runtime: Fix GC bug caused by Entersyscall modifying reg.
This patch fixes a rare but serious bug. The Go garbage
collector only examines Go stacks. When Go code calls a
function that is not written in Go, it first calls
syscall.Entersyscall. Entersyscall records the position of
the Go stack pointer and saves a copy of all the registers.
If the garbage collector runs while the thread is executing
the non-Go code, the garbage collector fetches the stack
pointer and registers from the saved location.
Entersyscall saves the registers using the getcontext
function. Unfortunately I didn't consider the possibility
that Entersyscall might itself change a register before
calling getcontext. This only matters for callee-saved
registers, as caller-saved registers would be visible on the
saved stack. And it only matters if Entersyscall is compiled
to save and modify a callee-saved register before it calls
getcontext. And it only matters if a garbage collection
occurs while the non-Go code is executing. And it only
matters if the only copy of a valid Go pointer happens to be
in the callee-saved register when Entersyscall is called.
When all those conditions are true, the Go pointer might get
collected incorrectly, leading to memory corruption.
This patch tries to avoid the problem by splitting
Entersyscall into two functions. The first is a simple
function that just calls getcontext and then calls the rest of
Entersyscall. This should fix the problem, provided the
simple Entersyscall function does not itself modify any
callee-saved registers before calling getcontext. That seems
to be true on the systems I checked. But since the argument
to getcontext is an offset from a TLS variable, it won't be
true on a system which needs to save callee-saved registers in
order to get the address of a TLS variable. I don't know why
any system would work that way, but I don't know how to rule
it out. I think that on any such system this will have to be
implemented in assembler. I can't put the ucontext_t
structure on the stack, because this function can not split
stacks, and the ucontext_t structure is large enough that it
could cause a stack overflow.
Jakub Jelinek [Thu, 6 Mar 2014 08:11:20 +0000 (09:11 +0100)]
backport: re PR c/37743 (Bogus printf format warning with __builtin_bswap32.)
Backport from mainline
2014-02-19 Jakub Jelinek <jakub@redhat.com>
PR c/37743
* c-common.c (c_common_nodes_and_builtins): When initializing
c_uint{16,32,64}_type_node, also set corresponding
uint{16,32,64}_type_node to the same value.
* g++.dg/ext/builtin-bswap1.C: New test.
* c-c++-common/pr37743.c: New test.
Jakub Jelinek [Thu, 6 Mar 2014 08:10:08 +0000 (09:10 +0100)]
backport: re PR preprocessor/58844 (ICE with invalid use of ##)
Backport from mainline
2014-02-19 Jakub Jelinek <jakub@redhat.com>
PR preprocessor/58844
* macro.c (enter_macro_context): Only push
macro_real_token_count (macro) tokens rather than
macro->count tokens, regardless of
CPP_OPTION (pfile, track-macro-expansion).
* c-c++-common/cpp/pr58844-1.c: New test.
* c-c++-common/cpp/pr58844-2.c: New test.
Jakub Jelinek [Thu, 6 Mar 2014 08:01:57 +0000 (09:01 +0100)]
backport: re PR c/60101 (Long compile times when mixed complex floating point datatypes are used in lengthy expressions)
Backport from mainline
2014-02-12 Jakub Jelinek <jakub@redhat.com>
PR c/60101
* c-common.c (merge_tlist): If copy is true, call new_tlist,
if false, add ADD itself, rather than vice versa.
(verify_tree): For COND_EXPR, don't call merge_tlist with non-zero
copy. For SAVE_EXPR, only call merge_tlist once.
Peter Bergner [Mon, 3 Mar 2014 22:13:28 +0000 (16:13 -0600)]
backport: re PR libitm/57643 (libitm.c/reentrant.c hangs on POWER8 with HTM)
Backport from mainline
2013-06-20 Torvald Riegel <triegel@redhat.com>
* query.cc (_ITM_inTransaction): Abort when using the HTM fastpath.
(_ITM_getTransactionId): Same.
* config/x86/target.h (htm_transaction_active): New.
2013-06-20 Torvald Riegel <triegel@redhat.com>
PR libitm/57643
* beginend.cc (gtm_thread::begin_transaction): Handle reentrancy in
the HTM fastpath.
Richard Biener [Tue, 25 Feb 2014 12:42:59 +0000 (12:42 +0000)]
backport: re PR tree-optimization/60276 (-O3 autovectorizer breaks on a particular loop)
2014-02-25 Richard Biener <rguenther@suse.de>
Backport from mainline
2014-02-21 Richard Biener <rguenther@suse.de>
PR tree-optimization/60276
* tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
(STMT_VINFO_MIN_NEG_DIST): New macro.
* tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
STMT_VINFO_MIN_NEG_DIST.
* tree-vect-stmts.c (vectorizable_load): Verify if assumptions
made for negative dependence distances still hold.
David Holsgrove [Sun, 23 Feb 2014 18:38:29 +0000 (18:38 +0000)]
predicates.md: Add cmp_op predicate.
2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
* config/microblaze/predicates.md: Add cmp_op predicate.
* config/microblaze/microblaze.md: Add branch_compare instruction
which uses cmp_op predicate and emits cmp insn before branch.
* config/microblaze/microblaze.c (microblaze_emit_compare): Rename
to microblaze_expand_conditional_branch and consolidate logic.
(microblaze_expand_conditional_branch): emit branch_compare
insn instead of handling cmp op separate from branch insn.
Jerry DeLisle [Sat, 22 Feb 2014 02:11:44 +0000 (02:11 +0000)]
backport: re PR fortran/59700 (Misleading/buggy runtime error message: Bad integer for item 0 in list input)
2014-02-21 Jerry DeLisle <jvdelisle@gcc.gnu>
Dominique d'Humieres <dominiq@lps.ens.fr>
Steven G. Kargl <kargl@gcc.gnu.org>
Backport from mainline
PR libfortran/59700
PR libfortran/59764
* io/io.h (struct st_parameter_dt): Assign expanded_read flag to
unused bit. Define new variable line_buffer_pos.
* io/list_read.c (free_saved, next_char, l_push_char,
read_logical, read_real): Replace use of item_count with
line_buffer_pos for line_buffer look ahead.
(read_logical, read_integer, parse_real, read_real, check_type):
Adjust location of free_line to after generating error messages
to retain the correct item count for the message.
Co-Authored-By: Dominique d'Humieres <dominiq@lps.ens.fr> Co-Authored-By: Steven G. Kargl <kargl@gcc.gnu.org>
From-SVN: r208038
re PR c++/60272 (atomic<>::compare_exchange_weak has spurious store and can cause race conditions)
PR c++/60272
gcc/
* builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
on failure the store back into EXPECT.
libatomic/
* cas_n.c (libat_compare_exchange): Conditionalize on failure
the store back to EPTR.
Jakub Jelinek [Thu, 20 Feb 2014 20:40:52 +0000 (21:40 +0100)]
re PR target/57896 (ICE in expand_expr_real_2)
PR target/57896
* config/i386/i386.c (expand_vec_perm_interleave2): Don't call
gen_reg_rtx if d->testing_p.
(expand_vec_perm_pshufb2, expand_vec_perm_even_odd_1,
expand_vec_perm_broadcast_1): Return early if d->testing_p and
we will certainly return true.
Kai Tietz [Tue, 18 Feb 2014 12:57:32 +0000 (13:57 +0100)]
backport: re PR target/60193 (ICE on big nested frame)
2014-02-18 Kai Tietz <ktietz@redhat.com>
Backport from mainline
PR target/60193
* config/i386/i386.c (ix86_expand_prologue): Use
rax register as displacement for restoring %r10, %rax.
Additional fix wrong offset for restoring both-registers.
2014-02-18 Kai Tietz <ktietz@redhat.com>
PR target/60193
* gcc.target/i386/nest-1.c: New testcase.
Jakub Jelinek [Tue, 18 Feb 2014 07:23:51 +0000 (08:23 +0100)]
re PR driver/60233 (AVX instructions emitted with -march=native on host without AVX support)
PR driver/60233
* config/i386/driver-i386.c (host_detect_local_cpu): If
YMM state is not saved by the OS, also clear has_f16c. Move
CPUID 0x80000001 handling before YMM state saving checking.
Roland McGrath [Fri, 14 Feb 2014 17:37:50 +0000 (17:37 +0000)]
Use ud2 assembly mnemonic when available.
gcc/
* configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
* configure: Regenerated.
* config.in: Regenerated.
* config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
instead of ASM_SHORT.
Dominik Vogt [Thu, 13 Feb 2014 12:34:11 +0000 (12:34 +0000)]
s390.c (s390_asm_output_function_label): Fix crash caused by bad second argument to warning_at() with -mhotpatch and...
2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
* config/s390/s390.c (s390_asm_output_function_label): Fix crash
caused by bad second argument to warning_at() with -mhotpatch and
nested functions (e.g. with gfortran).
PR tree-optimization/60115
* tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
MEM_REF handling. Properly verify that the accesses are not
out of the objects bound.
projects / thirdparty / gcc.git / log
