Re: [patch, fortran] PR 47359 - warnings for constant conversion
Am 01.06.2015 um 22:10 schrieb Steve Kargl: >> c) Warn for >> >> print *,3.1415926535 with -Wconversion-extra >> >> and don't warn for >> >> print *,3.141592653589_4 >> > > This would be my first choice. If a user actually specifies > a suffix, I assume that the user has given some thought > to the preceding digits. >> d) Like now: Warn with -Wconversion-extra for both >> >>print *,3.1415926535 >> >>and >> >>print *,3.14159265358979_4 > > This would be my second choice. I would actually prefer d). -Wconversion-extra is about warning for things that are often correct. I would like to have a chance to warn about this kind of construct. Any other comments? OK to commit? Thomas
genmatch: guess the type of a?b:c as b instead of a
Hello,
as discussed around
https://gcc.gnu.org/ml/gcc-patches/2015-06/msg00041.html
we are currently guessing the type of a?b:c incorrectly. This does not
affect current simplifications, because the only 'cond' in output patterns
are at the outermost level, so their type is forced to 'type' and never
guessed. Indeed, the patch does not change the generated *-match.c. It
would allow removing an explicit cond:itype in a patch posted by Jeff.
I tested it on a dummy .pd file containing:
(simplify
(plus @0 (plus @1 @2))
(negate (cond @0 @1 @2)))
and the generated files differ by:
- res = fold_build3_loc (loc, COND_EXPR, TREE_TYPE (ops1[0]), ops1[0],
ops1[1], ops1[2]);
+ res = fold_build3_loc (loc, COND_EXPR, TREE_TYPE (ops1[1]), ops1[0],
ops1[1], ops1[2]);
(and something similar for gimple)
I wondered about using something like
VOID_TYPE_P (TREE_TYPE (ops1[1])) ? TREE_TYPE (ops1[2]) : TREE_TYPE (ops1[1])
but I don't think that will be necessary.
Bootstrap is currently broken on many platforms with comparison failures,
but since it went that far and generated the same *-match.c files, that
seems sufficient testing.
2015-06-08 Marc Glisse
* genmatch.c (expr::gen_transform): For conditions, guess the type
from the second operand.
--
Marc GlisseIndex: gcc/genmatch.c
===
--- gcc/genmatch.c (revision 224186)
+++ gcc/genmatch.c (working copy)
@@ -1702,20 +1702,27 @@ expr::gen_transform (FILE *f, const char
type = optype;
}
else if (is_a (operation)
&& !strcmp (as_a (operation)->tcc, "tcc_comparison"))
{
/* comparisons use boolean_type_node (or what gets in), but
their operands need to figure out the types themselves. */
sprintf (optype, "boolean_type_node");
type = optype;
}
+ else if (*operation == COND_EXPR
+ || *operation == VEC_COND_EXPR)
+{
+ /* Conditions are of the same type as their first alternative. */
+ sprintf (optype, "TREE_TYPE (ops%d[1])", depth);
+ type = optype;
+}
else
{
/* Other operations are of the same type as their first operand. */
sprintf (optype, "TREE_TYPE (ops%d[0])", depth);
type = optype;
}
if (!type)
fatal ("two conversions in a row");
fprintf (f, "{\n");
[committed] Tighten some rtx-based variable and return types
These are the main rtx-related changes found by the -Wupcast-* warnings
that I posted last week.
Bootstrapped & regression-tested on x86_64-linux-gnu. Also tested on
config-list.mk, checking that all build failures were unrelated.
Applied as obvious.
Thanks,
Richard
gcc/
* emit-rtl.c, expr.c, gcse.c, optabs.c, optabs.h, print-rtl.c,
rtl.h, bb-reorder.c, builtins.c, calls.c, cfgbuild.c, cfgexpand.c,
cfgrtl.c, cilk-common.c, config/i386/i386.md, cse.c, dwarf2cfi.c,
except.c, final.c, function.c, gcse-common.c, genemit.c,
haifa-sched.c, ifcvt.c, jump.c, loop-invariant.c, loop-iv.c,
lra-constraints.c, lra.c, reload1.c, resource.c, rtlanal.c,
sched-deps.c, sched-ebb.c, sel-sched-ir.c, sel-sched.c,
shrink-wrap.c, stmt.c, store-motion.c: Replace rtx base types with
more derived ones.
