From: Patrick Marlier <patrick.marlier@gmail.com>
To: gcc-patches@gcc.gnu.org
Cc: Richard Henderson <rth@redhat.com>
Subject: [PATCH] PR53992 - openmp lower transaction code
Date: Mon, 20 Aug 2012 14:20:00 -0000 [thread overview]
Message-ID: <503247B1.5020809@gmail.com> (raw)
[-- Attachment #1: Type: text/plain, Size: 541 bytes --]
In this PR, OMP lowering is not going into the transaction code.
So if GIMPLE_TRANSACTION is found, we lower its body.
(Patch also fixes a format issue.)
Note that PR53992 component in Bugzilla must be change from c to libgomp
(I don't have bugzilla account with admin rights, who should I ask for
that?).
Tested on trunk / i686.
Ok for trunk? Ok to backport to 4.7 branch if no regression?
Thanks.
gcc/
2012-08-17 Patrick Marlier <patrick.marlier@gmail.com>
PR libgomp/53992
* omp-low.c (lower_omp_1): Handle GIMPLE_TRANSACTION.
[-- Attachment #2: pr53992.patch --]
[-- Type: text/x-patch, Size: 2265 bytes --]
Index: omp-low.c
===================================================================
--- omp-low.c (revision 190488)
+++ omp-low.c (working copy)
@@ -6827,6 +6827,9 @@ lower_omp_1 (gimple_stmt_iterator *gsi_p, omp_cont
lower_omp (gimple_try_eval_ptr (stmt), ctx);
lower_omp (gimple_try_cleanup_ptr (stmt), ctx);
break;
+ case GIMPLE_TRANSACTION:
+ lower_omp (gimple_transaction_body_ptr (stmt), ctx);
+ break;
case GIMPLE_BIND:
lower_omp (gimple_bind_body_ptr (stmt), ctx);
break;
@@ -7108,24 +7111,24 @@ diagnose_sb_2 (gimple_stmt_iterator *gsi_p, bool *
break;
case GIMPLE_COND:
- {
- tree lab = gimple_cond_true_label (stmt);
- if (lab)
- {
- n = splay_tree_lookup (all_labels,
- (splay_tree_key) lab);
- diagnose_sb_0 (gsi_p, context,
- n ? (gimple) n->value : NULL);
- }
- lab = gimple_cond_false_label (stmt);
- if (lab)
- {
- n = splay_tree_lookup (all_labels,
- (splay_tree_key) lab);
- diagnose_sb_0 (gsi_p, context,
- n ? (gimple) n->value : NULL);
- }
- }
+ {
+ tree lab = gimple_cond_true_label (stmt);
+ if (lab)
+ {
+ n = splay_tree_lookup (all_labels,
+ (splay_tree_key) lab);
+ diagnose_sb_0 (gsi_p, context,
+ n ? (gimple) n->value : NULL);
+ }
+ lab = gimple_cond_false_label (stmt);
+ if (lab)
+ {
+ n = splay_tree_lookup (all_labels,
+ (splay_tree_key) lab);
+ diagnose_sb_0 (gsi_p, context,
+ n ? (gimple) n->value : NULL);
+ }
+ }
break;
case GIMPLE_GOTO:
Index: testsuite/gcc.dg/gomp/pr53992.c
===================================================================
--- testsuite/gcc.dg/gomp/pr53992.c (revision 0)
+++ testsuite/gcc.dg/gomp/pr53992.c (working copy)
@@ -0,0 +1,20 @@
+/* { dg-do compile } */
+/* { dg-options "-fgnu-tm -fopenmp" } */
+/* { dg-require-effective-target fgnu_tm } */
+
+int main() {
+ long data[10000];
+ long i, min=10000;
+ for (i=0; i<10000; i++) data[i] = -i;
+
+#pragma omp parallel for
+ for (i=0; i<10000; i++) {
+ __transaction_atomic
+ {
+ if (data[i] < min)
+ min = data[i];
+ }
+ }
+
+ return !(min == -9999);
+}
next reply other threads:[~2012-08-20 14:20 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-08-20 14:20 Patrick Marlier [this message]
2012-08-20 14:33 ` Richard Henderson
2012-08-20 14:34 ` Jakub Jelinek
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=503247B1.5020809@gmail.com \
--to=patrick.marlier@gmail.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=rth@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).