subject:"\[Xen\-devel\] \[RFC PATCH v1 02\/21\] x86\: NUMA\: Refactor NUMA code"

Re: [Xen-devel] [RFC PATCH v1 02/21] x86: NUMA: Refactor NUMA code

2017-02-27 Thread Jan Beulich

>>> Vijay Kilari  02/27/17 12:43 PM >>>
>On Thu, Feb 9, 2017 at 9:41 PM, Jan Beulich  wrote:
> On 09.02.17 at 16:56,  wrote:
>>> +struct node_data node_data[MAX_NUMNODES];
>>> +
>>> +/* Mapping from pdx to node id */
>>> +int memnode_shift;
>>> +unsigned long memnodemapsize;
>>> +u8 *memnodemap;
>>> +typeof(*memnodemap) _memnodemap[64];
>>
>> Careful with removing any "static" please. For the last one in
>> particular you would need to change the name if it's really necessary
>> to make non-static. Even better would be though to keep it static
>> and provide suitable accessors.
>>
>> Also please sanitize types as you're moving stuff: memnode_shift
>> looks like it really wants to be unsigned int, and u8 should really
>> be uint8_t (as we're trying to phase out u8 & Co). This also applies
>> to code further down.
>
>You mean to change all occurrences of
>s/u8/uint8_t
>s/u32/uint32_t
>s/u64/uint64_t

Yes.

>Also, I see that xen/arch/x86/srat.c coding style is not adhere to xen
>coding style.
>Shall I clean up before I move the code?

If you want to take the time - sure. All I'd like to ask for is that at least 
the
file(s) you move the code _to_ end up with consistent style.

Jan


___
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel

Re: [Xen-devel] [RFC PATCH v1 02/21] x86: NUMA: Refactor NUMA code

2017-02-27 Thread Vijay Kilari

Hi Jan,

On Thu, Feb 9, 2017 at 9:41 PM, Jan Beulich  wrote:
 On 09.02.17 at 16:56,  wrote:
>> From: Vijaya Kumar K 
>>
>> Move common generic NUMA code to xen/common/numa.c from
>> xen/arch/x86/numa.c. Also move generic code in header file
>> xen/include/asm-x86/numa.h to xen/include/xen/numa.h
>>
>> This common code can be re-used later for ARM.
>>
>> Signed-off-by: Vijaya Kumar K 
>
> I would have been nice if you Cc-ed the maintainers of the code
> you're moving.
>
>> --- /dev/null
>> +++ b/xen/common/numa.c
>> @@ -0,0 +1,342 @@
>> +/*
>> + * Common NUMA handling functions for x86 and arm.
>> + * Original code extracted from arch/x86/numa.c
>> + *
>> + * This program is free software; you can redistribute it and/or modify
>> + * it under the terms of the GNU General Public License as published by
>> + * the Free Software Foundation; either version 2 of the License, or
>> + * (at your option) any later version.
>> + *
>> + * This program is distributed in the hope that it will be useful,
>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
>> + * GNU General Public License for more details.
>> + *
>> + * You should have received a copy of the GNU General Public License
>> + * along with this program; If not, see .
>> + */
>> +
>> +
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>
> This last one would better not be included here.
>
>> +struct node_data node_data[MAX_NUMNODES];
>> +
>> +/* Mapping from pdx to node id */
>> +int memnode_shift;
>> +unsigned long memnodemapsize;
>> +u8 *memnodemap;
>> +typeof(*memnodemap) _memnodemap[64];
>
> Careful with removing any "static" please. For the last one in
> particular you would need to change the name if it's really necessary
> to make non-static. Even better would be though to keep it static
> and provide suitable accessors.
>
> Also please sanitize types as you're moving stuff: memnode_shift
> looks like it really wants to be unsigned int, and u8 should really
> be uint8_t (as we're trying to phase out u8 & Co). This also applies
> to code further down.

You mean to change all occurrences of
s/u8/uint8_t
s/u32/uint32_t
s/u64/uint64_t

Also, I see that xen/arch/x86/srat.c coding style is not adhere to xen
coding style.
Shall I clean up before I move the code?

