579 [4.8/4.9/5 regression] i386_w32_fallback_frame_state should care ffi raw-closure stub function
60898 [4.8/4.9/5 Regression] model compile error with gfortran 4.7 and gcc 4.9
57059 [4.8/4.9/5 Regression] Host configuration of loose_warn breaks for build components for Canadian crosses
52664 [4.8 Regression]: gcc.dg/tree-ssa/pr31261.c fails
65220 [4.8/4.9/5 Regression] integer division in stack alignment for VLA allocation
43701 [4.8/4.9/5 Regression] ICE: SIGSEGV (too deep recursion) with -mno-sse and __float128
35330 [4.8/4.9/5 regression] ICE with invalid pragma weak
46102 [4.8/4.9/5 Regression] ICE: SIGSEGV in dwarf2out_finish (dwarf2out.c:8490) with -feliminate-dwarf2-dups when using precompiled headers
61634 [4.8 Regression] ICE in in vect_get_vec_def_for_operand
65063 [4.8 Regression] gcc.dg/vect/vect-double-reduc-6.c FAILs with -O3 -fno-tree-loop-ivcanon -fno-tree-vectorize
63593 ICE: verify_gimple failed: incompatible types in PHI argument 0 with -O3 -fno-tree-vectorize
63844 [4.8 Regression] open mp parallelization prevents vectorization
64493 [4.8 Regression] ICE at -O3 on x86_64-linux-gnu
64495 [4.8 Regression] ICE at -O3 for trunk and wrong code for 4.8/4.9 on x86_64-linux-gnu
64199 [4.8 Regression] ICE: tree check: expected class 'constant'
56273 [4.8 regression] Bogus -Warray-bounds warning
58123 [4.8/4.9/5 Regression] debug line not tracked for last autovariable dtor
45402 [4.8/4.9/5 Regression] ICE in extract_insn
64768 [4.8/4.9/5 Regression] internal compiler error: tree check: expected tree that contains ‘decl with RTL’ structure
63744 [4.8/4.9/5 Regression] Duplicate use-statement causes error
64979 [4.8 Regression] stdarg optimization not able to find escape sites in phi nodes
57653 [4.8/4.9 Regression] filename information discarded when using -imacros
60718 [4.8/4.9 regression] Test case gfortran.dg/select_type_4.f90 fails on ARM
64479 [4.8 Regression][SH] wrong optimization delayed-branch
39722 [4.8/4.9 Regression][cond-optab] worse code with bitfields on v850
64882 ICE on valid code at -O3 with -g enabled in simplify_subreg
61058 [4.8/4.9 Regression] ICE: RTL check: expected elt 3 type 'B'
64513 [4.8/4.9 Regression] ICE: in maybe_record_trace_start
64766 [4.8/4.9 Regression] internal compiler error: tree check: expected block
64487 [4.8/4.9 Regression] internal compiler error: in fold_offsetof_1
62044 [4.8/4.9 Regression] ICE in USE statement with RENAME for extended derived type
57748 [4.8 Regression] ICE when expanding assignment to unaligned zero-sized array
57023 [4.8 Regression] Not packing arrays with changing variable used for size
52306 [4.8 regression] ICE in cselib_record_set
52714 [4.8 regression] ICE in fixup_reorder_chain
58369 [4.8 regression] ICE in subreg_get_info when compiling boost for m68k-linux
63733 [4.8/4.9/5 Regression] [OOP] wrong resolution for OPERATOR generic
netbsd only: link against libc when building shared.
libssp is broken as it has headers like ssp/stdio.h which include
<ssp.h>, when only ssp/ssp.h exists.
simply building hello world (including stdio.h) with -D_FORTIFY_SOURCE
will fail.
netbsd building of shared was broken as it did not link against libc when
making shared. adjust the spec file so it will do this.
bump PKGREVISION
XXX - all the other versions of GCC likely need the same
thanks joerg for the help
The find-prefix infrastructure was required in a pkgviews world where
packages installed from pkgsrc could have different installation
prefixes, and this was a way for a dependency prefix to be determined.
Now that pkgviews has been removed there is no longer any need for the
overhead of this infrastructure. Instead we use BUILDLINK_PREFIX.pkg
for dependencies pulled in via buildlink, or LOCALBASE/PREFIX where the
dependency is coming from pkgsrc.
Provides a reasonable performance win due to the reduction of `pkg_info
-qp` calls, some of which were redundant anyway as they were duplicating
the same information provided by BUILDLINK_PREFIX.pkg.
* Pull in patch from MacPorts to handle 10.11 SDK header issue
with fixincludes.
* Backport upstream bugfix for GCC bug 66523, fixing an issue
with Xcode 7.
* Pull in --with-build-config=bootstrap-debug change from FreeBSD
ports to fix bootstrap compare issue seen with clang / Xcode 7.
necessarily compatible (E.g. binutils-2.25 does not define PARAMS). Adjust
the include path priority so the internal ansidecl.h gets precedence, allowing
cp-demangle.c to use libiberty.h without compilation errors.
The removed hunk is definitely wrong, but I can't figure out what it was
trying to do. I think the patch hunk was accidentally reintroduced in r1.4
after being correctly removed in r1.3.
The reason it is wrong is because it breaks later tests by introducing code
into confdefs.h. The following tests always break, because they have
duplicate main() definitions.
Build test on NetBSD works because the java support isn't in the suggested
options anyway.
* Fixes regressions, revive parts removed at "Fix build under OpenBSD/amd64 5.5"
* Drop default behaviour changes and local features from OpenBSD ports,
it break other platforms, at least, NetBSD-6.0/i386.
* Remove needless patches after "Restore monolithic gcc48 and gcc48-libs".
* Convert gcc48 to simple meta-pkg
* Potentially fix non-NetBSD packaging
* Fix gcc48-libs/gcc48-cc++ build under OpenBSD/amd64 5.5
* Enable graphite option, from wip/gcc48snapshot via Xiyue Deng
* Tweak nls option
Changelog:
GCC 4.8.3
This is the [35]list of problem reports (PRs) from GCC's bug tracking
system that are known to be fixed in the 4.8.3 release. This list might
not be complete (that is, it is possible that some PRs that have been
fixed are not listed here).
Support for the new powerpc64le-linux platform has been added. It
defaults to generating code that conforms to the ELFV2
Do it for all packages that
* mention perl, or
* have a directory name starting with p5-*, or
* depend on a package starting with p5-
like last time, for 5.18, where this didn't lead to complaints.
Let me know if you have any this time.
