sun4v/cpumem-retire/cma_cpu_sun4v.c

	cma_cpu_sun4v.c revision aab83bb83be7342f6cfccaed8d5fe0b2f404855d
/*
 * CDDL HEADER START
 *
 * The contents of this file are subject to the terms of the
 * Common Development and Distribution License (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
 * or http://www.opensolaris.org/os/licensing.
 * 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 2008 Sun Microsystems, Inc.  All rights reserved.
 * Use is subject to license terms.
 */

#include <cma.h>

#include <sys/fm/ldom.h>
#include <sys/fm/protocol.h>
#include <fm/fmd_fmri.h>
#include <fm/libtopo.h>

#include <assert.h>
#include <fcntl.h>
#include <unistd.h>
#include <errno.h>
#include <strings.h>

#include <sys/types.h>
#include <sys/processor.h>

extern ldom_hdl_t *cma_lhp;

/*ARGSUSED*/
int
cpu_blacklist_cmd(fmd_hdl_t *hdl, nvlist_t *fmri, boolean_t repair)
{
    if (repair)
        return (ldom_fmri_unblacklist(cma_lhp, fmri));
    else
        return (ldom_fmri_blacklist(cma_lhp, fmri));
}

int
cma_cpu_blacklist(fmd_hdl_t *hdl, nvlist_t *nvl, nvlist_t *asru,
    boolean_t repair)
{
    nvlist_t *fmri;
    int rc, err;

    /*
     * Some platforms have special unums for the E$ DIMMs.  If we're dealing
     * with a platform that has these unums, one will have been added to the
     * fault as the resource.  We'll use that for the blacklisting.  If we
     * can't find a resource, we'll fall back to the ASRU.
     */
    if (nvlist_lookup_nvlist(nvl, FM_FAULT_RESOURCE, &fmri) != 0)
        fmri = asru;

    rc = cpu_blacklist_cmd(hdl, fmri, repair);
    err = errno;

    if (rc < 0 && err != ENOTSUP) {
        errno = err;
        return (-1);
    }

    return (0);
}

/*ARGSUSED*/
static int
cpu_cmd(fmd_hdl_t *hdl, nvlist_t *fmri, int cmd)
{
    int rc = 0;
    char *scheme;

    /*
     * We're using topo retire if the fmri is in "hc" scheme.
     */
    if (nvlist_lookup_string(fmri, FM_FMRI_SCHEME, &scheme) == 0 &&
        strcmp(scheme, FM_FMRI_SCHEME_HC) == 0) {
        if (cmd != P_STATUS) {
            errno = EINVAL;
            return (-1);
        }
        rc = fmd_nvl_fmri_service_state(hdl, fmri);
        switch (rc) {
        case FMD_SERVICE_STATE_UNUSABLE:
            return (P_FAULTED);
        case -1:
            return (-1);
        default:
            return (P_ONLINE);
        }
    }

    switch (cmd & ~P_FORCED) {
    case P_STATUS:
        rc = ldom_fmri_status(cma_lhp, fmri);
        break;
    case P_FAULTED:
        rc = ldom_fmri_retire(cma_lhp, fmri);
        break;
    case P_ONLINE:
        rc = ldom_fmri_unretire(cma_lhp, fmri);
        break;
    default:
        errno = EINVAL;
        return (-1);
    }

    if (rc != P_OFFLINE && rc != P_ONLINE && rc != P_FAULTED) {
        errno = rc;
        return (-1);
    }

    return (rc);
}

void
cma_cpu_start_retry(fmd_hdl_t *hdl, nvlist_t *fmri, const char *uuid,
    boolean_t repair)
{
    cma_cpu_t *cpu;
    char *scheme;
    uint_t cpuid;
    nvlist_t *asru = NULL;
    topo_hdl_t *thp;
    int err;

    if (repair || nvlist_lookup_string(fmri, FM_FMRI_SCHEME, &scheme) != 0)
        return;
    if (strcmp(scheme, FM_FMRI_SCHEME_CPU) == 0) {
        if (nvlist_lookup_uint32(fmri, FM_FMRI_CPU_ID, &cpuid) != 0)
            return;
    } else if (strcmp(scheme, FM_FMRI_SCHEME_HC) != 0) {
        return;
    } else {
        /* lookup cpuid from ASRU */
        thp = fmd_fmri_topo_hold(TOPO_VERSION);
        if (thp != NULL) {
            (void) topo_fmri_asru(thp, fmri, &asru, &err);
            fmd_fmri_topo_rele(thp);
        }
        if (nvlist_lookup_uint32(asru, FM_FMRI_CPU_ID, &cpuid) != 0) {
            nvlist_free(asru);
            return;
        }
    }

    /*
     * check to see if the cpu has been offline.
     */
    fmd_hdl_debug(hdl, "cpu %u is not offline yet - sleeping\n", cpuid);

