Help with Ingo scheduler on IA64

Linux-ia64 Nick Pollitt 6 participants 34 messages Jan 12, 2002 – Jan 21, 2002

Subsystems (22)

Networking Network drivers Memory management VFS +18 more subsystems

lore

Share & tools

lore

Nick Pollitt

author

2002-01-12 02:23 UTC #0 lore raw

I'm trying to get Ingo's scheduler working on IA64 but I've hit a dead-end with the head.S code. Ingo's patch removes init_tasks, so I've modified the assembly in head.S to point at runqueues(cpu)->idle, I think - it dies very early in the boot, and I'm not familiar with ia64 assembly.

Other issues, I had to build offsets.h by hand, and I moved some stuff from sched.c to sched.h. Other than that, it's H6 + ia64.

Anyone have any feedback on getting this booting?

Thanks
Nick

Quoted 6 lines · Ingo Molnar

the -H6 patch is available:

    http://redhat.com/~mingo/O(1)-scheduler/sched-O1-2.5.2-pre11-H6.patch
    http://redhat.com/~mingo/O(1)-scheduler/sched-O1-2.4.17-H6.patch

Signature · 3 lines

Nick Pollitt                                   phone: 650.933.7406
Scalable Linux Project                           fax: 650.932.0317
Silicon Graphics, Inc.                       [email protected]

arch/i386/kernel/apic.c +2 -4

diff -X dontdiff -Nur origlinux/arch/i386/kernel/apic.c mylinux/arch/i386/kernel/apic.c
@@ -785,8 +785,7 @@
 	 */
 
 	slice = clocks / (smp_num_cpus+1);
-	printk("cpu: %d, clocks: %d, slice: %d\n",
-		smp_processor_id(), clocks, slice);
+	printk("cpu: %d, clocks: %d, slice: %d\n", smp_processor_id(), clocks, slice);
 
 	/*
 	 * Wait for IRQ0's slice:

@@ -809,8 +808,7 @@
 
 	__setup_APIC_LVTT(clocks);
 
-	printk("CPU%d<T0:%d,T1:%d,D:%d,S:%d,C:%d>\n",
-			smp_processor_id(), t0, t1, delta, slice, clocks);
+	printk("CPU%d<T0:%d,T1:%d,D:%d,S:%d,C:%d>\n", smp_processor_id(), t0, t1, delta, slice, clocks);
 
 	__restore_flags(flags);
 }

arch/i386/kernel/nmi.c +1 -1

diff -X dontdiff -Nur origlinux/arch/i386/kernel/nmi.c mylinux/arch/i386/kernel/nmi.c
@@ -283,7 +283,7 @@
 			 * to get a message out.
 			 */
 			bust_spinlocks(1);
-			printk("NMI Watchdog detected LOCKUP on CPU%d, registers:\n", cpu);
+			printk("NMI Watchdog detected LOCKUP on CPU%d, eip %08lx, registers:\n", cpu, regs->eip);
 			show_registers(regs);
 			printk("console shuts up ...\n");
 			console_silent();

arch/i386/kernel/process.c +1 -4

diff -X dontdiff -Nur origlinux/arch/i386/kernel/process.c mylinux/arch/i386/kernel/process.c
@@ -123,15 +123,12 @@
 void cpu_idle (void)
 {
 	/* endless idle loop with no priority at all */
-	init_idle();
-	current->nice = 20;
-	current->counter = -100;
 
 	while (1) {
 		void (*idle)(void) = pm_idle;
 		if (!idle)
 			idle = default_idle;
-		while (!current->need_resched)
+		if (!current->need_resched)
 			idle();
 		schedule();
 		check_pgt_cache();

arch/i386/kernel/smp.c +12 -2

diff -X dontdiff -Nur origlinux/arch/i386/kernel/smp.c mylinux/arch/i386/kernel/smp.c
@@ -105,7 +105,7 @@
 /* The 'big kernel lock' */
 spinlock_t kernel_flag __cacheline_aligned_in_smp = SPIN_LOCK_UNLOCKED;
 
-struct tlb_state cpu_tlbstate[NR_CPUS] = {[0 ... NR_CPUS-1] = { &init_mm, 0 }};
+struct tlb_state cpu_tlbstate[NR_CPUS] __cacheline_aligned = {[0 ... NR_CPUS-1] = { &init_mm, 0, }};
 
 /*
  * the following functions deal with sending IPIs between CPUs.

@@ -490,10 +490,20 @@
  * it goes straight through and wastes no time serializing
  * anything. Worst case is that we lose a reschedule ...
  */
-
 void smp_send_reschedule(int cpu)
 {
 	send_IPI_mask(1 << cpu, RESCHEDULE_VECTOR);
+}
+
+/*
+ * this function sends a reschedule IPI to all (other) CPUs.
+ * This should only be used if some 'global' task became runnable,
+ * such as a RT task, that must be handled now. The first CPU
+ * that manages to grab the task will run it.
+ */
+void smp_send_reschedule_all(void)
+{
+	send_IPI_allbutself(RESCHEDULE_VECTOR);
 }
 
 /*

arch/i386/kernel/smpboot.c +7 -9

diff -X dontdiff -Nur origlinux/arch/i386/kernel/smpboot.c mylinux/arch/i386/kernel/smpboot.c
@@ -308,14 +308,14 @@
 			if (tsc_values[i] < avg)
 				realdelta = -realdelta;
 
