head.s

xen虚拟机源代码安装包

END(_start)

GLOBAL_ENTRY(ia64_save_debug_regs)
	alloc r16=ar.pfs,1,0,0,0
	mov r20=ar.lc			// preserve ar.lc
	mov ar.lc=IA64_NUM_DBG_REGS-1
	mov r18=0
	add r19=IA64_NUM_DBG_REGS*8,in0
	;;
1:	mov r16=dbr[r18]
#ifdef CONFIG_ITANIUM
	;;
	srlz.d
#endif
	mov r17=ibr[r18]
	add r18=1,r18
	;;
	st8.nta [in0]=r16,8
	st8.nta [r19]=r17,8
	br.cloop.sptk.many 1b
	;;
	mov ar.lc=r20			// restore ar.lc
	br.ret.sptk.many rp
END(ia64_save_debug_regs)

GLOBAL_ENTRY(ia64_load_debug_regs)
	alloc r16=ar.pfs,1,0,0,0
	lfetch.nta [in0]
	mov r20=ar.lc			// preserve ar.lc
	add r19=IA64_NUM_DBG_REGS*8,in0
	mov ar.lc=IA64_NUM_DBG_REGS-1
	mov r18=-1
	;;
1:	ld8.nta r16=[in0],8
	ld8.nta r17=[r19],8
	add r18=1,r18
	;;
	mov dbr[r18]=r16
#ifdef CONFIG_ITANIUM
	;;
	srlz.d				// Errata 132 (NoFix status)
#endif
	mov ibr[r18]=r17
	br.cloop.sptk.many 1b
	;;
	mov ar.lc=r20			// restore ar.lc
	br.ret.sptk.many rp
END(ia64_load_debug_regs)

