Skip to contentSkip to conversation

ACPI: Set _PDC

BUG, REGRESSION Jan Pogadl 3 participants 5 messages Jun 16, 2010 – Jun 17, 2010 ACPI

lore
Share & tools
lore

Jan Pogadl

author
#0 lore raw

Hi,

with commit commit 5d554a7bb0643a6151a84319bfeba8270bf5269e (ACPI: processor:
add internal processor_physically_present()) a new function got introduced which always fails on machines with CONFIG_SMP disabled. Thus _PDC will never be initialized? set?.

The problem of the processor_physically_present() is, that it calls the acpi_get_cpuid(...) function which returns on NON-SMP machines always -1.

Note that i don't have any real knowledge what _PDC actualy realy do, all i gatherd so far is that its for switching C-Stats. I Stumbled across this by bisecting the problem, that the undervolting patches from linux-phc.org aren't working anymore with 2.6.34 (2.6.35-rc3). Whereas they worked fine with 2.6.33 on my Samsung X20 Notebook with a Pentium M precossor. The acpi-cpufreq is now using System-IO instead of MSR Registers.

Since i don't know what should be the correct behaviour here i can't provide any patches. But i doub't it's the correct behaviour of processor_physically_present() to always fail on non-SMP machines.

Deleting the function call, to this new function in early_init_pdc(...), and i get back the old behaviour.

cheers,
jan

#1 lore raw

Hi Jan,

* Jan Pogadl <[email protected]>:

Quoted 6 lines

Hi,

with commit commit 5d554a7bb0643a6151a84319bfeba8270bf5269e (ACPI: processor:
add internal processor_physically_present()) a new function got introduced which always fails on machines with CONFIG_SMP disabled. Thus _PDC will never be initialized? set?.

My apologies for the regression and the slow response. Please note my new email address.

Does this patch fix it?

---

drivers/acpi/processor_core.c +1 -1
index 5128435..bcf0cf8 100644
@@ -223,7 +223,7 @@ static bool processor_physically_present(acpi_handle handle)
 	type = (acpi_type == ACPI_TYPE_DEVICE) ? 1 : 0;
 	cpuid = acpi_get_cpuid(handle, type, acpi_id);
 
-	if (cpuid == -1)
+	if ((cpuid == -1) && (num_online_cpus() > 1))
 		return false;
 
 	return true;
author
#2 lore raw
Quoted 6 lines · On Thursday 17 June 2010 05:24:14 Alex Chiang wrote:
Quoted 4 lines

with commit commit 5d554a7bb0643a6151a84319bfeba8270bf5269e (ACPI: processor:
add internal processor_physically_present()) a new function got introduced which always fails on machines with CONFIG_SMP disabled. Thus _PDC will never be initialized? set?.

Does this patch fix it?

works great
thanks

Quoted 15 lines

---

drivers/acpi/processor_core.c +1 -1
index 5128435..bcf0cf8 100644
@@ -223,7 +223,7 @@ static bool processor_physically_present(acpi_handle handle)
 	type = (acpi_type == ACPI_TYPE_DEVICE) ? 1 : 0;
 	cpuid = acpi_get_cpuid(handle, type, acpi_id);
 
-	if (cpuid == -1)
+	if ((cpuid == -1) && (num_online_cpus() > 1))
 		return false;
 
 	return true;
#3 lore raw
Quoted 21 lines · Jan Pogadl
Quoted 6 lines · On Thursday 17 June 2010 05:24:14 Alex Chiang wrote:
Quoted 4 lines

with commit commit 5d554a7bb0643a6151a84319bfeba8270bf5269e (ACPI: processor:
add internal processor_physically_present()) a new function got introduced which always fails on machines with CONFIG_SMP disabled. Thus _PDC will never be initialized? set?.

Does this patch fix it?

works great
thanks

Quoted 9 lines

---

drivers/acpi/processor_core.c +0 -0
index 5128435..bcf0cf8 100644
@@ -223,7 +223,7 @@ static bool processor_physically_present(acpi_handle handle)
  	type = (acpi_type == ACPI_TYPE_DEVICE) ? 1 : 0;
  	cpuid = acpi_get_cpuid(handle, type, acpi_id);
 
Quoted 6 lines

drivers/acpi/processor_core.c static bool processor_physically_present(acpi_handle handle) · line 229

Quoted 6 lines
drivers/acpi/processor_core.c +1 -1
-	if (cpuid == -1)
+	if ((cpuid == -1)&&  (num_online_cpus()>  1))
  		return false;
 
  	return true;

I have a puzzle why num_online_cpus is used here, instead of num_possible_cpus. It will be possible under a hotplug scenario. --
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#4 lore raw

* Chen Gong <[email protected]>:

Quoted 9 lines
Quoted 9 lines · On Thursday 17 June 2010 05:24:14 Alex Chiang wrote:
Quoted 8 lines
drivers/acpi/processor_core.c +0 -0
index 5128435..bcf0cf8 100644
@@ -223,7 +223,7 @@ static bool processor_physically_present(acpi_handle handle)
 	type = (acpi_type == ACPI_TYPE_DEVICE) ? 1 : 0;
 	cpuid = acpi_get_cpuid(handle, type, acpi_id);
 
Quoted 8 lines
Quoted 6 lines

drivers/acpi/processor_core.c static bool processor_physically_present(acpi_handle handle) · line 229

Quoted 6 lines
drivers/acpi/processor_core.c +1 -1
-	if (cpuid == -1)
+	if ((cpuid == -1)&&  (num_online_cpus()>  1))
 		return false;
 
 	return true;

I have a puzzle why num_online_cpus is used here, instead of num_possible_cpus. It will be possible under a hotplug scenario.

Hm, you're probably right.

I was following the example of the other usage of acpi_get_cpuid() in acpi_processor_get_info(), but based on your observation, it's probably better to use num_possible_cpus.

I'll change it when I submit a proper patch.

Thanks for the review.

/ac