Re: PATCH: Properly set ix86_gen_leave and ix86_gen_monitor
On Mon, Mar 12, 2012 at 12:02 AM, H.J. Lu hongjiu...@intel.com wrote: leave_rex64 works on DImode and sse3_monitor64 works on Pmode. This patch properly sets ix86_gen_leave and ix86_gen_monitor, depending on TARGET_64BIT and Pmode. Tested on Linux/x86-64. OK for trunk? 2012-03-11 H.J. Lu hongjiu...@intel.com * config/i386/i386.c (ix86_option_override_internal): Properly set ix86_gen_leave and ix86_gen_monitor. Check Pmode == DImode. * config/i386/sse.md (sse3_monitor64): Renamed to ... (sse3_monitor64_mode): This. Please provide more descriptive ChangeLog. You are changing the check on quite some global variables, please state them explicitly. Otherwise the patch is OK. BTW: I sincerely hope you reviewed all the places where these globals are used... Uros.
Re: PATCH: Call gen_tls_global_dynamic_64_mode and en_tls_local_dynamic_base_64_mode
On Mon, Mar 12, 2012 at 12:26 AM, H.J. Lu hongjiu...@intel.com wrote: Pmode may be DImode for x32. This patch calls gen_tls_global_dynamic_64_mode and gen_tls_local_dynamic_base_64_mode, depending on Pmode. Tested on Linux/x86-64. OK for trunk? 2012-03-11 H.J. Lu hongjiu...@intel.com * config/i386/i386.c (legitimize_tls_address): Call gen_tls_global_dynamic_64_mode and gen_tls_local_dynamic_base_64_mode, depending on Pmode. * config/i386/i386.md (*tls_global_dynamic_64): Renamed to ... (*tls_global_dynamic_64_mode): This. (tls_global_dynamic_64): Renamed to ... (tls_global_dynamic_64_mode): This. (*tls_local_dynamic_base_64): Renamed to ... (*tls_local_dynamic_base_64mode): This. (tls_local_dynamic_base_64): Renamed to ... (tls_local_dynamic_base_64mode): This. diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c index d673101..fc8fa12 100644 --- a/gcc/config/i386/i386.c +++ b/gcc/config/i386/i386.c @@ -12533,7 +12543,12 @@ legitimize_tls_address (rtx x, enum tls_model model, bool for_mov) rtx rax = gen_rtx_REG (Pmode, AX_REG), insns; start_sequence (); - emit_call_insn (gen_tls_global_dynamic_64 (rax, x, caddr)); + if (Pmode == DImode) + emit_call_insn (gen_tls_global_dynamic_64_di (rax, x, + caddr)); + else + emit_call_insn (gen_tls_global_dynamic_64_si (rax, x, + caddr)); insns = get_insns (); end_sequence (); @@ -12581,7 +12596,12 @@ legitimize_tls_address (rtx x, enum tls_model model, bool for_mov) rtx rax = gen_rtx_REG (Pmode, AX_REG), insns, eqv; start_sequence (); - emit_call_insn (gen_tls_local_dynamic_base_64 (rax, caddr)); + if (Pmode == DImode) + emit_call_insn (gen_tls_local_dynamic_base_64_di (rax, + caddr)); + else + emit_call_insn (gen_tls_local_dynamic_base_64_si (rax, + caddr)); insns = get_insns (); end_sequence (); Please use indirect functions here, see many examples in the code. OK with this change. Thanks, Uros.
Re: [Patch, Fortran] PR 52542 - Fix PROCEDURE() with Bind(C)
Dear Tobias, Apart from s/Contribute/Contributed/ this is OK for trunk. In fact, I would say that it is obvious. Thanks for the patch. Paul On Sat, Mar 10, 2012 at 4:53 PM, Tobias Burnus bur...@net-b.de wrote: Tobias Burnus wrote: If the interface in a PROCEDURE() statement is Bind(C), also the procedure (pointer) declared in that statement is BIND(C). From the F2008 standard: A proc-language-binding-spec without a NAME= is allowed, but is redundant with the proc-interface required by C1222. Build and currently regtested on x86-64-linux. OK for the trunk (if regtesting succeeded)? Well, it didn't as I forgot to reset two variables - one then gets then an error that one has specified an binding name - or the wrong binding name might be used. Build and regtested on x86-64-linux. OK? Tobias -- The knack of flying is learning how to throw yourself at the ground and miss. --Hitchhikers Guide to the Galaxy
Re: [VMS/committed]: Do not define __size_t on VMS
On Mar 9, 2012, at 6:22 PM, Joseph S. Myers wrote: On Fri, 9 Mar 2012, Tristan Gingold wrote: as it is already defined and used by the system headers. No regressions for x86_64-darwin after full bootstrap. Tristan. 2012-03-09 Tristan Gingold ging...@adacore.com * ginclude/stddef.h: Do not define __size_t on VMS. Sorry, this patch is wrong. The identifier VMS is in the user's namespace and users on non-VMS platforms must be able to define it as a macro without affecting how stddef.h behaves. Please change this to test a macro in the implementation namespace instead. Thank you for catching that. Fixed with: 2012-03-12 Tristan Gingold ging...@adacore.com * ginclude/stddef.h: Adjust previous patch. Use __VMS__ instead of VMS. Index: gcc/ginclude/stddef.h === --- gcc/ginclude/stddef.h (revision 185212) +++ gcc/ginclude/stddef.h (working copy) @@ -203,7 +203,7 @@ #if (defined (__FreeBSD__) (__FreeBSD__ = 5)) \ || defined(__FreeBSD_kernel__) /* __size_t is a typedef on FreeBSD 5, must not trash it. */ -#elif defined (VMS) +#elif defined (__VMS__) /* __size_t is also a typedef on VMS. */ #else #define __size_t
Re: [PATCH] Move strip_float_extensions to tree.c
On Wed, 7 Mar 2012, Joseph S. Myers wrote: On Wed, 7 Mar 2012, Richard Guenther wrote: Now, convert.c is used from all frontends to implement convert () (that looks backwards - the language convert should be a langhook, called from convert implemented in convert.c). But well, I aint not touching this beast ;) I don't think convert () should be a langhook; it's all about language-specific semantics (the only non-front-end places calling it, outside of convert.c itself, now appear to be in arm.c). I'm not sure of the extent to which the recursive calls to convert inside convert.c need any language-specific semantics. To the extent that they do, I think front ends should be dealing with the semantics rather than having convert call back into the front end. (I also think all the errors from convert.c should be given by front ends instead; front ends should only call the convert_to_* helpers for code they have checked is valid.) I was thinking about having convert () be a middle-end function (not langhook) working on generic, that does all frontend specific stuff via a langhook (I didn't try to compare the convert implementations to see if there is even any language-specifics about them). I agree that errors should be raised from the frontends (also the remaining optimizations convert_to_* do should be either moved to fold or to later gimple optimization passes). But no, I'm not going to touch this - maybe apart from removing the obvious wrong users in arm.c. Richard.
Re: [PATCH] Fix PRs 52080, 52097 and 48124, rewrite bitfield expansion, enable the C++ memory model wrt bitfields everywhere
On Fri, 9 Mar 2012, Eric Botcazou wrote:
This patch also completely replaces get_bit_range (which is where
PR52097 ICEs) by a trivial implementation.
How does it short-circuit the decision made by get_best_mode exactly? By
making get_bit_range return non-zero in more cases?
It will make get_bit_range return non-zero in all cases (well, in
all cases where there is a COMPONENT_REF with a FIELD_DECL that
was part of a RECORD_TYPE at the time of finish_record_layout)
There is PR52134 which will make this patch cause 1 gnat regression.
This looks rather benign to me.
Yeah, it should be easy to fix - the question is whether we can
really rely on TYPE_SIZE_UNIT (DECL_CONTEXT (field)) - DECL_FIELD_OFFSET
(field) to fold to a constant if field is the first field of a bitfield
group. We can as well easily avoid the ICE by not computing a
DECL_BIT_FIELD_REPRESENTATIVE for such field in some way.
In the end how we divide bitfield groups will probably get some
control via a langhook.
* gimplify.c (gimplify_expr): Translate bitfield accesses
to BIT_FIELD_REFs of the representative.
This part isn't in the patch.
Oops. And it should not be (I did that for experimentation), ChangeLog
piece dropped.
+ /* Return a new underlying object for a bitfield started with FIELD. */
+
+ static tree
+ start_bitfield_representative (tree field)
+ {
+ tree repr = make_node (FIELD_DECL);
+ DECL_FIELD_OFFSET (repr) = DECL_FIELD_OFFSET (field);
+ /* Force the representative to begin at an BITS_PER_UNIT aligned
...at a BITS_PER_UNIT aligned...
Fixed.
+ boundary - C++ may use tail-padding of a base object to
+ continue packing bits so the bitfield region does not start
+ at bit zero (see g++.dg/abi/bitfield5.C for example).
+ Unallocated bits may happen for other reasons as well,
+ for example Ada which allows explicit bit-granular structure layout.
*/ + DECL_FIELD_BIT_OFFSET (repr)
+ = size_binop (BIT_AND_EXPR,
+ DECL_FIELD_BIT_OFFSET (field),
+ bitsize_int (~(BITS_PER_UNIT - 1)));
+ SET_DECL_OFFSET_ALIGN (repr, DECL_OFFSET_ALIGN (field));
+ DECL_SIZE (repr) = DECL_SIZE (field);
+ DECL_PACKED (repr) = DECL_PACKED (field);
+ DECL_CONTEXT (repr) = DECL_CONTEXT (field);
+ return repr;
Any particular reason to set DECL_SIZE but not DECL_SIZE_UNIT here? They are
generally set together.
No reason, fixed (I just set those fields I will use during updating
of 'repr', the other fields are set once the field has final size).
+
+ /* Finish up a bitfield group that was started by creating the underlying
+object REPR with the last fied in the bitfield group FIELD. */
...the last field...
Fixed.
+ /* Round up bitsize to multiples of BITS_PER_UNIT. */
+ bitsize = (bitsize + BITS_PER_UNIT - 1) ~(BITS_PER_UNIT - 1);
+
+ /* Find the smallest nice mode to use.
+ ??? Possibly use get_best_mode with appropriate arguments instead
+ (which would eventually require splitting representatives here). */
+ for (modesize = bitsize; modesize = maxbitsize; modesize +=
BITS_PER_UNIT) + {
+ mode = mode_for_size (modesize, MODE_INT, 1);
+ if (mode != BLKmode)
+ break;
+ }
The double loop looks superfluous. Why not re-using the implementation of
smallest_mode_for_size?
Ah, indeed. Matching the current implementation would be
for (mode = GET_CLASS_NARROWEST_MODE (MODE_INT); mode != VOIDmode;
mode = GET_MODE_WIDER_MODE (mode))
if (GET_MODE_PRECISION (mode) = bitsize)
break;
if (mode != VOIDmode
(GET_MODE_PRECISION (mode) maxbitsize
|| GET_MODE_PRECISION (mode) MAX_FIXED_MODE_SIZE))
mode = VOIDmode;
if (mode == VOIDmode)
...
Btw, I _think_ I want GET_MODE_BITSIZE here - we cannot allow
GET_MODE_BITSIZE GET_MODE_PRECISION as that would possibly
access memory that is not allowed. Thus, what GET_MODE_* would
identify the access size used for a MEM of that mode?
Anyway, fixed as you suggested, with the MAX_FIXED_MODE_SIZE check.
I'll split out the tree-sra.c piece, re-test and re-post.
Thanks,
Richard.
Re: [PATCH] Fix up register_edge_assert_for_2 (PR tree-optimization/52533)
On Fri, 9 Mar 2012, Jakub Jelinek wrote:
Hi!
My recent commit to tree-vrp.c on the trunk caused the following
testcase to fail, the problem is that we would insert = 255
assertion for unsigned char expression (which doesn't say anything)
and VRP insist that such ASSERT_EXPRs aren't added.
Fixed thusly, in addition the patch cleans the code slightly,
e.g. by using double_int instead of a pair of uHWIs.
Bootstrapped/regtested on x86_64-linux and i686-linux,
ok for trunk?
Ok.
Thanks,
Richard.
2012-03-09 Jakub Jelinek ja...@redhat.com
PR tree-optimization/52533
* tree-vrp.c (register_edge_assert_for_2): Use double_int
type for mask, only handle shifts by non-zero in-range
shift count, for LE_EXPR and GT_EXPR if new_val is
maximum, don't add the assertion.
* gcc.c-torture/compile/pr52533.c: New test.
--- gcc/tree-vrp.c.jj 2012-03-09 10:26:26.0 +0100
+++ gcc/tree-vrp.c2012-03-09 12:30:19.348501480 +0100
@@ -4470,7 +4470,8 @@ register_edge_assert_for_2 (tree name, e
gimple def_stmt = SSA_NAME_DEF_STMT (name);
tree name2 = NULL_TREE, cst2 = NULL_TREE;
tree val2 = NULL_TREE;
- unsigned HOST_WIDE_INT mask[2] = { 0, 0 };
+ double_int mask = double_int_zero;
+ unsigned int prec = TYPE_PRECISION (TREE_TYPE (val));
/* Extract CST2 from the right shift. */
if (is_gimple_assign (def_stmt)
@@ -4480,23 +4481,13 @@ register_edge_assert_for_2 (tree name, e
cst2 = gimple_assign_rhs2 (def_stmt);
if (TREE_CODE (name2) == SSA_NAME
host_integerp (cst2, 1)
-(unsigned HOST_WIDE_INT) tree_low_cst (cst2, 1)
- 2 * HOST_BITS_PER_WIDE_INT
INTEGRAL_TYPE_P (TREE_TYPE (name2))
+IN_RANGE (tree_low_cst (cst2, 1), 1, prec - 1)
+prec = 2 * HOST_BITS_PER_WIDE_INT
live_on_edge (e, name2)
!has_single_use (name2))
{
- if ((unsigned HOST_WIDE_INT) tree_low_cst (cst2, 1)
-HOST_BITS_PER_WIDE_INT)
- mask[0] = ((unsigned HOST_WIDE_INT) 1
- tree_low_cst (cst2, 1)) - 1;
- else
- {
- mask[1] = ((unsigned HOST_WIDE_INT) 1
- (tree_low_cst (cst2, 1)
- - HOST_BITS_PER_WIDE_INT)) - 1;
- mask[0] = -1;
- }
+ mask = double_int_mask (tree_low_cst (cst2, 1));
val2 = fold_binary (LSHIFT_EXPR, TREE_TYPE (val), val, cst2);
}
}
@@ -4515,37 +4506,40 @@ register_edge_assert_for_2 (tree name, e
{
if (!TYPE_UNSIGNED (TREE_TYPE (val)))
{
- unsigned int prec = TYPE_PRECISION (TREE_TYPE (val));
tree type = build_nonstandard_integer_type (prec, 1);
tmp = build1 (NOP_EXPR, type, name2);
val2 = fold_convert (type, val2);
}
tmp = fold_build2 (MINUS_EXPR, TREE_TYPE (tmp), tmp, val2);
- new_val = build_int_cst_wide (TREE_TYPE (tmp), mask[0], mask[1]);
+ new_val = double_int_to_tree (TREE_TYPE (tmp), mask);
new_comp_code = comp_code == EQ_EXPR ? LE_EXPR : GT_EXPR;
}
else if (comp_code == LT_EXPR || comp_code == GE_EXPR)
new_val = val2;
else
{
- new_val = build_int_cst_wide (TREE_TYPE (val2),
- mask[0], mask[1]);
- new_val = fold_binary (BIT_IOR_EXPR, TREE_TYPE (val2),
- val2, new_val);
+ mask = double_int_ior (tree_to_double_int (val2), mask);
+ if (double_int_minus_one_p (double_int_sext (mask, prec)))
+ new_val = NULL_TREE;
+ else
+ new_val = double_int_to_tree (TREE_TYPE (val2), mask);
}
- if (dump_file)
+ if (new_val)
{
- fprintf (dump_file, Adding assert for );
- print_generic_expr (dump_file, name2, 0);
- fprintf (dump_file, from );
- print_generic_expr (dump_file, tmp, 0);
- fprintf (dump_file, \n);
- }
+ if (dump_file)
+ {
+ fprintf (dump_file, Adding assert for );
+ print_generic_expr (dump_file, name2, 0);
+ fprintf (dump_file, from );
+ print_generic_expr (dump_file, tmp, 0);
+ fprintf (dump_file, \n);
+ }
- register_new_assert_for (name2, tmp, new_comp_code, new_val,
-NULL, e, bsi);
- retval = true;
+ register_new_assert_for (name2, tmp, new_comp_code, new_val,
+NULL, e, bsi);
+ retval = true;
+ }
}
}
---
Re: [committed] Fix failure of gcc.dg/torture/pr52407.c on hppa*-*-hpux*
On Sat, 10 Mar 2012, John David Anglin wrote: Work around limited alignment of common. Tested on hppa2.0w-hp-hpux11.11 and hppa64-hp-hpux11.11. Committed to trunk. Ok for 4.7? Ok. Thanks, Richard.
[Patch]: Add include hacks for VMS headers
Hi,
this patch adds a few fix include entries for VMS so that:
* it is possible to use a gcc for VMS to compile gcc (collisions between system
and gcc headers),
* no warnings when signal.h is included,
* no warnings when resolv.h is included,
* avoid the use of DEC-C built-ins in some standard macros,
* disable getopt/optinf/opterr/optopt in 64 bit modes,
* declare some structure tag outside of argument list,
* use the more portable version of setjmp.
Tested for a while to build gcc for ia64 and alpha VMS.
'make check' is clean.
Comments are welcome!
Ok for trunk ?
Tristan.
fixincludes/
2012-03-12 Tristan Gingold ging...@adacore.com
* inclhack.def (vms_use_quoted_include,vms_add_missing_braces)
(vms_do_not_redeclare_hostalias, vms_decc_builtin)
(vms_no_64bit_getopt, vms_forward_declare_struct)
(vms_use_fast_setjmp): New fixes.
* fixincl.x: Regenerate.
* tests/base/rtldef/string.h: Update.
* tests/base/rtldef/if.h, tests/base/rtldef/resolv.h,
* tests/base/rtldef/setjmp.h, tests/base/rtldef/signal.h,
* tests/base/rtldef/stdio.h, tests/base/rtldef/wait.h: New files.
diff --git a/fixincludes/inclhack.def b/fixincludes/inclhack.def
index 40e7805..1a5cd8d 100644
--- a/fixincludes/inclhack.def
+++ b/fixincludes/inclhack.def
@@ -4611,6 +4611,129 @@ fix = {
};
/*
+ * On VMS, change resource.h to sys/resource.h to avoid a
+ * conflict while building gcc. Likewise for builtins.h
+ */
+fix = {
+hackname = vms_use_quoted_include;
+select= (#[ \t]*include[ \t]+)(resource|builtins)\\.h;
+mach = *-*-*vms*;
+files = rtldef/wait.h, starlet_c/pthread.h;
+c_fix = format;
+
+c_fix_arg = '%1sys/%2.h';
+
+test_text = # include resource.h;
+};
+
+/*
+ * On VMS, add missing braces around sigset_t constants.
+ */
+fix = {
+hackname = vms_add_missing_braces;
+select= (_SIG_[A-Z]+_SET[ \t]+= \\{)(0x[0F]+, 0x[0F]+);
+mach = *-*-*vms*;
+files = rtldef/signal.h;
+c_fix = format;
+
+c_fix_arg = '%1 {%2} ';
+
+test_text = static const __sigset_t _SIG_EMPTY_SET =
+{0x, 0x},\n
+ _SIG_FULL_SET = {0x, 0x};\n;
+};
+
+/*
+ * On VMS, fix incompatible redeclaration of hostalias.
+ */
+fix = {
+hackname = vms_do_not_redeclare_hostalias;
+select= (void[ \t]+fp_nquery \\(const u_char \\*, int, FILE
\\*\\);)\n
+(__char_ptr32[ \t]+hostalias \\(const char \\*\\););
+mach = *-*-*vms*;
+files = rtldef/resolv.h;
+c_fix = format;
+
+c_fix_arg = %1\n
+/* %2 */;
+
+test_text = void fp_nquery (const u_char *, int, FILE *);\n
+__char_ptr32 hostalias (const char *);\n;
+};
+
+/*
+ * On VMS, some DEC-C builtins are directly used.
+ */
+fix = {
+hackname = vms_decc_builtin;
+select= (__MEMSET|__MEMMOVE|__MEMCPY|__STRLEN|__STRCPY);
+mach = *-*-*vms*;
+files = rtldef/string.h, rtldef/time.h, rtldef/strings.h,
+rtldef/socket.h;
+sed = s@__MEMSET@memset@;
+sed = s@__MEMMOVE@memmove@;
+sed = s@__MEMCPY@memcpy@;
+sed = s@__STRLEN@strlen@;
+sed = s@__STRCPY@strcpy@;
+
+test_text = define FD_ZERO(__p) __MEMSET((__p), 0, sizeof(*(__p)))\n;
+};
+
+/*
+ * On VMS, do not declare getopt and al if pointers are 64 bit.
+ */
+fix = {
+hackname = vms_no_64bit_getopt;
+select= ^[ \t]*(extern[ \t]*)?(int[
\t]*(getopt|optind|opterr|optopt)|(char \\*optarg))([ \t]*\\(.*\\))?;\n;
+/*select= ^[ \t](extern[ \t))?(int[
\t](getopt|optind|opterr|optopt))|(char *optarg).*;\n; */
+mach = *-*-*vms*;
+files = rtldef/stdio.h, rtldef/unistd.h;
+c_fix = format;
+
+c_fix_arg = #if __INITIAL_POINTER_SIZE != 64 /* getopt is short ptr only.
*/\n
+%0
+#endif\n;
+
+test_text = int getopt (int, char * const [], const char *);;
+};
+
+/*
+ * On VMS, forward declare structure before referencing them in prototypes.
+ */
+fix = {
+hackname = vms_forward_declare_struct;
+select= (/\\* forward decls for C\\+\\+ \\*/\n)
+#ifdef __cplusplus\n;
+mach = *-*-*vms*;
+files = rtldef/if.h;
+c_fix = format;
+
+c_fix_arg = %1
+#if defined (__cplusplus) || defined (__GNUC__)\n;
+
+test_text = /* forward decls for C++ */\n
+#ifdef __cplusplus\n
+struct foo;\n
+#endif\n;
+};
+
+/*
+ * On VMS, force the use of fast setjmp, as the normal setjmp uses conditions
+ * which is not yet fully supported by gcc.
+ */
+fix = {
+hackname = vms_use_fast_setjmp;
+select= (#[ \t]*if[ \t]*defined\\(__FAST_SETJMP\\)[ \t]*\\|\\|);
+mach = *-*-*vms*;
+files = rtldef/setjmp.h;
+c_fix = format;
+
+c_fix_arg = %0 defined (__GNUC__) ||;
+
+
Re: [PATCH, i386]: Avoid partial reg stall with arith insn + setCC + movzbl sequence
Il 12/03/2012 09:52, Uros Bizjak ha scritto: +(define_peephole2 + [(parallel [(set (reg FLAGS_REG) (match_operand 0 )) + (match_operand 4 )]) + (set (match_operand:QI 1 register_operand ) + (match_operator:QI 2 ix86_comparison_operator + [(reg FLAGS_REG) (const_int 0)])) + (set (match_operand 3 q_regs_operand ) + (zero_extend (match_dup 1)))] + (peep2_reg_dead_p (3, operands[1]) +|| operands_match_p (operands[1], operands[3])) +! reg_overlap_mentioned_p (operands[3], operands[0]) I understand that you're assuming the shape of operands[4] to be the same as operands[3], but would it be preferrable to add another overlap check on operands[4]? For example the transformation is invalid if you had an overlap between operands[3] and the destination of operands[4]. Paolo
Re: [PATCH] Fix PRs 52080, 52097 and 48124, rewrite bitfield expansion, enable the C++ memory model wrt bitfields everywhere
Btw, I _think_ I want GET_MODE_BITSIZE here - we cannot allow GET_MODE_BITSIZE GET_MODE_PRECISION as that would possibly access memory that is not allowed. Thus, what GET_MODE_* would identify the access size used for a MEM of that mode? I agree that GET_MODE_BITSIZE makes more sense than GET_MODE_PRECISION here. -- Eric Botcazou
Re: [PATCH] Do not handle SUBREG in apply_distributive_law (Re: RFC: allowing fwprop to propagate subregs)
On Wed, Mar 7, 2012 at 6:40 PM, Ulrich Weigand uweig...@de.ibm.com wrote: Richard Kenner wrote: Given the current set of results, since I do not have any way to verify whether my simplify_set changes would actually trigger correctly, I'd rather propose to just remove the SUBREG case in apply_distributive_law (i.e. only apply the first patch below). Thoughts? I think that's reasonable. But I'd replace it with a comment saying what used to be there and why it was removed. Now that we're back in stage 1, I'd like to try and move forward with this again. Here's a patch that implements your suggestion. Tested on arm-linux-gnueabi, i386-linux-gnu and s390-linux-gnu. OK for mainline? Ok. Thanks, Richard. Bye, Ulrich 2012-03-07 Ulrich Weigand ulrich.weig...@linaro.org gcc/ * combine.c (apply_distributive_law): Do not distribute SUBREG. === modified file 'gcc/combine.c' --- gcc/combine.c 2012-02-07 15:48:52 + +++ gcc/combine.c 2012-02-22 11:57:19 + @@ -9286,36 +9286,22 @@ /* This is also a multiply, so it distributes over everything. */ break; - case SUBREG: - /* Non-paradoxical SUBREGs distributes over all operations, - provided the inner modes and byte offsets are the same, this - is an extraction of a low-order part, we don't convert an fp - operation to int or vice versa, this is not a vector mode, - and we would not be converting a single-word operation into a - multi-word operation. The latter test is not required, but - it prevents generating unneeded multi-word operations. Some - of the previous tests are redundant given the latter test, - but are retained because they are required for correctness. - - We produce the result slightly differently in this case. */ - - if (GET_MODE (SUBREG_REG (lhs)) != GET_MODE (SUBREG_REG (rhs)) - || SUBREG_BYTE (lhs) != SUBREG_BYTE (rhs) - || ! subreg_lowpart_p (lhs) - || (GET_MODE_CLASS (GET_MODE (lhs)) - != GET_MODE_CLASS (GET_MODE (SUBREG_REG (lhs - || paradoxical_subreg_p (lhs) - || VECTOR_MODE_P (GET_MODE (lhs)) - || GET_MODE_SIZE (GET_MODE (SUBREG_REG (lhs))) UNITS_PER_WORD - /* Result might need to be truncated. Don't change mode if - explicit truncation is needed. */ - || !TRULY_NOOP_TRUNCATION_MODES_P (GET_MODE (x), - GET_MODE (SUBREG_REG (lhs - return x; - - tem = simplify_gen_binary (code, GET_MODE (SUBREG_REG (lhs)), - SUBREG_REG (lhs), SUBREG_REG (rhs)); - return gen_lowpart (GET_MODE (x), tem); + /* This used to handle SUBREG, but this turned out to be counter- + productive, since (subreg (op ...)) usually is not handled by + insn patterns, and this optimization therefore transformed + recognizable patterns into unrecognizable ones. Therefore the + SUBREG case was removed from here. + + It is possible that distributing SUBREG over arithmetic operations + leads to an intermediate result than can then be optimized further, + e.g. by moving the outer SUBREG to the other side of a SET as done + in simplify_set. This seems to have been the original intent of + handling SUBREGs here. + + However, with current GCC this does not appear to actually happen, + at least on major platforms. If some case is found where removing + the SUBREG case here prevents follow-on optimizations, distributing + SUBREGs ought to be re-added at that place, e.g. in simplify_set. */ default: return x; -- Dr. Ulrich Weigand GNU Toolchain for Linux on System z and Cell BE ulrich.weig...@de.ibm.com
[PATCH][ARM] Do not call covert during RTL expansion
convert is supposed to be called only from frontend code as it may raise errors. The call in neon_dereference_pointer is from RTL expansion (and not necessary). Fixed as follows, I built a cross-cc1 for checking. Ok? Thanks, Richard. 2012-03-12 Richard Guenther rguent...@suse.de * config/arm/arm.c (neon_dereference_pointer): Do not call covert during RTL expansion. Index: gcc/config/arm/arm.c === *** gcc/config/arm/arm.c(revision 185220) --- gcc/config/arm/arm.c(working copy) *** neon_dereference_pointer (tree exp, enum *** 20637,20645 array_type = build_array_type (elem_type, build_index_type (upper_bound)); /* Dereference EXP using that type. */ - exp = convert (build_pointer_type (array_type), exp); return fold_build2 (MEM_REF, array_type, exp, ! build_int_cst (TREE_TYPE (exp), 0)); } /* Expand a Neon builtin. */ --- 20637,20644 array_type = build_array_type (elem_type, build_index_type (upper_bound)); /* Dereference EXP using that type. */ return fold_build2 (MEM_REF, array_type, exp, ! build_int_cst (build_pointer_type (array_type), 0)); } /* Expand a Neon builtin. */
[PATCH] Fix some SRA and GIMPLE verification issues
This fixes issues I noticed when working on the bitfield expansion
issues.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied to trunk.
Richard.
2012-03-05 Richard Guenther rguent...@suse.de
* tree-sra.c (create_access_replacement): Only rename the
replacement if we can rewrite it into SSA form. Properly
mark register typed replacements that we cannot rewrite
with TREE_ADDRESSABLE.
* tree-cfg.c (verify_expr): Fix BIT_FIELD_REF verification
for aggregate or BLKmode results.