-			printk("BIOS BUG: CPU#%d improperly initialized, has %ld usecs TSC skew! FIXED.\n",
-				i, realdelta);
+			printk("BIOS BUG: CPU#%d improperly initialized, has %ld usecs TSC skew! FIXED.\n", i, realdelta);
 		}
 
 		sum += delta;
 	}
 	if (!buggy)
 		printk("passed.\n");
+		;
 }
 
 static void __init synchronize_tsc_ap (void)

@@ -365,7 +365,7 @@
 	 * (This works even if the APIC is not enabled.)
 	 */
 	phys_id = GET_APIC_ID(apic_read(APIC_ID));
-	cpuid = current->processor;
+	cpuid = cpu();
 	if (test_and_set_bit(cpuid, &cpu_online_map)) {
 		printk("huh, phys CPU#%d, CPU#%d already present??\n",
 					phys_id, cpuid);

@@ -471,6 +471,8 @@
 	 */
 	local_flush_tlb();
 
+	init_idle();
+	printk("cpu %d has done init idle, doing cpu_idle().\n", cpu());
 	return cpu_idle();
 }

@@ -803,16 +805,13 @@
 	if (!idle)
 		panic("No idle process for CPU %d", cpu);
 
-	idle->processor = cpu;
-	idle->cpus_runnable = 1 << cpu; /* we schedule the first task manually */
+	idle->cpu = cpu;
 
 	map_cpu_to_boot_apicid(cpu, apicid);
 
 	idle->thread.eip = (unsigned long) start_secondary;
 
-	del_from_runqueue(idle);
 	unhash_process(idle);
-	init_tasks[cpu] = idle;
 
 	/* start_eip had better be page-aligned! */
 	start_eip = setup_trampoline();

@@ -1020,8 +1019,7 @@
 	map_cpu_to_boot_apicid(0, boot_cpu_apicid);
 
 	global_irq_holder = 0;
-	current->processor = 0;
-	init_idle();
+	current->cpu = 0;
 	smp_tune_scheduling();
 
 	/*

arch/i386/mm/fault.c +2 -4

diff -X dontdiff -Nur origlinux/arch/i386/mm/fault.c mylinux/arch/i386/mm/fault.c
@@ -86,8 +86,7 @@
 
 out_of_memory:
 	if (current->pid == 1) {
-		current->policy |= SCHED_YIELD;
-		schedule();
+		yield();
 		goto survive;
 	}
 	goto bad_area;

@@ -342,8 +341,7 @@
 out_of_memory:
 	up_read(&mm->mmap_sem);
 	if (tsk->pid == 1) {
-		tsk->policy |= SCHED_YIELD;
-		schedule();
+		yield();
 		down_read(&mm->mmap_sem);
 		goto survive;
 	}

arch/ia64/kernel/head.S +8 -2

diff -X dontdiff -Nur origlinux/arch/ia64/kernel/head.S mylinux/arch/ia64/kernel/head.S
@@ -124,6 +124,7 @@
 #define isAP	p2	// are we an Application Processor?
 #define isBP	p3	// are we the Bootstrap Processor?
 
+
 #ifdef CONFIG_SMP
 	/*
 	 * Find the init_task for the currently booting CPU.  At poweron, and in

@@ -132,9 +133,14 @@
 	movl r3=cpucount
  	;;
 	ld4 r3=[r3]		// r3 <- smp_processor_id()
-	movl r2=init_tasks
+	movl r2=runqueues
+	movl r4=IA64_RUNQUEUE_SIZE
 	;;
-	shladd r2=r3,3,r2
+1:      add r2=r2,r4
+        ;;
+        br.cloop.sptk.many 1b
+        ;;
+        addl r2=IA64_RUNQUEUE_IDLE_OFFSET,r2
 	;;
 	ld8 r2=[r2]
 #else

arch/ia64/kernel/process.c +1 -4

diff -X dontdiff -Nur origlinux/arch/ia64/kernel/process.c mylinux/arch/ia64/kernel/process.c
@@ -125,9 +125,6 @@
 cpu_idle (void *unused)
 {
 	/* endless idle loop with no priority at all */
-	init_idle();
-	current->nice = 20;
-	current->counter = -100;
 
 
 	while (1) {

@@ -136,7 +133,7 @@
 			min_xtp();
 #endif
 
-		while (!current->need_resched) {
+		if (!current->need_resched) {
 #ifdef CONFIG_IA64_SGI_SN
 			snidle();
 #endif

arch/ia64/kernel/setup.c +3 -3

diff -X dontdiff -Nur origlinux/arch/ia64/kernel/setup.c mylinux/arch/ia64/kernel/setup.c
@@ -375,10 +375,10 @@
 {
 #ifdef CONFIG_SMP
 #	define lpj	c->loops_per_jiffy
-#	define cpu	c->processor
+#	define cpum	c->processor
 #else
 #	define lpj	loops_per_jiffy
-#	define cpu	0
+#	define cpum	0
 #endif
 	char family[32], features[128], *cp;
 	struct cpuinfo_ia64 *c = v;

