Forthcoming IBM boxes will be using Nocona and/or Opteron chips in
clustered mode to get beyond 8 CPUs. In fact, there are plans to try
for 128 CPUs when the Tulsa chip comes out. Thus, there are a fair
number of signed vs. unsigned changes in the patch.
Thanks to the HPET timer and some HW changes, I've been able to remove
the MPS/ACPI string comparisons from the detection code. Instead, it
scans bios_cpu_apicid and uses simple heuristics to select the correct
IRQ delivery mode. No need for a config option. Hurrah!
Likewise, I've been able to avoid the preprocessor tricks that the i386
sub-arch needed to build with one or more sub-arches.
This patch has already been run past Andi Kleen and should be coming to
a source tree near you soon.
diff -pruN 2.6.8.1-mm4/arch/i386/kernel/acpi/boot.c z8.1m4/arch/i386/
kernel/acpi/boot.c
--- 2.6.8.1-mm4/arch/i386/kernel/acpi/boot.c 2004-08-25
14:48:19.000000000 -0700
+++ z8.1m4/arch/i386/kernel/acpi/boot.c 2004-08-25 14:49:56.000000000
-0700
@@ -40,7 +40,7 @@
#ifdef CONFIG_X86_64
static inline void acpi_madt_oem_check(char *oem_id, char
*oem_table_id) { }
-static inline void clustered_apic_check(void) { }
+extern void __init clustered_apic_check(void);
static inline int ioapic_setup_disabled(void) { return 0; }
#include <asm/proto.h>
diff -pruN 2.6.8.1-mm4/arch/x86_64/kernel/Makefile z8.1m4/arch/x86_64/
kernel/Makefile
--- 2.6.8.1-mm4/arch/x86_64/kernel/Makefile 2004-08-25
14:48:21.000000000 -0700
+++ z8.1m4/arch/x86_64/kernel/Makefile 2004-08-25 15:00:59.000000000
-0700
@@ -17,7 +17,8 @@ obj-$(CONFIG_MICROCODE) += microcode.o
obj-$(CONFIG_X86_CPUID) += cpuid.o
obj-$(CONFIG_SMP) += smp.o smpboot.o trampoline.o
obj-$(CONFIG_X86_LOCAL_APIC) += apic.o nmi.o
-obj-$(CONFIG_X86_IO_APIC) += io_apic.o mpparse.o
+obj-$(CONFIG_X86_IO_APIC) += io_apic.o mpparse.o \
+ genapic.o genapic_cluster.o genapic_flat.o
obj-$(CONFIG_KEXEC) += machine_kexec.o relocate_kernel.o
obj-$(CONFIG_PM) += suspend.o
obj-$(CONFIG_SOFTWARE_SUSPEND) += suspend_asm.o
diff -pruN 2.6.8.1-mm4/arch/x86_64/kernel/apic.c z8.1m4/arch/x86_64/
kernel/apic.c
--- 2.6.8.1-mm4/arch/x86_64/kernel/apic.c 2004-08-25 14:48:21.000000000
-0700
+++ z8.1m4/arch/x86_64/kernel/apic.c 2004-08-25 14:49:56.000000000 -0700
@@ -32,6 +32,7 @@
#include <asm/mtrr.h>
#include <asm/mpspec.h>
#include <asm/pgalloc.h>
+#include <asm/mach_apic.h>
int disable_apic_timer __initdata;
@@ -328,8 +329,7 @@ void __init setup_local_APIC (void)
* Double-check whether this APIC is really registered.
* This is meaningless in clustered apic mode, so we skip it.
*/
- if (!clustered_apic_mode &&
- !physid_isset(GET_APIC_ID(apic_read(APIC_ID)), phys_cpu_present_map))
+ if (!apic_id_registered())
BUG();
/*
@@ -337,23 +337,7 @@ void __init setup_local_APIC (void)
* an APIC. See e.g. "AP-388 82489DX User's Manual" (Intel
* document number 292116). So here it goes...
*/
-
- if (!clustered_apic_mode) {
- /*
- * In clustered apic mode, the firmware does this for us
- * Put the APIC into flat delivery mode.
- * Must be "all ones" explicitly for 82489DX.
- */
- apic_write_around(APIC_DFR, 0xffffffff);
-
- /*
- * Set up the logical destination ID.
- */
- value = apic_read(APIC_LDR);
- value &= ~APIC_LDR_MASK;
- value |= (1<<(smp_processor_id()+24));
- apic_write_around(APIC_LDR, value);
- }
+ init_apic_ldr();
/*
* Set Task Priority to 'accept all'. We never change this
diff -pruN 2.6.8.1-mm4/arch/x86_64/kernel/genapic.c z8.1m4/arch/x86_64/
kernel/genapic.c
--- 2.6.8.1-mm4/arch/x86_64/kernel/genapic.c 1969-12-31
16:00:00.000000000 -0800
+++ z8.1m4/arch/x86_64/kernel/genapic.c 2004-08-25 15:17:01.000000000
-0700
@@ -0,0 +1,80 @@
+/*
+ * Copyright 2004 James Cleverdon, IBM.
+ * Subject to the GNU Public License, v.2
+ *
+ * Generic APIC sub-arch probe layer.
+ *
+ * Hacked for x86-64 by James Cleverdon from i386 architecture code by
+ * Martin Bligh, Andi Kleen, James Bottomley, John Stultz, and
+ * James Cleverdon.
+ */
+#include <linux/config.h>
+#include <linux/threads.h>
+#include <linux/cpumask.h>
+#include <linux/string.h>
+#include <linux/kernel.h>
+#include <linux/ctype.h>
+#include <linux/init.h>
+#include <asm/smp.h>
+#include <asm/ipi.h>
+
+/* which logical CPU number maps to which CPU (physical APIC ID) */
+u8 x86_cpu_to_apicid[NR_CPUS] = { [0 ... NR_CPUS-1] = BAD_APICID };
+EXPORT_SYMBOL(x86_cpu_to_apicid);
+u8 x86_cpu_to_log_apicid[NR_CPUS] = { [0 ... NR_CPUS-1] = BAD_APICID };
+
+extern struct genapic apic_cluster;
+extern struct genapic apic_flat;
+
+struct genapic *genapic;
+
+
+/*
+ * Check the APIC IDs in bios_cpu_apicid and choose the APIC mode.
