[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <1432832996-12129-2-git-send-email-hofrat@osadl.org>
Date: Thu, 28 May 2015 19:09:56 +0200
From: Nicholas Mc Guire <hofrat@...dl.org>
To: Michal Marek <mmarek@...e.cz>
Cc: Masahiro Yamada <yamada.m@...panasonic.com>,
Sam Ravnborg <sam@...nborg.org>,
Thomas Gleixner <tglx@...utronix.de>,
"H. Peter Alvin" <hpa@...or.com>, Joe Perches <joe@...ches.com>,
John Stultz <john.stultz@...aro.org>,
Andrew Hunter <ahh@...gle.com>, Paul Turner <pjt@...gle.com>,
linux-kernel@...r.kernel.org, Nicholas Mc Guire <hofrat@...dl.org>
Subject: [PATCH 2/2] time: allow gcc to fold constants when possible
To allow constant folding in usecs_to_jiffies() conditionally calls
the HZ dependent _usecs_to_jiffies() helpers or, when gcc can not
figure out constant folding, __usecs_to_jiffies, which is the renamed
original usecs_to_jiffies() function.
Patch was build tested for:
arm - imx_v6_v7_defconfig, vexpress_defconfig
cris - etrax-100lx_v2_defconfig
ia64 - generic_defconfig
(though with lots of build warnings both with and without patch)
m68k - multi_defconfig
mips32 - ar7_defconfig,
mips64 - loongson3_defconfig
powerpc - 44x/virtex5_defconfig
powerpc64 - ppc64_defconfig
s390 - failed to build for all defconfigs with toolchain from kernel.org
(cc1: error: unrecognized command line option '-mtune=zEC12')
sh - se7780_defconfig,
sparc32 - sparc32_defconfig
sparc64 - sparc64_defconfig
x86 64 - x86_64_defconfig,
patch is against 4.1-rc5 (localversion-next is -next-20150527)
Link: http://lkml.org/lkml/2015/5/18/274
Signed-off-by: Nicholas Mc Guire <hofrat@...dl.org>
---
Thanks to Joe Perches <joe@...ches.com> for suggesting this approach.
and helping with the initial msecs_to_jiffies cleanup patch.
Verification of const folding:
config: x86_64_defconfig + CONFIG_DRM_GMA500=m, CONFIG_DRM_GMA3600=y
drivers/gpu/drm/gma500/intel_gmbus.c
204 gpio->algo.timeout = usecs_to_jiffies(2200);
-> reduced to load instruction
config: x86_64_defconfig + CONFIG_INFINIBAND=m, CONFIG_INFINIBAND_QIB=m
drivers/infiniband/hw/qib/qib_qp.c
809 qp->timeout = attr->timeout;
810 qp->timeout_jiffies =
811 usecs_to_jiffies((4096UL * (1UL << qp->timeout)) /
812 1000UL);
-> call __usecs_to_jiffies
include/linux/jiffies.h | 30 +++++++++++++++++++++++++++++-
1 file changed, 29 insertions(+), 1 deletion(-)
diff --git a/include/linux/jiffies.h b/include/linux/jiffies.h
index cf6eaaf..c0aa150 100644
--- a/include/linux/jiffies.h
+++ b/include/linux/jiffies.h
@@ -383,9 +383,37 @@ static inline unsigned long _usecs_to_jiffies(const unsigned int u)
}
#endif
+/**
+ * usecs_to_jiffies: - convert microseconds to jiffies
+ * @u: time in microseconds
+ *
+ * conversion is done as follows:
+ *
+ * - 'too large' values [that would result in larger than
+ * MAX_JIFFY_OFFSET values] mean 'infinite timeout' too.
+ *
+ * - all other values are converted to jiffies by either multiplying
+ * the input value by a factor or dividing it with a factor and
+ * handling any 32-bit overflows as for msecs_to_jiffies.
+ *
+ * usecs_to_jiffies() checks for the passed in value being a constant
+ * via __builtin_constant_p() allowing gcc to eliminate most of the
+ * code, __usecs_to_jiffies() is called if the value passed does not
+ * allow constant folding and the actual conversion must be done at
+ * runtime.
+ * the HZ range specific helpers _usecs_to_jiffies() are called both
+ * directly here and from __msecs_to_jiffies() in the case where
+ * constant folding is not possible.
+ */
static inline unsigned long usecs_to_jiffies(const unsigned int u)
{
- return __usecs_to_jiffies(u);
+ if (__builtin_constant_p(u)) {
+ if (u > jiffies_to_usecs(MAX_JIFFY_OFFSET))
+ return MAX_JIFFY_OFFSET;
+ return _usecs_to_jiffies(u);
+ } else {
+ return __usecs_to_jiffies(u);
+ }
}
extern unsigned long timespec_to_jiffies(const struct timespec *value);
--
1.7.10.4
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists