pfmlib_core.c

Go to the documentation of this file.

/*
 * pfmlib_core.c : Intel Core PMU
 *
 * Copyright (c) 2006 Hewlett-Packard Development Company, L.P.
 * Contributed by Stephane Eranian <eranian@hpl.hp.com>
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
 * of the Software, and to permit persons to whom the Software is furnished to do so,
 * subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in all
 * copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED,
 * INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A
 * PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
 * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF
 * CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE
 * OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 *
 *
 * This file implements support for Intel Core PMU as specified in the following document:
 *  "IA-32 Intel Architecture Software Developer's Manual - Volume 3B: System
 *  Programming Guide"
 *
 * Core PMU = architectural perfmon v2 + PEBS
 */
#include <sys/types.h>
#include <ctype.h>
#include <string.h>
#include <stdlib.h>
#include <stdio.h>
 
/* public headers */
#include <perfmon/pfmlib_core.h>
 
/* private headers */
#include "pfmlib_priv.h"
#include "pfmlib_core_priv.h"
 
#include "core_events.h"
 
/* let's define some handy shortcuts! */
#define sel_event_select perfevtsel.sel_event_select
#define sel_unit_mask    perfevtsel.sel_unit_mask
#define sel_usr      perfevtsel.sel_usr
#define sel_os       perfevtsel.sel_os
#define sel_edge     perfevtsel.sel_edge
#define sel_pc       perfevtsel.sel_pc
#define sel_int      perfevtsel.sel_int
#define sel_en       perfevtsel.sel_en
#define sel_inv      perfevtsel.sel_inv
#define sel_cnt_mask     perfevtsel.sel_cnt_mask
 
#define is_pebs(i)  (core_pe[i].pme_flags & PFMLIB_CORE_PEBS)
 
 
/*
 * Description of the PMC register mappings:
 *
 * 0  -> PMC0  -> PERFEVTSEL0
 * 1  -> PMC1  -> PERFEVTSEL1 
 * 16 -> PMC16 -> FIXED_CTR_CTRL
 * 17 -> PMC17 -> PEBS_ENABLED
 *
 * Description of the PMD register mapping:
 *
 * 0  -> PMD0 -> PMC0
 * 1  -> PMD1 -> PMC1
 * 16 -> PMD2 -> FIXED_CTR0
 * 17 -> PMD3 -> FIXED_CTR1
 * 18 -> PMD4 -> FIXED_CTR2
 */
#define CORE_SEL_BASE       0x186
#define CORE_CTR_BASE       0xc1
#define FIXED_CTR_BASE      0x309
 
#define PFMLIB_CORE_ALL_FLAGS \
    (PFM_CORE_SEL_INV|PFM_CORE_SEL_EDGE)
 
static pfmlib_regmask_t core_impl_pmcs, core_impl_pmds;
static int highest_counter;
 
static int
pfm_core_detect(void)
{
    int ret;
    int family, model;
    char buffer[128];
 
    ret = __pfm_getcpuinfo_attr("vendor_id", buffer, sizeof(buffer));
    if (ret == -1)
        return PFMLIB_ERR_NOTSUPP;
 
    if (strcmp(buffer, "GenuineIntel"))
        return PFMLIB_ERR_NOTSUPP;
 
    ret = __pfm_getcpuinfo_attr("cpu family", buffer, sizeof(buffer));
    if (ret == -1)
        return PFMLIB_ERR_NOTSUPP;
 
    family = atoi(buffer);
 
    ret = __pfm_getcpuinfo_attr("model", buffer, sizeof(buffer));
    if (ret == -1)
        return PFMLIB_ERR_NOTSUPP;
 
    if (family != 6)
        return PFMLIB_ERR_NOTSUPP;
 
    model = atoi(buffer);
    switch(model) {
        case 15: /* Merom */
        case 23: /* Penryn */
        case 29: /* Dunnington */
              break;
        default:
            return PFMLIB_ERR_NOTSUPP;
    }
    return PFMLIB_SUCCESS;
}
 
static int
pfm_core_init(void)
{
    int i;
 
    pfm_regmask_set(&core_impl_pmcs, 0);
    pfm_regmask_set(&core_impl_pmcs, 1);
    pfm_regmask_set(&core_impl_pmcs, 16);
    pfm_regmask_set(&core_impl_pmcs, 17);
 
    pfm_regmask_set(&core_impl_pmds, 0);
    pfm_regmask_set(&core_impl_pmds, 1);
    pfm_regmask_set(&core_impl_pmds, 16);
    pfm_regmask_set(&core_impl_pmds, 17);
    pfm_regmask_set(&core_impl_pmds, 18);
 