+ */
+void __init clustered_apic_check(void)
+{
+ long i;
+ u8 clusters, max_cluster;
+ u8 id;
+ u8 cluster_cnt[NUM_APIC_CLUSTERS];
+
+ memset(cluster_cnt, 0, sizeof(cluster_cnt));
+
+ for (i = 0; i < NR_CPUS; i++) {
+ id = bios_cpu_apicid[i];
+ if (id != BAD_APICID)
+ cluster_cnt[APIC_CLUSTERID(id)]++;
+ }
+
+ clusters = 0;
+ max_cluster = 0;
+ for (i = 0; i < NUM_APIC_CLUSTERS; i++) {
+ if (cluster_cnt[i] > 0) {
+ ++clusters;
+ if (cluster_cnt[i] > max_cluster)
+ max_cluster = cluster_cnt[i];
+ }
+ }
+
+ /*
+ * If we have clusters <= 1 and CPUs <= 8 in cluster 0, then flat
mode,
+ * else if max_cluster <= 4 and cluster_cnt[15] == 0, clustered
logical
+ * else physical mode.
+ * (We don't use lowest priority delivery + HW APIC IRQ steering, so
+ * can ignore the clustered logical case and go straight to physical.)
+ */
+ if (clusters <= 1 && max_cluster <= 8 && cluster_cnt[0] ==
max_cluster)
+ genapic = &apic_flat;
+ else
+ genapic = &apic_cluster;
+
+ printk(KERN_INFO "Setting APIC routing to %s\n", genapic->name);
+}
+
+/* Same for both flat and clustered. */
+
+void send_IPI_self(int vector)
+{
+ __send_IPI_shortcut(APIC_DEST_SELF, vector, APIC_DEST_PHYSICAL);
+}
diff -pruN 2.6.8.1-mm4/arch/x86_64/kernel/genapic_cluster.c z8.1m4/arch/
x86_64/kernel/genapic_cluster.c
--- 2.6.8.1-mm4/arch/x86_64/kernel/genapic_cluster.c 1969-12-31
16:00:00.000000000 -0800
+++ z8.1m4/arch/x86_64/kernel/genapic_cluster.c 2004-08-25
15:17:07.000000000 -0700
@@ -0,0 +1,146 @@
+/*
+ * Copyright 2004 James Cleverdon, IBM.
+ * Subject to the GNU Public License, v.2
+ *
+ * Clustered APIC subarch code. Up to 255 CPUs, physical delivery.
+ * (A more realistic maximum is around 230 CPUs.)
+ *
+ * Hacked for x86-64 by James Cleverdon from i386 architecture code by
+ * Martin Bligh, Andi Kleen, James Bottomley, John Stultz, and
+ * James Cleverdon.
+ */
+#include <linux/config.h>
+#include <linux/threads.h>
+#include <linux/cpumask.h>
+#include <linux/string.h>
+#include <linux/kernel.h>
+#include <linux/ctype.h>
+#include <linux/init.h>
+#include <asm/smp.h>
+#include <asm/ipi.h>
+
+
+/*
+ * Set up the logical destination ID.
+ *
+ * Intel recommends to set DFR, LDR and TPR before enabling
+ * an APIC. See e.g. "AP-388 82489DX User's Manual" (Intel
+ * document number 292116). So here it goes...
+ */
+static void cluster_init_apic_ldr(void)
+{
+ unsigned long val, id;
+ long i, count;
+ u8 lid;
+ u8 my_id = hard_smp_processor_id();
+ u8 my_cluster = APIC_CLUSTER(my_id);
+
+ /* Create logical APIC IDs by counting CPUs already in cluster. */
+ for (count = 0, i = NR_CPUS; --i >= 0; ) {
+ lid = x86_cpu_to_log_apicid[i];
+ if (lid != BAD_APICID && APIC_CLUSTER(lid) == my_cluster)
+ ++count;
+ }
+ /*
+ * We only have a 4 wide bitmap in cluster mode. There's no way
+ * to get above 60 CPUs and still give each one it's own bit.
+ * But, we're using physical IRQ delivery, so we don't care.
+ * Use bit 3 for the 4th through Nth CPU in each cluster.
+ */
+ if (count >= XAPIC_DEST_CPUS_SHIFT)
+ count = 3;
+ id = my_cluster | (1UL << count);
+ x86_cpu_to_log_apicid[smp_processor_id()] = id;
+ apic_write_around(APIC_DFR, APIC_DFR_CLUSTER);
+ val = apic_read(APIC_LDR) & ~APIC_LDR_MASK;
+ val |= SET_APIC_LOGICAL_ID(id);
+ apic_write_around(APIC_LDR, val);
+}
+
+
+/* Mapping from cpu number to logical apicid */
+static int cluster_cpu_to_logical_apicid(int cpu)
+{
+ if ((unsigned)cpu >= NR_CPUS)
+ return BAD_APICID;
+ return x86_cpu_to_log_apicid[cpu];
+}
+
+static int cluster_cpu_present_to_apicid(int mps_cpu)
+{
+ if ((unsigned)mps_cpu < NR_CPUS)
+ return (int)bios_cpu_apicid[mps_cpu];
+ else
+ return BAD_APICID;
+}
+
+/* Distribute IRQ load with round-robin allocation */
+
+static u8 cluster_target_cpus(void)
+{
+ unsigned long i;
+ static unsigned long last_cpu = 0;
+
+ i = last_cpu;
+ do {
+ if (++i >= NR_CPUS)
+ i = 0;
+ } while (x86_cpu_to_apicid[i] == BAD_APICID);
+ last_cpu = i;
+
+ return x86_cpu_to_apicid[i];
+}
+
+static void cluster_send_IPI_mask(cpumask_t mask, int vector)
+{
+ send_IPI_mask_sequence(mask, vector);
+}
+
+static void cluster_send_IPI_allbutself(int vector)
+{
+ cpumask_t mask = cpu_online_map;
+ cpu_clear(smp_processor_id(), mask);
+
+ if (!cpus_empty(mask))
+ cluster_send_IPI_mask(mask, vector);
+}
+
+static void cluster_send_IPI_all(int vector)
+{
+ cluster_send_IPI_mask(cpu_online_map, vector);
+}
+
+static int cluster_apic_id_registered(void)
+{
+ return 1;
+}
+
+static unsigned int cluster_cpu_mask_to_apicid(cpumask_t cpumask)
+{
+ long cpu;
+
+ /*
+ * We're using fixed IRQ delivery, can only return one phys APIC ID.