Index: gcc/tree-sra.c
===
*** gcc/tree-sra.c.orig 2012-02-21 12:59:35.0 +0100
--- gcc/tree-sra.c 2012-02-24 12:52:39.0 +0100
*** create_access_replacement (struct access
*** 1922,1934
repl = create_tmp_var (access-type, SR);
add_referenced_var (repl);
! if (rename)
mark_sym_for_renaming (repl);
! if (!access-grp_partial_lhs
!(TREE_CODE (access-type) == COMPLEX_TYPE
! || TREE_CODE (access-type) == VECTOR_TYPE))
! DECL_GIMPLE_REG_P (repl) = 1;
DECL_SOURCE_LOCATION (repl) = DECL_SOURCE_LOCATION (access-base);
DECL_ARTIFICIAL (repl) = 1;
--- 1922,1940
repl = create_tmp_var (access-type, SR);
add_referenced_var (repl);
! if (!access-grp_partial_lhs
!rename)
mark_sym_for_renaming (repl);
! if (TREE_CODE (access-type) == COMPLEX_TYPE
! || TREE_CODE (access-type) == VECTOR_TYPE)
! {
! if (!access-grp_partial_lhs)
! DECL_GIMPLE_REG_P (repl) = 1;
! }
! else if (access-grp_partial_lhs
! is_gimple_reg_type (access-type))
! TREE_ADDRESSABLE (repl) = 1;
DECL_SOURCE_LOCATION (repl) = DECL_SOURCE_LOCATION (access-base);
DECL_ARTIFICIAL (repl) = 1;
Index: gcc/tree-cfg.c
===
*** gcc/tree-cfg.c.orig 2012-02-24 12:29:18.0 +0100
--- gcc/tree-cfg.c 2012-02-24 12:52:39.0 +0100
*** verify_expr (tree *tp, int *walk_subtree
*** 2879,2895
error (invalid position or size operand to BIT_FIELD_REF);
return t;
}
! else if (INTEGRAL_TYPE_P (TREE_TYPE (t))
! (TYPE_PRECISION (TREE_TYPE (t))
! != TREE_INT_CST_LOW (TREE_OPERAND (t, 1
{
error (integral result type precision does not match
field size of BIT_FIELD_REF);
return t;
}
! if (!INTEGRAL_TYPE_P (TREE_TYPE (t))
! (GET_MODE_PRECISION (TYPE_MODE (TREE_TYPE (t)))
! != TREE_INT_CST_LOW (TREE_OPERAND (t, 1
{
error (mode precision of non-integral result does not
match field size of BIT_FIELD_REF);
--- 2879,2897
error (invalid position or size operand to BIT_FIELD_REF);
return t;
}
! if (INTEGRAL_TYPE_P (TREE_TYPE (t))
! (TYPE_PRECISION (TREE_TYPE (t))
! != TREE_INT_CST_LOW (TREE_OPERAND (t, 1
{
error (integral result type precision does not match
field size of BIT_FIELD_REF);
return t;
}
! else if (!INTEGRAL_TYPE_P (TREE_TYPE (t))
! !AGGREGATE_TYPE_P (TREE_TYPE (t))
! TYPE_MODE (TREE_TYPE (t)) != BLKmode
! (GET_MODE_PRECISION (TYPE_MODE (TREE_TYPE (t)))
! != TREE_INT_CST_LOW (TREE_OPERAND (t, 1
{
error (mode precision of non-integral result does not
match field size of BIT_FIELD_REF);
Re: [PATCH] Improve SCEV for array element
On Thu, Mar 8, 2012 at 8:13 AM, Jiangning Liu jiangning@arm.com wrote:
-Original Message-
From: Richard Guenther [mailto:richard.guent...@gmail.com]
Sent: Tuesday, March 06, 2012 9:12 PM
To: Jiangning Liu
Cc: gcc-patches@gcc.gnu.org
Subject: Re: [PATCH] Improve SCEV for array element
On Fri, Jan 20, 2012 at 10:06 AM, Jiangning Liu jiangning@arm.com
wrote:
It's definitely not ok at this stage but at most for next stage1.
OK. I may wait until next stage1.
This is a very narrow pattern-match. It doesn't allow for a[i].x
for
example,
even if a[i] is a one-element structure. I think the canonical way
of
handling
ADDR_EXPR is to use sth like
base = get_inner_reference (TREE_OPERAND (rhs1, 0), ...,
offset, ...);
base = build1 (ADDR_EXPR, TREE_TYPE (rhs1), base);
chrec1 = analyze_scalar_evolution (loop, base);
chrec2 = analyze_scalar_evolution (loop, offset);
chrec1 = chrec_convert (type, chrec1, at_stmt);
chrec2 = chrec_convert (TREE_TYPE (offset), chrec2, at_stmt);
res = chrec_fold_plus (type, chrec1, chrec2);
where you probably need to handle scev_not_known when analyzing
offset
(which might be NULL). You also need to add bitpos to the base
address
(in bytes, of course). Note that the MEM_REF case would naturally
work
with this as well.
OK. New patch is like below, and bootstrapped on x86-32.
You want instead of
+ if (TREE_CODE (TREE_OPERAND (rhs1, 0)) == ARRAY_REF
+ || TREE_CODE (TREE_OPERAND (rhs1, 0)) == MEM_REF
+ || TREE_CODE (TREE_OPERAND (rhs1, 0)) == COMPONENT_REF)
+ {
if (TREE_CODE (TREE_OPERAND (rhs1, 0)) == MEM_REF
|| handled_component_p (TREE_OPERAND (rhs1, 0)))
{
+ base = build1 (ADDR_EXPR, TREE_TYPE (rhs1), base);
+ chrec1 = analyze_scalar_evolution (loop, base);
can you please add a wrapper
tree
analyze_scalar_evolution_for_address_of (struct loop *loop, tree var)
{
return analyze_scalar_evolution (loop, build_fold_addr_expr (var));
}
and call that instead of building the ADDR_EXPR there? We want
to avoid building that tree node, but even such a simple wrapper would
be prefered.
+ if (bitpos)
if (bitpos != 0)
+ chrec3 = build_int_cst (integer_type_node,
+ bitpos / BITS_PER_UNIT);
please use size_int (bitpos / BITS_PER_UNIT) instead. Using
integer_type_node is definitely wrong.
Ok with that changes.
Richard,
Modified as all you suggested, and new code is like below. Bootstrapped on
x86-32. OK for trunk now?
Ok.
Thanks,
Richard.
Thanks,
-Jiangning
ChangeLog:
2012-03-08 Jiangning Liu jiangning@arm.com
* tree-scalar-evolution (interpret_rhs_expr): generate chrec for
array reference and component reference.
(analyze_scalar_evolution_for_address_of): New.
ChangeLog for testsuite:
2012-03-08 Jiangning Liu jiangning@arm.com
* gcc.dg/tree-ssa/scev-3.c: New.
* gcc.dg/tree-ssa/scev-4.c: New.
diff --git a/gcc/testsuite/gcc.dg/tree-ssa/scev-3.c
b/gcc/testsuite/gcc.dg/tree-ssa/scev-3.c
new file mode 100644
index 000..28d5c93
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/tree-ssa/scev-3.c
@@ -0,0 +1,18 @@
+/* { dg-do compile } */
+/* { dg-options -O2 -fdump-tree-optimized } */
+
+int *a_p;
+int a[1000];
+
+f(int k)
+{
+ int i;
+
+ for (i=k; i1000; i+=k) {
+ a_p = a[i];
+ *a_p = 100;
+ }
+}
+
+/* { dg-final { scan-tree-dump-times a 1 optimized } } */
+/* { dg-final { cleanup-tree-dump optimized } } */
diff --git a/gcc/testsuite/gcc.dg/tree-ssa/scev-4.c
b/gcc/testsuite/gcc.dg/tree-ssa/scev-4.c
new file mode 100644
index 000..6c1e530
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/tree-ssa/scev-4.c
@@ -0,0 +1,23 @@
+/* { dg-do compile } */
+/* { dg-options -O2 -fdump-tree-optimized } */
+
+typedef struct {
+ int x;
+ int y;
+} S;
+
+int *a_p;
+S a[1000];
+
+f(int k)
+{
+ int i;
+
+ for (i=k; i1000; i+=k) {
+ a_p = a[i].y;
+ *a_p = 100;
+ }
+}
+
+/* { dg-final { scan-tree-dump-times a 1 optimized } } */
+/* { dg-final { cleanup-tree-dump optimized } } */
diff --git a/gcc/tree-scalar-evolution.c b/gcc/tree-scalar-evolution.c
index 2077c8d..c719984
--- a/gcc/tree-scalar-evolution.c
+++ b/gcc/tree-scalar-evolution.c
@@ -266,6 +266,8 @@ along with GCC; see the file COPYING3. If not see
#include params.h
static tree analyze_scalar_evolution_1 (struct loop *, tree, tree);
+static tree analyze_scalar_evolution_for_address_of (struct loop *loop,
+ tree var);
/* The cached information about an SSA name VAR, claiming that below
basic block INSTANTIATED_BELOW, the value of VAR can be expressed
@@ -1712,16 +1714,59 @@ interpret_rhs_expr (struct loop
Re: [PATCH, i386]: Avoid partial reg stall with arith insn + setCC + movzbl sequence
On Mon, Mar 12, 2012 at 11:13 AM, Paolo Bonzini bonz...@gnu.org wrote: Il 12/03/2012 09:52, Uros Bizjak ha scritto: +(define_peephole2 + [(parallel [(set (reg FLAGS_REG) (match_operand 0 )) + (match_operand 4 )]) + (set (match_operand:QI 1 register_operand ) + (match_operator:QI 2 ix86_comparison_operator + [(reg FLAGS_REG) (const_int 0)])) + (set (match_operand 3 q_regs_operand ) + (zero_extend (match_dup 1)))] + (peep2_reg_dead_p (3, operands[1]) + || operands_match_p (operands[1], operands[3])) + ! reg_overlap_mentioned_p (operands[3], operands[0]) I understand that you're assuming the shape of operands[4] to be the same as operands[3], but would it be preferrable to add another overlap check on operands[4]? For example the transformation is invalid if you had an overlap between operands[3] and the destination of operands[4]. The destination of operands[4] _always_ matches one of operands inside operand[0]. All arithmetic insn that set flags are destructive on x86. Uros.
Re: [PATCH] Proper use of decl_function_context in dwar2out.c
On Thu, Mar 8, 2012 at 12:18 PM, Jakub Jelinek ja...@redhat.com wrote: On Thu, Mar 08, 2012 at 12:06:46PM +0100, Martin Jambor wrote: /* For local statics lookup proper context die. */ - if (TREE_STATIC (decl) decl_function_context (decl)) - context_die = lookup_decl_die (DECL_CONTEXT (decl)); + if (TREE_STATIC (decl) + (ctx_fndecl = decl_function_context (decl)) != NULL_TREE) + context_die = lookup_decl_die (ctx_fndecl); The formatting is wrong, shouldn't be at the end of line. For the rest I'll defer to Jason, not sure what exactly we want to do there. This hunk has been added by Honza: I don't think the patch is right. Suppose you have a function-local class declaration with a static member. Surely the context DIE you want to use is still the class one, not the function one. So, I'm not sure why we should not be able to create a testcase that fails even without LTO. I think using get_context_die here would be more appropriate. Or restrict this to DECL_CONTEXT (decl) == FUNCTION_DECL. Richard. 2005-06-09 Jan Hubicka j...@suse.cz * cgraphunit.c (cgraph_create_edges): Do not walk BLOCK; finalize local statics when doing unit-at-a-time. (cgraph_varpool_assemble_pending_decls): Output debug info. * dwarf2out.c (decls_for_scope): Skip local statics. (dwarf2out_decl): Handle local statics. * passes.c (rest_of_decl_compilation): Do not differentiate local and global statics in unit-at-a-time. * tree-inline.c (remap_decls): Put local static into unexpanded_vars_list rather than introducing duplicated VAR_DECL node. Jakub
Re: [WIP PATCH] Re: Inefficient end-of-loop value computation - missed optimization somewhere?
On Thu, Mar 8, 2012 at 3:29 PM, Ulrich Weigand uweig...@de.ibm.com wrote: Richard Guenther wrote: On Tue, Feb 28, 2012 at 4:10 PM, Ulrich Weigand uweig...@de.ibm.com wrote: I'll still need to do proper testing and benchmarking, but I thought I'd post the patch anyway just as a heads-up ... Does this look reasonable to you? Yes, that looks reasonable. Though instead of unsigned_type_for please use build_nonstandard_integer_type instead (if the type is not already unsigned). unsigned_type_for does not necessarily preserve type-precision and may even return NULL. OK, I've changed the patch to use build_nonstandard_integer_type, and it still fixes the original problem. However, on further examination it turns out that this fix is not because the expression is actually simplified at tree level. Rather, it is just that because of some minor reassociation (the +1 is moved to the end), the *RTL* optimizers now see a (just as complex but) slightly different RTL sequence, and therefore combine now just happens to be able to fold everything together (using the RTL-level simplify_plus_minus machinery). Even worse, the patch causes a number of regressions: FAIL: gcc.dg/tree-ssa/loop-15.c scan-tree-dump-times optimized \\+ 0 FAIL: gcc.dg/tree-ssa/loop-15.c scan-tree-dump-times optimized n_. \\* n_. 1 FAIL: gcc.dg/vect/no-scevccp-noreassoc-outer-4.c scan-tree-dump-times vect OUTER LOOP VECTORIZED. 1 The loop-15.c problem is probably just a missed optimization elsewhere that can be fixed. The problem is that a loop-end value used to be computed as (n - 1) * n + n which got simplified via fold_plusminus_mult_expr to n * n. When computing in unsigned, however, the expression becomes something along the lines of [...] * (unsigned) n + (unsigned) n and the top-level of fold_binary_loc removes the *outermost* NOP_EXPRs resulting in [...] * (unsigned) n + n which fold_plusminus_mult_expr no longer recognizes as something it can handle (since (unsigned) n is not operand_equal_p to n). I think this can be fixed by having fold_plusminus_mult_expr call STRIP_NOPS on sub-operands, as is already done elsewhere in fold_binary_loc. This basically fixes the test case (due to some remaining casts, the scan-tree-dump patterns still don't match as-is, but the expected code is again generated). The no-scevccp-noreassoc-outer-4.c problem is more difficult, and would seem to point to a fundamental problem with performing the computation in unsigned mode. In this case, the end value of an inner loop forms part of a scalar evolution of the outer loop. When computing the end value in unsigned type, that scalar evolution is no longer detected. In part, this is because the scalar evolution machinery might need to be improved in handling casts. In particular, in some places it only does STRIP_USELESS_TYPE_CONVERSION; changing those to STRIP_NOPS makes it detect the evolution again. But I'm not fully convinced this is a safe change ... In any case, this still doesn't make the outer loop vectorizable, since it is no longer provably finite. Without the patch, this could be proven *because* the computation of the inner loop's end value used signed arithmetic which cannot overflow. By performing the computation in unsigned, this information is in fact lost. This would seem to indicate that unconditionally using unsigned arithmetic even if not strictly necessary might not always be the best solution either. I've gone back and looked that at the original problem that (start + 1) + (int)((unsigned int) ~start + (unsigned int) end) is not simplified by fold-const. Interestingly enough, it *is* valid to simplify this expression to just plain end, even with signed arithmetic, since the transformation does not introduce any new potential overflows. This is actually recognized in fold_binary_loc, but only one special case. See the code around the following comment: /* The only case we can still associate with two variables is if they are the same, modulo negation. */ Unfortunately, this handles only A and -A; it doesn't recognize that A+1 is in fact the negation of ~A ... There is also code in tree-ssa-forwprop.c:associate_plusminus that handles quite a number of special cases where re-association *is* allowed even for signed arithmetic: (A +- B) - A - +- B (A +- B) -+ B - A (CST +- A) +- CST - CST +- A (A + CST) +- CST - A + CST ~A + A - -1 ~A + 1 - -A A - (A +- B) - -+ B A +- (B +- A) - +- B CST +- (CST +- A) - CST +- A CST +- (A +- CST) - CST +- A A + ~A - -1 This handles some cases involving ~, but still not quite the case we need for this expression. In addition, the forward propagtion logic doesn't seem to handle casts very well (as
Re: [ping] Vectorizer patches for 4.8
On Thu, Mar 8, 2012 at 3:56 PM, Ulrich Weigand uweig...@de.ibm.com wrote: Hello, Ira Rosen posted a couple of vectorizer patches intended for 4.8: http://gcc.gnu.org/ml/gcc-patches/2012-02/msg00191.html http://gcc.gnu.org/ml/gcc-patches/2012-02/msg00223.html As she is no longer able to contribute to GCC, she has asked me to help push these upstream now that 4.8 is in stage 1. We've been carrying those patches for a while in our Linaro GCC builds for ARM, and they're showing significant improvements in certain embedded test cases. I've now retested both patches on i386-linux with no regressions; the PowerPC team has helped with retesting on powerpc64-linux with no regressions, and there are (as expected) no significant changes in PowerPC benchmark results either. OK to commit the two patches to mainline? Ok. Who will be able to help with problems with those patches? Thanks, Richard. Bye, Ulrich -- Dr. Ulrich Weigand GNU Toolchain for Linux on System z and Cell BE ulrich.weig...@de.ibm.com
Re: [PATCH] Further VRP improvements
On Thu, 8 Mar 2012, Jakub Jelinek wrote:
Hi!
This patch adds ASSERT_EXPRs for
if ((int) var cmp CST)
(for cmp , =, , =). As the testcase shows, various different tests
are folded into that form.
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
Ok.
Thanks,
Richard.
2012-03-08 Jakub Jelinek ja...@redhat.com
PR tree-optimization/51721
* tree-vrp.c (register_edge_assert_for_2): Add asserts for unsvar
if (int) unsvar cmp CST.
* gcc.dg/tree-ssa/vrp64.c: New test.
--- gcc/tree-vrp.c.jj 2012-03-06 17:02:12.0 +0100
+++ gcc/tree-vrp.c2012-03-08 11:41:17.749509390 +0100
@@ -4462,8 +4462,6 @@ register_edge_assert_for_2 (tree name, e
}
}
- /* Similarly add asserts for NAME == CST and NAME being defined as
- NAME = NAME2 CST2. */
if (TREE_CODE_CLASS (comp_code) == tcc_comparison
TREE_CODE (val) == INTEGER_CST)
{
@@ -4472,7 +4470,61 @@ register_edge_assert_for_2 (tree name, e
tree val2 = NULL_TREE;
unsigned HOST_WIDE_INT mask[2] = { 0, 0 };
- /* Extract CST2 from the right shift. */
+ /* Add asserts for NAME cmp CST and NAME being defined
+ as NAME = (int) NAME2. */
+ if (!TYPE_UNSIGNED (TREE_TYPE (val))
+(comp_code == LE_EXPR || comp_code == LT_EXPR
+ || comp_code == GT_EXPR || comp_code == GE_EXPR)
+gimple_assign_cast_p (def_stmt))
+ {
+ name2 = gimple_assign_rhs1 (def_stmt);
+ if (CONVERT_EXPR_CODE_P (gimple_assign_rhs_code (def_stmt))
+INTEGRAL_TYPE_P (TREE_TYPE (name2))
+TYPE_UNSIGNED (TREE_TYPE (name2))
+(TYPE_PRECISION (TREE_TYPE (val))
+ == TYPE_PRECISION (TREE_TYPE (name2)))
+(comp_code == LE_EXPR || comp_code == GT_EXPR
+ || !tree_int_cst_equal (val,
+ TYPE_MIN_VALUE (TREE_TYPE (val
+live_on_edge (e, name2)
+!has_single_use (name2))
+ {
+ tree tmp, cst;
+ enum tree_code new_comp_code = comp_code;
+
+ cst = fold_convert (TREE_TYPE (name2),
+ TYPE_MIN_VALUE (TREE_TYPE (val)));
+ /* Build an expression for the range test. */
+ tmp = build2 (PLUS_EXPR, TREE_TYPE (name2), name2, cst);
+ cst = fold_build2 (PLUS_EXPR, TREE_TYPE (name2), cst,
+ fold_convert (TREE_TYPE (name2), val));
+ if (comp_code == LT_EXPR || comp_code == GE_EXPR)
+ {
+ new_comp_code = comp_code == LT_EXPR ? LE_EXPR : GT_EXPR;
+ cst = fold_build2 (MINUS_EXPR, TREE_TYPE (name2), cst,
+ build_int_cst (TREE_TYPE (name2), 1));
+ }
+
+ if (dump_file)
+ {
+ fprintf (dump_file, Adding assert for );
+ print_generic_expr (dump_file, name2, 0);
+ fprintf (dump_file, from );
+ print_generic_expr (dump_file, tmp, 0);
+ fprintf (dump_file, \n);
+ }
+
+ register_new_assert_for (name2, tmp, new_comp_code, cst, NULL,
+e, bsi);
+
+ retval = true;
+ }
+ }
+
+ /* Add asserts for NAME cmp CST and NAME being defined as
+ NAME = NAME2 CST2.
+
+ Extract CST2 from the right shift. */
if (is_gimple_assign (def_stmt)
gimple_assign_rhs_code (def_stmt) == RSHIFT_EXPR)
{
@@ -4500,7 +4552,6 @@ register_edge_assert_for_2 (tree name, e
val2 = fold_binary (LSHIFT_EXPR, TREE_TYPE (val), val, cst2);
}
}
-
if (val2 != NULL_TREE
TREE_CODE (val2) == INTEGER_CST
simple_cst_equal (fold_build2 (RSHIFT_EXPR,
--- gcc/testsuite/gcc.dg/tree-ssa/vrp64.c.jj 2012-03-08 12:58:05.773797384
+0100
+++ gcc/testsuite/gcc.dg/tree-ssa/vrp64.c 2012-03-08 12:57:26.0
+0100
@@ -0,0 +1,152 @@
+/* PR tree-optimization/51721 */
+/* { dg-do link } */
+/* { dg-options -O2 } */
+
+extern void link_error (void);
+
+#define BITSM1 (sizeof (int) * __CHAR_BIT__ - 1)
+
+void
+f1 (unsigned int s)
+{
+ if (s BITSM1 != 0)
+{
+ if (s == 0 || s == 5 || s == __INT_MAX__)
+ link_error ();
+}
+ else
+{
+ if (s == 1U + __INT_MAX__ || s == 6U + __INT_MAX__ || s == -1U)
+ link_error ();
+}
+}
+
+void
+f2 (int s)
+{
+ if (s BITSM1 == 0)
+{
+ if (s == -1 || s == -5 || s == -__INT_MAX__ - 1)
+ link_error ();
+}
+ else
+{
+ if (s == 0 || s == 5 || s == __INT_MAX__)
+ link_error ();
+}
+}
+
+void
+f3 (unsigned int s)
+{
+ if ((s (1U BITSM1)) != 0)
+{
+ if (s == 0 || s == 5 || s == __INT_MAX__)
+ link_error ();
+}
+ else
+{
+ if (s == 1U + __INT_MAX__ || s == 6U +
Re: Support for Runtime CPU type detection via builtins (issue5754058)
On Thu, Mar 8, 2012 at 9:35 PM, Xinliang David Li davi...@google.com wrote: On Wed, Mar 7, 2012 at 5:51 AM, Richard Guenther richard.guent...@gmail.com wrote: On Wed, Mar 7, 2012 at 1:49 AM, Sriraman Tallam tmsri...@google.com wrote: Patch for CPU detection at run-time. === Patch for CPU detection at run-time, to be used in dispatching of multi-versioned functions. Please see this discussion: http://gcc.gnu.org/ml/gcc-patches/2011-08/msg01355.html when this patch for reviewed the last time. For more detailed description: http://gcc.gnu.org/ml/gcc/2012-03/msg00074.html One of the main concerns was about making CPU detection initialization a constructor. The main point raised was about constructor ordering. I have added a priority value to the CPU detection constructor to make it very high priority so that it is guaranteed to fire before every constructor without an explicitly marked priority value of 101. However, IFUNC initializers will still fire before this constructor, so the cpu initialization routine has to be explicitly called in such initializers for which I have added a builtin: __builtin_cpu_init (). This patch adds the following new builtins: * __builtin_cpu_init * __builtin_cpu_supports_cmov * __builtin_cpu_supports_mmx * __builtin_cpu_supports_popcount * __builtin_cpu_supports_sse * __builtin_cpu_supports_sse2 * __builtin_cpu_supports_sse3 * __builtin_cpu_supports_ssse3 * __builtin_cpu_supports_sse4_1 * __builtin_cpu_supports_sse4_2 * __builtin_cpu_is_amd * __builtin_cpu_is_intel_atom * __builtin_cpu_is_intel_core2 * __builtin_cpu_is_intel * __builtin_cpu_is_intel_corei7 * __builtin_cpu_is_intel_corei7_nehalem * __builtin_cpu_is_intel_corei7_westmere * __builtin_cpu_is_intel_corei7_sandybridge * __builtin_cpu_is_amdfam10 * __builtin_cpu_is_amdfam10_barcelona * __builtin_cpu_is_amdfam10_shanghai * __builtin_cpu_is_amdfam10_istanbul * __builtin_cpu_is_amdfam15_bdver1 * __builtin_cpu_is_amdfam15_bdver2 I think the non-feature detection functions are not necessary at all. They are useful if compiler needs to do auto versioning based on cpu model. Builtin functions are not exactly cheap, nor is the scheme you invent backward/forward compatible. Instead, why not add a single builtin function, __builtin_cpu_supports(const char *), and decode from a comma-separated list of features? Unknown features are simply not present. So I can write code with only a single configure check, This is a good idea. __builtin_is_cpu (const char* ); __builtin_cpu_supports (char char*); That looks good to me. Richard. thanks, David for __builtin_cpu_supports, and cater for future features or older compilers. And of course that builtin would be even cross-platform. Implementation-wise I'll leave this to x86 maintainers to comment on. Richard. * config/i386/i386.c (build_struct_with_one_bit_fields): New function. (make_var_decl): New function. (get_field_from_struct): New function. (fold_builtin_target): New function. (ix86_fold_builtin): New function. (ix86_expand_builtin): Expand new builtins by folding them. (make_platform_builtin): New functions. (ix86_init_platform_type_builtins): Make the new builtins. (ix86_init_builtins): Make new builtins to detect CPU type. (TARGET_FOLD_BUILTIN): New macro. (IX86_BUILTIN_CPU_SUPPORTS_CMOV): New enum value. (IX86_BUILTIN_CPU_SUPPORTS_MMX): New enum value. (IX86_BUILTIN_CPU_SUPPORTS_POPCOUNT): New enum value. (IX86_BUILTIN_CPU_SUPPORTS_SSE): New enum value. (IX86_BUILTIN_CPU_SUPPORTS_SSE2): New enum value. (IX86_BUILTIN_CPU_SUPPORTS_SSE3): New enum value. (IX86_BUILTIN_CPU_SUPPORTS_SSSE3): New enum value. (IX86_BUILTIN_CPU_SUPPORTS_SSE4_1): New enum value. (IX86_BUILTIN_CPU_SUPPORTS_SSE4_2): New enum value. (IX86_BUILTIN_CPU_INIT): New enum value. (IX86_BUILTIN_CPU_IS_AMD): New enum value. (IX86_BUILTIN_CPU_IS_INTEL): New enum value. (IX86_BUILTIN_CPU_IS_INTEL_ATOM): New enum value. (IX86_BUILTIN_CPU_IS_INTEL_CORE2): New enum value. (IX86_BUILTIN_CPU_IS_INTEL_COREI7_NEHALEM): New enum value. (IX86_BUILTIN_CPU_IS_INTEL_COREI7_WESTMERE): New enum value. (IX86_BUILTIN_CPU_IS_INTEL_COREI7_SANDYBRIDGE): New enum value. (IX86_BUILTIN_CPU_IS_AMDFAM10_BARCELONA): New enum value. (IX86_BUILTIN_CPU_IS_AMDFAM10_SHANGHAI): New enum value. (IX86_BUILTIN_CPU_IS_AMDFAM10_ISTANBUL): New enum value. (IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER1): New enum value. (IX86_BUILTIN_CPU_IS_AMDFAM15H_BDVER2): New enum value. * config/i386/i386-builtin-types.def: New function type. * testsuite/gcc.target/builtin_target.c: New testcase. * libgcc/config/i386/i386-cpuinfo.c: New file. * libgcc/config/i386/t-cpuinfo:
Re: [PATCH] Further VRP improvements (take 2)
On Fri, 9 Mar 2012, Jakub Jelinek wrote:
On Thu, Mar 08, 2012 at 04:54:31PM +0100, Jakub Jelinek wrote:
This patch adds ASSERT_EXPRs for
if ((int) var cmp CST)
(for cmp , =, , =). As the testcase shows, various different tests
are folded into that form.
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
This patch no longer applies after the PR tree-optimization/52533 fix,
so here is an updated patch that does apply and contains a tiny cleanup too:
Ok.
Thanks,
Richard.
2012-03-09 Jakub Jelinek ja...@redhat.com
PR tree-optimization/51721
* tree-vrp.c (register_edge_assert_for_2): Add asserts for unsvar
if (int) unsvar cmp CST.
* gcc.dg/tree-ssa/vrp64.c: New test.
