ceph: fix incorrect kmalloc size of pagevec mempool
PATCH 1/1 ceph: fix incorrect kmalloc size of pagevec mempool
The kmalloc size of pagevec mempool is incorrectly calculated.
It misses the size of page pointer and only accounts the number for the array.
--- fs/ceph/super.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
fs/ceph/super.c +2 -1
@@ -961,7 +961,8 @@ static int __init init_caches(void)if (!ceph_mds_request_cachep) goto bad_mds_req; - ceph_wb_pagevec_pool =mempool_create_kmalloc_pool (10 ,CEPH_MAX_WRITE_SIZE >>PAGE_SHIFT ); + ceph_wb_pagevec_pool =mempool_create_kmalloc_pool (10 , + (CEPH_MAX_WRITE_SIZE >>PAGE_SHIFT ) *sizeof (struct page *));if (!ceph_wb_pagevec_pool) goto bad_pagevec_pool;
Signature · 1 line
2.25.1
Quoted 14 lines · ethanwu
The kmalloc size of pagevec mempool is incorrectly calculated.
It misses the size of page pointer and only accounts the number for the array.Fixes a0102bda5bc0 ("ceph: move sb->wb_pagevec_pool to be a global mempool")Signed-off-by ethanwu <[email protected]>--- fs/ceph/super.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)fs/ceph/super.c +0
-0
fs/ceph/super.c static int __init init_caches(void) · line 968
Quoted 9 lines
fs/ceph/super.c +2
-1@@ -961,7 +961,8 @@ static int __init init_caches(void)if (!ceph_mds_request_cachep) goto bad_mds_req; - ceph_wb_pagevec_pool =mempool_create_kmalloc_pool (10 ,CEPH_MAX_WRITE_SIZE >>PAGE_SHIFT ); + ceph_wb_pagevec_pool =mempool_create_kmalloc_pool (10 , + (CEPH_MAX_WRITE_SIZE >>PAGE_SHIFT ) *sizeof (struct page *));if (!ceph_wb_pagevec_pool) goto bad_pagevec_pool;
Good cache, LGTM.
Ilya Dryomov
maintainerQuoted 20 lines · Xiubo Li
Quoted 17 lines · ethanwu
The kmalloc size of pagevec mempool is incorrectly calculated.
It misses the size of page pointer and only accounts the number for the array.Fixes a0102bda5bc0 ("ceph: move sb->wb_pagevec_pool to be a global mempool")Signed-off-by ethanwu <[email protected]>--- fs/ceph/super.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)fs/ceph/super.c +0
-0@@ -961,7 +961,8 @@ static int __init init_caches(void)if (!ceph_mds_request_cachep) goto bad_mds_req;
Quoted 8 lines
fs/ceph/super.c static int __init init_caches(void) · line 968
Quoted 6 lines
fs/ceph/super.c +2
-1- ceph_wb_pagevec_pool =mempool_create_kmalloc_pool (10 ,CEPH_MAX_WRITE_SIZE >>PAGE_SHIFT ); + ceph_wb_pagevec_pool =mempool_create_kmalloc_pool (10 , + (CEPH_MAX_WRITE_SIZE >>PAGE_SHIFT ) *sizeof (struct page *));if (!ceph_wb_pagevec_pool) goto bad_pagevec_pool;Good cache, LGTM.
I don't think this matters in practice since 64M / 4K = 16K * 10 = 160K of memory reserved just for page pointers is already quite a lot...
However, since commit a0102bda5bc0 is talking about moving a pool to be global without adjusting its parameters, I agree that this fix is valid.
Thanks,
Ilya