+ * May as well be the first.
+ */
+ cpu = first_cpu(cpumask);
+ if (cpu >= 0 && cpu < NR_CPUS)
+ return x86_cpu_to_apicid[cpu];
+ else
+ return BAD_APICID;
+}
+
+
+struct genapic apic_cluster = {
+ .name = "clustered",
+ .int_delivery_mode = dest_Fixed,
+ .int_dest_mode = (APIC_DEST_PHYSICAL != 0),
+ .int_delivery_dest = APIC_DEST_PHYSICAL | APIC_DM_FIXED,
+ .target_cpus = cluster_target_cpus,
+ .apic_id_registered = cluster_apic_id_registered,
+ .init_apic_ldr = cluster_init_apic_ldr,
+ .send_IPI_all = cluster_send_IPI_all,
+ .send_IPI_allbutself = cluster_send_IPI_allbutself,
+ .send_IPI_mask = cluster_send_IPI_mask,
+ .cpu_mask_to_apicid = cluster_cpu_mask_to_apicid,
+};
diff -pruN 2.6.8.1-mm4/arch/x86_64/kernel/genapic_flat.c z8.1m4/arch/
x86_64/kernel/genapic_flat.c
--- 2.6.8.1-mm4/arch/x86_64/kernel/genapic_flat.c 1969-12-31
16:00:00.000000000 -0800
+++ z8.1m4/arch/x86_64/kernel/genapic_flat.c 2004-08-25
15:17:11.000000000 -0700
@@ -0,0 +1,119 @@
+/*
+ * Copyright 2004 James Cleverdon, IBM.
+ * Subject to the GNU Public License, v.2
+ *
+ * Flat APIC subarch code. Maximum 8 CPUs, logical delivery.
+ *
+ * Hacked for x86-64 by James Cleverdon from i386 architecture code by
+ * Martin Bligh, Andi Kleen, James Bottomley, John Stultz, and
+ * James Cleverdon.
+ */
+#include <linux/config.h>
+#include <linux/threads.h>
+#include <linux/cpumask.h>
+#include <linux/string.h>
+#include <linux/kernel.h>
+#include <linux/ctype.h>
+#include <linux/init.h>
+#include <asm/smp.h>
+#include <asm/ipi.h>
+
+
+static u8 flat_target_cpus(void)
+{
+ return cpus_addr(cpu_online_map)[0];
+}
+
+/*
+ * Set up the logical destination ID.
+ *
+ * Intel recommends to set DFR, LDR and TPR before enabling
+ * an APIC. See e.g. "AP-388 82489DX User's Manual" (Intel
+ * document number 292116). So here it goes...
+ */
+static void flat_init_apic_ldr(void)
+{
+ unsigned long val;
+ unsigned long num, id;
+
+ num = smp_processor_id();
+ id = 1UL << num;
+ x86_cpu_to_log_apicid[num] = id;
+ apic_write_around(APIC_DFR, APIC_DFR_FLAT);
+ val = apic_read(APIC_LDR) & ~APIC_LDR_MASK;
+ val |= SET_APIC_LOGICAL_ID(id);
+ apic_write_around(APIC_LDR, val);
+}
+
+static void flat_send_IPI_allbutself(int vector)
+{
+ /*
+ * if there are no other CPUs in the system then
+ * we get an APIC send error if we try to broadcast.
+ * thus we have to avoid sending IPIs in this case.
+ */
+ if (num_online_cpus() > 1)
+ __send_IPI_shortcut(APIC_DEST_ALLBUT, vector, APIC_DEST_LOGICAL);
+}
+
+static void flat_send_IPI_all(int vector)
+{
+ __send_IPI_shortcut(APIC_DEST_ALLINC, vector, APIC_DEST_LOGICAL);
+}
+
+static void flat_send_IPI_mask(cpumask_t cpumask, int vector)
+{
+ unsigned long mask = cpus_addr(cpumask)[0];
+ unsigned long cfg;
+ unsigned long flags;
+
+ local_save_flags(flags);
+ local_irq_disable();
+
+ /*
+ * Wait for idle.
+ */
+ apic_wait_icr_idle();
+
+ /*
+ * prepare target chip field
+ */
+ cfg = __prepare_ICR2(mask);
+ apic_write_around(APIC_ICR2, cfg);
+
+ /*
+ * program the ICR
+ */
+ cfg = __prepare_ICR(0, vector, APIC_DEST_LOGICAL);
+
+ /*
+ * Send the IPI. The write to APIC_ICR fires this off.
