mm, sparse: do not waste pre allocated memmap space
Subsystems (2)
PATCH 1/1 mm, sparse: do not waste pre allocated memmap space
From: Michal Hocko <[email protected]>
Vincent has noticed [1] that there is something unusual with the memmap allocations going on on his platform
: I noticed this because on my ARM64 platform, with 1 GiB of memory the : first [and only] section is allocated from the zeroing path while with : 2 GiB of memory the first 1 GiB section is allocated from the : non-zeroing path.
The underlying problem is that although sparse_buffer_init allocates enough memory for all sections on the node sparse_buffer_alloc is not able to consume them due to mismatch in the expected allocation alignement. While sparse_buffer_init preallocation uses the PAGE_SIZE alignment the real memmap has to be aligned to section_map_size() this results in a wasted initial chunk of the preallocated memmap and unnecessary fallback allocation for a section.
While we are at it also change __populate_section_memmap to align to the requested size because at least VMEMMAP has constrains to have memmap properly aligned.
[1] http://lkml.kernel.org/r/[email protected]
--- mm/sparse.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-)
mm/sparse.c +7 -3
@@ -458,8 +458,7 @@ struct page __init *__populate_section_memmap(unsigned long pfn,if (map)return map; - map =memblock_alloc_try_nid (size, -PAGE_SIZE , addr, + map =memblock_alloc_try_nid (size, size, addr,MEMBLOCK_ALLOC_ACCESSIBLE , nid);if (!map) panic("%s: Failed to allocate %lu bytes align=0x%lx nid=%d from=%pa\n",
@@ -482,8 +481,13 @@ static void __init sparse_buffer_init(unsigned long size, int nid) {phys_addr_t addr =__pa (MAX_DMA_ADDRESS );WARN_ON (sparsemap_buf);/* forgot to call sparse_buffer_fini()? */ +/* +* Pre-allocated buffer is mainly used by __populate_section_memmap +* and we want it to be properly aligned to the section size - this is +* especially the case for VMEMMAP which maps memmap to PMDs +*/ sparsemap_buf = -memblock_alloc_try_nid_raw (size,PAGE_SIZE , +memblock_alloc_try_nid_raw (size,section_map_size (), addr,MEMBLOCK_ALLOC_ACCESSIBLE , nid); sparsemap_buf_end = sparsemap_buf + size;
Signature · 1 line
2.20.1
Quoted 47 lines · Michal Hocko
From: Michal Hocko <[email protected]>
Vincent has noticed [1] that there is something unusual with the memmap allocations going on on his platform
: I noticed this because on my ARM64 platform, with 1 GiB of memory the : first [and only] section is allocated from the zeroing path while with : 2 GiB of memory the first 1 GiB section is allocated from the : non-zeroing path.The underlying problem is that although
sparse_buffer_initallocates enough memory for all sections on the nodesparse_buffer_allocis not able to consume them due to mismatch in the expected allocation alignement. Whilesparse_buffer_initpreallocation uses thePAGE_SIZEalignment the real memmap has to be aligned tosection_map_size()this results in a wasted initial chunk of the preallocated memmap and unnecessary fallback allocation for a section.While we are at it also change
__populate_section_memmapto align to the requested size because at least VMEMMAP has constrains to have memmap properly aligned.[1] http://lkml.kernel.org/r/[email protected]
Reported-and-debugged-by Vincent Whitchurch <[email protected]>Fixes 35fd1eb1e821 ("mm/sparse: abstract sparse buffer allocations")Signed-off-by Michal Hocko <[email protected]>--- mm/sparse.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-)mm/sparse.c +1
-2@@ -458,8 +458,7 @@ struct page __init *__populate_section_memmap(unsigned long pfn,if (map)return map; - map =memblock_alloc_try_nid (size, -PAGE_SIZE , addr, + map =memblock_alloc_try_nid (size, size, addr,MEMBLOCK_ALLOC_ACCESSIBLE , nid);if (!map) panic("%s: Failed to allocate %lu bytes align=0x%lx nid=%d from=%pa\n",@@ -482,8 +481,13 @@ static void __init sparse_buffer_init(unsigned long size, int nid) {phys_addr_t addr =__pa (MAX_DMA_ADDRESS );WARN_ON (sparsemap_buf);/* forgot to call sparse_buffer_fini()? */
mm/sparse.c static void __init sparse_buffer_init(unsigned long size, int nid) · line 492
Quoted 10 lines
mm/sparse.c +6
-1+/* +* Pre-allocated buffer is mainly used by __populate_section_memmap +* and we want it to be properly aligned to the section size - this is +* especially the case for VMEMMAP which maps memmap to PMDs +*/ sparsemap_buf = -memblock_alloc_try_nid_raw (size,PAGE_SIZE , +memblock_alloc_try_nid_raw (size,section_map_size (), addr,MEMBLOCK_ALLOC_ACCESSIBLE , nid);
Wow, that alignment/layout gives me nightmares ^
None of your business, though :)
mm/sparse.c static void __init sparse_buffer_init(unsigned long size, int nid) · line 493
Quoted 2 lines
mm/sparse.c +0
-0sparsemap_buf_end = sparsemap_buf + size;
Signature · 3 lines
Thanks, David / dhildenb
Andrew Morton
maintainerQuoted 17 lines · On Tue, 19 Nov 2019 11:03:58 +0100 David Hildenbrand <[email protected]> wrote:
Quoted 4 lines
(quoted diff) +0
-0@@ -482,8 +481,13 @@ static void __init sparse_buffer_init(unsigned long size, int nid) { phys_addr_t addr = __pa(MAX_DMA_ADDRESS); WARN_ON(sparsemap_buf); /* forgot to call sparse_buffer_fini()? */Review static void __init sparse_buffer_init(unsigned long size, int nid) · line 492
Quoted 10 lines
(quoted diff) +6
-1+ /* + * Pre-allocated buffer is mainly used by __populate_section_memmap + * and we want it to be properly aligned to the section size - this is + * especially the case for VMEMMAP which maps memmap to PMDs + */ sparsemap_buf = - memblock_alloc_try_nid_raw(size, PAGE_SIZE, + memblock_alloc_try_nid_raw(size, section_map_size(), addr, MEMBLOCK_ALLOC_ACCESSIBLE, nid);Wow, that alignment/layout gives me nightmares ^
Quoted 1 line
None of your business, though :)
We're allowed to change it ;)
mm/sparse.c renamed from mm/sparse.c~mm-sparse-do-not-waste-pre-allocated-memmap-space-fix +2 -4
@@ -486,10 +486,8 @@ static void __init sparse_buffer_init(un* and we want it to be properly aligned to the section size - this is * especially the case for VMEMMAP which maps memmap to PMDs */ - sparsemap_buf = -memblock_alloc_try_nid_raw (size,section_map_size (), - addr, -MEMBLOCK_ALLOC_ACCESSIBLE , nid); + sparsemap_buf =memblock_alloc_try_nid_raw (size,section_map_size (), + addr,MEMBLOCK_ALLOC_ACCESSIBLE , nid); sparsemap_buf_end = sparsemap_buf + size; }
_
Quoted 27 lines · Andrew Morton
Quoted 17 lines · On Tue, 19 Nov 2019 11:03:58 +0100 David Hildenbrand <[email protected]> wrote:
Quoted 4 lines
(quoted diff) +0
-0@@ -482,8 +481,13 @@ static void __init sparse_buffer_init(unsigned long size, int nid) { phys_addr_t addr = __pa(MAX_DMA_ADDRESS); WARN_ON(sparsemap_buf); /* forgot to call sparse_buffer_fini()? */Review static void __init sparse_buffer_init(unsigned long size, int nid) · line 492
Quoted 10 lines
(quoted diff) +6
-1+ /* + * Pre-allocated buffer is mainly used by __populate_section_memmap + * and we want it to be properly aligned to the section size - this is + * especially the case for VMEMMAP which maps memmap to PMDs + */ sparsemap_buf = - memblock_alloc_try_nid_raw(size, PAGE_SIZE, + memblock_alloc_try_nid_raw(size, section_map_size(), addr, MEMBLOCK_ALLOC_ACCESSIBLE, nid);Wow, that alignment/layout gives me nightmares ^
Quoted 1 line
None of your business, though :)
We're allowed to change it ;)
mm/sparse.c renamed from mm/sparse.c~mm-sparse-do-not-waste-pre-allocated-memmap-space-fix +0
-0@@ -486,10 +486,8 @@ static void __init sparse_buffer_init(un* and we want it to be properly aligned to the section size - this is
mm/sparse.c static void __init sparse_buffer_init(un · line 492
Quoted 10 lines
mm/sparse.c renamed from mm/sparse.c~mm-sparse-do-not-waste-pre-allocated-memmap-space-fix +2
-4* especially the case for VMEMMAP which maps memmap to PMDs */ - sparsemap_buf = -memblock_alloc_try_nid_raw (size,section_map_size (), - addr, -MEMBLOCK_ALLOC_ACCESSIBLE , nid); + sparsemap_buf =memblock_alloc_try_nid_raw (size,section_map_size (), + addr,MEMBLOCK_ALLOC_ACCESSIBLE , nid); sparsemap_buf_end = sparsemap_buf + size; }
I didn't bother mostly because the creative code layout made the intention of the patch more obvious. But if it saves from nightmares then why not.
Signature · 2 lines
Michal Hocko SUSE Labs