LKML Archive on lore.kernel.org
help / color / mirror / Atom feed
* Fwd: [PATCH 7/9] ACPI: Only use IPI on known broken machines (AMD, Dothan/BaniasPentium M)
@ 2007-03-20 17:57 Len Brown
2007-03-20 19:23 ` Andi Kleen
0 siblings, 1 reply; 3+ messages in thread
From: Len Brown @ 2007-03-20 17:57 UTC (permalink / raw)
To: tglx, Andrew Morton, Ingo Molnar, Andi Kleen, Thomas Renninger
Cc: Linux Kernel Mailing List
[-- Attachment #1: Type: text/plain, Size: 4685 bytes --]
Thomas/Ingo/Andi/Andrew,
This one in my tree merits your review and comment.
thanks,
-Len
---------- Forwarded Message ----------
Subject: [PATCH 7/9] ACPI: Only use IPI on known broken machines (AMD, Dothan/BaniasPentium M)
Date: Tuesday 20 March 2007 11:27
From: Len Brown <len.brown@intel.com>
To: linux-acpi@vger.kernel.org
Cc: Thomas Renninger <trenn@suse.de>, Len Brown <len.brown@intel.com>
From: Thomas Renninger <trenn@suse.de>
Use IPI for blacklisted CPUs, add parameter IPI vs LAPIC
Currently, Linux disables lapic timer for all machines with C2 and higher
C-state support.
According to Intel only specific Intel models (Banias/Dothan) are broken
in respect of not waking up from C2 with lapic.
However, I am not sure about the naming of the parameter and how it
could/should get integrated into the dyntick part
(CONFIG_GENERIC_CLOCKEVENTS). There, a more fine grained check (TSC
still running?, ..) is needed? Does this make sense (always use
CLOCK_EVT_NOTIFY_BROADCAST_ON, but use OFF if forced by use_ipi=0:
clockevents_notify(use_ipi ? CLOCK_EVT_NOTIFY_BROADCAST_ON :
CLOCK_EVT_NOTIFY_BROADCAST_OFF, &pr->id);
Signed-off-by: Thomas Renninger <trenn@suse.de>
Signed-off-by: Len Brown <len.brown@intel.com>
---
drivers/acpi/processor_idle.c | 38 +++++++++++++++++++++++++++++---------
1 files changed, 29 insertions(+), 9 deletions(-)
diff --git a/drivers/acpi/processor_idle.c b/drivers/acpi/processor_idle.c
index 6077300..562124e 100644
--- a/drivers/acpi/processor_idle.c
+++ b/drivers/acpi/processor_idle.c
@@ -89,6 +89,12 @@ module_param(nocst, uint, 0000);
static unsigned int bm_history __read_mostly =
(HZ >= 800 ? 0xFFFFFFFF : ((1U << (HZ / 25)) - 1));
module_param(bm_history, uint, 0644);
+
+static unsigned use_ipi = 2;
+module_param(use_ipi, uint, 0644);
+MODULE_PARM_DESC(use_ipi, "IPI (vs. LAPIC) irqs for not waking up from C2/C3"
+ " machines. 0=apic, 1=ipi, 2=auto\n");
+
/* --------------------------------------------------------------------------
Power Management
-------------------------------------------------------------------------- */
@@ -260,9 +266,8 @@ static void acpi_cstate_enter(struct acpi_processor_cx *cstate)
/*
* Some BIOS implementations switch to C3 in the published C2 state.
- * This seems to be a common problem on AMD boxen, but other vendors
- * are affected too. We pick the most conservative approach: we assume
- * that the local APIC stops in both C2 and C3.
+ * This seems to be a common problem on AMD boxen and Intel Dothan/Banias
+ * Pentium M machines.