+ */
+ apic_write_around(APIC_ICR, cfg);
+ local_irq_restore(flags);
+}
+
+static int flat_apic_id_registered(void)
+{
+ return physid_isset(GET_APIC_ID(apic_read(APIC_ID)),
phys_cpu_present_map);
+}
+
+static unsigned int flat_cpu_mask_to_apicid(cpumask_t cpumask)
+{
+ return cpus_addr(cpumask)[0] & APIC_ALL_CPUS;
+}
+
+
+struct genapic apic_flat = {
+ .name = "flat",
+ .int_delivery_mode = dest_LowestPrio,
+ .int_dest_mode = (APIC_DEST_LOGICAL != 0),
+ .int_delivery_dest = APIC_DEST_LOGICAL | APIC_DM_LOWEST,
+ .target_cpus = flat_target_cpus,
+ .apic_id_registered = flat_apic_id_registered,
+ .init_apic_ldr = flat_init_apic_ldr,
+ .send_IPI_all = flat_send_IPI_all,
+ .send_IPI_allbutself = flat_send_IPI_allbutself,
+ .send_IPI_mask = flat_send_IPI_mask,
+ .cpu_mask_to_apicid = flat_cpu_mask_to_apicid,
+};
diff -pruN 2.6.8.1-mm4/arch/x86_64/kernel/io_apic.c z8.1m4/arch/x86_64/
kernel/io_apic.c
--- 2.6.8.1-mm4/arch/x86_64/kernel/io_apic.c 2004-08-25
14:48:21.000000000 -0700
+++ z8.1m4/arch/x86_64/kernel/io_apic.c 2004-08-25 14:49:56.000000000
-0700
@@ -35,6 +35,7 @@
#include <asm/smp.h>
#include <asm/desc.h>
#include <asm/proto.h>
+#include <asm/mach_apic.h>
int sis_apic_bug; /* not actually supported, dummy for compile */
@@ -729,8 +730,8 @@ void __init setup_IO_APIC_irqs(void)
*/
memset(&entry,0,sizeof(entry));
- entry.delivery_mode = dest_LowestPrio;
- entry.dest_mode = INT_DELIVERY_MODE;
+ entry.delivery_mode = INT_DELIVERY_MODE;
+ entry.dest_mode = INT_DEST_MODE;
entry.mask = 0; /* enable IRQ */
entry.dest.logical.logical_dest = TARGET_CPUS;
@@ -798,10 +799,10 @@ void __init setup_ExtINT_IRQ0_pin(unsign
* We use logical delivery to get the timer IRQ
* to the first CPU.
*/
- entry.dest_mode = INT_DELIVERY_MODE;
+ entry.dest_mode = INT_DEST_MODE;
entry.mask = 0; /* unmask IRQ now */
entry.dest.logical.logical_dest = TARGET_CPUS;
- entry.delivery_mode = dest_LowestPrio;
+ entry.delivery_mode = INT_DELIVERY_MODE;
entry.polarity = 0;
entry.trigger = 0;
entry.vector = vector;
@@ -1173,7 +1174,6 @@ void disable_IO_APIC(void)
static void __init setup_ioapic_ids_from_mpc (void)
{
union IO_APIC_reg_00 reg_00;
- physid_mask_t phys_id_present_map = phys_cpu_present_map;
int apic;
int i;
unsigned char old_id;
@@ -1199,28 +1199,7 @@ static void __init setup_ioapic_ids_from
mp_ioapics[apic].mpc_apicid = reg_00.bits.ID;
}
- /*
- * Sanity check, is the ID really free? Every APIC in a
- * system must have a unique ID or we get lots of nice
- * 'stuck on smp_invalidate_needed IPI wait' messages.
- */
- if (physid_isset(mp_ioapics[apic].mpc_apicid, phys_id_present_map)) {
- printk(KERN_ERR "BIOS bug, IO-APIC#%d ID %d is already used!...\n",
- apic, mp_ioapics[apic].mpc_apicid);
- for (i = 0; i < 0xf; i++)
- if (!physid_isset(i, phys_id_present_map))
- break;
- if (i >= 0xf)
- panic("Max APIC ID exceeded!\n");
- printk(KERN_ERR "... fixing up to %d. (tell your hw vendor)\n",
- i);
- physid_set(i, phys_id_present_map);
- mp_ioapics[apic].mpc_apicid = i;
- } else {
- printk(KERN_INFO
- "Using IO-APIC %d\n", mp_ioapics[apic].mpc_apicid);
- physid_set(mp_ioapics[apic].mpc_apicid, phys_id_present_map);
- }
+ printk(KERN_INFO "Using IO-APIC %d\n", mp_ioapics[apic].mpc_apicid);
/*
@@ -1429,9 +1408,9 @@ static void set_ioapic_affinity_irq(unsi
dest = cpu_mask_to_apicid(mask);
/*
- * Only the first 8 bits are valid.
+ * Only the high 8 bits are valid.
*/
- dest = dest << 24;
+ dest = SET_APIC_LOGICAL_ID(dest);
spin_lock_irqsave(&ioapic_lock, flags);
__DO_ACTION(1, = dest, )
@@ -1821,7 +1800,7 @@ void __init setup_IO_APIC(void)
#ifdef CONFIG_ACPI_BOOT
-#define IO_APIC_MAX_ID 15
+#define IO_APIC_MAX_ID 0xFE
int __init io_apic_get_unique_id (int ioapic, int apic_id)
{
@@ -1938,8 +1917,8 @@ int io_apic_set_pci_routing (int ioapic,
memset(&entry,0,sizeof(entry));
- entry.delivery_mode = dest_LowestPrio;
- entry.dest_mode = INT_DELIVERY_MODE;
+ entry.delivery_mode = INT_DELIVERY_MODE;
+ entry.dest_mode = INT_DEST_MODE;
entry.dest.logical.logical_dest = TARGET_CPUS;
entry.trigger = edge_level;
entry.polarity = active_high_low;
@@ -1981,20 +1960,3 @@ int io_apic_set_pci_routing (int ioapic,
#endif /*CONFIG_ACPI_BOOT*/
-#ifndef CONFIG_SMP
-void send_IPI_self(int vector)
-{
- unsigned int cfg;
-
- /*
- * Wait for idle.
- */
- apic_wait_icr_idle();
- cfg = APIC_DM_FIXED | APIC_DEST_SELF | vector | APIC_DEST_LOGICAL;
-
- /*
- * Send the IPI. The write to APIC_ICR fires this off.
- */
- apic_write_around(APIC_ICR, cfg);
-}
-#endif
diff -pruN 2.6.8.1-mm4/arch/x86_64/kernel/smp.c z8.1m4/arch/x86_64/
kernel/smp.c
--- 2.6.8.1-mm4/arch/x86_64/kernel/smp.c 2004-08-25 14:48:21.000000000
-0700
+++ z8.1m4/arch/x86_64/kernel/smp.c 2004-08-25 14:49:56.000000000 -0700
@@ -24,105 +24,7 @@
#include <asm/mtrr.h>
#include <asm/pgalloc.h>
#include <asm/tlbflush.h>
-
-/*
- * the following functions deal with sending IPIs between CPUs.
