linux/arch/x86/kernel/apic/numaq_32.c

526 lines
13 KiB
C
Raw Normal View History

/*
* Written by: Patricia Gaughen, IBM Corporation
*
* Copyright (C) 2002, IBM Corp.
* Copyright (C) 2009, Red Hat, Inc., Ingo Molnar
*
* All rights reserved.
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful, but
* WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
* NON INFRINGEMENT. See the GNU General Public License for more
* details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
*
* Send feedback to <gone@us.ibm.com>
*/
#include <linux/nodemask.h>
#include <linux/topology.h>
#include <linux/bootmem.h>
#include <linux/memblock.h>
#include <linux/threads.h>
#include <linux/cpumask.h>
#include <linux/kernel.h>
#include <linux/mmzone.h>
#include <linux/module.h>
#include <linux/string.h>
#include <linux/init.h>
#include <linux/numa.h>
#include <linux/smp.h>
#include <linux/io.h>
#include <linux/mm.h>
#include <asm/processor.h>
#include <asm/fixmap.h>
#include <asm/mpspec.h>
#include <asm/numaq.h>
#include <asm/setup.h>
#include <asm/apic.h>
#include <asm/e820.h>
#include <asm/ipi.h>
int found_numaq;
/*
* Have to match translation table entries to main table entries by counter
* hence the mpc_record variable .... can't see a less disgusting way of
* doing this ....
*/
struct mpc_trans {
unsigned char mpc_type;
unsigned char trans_len;
unsigned char trans_type;
unsigned char trans_quad;
unsigned char trans_global;
unsigned char trans_local;
unsigned short trans_reserved;
};
static int mpc_record;
static struct mpc_trans *translation_table[MAX_MPC_ENTRY];
int mp_bus_id_to_node[MAX_MP_BUSSES];
int mp_bus_id_to_local[MAX_MP_BUSSES];
int quad_local_to_mp_bus_id[NR_CPUS/4][4];
static inline void numaq_register_node(int node, struct sys_cfg_data *scd)
{
struct eachquadmem *eq = scd->eq + node;
u64 start = (u64)(eq->hi_shrd_mem_start - eq->priv_mem_size) << 20;
u64 end = (u64)(eq->hi_shrd_mem_start + eq->hi_shrd_mem_size) << 20;
int ret;
node_set(node, numa_nodes_parsed);
ret = numa_add_memblk(node, start, end);
BUG_ON(ret < 0);
}
/*
* Function: smp_dump_qct()
*
* Description: gets memory layout from the quad config table. This
* function also updates numa_nodes_parsed with the nodes (quads) present.
*/
static void __init smp_dump_qct(void)
{
struct sys_cfg_data *scd;
int node;
scd = (void *)__va(SYS_CFG_DATA_PRIV_ADDR);
for_each_node(node) {
if (scd->quads_present31_0 & (1 << node))
numaq_register_node(node, scd);
}
}
void __cpuinit numaq_tsc_disable(void)
{
if (!found_numaq)
return;
if (num_online_nodes() > 1) {
printk(KERN_DEBUG "NUMAQ: disabling TSC\n");
setup_clear_cpu_cap(X86_FEATURE_TSC);
}
}
static void __init numaq_tsc_init(void)
{
numaq_tsc_disable();
}
static inline int generate_logical_apicid(int quad, int phys_apicid)
{
return (quad << 4) + (phys_apicid ? phys_apicid << 1 : 1);
}
/* x86_quirks member */
static int mpc_apic_id(struct mpc_cpu *m)
{
int quad = translation_table[mpc_record]->trans_quad;
int logical_apicid = generate_logical_apicid(quad, m->apicid);
printk(KERN_DEBUG
"Processor #%d %u:%u APIC version %d (quad %d, apic %d)\n",
m->apicid, (m->cpufeature & CPU_FAMILY_MASK) >> 8,
(m->cpufeature & CPU_MODEL_MASK) >> 4,
m->apicver, quad, logical_apicid);
return logical_apicid;
}
/* x86_quirks member */
static void mpc_oem_bus_info(struct mpc_bus *m, char *name)
{
int quad = translation_table[mpc_record]->trans_quad;
int local = translation_table[mpc_record]->trans_local;
mp_bus_id_to_node[m->busid] = quad;
mp_bus_id_to_local[m->busid] = local;
printk(KERN_INFO "Bus #%d is %s (node %d)\n", m->busid, name, quad);
}
/* x86_quirks member */
static void mpc_oem_pci_bus(struct mpc_bus *m)
{
int quad = translation_table[mpc_record]->trans_quad;
int local = translation_table[mpc_record]->trans_local;
quad_local_to_mp_bus_id[quad][local] = m->busid;
}
/*
* Called from mpparse code.