GLOBAL_ENTRY(__ia64_save_fpu)
	alloc r2=ar.pfs,1,4,0,0
	adds loc0=96*16-16,in0
	adds loc1=96*16-16-128,in0
	;;
	stf.spill.nta [loc0]=f127,-256
	stf.spill.nta [loc1]=f119,-256
	;;
	stf.spill.nta [loc0]=f111,-256
	stf.spill.nta [loc1]=f103,-256
	;;
	stf.spill.nta [loc0]=f95,-256
	stf.spill.nta [loc1]=f87,-256
	;;
	stf.spill.nta [loc0]=f79,-256
	stf.spill.nta [loc1]=f71,-256
	;;
	stf.spill.nta [loc0]=f63,-256
	stf.spill.nta [loc1]=f55,-256
	adds loc2=96*16-32,in0
	;;
	stf.spill.nta [loc0]=f47,-256
	stf.spill.nta [loc1]=f39,-256
	adds loc3=96*16-32-128,in0
	;;
	stf.spill.nta [loc2]=f126,-256
	stf.spill.nta [loc3]=f118,-256
	;;
	stf.spill.nta [loc2]=f110,-256
	stf.spill.nta [loc3]=f102,-256
	;;
	stf.spill.nta [loc2]=f94,-256
	stf.spill.nta [loc3]=f86,-256
	;;
	stf.spill.nta [loc2]=f78,-256
	stf.spill.nta [loc3]=f70,-256
	;;
	stf.spill.nta [loc2]=f62,-256
	stf.spill.nta [loc3]=f54,-256
	adds loc0=96*16-48,in0
	;;
	stf.spill.nta [loc2]=f46,-256
	stf.spill.nta [loc3]=f38,-256
	adds loc1=96*16-48-128,in0
	;;
	stf.spill.nta [loc0]=f125,-256
	stf.spill.nta [loc1]=f117,-256
	;;
	stf.spill.nta [loc0]=f109,-256
	stf.spill.nta [loc1]=f101,-256
	;;
	stf.spill.nta [loc0]=f93,-256
	stf.spill.nta [loc1]=f85,-256
	;;
	stf.spill.nta [loc0]=f77,-256
	stf.spill.nta [loc1]=f69,-256
	;;
	stf.spill.nta [loc0]=f61,-256
	stf.spill.nta [loc1]=f53,-256
	adds loc2=96*16-64,in0
	;;
	stf.spill.nta [loc0]=f45,-256
	stf.spill.nta [loc1]=f37,-256
	adds loc3=96*16-64-128,in0
	;;
	stf.spill.nta [loc2]=f124,-256
	stf.spill.nta [loc3]=f116,-256
	;;
	stf.spill.nta [loc2]=f108,-256
	stf.spill.nta [loc3]=f100,-256
	;;
	stf.spill.nta [loc2]=f92,-256
	stf.spill.nta [loc3]=f84,-256
	;;
	stf.spill.nta [loc2]=f76,-256
	stf.spill.nta [loc3]=f68,-256
	;;
	stf.spill.nta [loc2]=f60,-256
	stf.spill.nta [loc3]=f52,-256
	adds loc0=96*16-80,in0
	;;
	stf.spill.nta [loc2]=f44,-256
	stf.spill.nta [loc3]=f36,-256
	adds loc1=96*16-80-128,in0
	;;
	stf.spill.nta [loc0]=f123,-256
	stf.spill.nta [loc1]=f115,-256
	;;
	stf.spill.nta [loc0]=f107,-256
	stf.spill.nta [loc1]=f99,-256
	;;
	stf.spill.nta [loc0]=f91,-256
	stf.spill.nta [loc1]=f83,-256
	;;
	stf.spill.nta [loc0]=f75,-256
	stf.spill.nta [loc1]=f67,-256
	;;
	stf.spill.nta [loc0]=f59,-256
	stf.spill.nta [loc1]=f51,-256
	adds loc2=96*16-96,in0
	;;
	stf.spill.nta [loc0]=f43,-256
	stf.spill.nta [loc1]=f35,-256
	adds loc3=96*16-96-128,in0
	;;
	stf.spill.nta [loc2]=f122,-256
	stf.spill.nta [loc3]=f114,-256
	;;
	stf.spill.nta [loc2]=f106,-256
	stf.spill.nta [loc3]=f98,-256
	;;
	stf.spill.nta [loc2]=f90,-256
	stf.spill.nta [loc3]=f82,-256
	;;
	stf.spill.nta [loc2]=f74,-256
	stf.spill.nta [loc3]=f66,-256
	;;
	stf.spill.nta [loc2]=f58,-256
	stf.spill.nta [loc3]=f50,-256
	adds loc0=96*16-112,in0
	;;
	stf.spill.nta [loc2]=f42,-256
	stf.spill.nta [loc3]=f34,-256
	adds loc1=96*16-112-128,in0
	;;
	stf.spill.nta [loc0]=f121,-256
	stf.spill.nta [loc1]=f113,-256
	;;
	stf.spill.nta [loc0]=f105,-256
	stf.spill.nta [loc1]=f97,-256
	;;
	stf.spill.nta [loc0]=f89,-256
	stf.spill.nta [loc1]=f81,-256
	;;
	stf.spill.nta [loc0]=f73,-256
	stf.spill.nta [loc1]=f65,-256
	;;
	stf.spill.nta [loc0]=f57,-256
	stf.spill.nta [loc1]=f49,-256
	adds loc2=96*16-128,in0
	;;
	stf.spill.nta [loc0]=f41,-256
	stf.spill.nta [loc1]=f33,-256
	adds loc3=96*16-128-128,in0
	;;
	stf.spill.nta [loc2]=f120,-256
	stf.spill.nta [loc3]=f112,-256
	;;
	stf.spill.nta [loc2]=f104,-256
	stf.spill.nta [loc3]=f96,-256
	;;
	stf.spill.nta [loc2]=f88,-256
	stf.spill.nta [loc3]=f80,-256
	;;
	stf.spill.nta [loc2]=f72,-256
	stf.spill.nta [loc3]=f64,-256
	;;
	stf.spill.nta [loc2]=f56,-256
	stf.spill.nta [loc3]=f48,-256
	;;
	stf.spill.nta [loc2]=f40
	stf.spill.nta [loc3]=f32
	br.ret.sptk.many rp
END(__ia64_save_fpu)