- *
- * We use 'broadcast', CPU->CPU IPIs and self-IPIs too.
- */
-
-static inline unsigned int __prepare_ICR (unsigned int shortcut, int
vector)
-{
- unsigned int icr = APIC_DM_FIXED | shortcut | vector |
APIC_DEST_LOGICAL;
- if (vector == KDB_VECTOR)
- icr = (icr & (~APIC_VECTOR_MASK)) | APIC_DM_NMI;
- return icr;
-}
-
-static inline int __prepare_ICR2 (unsigned int mask)
-{
- return SET_APIC_DEST_FIELD(mask);
-}
-
-static inline void __send_IPI_shortcut(unsigned int shortcut, int
vector)
-{
- /*
- * Subtle. In the case of the 'never do double writes' workaround
- * we have to lock out interrupts to be safe. As we don't care
- * of the value read we use an atomic rmw access to avoid costly
- * cli/sti. Otherwise we use an even cheaper single atomic write
- * to the APIC.
- */
- unsigned int cfg;
-
- /*
- * Wait for idle.
- */
- apic_wait_icr_idle();
-
- /*
- * No need to touch the target chip field
- */
- cfg = __prepare_ICR(shortcut, vector);
-
- /*
- * Send the IPI. The write to APIC_ICR fires this off.
- */
- apic_write_around(APIC_ICR, cfg);
-}
-
-static inline void send_IPI_allbutself(int vector)
-{
- /*
- * if there are no other CPUs in the system then
- * we get an APIC send error if we try to broadcast.
- * thus we have to avoid sending IPIs in this case.
- */
- if (num_online_cpus() > 1)
- __send_IPI_shortcut(APIC_DEST_ALLBUT, vector);
-}
-
-static inline void send_IPI_all(int vector)
-{
- __send_IPI_shortcut(APIC_DEST_ALLINC, vector);
-}
-
-void send_IPI_self(int vector)
-{
- __send_IPI_shortcut(APIC_DEST_SELF, vector);
-}
-
-static inline void send_IPI_mask(cpumask_t cpumask, int vector)
-{
- unsigned long mask = cpus_addr(cpumask)[0];
- unsigned long cfg;
- unsigned long flags;
-
- local_save_flags(flags);
- local_irq_disable();
-
- /*
- * Wait for idle.
- */
- apic_wait_icr_idle();
-
- /*
- * prepare target chip field
- */
- cfg = __prepare_ICR2(mask);
- apic_write_around(APIC_ICR2, cfg);
-
- /*
- * program the ICR
- */
- cfg = __prepare_ICR(0, vector);
-
- /*
- * Send the IPI. The write to APIC_ICR fires this off.
- */
- apic_write_around(APIC_ICR, cfg);
- local_irq_restore(flags);
-}
+#include <asm/mach_apic.h>
/*
* Smarter SMP flushing macros.
diff -pruN 2.6.8.1-mm4/arch/x86_64/kernel/smpboot.c z8.1m4/arch/x86_64/
kernel/smpboot.c
--- 2.6.8.1-mm4/arch/x86_64/kernel/smpboot.c 2004-08-25
14:48:21.000000000 -0700
+++ z8.1m4/arch/x86_64/kernel/smpboot.c 2004-08-25 14:58:01.000000000
-0700
@@ -56,15 +56,12 @@
/* Number of siblings per CPU package */
int smp_num_siblings = 1;
-char phys_proc_id[NR_CPUS]; /* Package ID of each logical CPU */
+/* Package ID of each logical CPU */
+u8 phys_proc_id[NR_CPUS] = { [0 ... NR_CPUS-1] = BAD_APICID };
/* Bitmask of currently online CPUs */
cpumask_t cpu_online_map;
-/* which logical CPU number maps to which CPU (physical APIC ID) */
-volatile char x86_cpu_to_apicid[NR_CPUS];
-EXPORT_SYMBOL(x86_cpu_to_apicid);
-
static cpumask_t cpu_callin_map;
cpumask_t cpu_callout_map;
static cpumask_t smp_commenced_mask;
@@ -658,6 +655,8 @@ static void __init do_boot_cpu (int apic
cpu_clear(cpu, cpu_callout_map); /* was set here (do_boot_cpu()) */
clear_bit(cpu, &cpu_initialized); /* was set by cpu_init() */
cpucount--;
+ x86_cpu_to_apicid[cpu] = BAD_APICID;
+ x86_cpu_to_log_apicid[cpu] = BAD_APICID;
}
}
diff -pruN 2.6.8.1-mm4/include/asm-x86_64/apic.h z8.1m4/include/
asm-x86_64/apic.h
--- 2.6.8.1-mm4/include/asm-x86_64/apic.h 2004-08-14 03:56:01.000000000
-0700
+++ z8.1m4/include/asm-x86_64/apic.h 2004-08-25 14:49:56.000000000 -0700
@@ -97,7 +97,6 @@ extern unsigned int nmi_watchdog;
#endif /* CONFIG_X86_LOCAL_APIC */
-#define clustered_apic_mode 0
#define esr_disable 0
extern unsigned boot_cpu_id;
diff -pruN 2.6.8.1-mm4/include/asm-x86_64/apicdef.h z8.1m4/include/
asm-x86_64/apicdef.h
--- 2.6.8.1-mm4/include/asm-x86_64/apicdef.h 2004-08-25
14:48:28.000000000 -0700
+++ z8.1m4/include/asm-x86_64/apicdef.h 2004-08-25 14:52:52.000000000
-0700
@@ -11,26 +11,26 @@
#define APIC_DEFAULT_PHYS_BASE 0xfee00000
