xref: /linux/arch/powerpc/kernel/misc_32.S (revision d53b8e36925256097a08d7cb749198d85cbf9b2b)
1/* SPDX-License-Identifier: GPL-2.0-or-later */
2/*
3 * This file contains miscellaneous low-level functions.
4 *    Copyright (C) 1995-1996 Gary Thomas (gdt@linuxppc.org)
5 *
6 * Largely rewritten by Cort Dougan (cort@cs.nmt.edu)
7 * and Paul Mackerras.
8 *
9 */
10
11#include <linux/export.h>
12#include <linux/sys.h>
13#include <asm/unistd.h>
14#include <asm/errno.h>
15#include <asm/reg.h>
16#include <asm/page.h>
17#include <asm/cache.h>
18#include <asm/cputable.h>
19#include <asm/mmu.h>
20#include <asm/ppc_asm.h>
21#include <asm/thread_info.h>
22#include <asm/asm-offsets.h>
23#include <asm/processor.h>
24#include <asm/bug.h>
25#include <asm/ptrace.h>
26#include <asm/feature-fixups.h>
27
28	.text
29
30/*
31 * This returns the high 64 bits of the product of two 64-bit numbers.
32 */
33_GLOBAL(mulhdu)
34	cmpwi	r6,0
35	cmpwi	cr1,r3,0
36	mr	r10,r4
37	mulhwu	r4,r4,r5
38	beq	1f
39	mulhwu	r0,r10,r6
40	mullw	r7,r10,r5
41	addc	r7,r0,r7
42	addze	r4,r4
431:	beqlr	cr1		/* all done if high part of A is 0 */
44	mullw	r9,r3,r5
45	mulhwu	r10,r3,r5
46	beq	2f
47	mullw	r0,r3,r6
48	mulhwu	r8,r3,r6
49	addc	r7,r0,r7
50	adde	r4,r4,r8
51	addze	r10,r10
522:	addc	r4,r4,r9
53	addze	r3,r10
54	blr
55
56/*
57 * reloc_got2 runs through the .got2 section adding an offset
58 * to each entry.
59 */
60_GLOBAL(reloc_got2)
61	mflr	r11
62	lis	r7,__got2_start@ha
63	addi	r7,r7,__got2_start@l
64	lis	r8,__got2_end@ha
65	addi	r8,r8,__got2_end@l
66	subf	r8,r7,r8
67	srwi.	r8,r8,2
68	beqlr
69	mtctr	r8
70	bcl	20,31,$+4
711:	mflr	r0
72	lis	r4,1b@ha
73	addi	r4,r4,1b@l
74	subf	r0,r4,r0
75	add	r7,r0,r7
762:	lwz	r0,0(r7)
77	add	r0,r0,r3
78	stw	r0,0(r7)
79	addi	r7,r7,4
80	bdnz	2b
81	mtlr	r11
82	blr
83
84/*
85 * call_setup_cpu - call the setup_cpu function for this cpu
86 * r3 = data offset, r24 = cpu number
87 *
88 * Setup function is called with:
89 *   r3 = data offset
90 *   r4 = ptr to CPU spec (relocated)
91 */
92_GLOBAL(call_setup_cpu)
93	addis	r4,r3,cur_cpu_spec@ha
94	addi	r4,r4,cur_cpu_spec@l
95	lwz	r4,0(r4)
96	add	r4,r4,r3
97	lwz	r5,CPU_SPEC_SETUP(r4)
98	cmpwi	0,r5,0
99	add	r5,r5,r3
100	beqlr
101	mtctr	r5
102	bctr
103
104#if defined(CONFIG_CPU_FREQ_PMAC) && defined(CONFIG_PPC_BOOK3S_32)
105
106/* This gets called by via-pmu.c to switch the PLL selection
107 * on 750fx CPU. This function should really be moved to some
108 * other place (as most of the cpufreq code in via-pmu
109 */
110_GLOBAL(low_choose_750fx_pll)
111	/* Clear MSR:EE */
112	mfmsr	r7
113	rlwinm	r0,r7,0,17,15
114	mtmsr	r0
115
116	/* If switching to PLL1, disable HID0:BTIC */
117	cmplwi	cr0,r3,0
118	beq	1f
119	mfspr	r5,SPRN_HID0
120	rlwinm	r5,r5,0,27,25