@@ -417,7 +417,7 @@
 		   "cpu MHz    : %lu.%06lu\n"
 		   "itc MHz    : %lu.%06lu\n"
 		   "BogoMIPS   : %lu.%02lu\n\n",
-		   cpu, c->vendor, family, c->model, c->revision, c->archrev,
+		   cpum, c->vendor, family, c->model, c->revision, c->archrev,
 		   features, c->ppn, c->number,
 		   c->proc_freq / 1000000, c->proc_freq % 1000000,
 		   c->itc_freq / 1000000, c->itc_freq % 1000000,

arch/ia64/kernel/smp.c +6 -0

diff -X dontdiff -Nur origlinux/arch/ia64/kernel/smp.c mylinux/arch/ia64/kernel/smp.c
@@ -186,6 +186,12 @@
 }
 
 void
+smp_send_reschedule_all(void)
+{
+	send_IPI_all(IA64_IPI_RESCHEDULE);
+}
+
+void
 smp_flush_tlb_all (void)
 {
 	smp_call_function ((void (*)(void *))__flush_tlb_all,0,1,1);

arch/ia64/kernel/smpboot.c +4 -5

diff -X dontdiff -Nur origlinux/arch/ia64/kernel/smpboot.c mylinux/arch/ia64/kernel/smpboot.c
@@ -23,6 +23,7 @@
 #include <linux/smp.h>
 #include <linux/smp_lock.h>
 #include <linux/spinlock.h>
+#include <linux/sched.h>
 
 #include <asm/atomic.h>
 #include <asm/bitops.h>

@@ -323,7 +324,7 @@
 	extern void perfmon_init_percpu(void);
 #endif
 
-	cpuid = smp_processor_id();
+	cpuid = cpu();
 	phys_id = hard_smp_processor_id();
 
 	if (test_and_set_bit(cpuid, &cpu_online_map)) {

@@ -416,13 +417,11 @@
 	if (!idle)
 		panic("No idle process for CPU %d", cpu);
 
-	task_set_cpu(idle, cpu);	/* we schedule the first task manually */
+	idle->cpu = cpu();
 
 	ia64_cpu_to_sapicid[cpu] = sapicid;
 
-	del_from_runqueue(idle);
 	unhash_process(idle);
-	init_tasks[cpu] = idle;
 
 	Dprintk("Sending wakeup vector %u to AP 0x%x/0x%x.\n", ap_wakeup_vector, cpu, sapicid);

@@ -481,7 +480,7 @@
 	printk("Boot processor id 0x%x/0x%x\n", 0, boot_cpu_id);
 
 	global_irq_holder = 0;
-	current->processor = 0;
+	current->cpu = 0;
 	init_idle();
 
 	/*

arch/ia64/mm/fault.c +0 -2

diff -X dontdiff -Nur origlinux/arch/ia64/mm/fault.c mylinux/arch/ia64/mm/fault.c
@@ -194,8 +194,6 @@
   out_of_memory:
 	up_read(&mm->mmap_sem);
 	if (current->pid == 1) {
-		current->policy |= SCHED_YIELD;
-		schedule();
 		down_read(&mm->mmap_sem);
 		goto survive;
 	}

arch/ia64/tools/print_offsets.c +3 -1

diff -X dontdiff -Nur origlinux/arch/ia64/tools/print_offsets.c mylinux/arch/ia64/tools/print_offsets.c
@@ -50,11 +50,12 @@
     { "IA64_CPU_SIZE",			sizeof (struct cpuinfo_ia64) },
     { "SIGFRAME_SIZE",			sizeof (struct sigframe) },
     { "UNW_FRAME_INFO_SIZE",		sizeof (struct unw_frame_info) },
+    { "IA64_RUNQUEUE_SIZE",		sizeof (struct runqueue) },
     { "", 0 },			/* spacer */
     { "IA64_TASK_PTRACE_OFFSET",	offsetof (struct task_struct, ptrace) },
     { "IA64_TASK_SIGPENDING_OFFSET",	offsetof (struct task_struct, sigpending) },
     { "IA64_TASK_NEED_RESCHED_OFFSET",	offsetof (struct task_struct, need_resched) },
-    { "IA64_TASK_PROCESSOR_OFFSET",	offsetof (struct task_struct, processor) },
+    { "IA64_TASK_PROCESSOR_OFFSET",	offsetof (struct task_struct, cpu) },
     { "IA64_TASK_THREAD_OFFSET",	offsetof (struct task_struct, thread) },
     { "IA64_TASK_THREAD_KSP_OFFSET",	offsetof (struct task_struct, thread.ksp) },
 #ifdef CONFIG_PERFMON

@@ -62,6 +63,7 @@
 #endif
     { "IA64_TASK_PID_OFFSET",		offsetof (struct task_struct, pid) },
     { "IA64_TASK_MM_OFFSET",		offsetof (struct task_struct, mm) },
+    { "IA64_RUNQUEUE_IDLE_OFFSET",	offsetof (struct runqueue, idle) },
     { "IA64_PT_REGS_CR_IPSR_OFFSET",	offsetof (struct pt_regs, cr_ipsr) },
     { "IA64_PT_REGS_CR_IIP_OFFSET",	offsetof (struct pt_regs, cr_iip) },
     { "IA64_PT_REGS_CR_IFS_OFFSET",	offsetof (struct pt_regs, cr_ifs) },

drivers/block/loop.c +0 -3

diff -X dontdiff -Nur origlinux/drivers/block/loop.c mylinux/drivers/block/loop.c
@@ -570,9 +570,6 @@
 	flush_signals(current);
 	spin_unlock_irq(&current->sigmask_lock);
 
