[gomp4] remove kernel-specific launch
Tom,
as we discussed this patch removes the GOACC_kernels call, which forwards to
GOACC_parallel. We simply call GOACC_parallel directly.
ok for gomp4 branch?
nathan
2015-07-19 Nathan Sidwell
gcc/
* omp-low.c (expand_omp_target): Convert to
BUILT_IN_GOACC_PARALLEL. Remove BUILT_IN_GOACC_KERNELS cases.
* omp-builtins.def (BUILT_IN_GOACC_KERNELS): Delete.
libgomp/
* oacc-parallel.c (GOACC_kernels): Delete.
Index: gcc/omp-builtins.def
===
--- gcc/omp-builtins.def (revision 225982)
+++ gcc/omp-builtins.def (working copy)
@@ -48,9 +48,6 @@ DEF_GOACC_BUILTIN_FNSPEC (BUILT_IN_GOACC
BT_FN_VOID_INT_OMPFN_SIZE_PTR_PTR_PTR_INT_INT_INT_SIZE_INT_INT_VAR,
ATTR_FNSPEC_DOT_DOT_DOT_DOT_r_r_r_NOTHROW_LIST,
ATTR_NOTHROW_LIST, "rrr")
-DEF_GOACC_BUILTIN (BUILT_IN_GOACC_KERNELS, "GOACC_kernels",
- BT_FN_VOID_INT_OMPFN_SIZE_PTR_PTR_PTR_INT_INT_INT_SIZE_INT_INT_VAR,
- ATTR_NOTHROW_LIST)
DEF_GOACC_BUILTIN (BUILT_IN_GOACC_PARALLEL, "GOACC_parallel",
BT_FN_VOID_INT_OMPFN_SIZE_PTR_PTR_PTR_INT_INT_INT_SIZE_INT_INT_VAR,
ATTR_NOTHROW_LIST)
Index: gcc/omp-low.c
===
--- gcc/omp-low.c (revision 225982)
+++ gcc/omp-low.c (working copy)
@@ -9339,14 +9339,15 @@ expand_omp_target (struct omp_region *re
/* Don't emit the library call. We've already done that. */
do_emit_library_call = false;
/* Transform BUILT_IN_GOACC_KERNELS_INTERNAL into
- BUILT_IN_GOACC_KERNELS_INTERNAL. Now that the function body will be
- split off, we can no longer regard the omp_data_array reference as
- non-escaping. */
+ BUILT_IN_GOACC_PARALLELL. Now that the function
+ body will be split off, we can no longer regard the
+ omp_data_array reference as non-escaping. */
gsi = gsi_last_bb (entry_bb);
gsi_prev (&gsi);
gcall *call = as_a (gsi_stmt (gsi));
- gcc_assert (gimple_call_builtin_p (call, BUILT_IN_GOACC_KERNELS_INTERNAL));
- tree fndecl = builtin_decl_explicit (BUILT_IN_GOACC_KERNELS);
+ gcc_assert (gimple_call_builtin_p
+ (call, BUILT_IN_GOACC_KERNELS_INTERNAL));
+ tree fndecl = builtin_decl_explicit (BUILT_IN_GOACC_PARALLEL);
gimple_call_set_fndecl (call, fndecl);
gimple_call_set_fntype (call, TREE_TYPE (fndecl));
gimple_call_reset_alias_info (call);
@@ -9723,7 +9724,6 @@ expand_omp_target (struct omp_region *re
case BUILT_IN_GOACC_DATA_START:
case BUILT_IN_GOACC_DECLARE:
case BUILT_IN_GOACC_ENTER_EXIT_DATA:
-case BUILT_IN_GOACC_KERNELS:
case BUILT_IN_GOACC_KERNELS_INTERNAL:
case BUILT_IN_GOACC_PARALLEL:
case BUILT_IN_GOACC_UPDATE:
@@ -9743,7 +9743,6 @@ expand_omp_target (struct omp_region *re
case BUILT_IN_GOMP_TARGET_DATA:
case BUILT_IN_GOMP_TARGET_UPDATE:
break;
-case BUILT_IN_GOACC_KERNELS:
case BUILT_IN_GOACC_KERNELS_INTERNAL:
case BUILT_IN_GOACC_PARALLEL:
{
Index: libgomp/oacc-parallel.c
===
--- libgomp/oacc-parallel.c (revision 225982)
+++ libgomp/oacc-parallel.c (working copy)
@@ -530,37 +530,6 @@ GOACC_enter_exit_data (int device, size_
acc_dev->openacc.async_set_async_func (acc_async_sync);
}
-void
-GOACC_kernels (int device, void (*fn) (void *),
- size_t mapnum, void **hostaddrs, size_t *sizes,
- unsigned short *kinds,
- int num_gangs, int num_workers, int vector_length,
- size_t shared_size, int async, int num_waits, ...)
-{
-#ifdef HAVE_INTTYPES_H
- gomp_debug (0, "%s: mapnum=%"PRIu64", hostaddrs=%p, sizes=%p, kinds=%p\n",
- __FUNCTION__, (uint64_t) mapnum, hostaddrs, sizes, kinds);
-#else
- gomp_debug (0, "%s: mapnum=%lu, hostaddrs=%p, sizes=%p, kinds=%p\n",
- __FUNCTION__, (unsigned long) mapnum, hostaddrs, sizes, kinds);
-#endif
-
- va_list ap;
-
- goacc_lazy_initialize ();
-
- va_start (ap, num_waits);
-
- if (num_waits > 0)
-goacc_wait (async, num_waits, ap);
-
- va_end (ap);
-
- GOACC_parallel (device, fn, mapnum, hostaddrs, sizes, kinds,
- num_gangs, num_workers, vector_length, shared_size,
- async, 0);
-}
-
static void
goacc_wait (int async, int num_waits, va_list ap)
{
Refactor openacc wait routine
this trunk patch refactors libgomp's goacc_wait, which is used for two different
purposes.
1) when openacc pragmas specify a (non-zero) waits.
2) when the wait pragma itself specifies a zero number of waits.
this leads to #2 calling goacc_wait with num_waits=0, and forces #1 to never do
that.
Fixed by breaking out the num_waits == 0 handling from goacc_wait into
GOACC_wait, the wait pragma handler. I have kept the num_wait=0 checks
elsewhere, but they are now for efficiency rather than correctness.
ok for trunk (& gomp4)
nathan
Index: libgomp/oacc-parallel.c
===
--- libgomp/oacc-parallel.c (revision 225959)
+++ libgomp/oacc-parallel.c (working copy)
@@ -105,13 +105,13 @@ GOACC_parallel (int device, void (*fn) (
return;
}
- va_start (ap, num_waits);
+ if (num_waits)
+{
+ va_start (ap, num_waits);
+ goacc_wait (async, num_waits, ap);
+ va_end (ap);
+}
- if (num_waits > 0)
-goacc_wait (async, num_waits, ap);
-
- va_end (ap);
-
acc_dev->openacc.async_set_async_func (async);
if (!(acc_dev->capabilities & GOMP_OFFLOAD_CAP_NATIVE_EXEC))
@@ -225,14 +225,12 @@ GOACC_enter_exit_data (int device, size_
|| host_fallback)
return;
- if (num_waits > 0)
+ if (num_waits)
{
va_list ap;
va_start (ap, num_waits);
-
goacc_wait (async, num_waits, ap);
-
va_end (ap);
}
@@ -350,47 +348,21 @@ goacc_wait (int async, int num_waits, va
{
struct goacc_thread *thr = goacc_thread ();
struct gomp_device_descr *acc_dev = thr->dev;
- int i;
-
- assert (num_waits >= 0);
-
- if (async == acc_async_sync && num_waits == 0)
-{
- acc_wait_all ();
- return;
-}
-
- if (async == acc_async_sync && num_waits)
-{
- for (i = 0; i < num_waits; i++)
-{
- int qid = va_arg (ap, int);
-
- if (acc_async_test (qid))
-continue;
- acc_wait (qid);
-}
- return;
-}
-
- if (async == acc_async_noval && num_waits == 0)
-{
- acc_dev->openacc.async_wait_all_async_func (acc_async_noval);
- return;
-}
-
- for (i = 0; i < num_waits; i++)
+ while (num_waits--)
{
int qid = va_arg (ap, int);
if (acc_async_test (qid))
continue;
- /* If we're waiting on the same asynchronous queue as we're launching on,
- the queue itself will order work as required, so there's no need to
- wait explicitly. */
- if (qid != async)
+ if (async == acc_async_sync)
+ acc_wait (qid);
+ else if (qid == async)
+ ;/* If we're waiting on the same asynchronous queue as we're
+ launching on, the queue itself will order work as
+ required, so there's no need to wait explicitly. */
+ else
acc_dev->openacc.async_wait_async_func (qid, async);
}
}
@@ -412,14 +384,12 @@ GOACC_update (int device, size_t mapnum,
|| host_fallback)
return;
- if (num_waits > 0)
+ if (num_waits)
{
va_list ap;
va_start (ap, num_waits);
-
goacc_wait (async, num_waits, ap);
-
va_end (ap);
}
@@ -455,13 +425,18 @@ GOACC_update (int device, size_t mapnum,
void
GOACC_wait (int async, int num_waits, ...)
