diff mbox

Generic idle causes boot hang on T1000

Message ID 20140323.203859.1533896357512062661.davem@davemloft.net
State RFC
Delegated to: David Miller
Headers show

Commit Message

David Miller March 24, 2014, 12:38 a.m. UTC 
From: Jan Engelhardt <jengelh@inai.de>
Date: Sun, 23 Mar 2014 23:47:47 +0100 (CET)

> 
> Trying to update beyond 3.9 gives me a boot hang on SPARC T1000.
> The following commit is the result of the bisect:
> 
> commit 87fa05aeb3a5e8e21b1a5510eef6983650eff092
> Author: Sam Ravnborg <sam@ravnborg.org>
> Date:   Thu Apr 11 21:38:50 2013 +0200
> 
>     sparc: Use generic idle loop

Interesting.

> mptbase: ioc0: Initiating bringup
> <hangs>
> 
> after ~1 minute:
> 
> INFO: rcu_sched self-detected stall on CPU { 0}  (t=30061 jiffies
> g=18446744073709551319 c=18446744073709551318 q=414)
> INFO: rcu_sched detected stalls on CPUs/tasks:* CPU[  0]:
> TSTATE[0000000080001603] TPC[000000000042b9b4] TNPC[000000000042b9b8]
> TASK[swapper/0:0]
>              TPC[arch_cpu_idle+0x74/0xa0] O7[arch_cpu_idle+0x5c/0xa0]
>  I7[cpu_startup_entry+0x114/0x1a0] RPC[start_kernel+0x384/0x394]

I suspect the cpu yield hypervisor call requires local cpu interrupts
to be enabled in order to function properly.  Prior to this commit
above, they did.

Please try this:

commit 87fa05aeb3a5e8e21b1a5510eef6983650eff092
Author: Sam Ravnborg <sam@ravnborg.org>
Date:   Thu Apr 11 21:38:50 2013 +0200

    sparc: Use generic idle loop
    
    Add generic cpu_idle support
    
    sparc32:
    - replace call to cpu_idle() with cpu_startup_entry()
    - add arch_cpu_idle()
    
    sparc64:
    - smp_callin() now include cpu_startup_entry() call so we can
      skip calling cpu_idle from assembler
    - add arch_cpu_idle() and arch_cpu_idle_dead()
    
    Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
    Reviewed-by: "Srivatsa S. Bhat" <srivatsa.bhat@linux.vnet.ibm.com>
    Cc: torvalds@linux-foundation.org
    Cc: rusty@rustcorp.com.au
    Cc: paulmck@linux.vnet.ibm.com
    Cc: peterz@infradead.org
    Cc: magnus.damm@gmail.com
    Acked-by: David Miller <davem@davemloft.net>
    Link: http://lkml.kernel.org/r/20130411193850.GA2330@merkur.ravnborg.org
    Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

--
To unsubscribe from this list: send the line "unsubscribe sparclinux" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Comments

Jan Engelhardt March 24, 2014, 10:33 a.m. UTC | #1 
On Monday 2014-03-24 01:38, David Miller wrote:
>> Trying to update beyond 3.9 gives me a boot hang on SPARC T1000.
>> The following commit is the result of the bisect:
>> 
>> commit 87fa05aeb3a5e8e21b1a5510eef6983650eff092
>> Author: Sam Ravnborg <sam@ravnborg.org>
>> Date:   Thu Apr 11 21:38:50 2013 +0200
>> 
>>     sparc: Use generic idle loop
>
>Interesting.
>
>> mptbase: ioc0: Initiating bringup
>> <hangs>
>> 
>> after ~1 minute:
>> 
>> INFO: rcu_sched self-detected stall on CPU { 0}  (t=30061 jiffies
>> g=18446744073709551319 c=18446744073709551318 q=414)
>> INFO: rcu_sched detected stalls on CPUs/tasks:* CPU[  0]:
>> TSTATE[0000000080001603] TPC[000000000042b9b4] TNPC[000000000042b9b8]
>> TASK[swapper/0:0]
>>              TPC[arch_cpu_idle+0x74/0xa0] O7[arch_cpu_idle+0x5c/0xa0]
>>  I7[cpu_startup_entry+0x114/0x1a0] RPC[start_kernel+0x384/0x394]
>
>I suspect the cpu yield hypervisor call requires local cpu interrupts
>to be enabled in order to function properly.  Prior to this commit
>above, they did.
>
>Please try this:
>
>commit 87fa05aeb3a5e8e21b1a5510eef6983650eff092

This is the same commit as before (both in ID, as well as patch content).
Did you intend to paste another patch?

--
To unsubscribe from this list: send the line "unsubscribe sparclinux" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/arch/sparc/Kconfig b/arch/sparc/Kconfig
index 3d361f2..ee5eacc 100644
--- a/arch/sparc/Kconfig
+++ b/arch/sparc/Kconfig
@@ -37,6 +37,7 @@  config SPARC
 	select GENERIC_SMP_IDLE_THREAD
 	select GENERIC_CMOS_UPDATE
 	select GENERIC_CLOCKEVENTS
+	select GENERIC_IDLE_LOOP
 	select GENERIC_STRNCPY_FROM_USER
 	select GENERIC_STRNLEN_USER
 	select MODULES_USE_ELF_RELA
diff --git a/arch/sparc/kernel/hvtramp.S b/arch/sparc/kernel/hvtramp.S
index 9365432..605c960 100644
--- a/arch/sparc/kernel/hvtramp.S
+++ b/arch/sparc/kernel/hvtramp.S
@@ -128,8 +128,7 @@  hv_cpu_startup:
 
 	call		smp_callin
 	 nop
-	call		cpu_idle
-	 mov		0, %o0
+
 	call		cpu_panic
 	 nop
 
diff --git a/arch/sparc/kernel/process_32.c b/arch/sparc/kernel/process_32.c
index 62eede1..c852410 100644
--- a/arch/sparc/kernel/process_32.c
+++ b/arch/sparc/kernel/process_32.c
@@ -64,23 +64,12 @@  extern void fpsave(unsigned long *, unsigned long *, void *, unsigned long *);
 struct task_struct *last_task_used_math = NULL;
 struct thread_info *current_set[NR_CPUS];
 
