xref: /openbmc/linux/drivers/edac/pasemi_edac.c (revision d0e22329)
1 /*
2  * Copyright (C) 2006-2007 PA Semi, Inc
3  *
4  * Author: Egor Martovetsky <egor@pasemi.com>
5  * Maintained by: Olof Johansson <olof@lixom.net>
6  *
7  * Driver for the PWRficient onchip memory controllers
8  *
9  * This program is free software; you can redistribute it and/or modify
10  * it under the terms of the GNU General Public License version 2 as
11  * published by the Free Software Foundation.
12  *
13  * This program is distributed in the hope that it will be useful,
14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
16  * GNU General Public License for more details.
17  *
18  * You should have received a copy of the GNU General Public License
19  * along with this program; if not, write to the Free Software
20  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
21  */
22 
23 
24 #include <linux/module.h>
25 #include <linux/init.h>
26 #include <linux/pci.h>
27 #include <linux/pci_ids.h>
28 #include <linux/edac.h>
29 #include "edac_module.h"
30 
31 #define MODULE_NAME "pasemi_edac"
32 
33 #define MCCFG_MCEN				0x300
34 #define   MCCFG_MCEN_MMC_EN			0x00000001
35 #define MCCFG_ERRCOR				0x388
36 #define   MCCFG_ERRCOR_RNK_FAIL_DET_EN		0x00000100
37 #define   MCCFG_ERRCOR_ECC_GEN_EN		0x00000010
38 #define   MCCFG_ERRCOR_ECC_CRR_EN		0x00000001
39 #define MCCFG_SCRUB				0x384
40 #define   MCCFG_SCRUB_RGLR_SCRB_EN		0x00000001
41 #define MCDEBUG_ERRCTL1				0x728
42 #define   MCDEBUG_ERRCTL1_RFL_LOG_EN		0x00080000
43 #define   MCDEBUG_ERRCTL1_MBE_LOG_EN		0x00040000
44 #define   MCDEBUG_ERRCTL1_SBE_LOG_EN		0x00020000
45 #define MCDEBUG_ERRSTA				0x730
46 #define   MCDEBUG_ERRSTA_RFL_STATUS		0x00000004
47 #define   MCDEBUG_ERRSTA_MBE_STATUS		0x00000002
48 #define   MCDEBUG_ERRSTA_SBE_STATUS		0x00000001
49 #define MCDEBUG_ERRCNT1				0x734
50 #define   MCDEBUG_ERRCNT1_SBE_CNT_OVRFLO	0x00000080
51 #define MCDEBUG_ERRLOG1A			0x738
52 #define   MCDEBUG_ERRLOG1A_MERR_TYPE_M		0x30000000
53 #define   MCDEBUG_ERRLOG1A_MERR_TYPE_NONE	0x00000000
54 #define   MCDEBUG_ERRLOG1A_MERR_TYPE_SBE	0x10000000
55 #define   MCDEBUG_ERRLOG1A_MERR_TYPE_MBE	0x20000000
56 #define   MCDEBUG_ERRLOG1A_MERR_TYPE_RFL	0x30000000
57 #define   MCDEBUG_ERRLOG1A_MERR_BA_M		0x00700000
58 #define   MCDEBUG_ERRLOG1A_MERR_BA_S		20
59 #define   MCDEBUG_ERRLOG1A_MERR_CS_M		0x00070000
60 #define   MCDEBUG_ERRLOG1A_MERR_CS_S		16
61 #define   MCDEBUG_ERRLOG1A_SYNDROME_M		0x0000ffff
62 #define MCDRAM_RANKCFG				0x114
63 #define   MCDRAM_RANKCFG_EN			0x00000001
64 #define   MCDRAM_RANKCFG_TYPE_SIZE_M		0x000001c0
65 #define   MCDRAM_RANKCFG_TYPE_SIZE_S		6
66 
67 #define PASEMI_EDAC_NR_CSROWS			8
68 #define PASEMI_EDAC_NR_CHANS			1
69 #define PASEMI_EDAC_ERROR_GRAIN			64
70 
71 static int last_page_in_mmc;
72 static int system_mmc_id;
73 
74 
75 static u32 pasemi_edac_get_error_info(struct mem_ctl_info *mci)
76 {
77 	struct pci_dev *pdev = to_pci_dev(mci->pdev);
78 	u32 tmp;
79 
80 	pci_read_config_dword(pdev, MCDEBUG_ERRSTA,
81 			      &tmp);
82 
83 	tmp &= (MCDEBUG_ERRSTA_RFL_STATUS | MCDEBUG_ERRSTA_MBE_STATUS
84 		| MCDEBUG_ERRSTA_SBE_STATUS);
85 
86 	if (tmp) {
87 		if (tmp & MCDEBUG_ERRSTA_SBE_STATUS)
88 			pci_write_config_dword(pdev, MCDEBUG_ERRCNT1,
89 					       MCDEBUG_ERRCNT1_SBE_CNT_OVRFLO);
90 		pci_write_config_dword(pdev, MCDEBUG_ERRSTA, tmp);
91 	}
92 
93 	return tmp;
94 }
95 
96 static void pasemi_edac_process_error_info(struct mem_ctl_info *mci, u32 errsta)
97 {
98 	struct pci_dev *pdev = to_pci_dev(mci->pdev);
99 	u32 errlog1a;
100 	u32 cs;
101 
102 	if (!errsta)
103 		return;
104 
105 	pci_read_config_dword(pdev, MCDEBUG_ERRLOG1A, &errlog1a);
106 
107 	cs = (errlog1a & MCDEBUG_ERRLOG1A_MERR_CS_M) >>
108 		MCDEBUG_ERRLOG1A_MERR_CS_S;
109 