{
- va_list ap;
-
- va_start (ap, num_waits);
-
- goacc_wait (async, num_waits, ap);
+ if (num_waits)
+{
+ va_list ap;
- va_end (ap);
+ va_start (ap, num_waits);
+ goacc_wait (async, num_waits, ap);
+ va_end (ap);
+}
+ else if (async == acc_async_sync)
+acc_wait_all ();
+ else if (async == acc_async_noval)
+acc_dev->openacc.async_wait_all_async_func (acc_async_noval);
}
int
Re: [gomp4] remove kernel-specific launch
On 19/07/15 14:27, Nathan Sidwell wrote: Tom, as we discussed this patch removes the GOACC_kernels call, which forwards to GOACC_parallel. We simply call GOACC_parallel directly. ok for gomp4 branch? Hi Nathan, yes, that looks good. Thanks, - Tom
[v3 PATCH] Implement N4089 and N4387
Tested on Linux-PPC64. 2015-07-19 Ville Voutilainen Implement N4089 Safe conversions in unique_ptr (LWG 2118) and N4387 LWG 2228: Missing SFINAE rule in unique_ptr templated assignment * include/bits/unique_ptr.h (__remove_cv, __is_derived_Tp): Remove. (default_delete::default_delete(const default_delete<_Up[]>)): Constrain with array convertibility. (default_delete::operator(_Up*)): Turn into a template, constrain with array convertibility. (__safe_conversion_up): New, single object version. (unique_ptr(unique_ptr<_Up, _Ep>&& __u)): Constrain with array convertibility. (unique_ptr::operator=(unique_ptr<_Up, _Ep>&& __u)): Likewise, and add is_assignable as a constraint. (__safe_conversion_up): Array version, renamed from __safe_conversion, updated to implement N4089. (__safe_conversion_raw): New. (unique_ptr(_Up __p)): Turn into a template, constrain with array convertibility. (unique_ptr(_Up __p, typename conditional::value, deleter_type, const deleter_type&>::type __d)): Likewise. (unique_ptr(_Up __p, typename remove_reference::type&& __d)): Likewise. (unique_ptr(unique_ptr<_Up, _Ep>&& __u)): Likewise. (operator=(unique_ptr<_Up, _Ep>&& __u)): Likewise, and add is_assignable as a constraint (array version). (reset(_Up __p)): Turn into a template, constrain with array convertibility. (reset(nullptr_t p)): New. * testsuite/20_util/default_delete/48631_neg.cc: Adjust. * testsuite/20_util/unique_ptr/assign/48635.cc: Likewise. * testsuite/20_util/unique_ptr/assign/48635_neg.cc: Likewise. * testsuite/20_util/unique_ptr/assign/cv_qual.cc: Likewise. * testsuite/20_util/unique_ptr/cons/cv_qual.cc: Likewise. * testsuite/20_util/unique_ptr/dr2228.cc: New. * testsuite/20_util/unique_ptr/modifiers/cv_qual.cc: Adjust. unique_ptr_fixes.diff.gz Description: GNU Zip compressed data
Re: [v3 PATCH] Implement N4089 and N4387
On 19 July 2015 at 19:56, Ville Voutilainen wrote: > Tested on Linux-PPC64. > > 2015-07-19 Ville Voutilainen > Implement N4089 Safe conversions in unique_ptr (LWG 2118) > and N4387 LWG 2228: Missing SFINAE rule in unique_ptr > templated assignment > * include/bits/unique_ptr.h > (__remove_cv, __is_derived_Tp): Remove. > (default_delete::default_delete(const default_delete<_Up[]>)): > Constrain with array convertibility. > (default_delete::operator(_Up*)): Turn into a template, > constrain with array convertibility. > (__safe_conversion_up): New, single object version. > (unique_ptr(unique_ptr<_Up, _Ep>&& __u)): Constrain with array > convertibility. > (unique_ptr::operator=(unique_ptr<_Up, _Ep>&& __u)): Likewise, and add > is_assignable as a constraint. > (__safe_conversion_up): Array version, renamed from __safe_conversion, > updated to implement N4089. > (__safe_conversion_raw): New. > (unique_ptr(_Up __p)): Turn into a template, constrain with array > convertibility. > (unique_ptr(_Up __p, > typename conditional::value, > deleter_type, const deleter_type&>::type __d)): Likewise. > (unique_ptr(_Up __p, typename > remove_reference::type&& __d)): Likewise. > (unique_ptr(unique_ptr<_Up, _Ep>&& __u)): Likewise. > (operator=(unique_ptr<_Up, _Ep>&& __u)): Likewise, and add > is_assignable as a constraint (array version). > (reset(_Up __p)): Turn into a template, constrain with array > convertibility. > (reset(nullptr_t p)): New. > * testsuite/20_util/default_delete/48631_neg.cc: Adjust. > * testsuite/20_util/unique_ptr/assign/48635.cc: Likewise. > * testsuite/20_util/unique_ptr/assign/48635_neg.cc: Likewise. > * testsuite/20_util/unique_ptr/assign/cv_qual.cc: Likewise. > * testsuite/20_util/unique_ptr/cons/cv_qual.cc: Likewise. > * testsuite/20_util/unique_ptr/dr2228.cc: New. > * testsuite/20_util/unique_ptr/modifiers/cv_qual.cc: Adjust. Minor changelog fix, the single-pointer functions aren't constrained with array compatibility, but with deleter compatibility. 2015-07-19 Ville Voutilainen Implement N4089 Safe conversions in unique_ptr (LWG 2118) and N4387 LWG 2228: Missing SFINAE rule in unique_ptr templated assignment * include/bits/unique_ptr.h (__remove_cv, __is_derived_Tp): Remove. (default_delete::default_delete(const default_delete<_Up[]>)): Constrain with array convertibility. (default_delete::operator(_Up*)): Turn into a template, constrain with array convertibility. (__safe_conversion_up): New, single object version. (unique_ptr(unique_ptr<_Up, _Ep>&& __u)): Constrain with deleter convertibility. (unique_ptr::operator=(unique_ptr<_Up, _Ep>&& __u)): Likewise, and add is_assignable as a constraint. (__safe_conversion_up): Array version, renamed from __safe_conversion, updated to implement N4089. (__safe_conversion_raw): New. (unique_ptr(_Up __p)): Turn into a template, constrain with array convertibility. (unique_ptr(_Up __p, typename conditional::value, deleter_type, const deleter_type&>::type __d)): Likewise. (unique_ptr(_Up __p, typename remove_reference::type&& __d)): Likewise. (unique_ptr(unique_ptr<_Up, _Ep>&& __u)): Likewise. (operator=(unique_ptr<_Up, _Ep>&& __u)): Likewise, and add is_assignable as a constraint (array version). (reset(_Up __p)): Turn into a template, constrain with array convertibility. (reset(nullptr_t p)): New. * testsuite/20_util/default_delete/48631_neg.cc: Adjust. * testsuite/20_util/unique_ptr/assign/48635.cc: Likewise. * testsuite/20_util/unique_ptr/assign/48635_neg.cc: Likewise. * testsuite/20_util/unique_ptr/assign/cv_qual.cc: Likewise. * testsuite/20_util/unique_ptr/cons/cv_qual.cc: Likewise. * testsuite/20_util/unique_ptr/dr2228.cc: New. * testsuite/20_util/unique_ptr/modifiers/cv_qual.cc: Adjust.
Re: [v3 PATCH] Implement N4089 and N4387
On 19 July 2015 at 20:13, Ville Voutilainen wrote: > Minor changelog fix, the single-pointer functions aren't constrained with > array compatibility, but with deleter compatibility. And that should be single-object, not single-pointer, not that it has any effect on the corrected changelog. :)
[patch] Fix wrong URL for N4279 in libstdc++ manual.