    /*
     * Create a cpu node and add to the head of the cpu list
     */
    cpu = fmd_hdl_zalloc(hdl, sizeof (cma_cpu_t), FMD_SLEEP);
    (void) nvlist_dup(fmri, &cpu->cpu_fmri, 0);
    if (uuid != NULL)
        cpu->cpu_uuid = fmd_hdl_strdup(hdl, uuid, FMD_SLEEP);

    cpu->cpuid = cpuid;
    cpu->cpu_next = cma.cma_cpus;
    cma.cma_cpus = cpu;

    if (cma.cma_cpu_timerid != 0)
        fmd_timer_remove(hdl, cma.cma_cpu_timerid);

    cma.cma_cpu_curdelay = cma.cma_cpu_mindelay;

    cma.cma_cpu_timerid =
        fmd_timer_install(hdl, NULL, NULL, cma.cma_cpu_curdelay);
}


int
cma_cpu_statechange(fmd_hdl_t *hdl, nvlist_t *asru, const char *uuid,
    int cpustate, boolean_t repair)
{
    int i;
    uint_t cpuid;

    if (nvlist_lookup_uint32(asru, FM_FMRI_CPU_ID, &cpuid) != 0) {
        fmd_hdl_debug(hdl, "missing '%s'\n", FM_FMRI_CPU_ID);
        cma_stats.bad_flts.fmds_value.ui64++;
        return (CMA_RA_FAILURE);
    }

    /*
     * cpu offlining using ldom_fmri_retire() may be asynchronous, so we
     * have to set the timer and check the cpu status later.
     */
    for (i = 0; i < cma.cma_cpu_tries;
        i++, (void) nanosleep(&cma.cma_cpu_delay, NULL)) {
        if (cpu_cmd(hdl, asru, cpustate) != -1) {
            if (repair)
                cma_stats.cpu_repairs.fmds_value.ui64++;
            else
                cma_stats.cpu_flts.fmds_value.ui64++;
            break;
        }
    }

    if (i >= cma.cma_cpu_tries) {
        cma_stats.cpu_fails.fmds_value.ui64++;
    }

    cma_cpu_start_retry(hdl, asru, uuid, repair);

    return (CMA_RA_FAILURE);
}

static int
cpu_retry(fmd_hdl_t *hdl, cma_cpu_t *cpu)
{
    int rc = 0;

    fmd_hdl_debug(hdl, "cpu_retry()\n");

    if (cpu->cpu_fmri == NULL) {
        return (1);
    }

    if (!fmd_nvl_fmri_present(hdl, cpu->cpu_fmri)) {
        fmd_hdl_debug(hdl, "cpu %u is not present", cpu->cpuid);
        return (1);
    }

    rc = cpu_cmd(hdl, cpu->cpu_fmri, P_STATUS);
    if (rc == P_FAULTED || rc == P_OFFLINE) {
        fmd_hdl_debug(hdl, "cpu %u is offlined on retry %u\n",
            cpu->cpuid, cpu->cpu_nretries);
        cma_stats.cpu_flts.fmds_value.ui64++;

        if (cpu->cpu_uuid != NULL)
            fmd_case_uuclose(hdl, cpu->cpu_uuid);
        return (1); /* success */
    }

    if (rc == -1) {
        fmd_hdl_debug(hdl, "failed to retry cpu %u\n", cpu->cpuid);
        cma_stats.page_fails.fmds_value.ui64++;
        return (1); /* give up */
    }

    return (0);
}

static void
cma_cpu_free(fmd_hdl_t *hdl, cma_cpu_t *cpu)
{
    nvlist_free(cpu->cpu_fmri);
    if (cpu->cpu_uuid != NULL)
        fmd_hdl_strfree(hdl, cpu->cpu_uuid);
    fmd_hdl_free(hdl, cpu, sizeof (cma_cpu_t));
}

void
cma_cpu_retry(fmd_hdl_t *hdl)
{
    cma_cpu_t **cpup;

    fmd_hdl_debug(hdl, "cma_cpu_retry: timer fired\n");

    cma.cma_cpu_timerid = 0;

    cpup = &cma.cma_cpus;
    while (*cpup != NULL) {
        cma_cpu_t *cpu = *cpup;

        if (cpu_retry(hdl, cpu)) {
            /*
             * Successful retry or we're giving up - remove from
             * the list
             */
            *cpup = cpu->cpu_next;

            cma_cpu_free(hdl, cpu);
        } else {
            cpu->cpu_nretries++;
            cpup = &cpu->cpu_next;
        }
    }

    if (cma.cma_cpus == NULL)
        return; /* no more cpus */

    /*
     * We still have cpus to check.  Back the delay
     * off, and schedule a retry.
     */
    cma.cma_cpu_curdelay = MIN(cma.cma_cpu_curdelay * 2,
        cma.cma_cpu_maxdelay);

    fmd_hdl_debug(hdl, "scheduled cpu offline retry for %llu secs\n",
        (u_longlong_t)(cma.cma_cpu_curdelay / NANOSEC));

    cma.cma_cpu_timerid =
        fmd_timer_install(hdl, NULL, NULL, cma.cma_cpu_curdelay);
}

void
cma_cpu_fini(fmd_hdl_t *hdl)
{
    cma_cpu_t *cpu;

    while ((cpu = cma.cma_cpus) != NULL) {
        cma.cma_cpus = cpu->cpu_next;
        cma_cpu_free(hdl, cpu);
    }
}