1.1 --- a/.rootkeys	Tue Feb 22 23:31:53 2005 +0000
     1.2 +++ b/.rootkeys	Tue Feb 22 23:46:11 2005 +0000
     1.3 @@ -258,7 +258,6 @@ 40f5623bgzm_9vwxpzJswlAxg298Gg linux-2.6
     1.4  40f5623bVdKP7Dt7qm8twu3NcnGNbA linux-2.6.10-xen-sparse/include/asm-xen/asm-i386/system.h
     1.5  40f5623bc8LKPRO09wY5dGDnY_YCpw linux-2.6.10-xen-sparse/include/asm-xen/asm-i386/tlbflush.h
     1.6  41062ab7uFxnCq-KtPeAm-aV8CicgA linux-2.6.10-xen-sparse/include/asm-xen/asm-i386/vga.h
     1.7 -40f5623bxUbeGjkRrjDguCy_Gm8RLw linux-2.6.10-xen-sparse/include/asm-xen/asm-i386/xor.h
     1.8  41af4017PDMuSmMWtSRU5UC9Vylw5g linux-2.6.10-xen-sparse/include/asm-xen/balloon.h
     1.9  40f5623bYNP7tHE2zX6YQxp9Zq2utQ linux-2.6.10-xen-sparse/include/asm-xen/ctrl_if.h
    1.10  40f5623b3Eqs8pAc5WpPX8_jTzV2qw linux-2.6.10-xen-sparse/include/asm-xen/evtchn.h

     2.1 --- a/linux-2.6.10-xen-sparse/arch/xen/i386/kernel/traps.c	Tue Feb 22 23:31:53 2005 +0000
     2.2 +++ b/linux-2.6.10-xen-sparse/arch/xen/i386/kernel/traps.c	Tue Feb 22 23:46:11 2005 +0000
     2.3 @@ -910,7 +910,7 @@ asmlinkage void math_state_restore(struc
     2.4  	if ((regs.xcs & 2) == 0)
     2.5  		return;
     2.6  
     2.7 -	clts();		/* Allow maths ops (or we recurse) */
     2.8 +	/* NB. 'clts' is done for us by Xen during virtual trap. */
     2.9  	if (!tsk->used_math)
    2.10  		init_fpu(tsk);
    2.11  	restore_fpu(tsk);

     3.1 --- a/linux-2.6.10-xen-sparse/include/asm-xen/asm-i386/system.h	Tue Feb 22 23:31:53 2005 +0000
     3.2 +++ b/linux-2.6.10-xen-sparse/include/asm-xen/asm-i386/system.h	Tue Feb 22 23:46:11 2005 +0000
     3.3 @@ -106,8 +106,7 @@ static inline unsigned long _get_base(ch
     3.4  /*
     3.5   * Clear and set 'TS' bit respectively
     3.6   */
     3.7 -/* NB. 'clts' is done for us by Xen during virtual trap. */
     3.8 -#define clts() ((void)0)
     3.9 +#define clts() __asm__ __volatile__ ("clts")
    3.10  #define read_cr0() \
    3.11  	BUG();
    3.12  #define write_cr0(x) \

     4.1 --- a/linux-2.6.10-xen-sparse/include/asm-xen/asm-i386/xor.h	Tue Feb 22 23:31:53 2005 +0000
     4.2 +++ /dev/null	Thu Jan 01 00:00:00 1970 +0000
     4.3 @@ -1,884 +0,0 @@
     4.4 -/*
     4.5 - * include/asm-i386/xor.h
     4.6 - *
     4.7 - * Optimized RAID-5 checksumming functions for MMX and SSE.
     4.8 - *
     4.9 - * This program is free software; you can redistribute it and/or modify
    4.10 - * it under the terms of the GNU General Public License as published by
    4.11 - * the Free Software Foundation; either version 2, or (at your option)
    4.12 - * any later version.
    4.13 - *
    4.14 - * You should have received a copy of the GNU General Public License
    4.15 - * (for example /usr/src/linux/COPYING); if not, write to the Free
    4.16 - * Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
    4.17 - */
    4.18 -
    4.19 -/*
    4.20 - * High-speed RAID5 checksumming functions utilizing MMX instructions.
    4.21 - * Copyright (C) 1998 Ingo Molnar.
    4.22 - */
    4.23 -
    4.24 -#define LD(x,y)		"       movq   8*("#x")(%1), %%mm"#y"   ;\n"
    4.25 -#define ST(x,y)		"       movq %%mm"#y",   8*("#x")(%1)   ;\n"
    4.26 -#define XO1(x,y)	"       pxor   8*("#x")(%2), %%mm"#y"   ;\n"
    4.27 -#define XO2(x,y)	"       pxor   8*("#x")(%3), %%mm"#y"   ;\n"
    4.28 -#define XO3(x,y)	"       pxor   8*("#x")(%4), %%mm"#y"   ;\n"
    4.29 -#define XO4(x,y)	"       pxor   8*("#x")(%5), %%mm"#y"   ;\n"
    4.30 -
    4.31 -#include <asm/i387.h>
    4.32 -
    4.33 -static void
    4.34 -xor_pII_mmx_2(unsigned long bytes, unsigned long *p1, unsigned long *p2)
    4.35 -{
    4.36 -	unsigned long lines = bytes >> 7;
    4.37 -
    4.38 -	kernel_fpu_begin();
    4.39 -
    4.40 -	__asm__ __volatile__ (
    4.41 -#undef BLOCK
    4.42 -#define BLOCK(i) \
    4.43 -	LD(i,0)					\
    4.44 -		LD(i+1,1)			\
    4.45 -			LD(i+2,2)		\
    4.46 -				LD(i+3,3)	\
    4.47 -	XO1(i,0)				\
    4.48 -	ST(i,0)					\
    4.49 -		XO1(i+1,1)			\
    4.50 -		ST(i+1,1)			\
    4.51 -			XO1(i+2,2)		\
    4.52 -			ST(i+2,2)		\
