From: "H.J. Lu" <hjl.tools@gmail.com>
To: gcc-patches@gcc.gnu.org
Cc: Uros Bizjak <ubizjak@gmail.com>
Subject: [PATCH] PR target/69225: gcc uses double precision instead of single float with -m32 -std=c99 -msoft-float
Date: Mon, 11 Jan 2016 14:52:00 -0000 [thread overview]
Message-ID: <20160111145237.GA7821@gmail.com> (raw)
When FPU isn't used, there is no excess precision. We should set
FLT_EVAL_METHOD to 0 if 387 is disabled.
OK for trunk?
Thanks.
H.J.
---
gcc/
PR target/69225
* config/i386/i386.h (TARGET_FLT_EVAL_METHOD): Set to 0 if
TARGET_80387 is false.
gcc/testsuite
PR target/69225
* gcc.target/i386/pr69225-1.c: New test.
* gcc.target/i386/pr69225-2.c: Likewise.
* gcc.target/i386/pr69225-3.c: Likewise.
* gcc.target/i386/pr69225-4.c: Likewise.
* gcc.target/i386/pr69225-5.c: Likewise.
* gcc.target/i386/pr69225-6.c: Likewise.
---
gcc/config/i386/i386.h | 4 ++--
gcc/testsuite/gcc.target/i386/pr69225-1.c | 8 ++++++++
gcc/testsuite/gcc.target/i386/pr69225-2.c | 8 ++++++++
gcc/testsuite/gcc.target/i386/pr69225-3.c | 8 ++++++++
gcc/testsuite/gcc.target/i386/pr69225-4.c | 8 ++++++++
gcc/testsuite/gcc.target/i386/pr69225-5.c | 8 ++++++++
gcc/testsuite/gcc.target/i386/pr69225-6.c | 8 ++++++++
7 files changed, 50 insertions(+), 2 deletions(-)
create mode 100644 gcc/testsuite/gcc.target/i386/pr69225-1.c
create mode 100644 gcc/testsuite/gcc.target/i386/pr69225-2.c
create mode 100644 gcc/testsuite/gcc.target/i386/pr69225-3.c
create mode 100644 gcc/testsuite/gcc.target/i386/pr69225-4.c
create mode 100644 gcc/testsuite/gcc.target/i386/pr69225-5.c
create mode 100644 gcc/testsuite/gcc.target/i386/pr69225-6.c
diff --git a/gcc/config/i386/i386.h b/gcc/config/i386/i386.h
index dcaa011..5c97fd1 100644
--- a/gcc/config/i386/i386.h
+++ b/gcc/config/i386/i386.h
@@ -692,9 +692,9 @@ extern const char *host_detect_local_cpu (int argc, const char **argv);
FPU, assume that the fpcw is set to extended precision; when using
only SSE, rounding is correct; when using both SSE and the FPU,
the rounding precision is indeterminate, since either may be chosen
- apparently at random. */
+ apparently at random. Also set to 0 if FPU isn't used. */
#define TARGET_FLT_EVAL_METHOD \
- (TARGET_MIX_SSE_I387 ? -1 : TARGET_SSE_MATH ? 0 : 2)
+ (TARGET_MIX_SSE_I387 ? -1 : (TARGET_SSE_MATH || !TARGET_80387) ? 0 : 2)
/* Whether to allow x87 floating-point arithmetic on MODE (one of
SFmode, DFmode and XFmode) in the current excess precision
diff --git a/gcc/testsuite/gcc.target/i386/pr69225-1.c b/gcc/testsuite/gcc.target/i386/pr69225-1.c
new file mode 100644
index 0000000..cc5b782
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/pr69225-1.c
@@ -0,0 +1,8 @@
+/* { dg-do compile} */
+/* { dg-options "-msse2 -mfpmath=sse" } */
+
+#include <float.h>
+
+#if FLT_EVAL_METHOD != 0
+# error FLT_EVAL_METHOD != 0
+#endif
diff --git a/gcc/testsuite/gcc.target/i386/pr69225-2.c b/gcc/testsuite/gcc.target/i386/pr69225-2.c
new file mode 100644
index 0000000..cd0a626
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/pr69225-2.c
@@ -0,0 +1,8 @@
+/* { dg-do compile} */
+/* { dg-options "-mno-sse -mno-80387" } */
+
+#include <float.h>
+
+#if FLT_EVAL_METHOD != 0
+# error FLT_EVAL_METHOD != 0
+#endif
diff --git a/gcc/testsuite/gcc.target/i386/pr69225-3.c b/gcc/testsuite/gcc.target/i386/pr69225-3.c
new file mode 100644
index 0000000..1440b74
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/pr69225-3.c
@@ -0,0 +1,8 @@
+/* { dg-do compile} */
+/* { dg-options "-m80387 -mfpmath=387" } */
+
+#include <float.h>
+
+#if FLT_EVAL_METHOD != 2
+# error FLT_EVAL_METHOD != 2
+#endif
diff --git a/gcc/testsuite/gcc.target/i386/pr69225-4.c b/gcc/testsuite/gcc.target/i386/pr69225-4.c
new file mode 100644
index 0000000..b032a5f
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/pr69225-4.c
@@ -0,0 +1,8 @@
+/* { dg-do compile} */
+/* { dg-options "-msse2 -mfancy-math-387 -mfpmath=sse" } */
+
+#include <float.h>
+
+#if FLT_EVAL_METHOD != 0
+# error FLT_EVAL_METHOD != 0
+#endif
diff --git a/gcc/testsuite/gcc.target/i386/pr69225-5.c b/gcc/testsuite/gcc.target/i386/pr69225-5.c
new file mode 100644
index 0000000..4162f77
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/pr69225-5.c
@@ -0,0 +1,8 @@
+/* { dg-do compile} */
+/* { dg-options "-msse2 -m80387 -mfpmath=sse,387" } */
+
+#include <float.h>
+
+#if FLT_EVAL_METHOD != -1
+# error FLT_EVAL_METHOD != -1
+#endif
diff --git a/gcc/testsuite/gcc.target/i386/pr69225-6.c b/gcc/testsuite/gcc.target/i386/pr69225-6.c
new file mode 100644
index 0000000..cd5f419
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/pr69225-6.c
@@ -0,0 +1,8 @@
+/* { dg-do compile} */
+/* { dg-options "-mno-80387" } */
+
+#include <float.h>
+
+#if FLT_EVAL_METHOD != 0
+# error FLT_EVAL_METHOD != 0
+#endif
--
2.5.0
next reply other threads:[~2016-01-11 14:52 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-01-11 14:52 H.J. Lu [this message]
2016-01-11 16:15 ` Uros Bizjak
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=20160111145237.GA7821@gmail.com \
--to=hjl.tools@gmail.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=ubizjak@gmail.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).