ia64/lib/memset.S

1 /* SPDX-License-Identifier: GPL-2.0 */
4    Copyright (c) 2002 Hewlett-Packard Co/CERN
15    we get to a 16B-aligned address, then loop on 128 B chunks using an
42 // This routine uses only scratch predicate registers (p6 - p15)
43 #define p_scr		p6			// default register for same-cycle branches
72 	and	ptr2 = -(MIN1+1), dest		// aligned address
93 (p_y)	add	cnt = -8, cnt			//
96 (p_y)	st8	[ptr2] = value,-4		//
100 (p_yy)	add	cnt = -4, cnt			//
103 (p_yy)	st4	[ptr2] = value,-2		//
108 (p_y)	add	cnt = -2, cnt			//
112 (p_y)	st2	[ptr2] = value,-1		//
120 (p_yy)	add	cnt = -1, cnt			//
130 	TEXT_ALIGN(32) // --------------------- //  L1A: store ahead into cache lines; fill later
132 	and	tmp = -(LINE_SIZE), cnt		// compute end of range
134 	and	cnt = (LINE_SIZE-1), cnt	// remainder
136 	mov	loopcnt = PREF_AHEAD-1		// default prefetch loop
140 (p_scr)	add	loopcnt = -1, linecnt		//
145 	add	tmp = -1, linecnt		// next loop count
194 	cmp.le  p_scr, p0 = 8, cnt		// just a few bytes left ?
200 .l1b:	// ------------------------------------ //  L1B: store ahead into cache lines; fill later
202 	and	tmp = -(LINE_SIZE), cnt		// compute end of range
204 	and	cnt = (LINE_SIZE-1), cnt	// remainder
206 	mov	loopcnt = PREF_AHEAD-1		// default prefetch loop
210 (p_scr)	add	loopcnt = -1, linecnt
215 	add	tmp = -1, linecnt		// next loop count
248 	cmp.gt  p_scr, p0 = 8, cnt		// just a few bytes left ?
259 	add	loopcnt = -1, loopcnt
267 .l2:	// ------------------------------------ //  L2A:  store 32B in 2 cycles
278 	cmp.gt	p_scr, p0 = 8, cnt		// just a few bytes left ?
285 	add	cnt = -8, cnt			// subtract
290 (p_y)	add	cnt = -8, cnt			// subtract
294 (p_yy)	add	cnt = -8, cnt			// subtract
331 (p_y)	st1	[ptr1] = value, 1		// fill 1 (odd-aligned) byte [15, 14 (or less) left]
332 (p_y)	add	cnt = -1, cnt
339 (p_yy)	st2	[ptr1] = value, 4		// fill 2 (aligned) bytes
340 (p_yy)	st2	[ptr2] = value, 4		// fill 2 (aligned) bytes [11, 10 (o less) left]
341 (p_yy)	add	cnt = -4, cnt
345 	add	ptr3 = -1, ptr3			// last store
348 (p_y)	st2	[ptr1] = value, 4		// fill 2 (aligned) bytes
349 (p_y)	st2	[ptr2] = value, 4		// fill 2 (aligned) bytes [7, 6 (or less) left]
350 (p_y)	add	cnt = -4, cnt
353 (p_yy)	st2	[ptr1] = value, 4		// fill 2 (aligned) bytes
354 (p_yy)	st2	[ptr2] = value, 4		// fill 2 (aligned) bytes [3, 2 (or less) left]
357 (p_yy)	add	cnt = -4, cnt
360 (p_scr)	st2	[ptr1] = value			// fill 2 (aligned) bytes