Committed to trunk. commit bfd6e14c25de5e2f97dd1ab037a53b35df4f Author: Jonathan Wakely Date: Sun Jul 19 19:31:16 2015 +0100 * doc/xml/manual/status_cxx2017.xml: Fix URL for N4279. * doc/html/manual/status.html: Regenerate. diff --git a/libstdc++-v3/doc/xml/manual/status_cxx2017.xml b/libstdc++-v3/doc/xml/manual/status_cxx2017.xml index 491e024..c0bca84 100644 --- a/libstdc++-v3/doc/xml/manual/status_cxx2017.xml +++ b/libstdc++-v3/doc/xml/manual/status_cxx2017.xml @@ -118,7 +118,7 @@ not in any particular release. - http://www.w3.org/1999/xlink"; xlink:href="http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2014/n4279.pdf";> + http://www.w3.org/1999/xlink"; xlink:href="http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2014/n4279.html";> N4279
[PATCH] Refactor graphite-isl-ast-to-gimple.c
Refactor graphite-isl-ast-to-gimple.c:
Refactor so that each function can access 'region'. This will help
maintain a parameter rename_map within a region. No functional change intended.
This patch will be followed by another set of patches
where translate_isl_ast_to_gimple::region is used to keep parameters which needs
renaming. Since we are planning to remove limit_scops, we now have to maintain a
set of parameters which needs renaming. This refactoring helps avoid passing
`region' to many functions.
gcc/ChangeLog:
2015-07-19 Aditya Kumar
* graphite-isl-ast-to-gimple.c:
Refactor so that each function can access 'region'. This will help
maintain a parameter rename_map within a region.
---
gcc/graphite-isl-ast-to-gimple.c | 153 +++
1 file changed, 122 insertions(+), 31 deletions(-)
diff --git a/gcc/graphite-isl-ast-to-gimple.c b/gcc/graphite-isl-ast-to-gimple.c
index b32781a..86a921b 100644
--- a/gcc/graphite-isl-ast-to-gimple.c
+++ b/gcc/graphite-isl-ast-to-gimple.c
@@ -124,9 +124,84 @@ void ivs_params_clear (ivs_params &ip)
}
}
-static tree
-gcc_expression_from_isl_expression (tree type, __isl_take isl_ast_expr *,
- ivs_params &ip);
+class translate_isl_ast_to_gimple
+{
+public:
+ translate_isl_ast_to_gimple (sese r)
+ : region (r)
+ { }
+
+ edge translate_isl_ast (loop_p context_loop, __isl_keep isl_ast_node *node,
+ edge next_e, ivs_params &ip);
+
+ edge translate_isl_ast_node_for (loop_p context_loop,
+ __isl_keep isl_ast_node *node,
+ edge next_e, ivs_params &ip);
+
+ edge translate_isl_ast_for_loop (loop_p context_loop,
+ __isl_keep isl_ast_node *node_for,
+ edge next_e,
+ tree type, tree lb, tree ub,
+ ivs_params &ip);
+
+ edge translate_isl_ast_node_if (loop_p context_loop,
+ __isl_keep isl_ast_node *node,
+ edge next_e, ivs_params &ip);
+
+ edge translate_isl_ast_node_user (__isl_keep isl_ast_node *node,
+ edge next_e, ivs_params &ip);
+
+ edge translate_isl_ast_node_block (loop_p context_loop,
+__isl_keep isl_ast_node *node,
+edge next_e, ivs_params &ip);
+
+ tree unary_op_to_tree (tree type, __isl_take isl_ast_expr *expr,
+ivs_params &ip);
+
+ tree binary_op_to_tree (tree type, __isl_take isl_ast_expr *expr,
+ ivs_params &ip);
+
+ tree ternary_op_to_tree (tree type, __isl_take isl_ast_expr *expr,
+ ivs_params &ip);
+
+ tree nary_op_to_tree (tree type, __isl_take isl_ast_expr *expr,
+ ivs_params &ip);
+
+ tree gcc_expression_from_isl_expression (tree type,
+ __isl_take isl_ast_expr *,
+ ivs_params &ip);
+
+ tree gcc_expression_from_isl_ast_expr_id (tree type,
+ __isl_keep isl_ast_expr *expr_id,
+ ivs_params &ip);
+
+ tree gcc_expression_from_isl_expr_int (tree type,
+__isl_take isl_ast_expr *expr);
+
+ tree gcc_expression_from_isl_expr_op (tree type,
+ __isl_take isl_ast_expr *expr,
+ ivs_params &ip);
+
+ struct loop *graphite_create_new_loop (edge entry_edge,
+__isl_keep isl_ast_node *node_for,
+loop_p outer, tree type,
+tree lb, tree ub, ivs_params &ip);
+
+ edge graphite_create_new_guard (edge entry_edge,
+ __isl_take isl_ast_expr *if_cond,
+ ivs_params &ip);
+
+ edge graphite_create_new_loop_guard (edge entry_edge,
+ __isl_keep isl_ast_node *node_for,
+ tree *type,
+ tree *lb, tree *ub, ivs_params &ip);
+
+ void build_iv_mapping (vec iv_map, gimple_bb_p gbb,
+__isl_keep isl_ast_expr *user_expr, ivs_params &ip,
+sese region);
+private:
+ sese region;
+};
/* Return the tree variable that corresponds to the given isl ast identifier
expression (an isl_ast_expr of type isl_ast_expr_id).
@@ -136,7 +211,8 @@ gcc_expression_from_isl_expression (tree type, __isl_take
isl_ast_expr *,
converting type sizes may be problematic when we switch to smaller
types. */
-static tree
+tree
+translate_isl_ast_to_gimple::
gcc_expression_from_isl_ast_expr_id (tree type,
RE: [PATCH] Rename parameters which are within scop
> Date: Sat, 18 Jul 2015 02:22:19 +0200
> From: tob...@grosser.es
> To: hiradi...@msn.com; gcc-patches@gcc.gnu.org
> CC: seb...@gmail.com; richard.guent...@gmail.com
> Subject: Re: [PATCH] Rename parameters which are within scop
>
> Hi Aditya,
>
> could you possible expand the commit message a little bit to explain
> what you are doing?
My bad,
I sent a different patch which did not have the updated commit message. I have
send another patch (https://gcc.gnu.org/ml/gcc-patches/2015-07/msg01614.html)
which has comments and the email title is also relevant to the change. Sorry
for the confusion.
-Aditya
>
> Tobias
>
>
> On 07/18/2015 01:00 AM, Aditya Kumar wrote:> ---
>> gcc/graphite-isl-ast-to-gimple.c | 153
>> +++
>> 1 file changed, 122 insertions(+), 31 deletions(-)
>>
>> diff --git a/gcc/graphite-isl-ast-to-gimple.c
>> b/gcc/graphite-isl-ast-to-gimple.c
>> index b32781a..3e2c1fa 100644
>> --- a/gcc/graphite-isl-ast-to-gimple.c
>> +++ b/gcc/graphite-isl-ast-to-gimple.c
>> @@ -124,9 +124,84 @@ void ivs_params_clear (ivs_params &ip)
>> }
>> }
>>
>> -static tree
>> -gcc_expression_from_isl_expression (tree type, __isl_take isl_ast_expr *,
>> - ivs_params &ip);
>> +class translate_isl_ast_to_gimple
>> +{
>> +public:
>> + translate_isl_ast_to_gimple (sese r)
>> + : region (r)
>> + { }
>> +
>> + edge translate_isl_ast (loop_p context_loop, __isl_keep isl_ast_node *node,
>> + edge next_e, ivs_params &ip);
>> +
>> + edge translate_isl_ast_node_for (loop_p context_loop,
>> + __isl_keep isl_ast_node *node,
>> + edge next_e, ivs_params &ip);
>> +
>> + edge translate_isl_ast_for_loop (loop_p context_loop,
>> + __isl_keep isl_ast_node *node_for,
>> + edge next_e,
>> + tree type, tree lb, tree ub,
>> + ivs_params &ip);
>> +
>> + edge translate_isl_ast_node_if (loop_p context_loop,
>> + __isl_keep isl_ast_node *node,
>> + edge next_e, ivs_params &ip);
>> +
>> + edge translate_isl_ast_node_user (__isl_keep isl_ast_node *node,
>> + edge next_e, ivs_params &ip);
>> +
>> + edge translate_isl_ast_node_block (loop_p context_loop,
>> + __isl_keep isl_ast_node *node,
>> + edge next_e, ivs_params &ip);
>> +
>> + tree unary_op_to_tree (tree type, __isl_take isl_ast_expr *expr,
>> + ivs_params &ip);
>> +
>> + tree binary_op_to_tree (tree type, __isl_take isl_ast_expr *expr,
>> + ivs_params &ip);
>> +
>> + tree ternary_op_to_tree (tree type, __isl_take isl_ast_expr *expr,
>> + ivs_params &ip);
>> +
>> + tree nary_op_to_tree (tree type, __isl_take isl_ast_expr *expr,
>> + ivs_params &ip);
>> +
>> + tree gcc_expression_from_isl_expression (tree type,
>> + __isl_take isl_ast_expr *,
>> + ivs_params &ip);
>> +
>> + tree gcc_expression_from_isl_ast_expr_id (tree type,
>> + __isl_keep isl_ast_expr *expr_id,
>> + ivs_params &ip);
>> +
>> + tree gcc_expression_from_isl_expr_int (tree type,
>> + __isl_take isl_ast_expr *expr);
>> +
>> + tree gcc_expression_from_isl_expr_op (tree type,
>> + __isl_take isl_ast_expr *expr,
>> + ivs_params &ip);
>> +
>> + struct loop *graphite_create_new_loop (edge entry_edge,
>> + __isl_keep isl_ast_node *node_for,
>> + loop_p outer, tree type,
>> + tree lb, tree ub, ivs_params &ip);
>> +
>> + edge graphite_create_new_guard (edge entry_edge,
>> + __isl_take isl_ast_expr *if_cond,
>> + ivs_params &ip);
>> +
>> + edge graphite_create_new_loop_guard (edge entry_edge,
>> + __isl_keep isl_ast_node *node_for,
>> + tree *type,
>> + tree *lb, tree *ub, ivs_params &ip);
>> +
>> + void build_iv_mapping (vec iv_map, gimple_bb_p gbb,
>> + __isl_keep isl_ast_expr *user_expr, ivs_params &ip,
>> + sese region);
>> +private:
>> + sese region;
>> +};
>>
>> /* Return the tree variable that corresponds to the given isl ast identifier
>> expression (an isl_ast_expr of type isl_ast_expr_id).