GLOBAL_ENTRY(__ia64_load_fpu)
	alloc r2=ar.pfs,1,2,0,0
	adds r3=128,in0
	adds r14=256,in0
	adds r15=384,in0
	mov loc0=512
	mov loc1=-1024+16
	;;
	ldf.fill.nta f32=[in0],loc0
	ldf.fill.nta f40=[ r3],loc0
	ldf.fill.nta f48=[r14],loc0
	ldf.fill.nta f56=[r15],loc0
	;;
	ldf.fill.nta f64=[in0],loc0
	ldf.fill.nta f72=[ r3],loc0
	ldf.fill.nta f80=[r14],loc0
	ldf.fill.nta f88=[r15],loc0
	;;
	ldf.fill.nta f96=[in0],loc1
	ldf.fill.nta f104=[ r3],loc1
	ldf.fill.nta f112=[r14],loc1
	ldf.fill.nta f120=[r15],loc1
	;;
	ldf.fill.nta f33=[in0],loc0
	ldf.fill.nta f41=[ r3],loc0
	ldf.fill.nta f49=[r14],loc0
	ldf.fill.nta f57=[r15],loc0
	;;
	ldf.fill.nta f65=[in0],loc0
	ldf.fill.nta f73=[ r3],loc0
	ldf.fill.nta f81=[r14],loc0
	ldf.fill.nta f89=[r15],loc0
	;;
	ldf.fill.nta f97=[in0],loc1
	ldf.fill.nta f105=[ r3],loc1
	ldf.fill.nta f113=[r14],loc1
	ldf.fill.nta f121=[r15],loc1
	;;
	ldf.fill.nta f34=[in0],loc0
	ldf.fill.nta f42=[ r3],loc0
	ldf.fill.nta f50=[r14],loc0
	ldf.fill.nta f58=[r15],loc0
	;;
	ldf.fill.nta f66=[in0],loc0
	ldf.fill.nta f74=[ r3],loc0
	ldf.fill.nta f82=[r14],loc0
	ldf.fill.nta f90=[r15],loc0
	;;
	ldf.fill.nta f98=[in0],loc1
	ldf.fill.nta f106=[ r3],loc1
	ldf.fill.nta f114=[r14],loc1
	ldf.fill.nta f122=[r15],loc1
	;;
	ldf.fill.nta f35=[in0],loc0
	ldf.fill.nta f43=[ r3],loc0
	ldf.fill.nta f51=[r14],loc0
	ldf.fill.nta f59=[r15],loc0
	;;
	ldf.fill.nta f67=[in0],loc0
	ldf.fill.nta f75=[ r3],loc0
	ldf.fill.nta f83=[r14],loc0
	ldf.fill.nta f91=[r15],loc0
	;;
	ldf.fill.nta f99=[in0],loc1
	ldf.fill.nta f107=[ r3],loc1
	ldf.fill.nta f115=[r14],loc1
	ldf.fill.nta f123=[r15],loc1
	;;
	ldf.fill.nta f36=[in0],loc0
	ldf.fill.nta f44=[ r3],loc0
	ldf.fill.nta f52=[r14],loc0
	ldf.fill.nta f60=[r15],loc0
	;;
	ldf.fill.nta f68=[in0],loc0
	ldf.fill.nta f76=[ r3],loc0
	ldf.fill.nta f84=[r14],loc0
	ldf.fill.nta f92=[r15],loc0
	;;
	ldf.fill.nta f100=[in0],loc1
	ldf.fill.nta f108=[ r3],loc1
	ldf.fill.nta f116=[r14],loc1
	ldf.fill.nta f124=[r15],loc1
	;;
	ldf.fill.nta f37=[in0],loc0
	ldf.fill.nta f45=[ r3],loc0
	ldf.fill.nta f53=[r14],loc0
	ldf.fill.nta f61=[r15],loc0
	;;
	ldf.fill.nta f69=[in0],loc0
	ldf.fill.nta f77=[ r3],loc0
	ldf.fill.nta f85=[r14],loc0
	ldf.fill.nta f93=[r15],loc0
	;;
	ldf.fill.nta f101=[in0],loc1
	ldf.fill.nta f109=[ r3],loc1
	ldf.fill.nta f117=[r14],loc1
	ldf.fill.nta f125=[r15],loc1
	;;
	ldf.fill.nta f38 =[in0],loc0
	ldf.fill.nta f46 =[ r3],loc0
	ldf.fill.nta f54 =[r14],loc0
	ldf.fill.nta f62 =[r15],loc0
	;;
	ldf.fill.nta f70 =[in0],loc0
	ldf.fill.nta f78 =[ r3],loc0
	ldf.fill.nta f86 =[r14],loc0
	ldf.fill.nta f94 =[r15],loc0
	;;
	ldf.fill.nta f102=[in0],loc1
	ldf.fill.nta f110=[ r3],loc1
	ldf.fill.nta f118=[r14],loc1
	ldf.fill.nta f126=[r15],loc1
	;;
	ldf.fill.nta f39 =[in0],loc0
	ldf.fill.nta f47 =[ r3],loc0
	ldf.fill.nta f55 =[r14],loc0
	ldf.fill.nta f63 =[r15],loc0
	;;
	ldf.fill.nta f71 =[in0],loc0
	ldf.fill.nta f79 =[ r3],loc0
	ldf.fill.nta f87 =[r14],loc0
	ldf.fill.nta f95 =[r15],loc0
	;;
	ldf.fill.nta f103=[in0]
	ldf.fill.nta f111=[ r3]
	ldf.fill.nta f119=[r14]
	ldf.fill.nta f127=[r15]
	br.ret.sptk.many rp
END(__ia64_load_fpu)