* mode = 0: prescan
* mode = 1: one mpc entry scanned
*/
static void numaq_mpc_record(unsigned int mode)
{
if (!mode)
mpc_record = 0;
else
mpc_record++;
}
static void __init MP_translation_info(struct mpc_trans *m)
{
printk(KERN_INFO
"Translation: record %d, type %d, quad %d, global %d, local %d\n",
mpc_record, m->trans_type, m->trans_quad, m->trans_global,
m->trans_local);
if (mpc_record >= MAX_MPC_ENTRY)
printk(KERN_ERR "MAX_MPC_ENTRY exceeded!\n");
else
translation_table[mpc_record] = m; /* stash this for later */
if (m->trans_quad < MAX_NUMNODES && !node_online(m->trans_quad))
node_set_online(m->trans_quad);
}
static int __init mpf_checksum(unsigned char *mp, int len)
{
int sum = 0;
while (len--)
sum += *mp++;
return sum & 0xFF;
}
/*
* Read/parse the MPC oem tables
*/
static void __init smp_read_mpc_oem(struct mpc_table *mpc)
{
struct mpc_oemtable *oemtable = (void *)(long)mpc->oemptr;
int count = sizeof(*oemtable); /* the header size */
unsigned char *oemptr = ((unsigned char *)oemtable) + count;
mpc_record = 0;
printk(KERN_INFO
"Found an OEM MPC table at %8p - parsing it...\n", oemtable);
if (memcmp(oemtable->signature, MPC_OEM_SIGNATURE, 4)) {
printk(KERN_WARNING
"SMP mpc oemtable: bad signature [%c%c%c%c]!\n",
oemtable->signature[0], oemtable->signature[1],
oemtable->signature[2], oemtable->signature[3]);
return;
}
if (mpf_checksum((unsigned char *)oemtable, oemtable->length)) {
printk(KERN_WARNING "SMP oem mptable: checksum error!\n");
return;
}
while (count < oemtable->length) {
switch (*oemptr) {
case MP_TRANSLATION:
{
struct mpc_trans *m = (void *)oemptr;
MP_translation_info(m);
oemptr += sizeof(*m);
count += sizeof(*m);
++mpc_record;
break;
}
default:
printk(KERN_WARNING
"Unrecognised OEM table entry type! - %d\n",
(int)*oemptr);
return;
}
}
}
static __init void early_check_numaq(void)
{
/*
* get boot-time SMP configuration:
*/
if (smp_found_config)
early_get_smp_config();
if (found_numaq) {
x86_init.mpparse.mpc_record = numaq_mpc_record;
x86_init.mpparse.setup_ioapic_ids = x86_init_noop;
x86_init.mpparse.mpc_apic_id = mpc_apic_id;
x86_init.mpparse.smp_read_mpc_oem = smp_read_mpc_oem;
x86_init.mpparse.mpc_oem_pci_bus = mpc_oem_pci_bus;
x86_init.mpparse.mpc_oem_bus_info = mpc_oem_bus_info;
x86_init.timers.tsc_pre_init = numaq_tsc_init;
x86_init.pci.init = pci_numaq_init;
}
}
int __init numaq_numa_init(void)
{
early_check_numaq();
if (!found_numaq)
return -ENOENT;
smp_dump_qct();
return 0;
}
#define NUMAQ_APIC_DFR_VALUE (APIC_DFR_CLUSTER)
static inline unsigned int numaq_get_apic_id(unsigned long x)
{
return (x >> 24) & 0x0F;
}
static inline void numaq_send_IPI_mask(const struct cpumask *mask, int vector)
{
default_send_IPI_mask_sequence_logical(mask, vector);
}
static inline void numaq_send_IPI_allbutself(int vector)
{
default_send_IPI_mask_allbutself_logical(cpu_online_mask, vector);
}
static inline void numaq_send_IPI_all(int vector)
{
numaq_send_IPI_mask(cpu_online_mask, vector);
}
#define NUMAQ_TRAMPOLINE_PHYS_LOW (0x8)
#define NUMAQ_TRAMPOLINE_PHYS_HIGH (0xa)
/*
* Because we use NMIs rather than the INIT-STARTUP sequence to
* bootstrap the CPUs, the APIC may be in a weird state. Kick it:
*/
static inline void numaq_smp_callin_clear_local_apic(void)
{
clear_local_APIC();
}
static inline const struct cpumask *numaq_target_cpus(void)
{
return cpu_all_mask;
}
static unsigned long numaq_check_apicid_used(physid_mask_t *map, int apicid)
{
return physid_isset(apicid, *map);
}
static inline unsigned long numaq_check_apicid_present(int bit)
{
return physid_isset(bit, phys_cpu_present_map);
