Home | History | Annotate | Line # | Download | only in dev
      1 /*	$NetBSD: ecc_plb.c,v 1.16 2021/02/27 20:43:58 rin Exp $	*/
      2 
      3 /*
      4  * Copyright 2001 Wasabi Systems, Inc.
      5  * All rights reserved.
      6  *
      7  * Written by Eduardo Horvath and Simon Burge for Wasabi Systems, Inc.
      8  *
      9  * Redistribution and use in source and binary forms, with or without
     10  * modification, are permitted provided that the following conditions
     11  * are met:
     12  * 1. Redistributions of source code must retain the above copyright
     13  *    notice, this list of conditions and the following disclaimer.
     14  * 2. Redistributions in binary form must reproduce the above copyright
     15  *    notice, this list of conditions and the following disclaimer in the
     16  *    documentation and/or other materials provided with the distribution.
     17  * 3. All advertising materials mentioning features or use of this software
     18  *    must display the following acknowledgement:
     19  *      This product includes software developed for the NetBSD Project by
     20  *      Wasabi Systems, Inc.
     21  * 4. The name of Wasabi Systems, Inc. may not be used to endorse
     22  *    or promote products derived from this software without specific prior
     23  *    written permission.
     24  *
     25  * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
     26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
     27  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
     28  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL WASABI SYSTEMS, INC
     29  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
     30  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
     31  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
     32  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
     33  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
     34  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
     35  * POSSIBILITY OF SUCH DAMAGE.
     36  */
     37 
     38 #include <sys/cdefs.h>
     39 __KERNEL_RCSID(0, "$NetBSD: ecc_plb.c,v 1.16 2021/02/27 20:43:58 rin Exp $");
     40 
     41 #include "locators.h"
     42 
     43 #include <sys/param.h>
     44 #include <sys/systm.h>
     45 #include <sys/device.h>
     46 #include <sys/cpu.h>
     47 
     48 #include <prop/proplib.h>
     49 
     50 #include <powerpc/ibm4xx/cpu.h>
     51 #include <powerpc/ibm4xx/dcr4xx.h>
     52 #include <powerpc/ibm4xx/dev/plbvar.h>
     53 
     54 
     55 struct ecc_plb_softc {
     56 	device_t sc_dev;
     57 	uint64_t sc_ecc_tb;
     58 	uint64_t sc_ecc_iv;	 /* Interval */
     59 	uint32_t sc_ecc_cnt;
     60 	u_int sc_memsize;
     61 	int sc_irq;
     62 };
     63 
     64 static int	ecc_plbmatch(device_t, cfdata_t, void *);
     65 static void	ecc_plbattach(device_t, device_t, void *);
     66 static void	ecc_plb_deferred(device_t);
     67 static int	ecc_plb_intr(void *);
     68 
     69 CFATTACH_DECL_NEW(ecc_plb, sizeof(struct ecc_plb_softc),
     70     ecc_plbmatch, ecc_plbattach, NULL, NULL);
     71 
     72 static int ecc_plb_found;
     73 
     74 static int
     75 ecc_plbmatch(device_t parent, cfdata_t cf, void *aux)
     76 {
     77 	struct plb_attach_args *paa = aux;
     78 
     79 	if (strcmp(paa->plb_name, cf->cf_name) != 0)
     80 		return (0);
     81 
     82 	if (cf->cf_loc[PLBCF_IRQ] == PLBCF_IRQ_DEFAULT)
     83 		panic("ecc_plbmatch: wildcard IRQ not allowed");
     84 
     85 	paa->plb_irq = cf->cf_loc[PLBCF_IRQ];
     86 
     87 	return (!ecc_plb_found);
     88 }
     89 
     90 static void
     91 ecc_plbattach(device_t parent, device_t self, void *aux)
     92 {
     93 	struct ecc_plb_softc *sc = device_private(self);
     94 	struct plb_attach_args *paa = aux;
     95 	unsigned int processor_freq;
     96 	unsigned int memsiz;
     97 	prop_number_t pn;
     98 
     99 	ecc_plb_found++;
    100 
    101 	pn = prop_dictionary_get(board_properties, "processor-frequency");
    102 	KASSERT(pn != NULL);
    103 	processor_freq = (unsigned int) prop_number_integer_value(pn);
    104 
    105 	pn = prop_dictionary_get(board_properties, "mem-size");
    106 	KASSERT(pn != NULL);
    107 	memsiz = (unsigned int) prop_number_integer_value(pn);
    108 
    109 	aprint_normal(": ECC controller\n");
    110 
    111 	sc->sc_dev = self;
    112 	sc->sc_ecc_tb = 0;
    113 	sc->sc_ecc_cnt = 0;
    114 	sc->sc_ecc_iv = processor_freq; /* Set interval */
    115 	sc->sc_memsize = memsiz;
    116 	sc->sc_irq = paa->plb_irq;
    117 
    118 	/*
    119 	 * Defer hooking the interrupt until all PLB devices have attached
    120 	 * since the interrupt controller may well be one of those devices...