___
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel

Re: [Xen-devel] [RFC PATCH v1 02/21] x86: NUMA: Refactor NUMA code

2017-02-22 Thread Julien Grall


Hello Vijay,

On 22/02/17 10:04, Vijay Kilari wrote:

On Mon, Feb 20, 2017 at 6:07 PM, Julien Grall  wrote:



 {
 int rr, i;
@@ -288,16 +145,6 @@ void __init numa_initmem_init(unsigned long
start_pfn, unsigned long end_pfn)
 (u64)end_pfn << PAGE_SHIFT);
 }

-void numa_add_cpu(int cpu)
-{
-cpumask_set_cpu(cpu, _to_cpumask[cpu_to_node(cpu)]);
-}
-
-void numa_set_node(int cpu, nodeid_t node)
-{
-cpu_to_node[cpu] = node;
-}
-
 /* [numa=off] */
 static __init int numa_setup(char *opt)



Same question here. Everything in numa_setup and the fake NUMA looks
valid for ARM too.


numa_setup() is taken in separate patch. fake numa case is not considered.
for x86 it is under separate config CONFIG_NUMA_EMU.


Why fake numa is not considered? This would help to test the series on 
non-NUMA platform.




[]


diff --git a/xen/common/Makefile b/xen/common/Makefile
index 0fed30b..c1bd2ff 100644
--- a/xen/common/Makefile
+++ b/xen/common/Makefile
@@ -63,6 +63,7 @@ obj-y += wait.o
 obj-bin-y += warning.init.o
 obj-$(CONFIG_XENOPROF) += xenoprof.o
 obj-y += xmalloc_tlsf.o
+obj-y += numa.o



This needs to be gated with CONFIG_NUMA.


As it is shared with x86 and prior this changes it is not gated under
any config for x86.


Because x86 code assume that CONFIG_NUMA is always enabled. If you look 
at the .config CONFIG_NUMA will be set so you can gate it here.



Looking at this comment, it looks like the NUMA support should depend on
HAS_PDX as this is not something that may be able on all the architecture.


yes it uses pfn_to_pdx() while updating memnodemap.
May be comment should be suffice.


A comment may be missed, gated in the Kconfig will prevent a new 
architecture to use NUMA without knowing PDX is necessary.


[...]


+#endif /* CONFIG_NUMA */
+
+extern void numa_add_cpu(int cpu);
+extern nodeid_t setup_node(unsigned int pxm);
+extern void numa_set_node(int cpu, nodeid_t node);
+extern void setup_node_bootmem(nodeid_t nodeid, u64 start, u64 end);



I am not sure to understand why those function are not protected by #ifdef
CONFIFG_NUMA.

As these defined in xen/common/numa.c which is not under CONFIG_NUMA,
I have kept them outside CONFIG_NUMA


xen/common/numa.c should be under CONFIG_NUMA. There is no reason to 
expose the code on non-NUMA platform.


