PPC64: use hwref instead of cpuid

On CHRP and PowerNV, use the interrupt server number in the cpuref and pcpu
hwref field instead of the device-tree phandle and make the CPU IDs reported
to the scheduler dense and with the BSP at 0.

Submitted by:          Wojciech Macek <wma@semihalf.com>
Obtained from:         Semihalf
Sponsored by:          IBM, QCM Technologies
Differential revision: https://reviews.freebsd.org/D14011
This commit is contained in:
Wojciech Macek 2018-01-29 09:15:38 +00:00
parent b74fb1e713
commit f0393bbf34
Notes: svn2git 2020-12-20 02:59:44 +00:00
svn path=/head/; revision=328534
3 changed files with 222 additions and 224 deletions

View file

@ -1,5 +1,6 @@
/*-
* Copyright (c) 2015 Nathan Whitehorn
* Copyright (c) 2017-2018 Semihalf
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
@ -73,6 +74,7 @@ static struct cpu_group *powernv_smp_topo(platform_t plat);
#endif
static void powernv_reset(platform_t);
static void powernv_cpu_idle(sbintime_t sbt);
static int powernv_cpuref_init(void);
static platform_method_t powernv_methods[] = {
PLATFORMMETHOD(platform_probe, powernv_probe),
@ -100,9 +102,13 @@ static platform_def_t powernv_platform = {
0
};
static struct cpuref platform_cpuref[MAXCPU];
static int platform_cpuref_cnt;
static int platform_cpuref_valid;
PLATFORM_DEF(powernv_platform);
static int powernv_boot_pir;
static uint64_t powernv_boot_pir;
static int
powernv_probe(platform_t plat)
@ -126,12 +132,20 @@ powernv_attach(platform_t plat)
/* Ping OPAL again just to make sure */
opal_check();
#if BYTE_ORDER == LITTLE_ENDIAN
opal_call(OPAL_REINIT_CPUS, 2 /* Little endian */);
#else
opal_call(OPAL_REINIT_CPUS, 1 /* Big endian */);
#endif
cpu_idle_hook = powernv_cpu_idle;
powernv_boot_pir = mfspr(SPR_PIR);
/* Init CPU bits */
powernv_smp_ap_init(plat);
powernv_cpuref_init();
/* Set SLB count from device tree */
cpu = OF_peer(0);
cpu = OF_child(cpu);
@ -218,29 +232,13 @@ powernv_mem_regions(platform_t plat, struct mem_region *phys, int *physsz,
static u_long
powernv_timebase_freq(platform_t plat, struct cpuref *cpuref)
{
phandle_t phandle;
int32_t ticks = -1;
phandle = cpuref->cr_hwref;
OF_getencprop(phandle, "timebase-frequency", &ticks, sizeof(ticks));
if (ticks <= 0)
panic("Unable to determine timebase frequency!");
return (ticks);
}
static int
powernv_smp_first_cpu(platform_t plat, struct cpuref *cpuref)
{
char buf[8];
phandle_t cpu, dev, root;
int res, cpuid;
int res;
int32_t ticks = -1;
root = OF_peer(0);
dev = OF_child(root);
while (dev != 0) {
res = OF_getprop(dev, "name", buf, sizeof(buf));
@ -248,35 +246,96 @@ powernv_smp_first_cpu(platform_t plat, struct cpuref *cpuref)
break;
dev = OF_peer(dev);
}
if (dev == 0) {
/*
* psim doesn't have a name property on the /cpus node,
* but it can be found directly
*/