#define APIC_ID 0x20
-#define APIC_ID_MASK (0x0F<<24)
-#define GET_APIC_ID(x) (((x)>>24)&0x0F)
+#define APIC_ID_MASK (0xFFu<<24)
+#define GET_APIC_ID(x) (((x)>>24)&0xFFu)
#define APIC_LVR 0x30
#define APIC_LVR_MASK 0xFF00FF
-#define GET_APIC_VERSION(x) ((x)&0xFF)
-#define GET_APIC_MAXLVT(x) (((x)>>16)&0xFF)
-#define APIC_INTEGRATED(x) ((x)&0xF0)
+#define GET_APIC_VERSION(x) ((x)&0xFFu)
+#define GET_APIC_MAXLVT(x) (((x)>>16)&0xFFu)
+#define APIC_INTEGRATED(x) ((x)&0xF0u)
#define APIC_TASKPRI 0x80
-#define APIC_TPRI_MASK 0xFF
+#define APIC_TPRI_MASK 0xFFu
#define APIC_ARBPRI 0x90
-#define APIC_ARBPRI_MASK 0xFF
+#define APIC_ARBPRI_MASK 0xFFu
#define APIC_PROCPRI 0xA0
#define APIC_EOI 0xB0
#define APIC_EIO_ACK 0x0 /* Write this to the EOI register */
#define APIC_RRR 0xC0
#define APIC_LDR 0xD0
-#define APIC_LDR_MASK (0xFF<<24)
-#define GET_APIC_LOGICAL_ID(x) (((x)>>24)&0xFF)
+#define APIC_LDR_MASK (0xFFu<<24)
+#define GET_APIC_LOGICAL_ID(x) (((x)>>24)&0xFFu)
#define SET_APIC_LOGICAL_ID(x) (((x)<<24))
-#define APIC_ALL_CPUS 0xFF
+#define APIC_ALL_CPUS 0xFFu
#define APIC_DFR 0xE0
#define APIC_DFR_CLUSTER 0x0FFFFFFFul
#define APIC_DFR_FLAT 0xFFFFFFFFul
@@ -60,6 +60,7 @@
#define APIC_INT_ASSERT 0x04000
#define APIC_ICR_BUSY 0x01000
#define APIC_DEST_LOGICAL 0x00800
+#define APIC_DEST_PHYSICAL 0x00000
#define APIC_DM_FIXED 0x00000
#define APIC_DM_LOWEST 0x00100
#define APIC_DM_SMI 0x00200
@@ -114,6 +115,18 @@
#define MAX_IO_APICS 16
/*
+ * All x86-64 systems are xAPIC compatible.
+ * In the following, "apicid" is a physical APIC ID.
+ */
+#define XAPIC_DEST_CPUS_SHIFT 4
+#define XAPIC_DEST_CPUS_MASK ((1u << XAPIC_DEST_CPUS_SHIFT) - 1)
+#define XAPIC_DEST_CLUSTER_MASK (XAPIC_DEST_CPUS_MASK <<
XAPIC_DEST_CPUS_SHIFT)
+#define APIC_CLUSTER(apicid) ((apicid) & XAPIC_DEST_CLUSTER_MASK)
+#define APIC_CLUSTERID(apicid) (APIC_CLUSTER(apicid) >>
XAPIC_DEST_CPUS_SHIFT)
+#define APIC_CPUID(apicid) ((apicid) & XAPIC_DEST_CPUS_MASK)
+#define NUM_APIC_CLUSTERS ((BAD_APICID + 1) >> XAPIC_DEST_CPUS_SHIFT)
+
+/*
* the local APIC register structure, memory mapped. Not terribly well
* tested, but we might eventually use this one in the future - the
* problem why we cannot use it right now is the P5 APIC, it has an
diff -pruN 2.6.8.1-mm4/include/asm-x86_64/genapic.h z8.1m4/include/
asm-x86_64/genapic.h
--- 2.6.8.1-mm4/include/asm-x86_64/genapic.h 1969-12-31
16:00:00.000000000 -0800
+++ z8.1m4/include/asm-x86_64/genapic.h 2004-08-25 15:17:20.000000000
-0700
@@ -0,0 +1,34 @@
+#ifndef _ASM_GENAPIC_H
+#define _ASM_GENAPIC_H 1
+
+/*
+ * Copyright 2004 James Cleverdon, IBM.
+ * Subject to the GNU Public License, v.2
+ *
+ * Generic APIC sub-arch data struct.
+ *
+ * Hacked for x86-64 by James Cleverdon from i386 architecture code by
+ * Martin Bligh, Andi Kleen, James Bottomley, John Stultz, and
+ * James Cleverdon.
+ */
+
+struct genapic {
+ char *name;
+ u32 int_delivery_mode;
+ u32 int_dest_mode;
+ u32 int_delivery_dest; /* for quick IPIs */
+ int (*apic_id_registered)(void);
+ u8 (*target_cpus)(void);
+ void (*init_apic_ldr)(void);
+ /* ipi */
+ void (*send_IPI_mask)(cpumask_t mask, int vector);
+ void (*send_IPI_allbutself)(int vector);
+ void (*send_IPI_all)(int vector);
+ /* */
+ unsigned int (*cpu_mask_to_apicid)(cpumask_t cpumask);
+};
+
+
+extern struct genapic *genapic;
+
+#endif
diff -pruN 2.6.8.1-mm4/include/asm-x86_64/ipi.h z8.1m4/include/
asm-x86_64/ipi.h
--- 2.6.8.1-mm4/include/asm-x86_64/ipi.h 1969-12-31 16:00:00.000000000
-0800
+++ z8.1m4/include/asm-x86_64/ipi.h 2004-08-25 15:19:44.000000000 -0700
@@ -0,0 +1,113 @@
+#ifndef __ASM_IPI_H
+#define __ASM_IPI_H
+
+/*
+ * Copyright 2004 James Cleverdon, IBM.
+ * Subject to the GNU Public License, v.2
+ *
+ * Generic APIC InterProcessor Interrupt code.
+ *
+ * Moved to include file by James Cleverdon from
+ * arch/x86-64/kernel/smp.c
+ *
+ * Copyrights from kernel/smp.c:
+ *
+ * (c) 1995 Alan Cox, Building #3 <[email protected]>
+ * (c) 1998-99, 2000 Ingo Molnar <[email protected]>
+ * (c) 2002,2003 Andi Kleen, SuSE Labs.