Regards,

--
Julien Grall

___
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel

Re: [Xen-devel] [RFC PATCH v1 02/21] x86: NUMA: Refactor NUMA code

2017-02-22 Thread Vijay Kilari

On Mon, Feb 20, 2017 at 6:07 PM, Julien Grall  wrote:
> Hello Vijay,
>
>
> On 09/02/17 15:56, vijay.kil...@gmail.com wrote:
>>
>> From: Vijaya Kumar K 
>>
>> Move common generic NUMA code to xen/common/numa.c from
>> xen/arch/x86/numa.c. Also move generic code in header file
>> xen/include/asm-x86/numa.h to xen/include/xen/numa.h
>>
>> This common code can be re-used later for ARM.
>>
>> Signed-off-by: Vijaya Kumar K 
>> ---
>>  xen/arch/x86/numa.c| 299 ---
>>  xen/common/Makefile|   1 +
>>  xen/common/numa.c  | 342
>> +
>>  xen/include/asm-x86/numa.h |  47 ---
>>  xen/include/xen/numa.h |  54 +++
>>  5 files changed, 397 insertions(+), 346 deletions(-)
>>
>> diff --git a/xen/arch/x86/numa.c b/xen/arch/x86/numa.c
>> index 6f4d438..bc787e0 100644
>> --- a/xen/arch/x86/numa.c
>> +++ b/xen/arch/x86/numa.c
>> @@ -25,27 +25,12 @@ custom_param("numa", numa_setup);
>>  #define Dprintk(x...)
>>  #endif
>>
>> -/* from proto.h */
>> -#define round_up(x,y) x)+(y))-1) & (~((y)-1)))
>> -
>> -struct node_data node_data[MAX_NUMNODES];
>> -
>> -/* Mapping from pdx to node id */
>> -int memnode_shift;
>> -static typeof(*memnodemap) _memnodemap[64];
>> -unsigned long memnodemapsize;
>> -u8 *memnodemap;
>> -
>> -nodeid_t cpu_to_node[NR_CPUS] __read_mostly = {
>> -[0 ... NR_CPUS-1] = NUMA_NO_NODE
>> -};
>>  /*
>>   * Keep BIOS's CPU2node information, should not be used for memory
>> allocaion
>>   */
>>  nodeid_t apicid_to_node[MAX_LOCAL_APIC] = {
>>  [0 ... MAX_LOCAL_APIC-1] = NUMA_NO_NODE
>>  };
>> -cpumask_t node_to_cpumask[MAX_NUMNODES] __read_mostly;
>>
>>  nodemask_t __read_mostly node_online_map = { { [0] = 1UL } };
>
>
> Why this variable is not moved?
Ok. Can be moved.
>
> [...]
>
>>  void __init numa_init_array(void)
>
>
> Same question for this function.

Initially I was suspicious about the comment in this function and thought
it might be valid of x86 alone. But looks like it is generic.
I will have a look.

>
>>  {
>>  int rr, i;
>> @@ -288,16 +145,6 @@ void __init numa_initmem_init(unsigned long
>> start_pfn, unsigned long end_pfn)
>>  (u64)end_pfn << PAGE_SHIFT);
>>  }
>>
>> -void numa_add_cpu(int cpu)
>> -{
>> -cpumask_set_cpu(cpu, _to_cpumask[cpu_to_node(cpu)]);
>> -}
>> -
>> -void numa_set_node(int cpu, nodeid_t node)
>> -{
>> -cpu_to_node[cpu] = node;
>> -}
>> -
>>  /* [numa=off] */
>>  static __init int numa_setup(char *opt)
>
>
> Same question here. Everything in numa_setup and the fake NUMA looks
> valid for ARM too.

numa_setup() is taken in separate patch. fake numa case is not considered.
for x86 it is under separate config CONFIG_NUMA_EMU.

>
> []
>
>> diff --git a/xen/common/Makefile b/xen/common/Makefile
>> index 0fed30b..c1bd2ff 100644
>> --- a/xen/common/Makefile
>> +++ b/xen/common/Makefile
>> @@ -63,6 +63,7 @@ obj-y += wait.o
>>  obj-bin-y += warning.init.o
>>  obj-$(CONFIG_XENOPROF) += xenoprof.o
>>  obj-y += xmalloc_tlsf.o
>> +obj-y += numa.o
>
>
> This needs to be gated with CONFIG_NUMA.

As it is shared with x86 and prior this changes it is not gated under
any config for x86.

>
>>
>>  obj-bin-$(CONFIG_X86) += $(foreach n,decompress bunzip2 unxz unlzma unlzo
>> unlz4 earlycpio,$(n).init.o)
>>
>> diff --git a/xen/common/numa.c b/xen/common/numa.c
>> new file mode 100644
>> index 000..59dcb63
>> --- /dev/null
>> +++ b/xen/common/numa.c
>> @@ -0,0 +1,342 @@
>> +/*
>> + * Common NUMA handling functions for x86 and arm.
>> + * Original code extracted from arch/x86/numa.c
>> + *
>> + * This program is free software; you can redistribute it and/or modify
>> + * it under the terms of the GNU General Public License as published by
>> + * the Free Software Foundation; either version 2 of the License, or
>> + * (at your option) any later version.
>
>
> Xen is GPLv2 only. Please update to:
>
> "This program is free software; you can redistribute it and/or
> modify it under the terms and conditions of the GNU General Public
> License, version 2, as published by the Free Software Foundation."
>
>
>> + *
>> + * This program is distributed in the hope that it will be useful,
>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
>> + * GNU General Public License for more details.
>> + *
>> + * You should have received a copy of the GNU General Public License
>> + * along with this program; If not, see .
>> + */
>> +
>> +
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +#include 
>> +
>> +struct node_data node_data[MAX_NUMNODES];
>> +
>> +/* Mapping from pdx to node id */
>
>
> Looking at this comment, it looks like the NUMA