*/
static void acpi_timer_check_state(int state, struct acpi_processor *pr,
struct acpi_processor_cx *cx)
@@ -276,8 +281,17 @@ static void acpi_timer_check_state(int state, struct acpi_processor *pr,
if (pwr->timer_broadcast_on_state < state)
return;
- if (cx->type >= ACPI_STATE_C2)
- pr->power.timer_broadcast_on_state = state;
+ if (cx->type >= ACPI_STATE_C2) {
+ if (boot_cpu_data.x86_vendor == X86_VENDOR_AMD)
+ pr->power.timer_broadcast_on_state = state;
+ else if ((boot_cpu_data.x86_vendor == X86_VENDOR_INTEL &&
+ boot_cpu_data.x86 == 6) &&
+ (boot_cpu_data.x86_model == 13 ||
+ boot_cpu_data.x86_model == 9))
+ {
+ pr->power.timer_broadcast_on_state = state;
+ }
+ }
}
static void acpi_propagate_timer_broadcast(struct acpi_processor *pr)
@@ -292,10 +306,16 @@ static void acpi_propagate_timer_broadcast(struct acpi_processor *pr)
#else
cpumask_t mask = cpumask_of_cpu(pr->id);
- if (pr->power.timer_broadcast_on_state < INT_MAX)
+ if (use_ipi == 0)
on_each_cpu(switch_APIC_timer_to_ipi, &mask, 1, 1);
- else
+ else if (use_ipi == 1)
on_each_cpu(switch_ipi_to_APIC_timer, &mask, 1, 1);
+ else {
+ if (pr->power.timer_broadcast_on_state < INT_MAX)
+ on_each_cpu(switch_APIC_timer_to_ipi, &mask, 1, 1);
+ else
+ on_each_cpu(switch_ipi_to_APIC_timer, &mask, 1, 1);
+ }
#endif
}
@@ -1013,13 +1033,13 @@ static int acpi_processor_power_verify(struct acpi_processor *pr)
case ACPI_STATE_C2:
acpi_processor_power_verify_c2(cx);
- if (cx->valid)
+ if (cx->valid && use_ipi != 0 && use_ipi != 1)
acpi_timer_check_state(i, pr, cx);
break;
case ACPI_STATE_C3:
acpi_processor_power_verify_c3(pr, cx);
- if (cx->valid)
+ if (cx->valid && use_ipi != 0 && use_ipi != 1)
acpi_timer_check_state(i, pr, cx);
break;
}
--
1.5.0.3.382.g34572
-
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
-------------------------------------------------------
[-- Attachment #2: [PATCH 7/9] ACPI: Only use IPI on known broken machines (AMD, Dothan/BaniasPentium M) --]
[-- Type: text/plain, Size: 4222 bytes --]
From: Thomas Renninger <trenn@suse.de>
Use IPI for blacklisted CPUs, add parameter IPI vs LAPIC
Currently, Linux disables lapic timer for all machines with C2 and higher
C-state support.
According to Intel only specific Intel models (Banias/Dothan) are broken
in respect of not waking up from C2 with lapic.
However, I am not sure about the naming of the parameter and how it
could/should get integrated into the dyntick part
(CONFIG_GENERIC_CLOCKEVENTS). There, a more fine grained check (TSC
still running?, ..) is needed? Does this make sense (always use
CLOCK_EVT_NOTIFY_BROADCAST_ON, but use OFF if forced by use_ipi=0:
clockevents_notify(use_ipi ? CLOCK_EVT_NOTIFY_BROADCAST_ON :
CLOCK_EVT_NOTIFY_BROADCAST_OFF, &pr->id);
Signed-off-by: Thomas Renninger <trenn@suse.de>
Signed-off-by: Len Brown <len.brown@intel.com>
---
drivers/acpi/processor_idle.c | 38 +++++++++++++++++++++++++++++---------
1 files changed, 29 insertions(+), 9 deletions(-)
diff --git a/drivers/acpi/processor_idle.c b/drivers/acpi/processor_idle.c
index 6077300..562124e 100644
--- a/drivers/acpi/processor_idle.c
+++ b/drivers/acpi/processor_idle.c
@@ -89,6 +89,12 @@ module_param(nocst, uint, 0000);
static unsigned int bm_history __read_mostly =
(HZ >= 800 ? 0xFFFFFFFF : ((1U << (HZ / 25)) - 1));
module_param(bm_history, uint, 0644);
+
+static unsigned use_ipi = 2;
+module_param(use_ipi, uint, 0644);
+MODULE_PARM_DESC(use_ipi, "IPI (vs. LAPIC) irqs for not waking up from C2/C3"
+ " machines. 0=apic, 1=ipi, 2=auto\n");
+
/* --------------------------------------------------------------------------
Power Management
-------------------------------------------------------------------------- */
@@ -260,9 +266,8 @@ static void acpi_cstate_enter(struct acpi_processor_cx *cstate)
/*
* Some BIOS implementations switch to C3 in the published C2 state.
- * This seems to be a common problem on AMD boxen, but other vendors
- * are affected too. We pick the most conservative approach: we assume
- * that the local APIC stops in both C2 and C3.
+ * This seems to be a common problem on AMD boxen and Intel Dothan/Banias
+ * Pentium M machines.