--- gcc/tree-vrp.c.jj 2012-03-09 12:30:19.348501480 +0100
+++ gcc/tree-vrp.c2012-03-09 12:53:41.844227149 +0100
@@ -4462,8 +4462,6 @@ register_edge_assert_for_2 (tree name, e
}
}
- /* Similarly add asserts for NAME == CST and NAME being defined as
- NAME = NAME2 CST2. */
if (TREE_CODE_CLASS (comp_code) == tcc_comparison
TREE_CODE (val) == INTEGER_CST)
{
@@ -4473,7 +4471,60 @@ register_edge_assert_for_2 (tree name, e
double_int mask = double_int_zero;
unsigned int prec = TYPE_PRECISION (TREE_TYPE (val));
- /* Extract CST2 from the right shift. */
+ /* Add asserts for NAME cmp CST and NAME being defined
+ as NAME = (int) NAME2. */
+ if (!TYPE_UNSIGNED (TREE_TYPE (val))
+(comp_code == LE_EXPR || comp_code == LT_EXPR
+ || comp_code == GT_EXPR || comp_code == GE_EXPR)
+gimple_assign_cast_p (def_stmt))
+ {
+ name2 = gimple_assign_rhs1 (def_stmt);
+ if (CONVERT_EXPR_CODE_P (gimple_assign_rhs_code (def_stmt))
+INTEGRAL_TYPE_P (TREE_TYPE (name2))
+TYPE_UNSIGNED (TREE_TYPE (name2))
+prec == TYPE_PRECISION (TREE_TYPE (name2))
+(comp_code == LE_EXPR || comp_code == GT_EXPR
+ || !tree_int_cst_equal (val,
+ TYPE_MIN_VALUE (TREE_TYPE (val
+live_on_edge (e, name2)
+!has_single_use (name2))
+ {
+ tree tmp, cst;
+ enum tree_code new_comp_code = comp_code;
+
+ cst = fold_convert (TREE_TYPE (name2),
+ TYPE_MIN_VALUE (TREE_TYPE (val)));
+ /* Build an expression for the range test. */
+ tmp = build2 (PLUS_EXPR, TREE_TYPE (name2), name2, cst);
+ cst = fold_build2 (PLUS_EXPR, TREE_TYPE (name2), cst,
+ fold_convert (TREE_TYPE (name2), val));
+ if (comp_code == LT_EXPR || comp_code == GE_EXPR)
+ {
+ new_comp_code = comp_code == LT_EXPR ? LE_EXPR : GT_EXPR;
+ cst = fold_build2 (MINUS_EXPR, TREE_TYPE (name2), cst,
+ build_int_cst (TREE_TYPE (name2), 1));
+ }
+
+ if (dump_file)
+ {
+ fprintf (dump_file, Adding assert for );
+ print_generic_expr (dump_file, name2, 0);
+ fprintf (dump_file, from );
+ print_generic_expr (dump_file, tmp, 0);
+ fprintf (dump_file, \n);
+ }
+
+ register_new_assert_for (name2, tmp, new_comp_code, cst, NULL,
+e, bsi);
+
+ retval = true;
+ }
+ }
+
+ /* Add asserts for NAME cmp CST and NAME being defined as
+ NAME = NAME2 CST2.
+
+ Extract CST2 from the right shift. */
if (is_gimple_assign (def_stmt)
gimple_assign_rhs_code (def_stmt) == RSHIFT_EXPR)
{
@@ -4491,7 +4542,6 @@ register_edge_assert_for_2 (tree name, e
val2 = fold_binary (LSHIFT_EXPR, TREE_TYPE (val), val, cst2);
}
}
-
if (val2 != NULL_TREE
TREE_CODE (val2) == INTEGER_CST
simple_cst_equal (fold_build2 (RSHIFT_EXPR,
--- gcc/testsuite/gcc.dg/tree-ssa/vrp64.c.jj 2012-03-09 12:45:23.287171268
+0100
+++ gcc/testsuite/gcc.dg/tree-ssa/vrp64.c 2012-03-09 12:45:23.287171268
+0100
@@ -0,0 +1,152 @@
+/* PR tree-optimization/51721 */
+/* { dg-do link } */
+/* { dg-options -O2 } */
+
+extern void link_error (void);
+
+#define BITSM1 (sizeof (int) * __CHAR_BIT__ - 1)
+
+void
+f1 (unsigned int s)
+{
+ if (s BITSM1 != 0)
+{
+ if (s == 0 || s == 5 || s == __INT_MAX__)
+ link_error ();
+}
+ else
+{
+ if (s == 1U + __INT_MAX__ || s == 6U + __INT_MAX__ || s == -1U)
+ link_error ();
+}
+}
+
+void
+f2 (int s)
+{
+ if (s BITSM1 == 0)
+{
+ if (s == -1 || s == -5 || s == -__INT_MAX__ - 1)
+ link_error ();
+}
+ else
+{
+ if (s == 0 || s == 5 || s == __INT_MAX__)
+ link_error ();
+}
+}
+
+void
+f3 (unsigned int s)
[PATCH] : Fix negative value in TEST_BIT from mem_overlaps_already_clobbered_arg_p()
Hi all, After applying patch from http://gcc.gnu.org/ml/gcc-patches/2012-02/msg00228.html we can have a negative value in call to TEST_BIT() macro in mem_overlaps_already_clobbered_arg_p() function for stored_args_map bitmap table. This can lead to reject a valid tail call optimization. This patch will ensure that i+k is positive before calling TEST_BIT() macro. 2012-03-12 Mickael Guene mickael.gu...@st.com * calls.c (mem_overlaps_already_clobbered_arg_p): Don't call TEST_BIT for negative value. Regards Mickael --- a/gcc/calls.c 2012-03-12 10:44:27.0 +0100 +++ b/gcc/calls.c 2012-03-12 10:46:16.0 +0100 @@ -1822,7 +1822,8 @@ unsigned HOST_WIDE_INT k; for (k = 0; k size; k++) - if (i + k stored_args_map-n_bits + if (i + k = 0 + i + k stored_args_map-n_bits TEST_BIT (stored_args_map, i + k)) return true; }
[VMS/committed]: Add -mpointer-size switch
Hi,
Now that '#pragma pointer_size' is supported by GCC for VMS, we can add the
'-mpointer-size' switch (modeled on /POINTER_SIZE=xx DEC-C qualifier) to set
the default pointer size. This also makes alpha64-dec-*vms* target obsolete
(but still supported by alpha*-dec-*vms*).
The patch has many chunks to adjust ™ declarations. Documentation will follow.
Tested by build gcc, committed on trunk.
Tristan.
2012-03-12 Tristan Gingold ging...@adacore.com
* config/vms/vms.opt: Add vms-opts.h header.
(mmalloc64): Use flag_vms_malloc64 flag instead of MALLOC64
target mask.
(-mvms-return-codes): Document.
(-mpointer-size): New option.
(vms_pointer_size): Add enumeration.
* config/vms/vms-opts.h: New file.
* config/vms/vms.h (TARGET_OS_CPP_BUILTINS): Define
__INITIAL_POINTER_SIZE.
(POINTER_SIZE, SIZE_TYPE, PTRDIFF_TYPE): Adjust definition.
(C_COMMON_OVERRIDE_OPTIONS): Define.
(DWARF2_ADDR_SIZE): Define.
* config/vms/vms.c (vms_patch_builtins): Adjust condition.
* config/vms/vms-protos.h (vms_c_common_override_options):
New prototype.
* config/vms/vms-c.c (vms_pragma_pointer_size): Ignore pragma
if -mno-pointer-size.
(vms_c_common_override_options): New function.
* config/ia64/vms.h (TARGET_DEFAULT): Remove MASK_MALLOC64.
* config/alpha/vms.h (TARGET_DEFAULT): Remove MASK_MALLOC64.
(MALLOC_ABI_ALIGNMENT): Use flag_vms_malloc64
and flag_vms_pointer_size.
(MASK_RETURN_ADDR): Set according to flag_vms_pointer_size.
* config.gcc (*-*-*vms*): Define xm_file.
(alpha*-dec-*vms*): Do not define xm_file.
(alpha64-dec-*vms*): Remove.
(ia64-hp-*vms*): Do not define xm_file. Simplify tm_file
and tmake_file.
ada/
2012-03-12 Tristan Gingold ging...@adacore.com
* gcc-interface/decl.c (gnat_to_gnu_param): Use flag_vms_malloc64
instead of TARGET_MALLOC64.
* gcc-interface/gigi.h (flag_vms_malloc64): Define instead
of TARGET_MALLOC64.
libgcc/
2012-03-12 Tristan Gingold ging...@adacore.com
* config/alpha/t-vms: Define HOST_LIBGCC2_CFLAGS.
* config/ia64/t-vms: Likewise.
diff --git a/gcc/ada/gcc-interface/decl.c b/gcc/ada/gcc-interface/decl.c
index 34183ba..120628e 100644
--- a/gcc/ada/gcc-interface/decl.c
+++ b/gcc/ada/gcc-interface/decl.c
@@ -5488,7 +5488,7 @@ gnat_to_gnu_param (Entity_Id gnat_param, Mechanism_Type
mech,
/* VMS descriptors are themselves passed by reference. */
if (mech == By_Short_Descriptor ||
- (mech == By_Descriptor TARGET_ABI_OPEN_VMS !TARGET_MALLOC64))
+ (mech == By_Descriptor TARGET_ABI_OPEN_VMS !flag_vms_malloc64))
gnu_param_type
= build_pointer_type (build_vms_descriptor32 (gnu_param_type,
Mechanism (gnat_param),
diff --git a/gcc/ada/gcc-interface/gigi.h b/gcc/ada/gcc-interface/gigi.h
index 5e6495a..f7d787b 100644
--- a/gcc/ada/gcc-interface/gigi.h
+++ b/gcc/ada/gcc-interface/gigi.h
@@ -999,13 +999,11 @@ extern void enumerate_modes (void (*f) (const char *,
int, int, int, int, int,
intrusive preprocessor directives. */
#ifndef TARGET_ABI_OPEN_VMS
#define TARGET_ABI_OPEN_VMS 0
-#endif
-/* VMS macro set by default, when clear forces 32bit mallocs and 32bit
+/* VMS option set by default, when clear forces 32bit mallocs and 32bit
Descriptors. Always used in combination with TARGET_ABI_OPEN_VMS
so no effect on non-VMS systems. */
-#ifndef TARGET_MALLOC64
-#define TARGET_MALLOC64 0
+#define flag_vms_malloc64 0
#endif
/* Convenient shortcuts. */
diff --git a/gcc/config.gcc b/gcc/config.gcc
index 99f0b47..053f6ec 100644
--- a/gcc/config.gcc
+++ b/gcc/config.gcc
@@ -735,6 +735,7 @@ case ${target} in
extra_objs=vms.o
target_gtfiles=$target_gtfiles \$(srcdir)/config/vms/vms.c
tm_p_file=${tm_p_file} vms/vms-protos.h
+ xm_file=vms/xm-vms.h
c_target_objs=vms-c.o
cxx_target_objs=vms-c.o
if test x$gnu_ld != xyes; then
@@ -809,14 +810,8 @@ alpha*-dec-osf5.1*)
;;
esac
;;
-alpha64-dec-*vms*)
- tm_file=${tm_file} vms/vms.h vms/vms64.h alpha/vms.h
- xm_file=alpha/xm-vms.h vms/xm-vms64.h
- tmake_file=${tmake_file} vms/t-vms64 alpha/t-vms
- ;;
alpha*-dec-*vms*)
tm_file=${tm_file} vms/vms.h alpha/vms.h
- xm_file=alpha/xm-vms.h
tmake_file=${tmake_file} alpha/t-vms
;;
arm-wrs-vxworks)
@@ -1554,9 +1549,8 @@ ia64*-*-hpux*)
esac
;;
ia64-hp-*vms*)
- tm_file=${tm_file} elfos.h ia64/sysv4.h ia64/elf.h vms/vms.h
vms/vms64.h ia64/vms.h
- xm_file=vms/xm-vms.h vms/xm-vms64.h
- tmake_file=${tmake_file} vms/t-vms64 ia64/t-ia64
+ tm_file=${tm_file} elfos.h ia64/sysv4.h vms/vms.h ia64/vms.h
+ tmake_file=${tmake_file} ia64/t-ia64
target_cpu_default=0
if test x$gas = xyes
then
diff --git a/gcc/config/alpha/vms.h
Re: [patch, RFA] delete obsolete -madjust-unroll option for SH
On Sat, Mar 10, 2012 at 5:25 AM, Sandra Loosemore san...@codesourcery.com wrote: When I was working on cleanups to invoke.texi I noticed this bit in the SH Options section: @item -madjust-unroll @opindex madjust-unroll Throttle unrolling to avoid thrashing target registers. This option only has an effect if the GCC code base supports the TARGET_ADJUST_UNROLL_MAX target hook. Well, the current manual is supposed to document what the current GCC code base does, so I looked up this target hook and found it was deleted in 2007 with a ChangeLog note that it has been dead since gcc 4.0.0. However the command-line option is still present. :-S Any reason why we should not delete this entirely, per the attached patch? I haven't built/tested this (except for the manual) but can do that if the patch isn't sufficiently obvious. Command-line options have to be preserved for backward-compatibility so old makefiles do not fail. Though the entry in the .opt file should match madjust-unroll Target Ignore Does nothing. Preserved for backward compatibility. The documentation can be removed, of course. Richard. -Sandra 2012-03-09 Sandra Loosemore san...@codesourcery.com gcc/ * config/sh/sh.opt (madjust-unroll): Delete. * config/sh/sh.h (TARGET_OPT_DEFAULT): Don't use MASK_ADJUST_UNROLL. * doc/invoke.texi (Option Summary): Remove -madjust-unroll. (SH Options): Likewise.
Re: [PATCH] Improve PHI-OPT when there are multiple phis
On Sat, Mar 10, 2012 at 8:55 AM, Andrew Pinski andrew.pin...@caviumnetworks.com wrote: Woops I forgot the patch. Ok. Thanks, Richard. Thanks, Andrew Pinski On Fri, Mar 9, 2012 at 11:45 AM, Andrew Pinski andrew.pin...@caviumnetworks.com wrote: On Tue, Jan 24, 2012 at 2:50 AM, Richard Guenther richard.guent...@gmail.com wrote: On Tue, Jan 24, 2012 at 7:34 AM, Andrew Pinski andrew.pin...@caviumnetworks.com wrote: Hi, Right now PHI-OPT does try to handle the case where we have multiple PHIs but the other PHIs have the same value for the edges we care about. This fixes the issue and allows PHI-OPT to handle a few more cases and it removes the TODO in the comments. OK For 4.8? Bootstrapped and tested on x86_64-linux-gnu with no regressions. Thanks, Andrew Pinski ChangeLog: * tree-ssa-phiopt.c (gimple_phi_singleton_for_edges): New function. The name is confusing I think, because it returns the single non-singleton PHI for the edge pair ... you can avoid choosing a better name by inlining it to its single call site. I'd maybe name it single_non_singleton_phi_for_edges. Otherwise ok. Here is the updated patch with one small change, value_replacement uses single_non_singleton_phi_for_edges now too. OK still? Bootstrapped and tested on x86_64-linux-gnu with no regressions. Thanks, Andrew Pinski ChangeLog: * tree-ssa-phiopt.c (single_non_singleton_phi_for_edges): New function. (tree_ssa_phiopt_worker): Use single_non_singleton_phi_for_edges. (value_replacement): Likewise. (empty_block_p): Check also if the PHIs for the block are empty. testsuite/ChangeLog: * gcc.dg/tree-ssa/phi-opt-7.c: New testcase. Thanks, Richard. (tree_ssa_phiopt_worker): Use gimple_phi_singleton_for_edges. (empty_block_p): Check also if the PHIs for the block are empty. testsuite/ChangeLog: * gcc.dg/tree-ssa/phi-opt-7.c: New testcase.
Re: [PR51752] publication safety violations in loop invariant motion pass
On Fri, Mar 9, 2012 at 10:48 PM, Aldy Hernandez al...@redhat.com wrote:
Note that partial PRE (enabled at -O3) can insert expressions into paths
that did _not_ execute the expression. For regular PRE you are right.
Richard.
I've thought about this some more, and Torvald's comment makes a lot of
sense. PRE can make things completely redundant, and a later pass may move
things before its publication. I believe it's wise to disable PRE inside
transactions as originally discussed. The attached patch does so.
Below is an example (from my patch) with an explanation of what may go
wrong.
Torvald is this what you were thinking of?
Richards, is this OK for the 4.7 branch and trunk (pending tests)?
+ if (flag_tm
+ gimple_in_transaction (stmt)
+ gimple_assign_single_p (stmt))
+ {
+ tree rhs = gimple_assign_rhs1 (stmt);
+ if (DECL_P (rhs) is_global_var (rhs))
+ continue;
this does not cover a read like 'a.b', nor a '*p' read. I think in some other
thread I sketched some ref_may_refer_to_global function, maybe you
remember.
Richard.
Thanks.
Aldy
+ /* Non local loads in a transaction cannot be hoisted out,
+ because they may make partially redundant expressions
+ totally redundant, which a later pass may move before its
+ publication by another thread.
+
+ For example:
+
+ __transaction_atomic {
+ if (flag)
+ y = x + 4;
+ else
+ // stuff
+ z = x + 4;
+ }
+
+ PRE can rewrite this into:
+
+ __transaction_atomic {
+ if (flag) {
+ tmp = x + 4;
+ y = tmp;
+ } else {
+ // stuff
+ tmp = x + 4;
+ }
+ z = tmp;
+ }
+
+ A later pass can move the now totally redundant [x + 4]
+ before its publication predicated by flag:
+
+ __transaction_atomic {
+ tmp = x + 4;
+ if (flag) {
+ } else {
+ // stuff
+ }
+ z = tmp;
+ */
Re: Many regressions with: [patch] Cleanup fortran/convert.c
On Sat, Mar 10, 2012 at 12:21 PM, Steven Bosscher stevenb@gmail.com wrote: On Sat, Mar 10, 2012 at 11:19 AM, Tobias Burnus bur...@net-b.de wrote: Steven Bosscher wrote: This cleans up some remnants of the ancestors of fortran's convert.c, which was copied from GNAT IIRC. I would bootstraptest this, but trunk appears to be broken for x86_64-linux right now (ICE in patch_jump_insn). But I can post this for review, at least. OK for trunk, after bootstrap+test? Your patch seems to have caused many Fortran regressions. At least I see with 185156 only one (known) failure, cf. http://gcc.gnu.org/ml/gcc-testresults/2012-03/msg01069.html While starting with 185160 there are many, many gfortran failures, cf. http://gcc.gnu.org/ml/gcc-testresults/2012-03/msg01073.html Yes, it seems that different boolean types aren't allowed. I must have looked at the wrong test results somehow. I'm testing this fix: Index: convert.c === --- convert.c (revision 185160) +++ convert.c (working copy) @@ -95,7 +95,8 @@ convert (tree type, tree expr) if (code == VOID_TYPE) return fold_build1_loc (input_location, CONVERT_EXPR, type, e); if (code == BOOLEAN_TYPE) - return truthvalue_conversion (e); + return fold_build1_loc (input_location, NOP_EXPR, type, + truthvalue_conversion (e)); return fold_convert_loc (input_location, type, truthvalue_conversion (e)); if (code == INTEGER_TYPE) return fold (convert_to_integer (type, e)); if (code == POINTER_TYPE || code == REFERENCE_TYPE)
Re: [committed] Skip gcc.dg/torture/pr52402.c execution on 32-bit hppa*-*-hpux*
On Sat, Mar 10, 2012 at 11:53 PM, John David Anglin
d...@hiauly1.hia.nrc.ca wrote:
Tested on hppa2.0w-hp-hpux11.11 and hppa64-hp-hpux11.11. Committed to trunk.
Ok for 4.7?
Ok.
Thanks,
Richard.
Dave
--
J. David Anglin dave.ang...@nrc-cnrc.gc.ca
National Research Council of Canada (613) 990-0752 (FAX:
952-6602)
2012-03-10 John David Anglin dave.ang...@nrc-cnrc.gc.ca
PR target/52450
* gcc.dg/torture/pr52402.c: Skip execution on 32-bit hppa*-*-hpux*.
Index: gcc.dg/torture/pr52402.c
===
--- gcc.dg/torture/pr52402.c (revision 185121)
+++ gcc.dg/torture/pr52402.c (working copy)
@@ -1,6 +1,7 @@
/* { dg-do run } */
/* { dg-options -w -Wno-psabi } */
/* { dg-require-effective-target int32plus } */
+/* { dg-xfail-run-if pr52450 { { hppa*-*-hpux* } { ! lp64 } } } */
typedef int v4si __attribute__((vector_size(16)));
struct T { v4si i[2]; int j; } __attribute__((packed));
Re: [C++ Patch] for c++/52465
Hello Fabien, Fabien Chêne fabien.ch...@gmail.com a écrit: [...] Index: gcc/cp/decl.c === --- gcc/cp/decl.c (revision 184891) +++ gcc/cp/decl.c (working copy) @@ -8686,6 +8686,9 @@ grokdeclarator (const cp_declarator *dec type = NULL_TREE; type_was_error_mark_node = true; } + + type = strip_using_decl (type); + I am a little bit curious as to why this change is necessary. It seems to me that the test case of your patch would pass even without this change, wouldn't it? Thanks. -- Dodji
Re: [PATCH] Don't insert white space in 'orig_option_with_args_text' for OPT_l
On Mon, 12 Mar 2012, Ludovic Court?s wrote: The patch below solves the problem in a gfortran-specific way. WDYT? I think that's the right approach for this issue. -- Joseph S. Myers jos...@codesourcery.com
Re: [PATCH] Do not use lang_hooks.types.type_for_size in signed_or_unsigned_type_for
On Wed, 7 Mar 2012, Richard Guenther wrote:
On Wed, 7 Mar 2012, Michael Matz wrote:
Hi,
On Wed, 7 Mar 2012, Richard Guenther wrote:
FAIL: gcc.dg/tree-ssa/pr31261.c scan-tree-dump-times original return
\\(char\\)
-\\(unsigned char\\) c 31; 1
FAIL: gcc.dg/tree-ssa/pr31261.c scan-tree-dump-times original return
\\(int\\)
\\(12 - \\(unsigned int\\) d\\) 7; 1
because we dump the unsigned type variant differently now. What do
people think - adjust the testcase? Adjust how we pretty-print
these non-standard integer types?
Adjusting the pretty printer would be nice anyway. Those unnamed:35
thingies hurt my eyes. Just printing int17 or uint18 would be perfectly
fine, with special casing of sizes that match the normal C types for
the target in question (so that input 'unsigned char' isn't converted to
'uint8' on one and 'uint16' on another target).
Ok, I'll do that (special-casing some precisions via *_TYPE_SIZE).
I won't touch the unnamed-unsigned-type:35 stuff, for now.
Like so.
Richard.
2012-03-12 Richard Guenther rguent...@suse.de
* tree.c (signed_or_unsigned_type_for): Use
build_nonstandard_integer_type.
(signed_type_for): Adjust documentation.
(unsigned_type_for): Likewise.
* tree-pretty-print.c (dump_generic_node): Use standard names
for non-standard integer types if available.
Index: gcc/tree.c
===
*** gcc/tree.c.orig 2012-03-12 11:11:36.0 +0100
--- gcc/tree.c 2012-03-12 12:25:37.0 +0100
*** widest_int_cst_value (const_tree x)
*** 10197,10228
return val;
}
! /* If TYPE is an integral type, return an equivalent type which is
! unsigned iff UNSIGNEDP is true. If TYPE is not an integral type,
! return TYPE itself. */
tree
signed_or_unsigned_type_for (int unsignedp, tree type)
{
! tree t = type;
! if (POINTER_TYPE_P (type))
! {
! /* If the pointer points to the normal address space, use the
!size_type_node. Otherwise use an appropriate size for the pointer
!based on the named address space it points to. */
! if (!TYPE_ADDR_SPACE (TREE_TYPE (t)))
! t = size_type_node;
! else
! return lang_hooks.types.type_for_size (TYPE_PRECISION (t), unsignedp);
! }
! if (!INTEGRAL_TYPE_P (t) || TYPE_UNSIGNED (t) == unsignedp)
! return t;
! return lang_hooks.types.type_for_size (TYPE_PRECISION (t), unsignedp);
}
! /* Returns unsigned variant of TYPE. */
tree
unsigned_type_for (tree type)
--- 10197,10222
return val;
}
! /* If TYPE is an integral or pointer type, return an integer type with
!the same precision which is unsigned iff UNSIGNEDP is true, or itself
!if TYPE is already an integer type of signedness UNSIGNEDP. */
tree
signed_or_unsigned_type_for (int unsignedp, tree type)
{
! if (TREE_CODE (type) == INTEGER_TYPE TYPE_UNSIGNED (type) == unsignedp)
! return type;
! if (!INTEGRAL_TYPE_P (type)
!!POINTER_TYPE_P (type))
! return NULL_TREE;
! return build_nonstandard_integer_type (TYPE_PRECISION (type), unsignedp);
}
! /* If TYPE is an integral or pointer type, return an integer type with
!the same precision which is unsigned, or itself if TYPE is already an
!unsigned integer type. */
tree
unsigned_type_for (tree type)
*** unsigned_type_for (tree type)
*** 10230,10236
return signed_or_unsigned_type_for (1, type);
}
! /* Returns signed variant of TYPE. */
tree
signed_type_for (tree type)
--- 10224,10232
return signed_or_unsigned_type_for (1, type);
}
! /* If TYPE is an integral or pointer type, return an integer type with
!the same precision which is signed, or itself if TYPE is already a
!signed integer type. */
tree
signed_type_for (tree type)
Index: gcc/tree-pretty-print.c
===
*** gcc/tree-pretty-print.c.orig2012-01-30 14:44:40.0 +0100
--- gcc/tree-pretty-print.c 2012-03-12 12:34:01.0 +0100
*** dump_generic_node (pretty_printer *buffe
*** 723,733
}
else if (TREE_CODE (node) == INTEGER_TYPE)
{
! pp_string (buffer, (TYPE_UNSIGNED (node)
! ? unnamed-unsigned:
! : unnamed-signed:));
! pp_decimal_int (buffer, TYPE_PRECISION (node));
! pp_string (buffer, );
}
else if (TREE_CODE (node) == COMPLEX_TYPE)
{
--- 723,763
}
else if (TREE_CODE (node) == INTEGER_TYPE)
{
! if (TYPE_PRECISION (node) == CHAR_TYPE_SIZE)
! pp_string (buffer, (TYPE_UNSIGNED (node)
!
[Fortran-dev, patch, committed] Minor fixes
I have committed a fix for one link issue (spurious
GFC_DESCRIPTOR_STRIDE_BYTES which escaped the renaming to
GFC_DESCRIPTOR_SM) - and changed data to base_addr for consistency.
Rev. 185216.
Additionally, I modified some tree-dump patterns for the extent and for
the base_addr change.
Result: 192 FAIL lines in 23 test-suite files of which 3 also exist on
the trunk.
At some point, the extent calculation should be updated. Dumps like the
following hurt, even if -O1 handles* them:
(((D.1871-dim[0].lower_bound + D.1871-dim[0].extent) + -1) -
D.1871-dim[0].lower_bound) + 1.
[* maybe -fstrict-overflow and/or -fno-protect-parens is required in
addition]
Regarding transpose_optimization_2.f90: Here,
atmp.58.dim[0].lbound = 0;
atmp.58.dim[0].ubound = 2;
is replaced by
atmp.58.dim[0].lower_bound = 0;
atmp.58.dim[0].extent = 3 - atmp.58.dim[0].lower_bound;
which increased the number of atmp by 2 (for two dims). Also something
which could be solved in the front end as both lbound and ubound are
known at compile time.
Tobias
Index: libgfortran/ChangeLog.fortran-dev
===
--- libgfortran/ChangeLog.fortran-dev (revision 185199)
+++ libgfortran/ChangeLog.fortran-dev (working copy)
@@ -1,3 +1,8 @@
+2012-03-12 Tobias Burnus bur...@net-b.de
+
+ * intrinsics/iso_c_binding.c (c_f_pointer_u0): Use
+ GFC_DESCRIPTOR_SM instead of GFC_DESCRIPTOR_STRIDE_BYTES.
+
2012-03-11 Tobias Burnus bur...@net-b.de
* ISO_Fortran_binding.h (CFI_dim_t): Remove stride/ubound,
Index: libgfortran/intrinsics/iso_c_binding.c
===
--- libgfortran/intrinsics/iso_c_binding.c (revision 185197)
+++ libgfortran/intrinsics/iso_c_binding.c (working copy)
@@ -103,7 +103,7 @@ ISO_C_BINDING_PREFIX (c_f_pointer_u0) (void *c_ptr
p = shape-base_addr;
size = GFC_DESCRIPTOR_SIZE(shape);
- source_stride = GFC_DESCRIPTOR_STRIDE_BYTES(shape,0);
+ source_stride = GFC_DESCRIPTOR_SM(shape,0);
/* shape's length (rank of the output array) */
shapeSize = GFC_DESCRIPTOR_EXTENT(shape,0);
Index: /home/tob/scratch/fortran-dev/gcc/fortran/trans-types.c
===
--- /home/tob/scratch/fortran-dev/gcc/fortran/trans-types.c (revision 185199)
+++ /home/tob/scratch/fortran-dev/gcc/fortran/trans-types.c (working copy)
@@ -1701,7 +1701,7 @@ gfc_get_array_descriptor_base (int dimen, int codi
/* Add the data member as the first element of the descriptor. */
decl = gfc_add_field_to_struct_1 (fat_type,
-get_identifier (data),
+get_identifier (base_addr),
(restricted
? prvoid_type_node
: ptr_type_node), chain);
Index: /home/tob/scratch/fortran-dev/gcc/fortran/ChangeLog.fortran-dev
===
--- /home/tob/scratch/fortran-dev/gcc/fortran/ChangeLog.fortran-dev (revision 185199)
+++ /home/tob/scratch/fortran-dev/gcc/fortran/ChangeLog.fortran-dev (working copy)
@@ -1,3 +1,8 @@
+2012-03-12 Tobias Burnus bur...@net-b.de
+
+ * trans-types.c (gfc_get_array_descriptor_base): Rename data
+ to base_addr.
+
2012-03-11 Tobias Burnus bur...@net-b.de
* trans-array.c (UBOUND_SUBFIELD, STRIDE_SUBFIELD): Remove.
Index: gfortran.dg/transpose_optimization_2.f90
===
--- gfortran.dg/transpose_optimization_2.f90 (revision 185197)
+++ gfortran.dg/transpose_optimization_2.f90 (working copy)
@@ -60,7 +60,7 @@
!
! The check below for temporaries gave 14 and 33 for parm and atmp.
!
