Re: [v6 13/15] mm: stop zeroing memory during allocation in vmemmap

[Pasha Tatashin]

On 08/11/2017 09:04 AM, Michal Hocko wrote:

On Mon 07-08-17 16:38:47, Pavel Tatashin wrote:

Replace allocators in sprase-vmemmap to use the non-zeroing version. So,
we will get the performance improvement by zeroing the memory in parallel
when struct pages are zeroed.


First of all this should be probably merged with the previous patch. The
I think vmemmap_alloc_block would be better to split up into
__vmemmap_alloc_block which doesn't zero and vmemmap_alloc_block which
does zero which would reduce the memset callsites and it would make it
slightly more robust interface.


Ok, I will add: vmemmap_alloc_block_zero() call, and merge this and the 
previous patches together.

Re: [v6 13/15] mm: stop zeroing memory during allocation in vmemmap

[Pasha Tatashin]

On 08/11/2017 09:04 AM, Michal Hocko wrote:

On Mon 07-08-17 16:38:47, Pavel Tatashin wrote:

Replace allocators in sprase-vmemmap to use the non-zeroing version. So,
we will get the performance improvement by zeroing the memory in parallel
when struct pages are zeroed.


First of all this should be probably merged with the previous patch. The
I think vmemmap_alloc_block would be better to split up into
__vmemmap_alloc_block which doesn't zero and vmemmap_alloc_block which
does zero which would reduce the memset callsites and it would make it
slightly more robust interface.


Ok, I will add: vmemmap_alloc_block_zero() call, and merge this and the 
previous patches together.

Re: [v6 13/15] mm: stop zeroing memory during allocation in vmemmap

[Michal Hocko]

On Mon 07-08-17 16:38:47, Pavel Tatashin wrote:
> Replace allocators in sprase-vmemmap to use the non-zeroing version. So,
> we will get the performance improvement by zeroing the memory in parallel
> when struct pages are zeroed.

First of all this should be probably merged with the previous patch. The
I think vmemmap_alloc_block would be better to split up into
__vmemmap_alloc_block which doesn't zero and vmemmap_alloc_block which
does zero which would reduce the memset callsites and it would make it
slightly more robust interface.
 
> Signed-off-by: Pavel Tatashin 
> Reviewed-by: Steven Sistare 
> Reviewed-by: Daniel Jordan 
> Reviewed-by: Bob Picco 
> ---
>  mm/sparse-vmemmap.c | 6 +++---
>  mm/sparse.c | 6 +++---
>  2 files changed, 6 insertions(+), 6 deletions(-)
> 
> diff --git a/mm/sparse-vmemmap.c b/mm/sparse-vmemmap.c
> index d40c721ab19f..3b646b5ce1b6 100644
> --- a/mm/sparse-vmemmap.c
> +++ b/mm/sparse-vmemmap.c
> @@ -41,7 +41,7 @@ static void * __ref __earlyonly_bootmem_alloc(int node,
>   unsigned long align,
>   unsigned long goal)
>  {
> - return memblock_virt_alloc_try_nid(size, align, goal,
> + return memblock_virt_alloc_try_nid_raw(size, align, goal,
>   BOOTMEM_ALLOC_ACCESSIBLE, node);
>  }
>  
> @@ -56,11 +56,11 @@ void * __meminit vmemmap_alloc_block(unsigned long size, 
> int node)
>  
>   if (node_state(node, N_HIGH_MEMORY))
>   page = alloc_pages_node(
> - node, GFP_KERNEL | __GFP_ZERO | 
> __GFP_RETRY_MAYFAIL,
> + node, GFP_KERNEL | __GFP_RETRY_MAYFAIL,
>   get_order(size));
>   else
>   page = alloc_pages(
> - GFP_KERNEL | __GFP_ZERO | __GFP_RETRY_MAYFAIL,
> + GFP_KERNEL | __GFP_RETRY_MAYFAIL,
>   get_order(size));
>   if (page)
>   return page_address(page);
> diff --git a/mm/sparse.c b/mm/sparse.c
> index 7b4be3fd5cac..0e315766ad11 100644
> --- a/mm/sparse.c
> +++ b/mm/sparse.c
> @@ -441,9 +441,9 @@ void __init sparse_mem_maps_populate_node(struct page 
> **map_map,
>   }
>  
>   size = PAGE_ALIGN(size);
> - map = memblock_virt_alloc_try_nid(size * map_count,
> -   PAGE_SIZE, __pa(MAX_DMA_ADDRESS),
> -   BOOTMEM_ALLOC_ACCESSIBLE, nodeid);
> + map = memblock_virt_alloc_try_nid_raw(size * map_count,
> +   PAGE_SIZE, __pa(MAX_DMA_ADDRESS),
> +   BOOTMEM_ALLOC_ACCESSIBLE, nodeid);
>   if (map) {
>   for (pnum = pnum_begin; pnum < pnum_end; pnum++) {
>   if (!present_section_nr(pnum))
> -- 
> 2.14.0

-- 
Michal Hocko
SUSE Labs

Re: [v6 13/15] mm: stop zeroing memory during allocation in vmemmap

[Michal Hocko]

On Mon 07-08-17 16:38:47, Pavel Tatashin wrote:
> Replace allocators in sprase-vmemmap to use the non-zeroing version. So,
> we will get the performance improvement by zeroing the memory in parallel
> when struct pages are zeroed.