}
static inline int numaq_apic_id_registered(void)
{
return 1;
}
static inline void numaq_init_apic_ldr(void)
{
/* Already done in NUMA-Q firmware */
}
static inline void numaq_setup_apic_routing(void)
{
printk(KERN_INFO
"Enabling APIC mode: NUMA-Q. Using %d I/O APICs\n",
nr_ioapics);
}
/*
* Skip adding the timer int on secondary nodes, which causes
* a small but painful rift in the time-space continuum.
*/
static inline int numaq_multi_timer_check(int apic, int irq)
{
return apic != 0 && irq == 0;
}
static inline void numaq_ioapic_phys_id_map(physid_mask_t *phys_map, physid_mask_t *retmap)
{
/* We don't have a good way to do this yet - hack */
return physids_promote(0xFUL, retmap);
}
/*
* Supporting over 60 cpus on NUMA-Q requires a locality-dependent
* cpu to APIC ID relation to properly interact with the intelligent
* mode of the cluster controller.
*/
static inline int numaq_cpu_present_to_apicid(int mps_cpu)
{
if (mps_cpu < 60)
return ((mps_cpu >> 2) << 4) | (1 << (mps_cpu & 0x3));
else
return BAD_APICID;
}
static inline int numaq_apicid_to_node(int logical_apicid)
{
return logical_apicid >> 4;
}
static int numaq_numa_cpu_node(int cpu)
{
int logical_apicid = early_per_cpu(x86_cpu_to_logical_apicid, cpu);
if (logical_apicid != BAD_APICID)
return numaq_apicid_to_node(logical_apicid);
return NUMA_NO_NODE;
}
static void numaq_apicid_to_cpu_present(int logical_apicid, physid_mask_t *retmap)
{
int node = numaq_apicid_to_node(logical_apicid);
int cpu = __ffs(logical_apicid & 0xf);
physid_set_mask_of_physid(cpu + 4*node, retmap);
}
/* Where the IO area was mapped on multiquad, always 0 otherwise */
void *xquad_portio;
static inline int numaq_check_phys_apicid_present(int phys_apicid)
{
return 1;
}
/*
* We use physical apicids here, not logical, so just return the default
* physical broadcast to stop people from breaking us
*/
x86/apic: Make cpu_mask_to_apicid() operations return error code Current cpu_mask_to_apicid() and cpu_mask_to_apicid_and() implementations have few shortcomings: 1. A value returned by cpu_mask_to_apicid() is written to hardware registers unconditionally. Should BAD_APICID get ever returned it will be written to a hardware too. But the value of BAD_APICID is not universal across all hardware in all modes and might cause unexpected results, i.e. interrupts might get routed to CPUs that are not configured to receive it. 2. Because the value of BAD_APICID is not universal it is counter- intuitive to return it for a hardware where it does not make sense (i.e. x2apic). 3. cpu_mask_to_apicid_and() operation is thought as an complement to cpu_mask_to_apicid() that only applies a AND mask on top of a cpumask being passed. Yet, as consequence of 18374d8 commit the two operations are inconsistent in that of: cpu_mask_to_apicid() should not get a offline CPU with the cpumask cpu_mask_to_apicid_and() should not fail and return BAD_APICID These limitations are impossible to realize just from looking at the operations prototypes. Most of these shortcomings are resolved by returning a error code instead of BAD_APICID. As the result, faults are reported back early rather than possibilities to cause a unexpected behaviour exist (in case of [1]). The only exception is setup_timer_IRQ0_pin() routine. Although obviously controversial to this fix, its existing behaviour is preserved to not break the fragile check_timer() and would better addressed in a separate fix. Signed-off-by: Alexander Gordeev <agordeev@redhat.com> Acked-by: Suresh Siddha <suresh.b.siddha@intel.com> Cc: Yinghai Lu <yinghai@kernel.org> Link: http://lkml.kernel.org/r/20120607131559.GF4759@dhcp-26-207.brq.redhat.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-06-07 13:15:59 +00:00