    /* lbr */
    pfm_regmask_set(&core_impl_pmds, 19);
    for(i=0; i < 8; i++)
        pfm_regmask_set(&core_impl_pmds, i);
 
    highest_counter = 18;
 
    return PFMLIB_SUCCESS;
}
 
static int
pfm_core_is_fixed(pfmlib_event_t *e, unsigned int f)
{
    unsigned int fl, flc, i;
    unsigned int mask = 0;
 
    fl = core_pe[e->event].pme_flags;
 
    /*
     * first pass: check if event as a whole supports fixed counters
     */
    switch(f) {
        case 0:
            mask = PFMLIB_CORE_FIXED0;
            break;
        case 1:
            mask = PFMLIB_CORE_FIXED1;
            break;
        case 2:
            mask = PFMLIB_CORE_FIXED2_ONLY;
            break;
        default:
            return 0;
    }
    if (fl & mask)
        return 1;
    /*
     * second pass: check if unit mask support fixed counter
     *
     * reject if mask not found OR if not all unit masks have
     * same fixed counter mask
     */
    flc = 0;
    for(i=0; i < e->num_masks; i++) {
        fl = core_pe[e->event].pme_umasks[e->unit_masks[i]].pme_flags;
        if (fl & mask)
            flc++;
    }
    return flc > 0 && flc == e->num_masks ? 1 : 0;
}
 
/*
 * IMPORTANT: the interface guarantees that pfp_pmds[] elements are returned in the order the events
 *        were submitted.
 */
static int
pfm_core_dispatch_counters(pfmlib_input_param_t *inp, pfmlib_core_input_param_t *param, pfmlib_output_param_t *outp)
{
#define HAS_OPTIONS(x)  (cntrs && (cntrs[x].flags || cntrs[x].cnt_mask))
#define is_fixed_pmc(a) (a == 16 || a == 17 || a == 18)
 
    pfmlib_core_counter_t *cntrs;
    pfm_core_sel_reg_t reg;
    pfmlib_event_t *e;
    pfmlib_reg_t *pc, *pd;
    pfmlib_regmask_t *r_pmcs;
    uint64_t val;
    unsigned long plm;
    unsigned long long fixed_ctr;
    unsigned int npc, npmc0, npmc1, nf2;
    unsigned int i, j, n, k, ucode, use_pebs = 0, done_pebs;
    unsigned int assign_pc[PMU_CORE_NUM_COUNTERS];
    unsigned int next_gen, last_gen;
 
    npc = npmc0 = npmc1 = nf2 = 0;
 
    e      = inp->pfp_events;
    pc     = outp->pfp_pmcs;
    pd     = outp->pfp_pmds;
    n      = inp->pfp_event_count;
    r_pmcs = &inp->pfp_unavail_pmcs;
    cntrs  = param ? param->pfp_core_counters : NULL;
    use_pebs = param ? param->pfp_core_pebs.pebs_used : 0;
 
    if (n > PMU_CORE_NUM_COUNTERS)
        return PFMLIB_ERR_TOOMANY;
 
    /*
     * initilize to empty
     */
    for(i=0; i < PMU_CORE_NUM_COUNTERS; i++)
        assign_pc[i] = -1;
 
    /*
     * error checking
     */
    for(i=0; i < n; i++) {
        /*
         * only supports two priv levels for perf counters
         */
        if (e[i].plm & (PFM_PLM1|PFM_PLM2))
            return PFMLIB_ERR_INVAL;
 
        /*
         * check for valid flags
         */
        if (cntrs && cntrs[i].flags & ~PFMLIB_CORE_ALL_FLAGS)
            return PFMLIB_ERR_INVAL;
 
        if (core_pe[e[i].event].pme_flags & PFMLIB_CORE_UMASK_NCOMBO
            && e[i].num_masks > 1) {
            DPRINT("events does not support unit mask combination\n");
                return PFMLIB_ERR_NOASSIGN;
        }
 
