[SPARC]: Try to start getting SMP back into shape.
Todo items:
- IRQ_INPROGRESS flag - use sparc64 irq buckets, or generic irq_desc?
- sun4d
- re-indent large chunks of sun4m_smp.c
- some places assume sequential cpu numbering (i.e. 0,1 instead of 0,2)
Last I checked (with 2.6.14), random programs segfault with dual
HyperSPARC. And with SuperSPARC II's, it seems stable but will
eventually die from a write lock error (wrong lock owner or something).
I haven't tried the HyperSPARC + highmem combination recently, so that
may still be a problem.
Signed-off-by: David S. Miller <davem@davemloft.net>
diff --git a/arch/sparc/kernel/sun4m_smp.c b/arch/sparc/kernel/sun4m_smp.c
index 1dde312..70b375a 100644
--- a/arch/sparc/kernel/sun4m_smp.c
+++ b/arch/sparc/kernel/sun4m_smp.c
@@ -40,15 +40,11 @@
extern void calibrate_delay(void);
extern volatile int smp_processors_ready;
-extern int smp_num_cpus;
extern volatile unsigned long cpu_callin_map[NR_CPUS];
extern unsigned char boot_cpu_id;
-extern int smp_activated;
-extern volatile int __cpu_number_map[NR_CPUS];
-extern volatile int __cpu_logical_map[NR_CPUS];
-extern volatile unsigned long ipi_count;
-extern volatile int smp_process_available;
-extern volatile int smp_commenced;
+
+extern cpumask_t smp_commenced_mask;
+
extern int __smp4m_processor_id(void);
/*#define SMP_DEBUG*/
@@ -77,8 +73,6 @@
local_flush_cache_all();
local_flush_tlb_all();
- set_irq_udt(boot_cpu_id);
-
/* Get our local ticker going. */
smp_setup_percpu_timer();
@@ -95,8 +89,9 @@
* to call the scheduler code.
*/
/* Allow master to continue. */
- swap((unsigned long *)&cpu_callin_map[cpuid], 1);
+ swap(&cpu_callin_map[cpuid], 1);
+ /* XXX: What's up with all the flushes? */
local_flush_cache_all();
local_flush_tlb_all();
@@ -111,13 +106,14 @@
atomic_inc(&init_mm.mm_count);
current->active_mm = &init_mm;
- while(!smp_commenced)
- barrier();
-
- local_flush_cache_all();
- local_flush_tlb_all();
+ while (!cpu_isset(cpuid, smp_commenced_mask))
+ mb();
local_irq_enable();
+
+ cpu_set(cpuid, cpu_online_map);
+ /* last one in gets all the interrupts (for testing) */
+ set_irq_udt(boot_cpu_id);
}
extern void init_IRQ(void);
@@ -134,102 +130,76 @@
void __init smp4m_boot_cpus(void)
{
- int cpucount = 0;
- int i, mid;
-
- printk("Entering SMP Mode...\n");
-
- local_irq_enable();
- cpus_clear(cpu_present_map);
-
- for (i = 0; !cpu_find_by_instance(i, NULL, &mid); i++)
- cpu_set(mid, cpu_present_map);
-
- for(i=0; i < NR_CPUS; i++) {
- __cpu_number_map[i] = -1;
- __cpu_logical_map[i] = -1;
- }
-
- __cpu_number_map[boot_cpu_id] = 0;
- __cpu_logical_map[0] = boot_cpu_id;
- current_thread_info()->cpu = boot_cpu_id;
-
- smp_store_cpu_info(boot_cpu_id);
- set_irq_udt(boot_cpu_id);
smp_setup_percpu_timer();
local_flush_cache_all();
- if(cpu_find_by_instance(1, NULL, NULL))
- return; /* Not an MP box. */
- for(i = 0; i < NR_CPUS; i++) {
- if(i == boot_cpu_id)
- continue;
+}
- if (cpu_isset(i, cpu_present_map)) {
- extern unsigned long sun4m_cpu_startup;
- unsigned long *entry = &sun4m_cpu_startup;
- struct task_struct *p;
- int timeout;
+int smp4m_boot_one_cpu(int i)
+{
+ extern unsigned long sun4m_cpu_startup;
+ unsigned long *entry = &sun4m_cpu_startup;
+ struct task_struct *p;
+ int timeout;
+ int cpu_node;
- /* Cook up an idler for this guy. */
- p = fork_idle(i);
- cpucount++;
- current_set[i] = task_thread_info(p);
- /* See trampoline.S for details... */
- entry += ((i-1) * 3);
+ cpu_find_by_mid(i, &cpu_node);
- /*
- * Initialize the contexts table
- * Since the call to prom_startcpu() trashes the structure,
- * we need to re-initialize it for each cpu
- */
- smp_penguin_ctable.which_io = 0;
- smp_penguin_ctable.phys_addr = (unsigned int) srmmu_ctx_table_phys;
- smp_penguin_ctable.reg_size = 0;
+ /* Cook up an idler for this guy. */
+ p = fork_idle(i);
+ current_set[i] = task_thread_info(p);