-	current->policy = SCHED_OTHER;
-	current->nice = -20;
-
 	spin_lock_irq(&lo->lo_lock);
 	lo->lo_state = Lo_bound;
 	atomic_inc(&lo->lo_pending);

drivers/ide/ataraid.c +4 ~~-10~~

diff -X dontdiff -Nur origlinux/drivers/ide/ataraid.c mylinux/drivers/ide/ataraid.c
@@ -121,11 +121,8 @@
 	void *ptr = NULL;
 	while (!ptr) {
 		ptr=kmalloc(sizeof(struct buffer_head),GFP_NOIO);
-		if (!ptr) {
-			__set_current_state(TASK_RUNNING);
-	                current->policy |= SCHED_YIELD;
-	                schedule();             
-		}
+		if (!ptr)
+			yield();
 	}
 	return ptr;
 }

@@ -137,11 +134,8 @@
 	void *ptr = NULL;
 	while (!ptr) {
 		ptr=kmalloc(sizeof(struct ataraid_bh_private),GFP_NOIO);
-		if (!ptr) {
-			__set_current_state(TASK_RUNNING);
-	                current->policy |= SCHED_YIELD;
-	                schedule();             
-		}
+		if (!ptr)
+			yield();
 	}
 	return ptr;
 }

drivers/md/md.c +1 ~~-11~~

diff -X dontdiff -Nur origlinux/drivers/md/md.c mylinux/drivers/md/md.c
@@ -2930,8 +2930,6 @@
 	 * bdflush, otherwise bdflush will deadlock if there are too
 	 * many dirty RAID5 blocks.
 	 */
-	current->policy = SCHED_OTHER;
-	current->nice = -20;
 	md_unlock_kernel();
 
 	complete(thread->event);

@@ -3381,11 +3379,6 @@
 	       "(but not more than %d KB/sec) for reconstruction.\n",
 	       sysctl_speed_limit_max);
 
-	/*
-	 * Resync has low priority.
-	 */
-	current->nice = 19;
-
 	is_mddev_idle(mddev); /* this also initializes IO event counters */
 	for (m = 0; m < SYNC_MARKS; m++) {
 		mark[m] = jiffies;

@@ -3463,16 +3456,13 @@
 		currspeed = (j-mddev->resync_mark_cnt)/2/((jiffies-mddev->resync_mark)/HZ +1) +1;
 
 		if (currspeed > sysctl_speed_limit_min) {
-			current->nice = 19;
-
 			if ((currspeed > sysctl_speed_limit_max) ||
 					!is_mddev_idle(mddev)) {
 				current->state = TASK_INTERRUPTIBLE;
 				md_schedule_timeout(HZ/4);
 				goto repeat;
 			}
-		} else
-			current->nice = -20;
+		}
 	}
 	printk(KERN_INFO "md: md%d: sync done.\n",mdidx(mddev));
 	err = 0;

drivers/net/slip.c +2 -4

diff -X dontdiff -Nur origlinux/drivers/net/slip.c mylinux/drivers/net/slip.c
@@ -1393,10 +1393,8 @@
 		/* First of all: check for active disciplines and hangup them.
 		 */
 		do {
-			if (busy) {
-				current->counter = 0;
-				schedule();
-			}
+			if (busy)
+				sys_sched_yield();
 
 			busy = 0;
 			local_bh_disable();

fs/binfmt_elf.c +1 -1

diff -X dontdiff -Nur origlinux/fs/binfmt_elf.c mylinux/fs/binfmt_elf.c
@@ -1143,7 +1143,7 @@
 	psinfo.pr_state = i;
 	psinfo.pr_sname = (i < 0 || i > 5) ? '.' : "RSDZTD"[i];
 	psinfo.pr_zomb = psinfo.pr_sname == 'Z';
-	psinfo.pr_nice = current->nice;
+	psinfo.pr_nice = current->__nice;
 	psinfo.pr_flag = current->flags;
 	psinfo.pr_uid = NEW_TO_OLD_UID(current->uid);
 	psinfo.pr_gid = NEW_TO_OLD_GID(current->gid);

fs/buffer.c +1 -3

diff -X dontdiff -Nur origlinux/fs/buffer.c mylinux/fs/buffer.c
@@ -725,9 +725,7 @@
 	wakeup_bdflush();
 	try_to_free_pages(zone, GFP_NOFS, 0);
 	run_task_queue(&tq_disk);
-	current->policy |= SCHED_YIELD;
-	__set_current_state(TASK_RUNNING);
-	schedule();
+	sys_sched_yield();
 }
 
 void init_buffer(struct buffer_head *bh, bh_end_io_t *handler, void *private)

fs/jbd/journal.c +3 -6

diff -X dontdiff -Nur origlinux/fs/jbd/journal.c mylinux/fs/jbd/journal.c
@@ -460,8 +460,7 @@
 			printk (KERN_NOTICE __FUNCTION__
 				": ENOMEM at get_unused_buffer_head, "
 				"trying again.\n");
-			current->policy |= SCHED_YIELD;
-			schedule();
+			yield();
 		}
 	} while (!new_bh);
 	/* keep subsequent assertions sane */