Re: [Xen-devel] [RFC PATCH v1 02/21] x86: NUMA: Refactor NUMA code

2017-02-20 Thread Julien Grall


Hello Vijay,

On 09/02/17 15:56, vijay.kil...@gmail.com wrote:

From: Vijaya Kumar K 

Move common generic NUMA code to xen/common/numa.c from
xen/arch/x86/numa.c. Also move generic code in header file
xen/include/asm-x86/numa.h to xen/include/xen/numa.h

This common code can be re-used later for ARM.

Signed-off-by: Vijaya Kumar K 
---
 xen/arch/x86/numa.c| 299 ---
 xen/common/Makefile|   1 +
 xen/common/numa.c  | 342 +
 xen/include/asm-x86/numa.h |  47 ---
 xen/include/xen/numa.h |  54 +++
 5 files changed, 397 insertions(+), 346 deletions(-)

diff --git a/xen/arch/x86/numa.c b/xen/arch/x86/numa.c
index 6f4d438..bc787e0 100644
--- a/xen/arch/x86/numa.c
+++ b/xen/arch/x86/numa.c
@@ -25,27 +25,12 @@ custom_param("numa", numa_setup);
 #define Dprintk(x...)
 #endif

-/* from proto.h */
-#define round_up(x,y) x)+(y))-1) & (~((y)-1)))
-
-struct node_data node_data[MAX_NUMNODES];
-
-/* Mapping from pdx to node id */
-int memnode_shift;
-static typeof(*memnodemap) _memnodemap[64];
-unsigned long memnodemapsize;
-u8 *memnodemap;
-
-nodeid_t cpu_to_node[NR_CPUS] __read_mostly = {
-[0 ... NR_CPUS-1] = NUMA_NO_NODE
-};
 /*
  * Keep BIOS's CPU2node information, should not be used for memory allocaion
  */
 nodeid_t apicid_to_node[MAX_LOCAL_APIC] = {
 [0 ... MAX_LOCAL_APIC-1] = NUMA_NO_NODE
 };
-cpumask_t node_to_cpumask[MAX_NUMNODES] __read_mostly;

 nodemask_t __read_mostly node_online_map = { { [0] = 1UL } };


Why this variable is not moved?

[...]


 void __init numa_init_array(void)


Same question for this function.


 {
 int rr, i;
@@ -288,16 +145,6 @@ void __init numa_initmem_init(unsigned long start_pfn, 
unsigned long end_pfn)
 (u64)end_pfn << PAGE_SHIFT);
 }

-void numa_add_cpu(int cpu)
-{
-cpumask_set_cpu(cpu, _to_cpumask[cpu_to_node(cpu)]);
-}
-
-void numa_set_node(int cpu, nodeid_t node)
-{
-cpu_to_node[cpu] = node;
-}
-
 /* [numa=off] */
 static __init int numa_setup(char *opt)


Same question here. Everything in numa_setup and the fake NUMA looks
valid for ARM too.

[]


diff --git a/xen/common/Makefile b/xen/common/Makefile
index 0fed30b..c1bd2ff 100644
--- a/xen/common/Makefile
+++ b/xen/common/Makefile
@@ -63,6 +63,7 @@ obj-y += wait.o
 obj-bin-y += warning.init.o
 obj-$(CONFIG_XENOPROF) += xenoprof.o
 obj-y += xmalloc_tlsf.o
+obj-y += numa.o


This needs to be gated with CONFIG_NUMA.



 obj-bin-$(CONFIG_X86) += $(foreach n,decompress bunzip2 unxz unlzma unlzo 
unlz4 earlycpio,$(n).init.o)

diff --git a/xen/common/numa.c b/xen/common/numa.c
new file mode 100644
index 000..59dcb63
--- /dev/null
+++ b/xen/common/numa.c
@@ -0,0 +1,342 @@
+/*
+ * Common NUMA handling functions for x86 and arm.
+ * Original code extracted from arch/x86/numa.c
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.