>> @@ -136,7 +211,8 @@ gcc_expression_from_isl_expression (tree type,
>> __isl_take isl_ast_expr *,
>> converting type sizes may be problematic when we switch to smaller
>> types. */
>>
>> -static tree
>> +tree
>> +translate_isl_ast_to_gimple::
>> gcc_expression_from_isl_ast_expr_id (tree type,
>> __isl_keep isl_ast_expr *expr_id,
>> ivs_params &ip)
>> @@ -147,7 +223,7 @@ gcc_expression_from_isl_ast_expr_id (tree type,
>> res = ip.find (tmp_isl_id);
>> isl_id_free (tmp_isl_id);
>> gcc_assert (res != ip.end () &&
>> - "Could not map isl_id to tree expression");
>> + "Could not map isl_id to tree expression");
>> isl_ast_expr_free (expr_id);
>> return fold_convert (type, res->second);
>> }
>> @@ -155,7 +231,8 @@ gcc_expression_from_isl_ast_expr_id (tree type,
>> /* Converts an isl_ast_expr_int expression E to a GCC expression tree of
>> type TYPE. */
>>
>> -static tree
>> +tree
>> +translate_isl_ast_to_gimple::
>> gcc_expression_from_isl_expr_int (tree type, __isl_take isl_ast_expr *expr)
>> {
>> gcc_assert (isl_ast_expr_get_type (expr) == isl_ast_expr_int);
>> @@ -176,7 +253,8 @@ gcc_expression_from_isl_expr_int (tree type, __isl_take
>> isl_ast_expr *expr
Re: [PATCH] Refactor graphite-isl-ast-to-gimple.c
On 07/19/2015 09:46 PM, Aditya Kumar wrote:
Refactor graphite-isl-ast-to-gimple.c:
Refactor so that each function can access 'region'. This will help
maintain a parameter rename_map within a region. No functional change intended.
This patch will be followed by another set of patches
where translate_isl_ast_to_gimple::region is used to keep parameters which needs
renaming. Since we are planning to remove limit_scops, we now have to maintain a
set of parameters which needs renaming. This refactoring helps avoid passing
`region' to many functions.
Ah, OK. Now I get it. Generally this looks fine to me (assuming it
passes a bootstrap), but I wonder if it would not make more sense to
move the comments inside the class right over the function declarations.
Best,
Tobias
gcc/ChangeLog:
2015-07-19 Aditya Kumar
* graphite-isl-ast-to-gimple.c:
Refactor so that each function can access 'region'. This will help
maintain a parameter rename_map within a region.
---
gcc/graphite-isl-ast-to-gimple.c | 153 +++
1 file changed, 122 insertions(+), 31 deletions(-)
diff --git a/gcc/graphite-isl-ast-to-gimple.c b/gcc/graphite-isl-ast-to-gimple.c
index b32781a..86a921b 100644
--- a/gcc/graphite-isl-ast-to-gimple.c
+++ b/gcc/graphite-isl-ast-to-gimple.c
@@ -124,9 +124,84 @@ void ivs_params_clear (ivs_params &ip)
}
}
-static tree
-gcc_expression_from_isl_expression (tree type, __isl_take isl_ast_expr *,
- ivs_params &ip);
+class translate_isl_ast_to_gimple
+{
+public:
+ translate_isl_ast_to_gimple (sese r)
+ : region (r)
+ { }
+
+ edge translate_isl_ast (loop_p context_loop, __isl_keep isl_ast_node *node,
+ edge next_e, ivs_params &ip);
+
+ edge translate_isl_ast_node_for (loop_p context_loop,
+ __isl_keep isl_ast_node *node,
+ edge next_e, ivs_params &ip);
+
+ edge translate_isl_ast_for_loop (loop_p context_loop,
+ __isl_keep isl_ast_node *node_for,
+ edge next_e,
+ tree type, tree lb, tree ub,
+ ivs_params &ip);
+
+ edge translate_isl_ast_node_if (loop_p context_loop,
+ __isl_keep isl_ast_node *node,
+ edge next_e, ivs_params &ip);
+
+ edge translate_isl_ast_node_user (__isl_keep isl_ast_node *node,
+ edge next_e, ivs_params &ip);
+
+ edge translate_isl_ast_node_block (loop_p context_loop,
+__isl_keep isl_ast_node *node,
+edge next_e, ivs_params &ip);
+
+ tree unary_op_to_tree (tree type, __isl_take isl_ast_expr *expr,
+ivs_params &ip);
+
+ tree binary_op_to_tree (tree type, __isl_take isl_ast_expr *expr,
+ ivs_params &ip);
+
+ tree ternary_op_to_tree (tree type, __isl_take isl_ast_expr *expr,
+ ivs_params &ip);
+
+ tree nary_op_to_tree (tree type, __isl_take isl_ast_expr *expr,
+ ivs_params &ip);
+
+ tree gcc_expression_from_isl_expression (tree type,
+ __isl_take isl_ast_expr *,
+ ivs_params &ip);
+
+ tree gcc_expression_from_isl_ast_expr_id (tree type,
+ __isl_keep isl_ast_expr *expr_id,
+ ivs_params &ip);
+
+ tree gcc_expression_from_isl_expr_int (tree type,
+__isl_take isl_ast_expr *expr);
+
+ tree gcc_expression_from_isl_expr_op (tree type,
+ __isl_take isl_ast_expr *expr,
+ ivs_params &ip);
+
+ struct loop *graphite_create_new_loop (edge entry_edge,
+__isl_keep isl_ast_node *node_for,
+loop_p outer, tree type,
+tree lb, tree ub, ivs_params &ip);
+
+ edge graphite_create_new_guard (edge entry_edge,
+ __isl_take isl_ast_expr *if_cond,
+ ivs_params &ip);
+
+ edge graphite_create_new_loop_guard (edge entry_edge,
+ __isl_keep isl_ast_node *node_for,
+ tree *type,
+ tree *lb, tree *ub, ivs_params &ip);
+
+ void build_iv_mapping (vec iv_map, gimple_bb_p gbb,
+__isl_keep isl_ast_expr *user_expr, ivs_params &ip,
+sese region);
+private:
+ sese region;
+};
/* Return the tree variable that corresponds to the given isl ast identifier
expression (an isl_ast_expr of type isl_ast_expr_id).