dev = OF_finddevice("/cpus");
if (dev == 0)
return (ENOENT);
}
cpu = OF_child(dev);
while (cpu != 0) {
for (cpu = OF_child(dev); cpu != 0; cpu = OF_peer(cpu)) {
res = OF_getprop(cpu, "device_type", buf, sizeof(buf));
if (res > 0 && strcmp(buf, "cpu") == 0)
break;
cpu = OF_peer(cpu);
}
if (cpu == 0)
return (ENOENT);
return (512000000);
cpuref->cr_hwref = cpu;
res = OF_getencprop(cpu, "ibm,ppc-interrupt-server#s", &cpuid,
sizeof(cpuid));
if (res <= 0)
res = OF_getencprop(cpu, "reg", &cpuid, sizeof(cpuid));
if (res <= 0)
cpuid = 0;
cpuref->cr_cpuid = cpuid;
OF_getencprop(cpu, "timebase-frequency", &ticks, sizeof(ticks));
if (ticks <= 0)
panic("Unable to determine timebase frequency!");
return (ticks);
}
static int
powernv_cpuref_init(void)
{
phandle_t cpu, dev;
char buf[32];
int a, res, tmp_cpuref_cnt;
static struct cpuref tmp_cpuref[MAXCPU];
cell_t interrupt_servers[32];
uint64_t bsp;
if (platform_cpuref_valid)
return (0);
dev = OF_peer(0);
dev = OF_child(dev);
while (dev != 0) {
res = OF_getprop(dev, "name", buf, sizeof(buf));
if (res > 0 && strcmp(buf, "cpus") == 0)
break;
dev = OF_peer(dev);
}
bsp = 0;
tmp_cpuref_cnt = 0;
for (cpu = OF_child(dev); cpu != 0; cpu = OF_peer(cpu)) {
res = OF_getprop(cpu, "device_type", buf, sizeof(buf));
if (res > 0 && strcmp(buf, "cpu") == 0) {
res = OF_getproplen(cpu, "ibm,ppc-interrupt-server#s");
if (res > 0) {
OF_getencprop(cpu, "ibm,ppc-interrupt-server#s",
interrupt_servers, res);
for (a = 0; a < res/sizeof(cell_t); a++) {
tmp_cpuref[tmp_cpuref_cnt].cr_hwref = interrupt_servers[a];
tmp_cpuref[tmp_cpuref_cnt].cr_cpuid = tmp_cpuref_cnt;
if (interrupt_servers[a] == (uint32_t)powernv_boot_pir)
bsp = tmp_cpuref_cnt;
tmp_cpuref_cnt++;
}
}
}
}
/* Map IDs, so BSP has CPUID 0 regardless of hwref */
for (a = bsp; a < tmp_cpuref_cnt; a++) {
platform_cpuref[platform_cpuref_cnt].cr_hwref = tmp_cpuref[a].cr_hwref;
platform_cpuref[platform_cpuref_cnt].cr_cpuid = platform_cpuref_cnt;
platform_cpuref_cnt++;
}
for (a = 0; a < bsp; a++) {
platform_cpuref[platform_cpuref_cnt].cr_hwref = tmp_cpuref[a].cr_hwref;
platform_cpuref[platform_cpuref_cnt].cr_cpuid = platform_cpuref_cnt;
platform_cpuref_cnt++;
}
platform_cpuref_valid = 1;
return (0);
}
static int
powernv_smp_first_cpu(platform_t plat, struct cpuref *cpuref)
{
if (platform_cpuref_valid == 0)
return (EINVAL);
cpuref->cr_cpuid = 0;
cpuref->cr_hwref = platform_cpuref[0].cr_hwref;
return (0);
}
@ -284,43 +343,17 @@ powernv_smp_first_cpu(platform_t plat, struct cpuref *cpuref)
static int
powernv_smp_next_cpu(platform_t plat, struct cpuref *cpuref)
{
char buf[8];
phandle_t cpu;
int i, res, cpuid;
int id;
/* Check for whether it should be the next thread */
res = OF_getproplen(cpuref->cr_hwref, "ibm,ppc-interrupt-server#s");
if (res > 0) {
cell_t interrupt_servers[res/sizeof(cell_t)];