@@ -1539,8 +1538,7 @@
 			last_warning = jiffies;
 		}
 		
-		current->policy |= SCHED_YIELD;
-		schedule();
+		yield();
 	}
 }

@@ -1598,8 +1596,7 @@
 			last_warning = jiffies;
 		}
 		while (ret == 0) {
-			current->policy |= SCHED_YIELD;
-			schedule();
+			yield();
 			ret = kmem_cache_alloc(journal_head_cache, GFP_NOFS);
 		}
 	}

fs/jbd/revoke.c +1 -2

diff -X dontdiff -Nur origlinux/fs/jbd/revoke.c mylinux/fs/jbd/revoke.c
@@ -137,8 +137,7 @@
 	if (!journal_oom_retry)
 		return -ENOMEM;
 	jbd_debug(1, "ENOMEM in " __FUNCTION__ ", retrying.\n");
-	current->policy |= SCHED_YIELD;
-	schedule();
+	yield();
 	goto repeat;
 }

fs/jbd/transaction.c +1 -2

diff -X dontdiff -Nur origlinux/fs/jbd/transaction.c mylinux/fs/jbd/transaction.c
@@ -1377,8 +1377,7 @@
 		do {
 			old_handle_count = transaction->t_handle_count;
 			set_current_state(TASK_RUNNING);
-			current->policy |= SCHED_YIELD;
-			schedule();
+			yield();
 		} while (old_handle_count != transaction->t_handle_count);
 	}

fs/jffs2/background.c +0 -3

diff -X dontdiff -Nur origlinux/fs/jffs2/background.c mylinux/fs/jffs2/background.c
@@ -106,9 +106,6 @@
 
         sprintf(current->comm, "jffs2_gcd_mtd%d", c->mtd->index);
 
-	/* FIXME in the 2.2 backport */
-	current->nice = 10;
-
 	for (;;) {
 		spin_lock_irq(&current->sigmask_lock);
 		siginitsetinv (&current->blocked, sigmask(SIGHUP) | sigmask(SIGKILL) | sigmask(SIGSTOP) | sigmask(SIGCONT));

fs/locks.c +1 -2

diff -X dontdiff -Nur origlinux/fs/locks.c mylinux/fs/locks.c
@@ -445,8 +445,7 @@
 			/* Let the blocked process remove waiter from the
 			 * block list when it gets scheduled.
 			 */
-			current->policy |= SCHED_YIELD;
-			schedule();
+			yield();
 		} else {
 			/* Remove waiter from the block list, because by the
 			 * time it wakes up blocker won't exist any more.

fs/nfs/pagelist.c +1 -2

diff -X dontdiff -Nur origlinux/fs/nfs/pagelist.c mylinux/fs/nfs/pagelist.c
@@ -96,8 +96,7 @@
 			continue;
 		if (signalled() && (server->flags & NFS_MOUNT_INTR))
 			return ERR_PTR(-ERESTARTSYS);
-		current->policy = SCHED_YIELD;
-		schedule();
+		yield();
 	}
 
 	/* Initialize the request struct. Initially, we assume a

fs/proc/array.c +7 -4

diff -X dontdiff -Nur origlinux/fs/proc/array.c mylinux/fs/proc/array.c
@@ -335,9 +335,12 @@
 
 	/* scale priority and nice values from timeslices to -20..20 */
 	/* to make it look like a "normal" Unix priority/nice value  */
-	priority = task->counter;
-	priority = 20 - (priority * 10 + DEF_COUNTER / 2) / DEF_COUNTER;
-	nice = task->nice;
+	priority = task->prio;
+	if (priority >= MAX_RT_PRIO)
+		priority -= MAX_RT_PRIO;
+	else
+		priority = priority-100;
+	nice = task->__nice;
 
 	read_lock(&tasklist_lock);
 	ppid = task->pid ? task->p_opptr->pid : 0;

@@ -387,7 +390,7 @@
 		task->nswap,
 		task->cnswap,
 		task->exit_signal,
-		task->processor);
+		task->cpu);
 	if(mm)
 		mmput(mm);
 	return res;

fs/proc/proc_misc.c +5 -5

diff -X dontdiff -Nur origlinux/fs/proc/proc_misc.c mylinux/fs/proc/proc_misc.c
@@ -85,11 +85,11 @@
 	a = avenrun[0] + (FIXED_1/200);
 	b = avenrun[1] + (FIXED_1/200);
 	c = avenrun[2] + (FIXED_1/200);
-	len = sprintf(page,"%d.%02d %d.%02d %d.%02d %d/%d %d\n",
+	len = sprintf(page,"%d.%02d %d.%02d %d.%02d %ld/%d %d\n",
 		LOAD_INT(a), LOAD_FRAC(a),
 		LOAD_INT(b), LOAD_FRAC(b),
 		LOAD_INT(c), LOAD_FRAC(c),
-		nr_running, nr_threads, last_pid);
+		nr_running(), nr_threads, last_pid);
 	return proc_calc_metrics(page, start, off, count, eof, len);
 }

@@ -101,7 +101,7 @@
 	int len;
 
 	uptime = jiffies;
-	idle = init_tasks[0]->times.tms_utime + init_tasks[0]->times.tms_stime;
+	idle = init_task.times.tms_utime + init_task.times.tms_stime;
 
 	/* The formula for the fraction parts really is ((t * 100) / HZ) % 100, but
 	   that would overflow about every five days at HZ == 100.