static int
numaq_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
x86/apic: Make cpu_mask_to_apicid() operations return error code Current cpu_mask_to_apicid() and cpu_mask_to_apicid_and() implementations have few shortcomings: 1. A value returned by cpu_mask_to_apicid() is written to hardware registers unconditionally. Should BAD_APICID get ever returned it will be written to a hardware too. But the value of BAD_APICID is not universal across all hardware in all modes and might cause unexpected results, i.e. interrupts might get routed to CPUs that are not configured to receive it. 2. Because the value of BAD_APICID is not universal it is counter- intuitive to return it for a hardware where it does not make sense (i.e. x2apic). 3. cpu_mask_to_apicid_and() operation is thought as an complement to cpu_mask_to_apicid() that only applies a AND mask on top of a cpumask being passed. Yet, as consequence of 18374d8 commit the two operations are inconsistent in that of: cpu_mask_to_apicid() should not get a offline CPU with the cpumask cpu_mask_to_apicid_and() should not fail and return BAD_APICID These limitations are impossible to realize just from looking at the operations prototypes. Most of these shortcomings are resolved by returning a error code instead of BAD_APICID. As the result, faults are reported back early rather than possibilities to cause a unexpected behaviour exist (in case of [1]). The only exception is setup_timer_IRQ0_pin() routine. Although obviously controversial to this fix, its existing behaviour is preserved to not break the fragile check_timer() and would better addressed in a separate fix. Signed-off-by: Alexander Gordeev <agordeev@redhat.com> Acked-by: Suresh Siddha <suresh.b.siddha@intel.com> Cc: Yinghai Lu <yinghai@kernel.org> Link: http://lkml.kernel.org/r/20120607131559.GF4759@dhcp-26-207.brq.redhat.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-06-07 13:15:59 +00:00
const struct cpumask *andmask,
unsigned int *apicid)
{
x86/apic: Make cpu_mask_to_apicid() operations return error code Current cpu_mask_to_apicid() and cpu_mask_to_apicid_and() implementations have few shortcomings: 1. A value returned by cpu_mask_to_apicid() is written to hardware registers unconditionally. Should BAD_APICID get ever returned it will be written to a hardware too. But the value of BAD_APICID is not universal across all hardware in all modes and might cause unexpected results, i.e. interrupts might get routed to CPUs that are not configured to receive it. 2. Because the value of BAD_APICID is not universal it is counter- intuitive to return it for a hardware where it does not make sense (i.e. x2apic). 3. cpu_mask_to_apicid_and() operation is thought as an complement to cpu_mask_to_apicid() that only applies a AND mask on top of a cpumask being passed. Yet, as consequence of 18374d8 commit the two operations are inconsistent in that of: cpu_mask_to_apicid() should not get a offline CPU with the cpumask cpu_mask_to_apicid_and() should not fail and return BAD_APICID These limitations are impossible to realize just from looking at the operations prototypes. Most of these shortcomings are resolved by returning a error code instead of BAD_APICID. As the result, faults are reported back early rather than possibilities to cause a unexpected behaviour exist (in case of [1]). The only exception is setup_timer_IRQ0_pin() routine. Although obviously controversial to this fix, its existing behaviour is preserved to not break the fragile check_timer() and would better addressed in a separate fix. Signed-off-by: Alexander Gordeev <agordeev@redhat.com> Acked-by: Suresh Siddha <suresh.b.siddha@intel.com> Cc: Yinghai Lu <yinghai@kernel.org> Link: http://lkml.kernel.org/r/20120607131559.GF4759@dhcp-26-207.brq.redhat.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-06-07 13:15:59 +00:00