OF_getencprop(cpuref->cr_hwref, "ibm,ppc-interrupt-server#s",
interrupt_servers, res);
for (i = 0; i < res/sizeof(cell_t) - 1; i++) {
if (interrupt_servers[i] == cpuref->cr_cpuid) {
cpuref->cr_cpuid = interrupt_servers[i+1];
return (0);
}
}
}
if (platform_cpuref_valid == 0)
return (EINVAL);
/* Next CPU core/package */
cpu = OF_peer(cpuref->cr_hwref);
while (cpu != 0) {
res = OF_getprop(cpu, "device_type", buf, sizeof(buf));
if (res > 0 && strcmp(buf, "cpu") == 0)
break;
cpu = OF_peer(cpu);
}
if (cpu == 0)
id = cpuref->cr_cpuid + 1;
if (id >= platform_cpuref_cnt)
return (ENOENT);
cpuref->cr_hwref = cpu;
res = OF_getencprop(cpu, "ibm,ppc-interrupt-server#s", &cpuid,
sizeof(cpuid));
if (res <= 0)
res = OF_getencprop(cpu, "reg", &cpuid, sizeof(cpuid));
if (res <= 0)
cpuid = 0;
cpuref->cr_cpuid = cpuid;
cpuref->cr_cpuid = platform_cpuref[id].cr_cpuid;
cpuref->cr_hwref = platform_cpuref[id].cr_hwref;
return (0);
}
@ -328,35 +361,9 @@ powernv_smp_next_cpu(platform_t plat, struct cpuref *cpuref)
static int
powernv_smp_get_bsp(platform_t plat, struct cpuref *cpuref)
{
phandle_t chosen;
int cpuid, res;
struct cpuref i;
chosen = OF_finddevice("/chosen");
if (chosen == 0)
return (ENOENT);
res = OF_getencprop(chosen, "fdtbootcpu", &cpuid, sizeof(cpuid));
if (res < 0)
return (ENOENT);
/* XXX: FDT from kexec lies sometimes. PIR seems not to. */
if (cpuid == 0)
cpuid = powernv_boot_pir;
cpuref->cr_cpuid = cpuid;
if (powernv_smp_first_cpu(plat, &i) != 0)
return (ENOENT);
cpuref->cr_hwref = i.cr_hwref;
do {
if (i.cr_cpuid == cpuid) {
cpuref->cr_hwref = i.cr_hwref;
break;
}
} while (powernv_smp_next_cpu(plat, &i) == 0);
cpuref->cr_cpuid = platform_cpuref[0].cr_cpuid;
cpuref->cr_hwref = platform_cpuref[0].cr_hwref;
return (0);
}
@ -369,10 +376,10 @@ powernv_smp_start_cpu(platform_t plat, struct pcpu *pc)
ap_pcpu = pc;
powerpc_sync();
result = opal_call(OPAL_START_CPU, pc->pc_cpuid, EXC_RST);
result = opal_call(OPAL_START_CPU, pc->pc_hwref, EXC_RST);
if (result != OPAL_SUCCESS) {
printf("OPAL error (%d): unable to start AP %d\n",
result, pc->pc_cpuid);
result, (int)pc->pc_hwref);
return (ENXIO);
}
@ -382,37 +389,49 @@ powernv_smp_start_cpu(platform_t plat, struct pcpu *pc)
static struct cpu_group *
powernv_smp_topo(platform_t plat)
{
struct pcpu *pc, *last_pc;
int i, ncores, ncpus;
char buf[8];
phandle_t cpu, dev, root;
int res, nthreads;
ncores = ncpus = 0;
last_pc = NULL;
CPU_FOREACH(i) {
pc = pcpu_find(i);
if (pc == NULL)
continue;
if (last_pc == NULL || pc->pc_hwref != last_pc->pc_hwref)
ncores++;
last_pc = pc;
ncpus++;
root = OF_peer(0);
dev = OF_child(root);
while (dev != 0) {
res = OF_getprop(dev, "name", buf, sizeof(buf));
if (res > 0 && strcmp(buf, "cpus") == 0)
break;
dev = OF_peer(dev);
}
if (ncpus % ncores != 0) {
nthreads = 1;
for (cpu = OF_child(dev); cpu != 0; cpu = OF_peer(cpu)) {