@@ -303,10 +303,10 @@
 	}
 
 	len += sprintf(page + len,
-		"\nctxt %u\n"
+		"\nctxt %lu\n"
 		"btime %lu\n"
 		"processes %lu\n",
-		kstat.context_swtch,
+		nr_context_switches(),
 		xtime.tv_sec - jif / HZ,
 		total_forks);

fs/reiserfs/buffer2.c +3 -4

diff -X dontdiff -Nur origlinux/fs/reiserfs/buffer2.c mylinux/fs/reiserfs/buffer2.c
@@ -33,8 +33,7 @@
 			buffer_journal_dirty(bh) ? ' ' : '!');
     }
     run_task_queue(&tq_disk);
-    current->policy |= SCHED_YIELD;
-    schedule();
+    yield();
   }
   if (repeat_counter > 30000000) {
     reiserfs_warning("vs-3051: done waiting, ignore vs-3050 messages for (%b)\n", bh) ;

@@ -52,11 +51,11 @@
 struct buffer_head  * reiserfs_bread (struct super_block *super, int n_block, int n_size) 
 {
     struct buffer_head  *result;
-    PROC_EXP( unsigned int ctx_switches = kstat.context_swtch );
+    PROC_EXP( unsigned int ctx_switches = nr_context_switches(); );
 
     result = bread (super -> s_dev, n_block, n_size);
     PROC_INFO_INC( super, breads );
-    PROC_EXP( if( kstat.context_swtch != ctx_switches ) 
+    PROC_EXP( if( nr_context_switches() != ctx_switches ) 
 	      PROC_INFO_INC( super, bread_miss ) );
     return result;
 }

fs/reiserfs/journal.c +1 -2

diff -X dontdiff -Nur origlinux/fs/reiserfs/journal.c mylinux/fs/reiserfs/journal.c
@@ -149,8 +149,7 @@
   }
   bn = allocate_bitmap_node(p_s_sb) ;
   if (!bn) {
-    current->policy |= SCHED_YIELD ;
-    schedule() ;
+    yield();
     goto repeat ;
   }
   return bn ;

fs/ufs/truncate.c +1 -4

diff -X dontdiff -Nur origlinux/fs/ufs/truncate.c mylinux/fs/ufs/truncate.c
@@ -448,10 +448,7 @@
 		if (IS_SYNC(inode) && (inode->i_state & I_DIRTY))
 			ufs_sync_inode (inode);
 		run_task_queue(&tq_disk);
-		current->policy |= SCHED_YIELD;
-		schedule ();
-
-
+		yield();
 	}
 	offset = inode->i_size & uspi->s_fshift;
 	if (offset) {

include/asm-i386/bitops.h +8 -0

diff -X dontdiff -Nur origlinux/include/asm-i386/bitops.h mylinux/include/asm-i386/bitops.h
@@ -75,6 +75,14 @@
 		:"=m" (ADDR)
 		:"Ir" (nr));
 }
+
+static __inline__ void __clear_bit(int nr, volatile void * addr)
+{
+	__asm__ __volatile__(
+		"btrl %1,%0"
+		:"=m" (ADDR)
+		:"Ir" (nr));
+}
 #define smp_mb__before_clear_bit()	barrier()
 #define smp_mb__after_clear_bit()	barrier()

include/asm-i386/mmu_context.h +22 -0

diff -X dontdiff -Nur origlinux/include/asm-i386/mmu_context.h mylinux/include/asm-i386/mmu_context.h
@@ -7,6 +7,28 @@
 #include <asm/pgalloc.h>
 
 /*
+ * Every architecture must define this function. It's the fastest
+ * way of searching a 168-bit bitmap where the first 128 bits are
+ * unlikely to be set. It's guaranteed that at least one of the 168
+ * bits is cleared.
+ */
+#if MAX_RT_PRIO != 128 || MAX_PRIO != 168
+# error update this function.
+#endif
+
+static inline int sched_find_first_zero_bit(unsigned long *b)
+{
+	unsigned int rt;
+
+	rt = b[0] & b[1] & b[2] & b[3];
+	if (unlikely(rt != 0xffffffff))
+		return find_first_zero_bit(b, MAX_RT_PRIO);
+
+	if (b[4] != ~0)
+		return ffz(b[4]) + MAX_RT_PRIO;
+	return ffz(b[5]) + 32 + MAX_RT_PRIO;
+}
+/*
  * possibly do the LDT unload here?
  */
 #define destroy_context(mm)		do { } while(0)

include/asm-i386/pgalloc.h +1 -0

diff -X dontdiff -Nur origlinux/include/asm-i386/pgalloc.h mylinux/include/asm-i386/pgalloc.h
@@ -224,6 +224,7 @@
 {
 	struct mm_struct *active_mm;
 	int state;
+	char __cacheline_padding[24];
 };
 extern struct tlb_state cpu_tlbstate[NR_CPUS];

include/asm-i386/smp.h +2 ~~-13~~

diff -X dontdiff -Nur origlinux/include/asm-i386/smp.h mylinux/include/asm-i386/smp.h
@@ -63,6 +63,7 @@
 extern void smp_flush_tlb(void);
 extern void smp_message_irq(int cpl, void *dev_id, struct pt_regs *regs);
 extern void smp_send_reschedule(int cpu);
+extern void smp_send_reschedule_all(void);
 extern void smp_invalidate_rcv(void);		/* Process an NMI */
 extern void (*mtrr_hook) (void);
 extern void zap_low_mappings (void);