        /*
         * check event-level single register constraint (PMC0, PMC1, FIXED_CTR2)
         * fail if more than two events requested for the same counter
         */
        if (core_pe[e[i].event].pme_flags & PFMLIB_CORE_PMC0) {
            if (++npmc0 > 1) {
                DPRINT("two events compete for a PMC0\n");
                return PFMLIB_ERR_NOASSIGN;
            }
        }
        /*
         * check if PMC1 is available and if only one event is dependent on it
         */
        if (core_pe[e[i].event].pme_flags & PFMLIB_CORE_PMC1) {
            if (++npmc1 > 1) {
                DPRINT("two events compete for a PMC1\n");
                return PFMLIB_ERR_NOASSIGN;
            }
        }
        /*
         * UNHALTED_REFERENCE_CYCLES can only be measured on FIXED_CTR2
         */
        if (core_pe[e[i].event].pme_flags & PFMLIB_CORE_FIXED2_ONLY) {
            if (++nf2 > 1) {
                DPRINT("two events compete for FIXED_CTR2\n");
                return PFMLIB_ERR_NOASSIGN;
            }
            if (HAS_OPTIONS(i)) {
                DPRINT("fixed counters do not support inversion/counter-mask\n");
                return PFMLIB_ERR_NOASSIGN;
            }
        }
        /*
         * unit-mask level constraint checking (PMC0, PMC1, FIXED_CTR2)
         */ 
        for(j=0; j < e[i].num_masks; j++) {
            unsigned int flags;
 
            flags = core_pe[e[i].event].pme_umasks[e[i].unit_masks[j]].pme_flags;
 
            if (flags & PFMLIB_CORE_FIXED2_ONLY) {
                if (++nf2 > 1) {
                    DPRINT("two events compete for FIXED_CTR2\n");
                    return PFMLIB_ERR_NOASSIGN;
                }
                if (HAS_OPTIONS(i)) {
                    DPRINT("fixed counters do not support inversion/counter-mask\n");
                    return PFMLIB_ERR_NOASSIGN;
                }
            }
                }
    }
 
    next_gen = 0; /* first generic counter */
    last_gen = 1; /* last generic counter */
 
    /*
     * strongest constraint first: works only in IA32_PMC0, IA32_PMC1, FIXED_CTR2
     *
     * When PEBS is used, we pick the first PEBS event and
     * place it into PMC0. Subsequent PEBS events, will go
     * in the other counters.
     */
    done_pebs = 0;
    for(i=0; i < n; i++) {
        if ((core_pe[e[i].event].pme_flags & PFMLIB_CORE_PMC0)
            || (use_pebs && pfm_core_is_pebs(e+i) && done_pebs == 0)) {
            if (pfm_regmask_isset(r_pmcs, 0))
                return PFMLIB_ERR_NOASSIGN;
            assign_pc[i] = 0;
            next_gen = 1;
            done_pebs = 1;
        }
        if (core_pe[e[i].event].pme_flags & PFMLIB_CORE_PMC1) {
            if (pfm_regmask_isset(r_pmcs, 1))
                return PFMLIB_ERR_NOASSIGN;
            assign_pc[i] = 1;
            if (next_gen == 1)
                next_gen = 2;
            else
                next_gen = 0;
        }
    }
    /*
     * next constraint: fixed counters
     *
     * We abuse the mapping here for assign_pc to make it easier
     * to provide the correct values for pd[].
     * We use:
     *  - 16 : fixed counter 0 (pmc16, pmd16)
     *  - 17 : fixed counter 1 (pmc16, pmd17)
     *  - 18 : fixed counter 1 (pmc16, pmd18)
     */
    fixed_ctr = pfm_regmask_isset(r_pmcs, 16) ? 0 : 0x7;
    if (fixed_ctr) {
        for(i=0; i < n; i++) {
            /* fixed counters do not support event options (filters) */
            if (HAS_OPTIONS(i) || (use_pebs && pfm_core_is_pebs(e+i)))
                continue;
 