-! { dg-final { scan-tree-dump-times parm 66 original } }
-! { dg-final { scan-tree-dump-times atmp 12 original } }
+! { dg-final { scan-tree-dump-times parm 78 original } }
+! { dg-final { scan-tree-dump-times atmp 14 original } }
! { dg-final { cleanup-tree-dump original } }
! { dg-final { cleanup-modules mod } }
Index: gfortran.dg/contiguous_3.f90
===
--- gfortran.dg/contiguous_3.f90 (revision 185197)
+++ gfortran.dg/contiguous_3.f90 (working copy)
@@ -31,10 +31,10 @@
c2 = d2
end subroutine t2
-! { dg-final { scan-tree-dump-times = a1-dim.0..stride; 0 original } }
-! { dg-final { scan-tree-dump-times = b1-dim.0..stride; 0 original } }
-! { dg-final { scan-tree-dump-times = c2-dim.0..stride; 1 original } }
-! { dg-final { scan-tree-dump-times = d2-dim.0..stride; 1 original } }
+! { dg-final { scan-tree-dump-times = a1-dim.0..sm 0 original } }
+! { dg-final { scan-tree-dump-times = b1-dim.0..sm 0 original } }
+! { dg-final { scan-tree-dump-times = c2-dim.0..sm /.fl. 4; 1 original } }
+! { dg-final { scan-tree-dump-times = d2-dim.0..sm /.fl. 4; 1 original } }
subroutine test3()
@@ -57,7 +57,7 @@
end subroutine test3
! Once for test1 (third call), once for test3 (second call)
-! { dg-final { scan-tree-dump-times
Re: [ARM] atomics for rtems, aka bare metal
Hello Richard, thank you very much for your implementation. On 03/09/2012 02:56 AM, Richard Henderson wrote: On 03/08/12 10:02, Joel Sherrill wrote: If so, it is just a matter of ifdef's to get the right code. Ok. Does arm-eabi have this support? We probably could just use the same code. There's scant bare metal support atm. For sufficiently new ARM core revisions, gcc will inline operations. In order to get all of it, you have to target ARMv7. I threw this together today. It likely needs quite a bit of spit and polish: [...] * This has received only compile testing with the arm-rtemseabi multilibs. There may well be bugs. [...] I run the test suite with this patch and now the link errors vanish. I have currently no time to test the C++ atomics, but it would be nice if we can use this work in GCC 4.7. === g++ tests === Schedule of variations: rtems-arm-edb7312 Running target rtems-arm-edb7312 Using /home/sh/rtems-testing/dejagnu/boards/rtems-arm-edb7312.exp as board description file for target. Using /usr/share/dejagnu/config/sim.exp as generic interface file for target. Using /usr/share/dejagnu/baseboards/basic-sim.exp as board description file for target. Using /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/config/default.exp as tool-and-target-specific interface file. Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/bprob/bprob.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/charset/charset.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/compat/compat.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/compat/struct-layout-1.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/debug/debug.exp ... FAIL: g++.dg/debug/const1.C -gdwarf-2 -g3 (test for excess errors) FAIL: g++.dg/debug/const1.C -gdwarf-2 -g3 -O2 (test for excess errors) FAIL: g++.dg/debug/const1.C -gdwarf-2 -g3 -O3 (test for excess errors) FAIL: g++.dg/debug/debug4.C -gdwarf-2 -g3 (test for excess errors) FAIL: g++.dg/debug/debug4.C -gdwarf-2 -g3 -O2 (test for excess errors) FAIL: g++.dg/debug/debug4.C -gdwarf-2 -g3 -O3 (test for excess errors) FAIL: g++.dg/debug/pr27657.C -gdwarf-2 -g3 (test for excess errors) FAIL: g++.dg/debug/pr27657.C -gdwarf-2 -g3 -O2 (test for excess errors) FAIL: g++.dg/debug/pr27657.C -gdwarf-2 -g3 -O3 (test for excess errors) FAIL: g++.dg/debug/trivial.C -gdwarf-2 -g3 (test for excess errors) WARNING: g++.dg/debug/trivial.C -gdwarf-2 -g3 compilation failed to produce executable FAIL: g++.dg/debug/trivial.C -gdwarf-2 -g3 -O2 (test for excess errors) WARNING: g++.dg/debug/trivial.C -gdwarf-2 -g3 -O2 compilation failed to produce executable FAIL: g++.dg/debug/trivial.C -gdwarf-2 -g3 -O3 (test for excess errors) WARNING: g++.dg/debug/trivial.C -gdwarf-2 -g3 -O3 compilation failed to produce executable Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/debug/dwarf2/dwarf2.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/dfp/dfp.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/dg.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/gcov/gcov.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/gomp/gomp.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/graphite/graphite.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/guality/guality.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/lto/lto.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/pch/pch.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/plugin/plugin.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/simulate-thread/simulate-thread.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/special/ecos.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/tls/tls.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/tm/tm.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/torture/dg-torture.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/torture/stackalign/stackalign.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/tree-prof/tree-prof.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.dg/vect/vect.exp ... Running /home/sh/archive/gcc-4.7.0-RC-20120302/gcc/testsuite/g++.old-deja/old-deja.exp ... === g++ Summary === # of expected passes 43093 # of unexpected failures 12 # of expected failures 277 # of unsupported tests 738 /home/sh/rtems-testing/gcc/b-arm-gcc/gcc/testsuite/g++/../../g++ version 4.7.0 20120307 (prerelease) (GCC) -- Sebastian Huber, embedded brains GmbH Address : Obere Lagerstr. 30, D-82178 Puchheim, Germany Phone : +49 89 18 90 80 79-6 Fax
Doc: Merge Alpha/VMS and IA-64/VMS options
Hi,
this patch merges Alpha/VMS and IA-64/VMS options as there is only one .opt for
them, so there is no need to duplicate the doc.
It also documents -mpointer-size.
Tested by 'make info'
Ok for trunk ?
Tristan.
2012-03-12 Tristan Gingold ging...@adacore.com
* doc/invoke.texi (VMS Options): Merge Alpha/VMS and IA-64/VMS
Options. Mention -mpointer-size.
diff --git a/gcc/doc/invoke.texi b/gcc/doc/invoke.texi
index a1196f0..2378ba5 100644
--- a/gcc/doc/invoke.texi
+++ b/gcc/doc/invoke.texi
@@ -569,9 +569,6 @@ Objective-C and Objective-C++ Dialects}.
-msmall-text -mlarge-text @gol
-mmemory-latency=@var{time}}
-@emph{DEC Alpha/VMS Options}
-@gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64}
-
@emph{FR30 Options}
@gccoptlist{-msmall-model -mno-lsim}
@@ -669,9 +666,6 @@ Objective-C and Objective-C++ Dialects}.
-msel-sched-dont-check-control-spec -msched-fp-mem-deps-zero-cost @gol
-msched-max-memory-insns-hard-limit -msched-max-memory-insns=@var{max-insns}}
-@emph{IA-64/VMS Options}
-@gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64}
-
@emph{LM32 Options}
@gccoptlist{-mbarrel-shift-enabled -mdivide-enabled -mmultiply-enabled @gol
-msign-extend-enabled -muser-enabled}
@@ -946,6 +940,10 @@ See RS/6000 and PowerPC Options.
@emph{VAX Options}
@gccoptlist{-mg -mgnu -munix}
+@emph{VMS Options}
+@gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64 @gol
+-mpointer-size=@var{size}}
+
@emph{VxWorks Options}
@gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
-Xbind-lazy -Xbind-now}
@@ -4816,8 +4814,8 @@ Allow using extensions of later DWARF standard version
than selected with
@item -gvms
@opindex gvms
-Produce debugging information in VMS debug format (if that is
-supported). This is the format used by DEBUG on VMS systems.
+Produce debugging information in Alpha/VMS debug format (if that is
+supported). This is the format used by DEBUG on Alpha/VMS systems.
@item -g@var{level}
@itemx -ggdb@var{level}
@@ -10331,7 +10329,6 @@ platform.
* CR16 Options::
* Darwin Options::
* DEC Alpha Options::
-* DEC Alpha/VMS Options::
* FR30 Options::
* FRV Options::
* GNU/Linux Options::
@@ -10340,7 +10337,6 @@ platform.
* i386 and x86-64 Options::
* i386 and x86-64 Windows Options::
* IA-64 Options::
-* IA-64/VMS Options::
* LM32 Options::
* M32C Options::
* M32R/D Options::
@@ -10368,6 +10364,7 @@ platform.
* TILEPro Options::
* V850 Options::
* VAX Options::
+* VMS Options::
* VxWorks Options::
* x86-64 Options::
* Xstormy16 Options::
@@ -12392,27 +12389,6 @@ Note that L3 is only valid for EV5.
@end table
@end table
-@node DEC Alpha/VMS Options
-@subsection DEC Alpha/VMS Options
-
-These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
-
-@table @gcctabopt
-@item -mvms-return-codes
-@opindex mvms-return-codes
-Return VMS condition codes from main. The default is to return POSIX
-style condition (e.g.@: error) codes.
-
-@item -mdebug-main=@var{prefix}
-@opindex mdebug-main=@var{prefix}
-Flag the first routine whose name starts with @var{prefix} as the main
-routine for the debugger.
-
-@item -mmalloc64
-@opindex mmalloc64
-Default to 64-bit memory allocation routines.
-@end table
-
@node FR30 Options
@subsection FR30 Options
@cindex FR30 Options
@@ -14325,27 +14301,6 @@ be scheduled.
@end table
-@node IA-64/VMS Options
-@subsection IA-64/VMS Options
-
-These @samp{-m} options are defined for the IA-64/VMS implementations:
-
-@table @gcctabopt
-@item -mvms-return-codes
-@opindex mvms-return-codes
-Return VMS condition codes from @code{main}. The default is to return
POSIX-style
-condition (e.g.@ error) codes.
-
-@item -mdebug-main=@var{prefix}
-@opindex mdebug-main=@var{prefix}
-Flag the first routine whose name starts with @var{prefix} as the main
-routine for the debugger.
-
-@item -mmalloc64
-@opindex mmalloc64
-Default to 64-bit memory allocation routines.
-@end table
-
@node LM32 Options
@subsection LM32 Options
@cindex LM32 options
@@ -18939,6 +18894,34 @@ will assemble with the GNU assembler.
Output code for G-format floating-point numbers instead of D-format.
@end table
+@node VMS Options
+@subsection VMS Options
+
+These @samp{-m} options are defined for the VMS implementations:
+
+@table @gcctabopt
+@item -mvms-return-codes
+@opindex mvms-return-codes
+Return VMS condition codes from @code{main}. The default is to return
POSIX-style
+condition (e.g.@ error) codes.
+
+@item -mdebug-main=@var{prefix}
+@opindex mdebug-main=@var{prefix}
+Flag the first routine whose name starts with @var{prefix} as the main
+routine for the debugger.
+
+@item -mmalloc64
+@opindex mmalloc64
+Default to 64-bit memory allocation routines.
+
+@item -mpointer-size=@var{size}
+@opindex -mpointer-size=@var{size}
+Set the default size of pointers. Possible options for @var{size} are
+@samp{32} or @samp{short} for 32 bit pointers, @samp{64} or @samp{long}
+for 64 bit pointers,
Re: [Fortran-dev, patch, committed] Minor fixes
Dear Tobias, At some point, the extent calculation should be updated. Dumps like the following hurt, even if -O1 handles* them: (((D.1871-dim[0].lower_bound + D.1871-dim[0].extent) + -1) - D.1871-dim[0].lower_bound) + 1. [* maybe -fstrict-overflow and/or -fno-protect-parens is required in addition] It was always the intention that the temporary API, which translates the present representation to the new, should be replaced in places where inefficient code is generated. Well done - you are making good progress on the first stage of changing the array descriptor. The next stage should be to get rid of the inefficiencies such as that above. Then we can add all the new fields out side of the dimension array. Cheers Paul
[v3] libstdc++/52562
Hi,
barring unexpected issues in mainline will go in 4.7.1 too.
Tested x86_64-linux.
Thanks,
Paolo.
///
2012-03-12 Paolo Carlini paolo.carl...@oracle.com
PR libstdc++/52562
* libsupc++/typeinfo (type_info::name, before, operator==,
operator!=): Mark noexcept in C++11 mode.
* libsupc++/tinfo.cc (type_info::operator==): Adjust.
* libsupc++/tinfo2.cc (type_info::before): Likewise.
* testsuite/18_support/type_info/52562.cc: New.
Index: libsupc++/tinfo2.cc
===
--- libsupc++/tinfo2.cc (revision 185225)
+++ libsupc++/tinfo2.cc (working copy)
@@ -1,6 +1,7 @@
// Methods for type_info for -*- C++ -*- Run Time Type Identification.
-// Copyright (C) 1994, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2009
+// Copyright (C) 1994, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2009,
+// 2012
// Free Software Foundation
//
// This file is part of GCC.
@@ -32,7 +33,7 @@
#if !__GXX_TYPEINFO_EQUALITY_INLINE
bool
-type_info::before (const type_info arg) const
+type_info::before (const type_info arg) const _GLIBCXX_NOEXCEPT
{
#if __GXX_MERGED_TYPEINFO_NAMES
return name () arg.name ();
Index: libsupc++/tinfo.cc
===
--- libsupc++/tinfo.cc (revision 185225)
+++ libsupc++/tinfo.cc (working copy)
@@ -1,6 +1,6 @@
// Methods for type_info for -*- C++ -*- Run Time Type Identification.
// Copyright (C) 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002,
-// 2003, 2004, 2005, 2006, 2007, 2009
+// 2003, 2004, 2005, 2006, 2007, 2009, 2012
// Free Software Foundation
//
// This file is part of GCC.
@@ -36,7 +36,7 @@
// We can't rely on common symbols being shared between shared objects.
bool std::type_info::
-operator== (const std::type_info arg) const
+operator== (const std::type_info arg) const _GLIBCXX_NOEXCEPT
{
#if __GXX_MERGED_TYPEINFO_NAMES
return name () == arg.name ();
Index: libsupc++/typeinfo
===
--- libsupc++/typeinfo (revision 185225)
+++ libsupc++/typeinfo (working copy)
@@ -98,15 +98,15 @@
/** Returns an @e implementation-defined byte string; this is not
* portable between compilers! */
-const char* name() const
+const char* name() const _GLIBCXX_NOEXCEPT
{ return __name[0] == '*' ? __name + 1 : __name; }
#if !__GXX_TYPEINFO_EQUALITY_INLINE
// In old abi, or when weak symbols are not supported, there can
// be multiple instances of a type_info object for one
// type. Uniqueness must use the _name value, not object address.
-bool before(const type_info __arg) const;
-bool operator==(const type_info __arg) const;
+bool before(const type_info __arg) const _GLIBCXX_NOEXCEPT;
+bool operator==(const type_info __arg) const _GLIBCXX_NOEXCEPT;
#else
#if !__GXX_MERGED_TYPEINFO_NAMES
/** Returns true if @c *this precedes @c __arg in the implementation's
@@ -114,12 +114,12 @@
// Even with the new abi, on systems that support dlopen
// we can run into cases where type_info names aren't merged,
// so we still need to do string comparison.
-bool before(const type_info __arg) const
+bool before(const type_info __arg) const _GLIBCXX_NOEXCEPT
{ return (__name[0] == '*' __arg.__name[0] == '*')
? __name __arg.__name
: __builtin_strcmp (__name, __arg.__name) 0; }
-bool operator==(const type_info __arg) const
+bool operator==(const type_info __arg) const _GLIBCXX_NOEXCEPT
{
return ((__name == __arg.__name)
|| (__name[0] != '*'
@@ -128,14 +128,14 @@
#else
// On some targets we can rely on type_info's NTBS being unique,
// and therefore address comparisons are sufficient.
-bool before(const type_info __arg) const
+bool before(const type_info __arg) const _GLIBCXX_NOEXCEPT
{ return __name __arg.__name; }
-bool operator==(const type_info __arg) const
+bool operator==(const type_info __arg) const _GLIBCXX_NOEXCEPT
{ return __name == __arg.__name; }
#endif
#endif
-bool operator!=(const type_info __arg) const
+bool operator!=(const type_info __arg) const _GLIBCXX_NOEXCEPT
{ return !operator==(__arg); }
#ifdef __GXX_EXPERIMENTAL_CXX0X__
Index: testsuite/18_support/type_info/52562.cc
===
--- testsuite/18_support/type_info/52562.cc (revision 0)
+++ testsuite/18_support/type_info/52562.cc (revision 0)
@@ -0,0 +1,30 @@
+// { dg-options -std=gnu++11 }
+// { dg-do compile }
+
+// Copyright (C) 2012 Free Software Foundation
+//
+// This file is part of the GNU ISO C++ Library. This library is free
+// software; you can redistribute it and/or modify it under the
+// terms of the GNU General Public License as published by the
+// Free Software Foundation; either version 3, or (at
Re: [patch, RFA] delete obsolete -madjust-unroll option for SH
Hi, Richard Guenther richard.guent...@gmail.com wrote: On Sat, Mar 10, 2012 at 5:25 AM, Sandra Loosemore san...@codesourcery.com wrote: When I was working on cleanups to invoke.texi I noticed this bit in the SH Options section: @item -madjust-unroll @opindex madjust-unroll Throttle unrolling to avoid thrashing target registers. This option only has an effect if the GCC code base supports the TARGET_ADJUST_UNROLL_MAX target hook. Well, the current manual is supposed to document what the current GCC code base does, so I looked up this target hook and found it was deleted in 2007 with a ChangeLog note that it has been dead since gcc 4.0.0. However the command-line option is still present. ,A (B:-S Any reason why we should not delete this entirely, per the attached patch? I haven't built/tested this (except for the manual) but can do that if the patch isn't sufficiently obvious. Command-line options have to be preserved for backward-compatibility so old makefiles do not fail. Though the entry in the .opt file should match madjust-unroll Target Ignore Does nothing. Preserved for backward compatibility. The documentation can be removed, of course. The patch is OK with the change suggested by Richard. Thanks! Regards, kaz
Re: [Patch]: Add include hacks for VMS headers
On Mon, Mar 12, 2012 at 3:05 AM, Tristan Gingold ging...@adacore.com wrote: Hi, this patch adds a few fix include entries for VMS so that: [things work] Looks good to me.
Re: [PATCH 07/10] addr32: Use word_mode instead of Pmode in loop expand
On Sun, Mar 11, 2012 at 2:06 AM, H.J. Lu hjl.to...@gmail.com wrote: On Thu, Mar 8, 2012 at 3:22 AM, Uros Bizjak ubiz...@gmail.com wrote: On Fri, Mar 2, 2012 at 10:02 PM, H.J. Lu hongjiu...@intel.com wrote: This patches uses word_mode instead of Pmode in loop expand since word_mode may have bigger size than Pmode. OK for trunk? Thanks. H.J. --- 2012-03-02 H.J. Lu hongjiu...@intel.com * config/i386/i386.c (ix86_expand_movmem): Use word_mode instead of Pmode on loop. (ix86_expand_setmem): Likwise. Jan, can you please comment on the changes in this patch? Here is a complete updated patch to use word_mode in ix86_expand_movmem and ix86_expand_setmem. It also fixes ix86_zero_extend_to_Pmode to handle Pmode != DImode. OK for trunk? Please rewrite ix86_zero_extend_to_Pmode to something like: rtx tmp; if (GET_MODE (exp) != Pmode) tmp = convert_to_mode (Pmode, exp, 1); return force_reg (Pmode, tmp)); Yes, the patch looks OK with this change. Thanks, Honza
[PATCH] More type_for_size call removals
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied as
obvious.
Richard.
2012-03-12 Richard Guenther rguent...@suse.de
lto/
* lto-lang.c (builtin_type_for_size): Use lto_type_for_size.
fortran/
* f95-lang.c (builtin_type_for_size): Use gfc_type_for_size.
c-common/
* c-common.c (c_common_get_narrower): Use c_common_type_for_size.
(builtin_type_for_size): Likewise.
* config/alpha/alpha.c (alpha_gimplify_va_arg): Use
build_nonstandard_integer_type.
Index: gcc/lto/lto-lang.c
===
*** gcc/lto/lto-lang.c (revision 185226)
--- gcc/lto/lto-lang.c (working copy)
*** along with GCC; see the file COPYING3.
*** 36,41
--- 36,43
#include toplev.h
#include lto-streamer.h
+ static tree lto_type_for_size (unsigned, int);
+
static tree handle_noreturn_attribute (tree *, tree, tree, int, bool *);
static tree handle_leaf_attribute (tree *, tree, tree, int, bool *);
static tree handle_const_attribute (tree *, tree, tree, int, bool *);
*** def_fn_type (builtin_type def, builtin_t
*** 523,529
static tree
builtin_type_for_size (int size, bool unsignedp)
{
! tree type = lang_hooks.types.type_for_size (size, unsignedp);
return type ? type : error_mark_node;
}
--- 525,531
static tree
builtin_type_for_size (int size, bool unsignedp)
{
! tree type = lto_type_for_size (size, unsignedp);
return type ? type : error_mark_node;
}
Index: gcc/fortran/f95-lang.c
===
*** gcc/fortran/f95-lang.c (revision 185226)
--- gcc/fortran/f95-lang.c (working copy)
*** build_builtin_fntypes (tree *fntype, tre
*** 605,611
static tree
builtin_type_for_size (int size, bool unsignedp)
{
! tree type = lang_hooks.types.type_for_size (size, unsignedp);
return type ? type : error_mark_node;
}
--- 605,611
static tree
builtin_type_for_size (int size, bool unsignedp)
{
! tree type = gfc_type_for_size (size, unsignedp);
return type ? type : error_mark_node;
}
Index: gcc/c-family/c-common.c
===
*** gcc/c-family/c-common.c (revision 185226)
--- gcc/c-family/c-common.c (working copy)
*** c_common_get_narrower (tree op, int *uns
*** 1849,1857
/* C++0x scoped enumerations don't implicitly convert to integral
type; if we stripped an explicit conversion to a larger type we
need to replace it so common_type will still work. */
! tree type = (lang_hooks.types.type_for_size
! (TYPE_PRECISION (TREE_TYPE (op)),
! TYPE_UNSIGNED (TREE_TYPE (op;
op = fold_convert (type, op);
}
return op;
--- 1849,1856
/* C++0x scoped enumerations don't implicitly convert to integral
type; if we stripped an explicit conversion to a larger type we
need to replace it so common_type will still work. */
! tree type = c_common_type_for_size (TYPE_PRECISION (TREE_TYPE (op)),
! TYPE_UNSIGNED (TREE_TYPE (op)));
op = fold_convert (type, op);
}
return op;
*** c_common_mark_addressable_vec (tree t)
*** 9259,9265
tree
builtin_type_for_size (int size, bool unsignedp)
{
! tree type = lang_hooks.types.type_for_size (size, unsignedp);
return type ? type : error_mark_node;
}
--- 9258,9264
tree
builtin_type_for_size (int size, bool unsignedp)
{
! tree type = c_common_type_for_size (size, unsignedp);
return type ? type : error_mark_node;
}
Index: gcc/config/alpha/alpha.c
===
*** gcc/config/alpha/alpha.c(revision 185226)
--- gcc/config/alpha/alpha.c(working copy)
*** alpha_gimplify_va_arg (tree valist, tree
*** 6228,6234
offset field so that it's the proper width for pointer arithmetic. */
base = get_formal_tmp_var (base_field, pre_p);
! t = fold_convert (lang_hooks.types.type_for_size (64, 0), offset_field);
offset = get_initialized_tmp_var (t, pre_p, NULL);
indirect = pass_by_reference (NULL, TYPE_MODE (type), type, false);
--- 6228,6234
offset field so that it's the proper width for pointer arithmetic. */
base = get_formal_tmp_var (base_field, pre_p);
! t = fold_convert (build_nonstandard_integer_type (64, 0), offset_field);
offset = get_initialized_tmp_var (t, pre_p, NULL);
indirect = pass_by_reference (NULL, TYPE_MODE (type), type, false);
[PATCH 0/3] Fixing expansion of misaligned MEM_REFs on strict-alignment targets
Hi, this is another iteration of my effort to fix expansion of misaligned memory accesses on strict-alignment platforms (which was suggested by Richi in http://gcc.gnu.org/ml/gcc-patches/2011-08/msg00931.html, my previous attempt was posted as http://gcc.gnu.org/ml/gcc-patches/2012-02/msg01378.html). Most importantly, I realized that the previous second patch was wrong because store_field, which is called in the same code path, itself calls store_bit_field just as I would. This then lead to unnecessary tripling of the code handling misalignment in some cases. The only problem with removing it was that this code path calls expand_normal to expand the base of the MEM_REF and that means the third patch to expand_expr_real_1 kicks in, extracts the value into a register which is returned to expand_assignment, the existing code then writes the value to a register but never stores it into memory. I came to conclusion that in this case expand_assignment should expand the MEM_REF itself rather passing it to expand_normal, just as it already does when expanding naked MEM_REF and introduced a new second patch doing just that. Because very similar code would now be twice in expand_assignment and once in expand_expr_real_1, I introduced a new function expanding it in all of these three cases (more details on that in the email with the actual patch). I have successfully bootstrapped the first patch and all three patches combined patch on x86_64-linux, i686-linux, ia64-linux (without Ada) and sparc64-linux (without Java). I will test the second on its own too, I have just not done that yet. Thanks in advance for any comments, Martin
[PATCH 1/3] Misaligned top level MEM_REFs on LHS of assignments
Hi,
this patch has not changed since I posted it the last time
(http://gcc.gnu.org/ml/gcc-patches/2012-02/msg01375.html) except for a
fixed formatting of a comment. It has been already approved by Richi
(http://gcc.gnu.org/ml/gcc-patches/2012-02/msg01389.html) and I intend
to commit it this week even if the other patches still undergo review.
Thanks,
Martin
2012-03-08 Martin Jambor mjam...@suse.cz
* expr.c (expand_assignment): Handle misaligned scalar writes to
memory through top-level MEM_REFs by calling store_bit_field.
* testsuite/gcc.dg/misaligned-expand-2.c: New test.
Index: src/gcc/expr.c
===
--- src.orig/gcc/expr.c
+++ src/gcc/expr.c
@@ -4593,10 +4593,12 @@ expand_assignment (tree to, tree from, b
if ((TREE_CODE (to) == MEM_REF
|| TREE_CODE (to) == TARGET_MEM_REF)
mode != BLKmode
+ !mem_ref_refers_to_non_mem_p (to)
((align = get_object_or_type_alignment (to))
GET_MODE_ALIGNMENT (mode))
- ((icode = optab_handler (movmisalign_optab, mode))
- != CODE_FOR_nothing))
+ (((icode = optab_handler (movmisalign_optab, mode))
+ != CODE_FOR_nothing)
+ || SLOW_UNALIGNED_ACCESS (mode, align)))
{
addr_space_t as
= TYPE_ADDR_SPACE (TREE_TYPE (TREE_TYPE (TREE_OPERAND (to, 0;
@@ -4639,11 +4641,17 @@ expand_assignment (tree to, tree from, b
if (TREE_THIS_VOLATILE (to))
MEM_VOLATILE_P (mem) = 1;
- create_fixed_operand (ops[0], mem);
- create_input_operand (ops[1], reg, mode);
- /* The movmisalignmode pattern cannot fail, else the assignment would
-silently be omitted. */
- expand_insn (icode, 2, ops);
+ if (icode != CODE_FOR_nothing)
+ {
+ create_fixed_operand (ops[0], mem);
+ create_input_operand (ops[1], reg, mode);
+ /* The movmisalignmode pattern cannot fail, else the assignment
+would silently be omitted. */
+ expand_insn (icode, 2, ops);
+ }
+ else
+ store_bit_field (mem, GET_MODE_BITSIZE (mode),
+0, 0, 0, mode, reg);
return;
}
Index: src/gcc/testsuite/gcc.dg/misaligned-expand-2.c
===
--- /dev/null
+++ src/gcc/testsuite/gcc.dg/misaligned-expand-2.c
@@ -0,0 +1,42 @@
+/* Test that expand can generate correct stores to misaligned data even on
+ strict alignment platforms. */
+
+/* { dg-do run } */
+/* { dg-options -O0 } */
+
+extern void abort ();
+
+typedef unsigned int myint __attribute__((aligned(1)));
+
+void
+foo (myint *p, unsigned int i)
+{
+ *p = i;
+}
+
+#define cst 0xdeadbeef
+#define NUM 8
+
+struct blah
+{
+ char c;
+ myint i[NUM];
+};
+
+struct blah g;
+
+#define cst 0xdeadbeef
+
+int
+main (int argc, char **argv)
+{
+ int k;
+
+ for (k = 0; k NUM; k++)
+{
+ foo (g.i[k], cst);
+ if (g.i[k] != cst)
+ abort ();
+}
+ return 0;
+}
[PATCH 2/3] Move MEM_REF expansion to a new function
Hi,
when we expand a misaligned MEM_REF on the LHS, we must not call the
code in expand_expr_real_1 if the subsequent patch is applied, because
the code generates code extracting the contents of the memory to a
register, which is of course bad if the intent is to write into that
memory. Therefore expand_assignment should expand MEM_REFs itself,
just as it do when it encounters naked misaligned ones.
In order not to have nearly identical code twice in expand_assignment
and once more in expand_expr_real_1, I put it into a separate function
expand_mem_ref_to_mem_rtx (I'll be happy to change the name to
anything more correct or fitting). Nevertheless, the existing code
pieces in expand_assignment and expand_expr_real_1 sre not exactly
identical, the differences are:
- expand_expr_real_1 handles a special case when the defining
statement of the MEM_REF base is a BIT_AND_EXPR, expand_assignment
does not. The changelog introducing the change says TER
BIT_AND_EXPRs into MEM_REFs which I suspect is a good thing for
LHSs as well, so I kept the code.
- When expanding the base, the two functions differ in the
expand_modifier they pass down to expand_expr. expand_assignment
uses EXPAND_NORMAL while expand_expr_real_1 passes EXPAND_SUM.
According to the comment in expr.h the latter seemed more permissive
and so I used that, even though I admit I do not really know what
the implications of this modifier are. Is it OK to use EXPAND_SUM
also on a LHS?
- expand_expr_real_1 calls memory_address_addr_space twice, whereas
expand_assignment replaces the first call with
convert_memory_address_addr_space. Looking at the two functions I
thought it might be OK to call memory_address_addr_space (which
itself calls convert_memory_address_addr_space) only once.
But again, my expertise in this area is limited, I'll be happy to be
shown I'm wrong here.