First of all this should be probably merged with the previous patch. The
I think vmemmap_alloc_block would be better to split up into
__vmemmap_alloc_block which doesn't zero and vmemmap_alloc_block which
does zero which would reduce the memset callsites and it would make it
slightly more robust interface.
 
> Signed-off-by: Pavel Tatashin 
> Reviewed-by: Steven Sistare 
> Reviewed-by: Daniel Jordan 
> Reviewed-by: Bob Picco 
> ---
>  mm/sparse-vmemmap.c | 6 +++---
>  mm/sparse.c | 6 +++---
>  2 files changed, 6 insertions(+), 6 deletions(-)
> 
> diff --git a/mm/sparse-vmemmap.c b/mm/sparse-vmemmap.c
> index d40c721ab19f..3b646b5ce1b6 100644
> --- a/mm/sparse-vmemmap.c
> +++ b/mm/sparse-vmemmap.c
> @@ -41,7 +41,7 @@ static void * __ref __earlyonly_bootmem_alloc(int node,
>   unsigned long align,
>   unsigned long goal)
>  {
> - return memblock_virt_alloc_try_nid(size, align, goal,
> + return memblock_virt_alloc_try_nid_raw(size, align, goal,
>   BOOTMEM_ALLOC_ACCESSIBLE, node);
>  }
>  
> @@ -56,11 +56,11 @@ void * __meminit vmemmap_alloc_block(unsigned long size, 
> int node)
>  
>   if (node_state(node, N_HIGH_MEMORY))
>   page = alloc_pages_node(
> - node, GFP_KERNEL | __GFP_ZERO | 
> __GFP_RETRY_MAYFAIL,
> + node, GFP_KERNEL | __GFP_RETRY_MAYFAIL,
>   get_order(size));
>   else
>   page = alloc_pages(
> - GFP_KERNEL | __GFP_ZERO | __GFP_RETRY_MAYFAIL,
> + GFP_KERNEL | __GFP_RETRY_MAYFAIL,
>   get_order(size));
>   if (page)
>   return page_address(page);
> diff --git a/mm/sparse.c b/mm/sparse.c
> index 7b4be3fd5cac..0e315766ad11 100644
> --- a/mm/sparse.c
> +++ b/mm/sparse.c
> @@ -441,9 +441,9 @@ void __init sparse_mem_maps_populate_node(struct page 
> **map_map,
>   }
>  
>   size = PAGE_ALIGN(size);
> - map = memblock_virt_alloc_try_nid(size * map_count,
> -   PAGE_SIZE, __pa(MAX_DMA_ADDRESS),
> -   BOOTMEM_ALLOC_ACCESSIBLE, nodeid);
> + map = memblock_virt_alloc_try_nid_raw(size * map_count,
> +   PAGE_SIZE, __pa(MAX_DMA_ADDRESS),
> +   BOOTMEM_ALLOC_ACCESSIBLE, nodeid);
>   if (map) {
>   for (pnum = pnum_begin; pnum < pnum_end; pnum++) {
>   if (!present_section_nr(pnum))
> -- 
> 2.14.0

-- 
Michal Hocko
SUSE Labs

[v6 13/15] mm: stop zeroing memory during allocation in vmemmap

[Pavel Tatashin]

Replace allocators in sprase-vmemmap to use the non-zeroing version. So,
we will get the performance improvement by zeroing the memory in parallel
when struct pages are zeroed.