            if ((fixed_ctr & 0x1) && pfm_core_is_fixed(e+i, 0)) {
                assign_pc[i] = 16;
                fixed_ctr &= ~1;
            }
            if ((fixed_ctr & 0x2) && pfm_core_is_fixed(e+i, 1)) {
                assign_pc[i] = 17;
                fixed_ctr &= ~2;
            }
            if ((fixed_ctr & 0x4) && pfm_core_is_fixed(e+i, 2)) {
                assign_pc[i] = 18;
                fixed_ctr &= ~4;
            }
        }
    }
    /*
     * assign what is left
     */
    for(i=0; i < n; i++) {
        if (assign_pc[i] == -1) {
            for(; next_gen <= last_gen; next_gen++) {
DPRINT("i=%d next_gen=%d last=%d isset=%d\n", i, next_gen, last_gen, pfm_regmask_isset(r_pmcs, next_gen));
                if (!pfm_regmask_isset(r_pmcs, next_gen))
                    break;
            }
            if (next_gen <= last_gen)
                assign_pc[i] = next_gen++;
            else {
                DPRINT("cannot assign generic counters\n");
                return PFMLIB_ERR_NOASSIGN;
            }
        }
    }
    j = 0;
 
    /* setup fixed counters */
    reg.val = 0;
    k = 0;
    for (i=0; i < n ; i++ ) {
        if (!is_fixed_pmc(assign_pc[i]))
            continue;
        val = 0;
        /* if plm is 0, then assume not specified per-event and use default */
        plm = e[i].plm ? e[i].plm : inp->pfp_dfl_plm;
        if (plm & PFM_PLM0)
            val |= 1ULL;
        if (plm & PFM_PLM3)
            val |= 2ULL;
        val |= 1ULL << 3;    /* force APIC int (kernel may force it anyway) */
 
        reg.val |= val << ((assign_pc[i]-16)<<2);
    }
 
    if (reg.val) {
        pc[npc].reg_num   = 16;
        pc[npc].reg_value = reg.val;
        pc[npc].reg_addr  = 0x38D;
        pc[npc].reg_alt_addr  = 0x38D;
 
        __pfm_vbprintf("[FIXED_CTRL(pmc%u)=0x%"PRIx64" pmi0=1 en0=0x%"PRIx64" pmi1=1 en1=0x%"PRIx64" pmi2=1 en2=0x%"PRIx64"] ",
                pc[npc].reg_num,
                reg.val,
                reg.val & 0x3ULL,
                (reg.val>>4) & 0x3ULL,
                (reg.val>>8) & 0x3ULL);
 
        if ((fixed_ctr & 0x1) == 0)
            __pfm_vbprintf("INSTRUCTIONS_RETIRED ");
        if ((fixed_ctr & 0x2) == 0)
            __pfm_vbprintf("UNHALTED_CORE_CYCLES ");
        if ((fixed_ctr & 0x4) == 0)
            __pfm_vbprintf("UNHALTED_REFERENCE_CYCLES ");
        __pfm_vbprintf("\n");
 
        npc++;
 
        if ((fixed_ctr & 0x1) == 0)
            __pfm_vbprintf("[FIXED_CTR0(pmd16)]\n");
        if ((fixed_ctr & 0x2) == 0)
            __pfm_vbprintf("[FIXED_CTR1(pmd17)]\n");
        if ((fixed_ctr & 0x4) == 0)
            __pfm_vbprintf("[FIXED_CTR2(pmd18)]\n");
    }
 
    for (i=0; i < n ; i++ ) {
        /* skip fixed counters */
        if (is_fixed_pmc(assign_pc[i]))
            continue;
 
        reg.val = 0; /* assume reserved bits are zerooed */
 
        /* if plm is 0, then assume not specified per-event and use default */
        plm = e[i].plm ? e[i].plm : inp->pfp_dfl_plm;
 
        val = core_pe[e[i].event].pme_code;
 
        reg.sel_event_select = val  & 0xff;
 
        ucode = (val >> 8) & 0xff;
 
        for(k=0; k < e[i].num_masks; k++) {
            ucode |= core_pe[e[i].event].pme_umasks[e[i].unit_masks[k]].pme_ucode;
        }
 