So far I have bootstrapped and tested this patch separately on
x86_64-linx and i686-linux. Additionally, it has also passed
bootstrap and testing on usparc64-linux and ia64-linux.
Thanks in advance for any comments,
Martin
2012-03-09 Martin Jambor mjam...@suse.cz
* expr.c (expand_mem_ref_to_mem_rtx): New function.
(expand_assignment): Call it when expanding a MEM_REF on the LHS.
(expand_expr_real_1): Likewise.
Index: src/gcc/expr.c
===
--- src.orig/gcc/expr.c
+++ src/gcc/expr.c
@@ -4565,6 +4565,47 @@ mem_ref_refers_to_non_mem_p (tree ref)
!MEM_P (DECL_RTL (base)));
}
+/* Expand a MEM_REF referring an object in memory to a MEM RTX. Any spacial
+ treatment of misalignment must be handled on top of the returned result. */
+
+static rtx
+expand_mem_ref_to_mem_rtx (tree ref)
+{
+ enum machine_mode address_mode, mode = TYPE_MODE (TREE_TYPE (ref));
+ tree base = TREE_OPERAND (ref, 0);
+ addr_space_t as
+= TYPE_ADDR_SPACE (TREE_TYPE (TREE_TYPE (TREE_OPERAND (ref, 0;
+ gimple def_stmt;
+ rtx mem, op0;
+
+ gcc_checking_assert (!mem_ref_refers_to_non_mem_p (ref));
+
+ address_mode = targetm.addr_space.address_mode (as);
+
+ if ((def_stmt = get_def_for_expr (base, BIT_AND_EXPR)))
+{
+ tree mask = gimple_assign_rhs2 (def_stmt);
+ base = build2 (BIT_AND_EXPR, TREE_TYPE (base),
+gimple_assign_rhs1 (def_stmt), mask);
+ TREE_OPERAND (ref, 0) = base;
+}
+ op0 = expand_expr (base, NULL_RTX, VOIDmode, EXPAND_SUM);
+ op0 = convert_memory_address_addr_space (address_mode, op0, as);
+ if (!integer_zerop (TREE_OPERAND (ref, 1)))
+{
+ rtx off
+= immed_double_int_const (mem_ref_offset (ref), address_mode);
+ op0 = simplify_gen_binary (PLUS, address_mode, op0, off);
+}
+ op0 = memory_address_addr_space (mode, op0, as);
+ mem = gen_rtx_MEM (mode, op0);
+ set_mem_attributes (mem, ref, 0);
+ set_mem_addr_space (mem, as);
+ if (TREE_THIS_VOLATILE (ref))
+MEM_VOLATILE_P (mem) = 1;
+ return mem;
+}
+
/* Expand an assignment that stores the value of FROM into TO. If NONTEMPORAL
is true, try generating a nontemporal store. */
@@ -4600,46 +4641,31 @@ expand_assignment (tree to, tree from, b
!= CODE_FOR_nothing)
|| SLOW_UNALIGNED_ACCESS (mode, align)))
{
- addr_space_t as
- = TYPE_ADDR_SPACE (TREE_TYPE (TREE_TYPE (TREE_OPERAND (to, 0;
struct expand_operand ops[2];
- enum machine_mode address_mode;
- rtx reg, op0, mem;
+ rtx reg, mem;
reg = expand_expr (from, NULL_RTX, VOIDmode, EXPAND_NORMAL);
reg = force_not_mem (reg);
if (TREE_CODE (to) == MEM_REF)
- {
- tree base = TREE_OPERAND (to, 0);
- address_mode = targetm.addr_space.address_mode (as);
- op0 = expand_expr (base, NULL_RTX, VOIDmode, EXPAND_NORMAL);
- op0 = convert_memory_address_addr_space (address_mode, op0, as);
- if (!integer_zerop (TREE_OPERAND (to, 1)))
- {
-
Re: [VMS/committed]: Add -mpointer-size switch
On 3/12/2012 4:48 AM, Tristan Gingold wrote: Hi, Now that '#pragma pointer_size' is supported by GCC for VMS, we can add the '-mpointer-size' switch (modeled on /POINTER_SIZE=xx DEC-C qualifier) to set the default pointer size. This also makes alpha64-dec-*vms* target obsolete (but still supported by alpha*-dec-*vms*). What about the size of Ada system.address which is also controlled by alpha64 triplet?
Re: [VMS/committed]: Add -mpointer-size switch
On Mar 12, 2012, at 3:14 PM, Douglas Rupp wrote: On 3/12/2012 4:48 AM, Tristan Gingold wrote: Hi, Now that '#pragma pointer_size' is supported by GCC for VMS, we can add the '-mpointer-size' switch (modeled on /POINTER_SIZE=xx DEC-C qualifier) to set the default pointer size. This also makes alpha64-dec-*vms* target obsolete (but still supported by alpha*-dec-*vms*). What about the size of Ada system.address which is also controlled by alpha64 triplet? You should need to configure as alpha64-dec-openvms. But indeed, I think we should always default to 64 for Ada. Tristan.
Re: [Patch]: Add include hacks for VMS headers
On Mar 12, 2012, at 3:00 PM, Bruce Korb wrote: On Mon, Mar 12, 2012 at 3:05 AM, Tristan Gingold ging...@adacore.com wrote: Hi, this patch adds a few fix include entries for VMS so that: [things work] Looks good to me. Thank you, now committed. Tristan.
Re: [PATCH] Fix PR49484, gthr requirements update (target maintainers have a looksee)
On Mon, Mar 5, 2012 at 1:17 PM, Richard Guenther richard.guent...@gmail.com wrote: On Wed, Jan 18, 2012 at 3:21 PM, Richard Guenther rguent...@suse.de wrote: This fixes PR49484 by protecting __gcov_flush against concurrent execution. To be able to use the gthread facility I have to introduce the requirement that __GTHREAD_MUTEX_INIT_FUNCTION is always available, even if __GTHREAD_MUTEX_INIT is available as otherwise no dynamic initialization of a mutex is possible. I have adjusted gthr-posix.h and gthr-single.h only - target maintainers, please update your ports accordingly in advance. Boostrap and regtest on x86_64-unknown-linux-gnu is ongoing, but this looks like something for stage1 anyway. Bootstrap and regtest finished ok. I am going to commit this without fixing eventually broken targets that I know nothing of. See above. Unless I hear objections until tomorrow. Done now, after waiting another week. Thanks, Richard. 2012-01-18 Richard Guenther rguent...@suse.de * gthr.h (__GTHREAD_MUTEX_INIT_FUNCTION): Adjust specification. * gthr-posix.h (__GTHREAD_MUTEX_INIT_FUNCTION): Define. (__gthread_mutex_init_function): New function. * gthr-single.h (__GTHREAD_MUTEX_INIT_FUNCTION): Define. PR gcov/49484 * libgcov.c: Include gthr.h. (__gcov_flush_mx): New global variable. (init_mx, init_mx_once): New functions. (__gcov_flush): Protect self with a mutex. (__gcov_fork): Re-initialize mutex after forking. * unwind-dw2-fde.c: Change condition under which to use __GTHREAD_MUTEX_INIT_FUNCTION.
Re: Doc: Merge Alpha/VMS and IA-64/VMS options
On Mar 12, 2012, at 2:07 PM, Richard Guenther wrote: On Mon, Mar 12, 2012 at 2:00 PM, Tristan Gingold ging...@adacore.com wrote: Hi, this patch merges Alpha/VMS and IA-64/VMS options as there is only one .opt for them, so there is no need to duplicate the doc. It also documents -mpointer-size. Tested by 'make info' Ok for trunk ? Ok. Thanks, committed. Tristan.
Re: [PATCH 2/3] Move MEM_REF expansion to a new function
On Mon, 12 Mar 2012, Martin Jambor wrote: Hi, when we expand a misaligned MEM_REF on the LHS, we must not call the code in expand_expr_real_1 if the subsequent patch is applied, because the code generates code extracting the contents of the memory to a register, which is of course bad if the intent is to write into that memory. Therefore expand_assignment should expand MEM_REFs itself, just as it do when it encounters naked misaligned ones. Just a quick comment here - the expand_expr_real_1 code needs to be guarded with exactly the same conditions as the misaligned LHS case to be able to call expand_expr on it and generate a naked MEM. So if that is not working you have a bug in the RHS side handling ;) Richard.
[doc] GCC 4.7 IRIX, Tru64 UNIX updates to install.texi
Reviewing install.texi for the 4.7 release, I found that it only needs
trivial updates for Tru64 UNIX and IRIX.
Tested with make doc/gccinstall.info doc/gccinstall.pdf.
Ok for 4.7 branch? I won't apply the patch to mainline since I'm about
to remove both ports there.
Thanks.
Rainer
2012-03-11 Rainer Orth r...@cebitec.uni-bielefeld.de
* doc/install.texi (Specific, alpha*-dec-osf5.1): Update binutils
reference.
(Specific, mips-sgi-irix6): Likewise.
# HG changeset patch
# Parent bf29b6f1cf94c37151c9da1d6014e7faf6d74984
GCC 4.7 IRIX, Tru64 UNIX updates to install.texi
diff --git a/gcc/doc/install.texi b/gcc/doc/install.texi
--- a/gcc/doc/install.texi
+++ b/gcc/doc/install.texi
@@ -3175,7 +3175,7 @@ or applying the patch in
the OS version used, you need a data segment size between 512 MB and
1 GB, so simply use @command{ulimit -Sd unlimited}.
-As of GNU binutils 2.21, neither GNU @command{as} nor GNU @command{ld}
+As of GNU binutils 2.22, neither GNU @command{as} nor GNU @command{ld}
are supported on Tru64 UNIX, so you must not configure GCC with
@option{--with-gnu-as} or @option{--with-gnu-ld}.
@@ -4066,7 +4066,7 @@ Look for @file{/usr/lib64/libc.so.1} to
have the 64-bit libraries installed.
GCC must be configured with GNU @command{as}. The latest version, from GNU
-binutils 2.21, is known to work. On the other hand, bootstrap fails
+binutils 2.22, is known to work. On the other hand, bootstrap fails
with GNU @command{ld} at least since GNU binutils 2.17.
The @option{--enable-libgcj}
--
-
Rainer Orth, Center for Biotechnology, Bielefeld University
Re: [doc] GCC 4.7 IRIX, Tru64 UNIX updates to install.texi
On Mon, 12 Mar 2012, Rainer Orth wrote: Reviewing install.texi for the 4.7 release, I found that it only needs trivial updates for Tru64 UNIX and IRIX. Tested with make doc/gccinstall.info doc/gccinstall.pdf. Ok for 4.7 branch? I won't apply the patch to mainline since I'm about to remove both ports there. Ok. Thanks, Richard. Thanks. Rainer 2012-03-11 Rainer Orth r...@cebitec.uni-bielefeld.de * doc/install.texi (Specific, alpha*-dec-osf5.1): Update binutils reference. (Specific, mips-sgi-irix6): Likewise.
[doc] GCC 4.7 Solaris updates to install.texi
The following patch updates install.texi for Solaris, and applies
equally to the 4.7 branch and mainline.
Tested with make doc/gccinstall.info doc/gccinstall.pdf, ok for mainline
and 4.7 branch?
Rainer
2012-03-11 Rainer Orth r...@cebitec.uni-bielefeld.de
* doc/install.texi (Specific, i?86-*-solaris2.[89]): Update
binutils reference.
(Specific, i?86-*-solaris2.10): Remove GCC 4.0 reference.
Update binutils references.
(Specific, *-*-solaris2*): Mention bundled GCC in Solaris 10 and 11.
Update binutils reference.
Update Sun as/GNU ld caveat.
Document binutils largefile requirement for LTO plugin.
Remove referent to alternate libpthread.
# HG changeset patch
# Parent 87d293abd3111de02ec54b8b863d495e7bf7e1f8
GCC 4.7 Solaris updates to install.texi
diff --git a/gcc/doc/install.texi b/gcc/doc/install.texi
--- a/gcc/doc/install.texi
+++ b/gcc/doc/install.texi
@@ -3612,7 +3612,7 @@ The Sun assembler in Solaris 8 and 9 has
While GCC works around them, several features are missing, so it is
@c FIXME: which ones?
recommended to use the GNU assembler instead. There is no bundled
-version, but the current version, from GNU binutils 2.21, is known to
+version, but the current version, from GNU binutils 2.22, is known to
work.
Solaris@tie{}2/x86 doesn't support the execution of SSE/SSE2 instructions
@@ -3628,8 +3628,7 @@ you have the patch installed, you can co
hr /
@end html
@heading @anchor{ix86-x-solaris210}i?86-*-solaris2.10
-Use this for Solaris 10 or later on x86 and x86-64 systems. This
-configuration is supported by GCC 4.0 and later versions only. Starting
+Use this for Solaris 10 or later on x86 and x86-64 systems. Starting
with GCC 4.7, there is also a 64-bit @samp{amd64-*-solaris2.1[0-9]*} or
@samp{x86_64-*-solaris2.1[0-9]*} configuration that corresponds to
@samp{sparcv9-sun-solaris2*}.
@@ -3638,7 +3637,7 @@ It is recommended that you configure GCC
@file{/usr/sfw/bin/gas}. The versions included in Solaris 10, from GNU
binutils 2.15, and Solaris 11, from GNU binutils 2.19, work fine,
although the current version, from GNU binutils
-2.21, is known to work, too. Recent versions of the Sun assembler in
+2.22, is known to work, too. Recent versions of the Sun assembler in
@file{/usr/ccs/bin/as} work almost as well, though.
@c FIXME: as patch requirements?
@@ -3646,7 +3645,7 @@ For linking, the Sun linker, is preferre
linker instead, which is available in @file{/usr/sfw/bin/gld}, note that
due to a packaging bug the version in Solaris 10, from GNU binutils
2.15, cannot be used, while the version in Solaris 11, from GNU binutils
-2.19, works, as does the latest version, from GNU binutils 2.21.
+2.19, works, as does the latest version, from GNU binutils 2.22.
To use GNU @command{as}, configure with the options
@option{--with-gnu-as --with-as=@//usr/@/sfw/@/bin/@/gas}. It may be necessary
@@ -4217,8 +4216,10 @@ Support for Solaris 8 has been obsoleted
enabled by configuring with @option{--enable-obsolete}. Support will be
removed in GCC 4.8. Support for Solaris 7 has been removed in GCC 4.6.
-Sun does not ship a C compiler with Solaris 2, though you can download
-the Sun Studio compilers for free. Alternatively,
+Sun does not ship a C compiler with Solaris 2 before Solaris 10, though
+you can download the Sun Studio compilers for free. In Solaris 10 and
+11, GCC 3.4.3 is available in @command{/usr/sfw/bin/gcc}. Solaris 11
+also provides GCC 4.5.2 in @command{/usr/gcc/4.5/bin/gcc}. Alternatively,
you can install a pre-built GCC to bootstrap and install GCC. See the
@uref{binaries.html,,binaries page} for details.
@@ -4261,17 +4262,19 @@ We recommend the use of the Sun assemble
conjunction with the Sun linker. The GNU @command{as}
versions included in Solaris 10, from GNU binutils 2.15, and Solaris 11,
from GNU binutils 2.19, are known to work. They can be found in
-@file{/usr/sfw/bin/gas}. Current versions of GNU binutils (2.21)
+@file{/usr/sfw/bin/gas}. Current versions of GNU binutils (2.22)
are known to work as well. Note that your mileage may vary
if you use a combination of the GNU tools and the Sun tools: while the
combination GNU @command{as} + Sun @command{ld} should reasonably work,
-the reverse combination Sun @command{as} + GNU @command{ld} is known to
-cause memory corruption at runtime in some cases for C++ programs.
+the reverse combination Sun @command{as} + GNU @command{ld} may fail to
+build or cause memory corruption at runtime in some cases for C++ programs.
@c FIXME: still?
GNU @command{ld} usually works as well, although the version included in
Solaris 10 cannot be used due to several bugs. Again, the current
-version (2.21) is known to work, but generally lacks platform specific
-features, so better stay with Sun @command{ld}.
+version (2.22) is known to work, but generally lacks platform specific
+features, so better stay with Sun
Re: [PATCH 2/3] Move MEM_REF expansion to a new function
Hi, On Mon, 12 Mar 2012, Martin Jambor wrote: when we expand a misaligned MEM_REF on the LHS, we must not call the code in expand_expr_real_1 if the subsequent patch is applied, because the code generates code extracting the contents of the memory to a register, which is of course bad if the intent is to write into that memory. Then expand_expr_real_1 should be called with EXPAND_WRITE modifier, instead of any of the others. Then it will (or should) return an lvalue. That might still be wrong for alignment reasons, but writing into the so returned rtx will change the original object. Therefore expand_assignment should expand MEM_REFs itself, just as it do when it encounters naked misaligned ones. I think this goes into the wrong direction. expand_assignment shouldn't create an lvalue rtx for any REFs itself. It should call expand_expr with EXPAND_WRITE, and that should do the right thing (i.e. what's now done directly in expand_assignment). I realize the docu of EXPAND_WRITE is lacking, but here's what I think it should do (and what I think it actually also mostly does already): Given EXPAND_WRITE expand_expr is required to be called on an (sub)object, i.e. an lvalue, and it should return an RTX lvalue (a REG or MEM) that if written into is changing the originally specified tree lvalue (i.e. not some temporary storage). That doesn't mean that the result of expand_expr(EXPAND_WRITE) is directly usable in a simple RTL (set) pattern as LHS in all cases. For instance it won't be directly usable when it's misalign. Dealing with this situation is left to the caller, i.e. expand_assignment mostly. - When expanding the base, the two functions differ in the expand_modifier they pass down to expand_expr. expand_assignment uses EXPAND_NORMAL while expand_expr_real_1 passes EXPAND_SUM. According to the comment in expr.h the latter seemed more permissive and so I used that, even though I admit I do not really know what the implications of this modifier are. Is it OK to use EXPAND_SUM also on a LHS? No, but it might not matter in the situations you are facing, haven't checked. EXPAND_SUM can return a PLUS rtx, e.g. (plus (p60) (const_int 4)) for an offsetted address. Naturally you can't assign into such a plus rtx. But if you only expand the base with that modifier (which for BLKmode bases actually means expanding to a MEM containing the address of base) you should be fine with EXPAND_SUM, it won't be used I think. Ciao, Michael.
Re: [doc] GCC 4.7 Solaris updates to install.texi
On Mon, 12 Mar 2012, Rainer Orth wrote: The following patch updates install.texi for Solaris, and applies equally to the 4.7 branch and mainline. Tested with make doc/gccinstall.info doc/gccinstall.pdf, ok for mainline and 4.7 branch? Ok. Thanks, Richard. Rainer 2012-03-11 Rainer Orth r...@cebitec.uni-bielefeld.de * doc/install.texi (Specific, i?86-*-solaris2.[89]): Update binutils reference. (Specific, i?86-*-solaris2.10): Remove GCC 4.0 reference. Update binutils references. (Specific, *-*-solaris2*): Mention bundled GCC in Solaris 10 and 11. Update binutils reference. Update Sun as/GNU ld caveat. Document binutils largefile requirement for LTO plugin. Remove referent to alternate libpthread. -- Richard Guenther rguent...@suse.de SUSE / SUSE Labs SUSE LINUX Products GmbH - Nuernberg - AG Nuernberg - HRB 16746 GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer
Re: [PATCH][1/n] No longer sign-extend sizetype constants, remove TYPE_IS_SIZETYPE
On Fri, 9 Mar 2012, Eric Botcazou wrote:
Well. I suppose fixing that negative DECL_FIELD_OFFSET thing should
be #1 priority.
Tentative patch attached (you need an up-to-date tree). We cannot really get
rid of the negative offsets for thin pointers, so the patch pulls them out of
record types and uses pointer arithmetics instead.
Thanks. That seems to get me into Ada bootstrap land together with
the following two patches applied (but not when I apply only the
first patch due to the stale TREE_OVERFLOW bits on type/decl sizes).
Richard.
2012-03-06 Richard Guenther rguent...@suse.de
* tree.c (valid_constant_size_p): New function.
* tree.h (valid_constant_size_p): Declare.
* cfgexpand.c (expand_one_var): Adjust check for too large
variables by using valid_constant_size_p.
* varasm.c (assemble_variable): Likewise.
c/
* c-decl.c (grokdeclarator): Properly check for sizes that
cover more than half of the address-space.
cp/
* decl.c (grokdeclarator): Properly check for sizes that
cover more than half of the address-space.
Index: trunk/gcc/tree.c
===
*** trunk.orig/gcc/tree.c 2012-03-06 13:54:25.0 +0100
--- trunk/gcc/tree.c2012-03-06 14:46:08.0 +0100
*** compare_tree_int (const_tree t, unsigned
*** 6829,6834
--- 6829,6848
return 1;
}
+ /* Return true if SIZE represents a constant size that is in bounds of
+what the middle-end and the backend accepts (covering not more than
+half of the address-space). */
+
+ bool
+ valid_constant_size_p (const_tree size)
+ {
+ if (! host_integerp (size, 1)
+ || TREE_OVERFLOW (size)
+ || tree_int_cst_sign_bit (size) != 0)
+ return false;
+ return true;
+ }
+
/* Return true if CODE represents an associative tree code. Otherwise
return false. */
bool
Index: trunk/gcc/varasm.c
===
*** trunk.orig/gcc/varasm.c 2012-03-06 13:54:25.0 +0100
--- trunk/gcc/varasm.c 2012-03-06 14:46:08.0 +0100
*** assemble_variable (tree decl, int top_le
*** 1987,1993
return;
if (! dont_output_data
!! host_integerp (DECL_SIZE_UNIT (decl), 1))
{
error (size of variable %q+D is too large, decl);
return;
--- 1987,1993
return;
if (! dont_output_data
!! valid_constant_size_p (DECL_SIZE_UNIT (decl)))
{
error (size of variable %q+D is too large, decl);
return;
Index: trunk/gcc/c-decl.c
===
*** trunk.orig/gcc/c-decl.c 2012-03-06 13:54:25.0 +0100
--- trunk/gcc/c-decl.c 2012-03-06 14:46:08.0 +0100
*** grokdeclarator (const struct c_declarato
*** 5798,5809
}
}
! /* Did array size calculations overflow? */
!
if (TREE_CODE (type) == ARRAY_TYPE
COMPLETE_TYPE_P (type)
TREE_CODE (TYPE_SIZE_UNIT (type)) == INTEGER_CST
!TREE_OVERFLOW (TYPE_SIZE_UNIT (type)))
{
if (name)
error_at (loc, size of array %qE is too large, name);
--- 5798,5809
}
}
! /* Did array size calculations overflow or does the array cover more
! than half of the address-space? */
if (TREE_CODE (type) == ARRAY_TYPE
COMPLETE_TYPE_P (type)
TREE_CODE (TYPE_SIZE_UNIT (type)) == INTEGER_CST
!! valid_constant_size_p (TYPE_SIZE_UNIT (type)))
{
if (name)
error_at (loc, size of array %qE is too large, name);
Index: trunk/gcc/cp/decl.c
===
*** trunk.orig/gcc/cp/decl.c2012-03-06 13:54:25.0 +0100
--- trunk/gcc/cp/decl.c 2012-03-06 14:46:08.0 +0100
*** grokdeclarator (const cp_declarator *dec
*** 9635,9646
error (non-parameter %qs cannot be a parameter pack, name);
}
! /* Did array size calculations overflow? */
!
if (TREE_CODE (type) == ARRAY_TYPE
COMPLETE_TYPE_P (type)
TREE_CODE (TYPE_SIZE_UNIT (type)) == INTEGER_CST
!TREE_OVERFLOW (TYPE_SIZE_UNIT (type)))
{
error (size of array %qs is too large, name);
/* If we proceed with the array type as it is, we'll eventually
--- 9635,9646
error (non-parameter %qs cannot be a parameter pack, name);
}
! /* Did array size calculations overflow or does the array cover more
! than half of the address-space? */
if (TREE_CODE (type) == ARRAY_TYPE
COMPLETE_TYPE_P (type)
TREE_CODE (TYPE_SIZE_UNIT (type)) == INTEGER_CST
!! valid_constant_size_p (TYPE_SIZE_UNIT (type)))
{
error (size of array %qs is too large, name);
/* If we proceed with the array type as it is, we'll
[VMS/committed]: Fix ABI issue on vms/ia64
Hi,
ia64/VMS doesn't closely follow the 'standard' ia64 ABI (in order to be
backward compatible with VAX).
We recently found one mismatch while passing FP parameters.
Manually tested and checked - no test case added as this require another
compiler to test.
Committed on trunk.
Tristan.
2012-03-12 Tristan Gingold ging...@adacore.com
* config/ia64/ia64.c (ia64_function_arg_1): Move code around.
(ia64_function_arg_advance): Ditto.
Index: gcc/config/ia64/ia64.c
===
--- gcc/config/ia64/ia64.c (revision 185231)
+++ gcc/config/ia64/ia64.c (working copy)
@@ -4491,6 +4491,15 @@
if (cum-words + offset = MAX_ARGUMENT_SLOTS)
return 0;
+ /* On OpenVMS argument is either in Rn or Fn. */
+ if (TARGET_ABI_OPEN_VMS)
+{
+ if (FLOAT_MODE_P (mode))
+ return gen_rtx_REG (mode, FR_ARG_FIRST + cum-words);
+ else
+ return gen_rtx_REG (mode, basereg + cum-words);
+}
+
/* Check for and handle homogeneous FP aggregates. */
if (type)
hfa_mode = hfa_element_mode (type, 0);
@@ -4577,15 +4586,6 @@
return gen_rtx_PARALLEL (mode, gen_rtvec_v (i, loc));
}
- /* On OpenVMS variable argument is either in Rn or Fn. */
- else if (TARGET_ABI_OPEN_VMS named == 0)
-{
- if (FLOAT_MODE_P (mode))
- return gen_rtx_REG (mode, FR_ARG_FIRST + cum-words);
- else
- return gen_rtx_REG (mode, basereg + cum-words);
-}
-
/* Integral and aggregates go in general registers. If we have run out of
FR registers, then FP values must also go in general registers. This can
happen when we have a SFmode HFA. */
@@ -4736,6 +4736,14 @@
cum-atypes[cum-words] = ia64_arg_type (mode);
cum-words += words + offset;
+ /* On OpenVMS argument is either in Rn or Fn. */
+ if (TARGET_ABI_OPEN_VMS named == 0)
+{
+ cum-int_regs = cum-words;
+ cum-fp_regs = cum-words;
+ return;
+}
+
/* Check for and handle homogeneous FP aggregates. */
if (type)
hfa_mode = hfa_element_mode (type, 0);
@@ -4776,13 +4784,6 @@
cum-fp_regs = fp_regs;
}
- /* On OpenVMS variable argument is either in Rn or Fn. */
- else if (TARGET_ABI_OPEN_VMS named == 0)
-{
- cum-int_regs = cum-words;
- cum-fp_regs = cum-words;
-}
-
/* Integral and aggregates go in general registers. So do TFmode FP values.
If we have run out of FR registers, then other FP values must also go in
general registers. This can happen when we have a SFmode HFA. */
Re: Remove obsolete OpenBSD/MIPS support
Richard Sandiford rdsandif...@googlemail.com writes:
Rainer Orth r...@cebitec.uni-bielefeld.de writes:
I'm currently working on removing the obsolete Tru64 UNIX and IRIX
ports. When IRIX is gone, the obsoleted OpenBSD/MIPS is the only
remaining port that uses MIPS_DEBUGGING_INFO (which I plan to remove as
a followup once IRIX is gone).
The following patch has been included in a i386-pc-solaris2.10 bootstrap
to make sure that no obvious syntactic errors have crept in.
Ok for mainline?
Definitely. Thanks for saving me the legwork :-)
There are also some SDB_OUTPUT_SOURCE_LINE references in mips.c and
mips.h that could go. Removing those is preapproved if you want to
do it as part of the same patch, otherwise I can do it this weekend.
Here's what I've installed. The patch was bootstrapped (together with a
bunch of other removal patches to be submitted shortly) on
i386-pc-solaris2.11 and tested on an i386-solaris2.11 x mips-elf C-only
cross to make sure cc1 still builds.
Rainer
2012-02-24 Rainer Orth r...@cebitec.uni-bielefeld.de
libgcc:
* config.host (mips*-*-openbsd*): Remove.
gcc:
* config.gcc (mips*-*-openbsd*): Remove.
* config/mips/openbsd.h: Remove.
* config/mips/sdb.h: Remove.
* config/mips/mips.h (SDB_OUTPUT_SOURCE_LINE): Remove.
* config/mips/mips.c (sdb_label_count): Remove.
(mips_debugger_offset): Remove #if 0 code.
(mips_output_function_prologue) [SDB_DEBUGGING_INFO]: Remove.
* config/mips/sde.h (SDB_DEBUGGING_INFO): Don't undef.
contrib:
* config-list.mk (LIST): Remove mips-openbsd.