@@ -136
Re: [gomp4] remove kernel-specific launch
Hi! On Sun, 19 Jul 2015 08:27:44 -0400, Nathan Sidwell wrote: > Tom, > as we discussed this patch removes the GOACC_kernels call, which forwards to > GOACC_parallel. We simply call GOACC_parallel directly. > > ok for gomp4 branch? > gcc/ > * omp-low.c (expand_omp_target): Convert to > BUILT_IN_GOACC_PARALLEL. Remove BUILT_IN_GOACC_KERNELS cases. > * omp-builtins.def (BUILT_IN_GOACC_KERNELS): Delete. > > libgomp/ > * oacc-parallel.c (GOACC_kernels): Delete. Yes, thanks for the cleanup. Please also adjust/remove the following: gcc/tree-parloops.c:/* Remove GOACC_kernels. */ libgomp/libgomp.map:GOACC_kernels; libgomp/libgomp_g.h:extern void GOACC_kernels (int, void (*) (void *), size_t, Does it make sense then to rename GOACC_kernels_internal to GOACC_kernels? Grüße, Thomas signature.asc Description: PGP signature
Re: [gomp4] remove kernel-specific launch
On 19/07/15 22:30, Thomas Schwinge wrote: Does it make sense then to rename GOACC_kernels_internal to GOACC_kernels? I prefer to keep the name GOACC_kernels_internal, IMO it's clearer. Thanks, - Tom
Fix PTX loaded module data
Hi,
this patch fixes the ptx plugin's management of offloaded functions. Currently
it has a single global list of functions. But it should track this information
per ptx device instance. I've moved the linked list into the per-device data.
We were also not managing the plugin-allocated vector of offloaded functions,
with some totally bogus code in the unload routine. This is currently not used,
because there is no unloading destructor. The patch fixes that.
A followup patch will then be able to enable unloading.
ok for trunk?
nathan
2015-07-18 Nathan Sidwell
* plugin/plugin-nvptx.c (struct targ_fn_descriptor): Move later.
(struct ptx_image_data): Move earlier, add fns field.
(struct ptx_device): Add images and image_lock fields.
(ptx_images, ptx_image_lock): Delete.
(nvptx_open_device): Initialize images and image_lock fields.
(nvptx_close_device): Destroy image_lock.
(GOMP_OFFLOAD_load_image): Register image to device-specific fields.
(GOMP_OFFLOAD_unload_image): Unregister image from device-specific
fields.
Index: libgomp/plugin/plugin-nvptx.c
===
--- libgomp/plugin/plugin-nvptx.c (revision 225984)
+++ libgomp/plugin/plugin-nvptx.c (working copy)
@@ -127,12 +127,6 @@ cuda_error (CUresult r)
return &errmsg[0];
}
-struct targ_fn_descriptor
-{
- CUfunction fn;
- const char *name;
-};
-
static unsigned int instantiated_devices = 0;
static pthread_mutex_t ptx_dev_lock = PTHREAD_MUTEX_INITIALIZER;
@@ -288,6 +282,25 @@ map_push (struct ptx_stream *s, int asyn
return;
}
+/* Descriptor of a loaded function. */
+
+struct targ_fn_descriptor
+{
+ CUfunction fn;
+ const char *name;
+};
+
+/* A loaded PTX image. */
+struct ptx_image_data
+{
+ const void *target_data;
+ CUmodule module;
+
+ struct targ_fn_descriptor *fns; /* Array of functions. */
+
+ struct ptx_image_data *next;
+};
+
struct ptx_device
{
CUcontext ctx;
@@ -311,6 +324,9 @@ struct ptx_device
int mode;
bool mkern;
+ struct ptx_image_data *images; /* Images loaded on device. */
+ pthread_mutex_t image_lock; /* Lock for above list. */
+
struct ptx_device *next;
};
@@ -332,21 +348,11 @@ struct ptx_event
struct ptx_event *next;
};
-struct ptx_image_data
-{
- const void *target_data;
- CUmodule module;
- struct ptx_image_data *next;
-};
-
static pthread_mutex_t ptx_event_lock;
static struct ptx_event *ptx_events;
static struct ptx_device **ptx_devices;
-static struct ptx_image_data *ptx_images = NULL;
-static pthread_mutex_t ptx_image_lock = PTHREAD_MUTEX_INITIALIZER;
-
#define _XSTR(s) _STR(s)
#define _STR(s) #s
@@ -590,6 +596,7 @@ select_stream_for_async (int async, pthr
/* Initialize the device. Return TRUE on success, else FALSE. PTX_DEV_LOCK
should be locked on entry and remains locked on exit. */
+
static bool
nvptx_init (void)
{
@@ -746,6 +753,9 @@ nvptx_open_device (int n)
if (r != CUDA_SUCCESS)
async_engines = 1;
+ ptx_dev->images = NULL;
+ pthread_mutex_init (&ptx_dev->image_lock, NULL);
+
init_streams_for_device (ptx_dev, async_engines);
return ptx_dev;
@@ -760,6 +770,8 @@ nvptx_close_device (struct ptx_device *p
return;
fini_streams_for_device (ptx_dev);
+
+ pthread_mutex_destroy (&ptx_dev->image_lock);
if (!ptx_dev->ctx_shared)
{
@@ -1632,6 +1644,9 @@ typedef struct nvptx_tdata
size_t fn_num;
} nvptx_tdata_t;
+/* Load the (partial) program described by TARGET_DATA to device
+ number ORD. Allocate and return TARGET_TABLE. */
+
int
GOMP_OFFLOAD_load_image (int ord, const void *target_data,
struct addr_pair **target_table)
@@ -1641,23 +1656,19 @@ GOMP_OFFLOAD_load_image (int ord, const
unsigned int fn_entries, var_entries, i, j;
CUresult r;
struct targ_fn_descriptor *targ_fns;
+ struct addr_pair *targ_tbl;
const nvptx_tdata_t *img_header = (const nvptx_tdata_t *) target_data;
struct ptx_image_data *new_image;
+ struct ptx_device *dev;
GOMP_OFFLOAD_init_device (ord);
+ dev = ptx_devices[ord];
+
nvptx_attach_host_thread_to_device (ord);
link_ptx (&module, img_header->ptx_src);
- pthread_mutex_lock (&ptx_image_lock);
- new_image = GOMP_PLUGIN_malloc (sizeof (struct ptx_image_data));
- new_image->target_data = target_data;
- new_image->module = module;
- new_image->next = ptx_images;
- ptx_images = new_image;
- pthread_mutex_unlock (&ptx_image_lock);
-
/* The mkoffload utility emits a struct of pointers/integers at the
start of each offload image. The array of kernel names and the
functions addresses form a one-to-one correspondence. */
@@ -1667,12 +1678,24 @@ GOMP_OFFLOAD_load_image (int ord, const
fn_entries = img_header->fn_num;
fn_names = img_header->fn_names;
- *target_table = GOMP_PLUGIN_malloc (sizeof (struct addr_pair)
- * (fn_entries + var_entries));
+ targ_tbl = GOMP_PLUGIN_malloc (sizeof (struct addr_pair)
+ * (fn_e
Re: [gomp4] remove kernel-specific launch
On 07/19/15 16:30, Thomas Schwinge wrote:
gcc/tree-parloops.c:/* Remove GOACC_kernels. */
libgomp/libgomp.map:GOACC_kernels;
libgomp/libgomp_g.h:extern void GOACC_kernels (int, void (*) (void *),
size_t,
I fixed all byt the parloops comment. That comment didn't really make sense to
me -- it seems to be doing something with the pragma not the call. Perhaps Tom
could correct/clarify it?
Does it make sense then to rename GOACC_kernels_internal to
GOACC_kernels?
I agree with Tom. But perhaps it should be an internal fn? IIUC those are for
pseudo-funcs that should be converted to something else before the end of
compilation.
nathan
2015-07-19 Nathan Sidwell
gcc/
* omp-low.c (expand_omp_target): Convert to
BUILT_IN_GOACC_PARALLEL. Remove BUILT_IN_GOACC_KERNELS cases.
* omp-builtins.def (BUILT_IN_GOACC_KERNELS): Delete.
libgomp/
* oacc-parallel.c (GOACC_kernels): Delete.
* libgomp.map: Remove GOACC_kernels.
* libgomp_g.h (GOACC_kernels): Remove.