res = OF_getprop(cpu, "device_type", buf, sizeof(buf));
if (res <= 0 || strcmp(buf, "cpu") != 0)
continue;
res = OF_getproplen(cpu, "ibm,ppc-interrupt-server#s");
if (res >= 0)
nthreads = res / sizeof(cell_t);
else
nthreads = 1;
break;
}
if (mp_ncpus % nthreads != 0) {
printf("WARNING: Irregular SMP topology. Performance may be "
"suboptimal (%d CPUS, %d cores)\n", ncpus, ncores);
"suboptimal (%d threads, %d on first core)\n",
mp_ncpus, nthreads);
return (smp_topo_none());
}
/* Don't do anything fancier for non-threaded SMP */
if (ncpus == ncores)
if (nthreads == 1)
return (smp_topo_none());
#ifdef NOTYET /* smp_topo_1level() fails with non-consecutive CPU IDs */
return (smp_topo_1level(CG_SHARE_L1, ncpus / ncores, CG_FLAG_SMT));
#else
return (smp_topo_none());
#endif
return (smp_topo_1level(CG_SHARE_L1, nthreads, CG_FLAG_SMT));
}
#endif
static void

View file

@ -75,6 +75,7 @@ static int chrp_smp_first_cpu(platform_t, struct cpuref *cpuref);
static int chrp_smp_next_cpu(platform_t, struct cpuref *cpuref);
static int chrp_smp_get_bsp(platform_t, struct cpuref *cpuref);
static void chrp_smp_ap_init(platform_t);
static int chrp_cpuref_init(void);
#ifdef SMP
static int chrp_smp_start_cpu(platform_t, struct pcpu *cpu);
static struct cpu_group *chrp_smp_topo(platform_t plat);
@ -85,6 +86,10 @@ static void chrp_reset(platform_t);
static void phyp_cpu_idle(sbintime_t sbt);
#endif
static struct cpuref platform_cpuref[MAXCPU];
static int platform_cpuref_cnt;
static int platform_cpuref_valid;
static platform_method_t chrp_methods[] = {
PLATFORMMETHOD(platform_probe, chrp_probe),
PLATFORMMETHOD(platform_attach, chrp_attach),
@ -159,6 +164,7 @@ chrp_attach(platform_t plat)
chrp_smp_ap_init(plat);
}
#endif
chrp_cpuref_init();
/* Some systems (e.g. QEMU) need Open Firmware to stand down */
ofw_quiesce();
@ -295,48 +301,12 @@ chrp_timebase_freq(platform_t plat, struct cpuref *cpuref)
static int
chrp_smp_first_cpu(platform_t plat, struct cpuref *cpuref)
{
char buf[8];
phandle_t cpu, dev, root;
int res, cpuid;
root = OF_peer(0);
if (platform_cpuref_valid == 0)
return (EINVAL);
dev = OF_child(root);
while (dev != 0) {
res = OF_getprop(dev, "name", buf, sizeof(buf));
if (res > 0 && strcmp(buf, "cpus") == 0)
break;
dev = OF_peer(dev);
}
if (dev == 0) {
/*
* psim doesn't have a name property on the /cpus node,
* but it can be found directly
*/
dev = OF_finddevice("/cpus");
if (dev == 0)
return (ENOENT);
}
cpu = OF_child(dev);
while (cpu != 0) {
res = OF_getprop(cpu, "device_type", buf, sizeof(buf));
if (res > 0 && strcmp(buf, "cpu") == 0)
break;
cpu = OF_peer(cpu);
}
if (cpu == 0)
return (ENOENT);
cpuref->cr_hwref = cpu;
res = OF_getencprop(cpu, "ibm,ppc-interrupt-server#s", &cpuid,
sizeof(cpuid));
if (res <= 0)
res = OF_getencprop(cpu, "reg", &cpuid, sizeof(cpuid));