        /*
         * for events supporting Core specificity (self, both), a value
         * of 0 for bits 15:14 (7:6 in our umask) is reserved, therefore we
         * force to SELF if user did not specify anything
         */
        if ((core_pe[e[i].event].pme_flags & PFMLIB_CORE_CSPEC)
            && ((ucode & (0x3 << 6)) == 0)) {
            ucode |= 1 << 6;
        }
        /*
         * for events supporting MESI, a value
         * of 0 for bits 11:8 (0-3 in our umask) means nothing will be
         * counted. Therefore, we force a default of 0xf (M,E,S,I).
         */
        if ((core_pe[e[i].event].pme_flags & PFMLIB_CORE_MESI)
            && ((ucode & 0xf) == 0)) {
            ucode |= 0xf;
        }
 
        val |= ucode << 8;
 
        reg.sel_unit_mask  = ucode;
        reg.sel_usr        = plm & PFM_PLM3 ? 1 : 0;
        reg.sel_os         = plm & PFM_PLM0 ? 1 : 0;
        reg.sel_en         = 1; /* force enable bit to 1 */
        reg.sel_int        = 1; /* force APIC int to 1 */
 
        reg.sel_cnt_mask = val >>24;
        reg.sel_inv = val >> 23;
        reg.sel_edge = val >> 18;
 
        if (cntrs) {
            if (!reg.sel_cnt_mask) {
                /*
                 * counter mask is 8-bit wide, do not silently
                 * wrap-around
                 */
                if (cntrs[i].cnt_mask > 255)
                    return PFMLIB_ERR_INVAL;
                reg.sel_cnt_mask = cntrs[i].cnt_mask;
            }
 
            if (!reg.sel_edge)
                reg.sel_edge = cntrs[i].flags & PFM_CORE_SEL_EDGE ? 1 : 0;
            if (!reg.sel_inv)
                reg.sel_inv = cntrs[i].flags & PFM_CORE_SEL_INV ? 1 : 0;
        }
 
        pc[npc].reg_num     = assign_pc[i];
        pc[npc].reg_value   = reg.val;
        pc[npc].reg_addr    = CORE_SEL_BASE+assign_pc[i];
        pc[npc].reg_alt_addr= CORE_SEL_BASE+assign_pc[i];
 
        __pfm_vbprintf("[PERFEVTSEL%u(pmc%u)=0x%"PRIx64" event_sel=0x%x umask=0x%x os=%d usr=%d en=%d int=%d inv=%d edge=%d cnt_mask=%d] %s\n",
                pc[npc].reg_num,
                pc[npc].reg_num,
                reg.val,
                reg.sel_event_select,
                reg.sel_unit_mask,
                reg.sel_os,
                reg.sel_usr,
                reg.sel_en,
                reg.sel_int,
                reg.sel_inv,
                reg.sel_edge,
                reg.sel_cnt_mask,
                core_pe[e[i].event].pme_name);
 
        __pfm_vbprintf("[PMC%u(pmd%u)]\n",
                pc[npc].reg_num,
                pc[npc].reg_num);
 
        npc++;
    }
    /*
     * setup pmds: must be in the same order as the events
     */
    for (i=0; i < n ; i++) {
        if (is_fixed_pmc(assign_pc[i])) {
            /* setup pd array */
            pd[i].reg_num = assign_pc[i];
            pd[i].reg_addr = FIXED_CTR_BASE+assign_pc[i]-16;
            pd[i].reg_alt_addr = 0x40000000+assign_pc[i]-16;
        } else {
            pd[i].reg_num  = assign_pc[i];
            pd[i].reg_addr = CORE_CTR_BASE+assign_pc[i];
            /* index to use with RDPMC */
            pd[i].reg_alt_addr  = assign_pc[i];
        }
    }
    outp->pfp_pmd_count = i;
 
    /*
     * setup PEBS_ENABLE
     */
    if (use_pebs && done_pebs) {
        /*
         * check that PEBS_ENABLE is available
         */
        if (pfm_regmask_isset(r_pmcs, 17))
            return PFMLIB_ERR_NOASSIGN;
        pc[npc].reg_num   = 17;
        pc[npc].reg_value = 1ULL;
        pc[npc].reg_addr  = 0x3f1; /* IA32_PEBS_ENABLE */
        pc[npc].reg_alt_addr  = 0x3f1; /* IA32_PEBS_ENABLE */
 
        __pfm_vbprintf("[PEBS_ENABLE(pmc%u)=0x%"PRIx64" ena=%d]\n",
                pc[npc].reg_num,
                pc[npc].reg_value,
                pc[npc].reg_value & 0x1ull);
 
        npc++;
 
