cmd_mem.h revision 7c478bd95313f5f23a4c958a745db2134aa03244
/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License, Version 1.0 only
* (the "License"). You may not use this file except in compliance
* with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Copyright 2005 Sun Microsystems, Inc. All rights reserved.
* Use is subject to license terms.
*/
#ifndef _CMD_MEM_H
#define _CMD_MEM_H
#pragma ident "%Z%%M% %I% %E% SMI"
/*
* Support routines for managing state related to memory modules.
*
* Correctable errors generally cause changes to the DIMM-related state (see
* cmd_dimm.c), whereas uncorrectable errors tend to use the bank-related
* routines (see cmd_bank.c). The primary exception to this division (though
* UltraSPARC-IIIi processors. With these errors, a complete pair must be
* received and matched before we know whether we're dealing with a CE or a UE.
*/
#include <cmd.h>
#include <cmd_state.h>
#include <cmd_fmri.h>
#include <sys/errclassify.h>
#ifdef __cplusplus
extern "C" {
#endif
#define CMD_MEM_F_FAULTING 0x1
/*
* Used to store as-yet unmatched IOxEs, RxEs, and FRxs. When a new IOxE,
* RxE or FRx arrives, we traverse the cmd.cmd_iorxefrx list, looking for
* matching entries. Matching has a cpuid-based component, as well as a
* temporal one. We can compare the cpuids directly, using the cmd_iorxefrx_t
* and the newly-received event. Temporal comparison isn't performed directly.
* Instead, we ensure that entries in the iorxefrx list are removed when they
* expire by means of timers. This frees the matching code from the need to
* worry about time.
*/
typedef struct cmd_iorxefrx {
typedef struct cmd_dimm cmd_dimm_t;
typedef struct cmd_bank cmd_bank_t;
/*
* Correctable and Uncorrectable memory errors
*
* CEs of "Unknown" or "Intermittent" classification are not used in diagnosis.
*
* "Persistent" CEs are added to per-DIMM SERD engines. When the
* engine for a given DIMM fires, the page corresponding to the CE that
* caused the engine to fire is retired, and the SERD engine for that
* DIMM is reset.
*
* "Possibly Persistent" CEs are at least Persistent and so are treated
* as "Persistent" errors above, being added to the same SERD engines.
*
* "Leaky" CEs and "Sticky" CEs trigger immediate page retirement.
*
* "Possibly Sticky" CEs to which no valid partner test has been applied
* are not used in diagnosis. Where a valid partner test has been applied
* but did not confirm "Sticky" status there is a _suggestion_ that the
* original cpu may be a bad reader or writer or suffering from other
* datapath issues. To avoid retiring pages for such non-DIMM problems
* these classifications are also not used in diagnosis.
*
* UEs immediately trigger page retirements, but do not affect the CE SERD
* engines. In addition, UEs are recorded in the UE caches of the detecting
* CPUs. When a page is to be retired, a fault.memory.page fault is
* generated.
*
*/
const char *, cmd_errcl_t);
const char *, cmd_errcl_t);
extern void cmd_dimm_close(fmd_hdl_t *, void *);
extern void cmd_bank_close(fmd_hdl_t *, void *);
/*
* US-IIIi I/O, Remote and Foreign Read memory errors
*
* When one processor or I/O bridge attempts to read memory local to
* are needed, as each contains data necessary to the diagnosis of the error.
* Upon receipt of one of the errors, we wait until we receive the other.
* When the pair has been successfully received and matched, a CE or UE,
* as appropriate, is synthesized from the data in the matched ereports.
* The synthesized ereports are handled by the normal CE and UE mechanisms.
*/
const char *, cmd_errcl_t);
const char *, cmd_errcl_t);
const char *, cmd_errcl_t);
const char *, cmd_errcl_t);
/*
* purposes (see above). These two routines manage the addition of new
* ereports, and the retrieval of existing ones. Pruning of the list is
* handled automatically.
*/
extern const char *cmd_fmri_get_unum(nvlist_t *);
const char *, const char *);
extern char *cmd_mem_serdnm_create(fmd_hdl_t *, const char *, const char *);
uint64_t);
extern void cmd_mem_gc(fmd_hdl_t *);
extern void cmd_mem_fini(fmd_hdl_t *);
#ifdef __cplusplus
}
#endif
#endif /* _CMD_MEM_H */