Revision 3ac805d2afd3fa4a07cb5bcf352fd7fa83f28935 authored by Ingo Molnar on 03 July 2009, 10:51:19 UTC, committed by Ingo Molnar on 03 July 2009, 11:26:43 UTC
cmpxchg8b is a huge instruction in terms of register footprint,
we almost never want to inline it, not even within the same
code module.

GCC 4.3 still messes up for two functions, under-judging the
true cost of this instruction - so annotate two key functions
to reduce the bloat:

arch/x86/lib/atomic64_32.o:

   text	   data	    bss	    dec	    hex	filename
   1763	      0	      0	   1763	    6e3	atomic64_32.o.before
    435	      0	      0	    435	    1b3	atomic64_32.o.after

Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Arnd Bergmann <arnd@arndb.de>
LKML-Reference: <alpine.LFD.2.01.0907021653030.3210@localhost.localdomain>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
1 parent 824975e
Raw File
raid6int.uc
/* -*- linux-c -*- ------------------------------------------------------- *
 *
 *   Copyright 2002-2004 H. Peter Anvin - All Rights Reserved
 *
 *   This program is free software; you can redistribute it and/or modify
 *   it under the terms of the GNU General Public License as published by
 *   the Free Software Foundation, Inc., 53 Temple Place Ste 330,
 *   Boston MA 02111-1307, USA; either version 2 of the License, or
 *   (at your option) any later version; incorporated herein by reference.
 *
 * ----------------------------------------------------------------------- */

/*
 * raid6int$#.c
 *
 * $#-way unrolled portable integer math RAID-6 instruction set
 *
 * This file is postprocessed using unroll.pl
 */

#include <linux/raid/pq.h>

/*
 * This is the C data type to use
 */

/* Change this from BITS_PER_LONG if there is something better... */
#if BITS_PER_LONG == 64
# define NBYTES(x) ((x) * 0x0101010101010101UL)
# define NSIZE  8
# define NSHIFT 3
# define NSTRING "64"
typedef u64 unative_t;
#else
# define NBYTES(x) ((x) * 0x01010101U)
# define NSIZE  4
# define NSHIFT 2
# define NSTRING "32"
typedef u32 unative_t;
#endif



/*
 * IA-64 wants insane amounts of unrolling.  On other architectures that
 * is just a waste of space.
 */
#if ($# <= 8) || defined(__ia64__)


/*
 * These sub-operations are separate inlines since they can sometimes be
 * specially optimized using architecture-specific hacks.
 */

/*
 * The SHLBYTE() operation shifts each byte left by 1, *not*
 * rolling over into the next byte
 */
static inline __attribute_const__ unative_t SHLBYTE(unative_t v)
{
	unative_t vv;

	vv = (v << 1) & NBYTES(0xfe);
	return vv;
}

/*
 * The MASK() operation returns 0xFF in any byte for which the high
 * bit is 1, 0x00 for any byte for which the high bit is 0.
 */
static inline __attribute_const__ unative_t MASK(unative_t v)
{
	unative_t vv;

	vv = v & NBYTES(0x80);
	vv = (vv << 1) - (vv >> 7); /* Overflow on the top bit is OK */
	return vv;
}


static void raid6_int$#_gen_syndrome(int disks, size_t bytes, void **ptrs)
{
	u8 **dptr = (u8 **)ptrs;
	u8 *p, *q;
	int d, z, z0;

	unative_t wd$$, wq$$, wp$$, w1$$, w2$$;

	z0 = disks - 3;		/* Highest data disk */
	p = dptr[z0+1];		/* XOR parity */
	q = dptr[z0+2];		/* RS syndrome */

	for ( d = 0 ; d < bytes ; d += NSIZE*$# ) {
		wq$$ = wp$$ = *(unative_t *)&dptr[z0][d+$$*NSIZE];
		for ( z = z0-1 ; z >= 0 ; z-- ) {
			wd$$ = *(unative_t *)&dptr[z][d+$$*NSIZE];
			wp$$ ^= wd$$;
			w2$$ = MASK(wq$$);
			w1$$ = SHLBYTE(wq$$);
			w2$$ &= NBYTES(0x1d);
			w1$$ ^= w2$$;
			wq$$ = w1$$ ^ wd$$;
		}
		*(unative_t *)&p[d+NSIZE*$$] = wp$$;
		*(unative_t *)&q[d+NSIZE*$$] = wq$$;
	}
}

const struct raid6_calls raid6_intx$# = {
	raid6_int$#_gen_syndrome,
	NULL,		/* always valid */
	"int" NSTRING "x$#",
	0
};

#endif
back to top