+ * Subject to the GNU Public License, v.2
+ */
+
+#include <asm/fixmap.h>
+#include <asm/hw_irq.h>
+#include <asm/apicdef.h>
+#include <asm/genapic.h>
+
+/*
+ * the following functions deal with sending IPIs between CPUs.
+ *
+ * We use 'broadcast', CPU->CPU IPIs and self-IPIs too.
+ */
+
+static inline unsigned int __prepare_ICR (unsigned int shortcut, int
vector, unsigned int dest)
+{
+ unsigned int icr = APIC_DM_FIXED | shortcut | vector | dest;
+ if (vector == KDB_VECTOR)
+ icr = (icr & (~APIC_VECTOR_MASK)) | APIC_DM_NMI;
+ return icr;
+}
+
+static inline int __prepare_ICR2 (unsigned int mask)
+{
+ return SET_APIC_DEST_FIELD(mask);
+}
+
+static inline void __send_IPI_shortcut(unsigned int shortcut, int
vector, unsigned int dest)
+{
+ /*
+ * Subtle. In the case of the 'never do double writes' workaround
+ * we have to lock out interrupts to be safe. As we don't care
+ * of the value read we use an atomic rmw access to avoid costly
+ * cli/sti. Otherwise we use an even cheaper single atomic write
+ * to the APIC.
+ */
+ unsigned int cfg;
+
+ /*
+ * Wait for idle.
+ */
+ apic_wait_icr_idle();
+
+ /*
+ * No need to touch the target chip field
+ */
+ cfg = __prepare_ICR(shortcut, vector, dest);
+
+ /*
+ * Send the IPI. The write to APIC_ICR fires this off.
+ */
+ apic_write_around(APIC_ICR, cfg);
+}
+
+
+static inline void send_IPI_mask_sequence(cpumask_t mask, int vector)
+{
+ unsigned long cfg, flags;
+ unsigned long query_cpu;
+
+ /*
+ * Hack. The clustered APIC addressing mode doesn't allow us to send
+ * to an arbitrary mask, so I do a unicast to each CPU instead.
+ * - mbligh
+ */
+ local_irq_save(flags);
+
+ for (query_cpu = 0; query_cpu < NR_CPUS; ++query_cpu) {
+ if (cpu_isset(query_cpu, mask)) {
+
+ /*
+ * Wait for idle.
+ */
+ apic_wait_icr_idle();
+
+ /*
+ * prepare target chip field
+ */
+ cfg = __prepare_ICR2(x86_cpu_to_apicid[query_cpu]);
+ apic_write_around(APIC_ICR2, cfg);
+
+ /*
+ * program the ICR
+ */
+ cfg = __prepare_ICR(0, vector, APIC_DEST_PHYSICAL);
+
+ /*
+ * Send the IPI. The write to APIC_ICR fires this off.
+ */
+ apic_write_around(APIC_ICR, cfg);
+ }
+ }
+ local_irq_restore(flags);
+}
+
+#endif /* __ASM_IPI_H */
diff -pruN 2.6.8.1-mm4/include/asm-x86_64/irq.h z8.1m4/include/
asm-x86_64/irq.h
--- 2.6.8.1-mm4/include/asm-x86_64/irq.h 2004-08-25 14:48:28.000000000
-0700
+++ z8.1m4/include/asm-x86_64/irq.h 2004-08-25 14:49:56.000000000 -0700
@@ -36,7 +36,7 @@
#define NR_IRQ_VECTORS NR_IRQS
#else
#define NR_IRQS 224
-#define NR_IRQ_VECTORS NR_IRQS
+#define NR_IRQ_VECTORS 1024
#endif
static __inline__ int irq_canonicalize(int irq)
diff -pruN 2.6.8.1-mm4/include/asm-x86_64/mach_apic.h z8.1m4/include/
asm-x86_64/mach_apic.h
--- 2.6.8.1-mm4/include/asm-x86_64/mach_apic.h 1969-12-31
16:00:00.000000000 -0800
+++ z8.1m4/include/asm-x86_64/mach_apic.h 2004-08-25 15:17:38.000000000
-0700
@@ -0,0 +1,28 @@
+#ifndef __ASM_MACH_APIC_H
+#define __ASM_MACH_APIC_H
+
+/*
+ * Copyright 2004 James Cleverdon, IBM.
+ * Subject to the GNU Public License, v.2
+ *
+ * Generic APIC sub-arch defines.
+ *
+ * Hacked for x86-64 by James Cleverdon from i386 architecture code by
+ * Martin Bligh, Andi Kleen, James Bottomley, John Stultz, and
+ * James Cleverdon.
+ */
+
+#include <asm/genapic.h>
+
+#define INT_DELIVERY_MODE (genapic->int_delivery_mode)
+#define INT_DEST_MODE (genapic->int_dest_mode)
+#define INT_DELIVERY_DEST (genapic->int_delivery_dest)
+#define TARGET_CPUS (genapic->target_cpus())
+#define apic_id_registered (genapic->apic_id_registered)
+#define init_apic_ldr (genapic->init_apic_ldr)
+#define send_IPI_mask (genapic->send_IPI_mask)
+#define send_IPI_allbutself (genapic->send_IPI_allbutself)
+#define send_IPI_all (genapic->send_IPI_all)
+#define cpu_mask_to_apicid (genapic->cpu_mask_to_apicid)
+
+#endif /* __ASM_MACH_APIC_H */
diff -pruN 2.6.8.1-mm4/include/asm-x86_64/mpspec.h z8.1m4/include/
asm-x86_64/mpspec.h
--- 2.6.8.1-mm4/include/asm-x86_64/mpspec.h 2004-08-25
14:48:28.000000000 -0700
+++ z8.1m4/include/asm-x86_64/mpspec.h 2004-08-25 14:49:56.000000000
-0700
@@ -14,9 +14,9 @@
#define SMP_MAGIC_IDENT (('_'<<24)|('P'<<16)|('M'<<8)|'_')
/*
- * a maximum of 16 APICs with the current APIC ID architecture.
+ * A maximum of 255 APICs with the current APIC ID architecture.