110 	/* uncorrectable/multi-bit errors */
111 	if (errsta & (MCDEBUG_ERRSTA_MBE_STATUS |
112 		      MCDEBUG_ERRSTA_RFL_STATUS)) {
113 		edac_mc_handle_error(HW_EVENT_ERR_UNCORRECTED, mci, 1,
114 				     mci->csrows[cs]->first_page, 0, 0,
115 				     cs, 0, -1, mci->ctl_name, "");
116 	}
117 
118 	/* correctable/single-bit errors */
119 	if (errsta & MCDEBUG_ERRSTA_SBE_STATUS)
120 		edac_mc_handle_error(HW_EVENT_ERR_CORRECTED, mci, 1,
121 				     mci->csrows[cs]->first_page, 0, 0,
122 				     cs, 0, -1, mci->ctl_name, "");
123 }
124 
125 static void pasemi_edac_check(struct mem_ctl_info *mci)
126 {
127 	u32 errsta;
128 
129 	errsta = pasemi_edac_get_error_info(mci);
130 	if (errsta)
131 		pasemi_edac_process_error_info(mci, errsta);
132 }
133 
134 static int pasemi_edac_init_csrows(struct mem_ctl_info *mci,
135 				   struct pci_dev *pdev,
136 				   enum edac_type edac_mode)
137 {
138 	struct csrow_info *csrow;
139 	struct dimm_info *dimm;
140 	u32 rankcfg;
141 	int index;
142 
143 	for (index = 0; index < mci->nr_csrows; index++) {
144 		csrow = mci->csrows[index];
145 		dimm = csrow->channels[0]->dimm;
146 
147 		pci_read_config_dword(pdev,
148 				      MCDRAM_RANKCFG + (index * 12),
149 				      &rankcfg);
150 
151 		if (!(rankcfg & MCDRAM_RANKCFG_EN))
152 			continue;
153 
154 		switch ((rankcfg & MCDRAM_RANKCFG_TYPE_SIZE_M) >>
155 			MCDRAM_RANKCFG_TYPE_SIZE_S) {
156 		case 0:
157 			dimm->nr_pages = 128 << (20 - PAGE_SHIFT);
158 			break;
159 		case 1:
160 			dimm->nr_pages = 256 << (20 - PAGE_SHIFT);
161 			break;
162 		case 2:
163 		case 3:
164 			dimm->nr_pages = 512 << (20 - PAGE_SHIFT);
165 			break;
166 		case 4:
167 			dimm->nr_pages = 1024 << (20 - PAGE_SHIFT);
168 			break;
169 		case 5:
170 			dimm->nr_pages = 2048 << (20 - PAGE_SHIFT);
171 			break;
172 		default:
173 			edac_mc_printk(mci, KERN_ERR,
174 				"Unrecognized Rank Config. rankcfg=%u\n",
175 				rankcfg);
176 			return -EINVAL;
177 		}
178 
179 		csrow->first_page = last_page_in_mmc;
180 		csrow->last_page = csrow->first_page + dimm->nr_pages - 1;
181 		last_page_in_mmc += dimm->nr_pages;
182 		csrow->page_mask = 0;
183 		dimm->grain = PASEMI_EDAC_ERROR_GRAIN;
184 		dimm->mtype = MEM_DDR;
185 		dimm->dtype = DEV_UNKNOWN;
186 		dimm->edac_mode = edac_mode;
187 	}
188 	return 0;
189 }
190 
191 static int pasemi_edac_probe(struct pci_dev *pdev,
192 			     const struct pci_device_id *ent)
193 {
194 	struct mem_ctl_info *mci = NULL;
195 	struct edac_mc_layer layers[2];
196 	u32 errctl1, errcor, scrub, mcen;
197 
198 	pci_read_config_dword(pdev, MCCFG_MCEN, &mcen);
199 	if (!(mcen & MCCFG_MCEN_MMC_EN))
200 		return -ENODEV;
201 
202 	/*
203 	 * We should think about enabling other error detection later on
204 	 */
205 
206 	pci_read_config_dword(pdev, MCDEBUG_ERRCTL1, &errctl1);
207 	errctl1 |= MCDEBUG_ERRCTL1_SBE_LOG_EN |
208 		MCDEBUG_ERRCTL1_MBE_LOG_EN |
209 		MCDEBUG_ERRCTL1_RFL_LOG_EN;
210 	pci_write_config_dword(pdev, MCDEBUG_ERRCTL1, errctl1);
211 
212 	layers[0].type = EDAC_MC_LAYER_CHIP_SELECT;
213 	layers[0].size = PASEMI_EDAC_NR_CSROWS;
214 	layers[0].is_virt_csrow = true;
215 	layers[1].type = EDAC_MC_LAYER_CHANNEL;
216 	layers[1].size = PASEMI_EDAC_NR_CHANS;
217 	layers[1].is_virt_csrow = false;
218 	mci = edac_mc_alloc(system_mmc_id++, ARRAY_SIZE(layers), layers,
219 			    0);
220 	if (mci == NULL)
221 		return -ENOMEM;
222 
223 	pci_read_config_dword(pdev, MCCFG_ERRCOR, &errcor);
224 	errcor |= MCCFG_ERRCOR_RNK_FAIL_DET_EN |
225 		MCCFG_ERRCOR_ECC_GEN_EN |
226 		MCCFG_ERRCOR_ECC_CRR_EN;
227 
228 	mci->pdev = &pdev->dev;
229 	mci->mtype_cap = MEM_FLAG_DDR | MEM_FLAG_RDDR;
230 	mci->edac_ctl_cap = EDAC_FLAG_NONE | EDAC_FLAG_EC | EDAC_FLAG_SECDED;
231 	mci->edac_cap = (errcor & MCCFG_ERRCOR_ECC_GEN_EN) ?