Index: gcc/omp-low.c
===
--- gcc/omp-low.c (revision 225982)
+++ gcc/omp-low.c (working copy)
@@ -9339,14 +9339,15 @@ expand_omp_target (struct omp_region *re
/* Don't emit the library call. We've already done that. */
do_emit_library_call = false;
/* Transform BUILT_IN_GOACC_KERNELS_INTERNAL into
- BUILT_IN_GOACC_KERNELS_INTERNAL. Now that the function body will be
- split off, we can no longer regard the omp_data_array reference as
- non-escaping. */
+ BUILT_IN_GOACC_PARALLELL. Now that the function
+ body will be split off, we can no longer regard the
+ omp_data_array reference as non-escaping. */
gsi = gsi_last_bb (entry_bb);
gsi_prev (&gsi);
gcall *call = as_a (gsi_stmt (gsi));
- gcc_assert (gimple_call_builtin_p (call, BUILT_IN_GOACC_KERNELS_INTERNAL));
- tree fndecl = builtin_decl_explicit (BUILT_IN_GOACC_KERNELS);
+ gcc_assert (gimple_call_builtin_p
+ (call, BUILT_IN_GOACC_KERNELS_INTERNAL));
+ tree fndecl = builtin_decl_explicit (BUILT_IN_GOACC_PARALLEL);
gimple_call_set_fndecl (call, fndecl);
gimple_call_set_fntype (call, TREE_TYPE (fndecl));
gimple_call_reset_alias_info (call);
@@ -9723,7 +9724,6 @@ expand_omp_target (struct omp_region *re
case BUILT_IN_GOACC_DATA_START:
case BUILT_IN_GOACC_DECLARE:
case BUILT_IN_GOACC_ENTER_EXIT_DATA:
-case BUILT_IN_GOACC_KERNELS:
case BUILT_IN_GOACC_KERNELS_INTERNAL:
case BUILT_IN_GOACC_PARALLEL:
case BUILT_IN_GOACC_UPDATE:
@@ -9743,7 +9743,6 @@ expand_omp_target (struct omp_region *re
case BUILT_IN_GOMP_TARGET_DATA:
case BUILT_IN_GOMP_TARGET_UPDATE:
break;
-case BUILT_IN_GOACC_KERNELS:
case BUILT_IN_GOACC_KERNELS_INTERNAL:
case BUILT_IN_GOACC_PARALLEL:
{
Index: gcc/omp-builtins.def
===
--- gcc/omp-builtins.def (revision 225982)
+++ gcc/omp-builtins.def (working copy)
@@ -48,9 +48,6 @@ DEF_GOACC_BUILTIN_FNSPEC (BUILT_IN_GOACC
BT_FN_VOID_INT_OMPFN_SIZE_PTR_PTR_PTR_INT_INT_INT_SIZE_INT_INT_VAR,
ATTR_FNSPEC_DOT_DOT_DOT_DOT_r_r_r_NOTHROW_LIST,
ATTR_NOTHROW_LIST, "rrr")
-DEF_GOACC_BUILTIN (BUILT_IN_GOACC_KERNELS, "GOACC_kernels",
- BT_FN_VOID_INT_OMPFN_SIZE_PTR_PTR_PTR_INT_INT_INT_SIZE_INT_INT_VAR,
- ATTR_NOTHROW_LIST)
DEF_GOACC_BUILTIN (BUILT_IN_GOACC_PARALLEL, "GOACC_parallel",
BT_FN_VOID_INT_OMPFN_SIZE_PTR_PTR_PTR_INT_INT_INT_SIZE_INT_INT_VAR,
ATTR_NOTHROW_LIST)
Index: libgomp/libgomp.map
===
--- libgomp/libgomp.map (revision 225982)
+++ libgomp/libgomp.map (working copy)
@@ -331,7 +331,6 @@ GOACC_2.0.GOMP_4_BRANCH {
global:
GOACC_deviceptr;
GOACC_get_ganglocal_ptr;
- GOACC_kernels;
GOACC_register_static;
} GOACC_2.0;
Index: libgomp/oacc-parallel.c
===
--- libgomp/oacc-parallel.c (revision 225982)
+++ libgomp/oacc-parallel.c (working copy)
@@ -530,37 +530,6 @@ GOACC_enter_exit_data (int device, size_
acc_dev->openacc.async_set_async_func (acc_async_sync);
}
-void
-GOACC_kernels (int device, void (*fn) (void *),
- size_t mapnum, void **hostaddrs, size_t *sizes,
- unsigned short *kinds,
- int num_gangs, int num_workers, int vector_length,
- size_t shared_size, int async, int num_waits, ...)
-{
-#ifdef HAVE_INTTYPES_H
- gomp_debug (0, "%s: mapnum=%"PRIu64", hostaddrs=%p, sizes=%p, kinds=%p\n",
- __FUNCTION__, (uint64_t) mapnum, hostaddrs, sizes, kinds);
-#else
- gomp_debug (0, "%s: mapnum=%lu, hostaddrs=%p, sizes=%p, kinds=%p\n",
- __FUNCTION__, (unsigned long) mapnum, hostaddrs, sizes, kinds);
-#endif
-
- va_list ap;
-
- goacc_lazy_initialize ();
-
- va_start (ap, num_waits);
-
- if (num_waits > 0)
-goacc_wait (async, num_waits, ap);
-
- va_end (ap);
-
Re: [PATCH] gcc: fix building w/isl-0.15
This patch and https://gcc.gnu.org/ml/gcc-patches/2015-07/msg01506.html
Appear to do the trick. I'm building with ISL 0.15 just fine now.
On Tue, Jul 14, 2015 at 2:45 PM, Mike Frysinger wrote:
> ---
> gcc/config.in | 6 ++
> gcc/configure | 31 +++
> gcc/configure.ac| 14 ++
> gcc/graphite-dependences.c | 14 +++---
> gcc/graphite-optimize-isl.c | 8 ++--
> gcc/graphite-poly.h | 5 +
> 6 files changed, 69 insertions(+), 9 deletions(-)
>
> diff --git a/gcc/config.in b/gcc/config.in
> index b031a62..23e1757 100644
> --- a/gcc/config.in
> +++ b/gcc/config.in
> @@ -1326,6 +1326,12 @@
> #endif
>
>
> +/* Define if isl_options_set_schedule_serialize_sccs exists. */
> +#ifndef USED_FOR_TARGET
> +#undef HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS
> +#endif
> +
> +
> /* Define if isl_schedule_constraints_compute_schedule exists. */
> #ifndef USED_FOR_TARGET
> #undef HAVE_ISL_SCHED_CONSTRAINTS_COMPUTE_SCHEDULE
> diff --git a/gcc/configure b/gcc/configure
> index 9561e5c..6e81298 100755
> --- a/gcc/configure
> +++ b/gcc/configure
> @@ -28456,6 +28456,8 @@ fi
>
> # Check whether isl_schedule_constraints_compute_schedule is available;
> # it's new in ISL-0.13.
> +# Check whether isl_options_set_schedule_serialize_sccs is available;
> +# it's new in ISL-0.15.
> if test "x${ISLLIBS}" != "x" ; then
>saved_CXXFLAGS="$CXXFLAGS"
>CXXFLAGS="$CXXFLAGS $ISLINC"
> @@ -28485,6 +28487,29 @@ rm -f core conftest.err conftest.$ac_objext \
>{ $as_echo "$as_me:${as_lineno-$LINENO}: result:
> $ac_has_isl_schedule_constraints_compute_schedule" >&5
> $as_echo "$ac_has_isl_schedule_constraints_compute_schedule" >&6; }
>
> + { $as_echo "$as_me:${as_lineno-$LINENO}: checking Checking for
> isl_options_set_schedule_serialize_sccs" >&5
> +$as_echo_n "checking Checking for isl_options_set_schedule_serialize_sccs...
> " >&6; }
> + cat confdefs.h - <<_ACEOF >conftest.$ac_ext
> +/* end confdefs.h. */
> +#include
> +int
> +main ()
> +{
> +isl_options_set_schedule_serialize_sccs (NULL, 0);
> + ;
> + return 0;
> +}
> +_ACEOF
> +if ac_fn_cxx_try_link "$LINENO"; then :
> + ac_has_isl_options_set_schedule_serialize_sccs=yes
> +else
> + ac_has_isl_options_set_schedule_serialize_sccs=no
> +fi
> +rm -f core conftest.err conftest.$ac_objext \
> +conftest$ac_exeext conftest.$ac_ext
> + { $as_echo "$as_me:${as_lineno-$LINENO}: result:
> $ac_has_isl_options_set_schedule_serialize_sccs" >&5
> +$as_echo "$ac_has_isl_options_set_schedule_serialize_sccs" >&6; }
> +
>LIBS="$saved_LIBS"
>CXXFLAGS="$saved_CXXFLAGS"
>
> @@ -28493,6 +28518,12 @@ $as_echo
> "$ac_has_isl_schedule_constraints_compute_schedule" >&6; }
> $as_echo "#define HAVE_ISL_SCHED_CONSTRAINTS_COMPUTE_SCHEDULE 1" >>confdefs.h
>
>fi
> +
> + if test x"$ac_has_isl_options_set_schedule_serialize_sccs" = x"yes"; then
> +
> +$as_echo "#define HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS 1"
> >>confdefs.h
> +
> + fi
> fi
>
> # Check for plugin support
> diff --git a/gcc/configure.ac b/gcc/configure.ac
> index cb14639..7fb964a 100644
> --- a/gcc/configure.ac
> +++ b/gcc/configure.ac
> @@ -5725,6 +5725,8 @@ fi
>
> # Check whether isl_schedule_constraints_compute_schedule is available;
> # it's new in ISL-0.13.