    }
    outp->pfp_pmc_count = npc;
 
    return PFMLIB_SUCCESS;
}
 
#if 0
static int
pfm_core_dispatch_pebs(pfmlib_input_param_t *inp, pfmlib_core_input_param_t *mod_in, pfmlib_output_param_t *outp)
{
    pfmlib_event_t *e;
    pfm_core_sel_reg_t reg;
    unsigned int umask, npc, npd, k, plm;
    pfmlib_regmask_t *r_pmcs;
    pfmlib_reg_t *pc, *pd;
    int event;
 
    npc = outp->pfp_pmc_count;
    npd = outp->pfp_pmd_count;
    pc  = outp->pfp_pmcs;
    pd  = outp->pfp_pmds;
    r_pmcs = &inp->pfp_unavail_pmcs;
 
    e = inp->pfp_events;
    /*
     * check for valid flags
     */
    if (e[0].flags & ~PFMLIB_CORE_ALL_FLAGS)
        return PFMLIB_ERR_INVAL;
 
    /*
     * check event supports PEBS
     */
    if (pfm_core_is_pebs(e) == 0)
        return PFMLIB_ERR_FEATCOMB;
 
    /*
     * check that PMC0 is available
     * PEBS works only on PMC0
     * Some PEBS at-retirement events do require PMC0 anyway
     */
    if (pfm_regmask_isset(r_pmcs, 0))
        return PFMLIB_ERR_NOASSIGN;
 
    /*
     * check that PEBS_ENABLE is available
     */
    if (pfm_regmask_isset(r_pmcs, 17))
        return PFMLIB_ERR_NOASSIGN;
 
    reg.val = 0; /* assume reserved bits are zerooed */
 
    event = e[0].event;
 
    /* if plm is 0, then assume not specified per-event and use default */
    plm = e[0].plm ? e[0].plm : inp->pfp_dfl_plm;
 
    reg.sel_event_select = core_pe[event].pme_code & 0xff;
 
    umask = (core_pe[event].pme_code >> 8) & 0xff;
 
    for(k=0; k < e[0].num_masks; k++) {
        umask |= core_pe[event].pme_umasks[e[0].unit_masks[k]].pme_ucode;
    }
    reg.sel_unit_mask  = umask;
    reg.sel_usr        = plm & PFM_PLM3 ? 1 : 0;
    reg.sel_os         = plm & PFM_PLM0 ? 1 : 0;
    reg.sel_en         = 1; /* force enable bit to 1 */
    reg.sel_int        = 0; /* not INT for PEBS counter */
 
    reg.sel_cnt_mask = mod_in->pfp_core_counters[0].cnt_mask;
    reg.sel_edge     = mod_in->pfp_core_counters[0].flags & PFM_CORE_SEL_EDGE ? 1 : 0;
    reg.sel_inv  = mod_in->pfp_core_counters[0].flags & PFM_CORE_SEL_INV ? 1 : 0;
 
    pc[npc].reg_num     = 0;
    pc[npc].reg_value   = reg.val;
    pc[npc].reg_addr    = CORE_SEL_BASE;
    pc[npc].reg_alt_addr= CORE_SEL_BASE;
 
    pd[npd].reg_num  = 0;
    pd[npd].reg_addr = CORE_CTR_BASE;
    pd[npd].reg_alt_addr = 0;
 
    __pfm_vbprintf("[PERFEVTSEL%u(pmc%u)=0x%"PRIx64" event_sel=0x%x umask=0x%x os=%d usr=%d en=%d int=%d inv=%d edge=%d cnt_mask=%d] %s\n",
        pc[npc].reg_num,
        pc[npc].reg_num,
        reg.val,
        reg.sel_event_select,
        reg.sel_unit_mask,
        reg.sel_os,
        reg.sel_usr,
        reg.sel_en,
        reg.sel_int,
        reg.sel_inv,
        reg.sel_edge,
        reg.sel_cnt_mask,
        core_pe[e[0].event].pme_name);
 
    __pfm_vbprintf("[PMC%u(pmd%u)]\n",
            pd[npd].reg_num,
            pd[npd].reg_num);
 
    npc++;
    npd++;
    /*
     * setup PEBS_ENABLE
     */
    pc[npc].reg_num   = 17;
    pc[npc].reg_value = 1ULL;
    pc[npc].reg_addr  = 0x3f1; /* IA32_PEBS_ENABLE */
    pc[npc].reg_alt_addr = 0x3f1; /* IA32_PEBS_ENABLE */
 