Xen is GPLv2 only. Please update to:

"This program is free software; you can redistribute it and/or
modify it under the terms and conditions of the GNU General Public
License, version 2, as published by the Free Software Foundation."


+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; If not, see .
+ */
+
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+
+struct node_data node_data[MAX_NUMNODES];
+
+/* Mapping from pdx to node id */


Looking at this comment, it looks like the NUMA support should depend on 
HAS_PDX as this is not something that may be able on all the architecture.



+int memnode_shift;
+unsigned long memnodemapsize;
+u8 *memnodemap;
+typeof(*memnodemap) _memnodemap[64];
+
+nodeid_t cpu_to_node[NR_CPUS] __read_mostly = {
+[0 ... NR_CPUS-1] = NUMA_NO_NODE
+};
+
+cpumask_t node_to_cpumask[MAX_NUMNODES] __read_mostly;


[...]


+void numa_add_cpu(int cpu)
+{
+cpumask_set_cpu(cpu, _to_cpumask[cpu_to_node(cpu)]);
+}
+
+void numa_set_node(int cpu, nodeid_t node)
+{
+cpu_to_node[cpu] = node;
+}
+
+EXPORT_SYMBOL(node_to_cpumask);
+EXPORT_SYMBOL(memnode_shift);
+EXPORT_SYMBOL(memnodemap);
+EXPORT_SYMBOL(node_data);


Those 4 lines are not part of the original code. Why did you add them?

To ease review I would like to see this patch split multiple one:
- multiple small to prepare the code (export function, change the 
type...)

Re: [Xen-devel] [RFC PATCH v1 02/21] x86: NUMA: Refactor NUMA code

2017-02-20 Thread Julien Grall


Hi,

On 09/02/17 16:11, Jan Beulich wrote:

On 09.02.17 at 16:56,  wrote:

From: Vijaya Kumar K 

Move common generic NUMA code to xen/common/numa.c from
xen/arch/x86/numa.c. Also move generic code in header file
xen/include/asm-x86/numa.h to xen/include/xen/numa.h

This common code can be re-used later for ARM.

Signed-off-by: Vijaya Kumar K 


I would have been nice if you Cc-ed the maintainers of the code
you're moving.


--- /dev/null
+++ b/xen/common/numa.c
@@ -0,0 +1,342 @@
+/*
+ * Common NUMA handling functions for x86 and arm.
+ * Original code extracted from arch/x86/numa.c
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; If not, see .
+ */
+
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 


This last one would better not be included here.


+struct node_data node_data[MAX_NUMNODES];
+
+/* Mapping from pdx to node id */
+int memnode_shift;
+unsigned long memnodemapsize;
+u8 *memnodemap;
+typeof(*memnodemap) _memnodemap[64];


Careful with removing any "static" please. For the last one in
particular you would need to change the name if it's really necessary
to make non-static. Even better would be though to keep it static
and provide suitable accessors.

Also please sanitize types as you're moving stuff: memnode_shift
looks like it really wants to be unsigned int, and u8 should really
be uint8_t (as we're trying to phase out u8 & Co). This also applies
to code further down.


I agree with the changes suggested. If possible I would prefer all those 
changes made in separate patches before the move. This would ease the 
review.


Cheers,

--
Julien Grall

___
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel

Re: [Xen-devel] [RFC PATCH v1 02/21] x86: NUMA: Refactor NUMA code

2017-02-09 Thread Jan Beulich

>>> On 09.02.17 at 16:56,  wrote:
> From: Vijaya Kumar K 
> 
> Move common generic NUMA code to xen/common/numa.c from
> xen/arch/x86/numa.c. Also move generic code in header file
> xen/include/asm-x86/numa.h to xen/include/xen/numa.h
> 
> This common code can be re-used later for ARM.
> 
> Signed-off-by: Vijaya Kumar K 

I would have been nice if you Cc-ed the maintainers of the code
you're moving.