GLOBAL_ENTRY(__ia64_init_fpu)
	stf.spill [sp]=f0		// M3
	mov	 f32=f0			// F
	nop.b	 0

	ldfps	 f33,f34=[sp]		// M0
	ldfps	 f35,f36=[sp]		// M1
	mov      f37=f0			// F
	;;

	setf.s	 f38=r0			// M2
	setf.s	 f39=r0			// M3
	mov      f40=f0			// F

	ldfps	 f41,f42=[sp]		// M0
	ldfps	 f43,f44=[sp]		// M1
	mov      f45=f0			// F

	setf.s	 f46=r0			// M2
	setf.s	 f47=r0			// M3
	mov      f48=f0			// F

	ldfps	 f49,f50=[sp]		// M0
	ldfps	 f51,f52=[sp]		// M1
	mov      f53=f0			// F

	setf.s	 f54=r0			// M2
	setf.s	 f55=r0			// M3
	mov      f56=f0			// F

	ldfps	 f57,f58=[sp]		// M0
	ldfps	 f59,f60=[sp]		// M1
	mov      f61=f0			// F

	setf.s	 f62=r0			// M2
	setf.s	 f63=r0			// M3
	mov      f64=f0			// F

	ldfps	 f65,f66=[sp]		// M0
	ldfps	 f67,f68=[sp]		// M1
	mov      f69=f0			// F

	setf.s	 f70=r0			// M2
	setf.s	 f71=r0			// M3
	mov      f72=f0			// F

	ldfps	 f73,f74=[sp]		// M0
	ldfps	 f75,f76=[sp]		// M1
	mov      f77=f0			// F

	setf.s	 f78=r0			// M2
	setf.s	 f79=r0			// M3
	mov      f80=f0			// F

	ldfps	 f81,f82=[sp]		// M0
	ldfps	 f83,f84=[sp]		// M1
	mov      f85=f0			// F

	setf.s	 f86=r0			// M2
	setf.s	 f87=r0			// M3
	mov      f88=f0			// F

	/*
	 * When the instructions are cached, it would be faster to initialize
	 * the remaining registers with simply mov instructions (F-unit).
	 * This gets the time down to ~29 cycles.  However, this would use up
	 * 33 bundles, whereas continuing with the above pattern yields
	 * 10 bundles and ~30 cycles.
	 */

	ldfps	 f89,f90=[sp]		// M0
	ldfps	 f91,f92=[sp]		// M1
	mov      f93=f0			// F

	setf.s	 f94=r0			// M2
	setf.s	 f95=r0			// M3
	mov      f96=f0			// F