    __pfm_vbprintf("[PEBS_ENABLE(pmc%u)=0x%"PRIx64" ena=%d]\n",
            pc[npc].reg_num,
            pc[npc].reg_value,
            pc[npc].reg_value & 0x1ull);
 
    npc++;
 
    /* number of evtsel/ctr registers programmed */
    outp->pfp_pmc_count = npc;
    outp->pfp_pmd_count = npd;
 
    return PFMLIB_SUCCESS;
}
#endif
 
static int
pfm_core_dispatch_events(pfmlib_input_param_t *inp, void *model_in, pfmlib_output_param_t *outp, void *model_out)
{
    pfmlib_core_input_param_t *mod_in  = (pfmlib_core_input_param_t *)model_in;
 
    if (inp->pfp_dfl_plm & (PFM_PLM1|PFM_PLM2)) {
        DPRINT("invalid plm=%x\n", inp->pfp_dfl_plm);
        return PFMLIB_ERR_INVAL;
    }
    return pfm_core_dispatch_counters(inp, mod_in, outp);
}
 
static int
pfm_core_get_event_code(unsigned int i, unsigned int cnt, int *code)
{
    if (cnt != PFMLIB_CNT_FIRST
        && (cnt > highest_counter ||
        !pfm_regmask_isset(&core_impl_pmds, cnt)))
        return PFMLIB_ERR_INVAL;
 
    *code = core_pe[i].pme_code;
 
    return PFMLIB_SUCCESS;
}
 
static void
pfm_core_get_event_counters(unsigned int j, pfmlib_regmask_t *counters)
{
    unsigned int n, i;
    unsigned int has_f0, has_f1, has_f2;
 
    memset(counters, 0, sizeof(*counters));
 
    n = core_pe[j].pme_numasks;
    has_f0 = has_f1 = has_f2 = 0;
 
    for (i=0; i < n; i++) {
        if (core_pe[j].pme_umasks[i].pme_flags & PFMLIB_CORE_FIXED0)
            has_f0 = 1;
        if (core_pe[j].pme_umasks[i].pme_flags & PFMLIB_CORE_FIXED1)
            has_f1 = 1;
        if (core_pe[j].pme_umasks[i].pme_flags & PFMLIB_CORE_FIXED2_ONLY)
            has_f2 = 1;
    }
 
    if (has_f0 == 0)
        has_f0 = core_pe[j].pme_flags & PFMLIB_CORE_FIXED0;
    if (has_f1 == 0)
        has_f1 = core_pe[j].pme_flags & PFMLIB_CORE_FIXED1;
    if (has_f2 == 0)
        has_f2 = core_pe[j].pme_flags & PFMLIB_CORE_FIXED2_ONLY;
 
    if (has_f0)
        pfm_regmask_set(counters, 16);
    if (has_f1)
        pfm_regmask_set(counters, 17);
    if (has_f2)
        pfm_regmask_set(counters, 18);
 
    /* the event on FIXED_CTR2 is exclusive CPU_CLK_UNHALTED:REF */
    if (!has_f2) {
        pfm_regmask_set(counters, 0);
        pfm_regmask_set(counters, 1);
 
        if (core_pe[j].pme_flags & PFMLIB_CORE_PMC0)
            pfm_regmask_clr(counters, 1);
        if (core_pe[j].pme_flags & PFMLIB_CORE_PMC1)
            pfm_regmask_clr(counters, 0);
    }
}
 
static void
pfm_core_get_impl_pmcs(pfmlib_regmask_t *impl_pmcs)
{
    *impl_pmcs = core_impl_pmcs;
}
 
static void
pfm_core_get_impl_pmds(pfmlib_regmask_t *impl_pmds)
{
    *impl_pmds = core_impl_pmds;
}
 
static void
pfm_core_get_impl_counters(pfmlib_regmask_t *impl_counters)
{
    pfm_regmask_set(impl_counters, 0);
    pfm_regmask_set(impl_counters, 1);
    pfm_regmask_set(impl_counters, 16);
    pfm_regmask_set(impl_counters, 17);
    pfm_regmask_set(impl_counters, 18);
}
 