121	sync
122	mtspr	SPRN_HID0,r5
123	isync
124	sync
125
1261:
127	/* Calc new HID1 value */
128	mfspr	r4,SPRN_HID1	/* Build a HID1:PS bit from parameter */
129	rlwinm	r5,r3,16,15,15	/* Clear out HID1:PS from value read */
130	rlwinm	r4,r4,0,16,14	/* Could have I used rlwimi here ? */
131	or	r4,r4,r5
132	mtspr	SPRN_HID1,r4
133
134#ifdef CONFIG_SMP
135	/* Store new HID1 image */
136	lwz	r6,TASK_CPU(r2)
137	slwi	r6,r6,2
138#else
139	li	r6, 0
140#endif
141	addis	r6,r6,nap_save_hid1@ha
142	stw	r4,nap_save_hid1@l(r6)
143
144	/* If switching to PLL0, enable HID0:BTIC */
145	cmplwi	cr0,r3,0
146	bne	1f
147	mfspr	r5,SPRN_HID0
148	ori	r5,r5,HID0_BTIC
149	sync
150	mtspr	SPRN_HID0,r5
151	isync
152	sync
153
1541:
155	/* Return */
156	mtmsr	r7
157	blr
158
159_GLOBAL(low_choose_7447a_dfs)
160	/* Clear MSR:EE */
161	mfmsr	r7
162	rlwinm	r0,r7,0,17,15
163	mtmsr	r0
164
165	/* Calc new HID1 value */
166	mfspr	r4,SPRN_HID1
167	insrwi	r4,r3,1,9	/* insert parameter into bit 9 */
168	sync
169	mtspr	SPRN_HID1,r4
170	sync
171	isync
172
173	/* Return */
174	mtmsr	r7
175	blr
176
177#endif /* CONFIG_CPU_FREQ_PMAC && CONFIG_PPC_BOOK3S_32 */
178
179/*
180 * Copy a whole page.  We use the dcbz instruction on the destination
181 * to reduce memory traffic (it eliminates the unnecessary reads of
182 * the destination into cache).  This requires that the destination
183 * is cacheable.
184 */
185#define COPY_16_BYTES		\
186	lwz	r6,4(r4);	\
187	lwz	r7,8(r4);	\
188	lwz	r8,12(r4);	\
189	lwzu	r9,16(r4);	\
190	stw	r6,4(r3);	\
191	stw	r7,8(r3);	\
192	stw	r8,12(r3);	\
193	stwu	r9,16(r3)
194
195_GLOBAL(copy_page)
196	rlwinm	r5, r3, 0, L1_CACHE_BYTES - 1
197	addi	r3,r3,-4
198
1990:	twnei	r5, 0	/* WARN if r3 is not cache aligned */
200	EMIT_WARN_ENTRY 0b,__FILE__,__LINE__, BUGFLAG_WARNING
201
202	addi	r4,r4,-4
203
204	li	r5,4
205
206#if MAX_COPY_PREFETCH > 1
207	li	r0,MAX_COPY_PREFETCH
208	li	r11,4
209	mtctr	r0
21011:	dcbt	r11,r4
211	addi	r11,r11,L1_CACHE_BYTES
212	bdnz	11b
213#else /* MAX_COPY_PREFETCH == 1 */
214	dcbt	r5,r4
215	li	r11,L1_CACHE_BYTES+4
216#endif /* MAX_COPY_PREFETCH */
217	li	r0,PAGE_SIZE/L1_CACHE_BYTES - MAX_COPY_PREFETCH
218	crclr	4*cr0+eq
2192:
220	mtctr	r0
2211:
222	dcbt	r11,r4
223	dcbz	r5,r3
224	COPY_16_BYTES
225#if L1_CACHE_BYTES >= 32
226	COPY_16_BYTES
227#if L1_CACHE_BYTES >= 64
228	COPY_16_BYTES
229	COPY_16_BYTES
230#if L1_CACHE_BYTES >= 128
231	COPY_16_BYTES
232	COPY_16_BYTES
233	COPY_16_BYTES
234	COPY_16_BYTES
235#endif
236#endif
237#endif
238	bdnz	1b
239	beqlr
240	crnot	4*cr0+eq,4*cr0+eq
241	li	r0,MAX_COPY_PREFETCH
242	li	r11,4
243	b	2b
244EXPORT_SYMBOL(copy_page)
245
246/*
247 * Extended precision shifts.
248 *
249 * Updated to be valid for shift counts from 0 to 63 inclusive.