*/
static void acpi_timer_check_state(int state, struct acpi_processor *pr,
struct acpi_processor_cx *cx)
@@ -276,8 +281,17 @@ static void acpi_timer_check_state(int state, struct acpi_processor *pr,
if (pwr->timer_broadcast_on_state < state)
return;
- if (cx->type >= ACPI_STATE_C2)
- pr->power.timer_broadcast_on_state = state;
+ if (cx->type >= ACPI_STATE_C2) {
+ if (boot_cpu_data.x86_vendor == X86_VENDOR_AMD)
+ pr->power.timer_broadcast_on_state = state;
+ else if ((boot_cpu_data.x86_vendor == X86_VENDOR_INTEL &&
+ boot_cpu_data.x86 == 6) &&
+ (boot_cpu_data.x86_model == 13 ||
+ boot_cpu_data.x86_model == 9))
+ {
+ pr->power.timer_broadcast_on_state = state;
+ }
+ }
}
static void acpi_propagate_timer_broadcast(struct acpi_processor *pr)
@@ -292,10 +306,16 @@ static void acpi_propagate_timer_broadcast(struct acpi_processor *pr)
#else
cpumask_t mask = cpumask_of_cpu(pr->id);
- if (pr->power.timer_broadcast_on_state < INT_MAX)
+ if (use_ipi == 0)
on_each_cpu(switch_APIC_timer_to_ipi, &mask, 1, 1);
- else
+ else if (use_ipi == 1)
on_each_cpu(switch_ipi_to_APIC_timer, &mask, 1, 1);
+ else {
+ if (pr->power.timer_broadcast_on_state < INT_MAX)
+ on_each_cpu(switch_APIC_timer_to_ipi, &mask, 1, 1);
+ else
+ on_each_cpu(switch_ipi_to_APIC_timer, &mask, 1, 1);
+ }
#endif
}
@@ -1013,13 +1033,13 @@ static int acpi_processor_power_verify(struct acpi_processor *pr)
case ACPI_STATE_C2:
acpi_processor_power_verify_c2(cx);
- if (cx->valid)
+ if (cx->valid && use_ipi != 0 && use_ipi != 1)
acpi_timer_check_state(i, pr, cx);
break;
case ACPI_STATE_C3:
acpi_processor_power_verify_c3(pr, cx);
- if (cx->valid)
+ if (cx->valid && use_ipi != 0 && use_ipi != 1)
acpi_timer_check_state(i, pr, cx);
break;
}
--
1.5.0.3.382.g34572
-
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 3+ messages in thread* Re: Fwd: [PATCH 7/9] ACPI: Only use IPI on known broken machines (AMD, Dothan/BaniasPentium M)
2007-03-20 17:57 Fwd: [PATCH 7/9] ACPI: Only use IPI on known broken machines (AMD, Dothan/BaniasPentium M) Len Brown
@ 2007-03-20 19:23 ` Andi Kleen
2007-03-21 12:40 ` Thomas Gleixner
0 siblings, 1 reply; 3+ messages in thread
From: Andi Kleen @ 2007-03-20 19:23 UTC (permalink / raw)
To: Len Brown
Cc: tglx, Andrew Morton, Ingo Molnar, Andi Kleen, Thomas Renninger,
Linux Kernel Mailing List
> + else if ((boot_cpu_data.x86_vendor == X86_VENDOR_INTEL &&
> + boot_cpu_data.x86 == 6) &&
> + (boot_cpu_data.x86_model == 13 ||
> + boot_cpu_data.x86_model == 9))
What is with 10..12 and > 13 ? I would just force it for all model 6s that
have >= C2 and definitely for all with C3.
-Andi
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: Fwd: [PATCH 7/9] ACPI: Only use IPI on known broken machines (AMD, Dothan/BaniasPentium M)
2007-03-20 19:23 ` Andi Kleen
@ 2007-03-21 12:40 ` Thomas Gleixner
0 siblings, 0 replies; 3+ messages in thread
From: Thomas Gleixner @ 2007-03-21 12:40 UTC (permalink / raw)
To: Andi Kleen
Cc: Len Brown, Andrew Morton, Ingo Molnar, Thomas Renninger,
Linux Kernel Mailing List
On Tue, 2007-03-20 at 20:23 +0100, Andi Kleen wrote:
> > + else if ((boot_cpu_data.x86_vendor == X86_VENDOR_INTEL &&
> > + boot_cpu_data.x86 == 6) &&
> > + (boot_cpu_data.x86_model == 13 ||
> > + boot_cpu_data.x86_model == 9))
>
> What is with 10..12 and > 13 ? I would just force it for all model 6s that
> have >= C2 and definitely for all with C3.
C3 is unconditinally anyway.
tglx
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2007-03-21 12:33 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2007-03-20 17:57 Fwd: [PATCH 7/9] ACPI: Only use IPI on known broken machines (AMD, Dothan/BaniasPentium M) Len Brown
2007-03-20 19:23 ` Andi Kleen
2007-03-21 12:40 ` Thomas Gleixner
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).