    4.53 -				XO1(i+3,3)	\
    4.54 -				ST(i+3,3)
    4.55 -
    4.56 -	" .align 32			;\n"
    4.57 -  	" 1:                            ;\n"
    4.58 -
    4.59 -	BLOCK(0)
    4.60 -	BLOCK(4)
    4.61 -	BLOCK(8)
    4.62 -	BLOCK(12)
    4.63 -
    4.64 -	"       addl $128, %1         ;\n"
    4.65 -	"       addl $128, %2         ;\n"
    4.66 -	"       decl %0               ;\n"
    4.67 -	"       jnz 1b                ;\n"
    4.68 -	: "+r" (lines),
    4.69 -	  "+r" (p1), "+r" (p2)
    4.70 -	:
    4.71 -	: "memory");
    4.72 -
    4.73 -	kernel_fpu_end();
    4.74 -}
    4.75 -
    4.76 -static void
    4.77 -xor_pII_mmx_3(unsigned long bytes, unsigned long *p1, unsigned long *p2,
    4.78 -	      unsigned long *p3)
    4.79 -{
    4.80 -	unsigned long lines = bytes >> 7;
    4.81 -
    4.82 -	kernel_fpu_begin();
    4.83 -
    4.84 -	__asm__ __volatile__ (
    4.85 -#undef BLOCK
    4.86 -#define BLOCK(i) \
    4.87 -	LD(i,0)					\
    4.88 -		LD(i+1,1)			\
    4.89 -			LD(i+2,2)		\
    4.90 -				LD(i+3,3)	\
    4.91 -	XO1(i,0)				\
    4.92 -		XO1(i+1,1)			\
    4.93 -			XO1(i+2,2)		\
    4.94 -				XO1(i+3,3)	\
    4.95 -	XO2(i,0)				\
    4.96 -	ST(i,0)					\
    4.97 -		XO2(i+1,1)			\
    4.98 -		ST(i+1,1)			\
    4.99 -			XO2(i+2,2)		\
   4.100 -			ST(i+2,2)		\
   4.101 -				XO2(i+3,3)	\
   4.102 -				ST(i+3,3)
   4.103 -
   4.104 -	" .align 32			;\n"
   4.105 -	" 1:                            ;\n"
   4.106 -
   4.107 -	BLOCK(0)
   4.108 -	BLOCK(4)
   4.109 -	BLOCK(8)
   4.110 -	BLOCK(12)
   4.111 -
   4.112 -	"       addl $128, %1         ;\n"
   4.113 -	"       addl $128, %2         ;\n"
   4.114 -	"       addl $128, %3         ;\n"
   4.115 -	"       decl %0               ;\n"
   4.116 -	"       jnz 1b                ;\n"
   4.117 -	: "+r" (lines),
   4.118 -	  "+r" (p1), "+r" (p2), "+r" (p3)
   4.119 -	:
   4.120 -	: "memory");
   4.121 -
   4.122 -	kernel_fpu_end();
   4.123 -}
   4.124 -
   4.125 -static void
   4.126 -xor_pII_mmx_4(unsigned long bytes, unsigned long *p1, unsigned long *p2,
   4.127 -	      unsigned long *p3, unsigned long *p4)
   4.128 -{
   4.129 -	unsigned long lines = bytes >> 7;
   4.130 -
   4.131 -	kernel_fpu_begin();
   4.132 -
   4.133 -	__asm__ __volatile__ (
   4.134 -#undef BLOCK
   4.135 -#define BLOCK(i) \
   4.136 -	LD(i,0)					\
   4.137 -		LD(i+1,1)			\
   4.138 -			LD(i+2,2)		\
   4.139 -				LD(i+3,3)	\
   4.140 -	XO1(i,0)				\
   4.141 -		XO1(i+1,1)			\
   4.142 -			XO1(i+2,2)		\
   4.143 -				XO1(i+3,3)	\
   4.144 -	XO2(i,0)				\
   4.145 -		XO2(i+1,1)			\
   4.146 -			XO2(i+2,2)		\
   4.147 -				XO2(i+3,3)	\
   4.148 -	XO3(i,0)				\
   4.149 -	ST(i,0)					\
   4.150 -		XO3(i+1,1)			\
   4.151 -		ST(i+1,1)			\
   4.152 -			XO3(i+2,2)		\
   4.153 -			ST(i+2,2)		\
   4.154 -				XO3(i+3,3)	\
   4.155 -				ST(i+3,3)
   4.156 -
   4.157 -	" .align 32			;\n"
   4.158 -	" 1:                            ;\n"
   4.159 -
   4.160 -	BLOCK(0)
   4.161 -	BLOCK(4)
   4.162 -	BLOCK(8)
   4.163 -	BLOCK(12)
   4.164 -
   4.165 -	"       addl $128, %1         ;\n"
   4.166 -	"       addl $128, %2         ;\n"
   4.167 -	"       addl $128, %3         ;\n"
   4.168 -	"       addl $128, %4         ;\n"
   4.169 -	"       decl %0               ;\n"
   4.170 -	"       jnz 1b                ;\n"
   4.171 -	: "+r" (lines),
   4.172 -	  "+r" (p1), "+r" (p2), "+r" (p3), "+r" (p4)
   4.173 -	:
   4.174 -	: "memory");
   4.175 -
   4.176 -	kernel_fpu_end();
   4.177 -}
   4.178 -
   4.179 -
   4.180 -static void
   4.181 -xor_pII_mmx_5(unsigned long bytes, unsigned long *p1, unsigned long *p2,
   4.182 -	      unsigned long *p3, unsigned long *p4, unsigned long *p5)
   4.183 -{
   4.184 -	unsigned long lines = bytes >> 7;
   4.185 -
   4.186 -	kernel_fpu_begin();
   4.187 -
   4.188 -	/* Make sure GCC forgets anything it knows about p4 or p5,
   4.189 -	   such that it won't pass to the asm volatile below a
   4.190 -	   register that is shared with any other variable.  That's
   4.191 -	   because we modify p4 and p5 there, but we can't mark them