Index: gcc/emit-rtl.c
===
--- gcc/emit-rtl.c 2015-06-06 11:33:59.788667360 +0100
+++ gcc/emit-rtl.c 2015-06-06 11:33:59.756667734 +0100
@@ -3662,7 +3662,7 @@ try_split (rtx pat, rtx_insn *trial, int
int probability;
rtx_insn *insn_last, *insn;
int njumps = 0;
- rtx call_insn = NULL_RTX;
+ rtx_insn *call_insn = NULL;
/* We're not good at redistributing frame information. */
if (RTX_FRAME_RELATED_P (trial))
@@ -4684,10 +4684,10 @@ emit_pattern_after_setloc (rtx pattern,
rtx_insn *(*make_raw) (rtx))
{
rtx_insn *after = safe_as_a (uncast_after);
- rtx last = emit_pattern_after_noloc (pattern, after, NULL, make_raw);
+ rtx_insn *last = emit_pattern_after_noloc (pattern, after, NULL, make_raw);
if (pattern == NULL_RTX || !loc)
-return safe_as_a (last);
+return last;
after = NEXT_INSN (after);
while (1)
@@ -4700,7 +4700,7 @@ emit_pattern_after_setloc (rtx pattern,
break;
after = NEXT_INSN (after);
}
- return safe_as_a (last);
+ return last;
}
/* Insert PATTERN after AFTER. MAKE_RAW indicates how to turn PATTERN
Index: gcc/expr.c
===
--- gcc/expr.c 2015-06-06 11:33:59.788667360 +0100
+++ gcc/expr.c 2015-06-06 11:33:59.760667688 +0100
@@ -10559,7 +10559,7 @@ expand_expr_real_1 (tree exp, rtx target
if ((icode = optab_handler (movmisalign_optab, mode))
!= CODE_FOR_nothing)
{
- rtx reg, insn;
+ rtx reg;
op0 = adjust_address (op0, mode, 0);
/* We've already validated the memory, and we're creating a
@@ -10568,7 +10568,7 @@ expand_expr_real_1 (tree exp, rtx target
reg = gen_reg_rtx (mode);
/* Nor can the insn generator. */
- insn = GEN_FCN (icode) (reg, op0);
+ rtx_insn *insn = GEN_FCN (icode) (reg, op0);
emit_insn (insn);
return reg;
}
Index: gcc/gcse.c
===
--- gcc/gcse.c 2015-06-06 11:33:59.788667360 +0100
+++ gcc/gcse.c 2015-06-06 11:33:59.760667688 +0100
@@ -540,7 +540,6 @@ static void trim_ld_motion_mems (void);
static void update_ld_motion_stores (struct gcse_expr *);
static void clear_modify_mem_tables (void);
static void free_modify_mem_tables (void);
-static rtx gcse_emit_move_after (rtx, rtx, rtx_insn *);
static bool is_too_expensive (const char *);
#define GNEW(T)((T *) gmalloc (sizeof (T)))
@@ -2434,7 +2433,7 @@ single_set_gcse (rtx_insn *insn)
/* Emit move from SRC to DEST noting the equivalence with expression computed
in INSN. */
-static rtx
+static rtx_insn *
gcse_emit_move_after (rtx dest, rtx src, rtx_insn *insn)
{
rtx_insn *new_rtx;
@@ -3960,7 +3959,6 @@ update_ld_motion_stores (struct gcse_exp
rtx pat = PATTERN (insn);
rtx src = SET_SRC (pat);
rtx reg = expr->reaching_reg;
- rtx copy;
/* If we've already copied it, continue. */
if (expr->reaching_reg == src)
@@ -3975,7 +3973,7 @@ update_ld_motion_stores (struct gcse_exp
fprintf (dump_file, "\n");
}
- copy = gen_move_insn (reg, copy_rtx (SET_SRC (pat)));
+ rtx_insn *copy = gen_move_insn (reg, copy_rtx (SET_SRC (pat)));
emit_insn_before (copy, insn);
SET_SRC (pat) = reg;
df_insn_rescan (insn);
Index: gcc/optabs.c
===
--- gcc/optabs.c2015-06-06 11:33:59.788667360 +0100
+++ gcc/optabs.c2015-06-06 11:33:59.764667641 +0100
@@ -2048,7 +2048,7 @@ expand_binop (machine_mode mode, optab b
if (optab_handler (mov_optab, mode) != CODE_FOR_nothing
|| ! rtx_equal_p (target, xtarget))
{
- rtx temp = emit_move_insn (target, xtarget);
+ rtx_insn *temp = emit_move_i
Re: [patch, fortran] PR 47359 - warnings for constant conversion
On Sat, Jun 06, 2015 at 12:22:56PM +0200, Thomas Koenig wrote: > > Any other comments? OK to commit? > No. Yes. -- Steve
[committed] Update HP-UX builtin predefines
The trunk was changed recently for a period to build with "gcc -std=c++98".