if (res <= 0)
cpuid = 0;
cpuref->cr_cpuid = cpuid;
cpuref->cr_cpuid = 0;
cpuref->cr_hwref = platform_cpuref[0].cr_hwref;
return (0);
}
@ -344,43 +314,17 @@ chrp_smp_first_cpu(platform_t plat, struct cpuref *cpuref)
static int
chrp_smp_next_cpu(platform_t plat, struct cpuref *cpuref)
{
char buf[8];
phandle_t cpu;
int i, res, cpuid;
int id;
/* Check for whether it should be the next thread */
res = OF_getproplen(cpuref->cr_hwref, "ibm,ppc-interrupt-server#s");
if (res > 0) {
cell_t interrupt_servers[res/sizeof(cell_t)];
OF_getencprop(cpuref->cr_hwref, "ibm,ppc-interrupt-server#s",
interrupt_servers, res);
for (i = 0; i < res/sizeof(cell_t) - 1; i++) {
if (interrupt_servers[i] == cpuref->cr_cpuid) {
cpuref->cr_cpuid = interrupt_servers[i+1];
return (0);
}
}
}
if (platform_cpuref_valid == 0)
return (EINVAL);
/* Next CPU core/package */
cpu = OF_peer(cpuref->cr_hwref);
while (cpu != 0) {
res = OF_getprop(cpu, "device_type", buf, sizeof(buf));
if (res > 0 && strcmp(buf, "cpu") == 0)
break;
cpu = OF_peer(cpu);
}
if (cpu == 0)
id = cpuref->cr_cpuid + 1;
if (id >= platform_cpuref_cnt)
return (ENOENT);
cpuref->cr_hwref = cpu;
res = OF_getencprop(cpu, "ibm,ppc-interrupt-server#s", &cpuid,
sizeof(cpuid));
if (res <= 0)
res = OF_getencprop(cpu, "reg", &cpuid, sizeof(cpuid));
if (res <= 0)
cpuid = 0;
cpuref->cr_cpuid = cpuid;
cpuref->cr_cpuid = platform_cpuref[id].cr_cpuid;
cpuref->cr_hwref = platform_cpuref[id].cr_hwref;
return (0);
}
@ -388,33 +332,60 @@ chrp_smp_next_cpu(platform_t plat, struct cpuref *cpuref)
static int
chrp_smp_get_bsp(platform_t plat, struct cpuref *cpuref)
{
ihandle_t inst;
phandle_t bsp, chosen;
int res, cpuid;
chosen = OF_finddevice("/chosen");
if (chosen == 0)
return (ENXIO);
cpuref->cr_cpuid = platform_cpuref[0].cr_cpuid;
cpuref->cr_hwref = platform_cpuref[0].cr_hwref;
return (0);
}
res = OF_getencprop(chosen, "cpu", &inst, sizeof(inst));
if (res < 0)
return (ENXIO);
static int
chrp_cpuref_init(void)
{
phandle_t cpu, dev;
char buf[32];
int a, res;
cell_t interrupt_servers[32];
uint64_t bsp;
bsp = OF_instance_to_package(inst);
if (platform_cpuref_valid)
return (0);
/* Pick the primary thread. Can it be any other? */
cpuref->cr_hwref = bsp;
res = OF_getencprop(bsp, "ibm,ppc-interrupt-server#s", &cpuid,
sizeof(cpuid));
if (res <= 0)
res = OF_getencprop(bsp, "reg", &cpuid, sizeof(cpuid));
if (res <= 0)
cpuid = 0;
cpuref->cr_cpuid = cpuid;
dev = OF_peer(0);
dev = OF_child(dev);
while (dev != 0) {
res = OF_getprop(dev, "name", buf, sizeof(buf));
if (res > 0 && strcmp(buf, "cpus") == 0)
break;
dev = OF_peer(dev);
}
bsp = 0;
for (cpu = OF_child(dev); cpu != 0; cpu = OF_peer(cpu)) {
res = OF_getprop(cpu, "device_type", buf, sizeof(buf));
if (res > 0 && strcmp(buf, "cpu") == 0) {
res = OF_getproplen(cpu, "ibm,ppc-interrupt-server#s");