   4.192 -	   as read/write, otherwise we'd overflow the 10-asm-operands
   4.193 -	   limit of GCC < 3.1.  */
   4.194 -	__asm__ ("" : "+r" (p4), "+r" (p5));
   4.195 -
   4.196 -	__asm__ __volatile__ (
   4.197 -#undef BLOCK
   4.198 -#define BLOCK(i) \
   4.199 -	LD(i,0)					\
   4.200 -		LD(i+1,1)			\
   4.201 -			LD(i+2,2)		\
   4.202 -				LD(i+3,3)	\
   4.203 -	XO1(i,0)				\
   4.204 -		XO1(i+1,1)			\
   4.205 -			XO1(i+2,2)		\
   4.206 -				XO1(i+3,3)	\
   4.207 -	XO2(i,0)				\
   4.208 -		XO2(i+1,1)			\
   4.209 -			XO2(i+2,2)		\
   4.210 -				XO2(i+3,3)	\
   4.211 -	XO3(i,0)				\
   4.212 -		XO3(i+1,1)			\
   4.213 -			XO3(i+2,2)		\
   4.214 -				XO3(i+3,3)	\
   4.215 -	XO4(i,0)				\
   4.216 -	ST(i,0)					\
   4.217 -		XO4(i+1,1)			\
   4.218 -		ST(i+1,1)			\
   4.219 -			XO4(i+2,2)		\
   4.220 -			ST(i+2,2)		\
   4.221 -				XO4(i+3,3)	\
   4.222 -				ST(i+3,3)
   4.223 -
   4.224 -	" .align 32			;\n"
   4.225 -	" 1:                            ;\n"
   4.226 -
   4.227 -	BLOCK(0)
   4.228 -	BLOCK(4)
   4.229 -	BLOCK(8)
   4.230 -	BLOCK(12)
   4.231 -
   4.232 -	"       addl $128, %1         ;\n"
   4.233 -	"       addl $128, %2         ;\n"
   4.234 -	"       addl $128, %3         ;\n"
   4.235 -	"       addl $128, %4         ;\n"
   4.236 -	"       addl $128, %5         ;\n"
   4.237 -	"       decl %0               ;\n"
   4.238 -	"       jnz 1b                ;\n"
   4.239 -	: "+r" (lines),
   4.240 -	  "+r" (p1), "+r" (p2), "+r" (p3)
   4.241 -	: "r" (p4), "r" (p5) 
   4.242 -	: "memory");
   4.243 -
   4.244 -	/* p4 and p5 were modified, and now the variables are dead.
   4.245 -	   Clobber them just to be sure nobody does something stupid
   4.246 -	   like assuming they have some legal value.  */
   4.247 -	__asm__ ("" : "=r" (p4), "=r" (p5));
   4.248 -
   4.249 -	kernel_fpu_end();
   4.250 -}
   4.251 -
   4.252 -#undef LD
   4.253 -#undef XO1
   4.254 -#undef XO2
   4.255 -#undef XO3
   4.256 -#undef XO4
   4.257 -#undef ST
   4.258 -#undef BLOCK
   4.259 -
   4.260 -static void
   4.261 -xor_p5_mmx_2(unsigned long bytes, unsigned long *p1, unsigned long *p2)
   4.262 -{
   4.263 -	unsigned long lines = bytes >> 6;
   4.264 -
   4.265 -	kernel_fpu_begin();
   4.266 -
   4.267 -	__asm__ __volatile__ (
   4.268 -	" .align 32	             ;\n"
   4.269 -	" 1:                         ;\n"
   4.270 -	"       movq   (%1), %%mm0   ;\n"
   4.271 -	"       movq  8(%1), %%mm1   ;\n"
   4.272 -	"       pxor   (%2), %%mm0   ;\n"
   4.273 -	"       movq 16(%1), %%mm2   ;\n"
   4.274 -	"       movq %%mm0,   (%1)   ;\n"
   4.275 -	"       pxor  8(%2), %%mm1   ;\n"
   4.276 -	"       movq 24(%1), %%mm3   ;\n"
   4.277 -	"       movq %%mm1,  8(%1)   ;\n"
   4.278 -	"       pxor 16(%2), %%mm2   ;\n"
   4.279 -	"       movq 32(%1), %%mm4   ;\n"
   4.280 -	"       movq %%mm2, 16(%1)   ;\n"
   4.281 -	"       pxor 24(%2), %%mm3   ;\n"
   4.282 -	"       movq 40(%1), %%mm5   ;\n"
   4.283 -	"       movq %%mm3, 24(%1)   ;\n"
   4.284 -	"       pxor 32(%2), %%mm4   ;\n"
   4.285 -	"       movq 48(%1), %%mm6   ;\n"
   4.286 -	"       movq %%mm4, 32(%1)   ;\n"
   4.287 -	"       pxor 40(%2), %%mm5   ;\n"
   4.288 -	"       movq 56(%1), %%mm7   ;\n"
   4.289 -	"       movq %%mm5, 40(%1)   ;\n"
   4.290 -	"       pxor 48(%2), %%mm6   ;\n"
   4.291 -	"       pxor 56(%2), %%mm7   ;\n"
   4.292 -	"       movq %%mm6, 48(%1)   ;\n"
   4.293 -	"       movq %%mm7, 56(%1)   ;\n"
   4.294 -	
   4.295 -	"       addl $64, %1         ;\n"
   4.296 -	"       addl $64, %2         ;\n"
   4.297 -	"       decl %0              ;\n"
   4.298 -	"       jnz 1b               ;\n"
   4.299 -	: "+r" (lines),
   4.300 -	  "+r" (p1), "+r" (p2)
   4.301 -	:
   4.302 -	: "memory");
   4.303 -
   4.304 -	kernel_fpu_end();
   4.305 -}
   4.306 -
   4.307 -static void
   4.308 -xor_p5_mmx_3(unsigned long bytes, unsigned long *p1, unsigned long *p2,
   4.309 -	     unsigned long *p3)
   4.310 -{
   4.311 -	unsigned long lines = bytes >> 6;
   4.312 -
   4.313 -	kernel_fpu_begin();
   4.314 -
   4.315 -	__asm__ __volatile__ (
   4.316 -	" .align 32,0x90             ;\n"