# HG changeset patch
# Parent 629d4a30bc2bbb27d12aa73bd6e0221cecf83144
Remove obsolete OpenBSD/MIPS support
diff --git a/contrib/config-list.mk b/contrib/config-list.mk
--- a/contrib/config-list.mk
+++ b/contrib/config-list.mk
@@ -39,7 +39,7 @@ LIST = alpha-linux-gnu alpha-freebsd6 al
mips-sgi-irix6.5OPT-with-stabsOPT-enable-threads=posixOPT-enable-obsolete \
mips-netbsd \
mips64el-st-linux-gnu mips64octeon-linux mipsisa64r2-linux \
- mipsisa32r2-linux-gnu mips-openbsd mipsisa64r2-sde-elf mipsisa32-elfoabi \
+ mipsisa32r2-linux-gnu mipsisa64r2-sde-elf mipsisa32-elfoabi \
mipsisa64-elfoabi mipsisa64r2el-elf mipsisa64sr71k-elf mipsisa64sb1-elf \
mipsel-elf mips64-elf mips64vr-elf mips64orion-elf mips-rtems \
mips-wrs-vxworks mipstx39-elf mmix-knuth-mmixware mn10300-elf moxie-elf \
diff --git a/gcc/config.gcc b/gcc/config.gcc
--- a/gcc/config.gcc
+++ b/gcc/config.gcc
@@ -251,7 +251,6 @@ case ${target} in
| arm*-*-uclinux* \
| arm*-wince-pe* \
| mips-sgi-irix6.5 \
- | mips*-*-openbsd* \
| score-*\
| *-*-solaris2.8* \
)
@@ -1744,17 +1743,6 @@ mips*-*-linux*)# Linux MIPS, either
esac
test x$with_llsc != x || with_llsc=yes
;;
-mips*-*-openbsd*)
- tm_defines=${tm_defines} OBSD_HAS_DECLARE_FUNCTION_NAME OBSD_HAS_DECLARE_OBJECT OBSD_HAS_CORRECT_SPECS
- target_cpu_default=MASK_ABICALLS
- tm_file=mips/mips.h openbsd.h openbsd-stdint.h openbsd-libpthread.h mips/openbsd.h mips/sdb.h
- extra_options=${extra_options} openbsd.opt
- case ${target} in
- mips*el-*-openbsd*)
- tm_defines=${tm_defines} TARGET_ENDIAN_DEFAULT=0;;
- *) tm_defines=${tm_defines} TARGET_ENDIAN_DEFAULT=MASK_BIG_ENDIAN;;
-esac
- ;;
mips*-sde-elf*)
tm_file=elfos.h newlib-stdint.h ${tm_file} mips/elf.h mips/sde.h
tmake_file=mips/t-sde
diff --git a/gcc/config/mips/mips.c b/gcc/config/mips/mips.c
--- a/gcc/config/mips/mips.c
+++ b/gcc/config/mips/mips.c
@@ -494,9 +494,6 @@ int num_source_filenames;
written anything yet. */
const char *current_function_file = ;
-/* A label counter used by PUT_SDB_BLOCK_START and PUT_SDB_BLOCK_END. */
-int sdb_label_count;
-
/* Arrays that map GCC register numbers to debugger register numbers. */
int mips_dbx_regno[FIRST_PSEUDO_REGISTER];
int mips_dwarf_regno[FIRST_PSEUDO_REGISTER];
@@ -8114,13 +8111,6 @@ mips_debugger_offset (rtx addr, HOST_WID
offset += cfun-machine-frame.hard_frame_pointer_offset;
}
- /* sdbout_parms does not want this to crash for unrecognized cases. */
-#if 0
- else if (reg != arg_pointer_rtx)
-fatal_insn (mips_debugger_offset called with non stack/frame/arg pointer,
- addr);
-#endif
-
return offset;
}
@@ -10129,11 +10119,6 @@ mips_output_function_prologue (FILE *fil
{
const char *fnname;
-#ifdef SDB_DEBUGGING_INFO
- if (debug_info_level != DINFO_LEVEL_TERSE write_symbols == SDB_DEBUG)
-SDB_OUTPUT_SOURCE_LINE (file, DECL_SOURCE_LINE (current_function_decl));
-#endif
-
/* In MIPS16 mode, we may need to generate a non-MIPS16 stub to handle
floating-point arguments. */
if (TARGET_MIPS16
diff --git a/gcc/config/mips/mips.h b/gcc/config/mips/mips.h
--- a/gcc/config/mips/mips.h
+++ b/gcc/config/mips/mips.h
@@ -2551,10 +2551,6 @@ do \
} \
while (0)
-/* Use .loc directives for SDB line numbers. */
-#define
Re: PATCH: Properly set ix86_gen_leave and ix86_gen_monitor
On Mon, Mar 12, 2012 at 12:40 AM, Uros Bizjak ubiz...@gmail.com wrote: On Mon, Mar 12, 2012 at 12:02 AM, H.J. Lu hongjiu...@intel.com wrote: leave_rex64 works on DImode and sse3_monitor64 works on Pmode. This patch properly sets ix86_gen_leave and ix86_gen_monitor, depending on TARGET_64BIT and Pmode. Tested on Linux/x86-64. OK for trunk? 2012-03-11 H.J. Lu hongjiu...@intel.com * config/i386/i386.c (ix86_option_override_internal): Properly set ix86_gen_leave and ix86_gen_monitor. Check Pmode == DImode. * config/i386/sse.md (sse3_monitor64): Renamed to ... (sse3_monitor64_mode): This. Please provide more descriptive ChangeLog. You are changing the check on quite some global variables, please state them explicitly. Here is the ChangeLog entry I checked in. Otherwise the patch is OK. BTW: I sincerely hope you reviewed all the places where these globals are used... Uros. I verified that those ix86_gen_xxx function pointers I changed to check Pmode == DImode are used to work on Pmode, like stack or pointers. Thanks. -- H.J. --- 2012-03-12 H.J. Lu hongjiu...@intel.com * config/i386/i386.c (ix86_option_override_internal): Properly set ix86_gen_leave and ix86_gen_monitor. Check Pmode == DImode, instead of TARGET_64BIT, to set ix86_gen_add3, ix86_gen_sub3, ix86_gen_one_cmpl2, ix86_gen_andsp, ix86_gen_allocate_stack_worker, ix86_gen_adjust_stack_and_probe and ix86_gen_probe_stack_range. * config/i386/sse.md (sse3_monitor64): Renamed to ... (sse3_monitor64_mode): This.
[build] Remove extra_passes
As suggested, this patch removes the now unused extra_passes mechanism. Bootstrapped with no regressions on i386-pc-solaris2.11, ok for mainline? Rainer 2012-03-09 Rainer Orth r...@cebitec.uni-bielefeld.de * config.gcc (extra_passes): Remove. * configure.ac (extra_passes): Don't substitute. * configure: Regenerate. * Makefile.in (EXTRA_PASSES): Remove. (GCC_PASSES): Remove $(EXTRA_PASSES). (MOSTLYCLEANFILES): Likewise. (native): Likewise. (install-common): Likewise. # HG changeset patch # Parent 72aa91c9cd9c14a295c4c4f7c8e60b83e9ce7f2f Remove extra_passes diff --git a/gcc/Makefile.in b/gcc/Makefile.in --- a/gcc/Makefile.in +++ b/gcc/Makefile.in @@ -670,11 +670,8 @@ INHIBIT_LIBC_CFLAGS = -Dinhibit_libc endif # List of extra executables that should be compiled for this target machine -# that are used for compiling from source code to object code. +# that are used when linking. # The rules for compiling them should be in the t-* file for the machine. -EXTRA_PASSES =@extra_passes@ - -# Like EXTRA_PASSES, but these are used when linking. EXTRA_PROGRAMS = @extra_programs@ # List of extra object files that should be compiled and linked with @@ -724,7 +721,7 @@ COMPILERS = cc1$(exeext) @all_compilers@ # List of things which should already be built whenever we try to use xgcc # to compile anything (without linking). -GCC_PASSES=xgcc$(exeext) cc1$(exeext) specs $(EXTRA_PASSES) +GCC_PASSES=xgcc$(exeext) cc1$(exeext) specs # Directory to link to, when using the target `maketest'. DIR = ../gcc @@ -1481,7 +1478,7 @@ MOSTLYCLEANFILES = insn-flags.h insn-con tm-preds.h tm-constrs.h checksum-options \ tree-check.h min-insn-modes.c insn-modes.c insn-modes.h \ genrtl.h gt-*.h gtype-*.h gtype-desc.c gtyp-input.list \ - xgcc$(exeext) cpp$(exeext) cc1$(exeext) $(EXTRA_PASSES) \ + xgcc$(exeext) cpp$(exeext) cc1$(exeext) \ $(EXTRA_PROGRAMS) gcc-cross$(exeext) \ $(SPECS) collect2$(exeext) gcc-ar$(exeext) gcc-nm$(exeext) \ gcc-ranlib$(exeext) \ @@ -1717,7 +1714,7 @@ rest.encap: lang.rest.encap # This is what is made with the host's compiler # whether making a cross compiler or not. native: config.status auto-host.h build-@POSUB@ $(LANGUAGES) \ - $(EXTRA_PASSES) $(EXTRA_PROGRAMS) $(COLLECT2) lto-wrapper$(exeext) \ + $(EXTRA_PROGRAMS) $(COLLECT2) lto-wrapper$(exeext) \ gcc-ar$(exeext) gcc-nm$(exeext) gcc-ranlib$(exeext) ifeq ($(enable_plugin),yes) @@ -4561,7 +4558,7 @@ install-common: native lang.install-comm else true; \ fi; \ done - for file in $(EXTRA_PASSES) $(EXTRA_PROGRAMS) $(COLLECT2) ..; do \ + for file in $(EXTRA_PROGRAMS) $(COLLECT2) ..; do \ if [ x$$file != x.. ]; then \ rm -f $(DESTDIR)$(libexecsubdir)/$$file; \ $(INSTALL_PROGRAM) $$file $(DESTDIR)$(libexecsubdir)/$$file; \ diff --git a/gcc/config.gcc b/gcc/config.gcc --- a/gcc/config.gcc +++ b/gcc/config.gcc @@ -129,10 +129,8 @@ # such a system header; otherwise none, do not # provide such a header at all. # -# extra_passes List of extra executables compiled for this target -# machine, used for compiling from source to object. -# -# extra_programs Like extra_passes, but these are used when linking. +# extra_programs List of extra executables compiled for this target +# machine, used when linking. # # extra_options List of target-dependent .opt files. # @@ -205,7 +203,6 @@ user_headers_inc_next_pre= user_headers_inc_next_post= use_gcc_tgmath=yes use_gcc_stdint=none -extra_passes= extra_programs= extra_objs= extra_gcc_objs= diff --git a/gcc/configure.ac b/gcc/configure.ac --- a/gcc/configure.ac +++ b/gcc/configure.ac @@ -4974,7 +4974,6 @@ AC_SUBST(user_headers_inc_next_pre) AC_SUBST(user_headers_inc_next_post) AC_SUBST(extra_headers_list) AC_SUBST(extra_objs) -AC_SUBST(extra_passes) AC_SUBST(extra_programs) AC_SUBST(float_h_file) AC_SUBST(gcc_config_arguments) -- - Rainer Orth, Center for Biotechnology, Bielefeld University
Re: Remove obsolete Tru64 UNIX V5.1B support
On 03/12/2012 05:06 PM, Rainer Orth wrote: I think the remaining changes are either obvious or covered by osf or testsuite maintainerships, so I've applied the patch. Thanks. Rainer Thanks, but please remove all the spurious 'libstdc++v3' from the ChangeLog entry and make sure it's wrapped to 80 columns. Thanks, Paolo.
Re: [C++ Patch] for c++/52465
Salut Dodji,
2012/3/12 Dodji Seketeli do...@seketeli.org:
[...]
Index: gcc/cp/decl.c
===
--- gcc/cp/decl.c (revision 184891)
+++ gcc/cp/decl.c (working copy)
@@ -8686,6 +8686,9 @@ grokdeclarator (const cp_declarator *dec
type = NULL_TREE;
type_was_error_mark_node = true;
}
+
+ type = strip_using_decl (type);
+
I am a little bit curious as to why this change is necessary. It seems
to me that the test case of your patch would pass even without this
change, wouldn't it?
Yes, this testcase would pass, but an existing testcase wouldn't. I
don't remeber which one, but I think it was related to using
declarations that refer to a typedef.
struct A { typedef int T; };
stuct B : A { using B::T; };
--
Fabien
Re: Remove obsolete Tru64 UNIX V5.1B support
Paolo Carlini paolo.carl...@oracle.com writes: On 03/12/2012 05:06 PM, Rainer Orth wrote: I think the remaining changes are either obvious or covered by osf or testsuite maintainerships, so I've applied the patch. Thanks. Rainer Thanks, but please remove all the spurious 'libstdc++v3' from the ChangeLog entry and make sure it's wrapped to 80 columns. Oops, sorry, will do ASAP. Rainer -- - Rainer Orth, Center for Biotechnology, Bielefeld University
Re: Remove non-GAS non-ELF support in alpha backend
On 03/12/12 09:22, Rainer Orth wrote: There are only two issues: * In alpha.c (alpha_option_override), it's unclear if the optimize 0 test can be removed completely. Yeah, I guess leave that for now. * During testing, I ran into this error: alpha.md:5861: `*builtin_setjmp_receiver_er_1' matches `*builtin_setjmp_receiver_er_sl_1' alpha.md:5856: previous definition of `*builtin_setjmp_receiver_er_sl_1' Delete builtin_setjmp_receiver_er_1, which ought to have had !TARGET_AS_CAN_SUBTRACT_LABELS, but since patterns are matched in order, was implied by the filtering of the previous pattern. r~
Re: Remove obsolete Tru64 UNIX V5.1B support
David Daney david.da...@cavium.com writes: On 03/06/2012 05:14 AM, Rainer Orth wrote: Joseph S. Myersjos...@codesourcery.com writes: There's one particular issue: the change to java/io/File.java required my to regenerate the .class file in classpath. I've used Sun javac -target 1.5 for that and hope I got it right. I'd have expected regeneration to use GCJ built to use ECJ, though I don't know. I've never tried this. Given that the .class file lives below libjava/classpath and has to be synced with upstream Classpath anyway, I hope the Java maintainers will take care of that. This it documented (although perhaps badly) in install/configure.html You should use --enable-java-maintainer-mode, this will cause the build to use ecj and gjavah to regenerate all the generated files in the 'standard' manner. I tried that route, but failed: even with ecj1 and gjavah scripts in PATH, I get /var/gcc/gcc-4.8.0-20120309/11-gcc/./gcc/gcj -B/var/gcc/gcc-4.8.0-20120309/11-gcc/i386-pc-solaris2.11/amd64/libjava/ -B/var/gcc/gcc-4.8.0-20120309/11-gcc/i386-pc-solaris2.11/amd64/libjava/ -B/var/gcc/gcc-4.8.0-20120309/11-gcc/./gcc/ -B/usr/local/i386-pc-solaris2.11/bin/ -B/usr/local/i386-pc-solaris2.11/lib/ -isystem /usr/local/i386-pc-solaris2.11/include -isystem /usr/local/i386-pc-solaris2.11/sys-include -m64 -C -g -fsource=1.5 -ftarget=1.5 --bootclasspath='' --classpath=/vol/gcc/src/hg/trunk/solaris/libjava:/var/gcc/gcc-4.8.0-20120309/11-gcc/i386-pc-solaris2.11/amd64/libjava:/vol/gcc/src/hg/trunk/solaris/libjava/classpath:/vol/gcc/src/hg/trunk/solaris/libjava/classpath/external/w3c_dom:/vol/gcc/src/hg/trunk/solaris/libjava/classpath/external/sax:/vol/gcc/src/hg/trunk/solaris/libjava/classpath/external/relaxngDatatype:/vol/gcc/src/hg/trunk/solaris/libjava/classpath/external/jsr166:.:: -d /vol/gcc/src/hg/trunk/solaris/libjava/classpath/lib @classes no classpath specified make[6]: *** [compile-classes] Error 1 I gave up and checked in the File.class file produced with Sun javac. Rainer -- - Rainer Orth, Center for Biotechnology, Bielefeld University
Re: [PR51752] publication safety violations in loop invariant motion pass
On 03/10/12 08:14, Torvald Riegel wrote:
On Fri, 2012-03-09 at 15:48 -0600, Aldy Hernandez wrote:
Torvald is this what you were thinking of?
Yes, but with an exit in the else branch or something that can cause x
not being read after the condition. I _suppose_ that your original
example would be an allowed transformation but just because x would be
read anyway independently of flag's value; we can assume data-race
freedom, and thus we must be able to read x in a data-race-free way even
if flag is false, so flag's value actually doesn't matter.
What about modifying the example like below? In this case, if flag2 is
true, flag's value will matter and we can't move the load to x before
it. Will PRE still introduce tmp = x + 4 in such an example?
Torvald
+ __transaction_atomic {
+if (flag)
+ y = x + 4;
+else
+ // stuff
if (flag2)
return;
+z = x + 4;
+ }
Hmmm, by adding the exit, PRE introduces the read of x + 4 correctly
*after* the read of flag2, so something like this:
__transaction_atomic {
if (flag) {
tmp = x + 4;
y = tmp;
} else {
if (flag2)
return;
tmp = x + 4;
}
z = tmp;
So... by your logic, this is allowed because the read of x would
happen anyway (it is not inserted in the flag2 != 0 case).
I'm back to having no testcase, so perhaps I should drop this patch
until we can come up with a PRE testcase that actually triggers a
publication safety violation.
Richi, you mentioned partial PRE inserting code into code paths that
previously did not have reads. Do you have an example? As much as I
tried, I could not trigger a partial PRE, but that may be because I
don't understand the algorithm very well.
Thanks guys.
Aldy
Re: PATCH: Call gen_tls_global_dynamic_64_mode and en_tls_local_dynamic_base_64_mode
On Mon, Mar 12, 2012 at 12:46 AM, Uros Bizjak ubiz...@gmail.com wrote:
On Mon, Mar 12, 2012 at 12:26 AM, H.J. Lu hongjiu...@intel.com wrote:
Pmode may be DImode for x32. This patch calls
gen_tls_global_dynamic_64_mode and gen_tls_local_dynamic_base_64_mode,
depending on Pmode. Tested on Linux/x86-64. OK for trunk?
2012-03-11 H.J. Lu hongjiu...@intel.com
* config/i386/i386.c (legitimize_tls_address): Call
gen_tls_global_dynamic_64_mode and
gen_tls_local_dynamic_base_64_mode, depending on Pmode.
* config/i386/i386.md (*tls_global_dynamic_64): Renamed to ...
(*tls_global_dynamic_64_mode): This.
(tls_global_dynamic_64): Renamed to ...
(tls_global_dynamic_64_mode): This.
(*tls_local_dynamic_base_64): Renamed to ...
(*tls_local_dynamic_base_64mode): This.
(tls_local_dynamic_base_64): Renamed to ...
(tls_local_dynamic_base_64mode): This.
diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index d673101..fc8fa12 100644
--- a/gcc/config/i386/i386.c
+++ b/gcc/config/i386/i386.c
@@ -12533,7 +12543,12 @@ legitimize_tls_address (rtx x, enum tls_model
model, bool for_mov)
rtx rax = gen_rtx_REG (Pmode, AX_REG), insns;
start_sequence ();
- emit_call_insn (gen_tls_global_dynamic_64 (rax, x, caddr));
+ if (Pmode == DImode)
+ emit_call_insn (gen_tls_global_dynamic_64_di (rax, x,
+ caddr));
+ else
+ emit_call_insn (gen_tls_global_dynamic_64_si (rax, x,
+ caddr));
insns = get_insns ();
end_sequence ();
@@ -12581,7 +12596,12 @@ legitimize_tls_address (rtx x, enum tls_model
model, bool for_mov)
rtx rax = gen_rtx_REG (Pmode, AX_REG), insns, eqv;
start_sequence ();
- emit_call_insn (gen_tls_local_dynamic_base_64 (rax, caddr));
+ if (Pmode == DImode)
+ emit_call_insn (gen_tls_local_dynamic_base_64_di (rax,
+ caddr));
+ else
+ emit_call_insn (gen_tls_local_dynamic_base_64_si (rax,
+ caddr));
insns = get_insns ();
end_sequence ();
Please use indirect functions here, see many examples in the code.
OK with this change.
Thanks,
Uros.
This is the patch I checked in.
Thanks.
--
H.J.
---
2012-03-12 H.J. Lu hongjiu...@intel.com
* config/i386/i386.c (ix86_gen_tls_global_dynamic_64): New.
(ix86_gen_tls_local_dynamic_base_64): Likewise.
(ix86_option_override_internal): Set ix86_gen_tls_global_dynamic_64
and ix86_gen_tls_local_dynamic_base_64.
(legitimize_tls_address): Use ix86_gen_tls_global_dynamic_64 and
ix86_gen_tls_local_dynamic_base_64.
* config/i386/i386.md (*tls_global_dynamic_64): Renamed to ...
(*tls_global_dynamic_64_mode): This.
(tls_global_dynamic_64): Renamed to ...
(tls_global_dynamic_64_mode): This.
(*tls_local_dynamic_base_64): Renamed to ...
(*tls_local_dynamic_base_64_mode): This.
(tls_local_dynamic_base_64): Renamed to ...
(tls_local_dynamic_base_64_mode): This.
2012-03-12 H.J. Lu hongjiu...@intel.com
* config/i386/i386.c (ix86_gen_tls_global_dynamic_64): New.
(ix86_gen_tls_local_dynamic_base_64): Likewise.
(ix86_option_override_internal): Set ix86_gen_tls_global_dynamic_64
and ix86_gen_tls_local_dynamic_base_64.
(legitimize_tls_address): Use ix86_gen_tls_global_dynamic_64 and
ix86_gen_tls_local_dynamic_base_64.
* config/i386/i386.md (*tls_global_dynamic_64): Renamed to ...
(*tls_global_dynamic_64_mode): This.
(tls_global_dynamic_64): Renamed to ...
(tls_global_dynamic_64_mode): This.
(*tls_local_dynamic_base_64): Renamed to ...
(*tls_local_dynamic_base_64_mode): This.
(tls_local_dynamic_base_64): Renamed to ...
(tls_local_dynamic_base_64_mode): This.
diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index d85dc25..dbcf5a9 100644
--- a/gcc/config/i386/i386.c
+++ b/gcc/config/i386/i386.c
@@ -2444,6 +2444,8 @@ static rtx (*ix86_gen_andsp) (rtx, rtx, rtx);
static rtx (*ix86_gen_allocate_stack_worker) (rtx, rtx);
static rtx (*ix86_gen_adjust_stack_and_probe) (rtx, rtx, rtx);
static rtx (*ix86_gen_probe_stack_range) (rtx, rtx, rtx);
+static rtx (*ix86_gen_tls_global_dynamic_64) (rtx, rtx, rtx);
+static rtx (*ix86_gen_tls_local_dynamic_base_64) (rtx, rtx);
/* Preferred alignment for stack boundary in bits. */
unsigned int ix86_preferred_stack_boundary;
@@ -3754,9 +3756,19 @@ ix86_option_override_internal (bool main_args_p)
{
ix86_gen_leave =
Re: Remove non-GAS non-ELF support in alpha backend
On Mon, 12 Mar 2012, Rainer Orth wrote: While having the osf removal patch reviewed, it turned out that with the last non-gas non-ELF alpha port gone, there's considerable cleanup potential in the alpha backend. The following patch implements this: I wonder whether it will be possible for a future cleanup to make alpha targets use toplevel elfos.h, and for Alpha GNU/Linux toplevel gnu-user.h and linux.h as well? -- Joseph S. Myers jos...@codesourcery.com
Remove DWARF_OFFSET_SIZE, DWARF_INITIAL_LENGTH_SIZE
With the IRIX port gone (shortly), there are no targets that define a
non-default DWARF_OFFSET_SIZE or DWARF_INITIAL_LENGTH_SIZE any longer.
I therefore propose to remove the code handling this.
The following (pretty mechanical) patch does this. Bootstrapped without
regressions on i386-pc-solaris2.11, ok for mainline?
Rainer
2012-03-11 Rainer Orth r...@cebitec.uni-bielefeld.de
* defaults.h (DWARF_OFFSET_SIZE): Remove.
* config/sparc/freebsd.h (DWARF_OFFSET_SIZE): Remove comment.
* config/sparc/linux64.h : Remove comment.
* config/sparc/linux64.h : Likewise.
* system.h (DWARF_OFFSET_SIZE): Poison.
* dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE): Remove.
(DWARF_CIE_ID): Remove non-default DWARF_OFFSET_SIZE,
DWARF_INITIAL_LENGTH_SIZE handling.
(output_fde): Likewise.
(output_call_frame_info): Likewise.
(DWARF_REF_SIZE): Likewise.
(DWARF_COMPILE_UNIT_HEADER_SIZE): Likewise.
(DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Likeiwise.
(DWARF_PUBNAMES_HEADER_SIZE): Likewise.
(DWARF_ARANGES_HEADER_SIZE): Likewise.
(DWARF_ARANGES_PAD_SIZE): Likewise.
(AT_string_form): Likewise.
(size_of_die): Likewise.
(size_of_pubnames): Likewise.
(value_format): Likewise.
(output_die): Likewise.
(output_compilation_unit_header): Likewise.
(output_comdat_type_unit): Likewise.
(output_pubnames): Likewise.
(output_aranges): Likewise.
(output_line_info): Likewise.
(output_macinfo_op): Likewise.
(optimize_macinfo_range): Likewise.
(output_macinfo): Likewise.
(DW_FORM_data): Remove.
(DW_FORM_ref): Remove.
(value_format): Replace DW_FORM_data by DW_FORM_data4.
Replace DW_FORM_ref by DW_FORM_ref4.
# HG changeset patch
# Parent 15a86d35ec84fea1404f4ca7d5f98f602c7dd70d
Remove DWARF_OFFSET_SIZE, DWARF_INITIAL_LENGTH_SIZE
diff --git a/gcc/config/sparc/freebsd.h b/gcc/config/sparc/freebsd.h
--- a/gcc/config/sparc/freebsd.h
+++ b/gcc/config/sparc/freebsd.h
@@ -1,5 +1,5 @@
/* Definitions for Sun SPARC64 running FreeBSD using the ELF format
- Copyright (C) 2001, 2002, 2004, 2005, 2006, 2007, 2010, 2011
+ Copyright (C) 2001, 2002, 2004, 2005, 2006, 2007, 2010, 2011, 2012
Free Software Foundation, Inc.
Contributed by David E. O'Brien obr...@freebsd.org and BSDi.
@@ -124,15 +124,6 @@ along with GCC; see the file COPYING3.
#undef DBX_CONTIN_CHAR
#define DBX_CONTIN_CHAR '?'
-/* DWARF bits. */
-
-/* Follow Irix 6 and not the Dwarf2 draft in using 64-bit offsets.
- Obviously the Dwarf2 folks havn't tried to actually build systems
- with their spec. On a 64-bit system, only 64-bit relocs become
- RELATIVE relocations. */
-
-/* #define DWARF_OFFSET_SIZE PTR_SIZE */
-
#ifdef HAVE_AS_TLS
#undef TARGET_SUN_TLS
#undef TARGET_GNU_TLS
diff --git a/gcc/config/sparc/linux64.h b/gcc/config/sparc/linux64.h
--- a/gcc/config/sparc/linux64.h
+++ b/gcc/config/sparc/linux64.h
@@ -1,6 +1,6 @@
/* Definitions for 64-bit SPARC running Linux-based GNU systems with ELF.
Copyright 1996, 1997, 1998, 2000, 2002, 2003, 2004, 2005, 2006, 2007, 2008,
- 2009, 2010, 2011 Free Software Foundation, Inc.
+ 2009, 2010, 2011, 2012 Free Software Foundation, Inc.
Contributed by David S. Miller (da...@caip.rutgers.edu)
This file is part of GCC.
@@ -236,15 +236,6 @@ do { \
#undef LOCAL_LABEL_PREFIX
#define LOCAL_LABEL_PREFIX .
-/* DWARF bits. */
-
-/* Follow Irix 6 and not the Dwarf2 draft in using 64-bit offsets.
- Obviously the Dwarf2 folks haven't tried to actually build systems
- with their spec. On a 64-bit system, only 64-bit relocs become
- RELATIVE relocations. */
-
-/* #define DWARF_OFFSET_SIZE PTR_SIZE */
-
#undef DITF_CONVERSION_LIBFUNCS
#define DITF_CONVERSION_LIBFUNCS 1
diff --git a/gcc/defaults.h b/gcc/defaults.h
--- a/gcc/defaults.h
+++ b/gcc/defaults.h
@@ -462,14 +462,6 @@ see the files COPYING3 and COPYING.RUNTI
#define DWARF2_ADDR_SIZE (POINTER_SIZE / BITS_PER_UNIT)
#endif
-/* The size in bytes of a DWARF field indicating an offset or length
- relative to a debug info section, specified to be 4 bytes in the
- DWARF-2 specification. The SGI/MIPS ABI defines it to be the same
- as PTR_SIZE. */
-#ifndef DWARF_OFFSET_SIZE
-#define DWARF_OFFSET_SIZE 4
-#endif
-
/* The size in bytes of a DWARF 4 type signature. */
#ifndef DWARF_TYPE_SIGNATURE_SIZE
#define DWARF_TYPE_SIGNATURE_SIZE 8
diff --git a/gcc/dwarf2out.c b/gcc/dwarf2out.c
--- a/gcc/dwarf2out.c
+++ b/gcc/dwarf2out.c
@@ -162,18 +162,6 @@ static GTY(()) section *debug_frame_sect
/* Maximum size (in bytes) of an artificially generated label. */
#define MAX_ARTIFICIAL_LABEL_BYTES 30
-/* According to the (draft) DWARF 3 specification, the initial length
- should either be 4 or 12 bytes. When it's 12 bytes, the first 4
- bytes are 0x, followed by the
[Patch,AVR]: Fix PR52488 (insane stack usage)
...and now to the right mailing list...
This patch cuts down insane stack offsets for 8-bit SP targets to a value the
involved insns can handle.
Background is that newlib generates such insane offsets (2050 bytes of stack
for a device with 128 bytes of RAM) because of extension PR51345, i.e. new
multilib variants for 8-bit SP targets.
The CFA notes still contain the original size as requested by the source so
that -Wstack-usage= will still work as intended.
The generated code will never be functional, of course, but there should never
be an ICE.
This patch is less aggressive than the patch proposed in the PR, i.e. it is
silent and does not issue an error so that newlib can still be built. If the
user wants additional sanity checking, he has to use -Wstack-usage= or similar
options.
Passed testsuite.
Ok for trunk?
Johann
PR target/52488
* config/avr/avr.c (avr_prologue_setup_frame): Cut down stack
offset (size) to a value the insns can deal with.
(expand_epilogue): Ditto.
