This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
soft-fp: Support floating-point extensions without quieting sNaNs
- From: Joseph Myers <joseph at codesourcery dot com>
- To: <libc-alpha at sourceware dot org>
- Date: Thu, 5 Feb 2015 21:18:29 +0000
- Subject: soft-fp: Support floating-point extensions without quieting sNaNs
- Authentication-results: sourceware.org; auth=none
One special case needed in soft-fp to replace the old version in the
Linux kernel is extending from a narrower floating-point format to a
wider one without quieting signaling NaNs. (This is for
arch/powerpc/math-emu/lfs.c, where previously it used the old FP_CONV
which didn't do anything special for NaNs, then handled packing
specially for NaNs to avoid quieting at packing time, and discarded
the exceptions from unpacking.)
This patch accordingly refactors FP_EXTEND, creating a separate
_FP_EXTEND_CNAN that offers a choice of how NaNs are handled, with
FP_EXTEND reimplemented as a wrapper that provides the common case of
the IEEE operation that does quiet signaling NaNs and raise exceptions
for them.
Tested for powerpc (e500) that installed stripped shared libraries are
unchanged by this patch. Will commit once master reopens for 2.22
development.
2015-02-05 Joseph Myers <joseph@codesourcery.com>
* soft-fp/op-common.h (FP_EXTEND): Rename to _FP_EXTEND_CNAN with
extra argument CHECK_NAN. Redefine as wrapper around
_FP_EXTEND_CNAN.
diff --git a/soft-fp/op-common.h b/soft-fp/op-common.h
index 6b8b52f..98cc721 100644
--- a/soft-fp/op-common.h
+++ b/soft-fp/op-common.h
@@ -1825,8 +1825,10 @@
/* Extend from a narrower floating-point format to a wider one. Input
- and output are raw. */
-#define FP_EXTEND(dfs, sfs, dwc, swc, D, S) \
+ and output are raw. If CHECK_NAN, then signaling NaNs are
+ converted to quiet with the "invalid" exception raised; otherwise
+ signaling NaNs remain signaling with no exception. */
+#define _FP_EXTEND_CNAN(dfs, sfs, dwc, swc, D, S, check_nan) \
do \
{ \
if (_FP_FRACBITS_##dfs < _FP_FRACBITS_##sfs \
@@ -1876,18 +1878,22 @@
D##_e = _FP_EXPMAX_##dfs; \
if (!_FP_FRAC_ZEROP_##swc (S)) \
{ \
- if (_FP_FRAC_SNANP (sfs, S)) \
+ if (check_nan && _FP_FRAC_SNANP (sfs, S)) \
FP_SET_EXCEPTION (FP_EX_INVALID \
| FP_EX_INVALID_SNAN); \
_FP_FRAC_SLL_##dwc (D, (_FP_FRACBITS_##dfs \
- _FP_FRACBITS_##sfs)); \
- _FP_SETQNAN (dfs, dwc, D); \
+ if (check_nan) \
+ _FP_SETQNAN (dfs, dwc, D); \
} \
} \
} \
} \
while (0)
+#define FP_EXTEND(dfs, sfs, dwc, swc, D, S) \
+ _FP_EXTEND_CNAN (dfs, sfs, dwc, swc, D, S, 1)
+
/* Truncate from a wider floating-point format to a narrower one.
Input and output are semi-raw. */
#define FP_TRUNC(dfs, sfs, dwc, swc, D, S) \
--
Joseph S. Myers
joseph@codesourcery.com