@@ -104,7 +105,7 @@
  * so this is correct in the x86 case.
  */
 
-#define smp_processor_id() (current->processor)
+#define smp_processor_id() (current->cpu)
 
 static __inline int hard_smp_processor_id(void)
 {

@@ -121,18 +122,6 @@
 #endif /* !__ASSEMBLY__ */
 
 #define NO_PROC_ID		0xFF		/* No processor magic marker */
-
-/*
- *	This magic constant controls our willingness to transfer
- *	a process across CPUs. Such a transfer incurs misses on the L1
- *	cache, and on a P6 or P5 with multiple L2 caches L2 hits. My
- *	gut feeling is this will vary by board in value. For a board
- *	with separate L2 cache it probably depends also on the RSS, and
- *	for a board with shared L2 cache it ought to decay fast as other
- *	processes are run.
- */
- 
-#define PROC_CHANGE_PENALTY	15		/* Schedule penalty */
 
 #endif
 #endif

include/asm-ia64/bitops.h +1 -0

diff -X dontdiff -Nur origlinux/include/asm-ia64/bitops.h mylinux/include/asm-ia64/bitops.h
@@ -368,6 +368,7 @@
 
 #ifdef __KERNEL__
 
+#define __clear_bit(nr, addr)		clear_bit(nr, addr)
 #define ext2_set_bit                 test_and_set_bit
 #define ext2_clear_bit               test_and_clear_bit
 #define ext2_test_bit                test_bit

include/asm-ia64/mmu_context.h +1 -0

diff -X dontdiff -Nur origlinux/include/asm-ia64/mmu_context.h mylinux/include/asm-ia64/mmu_context.h
@@ -118,6 +118,7 @@
 	reload_context(next);
 }
 
+#define sched_find_first_zero_bit(bitmap)	ffz(bitmap)
 #define switch_mm(prev_mm,next_mm,next_task,cpu)	activate_mm(prev_mm, next_mm)
 
 # endif /* ! __ASSEMBLY__ */

include/asm-ia64/smp.h +1 -7

diff -X dontdiff -Nur origlinux/include/asm-ia64/smp.h mylinux/include/asm-ia64/smp.h
@@ -27,7 +27,7 @@
 #define SMP_IRQ_REDIRECTION	(1 << 0)
 #define SMP_IPI_REDIRECTION	(1 << 1)
 
-#define smp_processor_id()	(current->processor)
+#define smp_processor_id()	(current->cpu)
 
 extern struct smp_boot_data {
 	int cpu_count;

@@ -109,12 +109,6 @@
 }
 
 #define NO_PROC_ID		0xffffffff	/* no processor magic marker */
-
-/*
- * Extra overhead to move a task from one cpu to another (due to TLB and cache misses).
- * Expressed in "negative nice value" units (larger number means higher priority/penalty).
- */
-#define PROC_CHANGE_PENALTY	20
 
 extern void __init init_smp_config (void);
 extern void smp_do_timer (struct pt_regs *regs);

include/linux/kernel_stat.h +2 -1

diff -X dontdiff -Nur origlinux/include/linux/kernel_stat.h mylinux/include/linux/kernel_stat.h
@@ -32,10 +32,11 @@
 	unsigned int ipackets, opackets;
 	unsigned int ierrors, oerrors;
 	unsigned int collisions;
-	unsigned int context_swtch;
 };
 
 extern struct kernel_stat kstat;
+
+extern unsigned long nr_context_switches(void);
 
 #if !defined(CONFIG_ARCH_S390)
 /*

include/linux/list.h +2 -0

diff -X dontdiff -Nur origlinux/include/linux/list.h mylinux/include/linux/list.h
@@ -19,6 +19,8 @@
 	struct list_head *next, *prev;
 };
 
+typedef struct list_head list_t;
+
 #define LIST_HEAD_INIT(name) { &(name), &(name) }
 
 #define LIST_HEAD(name) \

include/linux/sched.h +143 ~~-64~~

306 lines not on this page · raw message

include/linux/smp.h +8 -0

diff -X dontdiff -Nur origlinux/include/linux/smp.h mylinux/include/linux/smp.h
@@ -77,6 +77,14 @@
 #define cpu_number_map(cpu)			0
 #define smp_call_function(func,info,retry,wait)	({ 0; })
 #define cpu_online_map				1
+static inline void smp_send_reschedule(int cpu) { }
+static inline void smp_send_reschedule_all(void) { }
 
 #endif
+
+/*
+ * Common definitions:
+ */
+#define cpu()					smp_processor_id()
+
 #endif

init/main.c +20 ~~-23~~

73 lines not on this page · raw message

kernel/capability.c +2 -0

diff -X dontdiff -Nur origlinux/kernel/capability.c mylinux/kernel/capability.c
@@ -8,6 +8,8 @@
 #include <linux/mm.h>
 #include <asm/uaccess.h>
 
+unsigned securebits = SECUREBITS_DEFAULT; /* systemwide security settings */
+
 kernel_cap_t cap_bset = CAP_INIT_EFF_SET;
 
