Home | History | Annotate | Download | only in cpumem-diagnosis
      1 /*
      2  * CDDL HEADER START
      3  *
      4  * The contents of this file are subject to the terms of the
      5  * Common Development and Distribution License (the "License").
      6  * You may not use this file except in compliance with the License.
      7  *
      8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
      9  * or http://www.opensolaris.org/os/licensing.
     10  * See the License for the specific language governing permissions
     11  * and limitations under the License.
     12  *
     13  * When distributing Covered Code, include this CDDL HEADER in each
     14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
     15  * If applicable, add the following below this CDDL HEADER, with the
     16  * fields enclosed by brackets "[]" replaced with your own identifying
     17  * information: Portions Copyright [yyyy] [name of copyright owner]
     18  *
     19  * CDDL HEADER END
     20  */
     21 /*
     22  * Copyright 2008 Sun Microsystems, Inc.  All rights reserved.
     23  * Use is subject to license terms.
     24  */
     25 
     26 #ifndef _CMD_DIMM_H
     27 #define	_CMD_DIMM_H
     28 
     29 /*
     30  * Memory modules are described by the cmd_dimm general-purpose state structure.
     31  * Whereas banks are primarily used to track UEs, this structure is used to
     32  * track CEs, which can be associated with individual modules.  Each memory
     33  * module is part of a bank, and will have a link to the bank if the bank is
     34  * known to the diagnosis engine.  Banks will be known if UEs have occurred.
     35  *
     36  * Data structures:
     37  *
     38  *     ,--------.       ,--------.
     39  *     |dimm    | <---- |case_ptr| (CMD_PTR_DIMM_CASE)
     40  *     |        |       `--------'
     41  *     |,-------|       ,-------------.
     42  *  ,->||asru_t | ----> |packed nvlist|
     43  *  |  |`-------|       `-------------'
     44  *  `--| unum   |
     45  *     | bank   | ----> bank buffer
     46  *     `--------'
     47  *
     48  * Data structure	P?  Case? Notes
     49  * ----------------	--- ----- ----------------------------------------------
     50  * cmd_dimm_t		Yes No    Name is derived from the unum ("dimm_%s")
     51  * cmd_case_ptr_t	Yes Yes   Name is case's UUID
     52  * dimm_asru		Yes No    Name is derived from the unum ("dimm_asru_%d")
     53  * dimm_unum		No  No    Pointer into ASRU - relinked during restore
     54  * dimm_bank		No  No    Recreated during restore
     55  */
     56 
     57 #include <cmd_mem.h>
     58 #include <values.h>
     59 
     60 #ifdef __cplusplus
     61 extern "C" {
     62 #endif
     63 
     64 /*
     65  * CMD_MAX_CKWDS denotes the highest number, across all covered
     66  * SPARC architectures, of checkwords per cache line.
     67  */
     68 
     69 #define	CMD_MAX_CKWDS	4
     70 
     71 /*
     72  * The DIMM structure started life without a version number.  Making things more
     73  * complicated, the version number in the new struct occupies the space used for
     74  * the case pointer in the non-versioned struct.  We therefore have to use
     75  * somewhat unorthodox version numbers so as to allow us to easily tell the
     76  * difference between a version number and a case pointer.  Case pointers will
     77  * be zero or (this being SPARC), a value with the bottom two bits clear.  Our
     78  * version numbers will begin with 0x11, and will increase by 0x10 each time.
     79  */
     80 
     81 #define	DIMM_MKVERSION(version)	((version) << 4 | 1)
     82 
     83 #define	CMD_DIMM_VERSION_1	DIMM_MKVERSION(1)	/* 17 */
     84 #define	CMD_DIMM_VERSION	CMD_DIMM_VERSION_1
     85 
     86 #define	CMD_DIMM_VERSIONED(dimm)	((dimm)->dimm_version & 1)
     87 
     88 #define	CMD_DIMM_STAT_PREFIX		"d"	/* d = dimm */
     89 
     90 typedef struct cmd_dimm_0 {
     91 	cmd_header_t dimm0_header;	/* Nodetype must be CMD_NT_DIMM */
     92 	fmd_case_t *dimm0_case;		/* Open CE case against this DIMM */
     93 	cmd_fmri_t dimm0_asru;		/* ASRU for this DIMM */
     94 	const char *dimm0_unum;		/* This DIMM's name */
     95 	uint_t dimm0_wrnthresh;		/* # of pages retired before warning */
     96 	uint_t dimm0_nretired;		/* # ret'd pages for CEs in DIMM */
     97 	cmd_bank_t *dimm0_bank;		/* This DIMM's bank (if discovered) */
     98 } cmd_dimm_0_t;
     99 
    100 typedef struct cmd_dimm_pers {
    101 	cmd_header_t dimmp_header;	/* Nodetype must be CMD_NT_DIMM */
    102 	uint_t dimmp_version;
    103 	cmd_fmri_t dimmp_asru;		/* ASRU for this DIMM */
    104 	uint_t dimmp_flags;		/* CMD_MEM_F_* */
    105 	uint_t dimmp_nretired;		/* # ret'd pages for CEs in DIMM */
    106 } cmd_dimm_pers_t;
    107 
    108 /*
    109  * Index block for MQSC rules 4A and 4B correlation of memory CEs
    110  * on a single DIMM. "Unit Position" refers to bit or nibble depending
    111  * on the memory ECC.  This structure is not persisted.
