memerr.c revision 1.9 1 /* $NetBSD: memerr.c,v 1.9 1998/01/12 20:32:22 thorpej Exp $ */
2
3 /*
4 * Copyright (c) 1992, 1993
5 * The Regents of the University of California. All rights reserved.
6 *
7 * This software was developed by the Computer Systems Engineering group
8 * at Lawrence Berkeley Laboratory under DARPA contract BG 91-66 and
9 * contributed to Berkeley.
10 *
11 * All advertising materials mentioning features or use of this software
12 * must display the following acknowledgement:
13 * This product includes software developed by the University of
14 * California, Lawrence Berkeley Laboratory.
15 *
16 * Redistribution and use in source and binary forms, with or without
17 * modification, are permitted provided that the following conditions
18 * are met:
19 * 1. Redistributions of source code must retain the above copyright
20 * notice, this list of conditions and the following disclaimer.
21 * 2. Redistributions in binary form must reproduce the above copyright
22 * notice, this list of conditions and the following disclaimer in the
23 * documentation and/or other materials provided with the distribution.
24 * 3. All advertising materials mentioning features or use of this software
25 * must display the following acknowledgement:
26 * This product includes software developed by the University of
27 * California, Berkeley and its contributors.
28 * 4. Neither the name of the University nor the names of its contributors
29 * may be used to endorse or promote products derived from this software
30 * without specific prior written permission.
31 *
32 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
33 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
34 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
35 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
36 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
37 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
38 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
39 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
40 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
41 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
42 * SUCH DAMAGE.
43 *
44 * @(#)memreg.c 8.1 (Berkeley) 6/11/93
45 */
46
47 #include <sys/param.h>
48 #include <sys/systm.h>
49 #include <sys/device.h>
50
51 #include <machine/autoconf.h>
52 #include <machine/control.h>
53 #include <machine/cpu.h>
54 #include <machine/idprom.h>
55 #include <machine/obio.h>
56 #include <machine/pte.h>
57
58 #include <sun3/dev/memerr.h>
59 /* #include <sun3/dev/eccreg.h> - not yet */
60
61 #define ME_PRI 7 /* Interrupt level (NMI) */
62
63 extern unsigned char cpu_machine_id;
64
65 enum memerr_type { ME_PAR = 0, ME_ECC = 1 };
66
67 struct memerr_softc {
68 struct device sc_dev;
69 struct memerr *sc_reg;
70 enum memerr_type sc_type;
71 char *sc_typename; /* "Parity" or "ECC" */
72 char *sc_csrbits; /* how to print csr bits */
73 /* XXX: counters? */
74 };
75
76 static int memerr_match __P((struct device *, struct cfdata *, void *));
77 static void memerr_attach __P((struct device *, struct device *, void *));
78 static int memerr_interrupt __P((void *));
79 static void memerr_correctable __P((struct memerr_softc *));
80
81 struct cfattach memerr_ca = {
82 sizeof(struct memerr_softc), memerr_match, memerr_attach
83 };
84
85 static int
86 memerr_match(parent, cf, args)
87 struct device *parent;
88 struct cfdata *cf;
89 void *args;
90 {
91 struct confargs *ca = args;
92
93 /* This driver only supports one unit. */
94 if (cf->cf_unit != 0)
95 return (0);
96
97 /* We use obio_mapin(), so require OBIO. */
98 if (ca->ca_bustype != BUS_OBIO)
99 return (0);
100
101 /* Make sure there is something there... */
102 if (bus_peek(ca->ca_bustype, ca->ca_paddr, 1) == -1)
103 return (0);
104
105 /* Default interrupt priority. */
106 if (ca->ca_intpri == -1)
107 ca->ca_intpri = ME_PRI;
108
109 return (1);
110 }
111
112 static void
113 memerr_attach(parent, self, args)
114 struct device *parent;
115 struct device *self;
116 void *args;
117 {
118 struct memerr_softc *sc = (void *)self;
119 struct confargs *ca = args;
120 struct memerr *mer;
121
122 /*
123 * Which type of memory subsystem do we have?