 /* Note: never hold tasklist_lock while spinning for this one */

kernel/exit.c +104 ~~-41~~

159 lines not on this page · raw message

kernel/fork.c +42 ~~-13~~

108 lines not on this page · raw message

kernel/ksyms.c +4 -0

diff -X dontdiff -Nur origlinux/kernel/ksyms.c mylinux/kernel/ksyms.c
@@ -437,6 +437,9 @@
 EXPORT_SYMBOL(interruptible_sleep_on_timeout);
 EXPORT_SYMBOL(schedule);
 EXPORT_SYMBOL(schedule_timeout);
+EXPORT_SYMBOL(sys_sched_yield);
+EXPORT_SYMBOL(set_user_nice);
+EXPORT_SYMBOL(set_cpus_allowed);
 EXPORT_SYMBOL(jiffies);
 EXPORT_SYMBOL(xtime);
 EXPORT_SYMBOL(do_gettimeofday);

@@ -448,6 +451,7 @@
 
 EXPORT_SYMBOL(kstat);
 EXPORT_SYMBOL(nr_running);
+EXPORT_SYMBOL(nr_context_switches);
 
 /* misc */
 EXPORT_SYMBOL(panic);

kernel/printk.c +1 -0

diff -X dontdiff -Nur origlinux/kernel/printk.c mylinux/kernel/printk.c
@@ -25,6 +25,7 @@
 #include <linux/init.h>
 #include <linux/module.h>
 #include <linux/interrupt.h>			/* For in_interrupt() */
+#include <linux/delay.h>
 
 #include <asm/uaccess.h>

kernel/ptrace.c +1 ~~-14~~

diff -X dontdiff -Nur origlinux/kernel/ptrace.c mylinux/kernel/ptrace.c
@@ -31,20 +31,7 @@
 		if (child->state != TASK_STOPPED)
 			return -ESRCH;
 #ifdef CONFIG_SMP
-		/* Make sure the child gets off its CPU.. */
-		for (;;) {
-			task_lock(child);
-			if (!task_has_cpu(child))
-				break;
-			task_unlock(child);
-			do {
-				if (child->state != TASK_STOPPED)
-					return -ESRCH;
-				barrier();
-				cpu_relax();
-			} while (task_has_cpu(child));
-		}
-		task_unlock(child);
+		wait_task_inactive(child);
 #endif		
 	}

kernel/sched.c +850 ~~-719~~

1,597 lines not on this page · raw message

kernel/signal.c +3 -6

diff -X dontdiff -Nur origlinux/kernel/signal.c mylinux/kernel/signal.c

Lines of this file are not shown. View the raw message.

kernel/softirq.c +9 ~~-12~~

diff -X dontdiff -Nur origlinux/kernel/softirq.c mylinux/kernel/softirq.c

Lines of this file are not shown. View the raw message.

kernel/sys.c +3 -3

diff -X dontdiff -Nur origlinux/kernel/sys.c mylinux/kernel/sys.c

Lines of this file are not shown. View the raw message.

kernel/timer.c +91 -7

98 lines not on this page · raw message

mm/highmem.c +2 -6

diff -X dontdiff -Nur origlinux/mm/highmem.c mylinux/mm/highmem.c

Lines of this file are not shown. View the raw message.

mm/oom_kill.c +3 -4

diff -X dontdiff -Nur origlinux/mm/oom_kill.c mylinux/mm/oom_kill.c

Lines of this file are not shown. View the raw message.

mm/page_alloc.c +1 -3

diff -X dontdiff -Nur origlinux/mm/page_alloc.c mylinux/mm/page_alloc.c

Lines of this file are not shown. View the raw message.

net/ipv4/tcp_output.c +1 -2

diff -X dontdiff -Nur origlinux/net/ipv4/tcp_output.c mylinux/net/ipv4/tcp_output.c

Lines of this file are not shown. View the raw message.

net/sched/sch_generic.c +2 -4

diff -X dontdiff -Nur origlinux/net/sched/sch_generic.c mylinux/net/sched/sch_generic.c

Lines of this file are not shown. View the raw message.

net/socket.c +1 -2

diff -X dontdiff -Nur origlinux/net/socket.c mylinux/net/socket.c

Lines of this file are not shown. View the raw message.

net/sunrpc/sched.c +3 -6

diff -X dontdiff -Nur origlinux/net/sunrpc/sched.c mylinux/net/sunrpc/sched.c

Lines of this file are not shown. View the raw message.

net/unix/af_unix.c +2 -4

diff -X dontdiff -Nur origlinux/net/unix/af_unix.c mylinux/net/unix/af_unix.c

Lines of this file are not shown. View the raw message.

The rest of this diff is not shown. View the raw message.

Part of the diff is not shown here. View the raw message.

David Mosberger

2002-01-12 03:13 UTC #1 lore raw

Nick> I'm trying to get Ingo's scheduler working on IA64 but I've hit a Nick> dead-end with the head.S code. Ingo's patch removes init_tasks, Nick> so I've modified the assembly in head.S to point at Nick> runqueues(cpu)->idle, I think - it dies very early in the boot, Nick> and I'm not familiar with ia64 assembly.

Nick> Other issues, I had to build offsets.h by hand, and I moved some Nick> stuff from sched.c to sched.h. Other than that, it's H6 + ia64.

Nick> Anyone have any feedback on getting this booting?

This may not help you but just fyi: I'm planning to sync up with 2.5 next week (2.5.2 by then, hopefully). Of course, if someone else has a working patch by then, I'll take it. ;-)

	--david

Erich Focht

2002-01-14 18:23 UTC #2 lore raw