-/*
- * the idle loop on a Sparc... ;)
- */
-void cpu_idle(void)
+/* Idle loop support. */
+void arch_cpu_idle(void)
 {
-	set_thread_flag(TIF_POLLING_NRFLAG);
-
-	/* endless idle loop with no priority at all */
-	for (;;) {
-		while (!need_resched()) {
-			if (sparc_idle)
-				(*sparc_idle)();
-			else
-				cpu_relax();
-		}
-		schedule_preempt_disabled();
-	}
+	if (sparc_idle)
+		(*sparc_idle)();
+	local_irq_enable();
 }
 
 /* XXX cli/sti -> local_irq_xxx here, check this works once SMP is fixed. */
diff --git a/arch/sparc/kernel/process_64.c b/arch/sparc/kernel/process_64.c
index cdb80b2..9fbf0d1 100644
--- a/arch/sparc/kernel/process_64.c
+++ b/arch/sparc/kernel/process_64.c
@@ -52,20 +52,17 @@ 
 
 #include "kstack.h"
 
-static void sparc64_yield(int cpu)
+/* Idle loop support on sparc64. */
+void arch_cpu_idle(void)
 {
 	if (tlb_type != hypervisor) {
 		touch_nmi_watchdog();
-		return;
-	}
-
-	clear_thread_flag(TIF_POLLING_NRFLAG);
-	smp_mb__after_clear_bit();
-
-	while (!need_resched() && !cpu_is_offline(cpu)) {
+	} else {
 		unsigned long pstate;
 
-		/* Disable interrupts. */
+                /* The sun4v sleeping code requires that we have PSTATE.IE cleared over
+                 * the cpu sleep hypervisor call.
+                 */
 		__asm__ __volatile__(
 			"rdpr %%pstate, %0\n\t"
 			"andn %0, %1, %0\n\t"
@@ -73,7 +70,7 @@  static void sparc64_yield(int cpu)
 			: "=&r" (pstate)
 			: "i" (PSTATE_IE));
 
-		if (!need_resched() && !cpu_is_offline(cpu))
+		if (!need_resched() && !cpu_is_offline(smp_processor_id()))
 			sun4v_cpu_yield();
 
 		/* Re-enable interrupts. */
@@ -84,36 +81,16 @@  static void sparc64_yield(int cpu)
 			: "=&r" (pstate)
 			: "i" (PSTATE_IE));
 	}
-
-	set_thread_flag(TIF_POLLING_NRFLAG);
+	local_irq_enable();
 }
 
-/* The idle loop on sparc64. */
-void cpu_idle(void)
-{
-	int cpu = smp_processor_id();
-
-	set_thread_flag(TIF_POLLING_NRFLAG);
-
-	while(1) {
-		tick_nohz_idle_enter();
-		rcu_idle_enter();
-
-		while (!need_resched() && !cpu_is_offline(cpu))
-			sparc64_yield(cpu);
-
-		rcu_idle_exit();
-		tick_nohz_idle_exit();
-
 #ifdef CONFIG_HOTPLUG_CPU
-		if (cpu_is_offline(cpu)) {
-			sched_preempt_enable_no_resched();
-			cpu_play_dead();
-		}
-#endif
-		schedule_preempt_disabled();
-	}
+void arch_cpu_idle_dead()
+{
+	sched_preempt_enable_no_resched();
+	cpu_play_dead();
 }
+#endif
 
 #ifdef CONFIG_COMPAT
 static void show_regwindow32(struct pt_regs *regs)
diff --git a/arch/sparc/kernel/smp_32.c b/arch/sparc/kernel/smp_32.c
index 9e7e6d7..e3f2b81 100644
--- a/arch/sparc/kernel/smp_32.c
+++ b/arch/sparc/kernel/smp_32.c
@@ -369,7 +369,7 @@  void __cpuinit sparc_start_secondary(void *arg)
 	local_irq_enable();
 
 	wmb();
-	cpu_idle();
+	cpu_startup_entry(CPUHP_ONLINE);
 
 	/* We should never reach here! */
 	BUG();
diff --git a/arch/sparc/kernel/smp_64.c b/arch/sparc/kernel/smp_64.c
index 537eb66..c025ffc 100644
--- a/arch/sparc/kernel/smp_64.c
+++ b/arch/sparc/kernel/smp_64.c
@@ -127,6 +127,8 @@  void __cpuinit smp_callin(void)
 
 	/* idle thread is expected to have preempt disabled */
 	preempt_disable();
+
+	cpu_startup_entry(CPUHP_ONLINE);
 }
 
 void cpu_panic(void)
diff --git a/arch/sparc/kernel/trampoline_64.S b/arch/sparc/kernel/trampoline_64.S
index da1b781..2e973a2 100644
--- a/arch/sparc/kernel/trampoline_64.S
+++ b/arch/sparc/kernel/trampoline_64.S
@@ -407,8 +407,7 @@  after_lock_tlb:
 
 	call		smp_callin
 	 nop
-	call		cpu_idle
-	 mov		0, %o0
+
 	call		cpu_panic
 	 nop
 1:	b,a,pt		%xcc, 1b