   4.317 -	" 1:                         ;\n"
   4.318 -	"       movq   (%1), %%mm0   ;\n"
   4.319 -	"       movq  8(%1), %%mm1   ;\n"
   4.320 -	"       pxor   (%2), %%mm0   ;\n"
   4.321 -	"       movq 16(%1), %%mm2   ;\n"
   4.322 -	"       pxor  8(%2), %%mm1   ;\n"
   4.323 -	"       pxor   (%3), %%mm0   ;\n"
   4.324 -	"       pxor 16(%2), %%mm2   ;\n"
   4.325 -	"       movq %%mm0,   (%1)   ;\n"
   4.326 -	"       pxor  8(%3), %%mm1   ;\n"
   4.327 -	"       pxor 16(%3), %%mm2   ;\n"
   4.328 -	"       movq 24(%1), %%mm3   ;\n"
   4.329 -	"       movq %%mm1,  8(%1)   ;\n"
   4.330 -	"       movq 32(%1), %%mm4   ;\n"
   4.331 -	"       movq 40(%1), %%mm5   ;\n"
   4.332 -	"       pxor 24(%2), %%mm3   ;\n"
   4.333 -	"       movq %%mm2, 16(%1)   ;\n"
   4.334 -	"       pxor 32(%2), %%mm4   ;\n"
   4.335 -	"       pxor 24(%3), %%mm3   ;\n"
   4.336 -	"       pxor 40(%2), %%mm5   ;\n"
   4.337 -	"       movq %%mm3, 24(%1)   ;\n"
   4.338 -	"       pxor 32(%3), %%mm4   ;\n"
   4.339 -	"       pxor 40(%3), %%mm5   ;\n"
   4.340 -	"       movq 48(%1), %%mm6   ;\n"
   4.341 -	"       movq %%mm4, 32(%1)   ;\n"
   4.342 -	"       movq 56(%1), %%mm7   ;\n"
   4.343 -	"       pxor 48(%2), %%mm6   ;\n"
   4.344 -	"       movq %%mm5, 40(%1)   ;\n"
   4.345 -	"       pxor 56(%2), %%mm7   ;\n"
   4.346 -	"       pxor 48(%3), %%mm6   ;\n"
   4.347 -	"       pxor 56(%3), %%mm7   ;\n"
   4.348 -	"       movq %%mm6, 48(%1)   ;\n"
   4.349 -	"       movq %%mm7, 56(%1)   ;\n"
   4.350 -      
   4.351 -	"       addl $64, %1         ;\n"
   4.352 -	"       addl $64, %2         ;\n"
   4.353 -	"       addl $64, %3         ;\n"
   4.354 -	"       decl %0              ;\n"
   4.355 -	"       jnz 1b               ;\n"
   4.356 -	: "+r" (lines),
   4.357 -	  "+r" (p1), "+r" (p2), "+r" (p3)
   4.358 -	:
   4.359 -	: "memory" );
   4.360 -
   4.361 -	kernel_fpu_end();
   4.362 -}
   4.363 -
   4.364 -static void
   4.365 -xor_p5_mmx_4(unsigned long bytes, unsigned long *p1, unsigned long *p2,
   4.366 -	     unsigned long *p3, unsigned long *p4)
   4.367 -{
   4.368 -	unsigned long lines = bytes >> 6;
   4.369 -
   4.370 -	kernel_fpu_begin();
   4.371 -
   4.372 -	__asm__ __volatile__ (
   4.373 -	" .align 32,0x90             ;\n"
   4.374 -	" 1:                         ;\n"
   4.375 -	"       movq   (%1), %%mm0   ;\n"
   4.376 -	"       movq  8(%1), %%mm1   ;\n"
   4.377 -	"       pxor   (%2), %%mm0   ;\n"
   4.378 -	"       movq 16(%1), %%mm2   ;\n"
   4.379 -	"       pxor  8(%2), %%mm1   ;\n"
   4.380 -	"       pxor   (%3), %%mm0   ;\n"
   4.381 -	"       pxor 16(%2), %%mm2   ;\n"
   4.382 -	"       pxor  8(%3), %%mm1   ;\n"
   4.383 -	"       pxor   (%4), %%mm0   ;\n"
   4.384 -	"       movq 24(%1), %%mm3   ;\n"
   4.385 -	"       pxor 16(%3), %%mm2   ;\n"
   4.386 -	"       pxor  8(%4), %%mm1   ;\n"
   4.387 -	"       movq %%mm0,   (%1)   ;\n"
   4.388 -	"       movq 32(%1), %%mm4   ;\n"
   4.389 -	"       pxor 24(%2), %%mm3   ;\n"
   4.390 -	"       pxor 16(%4), %%mm2   ;\n"
   4.391 -	"       movq %%mm1,  8(%1)   ;\n"
   4.392 -	"       movq 40(%1), %%mm5   ;\n"
   4.393 -	"       pxor 32(%2), %%mm4   ;\n"
   4.394 -	"       pxor 24(%3), %%mm3   ;\n"
   4.395 -	"       movq %%mm2, 16(%1)   ;\n"
   4.396 -	"       pxor 40(%2), %%mm5   ;\n"
   4.397 -	"       pxor 32(%3), %%mm4   ;\n"
   4.398 -	"       pxor 24(%4), %%mm3   ;\n"
   4.399 -	"       movq %%mm3, 24(%1)   ;\n"
   4.400 -	"       movq 56(%1), %%mm7   ;\n"
   4.401 -	"       movq 48(%1), %%mm6   ;\n"
   4.402 -	"       pxor 40(%3), %%mm5   ;\n"
   4.403 -	"       pxor 32(%4), %%mm4   ;\n"
   4.404 -	"       pxor 48(%2), %%mm6   ;\n"
   4.405 -	"       movq %%mm4, 32(%1)   ;\n"
   4.406 -	"       pxor 56(%2), %%mm7   ;\n"
   4.407 -	"       pxor 40(%4), %%mm5   ;\n"
   4.408 -	"       pxor 48(%3), %%mm6   ;\n"
   4.409 -	"       pxor 56(%3), %%mm7   ;\n"
   4.410 -	"       movq %%mm5, 40(%1)   ;\n"
   4.411 -	"       pxor 48(%4), %%mm6   ;\n"
   4.412 -	"       pxor 56(%4), %%mm7   ;\n"
   4.413 -	"       movq %%mm6, 48(%1)   ;\n"