250 * -- Gabriel
251 *
252 * R3/R4 has 64 bit value
253 * R5    has shift count
254 * result in R3/R4
255 *
256 *  ashrdi3: arithmetic right shift (sign propagation)
257 *  lshrdi3: logical right shift
258 *  ashldi3: left shift
259 */
260_GLOBAL(__ashrdi3)
261	subfic	r6,r5,32
262	srw	r4,r4,r5	# LSW = count > 31 ? 0 : LSW >> count
263	addi	r7,r5,32	# could be xori, or addi with -32
264	slw	r6,r3,r6	# t1 = count > 31 ? 0 : MSW << (32-count)
265	rlwinm	r8,r7,0,32	# t3 = (count < 32) ? 32 : 0
266	sraw	r7,r3,r7	# t2 = MSW >> (count-32)
267	or	r4,r4,r6	# LSW |= t1
268	slw	r7,r7,r8	# t2 = (count < 32) ? 0 : t2
269	sraw	r3,r3,r5	# MSW = MSW >> count
270	or	r4,r4,r7	# LSW |= t2
271	blr
272EXPORT_SYMBOL(__ashrdi3)
273
274_GLOBAL(__ashldi3)
275	subfic	r6,r5,32
276	slw	r3,r3,r5	# MSW = count > 31 ? 0 : MSW << count
277	addi	r7,r5,32	# could be xori, or addi with -32
278	srw	r6,r4,r6	# t1 = count > 31 ? 0 : LSW >> (32-count)
279	slw	r7,r4,r7	# t2 = count < 32 ? 0 : LSW << (count-32)
280	or	r3,r3,r6	# MSW |= t1
281	slw	r4,r4,r5	# LSW = LSW << count
282	or	r3,r3,r7	# MSW |= t2
283	blr
284EXPORT_SYMBOL(__ashldi3)
285
286_GLOBAL(__lshrdi3)
287	subfic	r6,r5,32
288	srw	r4,r4,r5	# LSW = count > 31 ? 0 : LSW >> count
289	addi	r7,r5,32	# could be xori, or addi with -32
290	slw	r6,r3,r6	# t1 = count > 31 ? 0 : MSW << (32-count)
291	srw	r7,r3,r7	# t2 = count < 32 ? 0 : MSW >> (count-32)
292	or	r4,r4,r6	# LSW |= t1
293	srw	r3,r3,r5	# MSW = MSW >> count
294	or	r4,r4,r7	# LSW |= t2
295	blr
296EXPORT_SYMBOL(__lshrdi3)
297
298/*
299 * 64-bit comparison: __cmpdi2(s64 a, s64 b)
300 * Returns 0 if a < b, 1 if a == b, 2 if a > b.
301 */
302_GLOBAL(__cmpdi2)
303	cmpw	r3,r5
304	li	r3,1
305	bne	1f
306	cmplw	r4,r6
307	beqlr
3081:	li	r3,0
309	bltlr
310	li	r3,2
311	blr
312EXPORT_SYMBOL(__cmpdi2)
313/*
314 * 64-bit comparison: __ucmpdi2(u64 a, u64 b)
315 * Returns 0 if a < b, 1 if a == b, 2 if a > b.
316 */
317_GLOBAL(__ucmpdi2)
318	cmplw	r3,r5
319	li	r3,1
320	bne	1f
321	cmplw	r4,r6
322	beqlr
3231:	li	r3,0
324	bltlr
325	li	r3,2
326	blr
327EXPORT_SYMBOL(__ucmpdi2)
328
329_GLOBAL(__bswapdi2)
330	rotlwi  r9,r4,8
331	rotlwi  r10,r3,8
332	rlwimi  r9,r4,24,0,7
333	rlwimi  r10,r3,24,0,7
334	rlwimi  r9,r4,24,16,23
335	rlwimi  r10,r3,24,16,23
336	mr      r3,r9
337	mr      r4,r10
338	blr
339EXPORT_SYMBOL(__bswapdi2)
340
341#ifdef CONFIG_SMP
342_GLOBAL(start_secondary_resume)
343	/* Reset stack */
344	rlwinm	r1, r1, 0, 0, 31 - THREAD_SHIFT
345	addi	r1,r1,THREAD_SIZE-STACK_FRAME_MIN_SIZE
346	li	r3,0
347	stw	r3,0(r1)		/* Zero the stack frame pointer	*/
348	bl	start_secondary
349	b	.
350#endif /* CONFIG_SMP */
351