Re: [ACPI] PATCH-ACPI based CPU hotplug[3/6]-Mapping lsapic to cpu

[Keiichiro Tokunaga]

On Wed, 22 Sep 2004 22:15:38 +0900 Keiichiro Tokunaga wrote:
> On Mon, 20 Sep 2004 09:38:19 -0700 Keshavamurthy Anil S wrote:
> > ---
> > Name:acpi_hotplug_arch.patch
> > Status: Tested on 2.6.9-rc2
> > Signed-off-by: Anil S Keshavamurthy <anil.s.keshavamurthy@xxxxxxxxx>
> > Depends:	
> > Version: applies on 2.6.9-rc2	
> > Description: 
> > This patch provides the architecture specifice support for mapping lsapic to cpu array.
> > Currently this supports just IA64. Support for IA32 and x86_64 is in progress
> > ---
> 
> I would like to suggest introducing a new function 'acpi_get_pxm()'
> since other drivers might need it in the future.  Acutally, ACPI container
> hotplug will be use it soon.

I have made a patch to modify your code to use acpi_get_pxm()
that I just posted earlier.  I hope it does not break your code:)
What do you think of it?

Thanks,
Keiichiro Tokunaga



Name: acpi_cpu_pxm_fix.patch
Status: Tested on 2.6.9-rc2
Signed-off-by: Keiichiro Tokunaga <tokunaga.keiich@xxxxxxxxxxxxxx>
Description:
Modify acpi_map_cpu2node() to use acpi_get_pxm().


---

 linux-2.6.9-rc2-fix-kei/arch/ia64/kernel/acpi.c |   31 +++++-------------------
 1 files changed, 7 insertions(+), 24 deletions(-)

diff -puN arch/ia64/kernel/acpi.c~acpi_cpu_pxm_fix arch/ia64/kernel/acpi.c
--- linux-2.6.9-rc2-fix/arch/ia64/kernel/acpi.c~acpi_cpu_pxm_fix	2004-09-22 22:20:53.797821495 +0900
+++ linux-2.6.9-rc2-fix-kei/arch/ia64/kernel/acpi.c	2004-09-22 22:20:53.800751203 +0900
@@ -659,36 +659,20 @@ int
 acpi_map_cpu2node(acpi_handle handle, int cpu, long physid)
 {
 #ifdef CONFIG_ACPI_NUMA
-	int 			pxm_id = 0;
-	union acpi_object 	*obj;
-	struct acpi_buffer 	buffer = {ACPI_ALLOCATE_BUFFER, NULL};
+	int 			pxm_id;
 
-	if (ACPI_FAILURE(acpi_evaluate_object(handle, "_PXM", NULL, &buffer)))
-		goto pxm_id_0;
+	pxm_id = acpi_get_pxm(handle);
 
-	if ((!buffer.length) || (!buffer.pointer))
-		goto pxm_id_0;
-
-	obj = buffer.pointer;
-	if (obj->type != ACPI_TYPE_INTEGER) {
-		acpi_os_free(buffer.pointer);
-		goto pxm_id_0;
-	}
-
-	pxm_id = obj->integer.value;
-
-pxm_id_0:
 	/*
 	 * Assuming that the container driver would have set the proximity
 	 * domain and would have initialized pxm_to_nid_map[pxm_id] && pxm_flag
 	 */
 
-	/* Return Error if proximity domain is not set */
-	if (!pxm_bit_test(pxm_id))
-		return -EINVAL;
-
 	node_cpuid[cpu].phys_id =  physid;
-	node_cpuid[cpu].nid = pxm_to_nid_map[pxm_id];
+	if (pxm_id < 0)
+		node_cpuid[cpu].nid = 0;
+	else
+		node_cpuid[cpu].nid = pxm_to_nid_map[pxm_id];
 
 #endif
 	return(0);
@@ -737,8 +721,7 @@ acpi_map_lsapic(acpi_handle handle, int 
 	if(cpu >= NR_CPUS)
 		return -EINVAL;
 
-	if (ACPI_FAILURE(acpi_map_cpu2node(handle, cpu, physid)))
-	return -ENODEV;
+	acpi_map_cpu2node(handle, cpu, physid);
 
  	cpu_set(cpu, cpu_present_map);
 	ia64_cpu_to_sapicid[cpu] = physid;

_
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/