124 */
125 switch (cpu_machine_id) {
126 case SUN3_MACH_160: /* XXX: correct? */
127 case SUN3_MACH_260:
128 sc->sc_type = ME_ECC;
129 sc->sc_typename = "ECC";
130 sc->sc_csrbits = ME_ECC_STR;
131 break;
132
133 default:
134 sc->sc_type = ME_PAR;
135 sc->sc_typename = "Parity";
136 sc->sc_csrbits = ME_PAR_STR;
137 break;
138 }
139 printf(": (%s memory)\n", sc->sc_typename);
140
141 mer = (struct memerr *)
142 obio_mapin(ca->ca_paddr, sizeof(*mer));
143 if (mer == NULL)
144 panic("memerr: can not map register");
145 sc->sc_reg = mer;
146
147 /* Install interrupt handler. */
148 isr_add_autovect(memerr_interrupt,
149 (void *)sc, ca->ca_intpri);
150
151 /* Enable error interrupt (and checking). */
152 if (sc->sc_type == ME_PAR)
153 mer->me_csr = ME_CSR_IENA | ME_PAR_CHECK;
154 else {
155 /*
156 * XXX: Some day, figure out how to decode
157 * correctable errors and set ME_ECC_CE_ENA
158 * here so we can log them...
159 */
160 mer->me_csr = ME_CSR_IENA; /* | ME_ECC_CE_ENA */
161 }
162 }
163
164 /*****************************************************************
165 * Functions for ECC memory
166 *****************************************************************/
167
168 static int
169 memerr_interrupt(arg)
170 void *arg;
171 {
172 struct memerr_softc *sc = arg;
173 volatile struct memerr *me = sc->sc_reg;
174 u_char csr, ctx;
175 u_int pa, va;
176 int pte;
177 char bits[64];
178
179 csr = me->me_csr;
180 if ((csr & ME_CSR_IPEND) == 0)
181 return (0);
182
183 va = me->me_vaddr;
184 ctx = (va >> 28) & 0xF;
185 va &= 0x0FFFffff;
186 pte = get_pte(va);
187 pa = PG_PA(pte);
188
189 printf("\nMemory error on %s cycle!\n",
190 (ctx & 8) ? "DVMA" : "CPU");
191 printf(" ctx=%d, vaddr=0x%x, paddr=0x%x\n",
192 (ctx & 7), va, pa);
193 printf(" csr=%s\n", bitmask_snprintf(csr, sc->sc_csrbits,
194 bits, sizeof(bits)));
195
196 /*
197 * If we have parity-checked memory, there is
198 * not much to be done. Any error is fatal.
199 */
200 if (sc->sc_type == ME_PAR) {
201 if (csr & ME_PAR_EMASK) {
202 /* Parity errors are fatal. */
203 goto die;
204 }
205 /* The IPEND bit was set, but no error bits. */
206 goto noerror;
207 }
208
209 /*
210 * We have ECC memory. More complicated...
211 */
212 if (csr & (ME_ECC_WBTMO | ME_ECC_WBERR)) {
213 printf(" write-back failed, pte=0x%x\n", pte);
214 goto die;
215 }
216 if (csr & ME_ECC_UE) {
217 printf(" uncorrectable ECC error\n");
218 goto die;
219 }
220 if (csr & ME_ECC_CE) {
221 /* Just log this and continue. */
222 memerr_correctable(sc);
223 goto recover;
224 }
225 /* The IPEND bit was set, but no error bits. */
226 goto noerror;
227
228 die:
229 panic("all bets are off...");
230
231 noerror:
232 printf("memerr: no error bits set?\n");
233
234 recover:
235 /* Clear the error by writing the address register. */
236 me->me_vaddr = 0;
237 return (1);
238 }
239
240 /*
241 * Announce (and log) a correctable ECC error.
242 * Need to look at the ECC syndrome register on
243 * the memory board that caused the error...
244 */
245 void
246 memerr_correctable(sc)
247 struct memerr_softc *sc;
248 {
249 /* XXX: Not yet... */
250 }
251