> +# Check whether isl_options_set_schedule_serialize_sccs is available;
> +# it's new in ISL-0.15.
> if test "x${ISLLIBS}" != "x" ; then
>saved_CXXFLAGS="$CXXFLAGS"
>CXXFLAGS="$CXXFLAGS $ISLINC"
> @@ -5738,6 +5740,13 @@ if test "x${ISLLIBS}" != "x" ; then
>[ac_has_isl_schedule_constraints_compute_schedule=no])
>AC_MSG_RESULT($ac_has_isl_schedule_constraints_compute_schedule)
>
> + AC_MSG_CHECKING([Checking for isl_options_set_schedule_serialize_sccs])
> + AC_TRY_LINK([#include ],
> + [isl_options_set_schedule_serialize_sccs (NULL, 0);],
> + [ac_has_isl_options_set_schedule_serialize_sccs=yes],
> + [ac_has_isl_options_set_schedule_serialize_sccs=no])
> + AC_MSG_RESULT($ac_has_isl_options_set_schedule_serialize_sccs)
> +
>LIBS="$saved_LIBS"
>CXXFLAGS="$saved_CXXFLAGS"
>
> @@ -5745,6 +5754,11 @@ if test "x${ISLLIBS}" != "x" ; then
> AC_DEFINE(HAVE_ISL_SCHED_CONSTRAINTS_COMPUTE_SCHEDULE, 1,
> [Define if isl_schedule_constraints_compute_schedule exists.])
>fi
> +
> + if test x"$ac_has_isl_options_set_schedule_serialize_sccs" = x"yes"; then
> + AC_DEFINE(HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS, 1,
> + [Define if isl_options_set_schedule_serialize_sccs exists.])
> + fi
> fi
>
> GCC_ENABLE_PLUGINS
> diff --git a/gcc/graphite-dependences.c b/gcc/graphite-dependences.c
> index 50fe73e..9a0986d 100644
> --- a/gcc/graphite-dependences.c
> +++ b/gcc/graphite-dependences.c
> @@ -205,7 +205,7 @@ scop_get_transformed_schedule (scop_p scop,
> vec pbbs)
> /* Helper function used on each MAP of a isl_union_map. Compute
Re: [PATCH] gcc/: Fix building with isl-0.15.0; includes
These two patches appear to do the trick. I'm building with ISL 0.15 just fine now. On Fri, Jul 17, 2015 at 11:10 AM, Bernhard Reutner-Fischer wrote: > In addition to Mike's "[PATCH] gcc: fix building w/isl-0.15" in > https://gcc.gnu.org/ml/gcc-patches/2015-07/msg01162.html > I seem to also need a couple of more includes for isl-0.15.0: > > 2015-07-15 Bernhard Reutner-Fischer > > * graphite-blocking.c, graphite-dependences.c, > graphite-interchange.c, graphite-isl-ast-to-gimple.c, > graphite-optimize-isl.c, graphite-poly.c, > graphite-scop-detection.c, graphite-sese-to-poly.c, > graphite.c: Add missing isl includes. > --- > gcc/graphite-blocking.c |1 + > gcc/graphite-dependences.c |1 + > gcc/graphite-interchange.c |1 + > gcc/graphite-isl-ast-to-gimple.c |2 ++ > gcc/graphite-optimize-isl.c |2 ++ > gcc/graphite-poly.c |1 + > gcc/graphite-scop-detection.c|1 + > gcc/graphite-sese-to-poly.c |1 + > gcc/graphite.c |1 + > 9 files changed, 11 insertions(+) > > diff --git a/gcc/graphite-blocking.c b/gcc/graphite-blocking.c > index 73d7c59..172bdcd 100644 > --- a/gcc/graphite-blocking.c > +++ b/gcc/graphite-blocking.c > @@ -27,6 +27,7 @@ along with GCC; see the file COPYING3. If not see > /* Workaround for GMP 5.1.3 bug, see PR56019. */ > #include > > +#include > #include > #include > #include > diff --git a/gcc/graphite-dependences.c b/gcc/graphite-dependences.c > index 9a0986d..df52e54 100644 > --- a/gcc/graphite-dependences.c > +++ b/gcc/graphite-dependences.c > @@ -25,6 +25,7 @@ along with GCC; see the file COPYING3. If not see > /* Workaround for GMP 5.1.3 bug, see PR56019. */ > #include > > +#include > #include > #include > #include > diff --git a/gcc/graphite-interchange.c b/gcc/graphite-interchange.c > index aee51a8..bcf7f3b 100644 > --- a/gcc/graphite-interchange.c > +++ b/gcc/graphite-interchange.c > @@ -27,6 +27,7 @@ along with GCC; see the file COPYING3. If not see > /* Workaround for GMP 5.1.3 bug, see PR56019. */ > #include > > +#include > #include > #include > #include > diff --git a/gcc/graphite-isl-ast-to-gimple.c > b/gcc/graphite-isl-ast-to-gimple.c > index b32781a..7aafee2 100644 > --- a/gcc/graphite-isl-ast-to-gimple.c > +++ b/gcc/graphite-isl-ast-to-gimple.c > @@ -24,7 +24,9 @@ along with GCC; see the file COPYING3. If not see > /* Workaround for GMP 5.1.3 bug, see PR56019. */ > #include > > +#include > #include > +#include > #include > #include > #include > diff --git a/gcc/graphite-optimize-isl.c b/gcc/graphite-optimize-isl.c > index 388e25c..2b55783 100644 > --- a/gcc/graphite-optimize-isl.c > +++ b/gcc/graphite-optimize-isl.c > @@ -24,7 +24,9 @@ along with GCC; see the file COPYING3. If not see > /* Workaround for GMP 5.1.3 bug, see PR56019. */ > #include > > +#include > #include > +#include > #include > #include > #include > diff --git a/gcc/graphite-poly.c b/gcc/graphite-poly.c > index bcd08d8..e3ddd57 100644 > --- a/gcc/graphite-poly.c > +++ b/gcc/graphite-poly.c > @@ -25,6 +25,7 @@ along with GCC; see the file COPYING3. If not see > /* Workaround for GMP 5.1.3 bug, see PR56019. */ > #include > > +#include > #include > #include > #include > diff --git a/gcc/graphite-scop-detection.c b/gcc/graphite-scop-detection.c > index 47e0da0..22fdde3 100644 > --- a/gcc/graphite-scop-detection.c > +++ b/gcc/graphite-scop-detection.c > @@ -25,6 +25,7 @@ along with GCC; see the file COPYING3. If not see > /* Workaround for GMP 5.1.3 bug, see PR56019. */ > #include > > +#include > #include > #include > #include > diff --git a/gcc/graphite-sese-to-poly.c b/gcc/graphite-sese-to-poly.c > index 5064247..7b10289 100644 > --- a/gcc/graphite-sese-to-poly.c > +++ b/gcc/graphite-sese-to-poly.c > @@ -24,6 +24,7 @@ along with GCC; see the file COPYING3. If not see > /* Workaround for GMP 5.1.3 bug, see PR56019. */ > #include > > +#include > #include > #include > #include > diff --git a/gcc/graphite.c b/gcc/graphite.c > index a81ef6a..32f405a 100644 > --- a/gcc/graphite.c > +++ b/gcc/graphite.c > @@ -38,6 +38,7 @@ along with GCC; see the file COPYING3. If not see > /* Workaround for GMP 5.1.3 bug, see PR56019. */ > #include > > +#include > #include > #include > #include > -- > 1.7.10.4 >
Re: [v3 PATCH] Implement N4089 and N4387
On 19 July 2015 at 20:14, Ville Voutilainen wrote: > On 19 July 2015 at 20:13, Ville Voutilainen > wrote: >> Minor changelog fix, the single-pointer functions aren't constrained with >> array compatibility, but with deleter compatibility. > > And that should be single-object, not single-pointer, not that it has any > effect on the corrected changelog. :) Argh, the paper number in the status table was wrong, although the link was correct, so the paper number for the missing sfinae condition in the changelog was also wrong. 2015-07-19 Ville Voutilainen Implement N4089 Safe conversions in unique_ptr (LWG 2118) and N4366 LWG 2228: Missing SFINAE rule in unique_ptr templated assignment * include/bits/unique_ptr.h (__remove_cv, __is_derived_Tp): Remove. (default_delete::default_delete(const default_delete<_Up[]>)): Constrain with array convertibility. (default_delete::operator(_Up*)): Turn into a template, constrain with array convertibility. (__safe_conversion_up): New, single object version. (unique_ptr(unique_ptr<_Up, _Ep>&& __u)): Constrain with deleter convertibility. (unique_ptr::operator=(unique_ptr<_Up, _Ep>&& __u)): Likewise, and add is_assignable as a constraint. (__safe_conversion_up): Array version, renamed from __safe_conversion, updated to implement N4089. (__safe_conversion_raw): New. (unique_ptr(_Up __p)): Turn into a template, constrain with array convertibility. (unique_ptr(_Up __p, typename conditional::value, deleter_type, const deleter_type&>::type __d)): Likewise. (unique_ptr(_Up __p, typename remove_reference::type&& __d)): Likewise. (unique_ptr(unique_ptr<_Up, _Ep>&& __u)): Likewise. (operator=(unique_ptr<_Up, _Ep>&& __u)): Likewise, and add is_assignable as a constraint (array version). (reset(_Up __p)): Turn into a template, constrain with array convertibility. (reset(nullptr_t p)): New. * testsuite/20_util/default_delete/48631_neg.cc: Adjust. * testsuite/20_util/unique_ptr/assign/48635.cc: Likewise. * testsuite/20_util/unique_ptr/assign/48635_neg.cc: Likewise. * testsuite/20_util/unique_ptr/assign/cv_qual.cc: Likewise. * testsuite/20_util/unique_ptr/cons/cv_qual.cc: Likewise. * testsuite/20_util/unique_ptr/dr2228.cc: New. * testsuite/20_util/unique_ptr/modifiers/cv_qual.cc: Adjust.