    121 	 */
    122 	config_defer(self, ecc_plb_deferred);
    123 }
    124 
    125 static void
    126 ecc_plb_deferred(device_t self)
    127 {
    128 	struct ecc_plb_softc *sc = device_private(self);
    129 
    130 	intr_establish_xname(sc->sc_irq, IST_LEVEL, IPL_SERIAL, ecc_plb_intr,
    131 	    sc, device_xname(self));
    132 }
    133 
    134 /*
    135  * ECC fault handler.
    136  */
    137 static int
    138 ecc_plb_intr(void *arg)
    139 {
    140 	struct ecc_plb_softc *sc = arg;
    141 	u_int32_t		esr, ear;
    142 	int			ue;
    143 	u_quad_t		tb;
    144 	u_long			tmp, msr, dat;
    145 
    146 	/* This code needs to be improved to handle double-bit errors */
    147 	/* in some intelligent fashion. */
    148 
    149 	mtdcr(DCR_SDRAM0_CFGADDR, DCR_SDRAM0_ECCESR);
    150 	esr = mfdcr(DCR_SDRAM0_CFGDATA);
    151 
    152 	mtdcr(DCR_SDRAM0_CFGADDR, DCR_SDRAM0_BEAR);
    153 	ear = mfdcr(DCR_SDRAM0_CFGDATA);
    154 
    155 	/* Always clear the error to stop the intr ASAP. */
    156 
    157 	mtdcr(DCR_SDRAM0_CFGADDR, DCR_SDRAM0_ECCESR);
    158 	mtdcr(DCR_SDRAM0_CFGDATA, 0xffffffff);
    159 
    160 	if (esr == 0x00) {
    161 		/* No current error.  Could happen due to intr. nesting */
    162 		return(1);
    163 	}
    164 
    165 	/*
    166 	 * Only report errors every once per second max. Do this using the TB,
    167 	 * because the system time (via microtime) may be adjusted when the
    168 	 * date is set and can't reliably be used to measure intervals.
    169 	 */
    170 
    171 	__asm ("1: mftbu %0; mftb %0+1; mftbu %1; cmpw %0,%1; bne 1b"
    172 		: "=r"(tb), "=r"(tmp));
    173 	sc->sc_ecc_cnt++;
    174 
    175 	if ((tb - sc->sc_ecc_tb) < sc->sc_ecc_iv)
    176 		return(1);
    177 
    178 	ue = (esr & SDRAM0_ECCESR_UE) != 0x00;
    179 
    180 	printf("ECC: Error CNT=%d ESR=%x EAR=%x %s BKNE=%d%d%d%d "
    181 		"BLCE=%d%d%d%d CBE=%d%d.\n",
    182 		sc->sc_ecc_cnt, esr, ear,
    183 		(ue) ? "Uncorrectable" : "Correctable",
    184 		((esr & SDRAM0_ECCESR_BKEN(0)) != 0x00),
    185 		((esr & SDRAM0_ECCESR_BKEN(1)) != 0x00),
    186 		((esr & SDRAM0_ECCESR_BKEN(2)) != 0x00),
    187 		((esr & SDRAM0_ECCESR_BKEN(3)) != 0x00),
    188 		((esr & SDRAM0_ECCESR_BLCEN(0)) != 0x00),
    189 		((esr & SDRAM0_ECCESR_BLCEN(1)) != 0x00),
    190 		((esr & SDRAM0_ECCESR_BLCEN(2)) != 0x00),
    191 		((esr & SDRAM0_ECCESR_BLCEN(3)) != 0x00),
    192 		((esr & SDRAM0_ECCESR_CBEN(0)) != 0x00),
    193 		((esr & SDRAM0_ECCESR_CBEN(1)) != 0x00));
    194 
    195 	/* Should check for uncorrectable errors and panic... */
    196 
    197 	if (sc->sc_ecc_cnt > 1000) {
    198 		printf("ECC: Too many errors, recycling entire "
    199 			"SDRAM (size = %d).\n", sc->sc_memsize);
    200 
    201 		/*
    202 		 * Can this code be changed to run without disabling data MMU
    203 		 * and disabling intrs?