    112  */
    113 
    114 typedef struct cmd_mq {
    115 	cmd_list_t mq_l;		/* pointers to prev and next */
    116 	uint64_t mq_tstamp;		/* timestamp of ereport in secs */
    117 	uint16_t mq_ckwd;		/* phys addr mod 64 */
    118 	uint64_t mq_phys_addr;		/* from ereport */
    119 	uint16_t mq_unit_position;	/* bit for sun4u, nibble for sun4v */
    120 	int16_t mq_dram;		/* by table lookup from unit pos */
    121 	fmd_event_t *mq_ep;		/* ereport - for potential fault */
    122 	char *mq_serdnm;		/* serd eng to retain CE events */
    123 } cmd_mq_t;
    124 
    125 struct cmd_dimm {
    126 	cmd_dimm_pers_t dimm_pers;
    127 	cmd_bank_t *dimm_bank;		/* This DIMM's bank (if discovered) */
    128 	const char *dimm_unum;		/* This DIMM's name */
    129 	cmd_case_t dimm_case;		/* Open CE case against this DIMM */
    130 	fmd_stat_t dimm_retstat;	/* retirement statistics, this DIMM */
    131 	cmd_list_t
    132 	    mq_root[CMD_MAX_CKWDS];	/* per-checkword CEs to correlate */
    133 };
    134 
    135 #define	CMD_MQ_TIMELIM	(72*60*60)	/* 72 hours */
    136 #define	CMD_MQ_SERDT	MAXINT		/* Never expected to fire */
    137 #define	CMD_MQ_SERDN	2		/* Dup CEs not allowed */
    138 
    139 #define	CMD_DIMM_MAXSIZE \
    140 	MAX(sizeof (cmd_dimm_0_t), sizeof (cmd_dimm_pers_t))
    141 #define	CMD_DIMM_MINSIZE \
    142 	MIN(sizeof (cmd_dimm_0_t), sizeof (cmd_dimm_pers_t))
    143 
    144 #define	dimm_header		dimm_pers.dimmp_header
    145 #define	dimm_nodetype		dimm_pers.dimmp_header.hdr_nodetype
    146 #define	dimm_bufname		dimm_pers.dimmp_header.hdr_bufname
    147 #define	dimm_version		dimm_pers.dimmp_version
    148 #define	dimm_asru		dimm_pers.dimmp_asru
    149 #define	dimm_asru_nvl		dimm_pers.dimmp_asru.fmri_nvl
    150 #define	dimm_flags		dimm_pers.dimmp_flags
    151 #define	dimm_nretired		dimm_pers.dimmp_nretired
    152 
    153 extern cmd_dimm_t *cmd_dimm_lookup(fmd_hdl_t *, nvlist_t *);
    154 extern cmd_dimm_t *cmd_dimm_create(fmd_hdl_t *, nvlist_t *);
    155 
    156 extern nvlist_t *cmd_dimm_fru(cmd_dimm_t *);
    157 extern nvlist_t *cmd_dimm_create_fault(fmd_hdl_t *, cmd_dimm_t *, const char *,
    158     uint_t);
    159 #ifdef sun4v
    160 extern nvlist_t *cmd_mem2hc(fmd_hdl_t *, nvlist_t *);
    161 #endif /* sun4v */
    162 
    163 extern nvlist_t *cmd_dimm_fmri_derive(fmd_hdl_t *, uint64_t, uint16_t,
    164     uint64_t);
    165 extern int cmd_dimm_thresh_reached(fmd_hdl_t *, cmd_dimm_t *, uint64_t,
    166     uint16_t);
    167 
    168 extern void cmd_dimm_dirty(fmd_hdl_t *, cmd_dimm_t *);
    169 extern void *cmd_dimm_restore(fmd_hdl_t *, fmd_case_t *, cmd_case_ptr_t *);
    170 extern void cmd_dimm_destroy(fmd_hdl_t *, cmd_dimm_t *);
    171 extern void cmd_dimm_validate(fmd_hdl_t *);
    172 extern void cmd_dimm_gc(fmd_hdl_t *);
    173 extern void cmd_dimm_fini(fmd_hdl_t *);
    174 
    175 extern void cmd_dimmlist_free(fmd_hdl_t *);
    176 
    177 #ifdef __cplusplus
    178 }
    179 #endif
    180 
    181 #endif /* _CMD_DIMM_H */
    182