Re: [PATCH] gcc/: Fix building with isl-0.15.0; includes
On 07/19/2015 11:10 PM, james harvey wrote: These two patches appear to do the trick. I'm building with ISL 0.15 just fine now. Nice. Tobias
[PATCH] config/bfin/bfin.c (hwloop_optimize): Recognize direct jump case and emit lsetup at loop head.
2015-07-20 Chen Gang
* config/bfin/bfin.c (hwloop_optimize): Recognize direct jump
case and emit lsetup at loop head.
---
gcc/config/bfin/bfin.c | 12 +---
1 file changed, 9 insertions(+), 3 deletions(-)
diff --git a/gcc/config/bfin/bfin.c b/gcc/config/bfin/bfin.c
index a131053..1c70040 100644
--- a/gcc/config/bfin/bfin.c
+++ b/gcc/config/bfin/bfin.c
@@ -3456,7 +3456,7 @@ hwloop_optimize (hwloop_info loop)
rtx seq_end;
rtx_insn *seq;
int length;
- bool clobber0, clobber1;
+ bool clobber0, clobber1, direct_jmp = false;
if (loop->depth > MAX_LOOP_DEPTH)
{
@@ -3519,7 +3519,13 @@ hwloop_optimize (hwloop_info loop)
|| !(loop->incoming->last ()->flags & EDGE_FALLTHRU))
{
gcc_assert (JUMP_P (insn));
- insn = PREV_INSN (insn);
+ if (JUMP_LABEL (insn) != loop->start_label)
+ insn = PREV_INSN (insn);
+ else
+ {
+ direct_jmp = true;
+ insn = loop->start_label;
+ }
}
for (; insn && insn != loop->start_label; insn = NEXT_INSN (insn))
@@ -3783,7 +3789,7 @@ hwloop_optimize (hwloop_info loop)
seq = get_insns ();
end_sequence ();
- if (loop->incoming_src)
+ if (loop->incoming_src && !direct_jmp)
{
rtx_insn *prev = BB_END (loop->incoming_src);
if (vec_safe_length (loop->incoming) > 1
--
1.9.3
Re: [PATCH 1/2] Allow REG_EQUAL for ZERO_EXTRACT
I have made a mistake while addressing the review comments for this
patch. Unfortunately, It was not detected in my earlier testing. My
sincere graphology for the mistake.
I have basically missed the STRICT_LOW_PART check for the first if-check
thus the second part (which is the ZERO_EXTRACT part) will never get
executed. Attached patch fixes this along with some minor changes.
Bootstrapped and regression tested on arm-none-linux (Chromebook) and
x86-64-linux-gnu with no new regression along with the ARM ennoblement
patch.
Also did a complete arm qemu regression testing with Chriophe's scripts
with no new regression.
(http://people.linaro.org/~christophe.lyon/cross-validation/gcc-test-patches/225987-reg4/report-build-info.html)
Is this OK for trunk,
Thanks,
Kugan
gcc/ChangeLog:
2015-07-20 Kugan Vivekanandarajah
* cse.c (cse_insn): Fix missing check for STRICT_LOW_PART and minor
clean up.
diff --git a/gcc/cse.c b/gcc/cse.c
index 1c14d83..96adf18 100644
--- a/gcc/cse.c
+++ b/gcc/cse.c
@@ -4529,10 +4529,10 @@ cse_insn (rtx_insn *insn)
this case, and if it isn't set, then there will be no equivalence
for the destination. */
if (n_sets == 1 && REG_NOTES (insn) != 0
- && (tem = find_reg_note (insn, REG_EQUAL, NULL_RTX)) != 0)
+ && (tem = find_reg_note (insn, REG_EQUAL, NULL_RTX)) != 0
+ && (! rtx_equal_p (XEXP (tem, 0), SET_SRC (sets[0].rtl
{
- if ((! rtx_equal_p (XEXP (tem, 0), SET_SRC (sets[0].rtl)))
- || GET_CODE (SET_DEST (sets[0].rtl)) == STRICT_LOW_PART)
+ if (GET_CODE (SET_DEST (sets[0].rtl)) == STRICT_LOW_PART)
src_eqv = copy_rtx (XEXP (tem, 0));
/* If DEST is of the form ZERO_EXTACT, as in:
@@ -4544,14 +4544,14 @@ cse_insn (rtx_insn *insn)
point. Note that this is different from SRC_EQV. We can however
calculate SRC_EQV with the position and width of ZERO_EXTRACT. */
else if (GET_CODE (SET_DEST (sets[0].rtl)) == ZERO_EXTRACT
- && CONST_INT_P (src_eqv)
+ && CONST_INT_P (XEXP (tem, 0))
&& CONST_INT_P (XEXP (SET_DEST (sets[0].rtl), 1))
&& CONST_INT_P (XEXP (SET_DEST (sets[0].rtl), 2)))
{
rtx dest_reg = XEXP (SET_DEST (sets[0].rtl), 0);
rtx width = XEXP (SET_DEST (sets[0].rtl), 1);
rtx pos = XEXP (SET_DEST (sets[0].rtl), 2);
- HOST_WIDE_INT val = INTVAL (src_eqv);
+ HOST_WIDE_INT val = INTVAL (XEXP (tem, 0));
HOST_WIDE_INT mask;
unsigned int shift;
if (BITS_BIG_ENDIAN)
[PATCH] libierty: change _ANSIDECL_H to ANSIDECL_H
ansidecl.h is using _ANSIDECL_H as it's guard #ifdef. But names that start with
underscore are reserved, and the latest version of Xcode seems to be exercising
its prerogative to define _ANSIDECL_H.
This patch changes _ANSIDECL_H to ANSIDECL_H so Xcode can still be used to
bootstrap gcc.
Signed-off-by: Lawrence D'Anna
---
include/ansidecl.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/include/ansidecl.h b/include/ansidecl.h
index 224627d..0d6cd89 100644
--- a/include/ansidecl.h
+++ b/include/ansidecl.h
@@ -33,8 +33,8 @@ Foundation, Inc., 51 Franklin Street - Fifth Floor, Boston,
MA 02110-1301, USA.
wrappers around __attribute__. Also, __extension__ will be #defined
to nothing if it doesn't work. See below. */
-#ifndef_ANSIDECL_H
-#define _ANSIDECL_H1
+#ifndefANSIDECL_H
+#define ANSIDECL_H 1
#ifdef __cplusplus
extern "C" {
--
2.3.8 (Apple Git-58)
