[PATCH RFC 4/6] SUNRPC: svc_fill_write_vector() must handle non-zero page_bases
STALE1857d
From: Chuck Lever <hidden>
Date: 2021-08-31 19:05:37
Also in:
linux-nfs
Subsystem:
kernel nfsd, sunrpc, and lockd servers, networking [general], nfs, sunrpc, and lockd clients, the rest · Maintainers:
Chuck Lever, Jeff Layton, "David S. Miller", Eric Dumazet, Jakub Kicinski, Paolo Abeni, Trond Myklebust, Anna Schumaker, Linus Torvalds
svc_fill_write_vector() was introduced in commit 8154ef2776aa
("NFSD: Clean up legacy NFS WRITE argument XDR decoders").
I'm about to add a case where page_base will sometimes not be zero
when the NFS WRITE proc functions invoke this API. Refactor the
API and function internals to handle this case properly.
Signed-off-by: Chuck Lever <redacted>
---
net/sunrpc/svc.c | 7 +++++--
1 file changed, 5 insertions(+), 2 deletions(-)
diff --git a/net/sunrpc/svc.c b/net/sunrpc/svc.c
index 34ced7f538ee..3d9f9da98aed 100644
--- a/net/sunrpc/svc.c
+++ b/net/sunrpc/svc.c@@ -1705,6 +1705,7 @@ unsigned int svc_fill_write_vector(struct svc_rqst *rqstp, struct kvec *vec = rqstp->rq_vec; size_t total = payload->len; unsigned int i; + size_t offset; /* Some types of transport can present the write payload * entirely in rq_arg.pages. In this case, @first is empty.
@@ -1717,12 +1718,14 @@ unsigned int svc_fill_write_vector(struct svc_rqst *rqstp, ++i; } + offset = payload->page_base; while (total) { - vec[i].iov_base = page_address(*pages); - vec[i].iov_len = min_t(size_t, total, PAGE_SIZE); + vec[i].iov_base = page_address(*pages) + offset; + vec[i].iov_len = min_t(size_t, total, PAGE_SIZE - offset); total -= vec[i].iov_len; ++i; ++pages; + offset = 0; } WARN_ON_ONCE(i > ARRAY_SIZE(rqstp->rq_vec));