*/
-#define MAX_APICS 16
+#define MAX_APICS 128
struct intel_mp_floating
{
@@ -157,7 +157,11 @@ struct mpc_config_lintsrc
*/
#define MAX_IRQ_SOURCES 256
-#define MAX_MP_BUSSES 32
+/*
+ * Assume max PCI and 8 chassis x366 system (with 8 ISA): 256 + 8 =
264.
+ * Call it 270 for safety.
+ */
+#define MAX_MP_BUSSES 270
enum mp_bustype {
MP_BUS_ISA = 1,
MP_BUS_EISA,
diff -pruN 2.6.8.1-mm4/include/asm-x86_64/msi.h z8.1m4/include/
asm-x86_64/msi.h
--- 2.6.8.1-mm4/include/asm-x86_64/msi.h 2004-08-14 03:56:00.000000000
-0700
+++ z8.1m4/include/asm-x86_64/msi.h 2004-08-25 14:49:56.000000000 -0700
@@ -7,6 +7,7 @@
#define ASM_MSI_H
#include <asm/desc.h>
+#include <asm/mach_apic.h>
#define LAST_DEVICE_VECTOR 232
#define MSI_DEST_MODE MSI_LOGICAL_MODE
diff -pruN 2.6.8.1-mm4/include/asm-x86_64/smp.h z8.1m4/include/
asm-x86_64/smp.h
--- 2.6.8.1-mm4/include/asm-x86_64/smp.h 2004-08-14 03:54:46.000000000
-0700
+++ z8.1m4/include/asm-x86_64/smp.h 2004-08-25 14:49:56.000000000 -0700
@@ -48,7 +48,7 @@ extern void (*mtrr_hook) (void);
extern void zap_low_mappings(void);
void smp_stop_cpu(void);
extern cpumask_t cpu_sibling_map[NR_CPUS];
-extern char phys_proc_id[NR_CPUS];
+extern u8 phys_proc_id[NR_CPUS];
#define SMP_TRAMPOLINE_BASE 0x6000
@@ -74,14 +74,29 @@ extern __inline int hard_smp_processor_i
return GET_APIC_ID(*(unsigned int *)(APIC_BASE+APIC_ID));
}
+#define safe_smp_processor_id() (disable_apic ? 0 :
x86_apicid_to_cpu(hard_smp_processor_id()))
+
+#endif /* !ASSEMBLY */
+
+#define NO_PROC_ID 0xFF /* No processor magic marker */
+
+#endif
+
+#ifndef ASSEMBLY
/*
* Some lowlevel functions might want to know about
* the real APIC ID <-> CPU # mapping.
- * AK: why is this volatile?
*/
-extern volatile char x86_cpu_to_apicid[NR_CPUS];
+extern u8 x86_cpu_to_apicid[NR_CPUS]; /* physical ID */
+extern u8 x86_cpu_to_log_apicid[NR_CPUS];
+extern u8 bios_cpu_apicid[];
+
+static inline unsigned int cpu_mask_to_apicid(cpumask_t cpumask)
+{
+ return cpus_addr(cpumask)[0];
+}
-static inline char x86_apicid_to_cpu(char apicid)
+static inline int x86_apicid_to_cpu(u8 apicid)
{
int i;
@@ -92,10 +107,6 @@ static inline char x86_apicid_to_cpu(cha
return -1;
}
-#define safe_smp_processor_id() (disable_apic ? 0 :
x86_apicid_to_cpu(hard_smp_processor_id()))
-
-extern u8 bios_cpu_apicid[];
-
static inline int cpu_present_to_apicid(int mps_cpu)
{
if (mps_cpu < NR_CPUS)
@@ -103,20 +114,6 @@ static inline int cpu_present_to_apicid(
else
return BAD_APICID;
}
-
-#endif /* !ASSEMBLY */
-
-#define NO_PROC_ID 0xFF /* No processor magic marker */
-
-#endif
-#define INT_DELIVERY_MODE 1 /* logical delivery */
-#define TARGET_CPUS 1
-
-#ifndef ASSEMBLY
-static inline unsigned int cpu_mask_to_apicid(cpumask_t cpumask)
-{
- return cpus_addr(cpumask)[0];
-}
#endif
#ifndef CONFIG_SMP
--
James Cleverdon
IBM LTC (xSeries Linux Solutions)
{jamesclv(Unix, preferred), cleverdj(Notes)} at us dot ibm dot comm
The value that cpuinfo returns for command 1 in ebx is the
physical APIC ID latched when the system comes out of reset.
Ordinarily, this is identical to the value in the local APIC's ID
register, because nearly all BIOSes accept the HW assigned value.
Our systems, made up of individual building blocks, can't do
that. Each node boots as a separate system and is joined
together by the BIOS. Thus, the BIOS rewrites the local APIC ID
register with a new value.
Potomac and Nocona chips have a mechanism by which the BIOS
writer can change bits 7:5 to match the assigned cluster ID.
Bits 2:0 come from the thread ID. However, bits 4:3 are still
those latched at reset. Oops!
Summary: Large clustered systems can't use cpuid to derive the
sibling information.
Fix: Use the local APIC ID. That's the value we use to online
the CPUs, so it had better be OK. For non-clustered systems,
cpuid == local APIC, so nothing but large boxes should be
affected.
diff -pruN 2.6.8.1-mm4/arch/x86_64/kernel/setup.c w8.1m4/arch/x86_64/kernel/setup.c
--- 2.6.8.1-mm4/arch/x86_64/kernel/setup.c 2004-08-25 14:48:21.000000000 -0700
+++ w8.1m4/arch/x86_64/kernel/setup.c 2004-09-02 15:51:55.000000000 -0700
@@ -724,7 +724,7 @@ static void __init detect_ht(struct cpui
}
if (index_lsb != index_msb )
index_msb++;
- initial_apic_id = ebx >> 24 & 0xff;
+ initial_apic_id = hard_smp_processor_id();
phys_proc_id[cpu] = initial_apic_id >> index_msb;
printk(KERN_INFO "CPU: Physical Processor ID: %d\n",
--
James Cleverdon
IBM LTC (xSeries Linux Solutions)
{jamesclv(Unix, preferred), cleverdj(Notes)} at us dot ibm dot comm