This caused a build failure
on HP-UX because a non standard HP variant of the nftw function was selected.
The attached change
revises the hpux predefines so this won't happen. In particular, we now define
_XOPEN_SOURCE_EXTENDED
when building with c++.
I also reviewed the existing predefines and added some new ones for c++ and
hpux11.31.
Tested on hppa2.0w-hp-hpux11.11 and hppa64-hp-hpux11.11. Committed to trunk
and gcc-5 branch.
Dave
--
John David Anglin dave.ang...@bell.net
2015-06-06 John David Anglin
PR bootstrap/66319
* config/pa/pa-hpux10.h (TARGET_OS_CPP_BUILTINS): Rearrange builtin
defines. Define _LARGEFILE_SOURCE and _LARGEFILE64_SOURCE for c++.
Define _XOPEN_UNIX and _XOPEN_SOURCE_EXTENDED for c++ if unix95 or
later.
* config/pa/pa-hpux11.h (TARGET_OS_CPP_BUILTINS): Likewise.
Define _INCLUDE_STDC_SOURCE_PRE_199901, _INCLUDE_STDC_SOURCE_199901,
_INCLUDE_XOPEN_SOURCE_PRE_500, _INCLUDE_XOPEN_SOURCE_520,
_INCLUDE_XOPEN_SOURCE_PRE_600 and _INCLUDE_XOPEN_SOURCE_600 for c++
and non iso if unix2003.
Index: config/pa/pa-hpux10.h
===
--- config/pa/pa-hpux10.h (revision 224079)
+++ config/pa/pa-hpux10.h (working copy)
@@ -24,7 +24,9 @@
the definition of __cplusplus. We define _INCLUDE_LONGLONG
to prevent nlist.h from defining __STDC_32_MODE__ (no longlong
support). We define __STDCPP__ to get certain system headers
- (notably assert.h) to assume standard preprocessor behavior in C++. */
+ (notably assert.h) to assume standard preprocessor behavior in C++.
+ We define _XOPEN_SOURCE_EXTENDED when we define _HPUX_SOURCE to avoid
+ non standard hpux variants in _INCLUDE_XOPEN_SOURCE_EXTENDED. */
#undef TARGET_OS_CPP_BUILTINS
#define TARGET_OS_CPP_BUILTINS() \
do \
@@ -44,9 +46,21 @@
builtin_define ("_REENTRANT"); \
builtin_define ("_INCLUDE_LONGLONG"); \
builtin_define ("__STDCPP__"); \
+ builtin_define ("_LARGEFILE_SOURCE"); \
+ builtin_define ("_LARGEFILE64_SOURCE"); \
+ if (flag_pa_unix >= 1995) \
+ { \
+ builtin_define ("_XOPEN_UNIX"); \
+ builtin_define ("_XOPEN_SOURCE_EXTENDED"); \
+ } \
} \
- else if (!flag_iso) \
+ else if (flag_iso) \
{ \
+ if (flag_isoc94)\
+ builtin_define ("_INCLUDE__STDC_A1_SOURCE"); \
+ } \
+ else\
+ { \
builtin_define ("_HPUX_SOURCE");\
builtin_define ("_REENTRANT"); \
if (preprocessing_trad_p ())\
@@ -59,12 +73,12 @@
builtin_define ("_PWB");\
builtin_define ("PWB"); \
} \
+ if (flag_pa_unix >= 1995) \
+ { \
+ builtin_define ("_XOPEN_UNIX"); \
+ builtin_define ("_XOPEN_SOURCE_EXTENDED"); \
+ } \
} \
- if (flag_pa_unix >= 1995) \
- { \
- builtin_define ("_XOPEN_UNIX"); \
- builtin_define ("_XOPEN_SOURCE_EXTENDED"); \
- } \
if (TARGET_SIO) \
builtin_define ("_SIO"); \
else
[C++/58583] ICE instantiating NSDMIs
This patch fixes 58582, a set of ICEs that happen instantiating NSDMIs. There
are a couple of causes, both fixed.