   4.414 -	"       movq %%mm7, 56(%1)   ;\n"
   4.415 -      
   4.416 -	"       addl $64, %1         ;\n"
   4.417 -	"       addl $64, %2         ;\n"
   4.418 -	"       addl $64, %3         ;\n"
   4.419 -	"       addl $64, %4         ;\n"
   4.420 -	"       decl %0              ;\n"
   4.421 -	"       jnz 1b               ;\n"
   4.422 -	: "+r" (lines),
   4.423 -	  "+r" (p1), "+r" (p2), "+r" (p3), "+r" (p4)
   4.424 -	:
   4.425 -	: "memory");
   4.426 -
   4.427 -	kernel_fpu_end();
   4.428 -}
   4.429 -
   4.430 -static void
   4.431 -xor_p5_mmx_5(unsigned long bytes, unsigned long *p1, unsigned long *p2,
   4.432 -	     unsigned long *p3, unsigned long *p4, unsigned long *p5)
   4.433 -{
   4.434 -	unsigned long lines = bytes >> 6;
   4.435 -
   4.436 -	kernel_fpu_begin();
   4.437 -
   4.438 -	/* Make sure GCC forgets anything it knows about p4 or p5,
   4.439 -	   such that it won't pass to the asm volatile below a
   4.440 -	   register that is shared with any other variable.  That's
   4.441 -	   because we modify p4 and p5 there, but we can't mark them
   4.442 -	   as read/write, otherwise we'd overflow the 10-asm-operands
   4.443 -	   limit of GCC < 3.1.  */
   4.444 -	__asm__ ("" : "+r" (p4), "+r" (p5));
   4.445 -
   4.446 -	__asm__ __volatile__ (
   4.447 -	" .align 32,0x90             ;\n"
   4.448 -	" 1:                         ;\n"
   4.449 -	"       movq   (%1), %%mm0   ;\n"
   4.450 -	"       movq  8(%1), %%mm1   ;\n"
   4.451 -	"       pxor   (%2), %%mm0   ;\n"
   4.452 -	"       pxor  8(%2), %%mm1   ;\n"
   4.453 -	"       movq 16(%1), %%mm2   ;\n"
   4.454 -	"       pxor   (%3), %%mm0   ;\n"
   4.455 -	"       pxor  8(%3), %%mm1   ;\n"
   4.456 -	"       pxor 16(%2), %%mm2   ;\n"
   4.457 -	"       pxor   (%4), %%mm0   ;\n"
   4.458 -	"       pxor  8(%4), %%mm1   ;\n"
   4.459 -	"       pxor 16(%3), %%mm2   ;\n"
   4.460 -	"       movq 24(%1), %%mm3   ;\n"
   4.461 -	"       pxor   (%5), %%mm0   ;\n"
   4.462 -	"       pxor  8(%5), %%mm1   ;\n"
   4.463 -	"       movq %%mm0,   (%1)   ;\n"
   4.464 -	"       pxor 16(%4), %%mm2   ;\n"
   4.465 -	"       pxor 24(%2), %%mm3   ;\n"
   4.466 -	"       movq %%mm1,  8(%1)   ;\n"
   4.467 -	"       pxor 16(%5), %%mm2   ;\n"
   4.468 -	"       pxor 24(%3), %%mm3   ;\n"
   4.469 -	"       movq 32(%1), %%mm4   ;\n"
   4.470 -	"       movq %%mm2, 16(%1)   ;\n"
   4.471 -	"       pxor 24(%4), %%mm3   ;\n"
   4.472 -	"       pxor 32(%2), %%mm4   ;\n"
   4.473 -	"       movq 40(%1), %%mm5   ;\n"
   4.474 -	"       pxor 24(%5), %%mm3   ;\n"
   4.475 -	"       pxor 32(%3), %%mm4   ;\n"
   4.476 -	"       pxor 40(%2), %%mm5   ;\n"
   4.477 -	"       movq %%mm3, 24(%1)   ;\n"
   4.478 -	"       pxor 32(%4), %%mm4   ;\n"
   4.479 -	"       pxor 40(%3), %%mm5   ;\n"
   4.480 -	"       movq 48(%1), %%mm6   ;\n"
   4.481 -	"       movq 56(%1), %%mm7   ;\n"
   4.482 -	"       pxor 32(%5), %%mm4   ;\n"
   4.483 -	"       pxor 40(%4), %%mm5   ;\n"
   4.484 -	"       pxor 48(%2), %%mm6   ;\n"
   4.485 -	"       pxor 56(%2), %%mm7   ;\n"
   4.486 -	"       movq %%mm4, 32(%1)   ;\n"
   4.487 -	"       pxor 48(%3), %%mm6   ;\n"
   4.488 -	"       pxor 56(%3), %%mm7   ;\n"
   4.489 -	"       pxor 40(%5), %%mm5   ;\n"
   4.490 -	"       pxor 48(%4), %%mm6   ;\n"
   4.491 -	"       pxor 56(%4), %%mm7   ;\n"
   4.492 -	"       movq %%mm5, 40(%1)   ;\n"
   4.493 -	"       pxor 48(%5), %%mm6   ;\n"
   4.494 -	"       pxor 56(%5), %%mm7   ;\n"
   4.495 -	"       movq %%mm6, 48(%1)   ;\n"
   4.496 -	"       movq %%mm7, 56(%1)   ;\n"
   4.497 -      
   4.498 -	"       addl $64, %1         ;\n"
   4.499 -	"       addl $64, %2         ;\n"
   4.500 -	"       addl $64, %3         ;\n"
   4.501 -	"       addl $64, %4         ;\n"
   4.502 -	"       addl $64, %5         ;\n"
   4.503 -	"       decl %0              ;\n"
   4.504 -	"       jnz 1b               ;\n"
   4.505 -	: "+r" (lines),
   4.506 -	  "+r" (p1), "+r" (p2), "+r" (p3)
   4.507 -	: "r" (p4), "r" (p5)
   4.508 -	: "memory");
   4.509 -
   4.510 -	/* p4 and p5 were modified, and now the variables are dead.