if (res > 0) {
OF_getencprop(cpu, "ibm,ppc-interrupt-server#s",
interrupt_servers, res);
for (a = 0; a < res/sizeof(cell_t); a++) {
platform_cpuref[platform_cpuref_cnt].cr_hwref = interrupt_servers[a];
platform_cpuref[platform_cpuref_cnt].cr_cpuid = platform_cpuref_cnt;
platform_cpuref_cnt++;
}
}
}
}
platform_cpuref_valid = 1;
return (0);
}
#ifdef SMP
static int
chrp_smp_start_cpu(platform_t plat, struct pcpu *pc)
@ -438,7 +409,7 @@ chrp_smp_start_cpu(platform_t plat, struct pcpu *pc)
ap_pcpu = pc;
powerpc_sync();
result = rtas_call_method(start_cpu, 3, 1, pc->pc_cpuid, EXC_RST, pc,
result = rtas_call_method(start_cpu, 3, 1, pc->pc_hwref, EXC_RST, pc,
&err);
if (result < 0 || err != 0) {
printf("RTAS error (%d/%d): unable to start AP %d\n",
@ -514,8 +485,8 @@ chrp_smp_ap_init(platform_t platform)
{
if (!(mfmsr() & PSL_HV)) {
/* Register VPA */
phyp_hcall(H_REGISTER_VPA, 1UL, PCPU_GET(cpuid),
splpar_vpa[PCPU_GET(cpuid)]);
phyp_hcall(H_REGISTER_VPA, 1UL, PCPU_GET(hwref),
splpar_vpa[PCPU_GET(hwref)]);
/* Set interrupt priority */
phyp_hcall(H_CPPR, 0xff);

View file

@ -263,6 +263,10 @@ xicp_bind(device_t dev, u_int irq, cpuset_t cpumask)
cell_t status, cpu;
int ncpus, i, error;
/* Ignore IPIs */
if (irq == MAX_XICP_IRQS)
return;
/*
* This doesn't appear to actually support affinity groups, so pick a
* random CPU.
@ -281,6 +285,8 @@ xicp_bind(device_t dev, u_int irq, cpuset_t cpumask)
ncpus++;
}
cpu = pcpu_find(cpu)->pc_hwref;
/* XXX: super inefficient */
for (i = 0; i < sc->nintvecs; i++) {
if (sc->intvecs[i].irq == irq) {
@ -312,9 +318,9 @@ xicp_dispatch(device_t dev, struct trapframe *tf)
#ifdef POWERNV
if (mfmsr() & PSL_HV) {
regs = xicp_mem_for_cpu(PCPU_GET(cpuid));
regs = xicp_mem_for_cpu(PCPU_GET(hwref));
KASSERT(regs != NULL,
("Can't find regs for CPU %d", PCPU_GET(cpuid)));
("Can't find regs for CPU %ld", (uintptr_t)PCPU_GET(hwref)));
}
#endif
@ -343,7 +349,7 @@ xicp_dispatch(device_t dev, struct trapframe *tf)
if (regs)
bus_write_1(regs, 12, 0xff);
else
phyp_hcall(H_IPI, (uint64_t)(PCPU_GET(cpuid)),
phyp_hcall(H_IPI, (uint64_t)(PCPU_GET(hwref)),
0xff);
}
@ -370,7 +376,7 @@ xicp_enable(device_t dev, u_int irq, u_int vector)
("Too many XICP interrupts"));
/* Bind to this CPU to start: distrib. ID is last entry in gserver# */
cpu = PCPU_GET(cpuid);
cpu = PCPU_GET(hwref);
mtx_lock(&sc->sc_mtx);
sc->intvecs[sc->nintvecs].irq = irq;
@ -411,7 +417,7 @@ xicp_eoi(device_t dev, u_int irq)
#ifdef POWERNV
if (mfmsr() & PSL_HV)
bus_write_4(xicp_mem_for_cpu(PCPU_GET(cpuid)), 4, xirr);
bus_write_4(xicp_mem_for_cpu(PCPU_GET(hwref)), 4, xirr);
else
#endif
phyp_hcall(H_EOI, xirr);
@ -422,6 +428,8 @@ xicp_ipi(device_t dev, u_int cpu)
{
#ifdef POWERNV
cpu = pcpu_find(cpu)->pc_hwref;
if (mfmsr() & PSL_HV)
bus_write_1(xicp_mem_for_cpu(cpu), 12, XICP_PRIORITY);
else