    204 		 * Does kernel always map all of physical RAM VA=PA? If so,
    205 		 * just loop over lowmem.
    206 		 */
    207 		__asm volatile(
    208 			"mfmsr 	%0;"
    209 			"li	%1, 0x00;"
    210 			"ori	%1, %1, 0x8010;"
    211 			"andc	%1, %0, %1;"
    212 			"mtmsr	%1;"
    213 			"sync;isync;"
    214 			"li	%1, 0x00;"
    215 			"1:"
    216 			"dcbt	0, %1;"
    217 			"sync;isync;"
    218 			"lwz	%2, 0(%1);"
    219 			"stw	%2, 0(%1);"
    220 			"sync;isync;"
    221 			"dcbf	0, %1;"
    222 			"sync;isync;"
    223 			"addi	%1, %1, 0x20;"
    224 			"addic.	%3, %3, -0x20;"
    225 			"bge 	1b;"
    226 			"mtmsr %0;"
    227 			"sync;isync;"
    228 		: "=&r" (msr), "=&r" (tmp), "=&r" (dat)
    229 		: "r" (sc->sc_memsize) : "0" );
    230 
    231 		mtdcr(DCR_SDRAM0_CFGADDR, DCR_SDRAM0_ECCESR);
    232 		esr = mfdcr(DCR_SDRAM0_CFGDATA);
    233 
    234 		mtdcr(DCR_SDRAM0_CFGADDR, DCR_SDRAM0_ECCESR);
    235 		mtdcr(DCR_SDRAM0_CFGDATA, 0xffffffff);
    236 
    237 		/*
    238 		 * Correctable errors here are OK, mem should be clean now.
    239 		 *
    240 		 * Should check for uncorrectable errors and panic...
    241 		 */
    242 		printf("ECC: Recycling complete, ESR=%x. "
    243 			"Checking for persistent errors.\n", esr);
    244 
    245 		__asm volatile(
    246 			"mfmsr 	%0;"
    247 			"li	%1, 0x00;"
    248 			"ori	%1, %1, 0x8010;"
    249 			"andc	%1, %0, %1;"
    250 			"mtmsr	%1;"
    251 			"sync;isync;"
    252 			"li	%1, 0x00;"
    253 			"1:"
    254 			"dcbt	0, %1;"
    255 			"sync;isync;"
    256 			"lwz	%2, 0(%1);"
    257 			"stw	%2, 0(%1);"
    258 			"sync;isync;"
    259 			"dcbf	0, %1;"
    260 			"sync;isync;"
    261 			"addi	%1, %1, 0x20;"
    262 			"addic.	%3, %3, -0x20;"
    263 			"bge 	1b;"
    264 			"mtmsr %0;"
    265 			"sync;isync;"
    266 		: "=&r" (msr), "=&r" (tmp), "=&r" (dat)
    267 		: "r" (sc->sc_memsize) : "0" );
    268 
    269 		mtdcr(DCR_SDRAM0_CFGADDR, DCR_SDRAM0_ECCESR);
    270 		esr = mfdcr(DCR_SDRAM0_CFGDATA);
    271 
    272 		/*
    273 		 * If esr is non zero here, we're screwed.
    274 		 * Should check this and panic.
    275 		 */
    276 		printf("ECC: Persistent error check complete, "
    277 			"final ESR=%x.\n", esr);
    278 	}
    279 
    280 	sc->sc_ecc_tb = tb;
    281 	sc->sc_ecc_cnt = 0;
    282 
    283 	return(1);
    284 }
    285