Signed-off-by: Pavel Tatashin 
Reviewed-by: Steven Sistare 
Reviewed-by: Daniel Jordan 
Reviewed-by: Bob Picco 
---
 mm/sparse-vmemmap.c | 6 +++---
 mm/sparse.c | 6 +++---
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/mm/sparse-vmemmap.c b/mm/sparse-vmemmap.c
index d40c721ab19f..3b646b5ce1b6 100644
--- a/mm/sparse-vmemmap.c
+++ b/mm/sparse-vmemmap.c
@@ -41,7 +41,7 @@ static void * __ref __earlyonly_bootmem_alloc(int node,
unsigned long align,
unsigned long goal)
 {
-   return memblock_virt_alloc_try_nid(size, align, goal,
+   return memblock_virt_alloc_try_nid_raw(size, align, goal,
BOOTMEM_ALLOC_ACCESSIBLE, node);
 }
 
@@ -56,11 +56,11 @@ void * __meminit vmemmap_alloc_block(unsigned long size, 
int node)
 
if (node_state(node, N_HIGH_MEMORY))
page = alloc_pages_node(
-   node, GFP_KERNEL | __GFP_ZERO | 
__GFP_RETRY_MAYFAIL,
+   node, GFP_KERNEL | __GFP_RETRY_MAYFAIL,
get_order(size));
else
page = alloc_pages(
-   GFP_KERNEL | __GFP_ZERO | __GFP_RETRY_MAYFAIL,
+   GFP_KERNEL | __GFP_RETRY_MAYFAIL,
get_order(size));
if (page)
return page_address(page);
diff --git a/mm/sparse.c b/mm/sparse.c
index 7b4be3fd5cac..0e315766ad11 100644
--- a/mm/sparse.c
+++ b/mm/sparse.c
@@ -441,9 +441,9 @@ void __init sparse_mem_maps_populate_node(struct page 
**map_map,
}
 
size = PAGE_ALIGN(size);
-   map = memblock_virt_alloc_try_nid(size * map_count,
- PAGE_SIZE, __pa(MAX_DMA_ADDRESS),
- BOOTMEM_ALLOC_ACCESSIBLE, nodeid);
+   map = memblock_virt_alloc_try_nid_raw(size * map_count,
+ PAGE_SIZE, __pa(MAX_DMA_ADDRESS),
+ BOOTMEM_ALLOC_ACCESSIBLE, nodeid);
if (map) {
for (pnum = pnum_begin; pnum < pnum_end; pnum++) {
if (!present_section_nr(pnum))
-- 
2.14.0

[v6 13/15] mm: stop zeroing memory during allocation in vmemmap

[Pavel Tatashin]

Replace allocators in sprase-vmemmap to use the non-zeroing version. So,
we will get the performance improvement by zeroing the memory in parallel
when struct pages are zeroed.

Signed-off-by: Pavel Tatashin 
Reviewed-by: Steven Sistare 
Reviewed-by: Daniel Jordan 
Reviewed-by: Bob Picco 
---
 mm/sparse-vmemmap.c | 6 +++---
 mm/sparse.c | 6 +++---
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/mm/sparse-vmemmap.c b/mm/sparse-vmemmap.c
index d40c721ab19f..3b646b5ce1b6 100644
--- a/mm/sparse-vmemmap.c
+++ b/mm/sparse-vmemmap.c
@@ -41,7 +41,7 @@ static void * __ref __earlyonly_bootmem_alloc(int node,
unsigned long align,
unsigned long goal)
 {
-   return memblock_virt_alloc_try_nid(size, align, goal,
+   return memblock_virt_alloc_try_nid_raw(size, align, goal,
BOOTMEM_ALLOC_ACCESSIBLE, node);
 }
 
@@ -56,11 +56,11 @@ void * __meminit vmemmap_alloc_block(unsigned long size, 
int node)
 
if (node_state(node, N_HIGH_MEMORY))
page = alloc_pages_node(
-   node, GFP_KERNEL | __GFP_ZERO | 
__GFP_RETRY_MAYFAIL,
+   node, GFP_KERNEL | __GFP_RETRY_MAYFAIL,
get_order(size));
else
page = alloc_pages(
-   GFP_KERNEL | __GFP_ZERO | __GFP_RETRY_MAYFAIL,
+   GFP_KERNEL | __GFP_RETRY_MAYFAIL,
get_order(size));
if (page)
return page_address(page);
diff --git a/mm/sparse.c b/mm/sparse.c
index 7b4be3fd5cac..0e315766ad11 100644
--- a/mm/sparse.c
+++ b/mm/sparse.c
@@ -441,9 +441,9 @@ void __init sparse_mem_maps_populate_node(struct page 
**map_map,
}
 
size = PAGE_ALIGN(size);
-   map = memblock_virt_alloc_try_nid(size * map_count,
- PAGE_SIZE, __pa(MAX_DMA_ADDRESS),
- BOOTMEM_ALLOC_ACCESSIBLE, nodeid);
+   map = memblock_virt_alloc_try_nid_raw(size * map_count,
+ PAGE_SIZE, __pa(MAX_DMA_ADDRESS),
+ BOOTMEM_ALLOC_ACCESSIBLE, nodeid);
if (map) {
for (pnum = pnum_begin; pnum < pnum_end; pnum++) {
if (!present_section_nr(pnum))
-- 
2.14.0