*apicid = 0x0F;
return 0;
}
/* No NUMA-Q box has a HT CPU, but it can't hurt to use the default code. */
static inline int numaq_phys_pkg_id(int cpuid_apic, int index_msb)
{
return cpuid_apic >> index_msb;
}
static int
numaq_mps_oem_check(struct mpc_table *mpc, char *oem, char *productid)
{
if (strncmp(oem, "IBM NUMA", 8))
printk(KERN_ERR "Warning! Not a NUMA-Q system!\n");
else
found_numaq = 1;
return found_numaq;
}
static int probe_numaq(void)
{
/* already know from get_memcfg_numaq() */
return found_numaq;
}
static void numaq_setup_portio_remap(void)
{
int num_quads = num_online_nodes();
if (num_quads <= 1)
return;
printk(KERN_INFO
"Remapping cross-quad port I/O for %d quads\n", num_quads);
xquad_portio = ioremap(XQUAD_PORTIO_BASE, num_quads*XQUAD_PORTIO_QUAD);
printk(KERN_INFO
"xquad_portio vaddr 0x%08lx, len %08lx\n",
(u_long) xquad_portio, (u_long) num_quads*XQUAD_PORTIO_QUAD);
}
/* Use __refdata to keep false positive warning calm. */
static struct apic __refdata apic_numaq = {
.name = "NUMAQ",
.probe = probe_numaq,
.acpi_madt_oem_check = NULL,
.apic_id_valid = default_apic_id_valid,
.apic_id_registered = numaq_apic_id_registered,
.irq_delivery_mode = dest_LowestPrio,
/* physical delivery on LOCAL quad: */
.irq_dest_mode = 0,
.target_cpus = numaq_target_cpus,
.disable_esr = 1,
.dest_logical = APIC_DEST_LOGICAL,
.check_apicid_used = numaq_check_apicid_used,
.check_apicid_present = numaq_check_apicid_present,
.vector_allocation_domain = flat_vector_allocation_domain,
.init_apic_ldr = numaq_init_apic_ldr,
.ioapic_phys_id_map = numaq_ioapic_phys_id_map,
.setup_apic_routing = numaq_setup_apic_routing,
.multi_timer_check = numaq_multi_timer_check,
.cpu_present_to_apicid = numaq_cpu_present_to_apicid,
.apicid_to_cpu_present = numaq_apicid_to_cpu_present,
.setup_portio_remap = numaq_setup_portio_remap,
.check_phys_apicid_present = numaq_check_phys_apicid_present,
.enable_apic_mode = NULL,
.phys_pkg_id = numaq_phys_pkg_id,
.mps_oem_check = numaq_mps_oem_check,
.get_apic_id = numaq_get_apic_id,
.set_apic_id = NULL,
.apic_id_mask = 0x0F << 24,
.cpu_mask_to_apicid_and = numaq_cpu_mask_to_apicid_and,
.send_IPI_mask = numaq_send_IPI_mask,
.send_IPI_mask_allbutself = NULL,
.send_IPI_allbutself = numaq_send_IPI_allbutself,
.send_IPI_all = numaq_send_IPI_all,
.send_IPI_self = default_send_IPI_self,
.wakeup_secondary_cpu = wakeup_secondary_cpu_via_nmi,
.trampoline_phys_low = NUMAQ_TRAMPOLINE_PHYS_LOW,
.trampoline_phys_high = NUMAQ_TRAMPOLINE_PHYS_HIGH,
/* We don't do anything here because we use NMI's to boot instead */
.wait_for_init_deassert = NULL,
.smp_callin_clear_local_apic = numaq_smp_callin_clear_local_apic,
.inquire_remote_apic = NULL,
.read = native_apic_mem_read,
.write = native_apic_mem_write,
.eoi_write = native_apic_mem_write,
.icr_read = native_apic_icr_read,
.icr_write = native_apic_icr_write,
.wait_icr_idle = native_apic_wait_icr_idle,
.safe_wait_icr_idle = native_safe_apic_wait_icr_idle,
.x86_32_early_logical_apicid = noop_x86_32_early_logical_apicid,
.x86_32_numa_cpu_node = numaq_numa_cpu_node,
};
apic_driver(apic_numaq);