Index: config/avr/avr.c
===
--- config/avr/avr.c (revision 185100)
+++ config/avr/avr.c (working copy)
@@ -827,7 +827,11 @@ avr_prologue_setup_frame (HOST_WIDE_INT
bool isr_p = cfun-machine-is_interrupt || cfun-machine-is_signal;
int live_seq = sequent_regs_live ();
+ HOST_WIDE_INT size_max
+= (HOST_WIDE_INT) GET_MODE_MASK (AVR_HAVE_8BIT_SP ? QImode : Pmode);
+
bool minimize = (TARGET_CALL_PROLOGUES
+size size_max
live_seq
!isr_p
!cfun-machine-is_OS_task
@@ -933,6 +937,7 @@ avr_prologue_setup_frame (HOST_WIDE_INT
leaf function and thus X has already been saved. */
int irq_state = -1;
+ HOST_WIDE_INT size_cfa = size;
rtx fp_plus_insns, fp, my_fp;
gcc_assert (frame_pointer_needed
@@ -951,6 +956,27 @@ avr_prologue_setup_frame (HOST_WIDE_INT
my_fp = all_regs_rtx[FRAME_POINTER_REGNUM];
}
+ /* Cut down size and avoid size = 0 so that we don't run
+ into ICE like PR52488 in the remainder. */
+
+ if (size size_max)
+{
+ /* Don't error so that insane code from newlib still compiles
+ and does not break building newlib. As PR51345 is implemented
+ now, there are multilib variants with -mtiny-stack.
+
+ If user wants sanity checks he can use -Wstack-usage=
+ or similar options.
+
+ For CFA we emit the original, non-saturated size so that
+ the generic machinery is aware of the real stack usage and
+ will print the above diagnostic as expected. */
+
+ size = size_max;
+}
+
+ size = trunc_int_for_mode (size, GET_MODE (my_fp));
+
/ Method 1: Adjust frame pointer /
start_sequence ();
@@ -975,7 +1001,7 @@ avr_prologue_setup_frame (HOST_WIDE_INT
RTX_FRAME_RELATED_P (insn) = 1;
add_reg_note (insn, REG_CFA_ADJUST_CFA,
gen_rtx_SET (VOIDmode, fp,
- plus_constant (fp, -size)));
+ plus_constant (fp, -size_cfa)));
}
/* Copy to stack pointer. Note that since we've already
@@ -1003,7 +1029,7 @@ avr_prologue_setup_frame (HOST_WIDE_INT
add_reg_note (insn, REG_CFA_ADJUST_CFA,
gen_rtx_SET (VOIDmode, stack_pointer_rtx,
plus_constant (stack_pointer_rtx,
--size)));
+-size_cfa)));
}
fp_plus_insns = get_insns ();
@@ -1026,7 +1052,7 @@ avr_prologue_setup_frame (HOST_WIDE_INT
add_reg_note (insn, REG_CFA_ADJUST_CFA,
gen_rtx_SET (VOIDmode, stack_pointer_rtx,
plus_constant (stack_pointer_rtx,
--size)));
+-size_cfa)));
if (frame_pointer_needed)
{
insn = emit_move_insn (fp, stack_pointer_rtx);
@@ -1048,7 +1074,7 @@ avr_prologue_setup_frame (HOST_WIDE_INT
emit_insn (fp_plus_insns);
}
- cfun-machine-stack_usage += size;
+ cfun-machine-stack_usage += size_cfa;
} /* !minimize size != 0 */
} /* !minimize */
}
@@ -1261,6 +1287,7 @@ expand_epilogue (bool sibcall_p)
int irq_state = -1;
rtx fp, my_fp;
rtx fp_plus_insns;
+ HOST_WIDE_INT size_max;
Fix sgi typos
While working on the IRIX 6.5 removal patch, I came across a couple of typos. Fixed as follows. Will install as obvious, I only wonder what to do about the typo in ACATS? Rainer 2012-03-10 Rainer Orth r...@cebitec.uni-bielefeld.de libgcc: * config/arm/crtn.S: Fix typo. gcc/testsuite: * ada/acats/tests/cd/cdd1001.a (CDD1001): Fix typo. gcc: * config/microblaze/microblaze.md: Fix typo. * tree-if-conv.c: Likewise. * tree-vect-patterns.c: Likewise. # HG changeset patch # Parent e0cba5670633af032245b95c9efe3254a8cf9eed Fix sgi typos diff --git a/gcc/config/microblaze/microblaze.md b/gcc/config/microblaze/microblaze.md --- a/gcc/config/microblaze/microblaze.md +++ b/gcc/config/microblaze/microblaze.md @@ -1,5 +1,5 @@ ;; microblaze.md -- Machine description for Xilinx MicroBlaze processors. -;; Copyright 2009, 2010 Free Software Foundation, Inc. +;; Copyright 2009, 2010, 2012 Free Software Foundation, Inc. ;; Contributed by Michael Eager ea...@eagercon.com. @@ -1024,7 +1024,7 @@ } ) -;; Added for status resgisters +;; Added for status registers (define_insn movsi_status [(set (match_operand:SI 0 register_operand =d,d,z) (match_operand:SI 1 register_operand z,d,d))] diff --git a/gcc/testsuite/ada/acats/tests/cd/cdd1001.a b/gcc/testsuite/ada/acats/tests/cd/cdd1001.a --- a/gcc/testsuite/ada/acats/tests/cd/cdd1001.a +++ b/gcc/testsuite/ada/acats/tests/cd/cdd1001.a @@ -84,7 +84,7 @@ begin for I in B'Range loop if B (I).all /= Stream_Element (Ident_Int (Integer (I)) * Ident_Int (3)) then - Failed (Unable to build access values desginating elements + Failed (Unable to build access values designating elements of a Stream_Element_Array); end if; end loop; diff --git a/gcc/tree-if-conv.c b/gcc/tree-if-conv.c --- a/gcc/tree-if-conv.c +++ b/gcc/tree-if-conv.c @@ -1,5 +1,5 @@ /* If-conversion for vectorizer. - Copyright (C) 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011 + Copyright (C) 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012 Free Software Foundation, Inc. Contributed by Devang Patel dpa...@apple.com @@ -712,7 +712,7 @@ if_convertible_gimple_assign_stmt_p (gim /* Return true when STMT is if-convertible. A statement is if-convertible if: - - it is an if-convertible GIMPLE_ASSGIN, + - it is an if-convertible GIMPLE_ASSIGN, - it is a GIMPLE_LABEL or a GIMPLE_COND. */ static bool diff --git a/gcc/tree-vect-patterns.c b/gcc/tree-vect-patterns.c --- a/gcc/tree-vect-patterns.c +++ b/gcc/tree-vect-patterns.c @@ -462,7 +462,7 @@ vect_handle_widen_op_by_const (gimple st where type 'TYPE' is at least double the size of type 'type'. - Also detect unsgigned cases: + Also detect unsigned cases: unsigned type a_t, b_t; unsigned TYPE u_prod_T; diff --git a/libgcc/config/arm/crtn.S b/libgcc/config/arm/crtn.S --- a/libgcc/config/arm/crtn.S +++ b/libgcc/config/arm/crtn.S @@ -1,4 +1,4 @@ -# Copyright (C) 2001, 2004, 2008, 2009, 2010, 2011 +# Copyright (C) 2001, 2004, 2008, 2009, 2010, 2011, 2012 # Free Software Foundation, Inc. # Written By Nick Clifton # @@ -47,7 +47,7 @@ # # Note - we do not try any fancy optimizations of the return # sequences here, it is just not worth it. Instead keep things - # simple. Restore all the save resgisters, including the link + # simple. Restore all the save registers, including the link # register and then perform the correct function return instruction. # We also save/restore r3 to ensure stack alignment. .macro FUNC_END -- - Rainer Orth, Center for Biotechnology, Bielefeld University
Re: Remove obsolete IRIX 6.5 support
On Mon, Mar 12, 2012 at 10:03 AM, Rainer Orth r...@cebitec.uni-bielefeld.de wrote: I'm happy with the stuff going away! :)
[RFA/ARM]: Correct Neon testsuite generation
All, The commit to fix PR51534 did not update the testsuite (as no changes were expected there). Unfortunately, this means that I didn't notice that the Neon testsuite generator is broken. The attached patch fixes the generator. Checked by re-running the Neon testsuite and arm_neon.h generators and ensuring no changes in the generated testsuite/header. OK for trunk? OK for backporting to GCC 4.7? Thanks, Matt gcc/ChangeLog: 2012-03-12 Matthew Gretton-Dann matthew.gretton-d...@arm.com * config/arm/neon.ml (ops): Fixup expected instructions for unsigned vector compares. -- Matthew Gretton-Dann Principal Engineer, PD Software, ARM Ltd.diff --git a/gcc/config/arm/neon.ml b/gcc/config/arm/neon.ml index 363e55c..85eb5ec 100644 --- a/gcc/config/arm/neon.ml +++ b/gcc/config/arm/neon.ml @@ -780,14 +780,19 @@ let ops = (* Comparison, greater-than or equal. *) Vcge, [], All (3, Dreg), vcge, cmp_sign_matters, F32 :: s_8_32; -Vcge, [Builtin_name vcgeu], All (3, Dreg), vcge, cmp_sign_matters, u_8_32; +Vcge, [Instruction_name [vcge]; Builtin_name vcgeu], + All (3, Dreg), vcge, cmp_sign_matters, + u_8_32; Vcge, [], All (3, Qreg), vcgeQ, cmp_sign_matters, F32 :: s_8_32; -Vcge, [Builtin_name vcgeu], All (3, Qreg), vcgeQ, cmp_sign_matters, u_8_32; +Vcge, [Instruction_name [vcge]; Builtin_name vcgeu], + All (3, Qreg), vcgeQ, cmp_sign_matters, + u_8_32; (* Comparison, less-than or equal. *) Vcle, [Flipped vcge], All (3, Dreg), vcle, cmp_sign_matters, F32 :: s_8_32; -Vcle, [Flipped vcgeu], All (3, Dreg), vcle, cmp_sign_matters, +Vcle, [Instruction_name [vcge]; Flipped vcgeu], + All (3, Dreg), vcle, cmp_sign_matters, u_8_32; Vcle, [Instruction_name [vcge]; Flipped vcgeQ], All (3, Qreg), vcleQ, cmp_sign_matters, @@ -798,14 +803,19 @@ let ops = (* Comparison, greater-than. *) Vcgt, [], All (3, Dreg), vcgt, cmp_sign_matters, F32 :: s_8_32; -Vcgt, [Builtin_name vcgtu], All (3, Dreg), vcgt, cmp_sign_matters, u_8_32; +Vcgt, [Instruction_name [vcgt]; Builtin_name vcgtu], + All (3, Dreg), vcgt, cmp_sign_matters, + u_8_32; Vcgt, [], All (3, Qreg), vcgtQ, cmp_sign_matters, F32 :: s_8_32; -Vcgt, [Builtin_name vcgtu], All (3, Qreg), vcgtQ, cmp_sign_matters, u_8_32; +Vcgt, [Instruction_name [vcgt]; Builtin_name vcgtu], + All (3, Qreg), vcgtQ, cmp_sign_matters, + u_8_32; (* Comparison, less-than. *) Vclt, [Flipped vcgt], All (3, Dreg), vclt, cmp_sign_matters, F32 :: s_8_32; -Vclt, [Flipped vcgtu], All (3, Dreg), vclt, cmp_sign_matters, +Vclt, [Instruction_name [vcgt]; Flipped vcgtu], + All (3, Dreg), vclt, cmp_sign_matters, u_8_32; Vclt, [Instruction_name [vcgt]; Flipped vcgtQ], All (3, Qreg), vcltQ, cmp_sign_matters,
RFC: PATCH: Add -maddress-mode=short|long for x86
Hi,
Trunk can now use SImode or DImode for Pmode with -mx32. Pmode == SImode
will avoid many conversions from SImode to DImode:
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50797
It will use 0x67 address size prefix instead. This patch adds a new
command line option, -maddress-mode=short|long, for x86. Any comments?
Thanks.
H.J.
2012-03-12 H.J. Lu hongjiu...@intel.com
* config/i386/i386-opts.h (pmode): New.
* config/i386/i386.c (ix86_option_override_internal): Properly
check and set ix86_pmode.
* config/i386/i386.h (Pmode): Check ix86_pmode instead of
TARGET_64BIT.
* config/i386/i386.opt (maddress-mode=): New.
* doc/invoke.texi: Document -maddress-mode=short|long for x86.
diff --git a/gcc/config/i386/i386-opts.h b/gcc/config/i386/i386-opts.h
index 3cc2253..eea85fd 100644
--- a/gcc/config/i386/i386-opts.h
+++ b/gcc/config/i386/i386-opts.h
@@ -71,6 +71,11 @@ enum cmodel {
CM_LARGE_PIC /* No assumptions. */
};
+enum pmode {
+ PMODE_SI,/* Pmode == SImode. */
+ PMODE_DI /* Pmode == DImode. */
+};
+
enum asm_dialect {
ASM_ATT,
ASM_INTEL
diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index dbcf5a9..8d9c801 100644
--- a/gcc/config/i386/i386.c
+++ b/gcc/config/i386/i386.c
@@ -3172,6 +3172,17 @@ ix86_option_override_internal (bool main_args_p)
else
ix86_arch_specified = 1;
+ if (global_options_set.x_ix86_pmode)
+{
+ if ((TARGET_LP64 ix86_pmode == PMODE_SI)
+ || (!TARGET_64BIT ix86_pmode == PMODE_DI))
+ error (address mode %qs not supported in the %s bit mode,
+ TARGET_64BIT ? short : long,
+ TARGET_64BIT ? 64 : 32);
+}
+ else
+ix86_pmode = TARGET_LP64 ? PMODE_DI : PMODE_SI;
+
if (!global_options_set.x_ix86_abi)
ix86_abi = DEFAULT_ABI;
diff --git a/gcc/config/i386/i386.h b/gcc/config/i386/i386.h
index c77070b..3fcd209 100644
--- a/gcc/config/i386/i386.h
+++ b/gcc/config/i386/i386.h
@@ -1745,7 +1745,7 @@ do {
\
/* Specify the machine mode that pointers have.
After generation of rtl, the compiler makes no further distinction
between pointers and any other objects of this machine mode. */
-#define Pmode (TARGET_64BIT ? DImode : SImode)
+#define Pmode (ix86_pmode == PMODE_DI ? DImode : SImode)
/* A C expression whose value is zero if pointers that need to be extended
from being `POINTER_SIZE' bits wide to `Pmode' are sign-extended and
diff --git a/gcc/config/i386/i386.opt b/gcc/config/i386/i386.opt
index 1d8dc44..29f1082 100644
--- a/gcc/config/i386/i386.opt
+++ b/gcc/config/i386/i386.opt
@@ -159,6 +159,20 @@ Enum(cmodel) String(32) Value(CM_32)
EnumValue
Enum(cmodel) String(kernel) Value(CM_KERNEL)
+maddress-mode=
+Target RejectNegative Joined Enum(pmode) Var(ix86_pmode) Init(PMODE_SI)
+Use given address mode
+
+Enum
+Name(pmode) Type(enum pmode)
+Known address mode (for use with the -maddress-mode= option):
+
+EnumValue
+Enum(pmode) String(short) Value(PMODE_SI)
+
+EnumValue
+Enum(pmode) String(long) Value(PMODE_DI)
+
mcpu=
Target RejectNegative Joined Undocumented Alias(mtune=) Warn(%-mcpu=% is
deprecated; use %-mtune=% or %-march=% instead)
diff --git a/gcc/doc/invoke.texi b/gcc/doc/invoke.texi
index 5f5397b..e9188db 100644
--- a/gcc/doc/invoke.texi
+++ b/gcc/doc/invoke.texi
@@ -634,7 +634,7 @@ Objective-C and Objective-C++ Dialects}.
-mveclibabi=@var{type} -mvect8-ret-in-mem @gol
-mpc32 -mpc64 -mpc80 -mstackrealign @gol
-momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
--mcmodel=@var{code-model} -mabi=@var{name} @gol
+-mcmodel=@var{code-model} -mabi=@var{name} -maddress-mode=@var{mode} @gol
-m32 -m64 -mx32 -mlarge-data-threshold=@var{num} @gol
-msse2avx -mfentry -m8bit-idiv @gol
-mavx256-split-unaligned-load -mavx256-split-unaligned-store}
@@ -13938,6 +13938,18 @@ be statically or dynamically linked.
@opindex mcmodel=large
Generate code for the large model. This model makes no assumptions
about addresses and sizes of sections.
+
+@item -maddress-mode=long
+@opindex maddress-mode=long
+Generate code for long address mode. It is only supported for 64-bit
+and x32 environments. This is the default address mode for 64-bit
+environment.
+
+@item -maddress-mode=short
+@opindex maddress-mode=short
+Generate code for short address mode. It is only supported for 32-bit
+and x32 environments. This is the default address mode for 32-bit and
+x32 environments.
@end table
@node i386 and x86-64 Windows Options
Re: [Fortran-dev, patch, committed] Minor fixes
On 03/12/2012 01:50 PM, Tobias Burnus wrote:
Additionally, I modified some tree-dump patterns for the extent and
for the base_addr change.
Three more changes: The libgfortran change should have been just a
cleanup, but it seems to also fix a failure. Plus two dump changes. The
number of failures is now down to 160 FAIL lines - or to the following
18 test cases (plus three which fail also on the trunk):
gfortran.dg/c_f_pointer_complex.f03
gfortran.dg/c_f_pointer_shape_tests_2.f03
gfortran.dg/c_f_pointer_shape_tests_4.f03
gfortran.dg/c_f_pointer_tests.f90
gfortran.dg/c_f_pointer_tests_4.f90
gfortran.dg/optional_dim_3.f90
gfortran.dg/associated_2.f90
gfortran.dg/auto_char_dummy_array_1.f90
gfortran.dg/class_array_1.f03
gfortran.dg/auto_char_len_3.f90
gfortran.dg/class_array_1.f03
gfortran.dg/class_array_2.f03
gfortran.dg/class_array_3.f03
gfortran.dg/class_to_type_1.f03
gfortran.dg/proc_decl_23.f90
gfortran.dg/read_eof_all.f90
gfortran.dg/transfer_intrinsic_3.f90
gfortran.dg/subref_array_pointer_2.f90
The array c_f_pointer function fails because the fptr's size is not set
(it is needed to set the sm). Solution: Handle everything in the FE.
Tobias
Index: libgfortran/ChangeLog.fortran-dev
===
--- libgfortran/ChangeLog.fortran-dev (revision 185216)
+++ libgfortran/ChangeLog.fortran-dev (working copy)
@@ -1,5 +1,9 @@
2012-03-12 Tobias Burnus bur...@net-b.de
+ * intrinsics/spread_generic.c (spread_internal): Properly use sm.
+
+2012-03-12 Tobias Burnus bur...@net-b.de
+
* intrinsics/iso_c_binding.c (c_f_pointer_u0): Use
GFC_DESCRIPTOR_SM instead of GFC_DESCRIPTOR_STRIDE_BYTES.
Index: libgfortran/intrinsics/spread_generic.c
===
--- libgfortran/intrinsics/spread_generic.c (revision 185199)
+++ libgfortran/intrinsics/spread_generic.c (working copy)
@@ -71,18 +71,18 @@ spread_internal (gfc_array_char *ret, const gfc_ar
/* The front end has signalled that we need to populate the
return array descriptor. */
- size_t ub, stride;
+ size_t ext, sm;
ret-dtype = (source-dtype ~GFC_DTYPE_RANK_MASK) | rrank;
dim = 0;
- rs = 1;
+ rs = size;
for (n = 0; n rrank; n++)
{
- stride = rs;
+ sm = rs;
if (n == *along - 1)
{
- ub = ncopies - 1;
- rdelta = rs * size;
+ ext = ncopies;
+ rdelta = rs;
rs *= ncopies;
}
else
@@ -90,17 +90,17 @@ spread_internal (gfc_array_char *ret, const gfc_ar
count[dim] = 0;
extent[dim] = GFC_DESCRIPTOR_EXTENT(source,dim);
sstride[dim] = GFC_DESCRIPTOR_SM(source,dim);
- rstride[dim] = rs * size;
+ rstride[dim] = rs;
- ub = extent[dim]-1;
+ ext = extent[dim];
rs *= extent[dim];
dim++;
}
- GFC_DIMENSION_SET (ret-dim[n], 0, ub, stride*size);
+ GFC_DIMENSION_SET (ret-dim[n], 0, ext, sm);
}
ret-offset = 0;
- ret-base_addr = internal_malloc_size (rs * size);
+ ret-base_addr = internal_malloc_size (rs);
if (rs = 0)
return;
Index: gcc/testsuite/gfortran.dg/internal_pack_4.f90
===
--- gcc/testsuite/gfortran.dg/internal_pack_4.f90 (revision 185197)
+++ gcc/testsuite/gfortran.dg/internal_pack_4.f90 (working copy)
@@ -26,7 +26,7 @@ USE M1
CALL S2()
END
-! { dg-final { scan-tree-dump-times a != 0B \\? \\\(.*\\\) _gfortran_internal_pack 1 original } }
-! { dg-final { scan-tree-dump-times if \\(a != 0B 1 original } }
+! { dg-final { scan-tree-dump-times a != 0B \\(real\\(kind=.\\).0:. . restrict\\) a-base_addr != 0B \\? \\\(.*\\\) _gfortran_internal_pack 1 original } }
+! { dg-final { scan-tree-dump-times if \\(\\(a != 0B \\(real\\(kind=.\\).0:. . restrict\\) a-base_addr != 0B\\) 1 original } }
! { dg-final { cleanup-tree-dump original } }
! { dg-final { cleanup-modules m1 } }
Index: gcc/testsuite/gfortran.dg/assign_10.f90
===
--- gcc/testsuite/gfortran.dg/assign_10.f90 (revision 185197)
+++ gcc/testsuite/gfortran.dg/assign_10.f90 (working copy)
@@ -23,6 +23,6 @@ end
! cases will all yield a temporary, so that atmp appears 18 times.
! Note that it is the kind conversion that generates the temp.
!
-! { dg-final { scan-tree-dump-times parm 18 original } }
-! { dg-final { scan-tree-dump-times atmp 18 original } }
+! { dg-final { scan-tree-dump-times parm 20 original } }
+! { dg-final { scan-tree-dump-times atmp 20 original } }
! { dg-final { cleanup-tree-dump original } }
Index: gcc/testsuite/ChangeLog.fortran-dev
===
--- gcc/testsuite/ChangeLog.fortran-dev (revision 185225)
+++ gcc/testsuite/ChangeLog.fortran-dev (working copy)
@@ -1,5 +1,10 @@
2012-03-12 Tobias Burnus bur...@net-b.de
+ * gfortran.dg/assign_10.f90: Update scan-tree-dump-times.
[Patch,AVR.testsuite,committed]: More test for address spaces (PR49868)
http://gcc.gnu.org/viewcvs?view=revisionrevision=185255 This adds some test cases for the address space support. Some tests fail because of PR other/52545. Moreover, you will have to update the avrtest simulator because the test case for __flash1 needs its own linker script in order to locate .progmem1.data* at 0x1. Johann
Re: [Patch ARM/ configury] Add fall-back check for gnu_unique_object
Looks OK to me.
[PATCH][Cilkplus] Elemental Function Mangling
Hello Everyone,
This patch is for the Cilkplus branch mainly affecting the C compiler.
This patch will do the vector function mangling correctly for elemental
functions.
Thanking You,
Yours Sincerely,
Balaji V. Iyer.diff --git a/gcc/ChangeLog.cilk b/gcc/ChangeLog.cilk
index 1398870..c1b1d71 100644
--- a/gcc/ChangeLog.cilk
+++ b/gcc/ChangeLog.cilk
@@ -1,3 +1,20 @@
+2012-03-11 Balaji V. Iyer balaji.v.i...@intel.com
+
+ * attribs.c (decl_attributes): Concatinated existing attributes with
+ vector attributes.
+ * c-decl.c (bind): Added a check if scope is not null.
+ * elem-function.c (rename_elm_fn): New function.
+ (is_elem_fn): Likewise.
+ (find_processor_code): Likewise.
+ (find_vlength_code): Likewise.
+ (create_processor_attribute): Likewise.
+ (create_optimize_attribute): Likewise.
+ (find_suffix): Likewise.
+ (create_elem_fn_nodes): Likewise.
+ (extract_elem_fn_values): Likewise.
+ (create_elem_vec_fn): Likewise.
+ * passes.c (init_optimization_passes): Added elemental function pass.
+
2012-03-09 Balaji V. Iyer balaji.v.i...@intel.com
* attribs.c (decl_attributes): Added a check for elemental function
diff --git a/gcc/Makefile.in b/gcc/Makefile.in
index 0da06b3..ada4090 100644
--- a/gcc/Makefile.in
+++ b/gcc/Makefile.in
@@ -1453,6 +1453,7 @@ OBJS = \
cilk.o \
cilk-low.o \
array-notation-common.o \
+ elem-function.o \
$(out_object_file) \
$(EXTRA_OBJS) \
$(host_hook_obj)
@@ -3436,7 +3437,8 @@ lower-subreg.o : lower-subreg.c $(CONFIG_H) $(SYSTEM_H)
coretypes.h \
insn-config.h $(BASIC_BLOCK_H) $(RECOG_H) $(OBSTACK_H) $(BITMAP_H) \
$(EXPR_H) $(EXCEPT_H) $(REGS_H) $(TREE_PASS_H) $(DF_H) dce.h
-
+elem-function.o: elem-function.c $(CONFIG_H) $(SYSTEM_H) $(TREE_H) $(GIMPLE_H)
\
+ $(OPTABS_H) $(RECOG_H)
array-notation-common.o: array-notation-common.c $(CONFIG_H) $(SYSTEM_H) \
$(TREE_H) $(RTL_H) $(OPTABS_H) $(GIMPLE_H) $(RECOG_H)
diff --git a/gcc/attribs.c b/gcc/attribs.c
index 2ececc4..13c1417 100644
--- a/gcc/attribs.c
+++ b/gcc/attribs.c
@@ -327,6 +327,12 @@ decl_attributes (tree *node, tree attributes, int flags)
if (!is_elem_fn_attribute_p (name))
warning (OPT_Wattributes, %qE attribute directive ignored,
name);
+ else
+ {
+ returned_attrs = tree_cons (name, args, returned_attrs);
+ DECL_ATTRIBUTES (*anode) = tree_cons (name, args,
+ DECL_ATTRIBUTES (*anode));
+ }
continue;
}
else if (list_length (args) spec-min_length
diff --git a/gcc/c-decl.c b/gcc/c-decl.c
index 4abf738..6251bee 100644
--- a/gcc/c-decl.c
+++ b/gcc/c-decl.c
@@ -620,7 +620,8 @@ bind (tree name, tree decl, struct c_scope *scope, bool
invisible,
b-shadowed = 0;
b-decl = decl;
b-id = name;
- b-depth = scope-depth;
+ if (scope)
+b-depth = scope-depth;
b-invisible = invisible;
b-nested = nested;
b-inner_comp = 0;
@@ -629,8 +630,11 @@ bind (tree name, tree decl, struct c_scope *scope, bool
invisible,
b-u.type = NULL;
- b-prev = scope-bindings;
- scope-bindings = b;
+ if (scope)
+{
+ b-prev = scope-bindings;
+ scope-bindings = b;
+}
if (decl_jump_unsafe (decl))
scope-has_jump_unsafe_decl = 1;
@@ -658,9 +662,11 @@ bind (tree name, tree decl, struct c_scope *scope, bool
invisible,
/* Locate the appropriate place in the chain of shadowed decls
to insert this binding. Normally, scope == current_scope and
this does nothing. */
- while (*here (*here)-depth scope-depth)
-here = (*here)-shadowed;
-
+ if (scope)
+{
+ while (*here (*here)-depth scope-depth)
+ here = (*here)-shadowed;
+}
b-shadowed = *here;
*here = b;
}
diff --git a/gcc/c-family/ChangeLog.cilk b/gcc/c-family/ChangeLog.cilk
index a4049d0..08d4c2d 100644
--- a/gcc/c-family/ChangeLog.cilk
+++ b/gcc/c-family/ChangeLog.cilk
@@ -1,3 +1,7 @@
+2012-03-11 Balaji V. Iyer balaji.v.i...@intel.com
+
+ * c-common.c (handle_vector_attribute): New function.
+
2012-01-20 Balaji V. Iyer balaji.v.i...@intel.com
* c-common.c (c_define_builtins): Added a call to
diff --git a/gcc/c-family/c-common.c b/gcc/c-family/c-common.c
index f84ccb9..47b5c54 100644
--- a/gcc/c-family/c-common.c
+++ b/gcc/c-family/c-common.c
@@ -381,6 +381,7 @@ static tree handle_type_generic_attribute (tree *, tree,
tree, int, bool *);
static tree handle_alloc_size_attribute (tree *, tree, tree, int, bool *);
static tree handle_target_attribute (tree *, tree, tree, int, bool *);
static tree handle_optimize_attribute (tree *, tree, tree, int, bool *);
+static tree handle_vector_attribute (tree *, tree, tree, int, bool *);
static tree ignore_attribute (tree *, tree, tree, int, bool *);
static tree handle_no_split_stack_attribute (tree *, tree,
Re: Swap SECTION_EXCLUDE and SECTION_MACH_DEP
Already submitted by gjl in rev. 185259. -Sri. On Mon, Mar 12, 2012 at 10:27 AM, Richard Henderson r...@redhat.com wrote: On 03/12/12 10:26, Sriraman Tallam wrote: Hi Richard, Is this ok for trunk? Asking you because you approved the patch the first time around. Thanks, -Sri. On Fri, Mar 9, 2012 at 3:10 PM, Sriraman Tallam tmsri...@google.com wrote: Please see: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52545 SECTION_EXCLUDE should not take up bits reserved for SECTION_MACH_DEP Patch: PR 52545 * output.h (SECTION_MACH_DEP) (SECTION_EXCLUDE): Swap values. Yes, this patch is ok r~
[PATCH,testsuite,committed] Skip some tests for powerpc*-*-darwin*
Following in response to a report on PR46728. Committed as obvious.
2012-03-12 Bill Schmidt wschm...@linux.vnet.ibm.com
PR tree-optimization/46728
* gcc.target/powerpc/pr46728-4.c: Skip for powerpc*-*-darwin*.
