summaryrefslogtreecommitdiffstats
path: root/fs/nfs
diff options
context:
space:
mode:
authorNeilBrown <neilb@suse.com>2017-04-10 04:22:09 +0200
committerTrond Myklebust <trond.myklebust@primarydata.com>2017-04-20 19:44:05 +0200
commit518662e0fcb9fa241fe90a337b59bc5066b2a930 (patch)
tree828d198fc53348eb87734bd689abe12cb0679287 /fs/nfs
parentNFS: Clean up nfs4_proc_get_lease_time() (diff)
downloadlinux-518662e0fcb9fa241fe90a337b59bc5066b2a930.tar.xz
linux-518662e0fcb9fa241fe90a337b59bc5066b2a930.zip
NFS: fix usage of mempools.
When passed GFP flags that allow sleeping (such as GFP_NOIO), mempool_alloc() will never return NULL, it will wait until memory is available. This means that we don't need to handle failure, but that we do need to ensure one thread doesn't call mempool_alloc() twice on the one pool without queuing or freeing the first allocation. If multiple threads did this during times of high memory pressure, the pool could be exhausted and a deadlock could result. pnfs_generic_alloc_ds_commits() attempts to allocate from the nfs_commit_mempool while already holding an allocation from that pool. This is not safe. So change nfs_commitdata_alloc() to take a flag that indicates whether failure is acceptable. In pnfs_generic_alloc_ds_commits(), accept failure and handle it as we currently do. Else where, do not accept failure, and do not handle it. Even when failure is acceptable, we want to succeed if possible. That means both - using an entry from the pool if there is one - waiting for direct reclaim is there isn't. We call mempool_alloc(GFP_NOWAIT) to achieve the first, then kmem_cache_alloc(GFP_NOIO|__GFP_NORETRY) to achieve the second. Each of these can fail, but together they do the best they can without blocking indefinitely. The objects returned by kmem_cache_alloc() will still be freed by mempool_free(). This is safe as mempool_alloc() uses exactly the same function to allocate objects (since the mempool was created with mempool_create_slab_pool()). The object returned by mempool_alloc() and kmem_cache_alloc() are indistinguishable so mempool_free() will handle both identically, either adding to the pool or calling kmem_cache_free(). Also, don't test for failure when allocating from nfs_wdata_mempool. Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
Diffstat (limited to 'fs/nfs')
-rw-r--r--fs/nfs/pnfs_nfs.c16
-rw-r--r--fs/nfs/write.c35
2 files changed, 26 insertions, 25 deletions
diff --git a/fs/nfs/pnfs_nfs.c b/fs/nfs/pnfs_nfs.c
index 7250b95549ec..1edf5b84aba5 100644
--- a/fs/nfs/pnfs_nfs.c
+++ b/fs/nfs/pnfs_nfs.c
@@ -217,7 +217,7 @@ pnfs_generic_alloc_ds_commits(struct nfs_commit_info *cinfo,
for (i = 0; i < fl_cinfo->nbuckets; i++, bucket++) {
if (list_empty(&bucket->committing))
continue;
- data = nfs_commitdata_alloc();
+ data = nfs_commitdata_alloc(false);
if (!data)
break;
data->ds_commit_index = i;
@@ -283,16 +283,10 @@ pnfs_generic_commit_pagelist(struct inode *inode, struct list_head *mds_pages,
unsigned int nreq = 0;
if (!list_empty(mds_pages)) {
- data = nfs_commitdata_alloc();
- if (data != NULL) {
- data->ds_commit_index = -1;
- list_add(&data->pages, &list);
- nreq++;
- } else {
- nfs_retry_commit(mds_pages, NULL, cinfo, 0);
- pnfs_generic_retry_commit(cinfo, 0);
- return -ENOMEM;
- }
+ data = nfs_commitdata_alloc(true);
+ data->ds_commit_index = -1;
+ list_add(&data->pages, &list);
+ nreq++;
}
nreq += pnfs_generic_alloc_ds_commits(cinfo, &list);
diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index abb2c8a3be42..bdfe5a7c5874 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -60,14 +60,28 @@ static mempool_t *nfs_wdata_mempool;
static struct kmem_cache *nfs_cdata_cachep;
static mempool_t *nfs_commit_mempool;
-struct nfs_commit_data *nfs_commitdata_alloc(void)
+struct nfs_commit_data *nfs_commitdata_alloc(bool never_fail)
{
- struct nfs_commit_data *p = mempool_alloc(nfs_commit_mempool, GFP_NOIO);
+ struct nfs_commit_data *p;
- if (p) {
- memset(p, 0, sizeof(*p));
- INIT_LIST_HEAD(&p->pages);
+ if (never_fail)
+ p = mempool_alloc(nfs_commit_mempool, GFP_NOIO);
+ else {
+ /* It is OK to do some reclaim, not no safe to wait
+ * for anything to be returned to the pool.
+ * mempool_alloc() cannot handle that particular combination,
+ * so we need two separate attempts.
+ */
+ p = mempool_alloc(nfs_commit_mempool, GFP_NOWAIT);
+ if (!p)
+ p = kmem_cache_alloc(nfs_cdata_cachep, GFP_NOIO |
+ __GFP_NOWARN | __GFP_NORETRY);
+ if (!p)
+ return NULL;
}
+
+ memset(p, 0, sizeof(*p));
+ INIT_LIST_HEAD(&p->pages);
return p;
}
EXPORT_SYMBOL_GPL(nfs_commitdata_alloc);
@@ -82,8 +96,7 @@ static struct nfs_pgio_header *nfs_writehdr_alloc(void)
{
struct nfs_pgio_header *p = mempool_alloc(nfs_wdata_mempool, GFP_NOIO);
- if (p)
- memset(p, 0, sizeof(*p));
+ memset(p, 0, sizeof(*p));
return p;
}
@@ -1705,19 +1718,13 @@ nfs_commit_list(struct inode *inode, struct list_head *head, int how,
if (list_empty(head))
return 0;
- data = nfs_commitdata_alloc();
-
- if (!data)
- goto out_bad;
+ data = nfs_commitdata_alloc(true);
/* Set up the argument struct */
nfs_init_commit(data, head, NULL, cinfo);
atomic_inc(&cinfo->mds->rpcs_out);
return nfs_initiate_commit(NFS_CLIENT(inode), data, NFS_PROTO(inode),
data->mds_ops, how, 0);
- out_bad:
- nfs_retry_commit(head, NULL, cinfo, 0);
- return -ENOMEM;
}
int nfs_commit_file(struct file *file, struct nfs_write_verifier *verf)