1) instantiating the template while parsing an NSDMI of the template itself.
We see a DEFAULT_ARG in get_nsdmi. Fixed by jumping to the existing error
handling for the non-template case.
2) recursive instantiation of the NSDMI itself. As we instantiate lazily, we
end up running out of stack. Fixed by creating a temporary DEFAULT_ARG and then
detecting it on a subsequent recursion. I did contemplate having this direct to
the same error as above, but decided against it, because this really is an
instantiation problem not a parsing problem.
built & tested on x86_64-linux, ok?
nathan
2015-06-05 Nathan Sidwell
cp/
PR c++/58583
* init.c (get_nsdmi): Check for DEFAULT_ARG in template case and
protect it from recursive instantiation.
testsuite/
PR c++/58583
* g++.dg/cpp0x/nsdmi-template14.C: New test.
Index: cp/init.c
===
--- cp/init.c (revision 224152)
+++ cp/init.c (working copy)
@@ -544,6 +544,7 @@ get_nsdmi (tree member, bool in_ctor)
tree init;
tree save_ccp = current_class_ptr;
tree save_ccr = current_class_ref;
+
if (!in_ctor)
{
/* Use a PLACEHOLDER_EXPR when we don't have a 'this' parameter to
@@ -551,22 +552,41 @@ get_nsdmi (tree member, bool in_ctor)
current_class_ref = build0 (PLACEHOLDER_EXPR, DECL_CONTEXT (member));
current_class_ptr = build_address (current_class_ref);
}
+
if (DECL_LANG_SPECIFIC (member) && DECL_TEMPLATE_INFO (member))
{
- /* Do deferred instantiation of the NSDMI. */
- init = (tsubst_copy_and_build
- (DECL_INITIAL (DECL_TI_TEMPLATE (member)),
- DECL_TI_ARGS (member),
- tf_warning_or_error, member, /*function_p=*/false,
- /*integral_constant_expression_p=*/false));
+ init = DECL_INITIAL (DECL_TI_TEMPLATE (member));
+ if (TREE_CODE (init) == DEFAULT_ARG)
+ goto unparsed;
- init = digest_nsdmi_init (member, init);
+ /* Check recursive instantiation. */
+ if (TREE_CODE (DECL_INITIAL (member)) == DEFAULT_ARG)
+ {
+ error ("recursive instantiation of non-static data member "
+ "initializer for %qD", member);
+ init = error_mark_node;
+ }
+ else
+ {
+ DECL_INITIAL (member) = make_node (DEFAULT_ARG);
+
+ /* Do deferred instantiation of the NSDMI. */
+ init = (tsubst_copy_and_build
+ (init, DECL_TI_ARGS (member),
+ tf_warning_or_error, member, /*function_p=*/false,
+ /*integral_constant_expression_p=*/false));
+ init = digest_nsdmi_init (member, init);
+
+ if (TREE_CODE (DECL_INITIAL (member)) == DEFAULT_ARG)
+ DECL_INITIAL (member) = void_node;
+ }
}
else
{
init = DECL_INITIAL (member);
if (init && TREE_CODE (init) == DEFAULT_ARG)
{
+ unparsed:
error ("constructor required before non-static data member "
"for %qD has been parsed", member);
DECL_INITIAL (member) = error_mark_node;
Index: testsuite/g++.dg/cpp0x/nsdmi-template14.C
===
--- testsuite/g++.dg/cpp0x/nsdmi-template14.C (revision 0)
+++ testsuite/g++.dg/cpp0x/nsdmi-template14.C (working copy)
@@ -0,0 +1,23 @@
+// PR c++/58583
+// { dg-do compile { target c++11 } }
+
+template struct A // {dg-error "non-static data member initializer" }
+{
+ int i = (A<0>(), 0); // { dg-error "non-static data member initializer required before parsing" } { dg-error "synthesized method" }
+};
+
+template struct B
+{
+ B* p = new B;
+};
+
+B<1> x; // { dg-error "constructor required before non-static data member" }
+
+struct C
+{
+ template struct D
+ {
+D* p = new D<0>;
+ };
+};
+
[v3 PATCH] Implement N4387 and LWG 2367