   4.511 -	   Clobber them just to be sure nobody does something stupid
   4.512 -	   like assuming they have some legal value.  */
   4.513 -	__asm__ ("" : "=r" (p4), "=r" (p5));
   4.514 -
   4.515 -	kernel_fpu_end();
   4.516 -}
   4.517 -
   4.518 -static struct xor_block_template xor_block_pII_mmx = {
   4.519 -	.name = "pII_mmx",
   4.520 -	.do_2 = xor_pII_mmx_2,
   4.521 -	.do_3 = xor_pII_mmx_3,
   4.522 -	.do_4 = xor_pII_mmx_4,
   4.523 -	.do_5 = xor_pII_mmx_5,
   4.524 -};
   4.525 -
   4.526 -static struct xor_block_template xor_block_p5_mmx = {
   4.527 -	.name = "p5_mmx",
   4.528 -	.do_2 = xor_p5_mmx_2,
   4.529 -	.do_3 = xor_p5_mmx_3,
   4.530 -	.do_4 = xor_p5_mmx_4,
   4.531 -	.do_5 = xor_p5_mmx_5,
   4.532 -};
   4.533 -
   4.534 -/*
   4.535 - * Cache avoiding checksumming functions utilizing KNI instructions
   4.536 - * Copyright (C) 1999 Zach Brown (with obvious credit due Ingo)
   4.537 - */
   4.538 -
   4.539 -#define XMMS_SAVE do {				\
   4.540 -	preempt_disable();			\
   4.541 -	if (!(current_thread_info()->status & TS_USEDFPU))	\
   4.542 -		clts();				\
   4.543 -	__asm__ __volatile__ ( 			\
   4.544 -		"movups %%xmm0,(%1)	;\n\t"	\
   4.545 -		"movups %%xmm1,0x10(%1)	;\n\t"	\
   4.546 -		"movups %%xmm2,0x20(%1)	;\n\t"	\
   4.547 -		"movups %%xmm3,0x30(%1)	;\n\t"	\
   4.548 -		: "=&r" (cr0)			\
   4.549 -		: "r" (xmm_save) 		\
   4.550 -		: "memory");			\
   4.551 -} while(0)
   4.552 -
   4.553 -#define XMMS_RESTORE do {			\
   4.554 -	__asm__ __volatile__ ( 			\
   4.555 -		"sfence			;\n\t"	\
   4.556 -		"movups (%1),%%xmm0	;\n\t"	\
   4.557 -		"movups 0x10(%1),%%xmm1	;\n\t"	\
   4.558 -		"movups 0x20(%1),%%xmm2	;\n\t"	\
   4.559 -		"movups 0x30(%1),%%xmm3	;\n\t"	\
   4.560 -		:				\
   4.561 -		: "r" (cr0), "r" (xmm_save)	\
   4.562 -		: "memory");			\
   4.563 -	if (!(current_thread_info()->status & TS_USEDFPU))	\
   4.564 -		stts();				\
   4.565 -	preempt_enable();			\
   4.566 -} while(0)
   4.567 -
   4.568 -#define ALIGN16 __attribute__((aligned(16)))
   4.569 -
   4.570 -#define OFFS(x)		"16*("#x")"
   4.571 -#define PF_OFFS(x)	"256+16*("#x")"
   4.572 -#define	PF0(x)		"	prefetchnta "PF_OFFS(x)"(%1)		;\n"
   4.573 -#define LD(x,y)		"       movaps   "OFFS(x)"(%1), %%xmm"#y"	;\n"
   4.574 -#define ST(x,y)		"       movaps %%xmm"#y",   "OFFS(x)"(%1)	;\n"
   4.575 -#define PF1(x)		"	prefetchnta "PF_OFFS(x)"(%2)		;\n"
   4.576 -#define PF2(x)		"	prefetchnta "PF_OFFS(x)"(%3)		;\n"
   4.577 -#define PF3(x)		"	prefetchnta "PF_OFFS(x)"(%4)		;\n"
   4.578 -#define PF4(x)		"	prefetchnta "PF_OFFS(x)"(%5)		;\n"
   4.579 -#define PF5(x)		"	prefetchnta "PF_OFFS(x)"(%6)		;\n"
   4.580 -#define XO1(x,y)	"       xorps   "OFFS(x)"(%2), %%xmm"#y"	;\n"
   4.581 -#define XO2(x,y)	"       xorps   "OFFS(x)"(%3), %%xmm"#y"	;\n"
   4.582 -#define XO3(x,y)	"       xorps   "OFFS(x)"(%4), %%xmm"#y"	;\n"
   4.583 -#define XO4(x,y)	"       xorps   "OFFS(x)"(%5), %%xmm"#y"	;\n"
   4.584 -#define XO5(x,y)	"       xorps   "OFFS(x)"(%6), %%xmm"#y"	;\n"
   4.585 -
   4.586 -
   4.587 -static void
   4.588 -xor_sse_2(unsigned long bytes, unsigned long *p1, unsigned long *p2)
   4.589 -{
   4.590 -        unsigned long lines = bytes >> 8;
   4.591 -	char xmm_save[16*4] ALIGN16;
   4.592 -	int cr0;
   4.593 -
   4.594 -	XMMS_SAVE;
   4.595 -
   4.596 -        __asm__ __volatile__ (
   4.597 -#undef BLOCK
   4.598 -#define BLOCK(i) \
   4.599 -		LD(i,0)					\
   4.600 -			LD(i+1,1)			\
   4.601 -		PF1(i)					\
   4.602 -				PF1(i+2)		\
   4.603 -				LD(i+2,2)		\
   4.604 -					LD(i+3,3)	\
   4.605 -		PF0(i+4)				\
   4.606 -				PF0(i+6)		\
   4.607 -		XO1(i,0)				\
   4.608 -			XO1(i+1,1)			\
   4.609 -				XO1(i+2,2)		\
   4.610 -					XO1(i+3,3)	\
   4.611 -		ST(i,0)					\
   4.612 -			ST(i+1,1)			\
   4.613 -				ST(i+2,2)		\
   4.614 -					ST(i+3,3)	\
   4.615 -
   4.616 -
   4.617 -		PF0(0)
   4.618 -				PF0(2)
   4.619 -
   4.620 -	" .align 32			;\n"
   4.621 -        " 1:                            ;\n"
   4.622 -
   4.623 -		BLOCK(0)
   4.624 -		BLOCK(4)
   4.625 -		BLOCK(8)
   4.626 -		BLOCK(12)
   4.627 -
   4.628 -        "       addl $256, %1           ;\n"
   4.629 -        "       addl $256, %2           ;\n"
   4.630 -        "       decl %0                 ;\n"
   4.631 -        "       jnz 1b                  ;\n"
   4.632 -	: "+r" (lines),
   4.633 -	  "+r" (p1), "+r" (p2)
   4.634 -	:
   4.635 -        : "memory");
   4.636 -
   4.637 -	XMMS_RESTORE;
   4.638 -}
   4.639 -
   4.640 -static void