> --- /dev/null
> +++ b/xen/common/numa.c
> @@ -0,0 +1,342 @@
> +/*
> + * Common NUMA handling functions for x86 and arm.
> + * Original code extracted from arch/x86/numa.c
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation; either version 2 of the License, or
> + * (at your option) any later version.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program; If not, see .
> + */
> +
> +
> +#include 
> +#include 
> +#include 
> +#include 
> +#include 
> +#include 
> +#include 
> +#include 
> +#include 
> +#include 
> +#include 
> +#include 
> +#include 
> +#include 

This last one would better not be included here.

> +struct node_data node_data[MAX_NUMNODES];
> +
> +/* Mapping from pdx to node id */
> +int memnode_shift;
> +unsigned long memnodemapsize;
> +u8 *memnodemap;
> +typeof(*memnodemap) _memnodemap[64];

Careful with removing any "static" please. For the last one in
particular you would need to change the name if it's really necessary
to make non-static. Even better would be though to keep it static
and provide suitable accessors.

Also please sanitize types as you're moving stuff: memnode_shift
looks like it really wants to be unsigned int, and u8 should really
be uint8_t (as we're trying to phase out u8 & Co). This also applies
to code further down.

Jan


___
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel

[Xen-devel] [RFC PATCH v1 02/21] x86: NUMA: Refactor NUMA code

2017-02-09 Thread vijay . kilari

From: Vijaya Kumar K 

Move common generic NUMA code to xen/common/numa.c from
xen/arch/x86/numa.c. Also move generic code in header file
xen/include/asm-x86/numa.h to xen/include/xen/numa.h

This common code can be re-used later for ARM.

Signed-off-by: Vijaya Kumar K 
---
 xen/arch/x86/numa.c| 299 ---
 xen/common/Makefile|   1 +
 xen/common/numa.c  | 342 +
 xen/include/asm-x86/numa.h |  47 ---
 xen/include/xen/numa.h |  54 +++
 5 files changed, 397 insertions(+), 346 deletions(-)

diff --git a/xen/arch/x86/numa.c b/xen/arch/x86/numa.c
index 6f4d438..bc787e0 100644
--- a/xen/arch/x86/numa.c
+++ b/xen/arch/x86/numa.c
@@ -25,27 +25,12 @@ custom_param("numa", numa_setup);
 #define Dprintk(x...)
 #endif
 
-/* from proto.h */
-#define round_up(x,y) x)+(y))-1) & (~((y)-1)))
-
-struct node_data node_data[MAX_NUMNODES];
-
-/* Mapping from pdx to node id */
-int memnode_shift;
-static typeof(*memnodemap) _memnodemap[64];
-unsigned long memnodemapsize;
-u8 *memnodemap;
-
-nodeid_t cpu_to_node[NR_CPUS] __read_mostly = {
-[0 ... NR_CPUS-1] = NUMA_NO_NODE
-};
 /*
  * Keep BIOS's CPU2node information, should not be used for memory allocaion
  */
 nodeid_t apicid_to_node[MAX_LOCAL_APIC] = {
 [0 ... MAX_LOCAL_APIC-1] = NUMA_NO_NODE
 };
-cpumask_t node_to_cpumask[MAX_NUMNODES] __read_mostly;
 
 nodemask_t __read_mostly node_online_map = { { [0] = 1UL } };
 
@@ -57,134 +42,6 @@ int srat_disabled(void)
 return numa_off || acpi_numa < 0;
 }
 