* gcc.target/powerpc/pr46728-5.c: Likewise.
* gcc.target/powerpc/pr46728-8.c: Likewise.
* gcc.target/powerpc/pr46728-10.c: Likewise.
* gcc.target/powerpc/pr46728-11.c: Likewise.
* gcc.target/powerpc/pr46728-13.c: Likewise.
* gcc.target/powerpc/pr46728-14.c: Likewise.
* gcc.target/powerpc/pr46728-15.c: Likewise.
Index: gcc/testsuite/gcc.target/powerpc/pr46728-13.c
===
--- gcc/testsuite/gcc.target/powerpc/pr46728-13.c (revision 185247)
+++ gcc/testsuite/gcc.target/powerpc/pr46728-13.c (working copy)
@@ -1,4 +1,5 @@
-/* { dg-do run } */
+/* { dg-do run { target { powerpc*-*-* } } } */
+/* { dg-skip-if { powerpc*-*-darwin* } { * } { } } */
/* { dg-options -O2 -ffast-math -fno-inline -fno-unroll-loops -lm
-mpowerpc-gpopt } */
#include math.h
Index: gcc/testsuite/gcc.target/powerpc/pr46728-14.c
===
--- gcc/testsuite/gcc.target/powerpc/pr46728-14.c (revision 185247)
+++ gcc/testsuite/gcc.target/powerpc/pr46728-14.c (working copy)
@@ -1,4 +1,5 @@
-/* { dg-do run } */
+/* { dg-do run { target { powerpc*-*-* } } } */
+/* { dg-skip-if { powerpc*-*-darwin* } { * } { } } */
/* { dg-options -O2 -ffast-math -fno-inline -fno-unroll-loops -lm
-mpowerpc-gpopt } */
#include math.h
Index: gcc/testsuite/gcc.target/powerpc/pr46728-4.c
===
--- gcc/testsuite/gcc.target/powerpc/pr46728-4.c(revision 185247)
+++ gcc/testsuite/gcc.target/powerpc/pr46728-4.c(working copy)
@@ -1,4 +1,5 @@
-/* { dg-do compile } */
+/* { dg-do compile { target { powerpc*-*-* } } } */
+/* { dg-skip-if { powerpc*-*-darwin* } { * } { } } */
/* { dg-options -O2 -ffast-math -fno-inline -fno-unroll-loops -lm
-mpowerpc-gpopt } */
#include math.h
Index: gcc/testsuite/gcc.target/powerpc/pr46728-15.c
===
--- gcc/testsuite/gcc.target/powerpc/pr46728-15.c (revision 185247)
+++ gcc/testsuite/gcc.target/powerpc/pr46728-15.c (working copy)
@@ -1,4 +1,5 @@
-/* { dg-do run } */
+/* { dg-do run { target { powerpc*-*-* } } } */
+/* { dg-skip-if { powerpc*-*-darwin* } { * } { } } */
/* { dg-options -O2 -ffast-math -fno-inline -fno-unroll-loops -lm
-mpowerpc-gpopt } */
#include math.h
Index: gcc/testsuite/gcc.target/powerpc/pr46728-5.c
===
--- gcc/testsuite/gcc.target/powerpc/pr46728-5.c(revision 185247)
+++ gcc/testsuite/gcc.target/powerpc/pr46728-5.c(working copy)
@@ -1,4 +1,5 @@
-/* { dg-do compile } */
+/* { dg-do compile { target { powerpc*-*-* } } } */
+/* { dg-skip-if { powerpc*-*-darwin* } { * } { } } */
/* { dg-options -O2 -ffast-math -fno-inline -fno-unroll-loops -lm
-mpowerpc-gpopt } */
#include math.h
Index: gcc/testsuite/gcc.target/powerpc/pr46728-10.c
===
--- gcc/testsuite/gcc.target/powerpc/pr46728-10.c (revision 185247)
+++ gcc/testsuite/gcc.target/powerpc/pr46728-10.c (working copy)
@@ -1,4 +1,5 @@
-/* { dg-do run } */
+/* { dg-do run { target { powerpc*-*-* } } } */
+/* { dg-skip-if { powerpc*-*-darwin* } { * } { } } */
/* { dg-options -O2 -ffast-math -fno-inline -fno-unroll-loops -lm
-mpowerpc-gpopt } */
#include math.h
Index: gcc/testsuite/gcc.target/powerpc/pr46728-8.c
===
--- gcc/testsuite/gcc.target/powerpc/pr46728-8.c(revision 185247)
+++ gcc/testsuite/gcc.target/powerpc/pr46728-8.c(working copy)
@@ -1,4 +1,5 @@
-/* { dg-do compile } */
+/* { dg-do compile { target { powerpc*-*-* } } } */
+/* { dg-skip-if { powerpc*-*-darwin* } { * } { } } */
/* { dg-options -O2 -ffast-math -fno-inline -fno-unroll-loops -lm
-mpowerpc-gpopt } */
#include math.h
Index: gcc/testsuite/gcc.target/powerpc/pr46728-11.c
===
--- gcc/testsuite/gcc.target/powerpc/pr46728-11.c (revision 185247)
+++ gcc/testsuite/gcc.target/powerpc/pr46728-11.c (working copy)
@@ -1,4 +1,5 @@
-/* { dg-do run } */
+/* { dg-do run { target { powerpc*-*-* } } } */
+/* { dg-skip-if { powerpc*-*-darwin* } { * } { } } */
/* { dg-options -O2 -ffast-math -fno-inline -fno-unroll-loops -lm
-mpowerpc-gpopt } */
#include math.h
Re: Remove obsolete Solaris 8 support
Rainer Orth r...@cebitec.uni-bielefeld.de writes: libgo: * configure.ac (OSCFLAGS): Remove *-*-solaris2.8 handling. (libgo_cv_lib_makecontext_stack_top): Remove sparc*-*-solaris2.8* handling. * configure: Regenerate. As with the Irix 5 changes, just send the libgo patches to me rather than committing them directly to the repository. Thanks. Ian
[C++ Patch] PR 52299, bogus div by zero warning
Hi,
I handled this issue as outlined by Jakub in the audit trail. For the
purpose of the bogus div by zero warning just using
c_inhibit_evaluation_warnings appears to work fine.
Tested x86_64-linux.
Thanks,
Paolo.
///
/cp
2012-03-12 Paolo Carlini paolo.carl...@oracle.com
PR c++/52299
* pt.c (tsubst_copy_and_build, case COND_EXPR): Avoid bogus
division by zero warnings.
/testsuite
2012-03-12 Paolo Carlini paolo.carl...@oracle.com
PR c++/52299
* g++.dg/warn/Wdiv-by-zero-bogus.C: New.
Index: testsuite/g++.dg/warn/Wdiv-by-zero-bogus.C
===
--- testsuite/g++.dg/warn/Wdiv-by-zero-bogus.C (revision 0)
+++ testsuite/g++.dg/warn/Wdiv-by-zero-bogus.C (revision 0)
@@ -0,0 +1,30 @@
+// PR c++/52299
+
+templateunsigned x
+struct test0 {
+ static const unsigned a_
+ = x ? 10 / x : 10;
+};
+
+templateunsigned x
+struct test1 {
+ static const unsigned a_
+ = !x ? 10 : 10 / x;
+};
+
+templatebool x
+struct test2 {
+ static const unsigned a_
+ = x ? 10 / x : 10;
+};
+
+templatebool x
+struct test3 {
+ static const unsigned a_
+ = !x ? 10 : 10 / x;
+};
+
+unsigned i0 = test00::a_;
+unsigned i1 = test10::a_;
+unsigned i2 = test2false::a_;
+unsigned i3 = test3false::a_;
Index: cp/pt.c
===
--- cp/pt.c (revision 185247)
+++ cp/pt.c (working copy)
@@ -13943,12 +13943,36 @@ tsubst_copy_and_build (tree t,
}
case COND_EXPR:
- return build_x_conditional_expr
- (RECUR (TREE_OPERAND (t, 0)),
-RECUR (TREE_OPERAND (t, 1)),
-RECUR (TREE_OPERAND (t, 2)),
- complain);
+ {
+ tree cond = RECUR (TREE_OPERAND (t, 0));
+ tree op1, op2;
+ if (TREE_CODE (cond) == INTEGER_CST)
+ {
+ if (integer_zerop (cond))
+ {
+ ++c_inhibit_evaluation_warnings;
+ op1 = RECUR (TREE_OPERAND (t, 1));
+ --c_inhibit_evaluation_warnings;
+ op2 = RECUR (TREE_OPERAND (t, 2));
+ }
+ else
+ {
+ op1 = RECUR (TREE_OPERAND (t, 1));
+ ++c_inhibit_evaluation_warnings;
+ op2 = RECUR (TREE_OPERAND (t, 2));
+ --c_inhibit_evaluation_warnings;
+ }
+ }
+ else
+ {
+ op1 = RECUR (TREE_OPERAND (t, 1));
+ op2 = RECUR (TREE_OPERAND (t, 2));
+ }
+
+ return build_x_conditional_expr (cond, op1, op2, complain);
+ }
+
case PSEUDO_DTOR_EXPR:
return finish_pseudo_destructor_expr
(RECUR (TREE_OPERAND (t, 0)),
Re: [PATCH,testsuite,committed] Skip some tests for powerpc*-*-darwin*
William J. Schmidt wschm...@linux.vnet.ibm.com writes:
Index: gcc/testsuite/gcc.target/powerpc/pr46728-13.c
===
--- gcc/testsuite/gcc.target/powerpc/pr46728-13.c (revision 185247)
+++ gcc/testsuite/gcc.target/powerpc/pr46728-13.c (working copy)
@@ -1,4 +1,5 @@
-/* { dg-do run } */
+/* { dg-do run { target { powerpc*-*-* } } } */
Why is the target necessary? gcc.target/powerpc is only run on
powerpc*-*-* or rs6000-*-*-* anyway (cf. powerpc.exp).
+/* { dg-skip-if { powerpc*-*-darwin* } { * } { } } */
Please omit defaults to dg-skip-if: { * } { } are unnecessary and
make the directive harder to read.
Thanks.
Rainer
--
-
Rainer Orth, Center for Biotechnology, Bielefeld University
Re: [C++ Patch] PR 52299, bogus div by zero warning
OK. Jason
Re: Remove obsolete Solaris 8 support
On Mar 12, 2012, at 10:44 AM, Rainer Orth wrote: Since even extended support for Solaris 8 ends by March 31st, this patch removes Solaris 8 support from mainline. One of the nice things about gcc is that gcc usually still works, long after a vendor has abandoned a machine. I rather like that gcc will just work, unlike vendor software, which often says, please buy a new machine. One doesn't have to remove support in gcc for something, just because a vendor doesn't support it. That said, truly crufty things, should go.
Re: Remove obsolete IRIX 6.5 support
OK from my point of view, except... Rainer Orth r...@cebitec.uni-bielefeld.de writes: * config/mips/mips.h (TARGET_GPWORD): Remove IRIX 6 N64 handling. (TARGET_IRIX6): Remove. (TARGET_CPU_CPP_BUILTINS): Remove IRIX 6 handling. Don't define LANGUAGE_C, _LANGUAGE_C for Objective-C. ...looking at SVN history, defining _LANGUAGE_C* for Objective-C predates the Irix comment, so I think we'd better keep it. How about changing the comment to: /* Bizarre, but retained for backwards compatibility. */ Thanks for all your work on this port, and for putting up with those long bootstrap times :-) Richard
Re: [PATCH,testsuite,committed] Skip some tests for powerpc*-*-darwin*
Rainer Orth r...@cebitec.uni-bielefeld.de writes:
+/* { dg-skip-if { powerpc*-*-darwin* } { * } { } } */
Please omit defaults to dg-skip-if: { * } { } are unnecessary and
make the directive harder to read.
Oh, I forgot: please indicate why you are skipping the test in the
comment field, like:
/* { dg-skip-if PR tree-optimization/46728 { powerpc*-*-darwin* } } */
Thanks.
Rainer
--
-
Rainer Orth, Center for Biotechnology, Bielefeld University
Re: PATCH: Call gen_tls_global_dynamic_64_mode and en_tls_local_dynamic_base_64_mode
On Mon, Mar 12, 2012 at 9:50 AM, H.J. Lu hjl.to...@gmail.com wrote: On Mon, Mar 12, 2012 at 12:46 AM, Uros Bizjak ubiz...@gmail.com wrote: On Mon, Mar 12, 2012 at 12:26 AM, H.J. Lu hongjiu...@intel.com wrote: Pmode may be DImode for x32. This patch calls gen_tls_global_dynamic_64_mode and gen_tls_local_dynamic_base_64_mode, depending on Pmode. Tested on Linux/x86-64. OK for trunk? 2012-03-11 H.J. Lu hongjiu...@intel.com * config/i386/i386.c (legitimize_tls_address): Call gen_tls_global_dynamic_64_mode and gen_tls_local_dynamic_base_64_mode, depending on Pmode. * config/i386/i386.md (*tls_global_dynamic_64): Renamed to ... (*tls_global_dynamic_64_mode): This. (tls_global_dynamic_64): Renamed to ... (tls_global_dynamic_64_mode): This. (*tls_local_dynamic_base_64): Renamed to ... (*tls_local_dynamic_base_64mode): This. (tls_local_dynamic_base_64): Renamed to ... (tls_local_dynamic_base_64mode): This. diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c index d673101..fc8fa12 100644 --- a/gcc/config/i386/i386.c +++ b/gcc/config/i386/i386.c @@ -12533,7 +12543,12 @@ legitimize_tls_address (rtx x, enum tls_model model, bool for_mov) rtx rax = gen_rtx_REG (Pmode, AX_REG), insns; start_sequence (); - emit_call_insn (gen_tls_global_dynamic_64 (rax, x, caddr)); + if (Pmode == DImode) + emit_call_insn (gen_tls_global_dynamic_64_di (rax, x, + caddr)); + else + emit_call_insn (gen_tls_global_dynamic_64_si (rax, x, + caddr)); insns = get_insns (); end_sequence (); @@ -12581,7 +12596,12 @@ legitimize_tls_address (rtx x, enum tls_model model, bool for_mov) rtx rax = gen_rtx_REG (Pmode, AX_REG), insns, eqv; start_sequence (); - emit_call_insn (gen_tls_local_dynamic_base_64 (rax, caddr)); + if (Pmode == DImode) + emit_call_insn (gen_tls_local_dynamic_base_64_di (rax, + caddr)); + else + emit_call_insn (gen_tls_local_dynamic_base_64_si (rax, + caddr)); insns = get_insns (); end_sequence (); Please use indirect functions here, see many examples in the code. OK with this change. Thanks, Uros. This is the patch I checked in. Thanks. -- H.J. --- 2012-03-12 H.J. Lu hongjiu...@intel.com * config/i386/i386.c (ix86_gen_tls_global_dynamic_64): New. (ix86_gen_tls_local_dynamic_base_64): Likewise. (ix86_option_override_internal): Set ix86_gen_tls_global_dynamic_64 and ix86_gen_tls_local_dynamic_base_64. (legitimize_tls_address): Use ix86_gen_tls_global_dynamic_64 and ix86_gen_tls_local_dynamic_base_64. * config/i386/i386.md (*tls_global_dynamic_64): Renamed to ... (*tls_global_dynamic_64_mode): This. (tls_global_dynamic_64): Renamed to ... (tls_global_dynamic_64_mode): This. (*tls_local_dynamic_base_64): Renamed to ... (*tls_local_dynamic_base_64_mode): This. (tls_local_dynamic_base_64): Renamed to ... (tls_local_dynamic_base_64_mode): This. This patch caused x32 libgcc build failure: ../../../../src-trunk/libgcc/generic-morestack-thread.c: In function 'stack_split_initialize_thread': ../../../../src-trunk/libgcc/generic-morestack-thread.c:128:1: error: unrecognizable insn: (call_insn/u 8 7 9 3 (parallel [ (set (reg:DI 0 ax) (call:DI (mem:QI (symbol_ref:DI (__tls_get_addr)) [0 S1 A8]) (const_int 0 [0]))) (unspec:DI [ (symbol_ref:SI (__morestack_segments) [flags 0x50] var_decl 0x77725780 __morestack_segments) ] UNSPEC_TLS_GD) ]) ../../../../src-trunk/libgcc/generic-morestack-thread.c:117 -1 (expr_list:REG_EH_REGION (const_int -2147483648 [0x8000]) (nil)) (nil)) ../../../../src-trunk/libgcc/generic-morestack-thread.c:128:1: internal compiler error: in extract_insn, at recog.c:2123 Please submit a full bug report, with preprocessed source if appropriate. See http://gcc.gnu.org/bugs.html for instructions. make[8]: *** [generic-morestack-thread.o] Error 1 make[8]: *** Waiting for unfinished jobs __morestack_segments is always in SImode for x32. Adding :P to (unspec:P [(match_operand 1 tls_symbolic_operand )] causes it to fail when Pmode == DImode I checked in this patch to revert the part of that change. Sorry for that. -- H.J. --- Index: ChangeLog === --- ChangeLog (revision
Re: Remove DWARF_OFFSET_SIZE, DWARF_INITIAL_LENGTH_SIZE
Jason Merrill ja...@redhat.com writes: On 03/12/2012 01:41 PM, Joseph S. Myers wrote: As a *target macro* it makes sense to remove it - reomve it from defaults.h, make it purely internal to dwarf2out.c. But I think it makes sense to have it inside dwarf2out.c Agreed. Fine with me. I've just been following the lead of several other places in dwarf2out.c where hardcoded numbers were used. What about DWARF_INITIAL_LENGTH_SIZE? Keep it as a macro inside dwarf2out.c, too? Rainer -- - Rainer Orth, Center for Biotechnology, Bielefeld University
Re: Remove obsolete Solaris 8 support
One of the nice things about gcc is that gcc usually still works, long after a vendor has abandoned a machine. I rather like that gcc will just work, unlike vendor software, which often says, please buy a new machine. One doesn't have to remove support in gcc for something, just because a vendor doesn't support it. That said, truly crufty things, should go. I agree. If it's not causing a serious maintenance issue, I see no reason to remove it.
Re: Remove obsolete Solaris 8 support
Mike Stump mikest...@comcast.net writes: On Mar 12, 2012, at 10:44 AM, Rainer Orth wrote: Since even extended support for Solaris 8 ends by March 31st, this patch removes Solaris 8 support from mainline. One of the nice things about gcc is that gcc usually still works, long after a vendor has abandoned a machine. I rather like that gcc will just work, unlike vendor software, which often says, please buy a new machine. One doesn't have to remove support in gcc for something, just because a vendor doesn't support it. That said, truly crufty things, should go. No need to tell me, that's why I've been keeping the IRIX and OSF ports going for so long, sort of like an OS archaeologist :-) But both for those and Solaris 8, the time has come where maintaining them is more trouble than it's worth, as can be seen by the contortions I had to go through to even find a machine still capable of running Solaris 8/x86. Rainer -- - Rainer Orth, Center for Biotechnology, Bielefeld University
Re: [Ping][PATCH, libstdc++-v3] Enable to cross-test libstdc++ on simulator
On Mar 10, 2012, at 3:25 PM, Jonathan Wakely wrote: On 7 March 2012 05:22, Terry Guo wrote: Hello, Can anybody please review and approve the following simple patch? Thanks very much. http://gcc.gnu.org/ml/libstdc++/2011-08/msg00063.html Ok. Ok for the release branches as relevant, if they are open for changes, after testing. If you backport it, let it brew for at least a week on trunk before any backports. Thanks.
Re: Remove obsolete Solaris 8 support
ken...@vlsi1.ultra.nyu.edu (Richard Kenner) writes: One of the nice things about gcc is that gcc usually still works, long after a vendor has abandoned a machine. I rather like that gcc will just work, unlike vendor software, which often says, please buy a new machine. One doesn't have to remove support in gcc for something, just because a vendor doesn't support it. That said, truly crufty things, should go. I agree. If it's not causing a serious maintenance issue, I see no reason to remove it. It does, unfortunately: Solaris 8 is even worse wrt. to C99 and POSIX.1 conformance than Solaris 9, especially in the areas of TLS and thread support, as I've found the hard way during the GCC 4.7 cycle. There are all sorts of quirks, and the OS is now 12 years old. I see better ways to spend my (volunteer) time. Rainer -- - Rainer Orth, Center for Biotechnology, Bielefeld University
Re: Remove obsolete Solaris 8 support
On Mon, Mar 12, 2012 at 7:37 PM, Mike Stump mikest...@comcast.net wrote: On Mar 12, 2012, at 10:44 AM, Rainer Orth wrote: Since even extended support for Solaris 8 ends by March 31st, this patch removes Solaris 8 support from mainline. One of the nice things about gcc is that gcc usually still works, long after a vendor has abandoned a machine. I rather like that gcc will just work, unlike vendor software, which often says, please buy a new machine. One doesn't have to remove support in gcc for something, just because a vendor doesn't support it. That said, truly crufty things, should go. Personally, I'd be in favor of removing any target for which no test results have been posted for the last, say, 2 GCC release series. Otherwise there isn't any measure for the quality for such a target. There've been GCC release series where a complete port failed to build... Ciao! Steven
Re: Remove DWARF_OFFSET_SIZE, DWARF_INITIAL_LENGTH_SIZE
On 03/12/2012 02:45 PM, Rainer Orth wrote: What about DWARF_INITIAL_LENGTH_SIZE? Keep it as a macro inside dwarf2out.c, too? Please. Jason
[google/4.6] Add xfails for arm-grtev2-linux-gnueabi (issue5794063)
Hi Diego This patch adds arm-grtev2-linux-gnueabi.xfail for our 4.6 branch so that we can track regressions. This just established the test baseline. The failures need to be investigated. -Doug 2012-03-12 Doug Kwan dougk...@google.com * contrib/testsuite-management/arm-grtev2-linux-gnueabi.xfail: New file. Index: contrib/testsuite-management/arm-grtev2-linux-gnueabi.xfail === --- contrib/testsuite-management/arm-grtev2-linux-gnueabi.xfail (revision 0) +++ contrib/testsuite-management/arm-grtev2-linux-gnueabi.xfail (revision 0) @@ -0,0 +1,74 @@ +# Failures in ./gcc/testsuite/gcc/gcc.sum: +# *** gcc: +FAIL: gcc.dg/automversn_1.c (test for excess errors) +UNRESOLVED: gcc.dg/automversn_1.c compilation failed to produce executable +UNRESOLVED: gcc.dg/automversn_1.c scan-tree-dump auto_clone foo.autoclone.original +UNRESOLVED: gcc.dg/automversn_1.c scan-tree-dump auto_clone foo.autoclone.0 +FAIL: gcc.dg/builtin-apply2.c execution test +FAIL: gcc.dg/tls/pr42894.c (test for excess errors) +FAIL: gcc.dg/torture/stackalign/builtin-apply-2.c -O0 execution test +FAIL: gcc.dg/torture/stackalign/builtin-apply-2.c -O0 execution test +FAIL: gcc.dg/torture/stackalign/builtin-apply-2.c -O1 execution test +FAIL: gcc.dg/torture/stackalign/builtin-apply-2.c -Os execution test +FAIL: gcc.dg/torture/tls/tls-test.c -O0 execution test +FAIL: gcc.dg/torture/tls/tls-test.c -O1 execution test +FAIL: gcc.dg/torture/tls/tls-test.c -O2 execution test +FAIL: gcc.dg/torture/tls/tls-test.c -O3 -fomit-frame-pointer execution test +FAIL: gcc.dg/torture/tls/tls-test.c -O3 -g execution test +FAIL: gcc.dg/torture/tls/tls-test.c -Os execution test +FAIL: gcc.dg/torture/tls/tls-test.c -O2 -flto -flto-partition=none execution test +FAIL: gcc.dg/torture/tls/tls-test.c -O2 -flto execution test +FAIL: gcc.dg/tree-prof/switch-case-1.c scan-rtl-dump-times expand Succ edge[^\n]*count:2000 1 +FAIL: gcc.dg/vect/vect-multitypes-11.c scan-tree-dump-times vect vectorized 1 loops 1 +FAIL: gcc.dg/vect/vect-multitypes-12.c scan-tree-dump-times vect vectorized 1 loops 1 +FAIL: gcc.dg/vect/vect-reduc-dot-s16b.c scan-tree-dump-times vect vectorized 1 loops 0 +FAIL: gcc.dg/vect/vect-reduc-pattern-1a.c scan-tree-dump-times vect vectorized 1 loops 0 +FAIL: gcc.dg/vect/vect-reduc-pattern-1b.c scan-tree-dump-times vect vectorized 1 loops 0 +FAIL: gcc.dg/vect/vect-reduc-pattern-1c.c scan-tree-dump-times vect vectorized 1 loops 0 +FAIL: gcc.dg/vect/vect-reduc-pattern-2a.c scan-tree-dump-times vect vectorized 1 loops 0 +FAIL: gcc.dg/vect/vect-reduc-pattern-2b.c scan-tree-dump-times vect vectorized 1 loops 0 +XPASS: gcc.dg/vect/slp-reduc-3.c scan-tree-dump-times vect vectorizing stmts using SLP 1 +FAIL: gcc.dg/vect/wrapv-vect-reduc-pattern-2c.c scan-tree-dump-times vect vectorized 1 loops 0 +XPASS: gcc.dg/vect/no-scevccp-outer-16.c scan-tree-dump-times vect OUTER LOOP VECTORIZED. 1 +XPASS: gcc.dg/vect/no-scevccp-outer-17.c scan-tree-dump-times vect OUTER LOOP VECTORIZED. 1 +XPASS: gcc.dg/vect/no-scevccp-outer-19.c scan-tree-dump-times vect OUTER LOOP VECTORIZED. 1 +XPASS: gcc.dg/vect/no-scevccp-outer-21.c scan-tree-dump-times vect OUTER LOOP VECTORIZED. 1 +FAIL: gcc.target/arm/pr42575.c scan-assembler-not mov +FAIL: gcc.target/arm/thumb-ltu.c (test for excess errors) +UNRESOLVED: gcc.target/arm/thumb-ltu.c scan-assembler-not uxtb + +# Failures in ./gcc/testsuite/gfortran/gfortran.sum: +# *** gfortran: +FAIL: gfortran.dg/pr45636.f90 -O scan-tree-dump-times forwprop2 memset 0 +FAIL: gfortran.dg/vect/fast-math-pr38968.f90 execution test +FAIL: gfortran.dg/vect/fast-math-pr38968.f90 scan-tree-dump vect vectorized 1 loops + +# Failures in ./gcc/testsuite/g++/g++.sum: +# *** g++: +FAIL: g++.dg/abi/forced.C execution test +FAIL: g++.dg/abi/local1.C execution test +FAIL: g++.dg/thread-ann/thread_annot_lock-82.C (test for warnings, line 47) +XPASS: g++.dg/uninit-pred-3_b.C (test for excess errors) +FAIL: g++.dg/tree-prof/mversn13.C execution,-g -fprofile-generate +UNRESOLVED: g++.dg/tree-prof/mversn13.C compilation, -g -fprofile-use +UNRESOLVED: g++.dg/tree-prof/mversn13.C execution,-g -fprofile-use +FAIL: g++.dg/tree-prof/mversn13.C execution,-O0 -fprofile-generate +UNRESOLVED: g++.dg/tree-prof/mversn13.C compilation, -O0 -fprofile-use +UNRESOLVED: g++.dg/tree-prof/mversn13.C execution,-O0 -fprofile-use +FAIL: g++.dg/tree-prof/mversn13.C execution,-O1 -fprofile-generate +UNRESOLVED: g++.dg/tree-prof/mversn13.C compilation, -O1 -fprofile-use +UNRESOLVED: g++.dg/tree-prof/mversn13.C execution,-O1 -fprofile-use +FAIL: g++.dg/tree-prof/mversn13.C execution,-O2 -fprofile-generate +UNRESOLVED: g++.dg/tree-prof/mversn13.C compilation, -O2 -fprofile-use +UNRESOLVED: g++.dg/tree-prof/mversn13.C execution,-O2 -fprofile-use +FAIL: g++.dg/tree-prof/mversn13.C execution,-O3
Re: Remove obsolete Solaris 8 support
On Mar 12, 2012, at 11:48 AM, Rainer Orth wrote: But both for those and Solaris 8, the time has come where maintaining them is more trouble than it's worth, The nice thing is, anyone that disagrees with you, will step forward. :-) Notice, I didn't disagree with you... I was only railing against the idea that we remove support, simply because a vendor did. In this case, we're removing support, because it is more trouble than it is worth, which makes for a fine reason to remove support.
Re: [google/4.6] Add xfails for arm-grtev2-linux-gnueabi (issue5794063)
On 12/03/12 14:58 , Doug Kwan wrote: 2012-03-12 Doug Kwandougk...@google.com * contrib/testsuite-management/arm-grtev2-linux-gnueabi.xfail: New file. OK. Diego.
Re: [PATCH,testsuite,committed] Skip some tests for
OK, will do. I just copied the style of some other test cases that
skipped darwin, assuming that was the preferred syntax.
Thanks,
Bill
On Mon, 2012-03-12 at 19:40 +0100, Rainer Orth wrote:
Rainer Orth r...@cebitec.uni-bielefeld.de writes:
+/* { dg-skip-if { powerpc*-*-darwin* } { * } { } } */
Please omit defaults to dg-skip-if: { * } { } are unnecessary and
make the directive harder to read.
Oh, I forgot: please indicate why you are skipping the test in the
comment field, like:
/* { dg-skip-if PR tree-optimization/46728 { powerpc*-*-darwin* } } */
Thanks.
Rainer
Re: [PATCH,testsuite,committed] Skip some tests for
William J. Schmidt wschm...@linux.vnet.ibm.com writes: OK, will do. I just copied the style of some other test cases that skipped darwin, assuming that was the preferred syntax. No problem, I'll probably make a pass over the whole testsuite to remove that cruft to avoid misleading others. Rainer -- - Rainer Orth, Center for Biotechnology, Bielefeld University