   4.641 -xor_sse_3(unsigned long bytes, unsigned long *p1, unsigned long *p2,
   4.642 -	  unsigned long *p3)
   4.643 -{
   4.644 -        unsigned long lines = bytes >> 8;
   4.645 -	char xmm_save[16*4] ALIGN16;
   4.646 -	int cr0;
   4.647 -
   4.648 -	XMMS_SAVE;
   4.649 -
   4.650 -        __asm__ __volatile__ (
   4.651 -#undef BLOCK
   4.652 -#define BLOCK(i) \
   4.653 -		PF1(i)					\
   4.654 -				PF1(i+2)		\
   4.655 -		LD(i,0)					\
   4.656 -			LD(i+1,1)			\
   4.657 -				LD(i+2,2)		\
   4.658 -					LD(i+3,3)	\
   4.659 -		PF2(i)					\
   4.660 -				PF2(i+2)		\
   4.661 -		PF0(i+4)				\
   4.662 -				PF0(i+6)		\
   4.663 -		XO1(i,0)				\
   4.664 -			XO1(i+1,1)			\
   4.665 -				XO1(i+2,2)		\
   4.666 -					XO1(i+3,3)	\
   4.667 -		XO2(i,0)				\
   4.668 -			XO2(i+1,1)			\
   4.669 -				XO2(i+2,2)		\
   4.670 -					XO2(i+3,3)	\
   4.671 -		ST(i,0)					\
   4.672 -			ST(i+1,1)			\
   4.673 -				ST(i+2,2)		\
   4.674 -					ST(i+3,3)	\
   4.675 -
   4.676 -
   4.677 -		PF0(0)
   4.678 -				PF0(2)
   4.679 -
   4.680 -	" .align 32			;\n"
   4.681 -        " 1:                            ;\n"
   4.682 -
   4.683 -		BLOCK(0)
   4.684 -		BLOCK(4)
   4.685 -		BLOCK(8)
   4.686 -		BLOCK(12)
   4.687 -
   4.688 -        "       addl $256, %1           ;\n"
   4.689 -        "       addl $256, %2           ;\n"
   4.690 -        "       addl $256, %3           ;\n"
   4.691 -        "       decl %0                 ;\n"
   4.692 -        "       jnz 1b                  ;\n"
   4.693 -	: "+r" (lines),
   4.694 -	  "+r" (p1), "+r"(p2), "+r"(p3)
   4.695 -	:
   4.696 -        : "memory" );
   4.697 -
   4.698 -	XMMS_RESTORE;
   4.699 -}
   4.700 -
   4.701 -static void
   4.702 -xor_sse_4(unsigned long bytes, unsigned long *p1, unsigned long *p2,
   4.703 -	  unsigned long *p3, unsigned long *p4)
   4.704 -{
   4.705 -        unsigned long lines = bytes >> 8;
   4.706 -	char xmm_save[16*4] ALIGN16;
   4.707 -	int cr0;
   4.708 -
   4.709 -	XMMS_SAVE;
   4.710 -
   4.711 -        __asm__ __volatile__ (
   4.712 -#undef BLOCK
   4.713 -#define BLOCK(i) \
   4.714 -		PF1(i)					\
   4.715 -				PF1(i+2)		\
   4.716 -		LD(i,0)					\
   4.717 -			LD(i+1,1)			\
   4.718 -				LD(i+2,2)		\
   4.719 -					LD(i+3,3)	\
   4.720 -		PF2(i)					\
   4.721 -				PF2(i+2)		\
   4.722 -		XO1(i,0)				\
   4.723 -			XO1(i+1,1)			\
   4.724 -				XO1(i+2,2)		\
   4.725 -					XO1(i+3,3)	\
   4.726 -		PF3(i)					\
   4.727 -				PF3(i+2)		\
   4.728 -		PF0(i+4)				\
   4.729 -				PF0(i+6)		\
   4.730 -		XO2(i,0)				\
   4.731 -			XO2(i+1,1)			\
   4.732 -				XO2(i+2,2)		\
   4.733 -					XO2(i+3,3)	\
   4.734 -		XO3(i,0)				\
   4.735 -			XO3(i+1,1)			\
   4.736 -				XO3(i+2,2)		\
   4.737 -					XO3(i+3,3)	\
   4.738 -		ST(i,0)					\
   4.739 -			ST(i+1,1)			\
   4.740 -				ST(i+2,2)		\
   4.741 -					ST(i+3,3)	\
   4.742 -
   4.743 -
   4.744 -		PF0(0)
   4.745 -				PF0(2)
   4.746 -
   4.747 -	" .align 32			;\n"
   4.748 -        " 1:                            ;\n"
   4.749 -
   4.750 -		BLOCK(0)
   4.751 -		BLOCK(4)
   4.752 -		BLOCK(8)
   4.753 -		BLOCK(12)
   4.754 -
   4.755 -        "       addl $256, %1           ;\n"
   4.756 -        "       addl $256, %2           ;\n"
   4.757 -        "       addl $256, %3           ;\n"
   4.758 -        "       addl $256, %4           ;\n"
   4.759 -        "       decl %0                 ;\n"
   4.760 -        "       jnz 1b                  ;\n"
   4.761 -	: "+r" (lines),
   4.762 -	  "+r" (p1), "+r" (p2), "+r" (p3), "+r" (p4)
   4.763 -	:
   4.764 -        : "memory" );
   4.765 -
   4.766 -	XMMS_RESTORE;
   4.767 -}
   4.768 -
   4.769 -static void
   4.770 -xor_sse_5(unsigned long bytes, unsigned long *p1, unsigned long *p2,
   4.771 -	  unsigned long *p3, unsigned long *p4, unsigned long *p5)
   4.772 -{
   4.773 -        unsigned long lines = bytes >> 8;
   4.774 -	char xmm_save[16*4] ALIGN16;
   4.775 -	int cr0;
   4.776 -
   4.777 -	XMMS_SAVE;
   4.778 -
   4.779 -	/* Make sure GCC forgets anything it knows about p4 or p5,
   4.780 -	   such that it won't pass to the asm volatile below a
   4.781 -	   register that is shared with any other variable.  That's
   4.782 -	   because we modify p4 and p5 there, but we can't mark them
   4.783 -	   as read/write, otherwise we'd overflow the 10-asm-operands
   4.784 -	   limit of GCC < 3.1.  */
   4.785 -	__asm__ ("" : "+r" (p4), "+r" (p5));
   4.786 -
   4.787 -        __asm__ __volatile__ (
   4.788 -#undef BLOCK
   4.789 -#define BLOCK(i) \