This patch implements the so-called conditionally-explicit constructors for tuple and pair, and also constrains the default constructors of tuple and pair. The patch does not try to implement the part in N4387 that allows constructing tuples from shorter packs than the tuple's element count, I suggest we look at that separately. The constraints involve fair amounts of evil black magic, but they do the job. Chances are they can be simplified, but I again suggest we look at that separately. Tested on Linux-PPC64. Patch gzipped since it's 50kB otherwise. :) 2015-06-07 Ville Voutilainen Implement N4387, "improving pair and tuple" and LWG 2367, "pair and tuple are not correctly implemented for is_constructible with no args". * include/bits/stl_pair.h (_ConstructiblePair, _ImplicitlyConvertiblePair, _MoveConstructiblePair, _ImplicitlyMoveConvertiblePair): New. * include/bits/stl_pair.h (pair()): Constrain it. * include/bits/stl_pair.h (pair(const _T1&, const _T2&), pair(const pair<_U1, _U2>&), pair(_U1&&, const _T2&), pair(const _T1&, _U2&&), pair(_U1&&, _U2&&), pair(pair<_U1, _U2>&&)): Make conditionally explicit. * include/std/tuple (_TC, _TC2): New. * include/std/tuple (tuple()): Constrain it. * include/std/tuple (tuple(const _UElements&...), tuple(_UElements&&...), tuple(const tuple<_UElements...>&), tuple(tuple<_UElements...>&&), tuple(allocator_arg_t, const _Alloc&, const _UElements&...), tuple(allocator_arg_t, const _Alloc&, _UElements&&...), tuple(allocator_arg_t, const _Alloc&, const tuple<_UElements...>&), tuple(allocator_arg_t, const _Alloc&, tuple<_UElements...>&&), tuple(const pair<_U1, _U2>&), tuple(pair<_U1, _U2>&&), tuple(allocator_arg_t, const _Alloc&, const pair<_U1, _U2>&), tuple(allocator_arg_t, const _Alloc&, pair<_U1, _U2>&&)): Make conditionally explicit. * testsuite/20_util/pair/cons/explicit_construct.cc: New. * testsuite/20_util/pair/requirements/dr2367.cc: New. * testsuite/20_util/tuple/cons/explicit_construct.cc: New. * testsuite/20_util/tuple/requirements/dr2367.cc: New. explicit-tuple-pair.diff.gz Description: GNU Zip compressed data
Re: [v3 PATCH] Implement N4387 and LWG 2367
On Sun, 7 Jun 2015, Ville Voutilainen wrote:
- explicit
- constexpr tuple(const _Elements&... __elements)
+ template...>::value
+ && _TC<(sizeof...(_Elements) == sizeof...(_UElements)),
+_Elements...>::template
+ _ConstructibleTuple<_UElements...>()
+ && _TC<(sizeof...(_Elements) == sizeof...(_UElements)),
+_Elements...>::template
+ _ImplicitlyConvertibleTuple<_UElements...>()
+ && !_TC<(sizeof...(_Elements) == sizeof...(_UElements)),
+_Elements...>::template
+ _MoveConstructibleTuple<_UElements...>()
+ && !_TC<(sizeof...(_Elements) == sizeof...(_UElements)),
+_Elements...>::template
+ _ImplicitlyMoveConvertibleTuple<_UElements...>()
+ && (sizeof...(_Elements) >= 1),
+bool>::type=true>
+ constexpr tuple(const _UElements&... __elements)
: _Inherited(__elements...) { }
Since the paper does not mention looking at _MoveConstructibleTuple or
_ImplicitlyMoveConvertibleTuple here, could you add a comment explaining
why that is needed?
Does the following code still compile with your patch?
struct A { int a,b; };
std::tuple a(3,4,{1,2});
IMO the parts with is_default_constructible point to a core issue, we
should not have to duplicate information, especially in such a convoluted
way. But I guess that has lower priority than noexcept(auto), and I
haven't yet looked if concepts will help.
You use a lot: typename enable_if::type=true
while the current code seems to favor: class=typename enable_if::type.
I don't really care which one is used, but it is easier to read when the
style is consistent through the library.
Introducing
typename _XXX = _TC<(sizeof...(_Elements) == sizeof...(_UElements)),
_Elements...>
and then using _XXX::template thing() might give less clutter when you
have to repeat it 4 times.
--
Marc Glisse