+ /* See trampoline.S for details... */
+ entry += ((i-1) * 3);
- /* whirrr, whirrr, whirrrrrrrrr... */
- printk("Starting CPU %d at %p\n", i, entry);
- local_flush_cache_all();
- prom_startcpu(cpu_data(i).prom_node,
- &smp_penguin_ctable, 0, (char *)entry);
+ /*
+ * Initialize the contexts table
+ * Since the call to prom_startcpu() trashes the structure,
+ * we need to re-initialize it for each cpu
+ */
+ smp_penguin_ctable.which_io = 0;
+ smp_penguin_ctable.phys_addr = (unsigned int) srmmu_ctx_table_phys;
+ smp_penguin_ctable.reg_size = 0;
- /* wheee... it's going... */
- for(timeout = 0; timeout < 10000; timeout++) {
- if(cpu_callin_map[i])
- break;
- udelay(200);
- }
- if(cpu_callin_map[i]) {
- /* Another "Red Snapper". */
- __cpu_number_map[i] = i;
- __cpu_logical_map[i] = i;
- } else {
- cpucount--;
- printk("Processor %d is stuck.\n", i);
- }
- }
- if(!(cpu_callin_map[i])) {
- cpu_clear(i, cpu_present_map);
- __cpu_number_map[i] = -1;
- }
- }
+ /* whirrr, whirrr, whirrrrrrrrr... */
+ printk("Starting CPU %d at %p\n", i, entry);
local_flush_cache_all();
- if(cpucount == 0) {
- printk("Error: only one Processor found.\n");
- cpu_present_map = cpumask_of_cpu(smp_processor_id());
- } else {
- unsigned long bogosum = 0;
- for_each_present_cpu(i)
- bogosum += cpu_data(i).udelay_val;
- printk("Total of %d Processors activated (%lu.%02lu BogoMIPS).\n",
- cpucount + 1,
- bogosum/(500000/HZ),
- (bogosum/(5000/HZ))%100);
- smp_activated = 1;
- smp_num_cpus = cpucount + 1;
+ prom_startcpu(cpu_node,
+ &smp_penguin_ctable, 0, (char *)entry);
+
+ /* wheee... it's going... */
+ for(timeout = 0; timeout < 10000; timeout++) {
+ if(cpu_callin_map[i])
+ break;
+ udelay(200);
}
+ if (!(cpu_callin_map[i])) {
+ printk("Processor %d is stuck.\n", i);
+ return -ENODEV;
+ }
+
+ local_flush_cache_all();
+ return 0;
+}
+
+void __init smp4m_smp_done(void)
+{
+ int i, first;
+ int *prev;
+
+ /* setup cpu list for irq rotation */
+ first = 0;
+ prev = &first;
+ for (i = 0; i < NR_CPUS; i++) {
+ if (cpu_online(i)) {
+ *prev = i;
+ prev = &cpu_data(i).next;
+ }
+ }
+ *prev = first;
+ local_flush_cache_all();
+
/* Free unneeded trap tables */
- if (!cpu_isset(i, cpu_present_map)) {
+ if (!cpu_isset(1, cpu_present_map)) {
ClearPageReserved(virt_to_page(trapbase_cpu1));
init_page_count(virt_to_page(trapbase_cpu1));
free_page((unsigned long)trapbase_cpu1);
@@ -263,6 +233,9 @@
*/
void smp4m_irq_rotate(int cpu)
{
+ int next = cpu_data(cpu).next;
+ if (next != cpu)
+ set_irq_udt(next);
}
/* Cross calls, in order to work efficiently and atomically do all
@@ -289,7 +262,7 @@
smp_cpu_in_msg[me]++;
if(target == MSG_ALL_BUT_SELF || target == MSG_ALL) {
- mask = cpu_present_map;
+ mask = cpu_online_map;
if(target == MSG_ALL_BUT_SELF)
cpu_clear(me, mask);
for(i = 0; i < 4; i++) {
@@ -314,8 +287,8 @@
unsigned long arg3;
unsigned long arg4;
unsigned long arg5;
- unsigned long processors_in[NR_CPUS]; /* Set when ipi entered. */
- unsigned long processors_out[NR_CPUS]; /* Set when ipi exited. */
+ unsigned long processors_in[SUN4M_NCPUS]; /* Set when ipi entered. */
+ unsigned long processors_out[SUN4M_NCPUS]; /* Set when ipi exited. */
} ccall_info;
static DEFINE_SPINLOCK(cross_call_lock);
@@ -324,8 +297,7 @@
void smp4m_cross_call(smpfunc_t func, unsigned long arg1, unsigned long arg2,
unsigned long arg3, unsigned long arg4, unsigned long arg5)
{
- if(smp_processors_ready) {
- register int ncpus = smp_num_cpus;
+ register int ncpus = SUN4M_NCPUS;
unsigned long flags;
spin_lock_irqsave(&cross_call_lock, flags);
@@ -340,7 +312,7 @@
/* Init receive/complete mapping, plus fire the IPI's off. */
{
- cpumask_t mask = cpu_present_map;
+ cpumask_t mask = cpu_online_map;
register int i;
cpu_clear(smp_processor_id(), mask);
@@ -373,7 +345,6 @@
}
spin_unlock_irqrestore(&cross_call_lock, flags);
- }
}
/* Running cross calls. */