232 		((errcor & MCCFG_ERRCOR_ECC_CRR_EN) ?
233 		 (EDAC_FLAG_EC | EDAC_FLAG_SECDED) : EDAC_FLAG_EC) :
234 		EDAC_FLAG_NONE;
235 	mci->mod_name = MODULE_NAME;
236 	mci->dev_name = pci_name(pdev);
237 	mci->ctl_name = "pasemi,pwrficient-mc";
238 	mci->edac_check = pasemi_edac_check;
239 	mci->ctl_page_to_phys = NULL;
240 	pci_read_config_dword(pdev, MCCFG_SCRUB, &scrub);
241 	mci->scrub_cap = SCRUB_FLAG_HW_PROG | SCRUB_FLAG_HW_SRC;
242 	mci->scrub_mode =
243 		((errcor & MCCFG_ERRCOR_ECC_CRR_EN) ? SCRUB_FLAG_HW_SRC : 0) |
244 		((scrub & MCCFG_SCRUB_RGLR_SCRB_EN) ? SCRUB_FLAG_HW_PROG : 0);
245 
246 	if (pasemi_edac_init_csrows(mci, pdev,
247 				    (mci->edac_cap & EDAC_FLAG_SECDED) ?
248 				    EDAC_SECDED :
249 				    ((mci->edac_cap & EDAC_FLAG_EC) ?
250 				     EDAC_EC : EDAC_NONE)))
251 		goto fail;
252 
253 	/*
254 	 * Clear status
255 	 */
256 	pasemi_edac_get_error_info(mci);
257 
258 	if (edac_mc_add_mc(mci))
259 		goto fail;
260 
261 	/* get this far and it's successful */
262 	return 0;
263 
264 fail:
265 	edac_mc_free(mci);
266 	return -ENODEV;
267 }
268 
269 static void pasemi_edac_remove(struct pci_dev *pdev)
270 {
271 	struct mem_ctl_info *mci = edac_mc_del_mc(&pdev->dev);
272 
273 	if (!mci)
274 		return;
275 
276 	edac_mc_free(mci);
277 }
278 
279 
280 static const struct pci_device_id pasemi_edac_pci_tbl[] = {
281 	{ PCI_DEVICE(PCI_VENDOR_ID_PASEMI, 0xa00a) },
282 	{ }
283 };
284 
285 MODULE_DEVICE_TABLE(pci, pasemi_edac_pci_tbl);
286 
287 static struct pci_driver pasemi_edac_driver = {
288 	.name = MODULE_NAME,
289 	.probe = pasemi_edac_probe,
290 	.remove = pasemi_edac_remove,
291 	.id_table = pasemi_edac_pci_tbl,
292 };
293 
294 static int __init pasemi_edac_init(void)
295 {
296        /* Ensure that the OPSTATE is set correctly for POLL or NMI */
297        opstate_init();
298 
299 	return pci_register_driver(&pasemi_edac_driver);
300 }
301 
302 static void __exit pasemi_edac_exit(void)
303 {
304 	pci_unregister_driver(&pasemi_edac_driver);
305 }
306 
307 module_init(pasemi_edac_init);
308 module_exit(pasemi_edac_exit);
309 
310 MODULE_LICENSE("GPL");
311 MODULE_AUTHOR("Egor Martovetsky <egor@pasemi.com>");
312 MODULE_DESCRIPTION("MC support for PA Semi PWRficient memory controller");
313 module_param(edac_op_state, int, 0444);
314 MODULE_PARM_DESC(edac_op_state, "EDAC Error Reporting state: 0=Poll,1=NMI");
315 
316