/*
 * Even though, CPUID 0xa returns in eax the actual counter
 * width, the architecture specifies that writes are limited
 * to lower 32-bits. As such, only the lower 32-bit have full
 * degree of freedom. That is the "useable" counter width.
 */
#define PMU_CORE_COUNTER_WIDTH       32
 
static void
pfm_core_get_hw_counter_width(unsigned int *width)
{
    /*
     * Even though, CPUID 0xa returns in eax the actual counter
     * width, the architecture specifies that writes are limited
     * to lower 32-bits. As such, only the lower 31 bits have full
     * degree of freedom. That is the "useable" counter width.
     */
    *width = PMU_CORE_COUNTER_WIDTH;
}
 
static char *
pfm_core_get_event_name(unsigned int i)
{
    return core_pe[i].pme_name;
}
 
static int
pfm_core_get_event_description(unsigned int ev, char **str)
{
    char *s;
    s = core_pe[ev].pme_desc;
    if (s) {
        *str = strdup(s);
    } else {
        *str = NULL;
    }
    return PFMLIB_SUCCESS;
}
 
static char *
pfm_core_get_event_mask_name(unsigned int ev, unsigned int midx)
{
    return core_pe[ev].pme_umasks[midx].pme_uname;
}
 
static int
pfm_core_get_event_mask_desc(unsigned int ev, unsigned int midx, char **str)
{
    char *s;
 
    s = core_pe[ev].pme_umasks[midx].pme_udesc;
    if (s) {
        *str = strdup(s);
    } else {
        *str = NULL;
    }
    return PFMLIB_SUCCESS;
}
 
static unsigned int
pfm_core_get_num_event_masks(unsigned int ev)
{
    return core_pe[ev].pme_numasks;
}
 
static int
pfm_core_get_event_mask_code(unsigned int ev, unsigned int midx, unsigned int *code)
{
    *code =core_pe[ev].pme_umasks[midx].pme_ucode;
    return PFMLIB_SUCCESS;
}
 
static int
pfm_core_get_cycle_event(pfmlib_event_t *e)
{
    e->event = PME_CORE_UNHALTED_CORE_CYCLES;
    return PFMLIB_SUCCESS;
}
 
static int
pfm_core_get_inst_retired(pfmlib_event_t *e)
{
    e->event = PME_CORE_INSTRUCTIONS_RETIRED;
    return PFMLIB_SUCCESS;
}
 
int
pfm_core_is_pebs(pfmlib_event_t *e)
{
    unsigned int i, n=0;
 
    if (e == NULL || e->event >= PME_CORE_EVENT_COUNT)
        return 0;
 
    if (core_pe[e->event].pme_flags & PFMLIB_CORE_PEBS)
        return 1;
 
    /*
     * ALL unit mask must support PEBS for this test to return true
     */
    for(i=0; i < e->num_masks; i++) {
        /* check for valid unit mask */
        if (e->unit_masks[i] >= core_pe[e->event].pme_numasks)
            return 0;
        if (core_pe[e->event].pme_umasks[e->unit_masks[i]].pme_flags & PFMLIB_CORE_PEBS)
            n++;
    }
    return n > 0 && n == e->num_masks;
}
 
pfm_pmu_support_t core_support={
    .pmu_name       = "Intel Core",
    .pmu_type       = PFMLIB_CORE_PMU,
    .pme_count      = PME_CORE_EVENT_COUNT,
    .pmc_count      = 4,
    .pmd_count      = 14,
    .num_cnt        = 5,
    .get_event_code     = pfm_core_get_event_code,
    .get_event_name     = pfm_core_get_event_name,
    .get_event_counters = pfm_core_get_event_counters,
    .dispatch_events    = pfm_core_dispatch_events,
    .pmu_detect     = pfm_core_detect,
    .pmu_init       = pfm_core_init,
    .get_impl_pmcs      = pfm_core_get_impl_pmcs,
    .get_impl_pmds      = pfm_core_get_impl_pmds,
    .get_impl_counters  = pfm_core_get_impl_counters,
    .get_hw_counter_width   = pfm_core_get_hw_counter_width,
    .get_event_desc         = pfm_core_get_event_description,
    .get_num_event_masks    = pfm_core_get_num_event_masks,
    .get_event_mask_name    = pfm_core_get_event_mask_name,
    .get_event_mask_code    = pfm_core_get_event_mask_code,
    .get_event_mask_desc    = pfm_core_get_event_mask_desc,
    .get_cycle_event    = pfm_core_get_cycle_event,
    .get_inst_retired_event = pfm_core_get_inst_retired
};