-/*
- * Given a shift value, try to populate memnodemap[]
- * Returns :
- * 1 if OK
- * 0 if memnodmap[] too small (of shift too small)
- * -1 if node overlap or lost ram (shift too big)
- */
-static int __init populate_memnodemap(const struct node *nodes,
-  int numnodes, int shift, nodeid_t 
*nodeids)
-{
-unsigned long spdx, epdx;
-int i, res = -1;
-
-memset(memnodemap, NUMA_NO_NODE, memnodemapsize * sizeof(*memnodemap));
-for ( i = 0; i < numnodes; i++ )
-{
-spdx = paddr_to_pdx(nodes[i].start);
-epdx = paddr_to_pdx(nodes[i].end - 1) + 1;
-if ( spdx >= epdx )
-continue;
-if ( (epdx >> shift) >= memnodemapsize )
-return 0;
-do {
-if ( memnodemap[spdx >> shift] != NUMA_NO_NODE )
-return -1;
-
-if ( !nodeids )
-memnodemap[spdx >> shift] = i;
-else
-memnodemap[spdx >> shift] = nodeids[i];
-
-spdx += (1UL << shift);
-} while ( spdx < epdx );
-res = 1;
-}
-
-return res;
-}
-
-static int __init allocate_cachealigned_memnodemap(void)
-{
-unsigned long size = PFN_UP(memnodemapsize * sizeof(*memnodemap));
-unsigned long mfn = alloc_boot_pages(size, 1);
-
-if ( !mfn )
-{
-printk(KERN_ERR
-   "NUMA: Unable to allocate Memory to Node hash map\n");
-memnodemapsize = 0;
-return -1;
-}
-
-memnodemap = mfn_to_virt(mfn);
-mfn <<= PAGE_SHIFT;
-size <<= PAGE_SHIFT;
-printk(KERN_DEBUG "NUMA: Allocated memnodemap from %lx - %lx\n",
-   mfn, mfn + size);
-memnodemapsize = size / sizeof(*memnodemap);
-
-return 0;
-}
-
-/*
- * The LSB of all start and end addresses in the node map is the value of the
- * maximum possible shift.
- */
-static int __init extract_lsb_from_nodes(const struct node *nodes,
- int numnodes)
-{
-int i, nodes_used = 0;
-unsigned long spdx, epdx;
-unsigned long bitfield = 0, memtop = 0;
-
-for ( i = 0; i < numnodes; i++ )
-{
-spdx = paddr_to_pdx(nodes[i].start);
-epdx = paddr_to_pdx(nodes[i].end - 1) + 1;
-if ( spdx >= epdx )
-continue;
-bitfield |= spdx;
-nodes_used++;
-if ( epdx > memtop )
-memtop = epdx;
-}
-if ( nodes_used <= 1 )
-i = BITS_PER_LONG - 1;
-else
-i = find_first_bit(, sizeof(unsigned long)*8);
-memnodemapsize = (memtop >> i) + 1;
-return i;
-}
-
-int __init compute_hash_shift(struct node *nodes, int numnodes,
-  nodeid_t *nodeids)
-{
-int shift;
-
-shift = extract_lsb_from_nodes(nodes, numnodes);
-if ( memnodemapsize <= ARRAY_SIZE(_memnodemap) )
-memnodemap = _memnodemap;
-else if ( allocate_cachealigned_memnodemap() )
-return -1;
-printk(KERN_DEBUG "NUMA: Using %d for the hash shift.\n", shift);
-
-if ( populate_memnodemap(nodes, numnodes, shift, nodeids) != 1 )
-{
-printk(KERN_INFO "Your memory is not aligned you need to "
-   "rebuild your hypervisor with a bigger NODEMAPSIZE "
-   "shift=%d\n", shift);
-return -1;
-}
-
-return shift;
-}
-/* initialize NODE_DATA given nodeid and

Re: [Xen-devel] [RFC PATCH v1 02/21] x86: NUMA: Refactor NUMA code

Re: [Xen-devel] [RFC PATCH v1 02/21] x86: NUMA: Refactor NUMA code

Re: [Xen-devel] [RFC PATCH v1 02/21] x86: NUMA: Refactor NUMA code

Re: [Xen-devel] [RFC PATCH v1 02/21] x86: NUMA: Refactor NUMA code

Re: [Xen-devel] [RFC PATCH v1 02/21] x86: NUMA: Refactor NUMA code

Re: [Xen-devel] [RFC PATCH v1 02/21] x86: NUMA: Refactor NUMA code

Re: [Xen-devel] [RFC PATCH v1 02/21] x86: NUMA: Refactor NUMA code

[Xen-devel] [RFC PATCH v1 02/21] x86: NUMA: Refactor NUMA code

8 matches

Site Navigation

Mail list logo

Footer information