   4.790 -		PF1(i)					\
   4.791 -				PF1(i+2)		\
   4.792 -		LD(i,0)					\
   4.793 -			LD(i+1,1)			\
   4.794 -				LD(i+2,2)		\
   4.795 -					LD(i+3,3)	\
   4.796 -		PF2(i)					\
   4.797 -				PF2(i+2)		\
   4.798 -		XO1(i,0)				\
   4.799 -			XO1(i+1,1)			\
   4.800 -				XO1(i+2,2)		\
   4.801 -					XO1(i+3,3)	\
   4.802 -		PF3(i)					\
   4.803 -				PF3(i+2)		\
   4.804 -		XO2(i,0)				\
   4.805 -			XO2(i+1,1)			\
   4.806 -				XO2(i+2,2)		\
   4.807 -					XO2(i+3,3)	\
   4.808 -		PF4(i)					\
   4.809 -				PF4(i+2)		\
   4.810 -		PF0(i+4)				\
   4.811 -				PF0(i+6)		\
   4.812 -		XO3(i,0)				\
   4.813 -			XO3(i+1,1)			\
   4.814 -				XO3(i+2,2)		\
   4.815 -					XO3(i+3,3)	\
   4.816 -		XO4(i,0)				\
   4.817 -			XO4(i+1,1)			\
   4.818 -				XO4(i+2,2)		\
   4.819 -					XO4(i+3,3)	\
   4.820 -		ST(i,0)					\
   4.821 -			ST(i+1,1)			\
   4.822 -				ST(i+2,2)		\
   4.823 -					ST(i+3,3)	\
   4.824 -
   4.825 -
   4.826 -		PF0(0)
   4.827 -				PF0(2)
   4.828 -
   4.829 -	" .align 32			;\n"
   4.830 -        " 1:                            ;\n"
   4.831 -
   4.832 -		BLOCK(0)
   4.833 -		BLOCK(4)
   4.834 -		BLOCK(8)
   4.835 -		BLOCK(12)
   4.836 -
   4.837 -        "       addl $256, %1           ;\n"
   4.838 -        "       addl $256, %2           ;\n"
   4.839 -        "       addl $256, %3           ;\n"
   4.840 -        "       addl $256, %4           ;\n"
   4.841 -        "       addl $256, %5           ;\n"
   4.842 -        "       decl %0                 ;\n"
   4.843 -        "       jnz 1b                  ;\n"
   4.844 -	: "+r" (lines),
   4.845 -	  "+r" (p1), "+r" (p2), "+r" (p3)
   4.846 -	: "r" (p4), "r" (p5)
   4.847 -	: "memory");
   4.848 -
   4.849 -	/* p4 and p5 were modified, and now the variables are dead.
   4.850 -	   Clobber them just to be sure nobody does something stupid
   4.851 -	   like assuming they have some legal value.  */
   4.852 -	__asm__ ("" : "=r" (p4), "=r" (p5));
   4.853 -
   4.854 -	XMMS_RESTORE;
   4.855 -}
   4.856 -
   4.857 -static struct xor_block_template xor_block_pIII_sse = {
   4.858 -        .name = "pIII_sse",
   4.859 -        .do_2 =  xor_sse_2,
   4.860 -        .do_3 =  xor_sse_3,
   4.861 -        .do_4 =  xor_sse_4,
   4.862 -        .do_5 = xor_sse_5,
   4.863 -};
   4.864 -
   4.865 -/* Also try the generic routines.  */
   4.866 -#include <asm-generic/xor.h>
   4.867 -
   4.868 -#undef XOR_TRY_TEMPLATES
   4.869 -#define XOR_TRY_TEMPLATES				\
   4.870 -	do {						\
   4.871 -		xor_speed(&xor_block_8regs);		\
   4.872 -		xor_speed(&xor_block_8regs_p);		\
   4.873 -		xor_speed(&xor_block_32regs);		\
   4.874 -		xor_speed(&xor_block_32regs_p);		\
   4.875 -	        if (cpu_has_xmm)			\
   4.876 -			xor_speed(&xor_block_pIII_sse);	\
   4.877 -	        if (cpu_has_mmx) {			\
   4.878 -	                xor_speed(&xor_block_pII_mmx);	\
   4.879 -	                xor_speed(&xor_block_p5_mmx);	\
   4.880 -	        }					\
   4.881 -	} while (0)
   4.882 -
   4.883 -/* We force the use of the SSE xor block because it can write around L2.
   4.884 -   We may also be able to load into the L1 only depending on how the cpu
   4.885 -   deals with a load to a line that is being prefetched.  */
   4.886 -#define XOR_SELECT_TEMPLATE(FASTEST) \
   4.887 -	(cpu_has_xmm ? &xor_block_pIII_sse : FASTEST)

     5.1 --- a/xen/arch/x86/traps.c	Tue Feb 22 23:31:53 2005 +0000
     5.2 +++ b/xen/arch/x86/traps.c	Tue Feb 22 23:46:11 2005 +0000
     5.3 @@ -351,7 +351,6 @@ asmlinkage int do_page_fault(struct xen_
     5.4  
     5.5  static int emulate_privileged_op(struct xen_regs *regs)
     5.6  {
     5.7 -    extern long do_fpu_taskswitch(void);
     5.8      extern void *decode_reg(struct xen_regs *regs, u8 b);
     5.9  
    5.10      struct exec_domain *ed = current;
    5.11 @@ -423,7 +422,16 @@ static int emulate_privileged_op(struct 
    5.12          {
    5.13          case 0: /* Write CR0 */
    5.14              if ( *reg & X86_CR0_TS )
    5.15 -                (void)do_fpu_taskswitch();
    5.16 +            {
    5.17 +                set_bit(EDF_GUEST_STTS, &ed->ed_flags);
    5.18 +                stts();
    5.19 +            }
    5.20 +            else
    5.21 +            {
    5.22 +                clear_bit(EDF_GUEST_STTS, &ed->ed_flags);
    5.23 +                if ( test_bit(EDF_USEDFPU, &ed->ed_flags) )
    5.24 +                    clts();
    5.25 +            }
    5.26              break;
    5.27  
    5.28          case 2: /* Write CR2 */