NFSD: Free async copy information in nfsd4_cb_offload_release()

Browse source

RFC 7862 Section 4.8 states:

> A copy offload stateid will be valid until either (A) the client
> or server restarts or (B) the client returns the resource by
> issuing an OFFLOAD_CANCEL operation or the client replies to a
> CB_OFFLOAD operation.

Currently, NFSD purges the metadata for an async COPY operation as
soon as the CB_OFFLOAD callback has been sent. It does not wait even
for the client's CB_OFFLOAD response, as the paragraph above
suggests that it should.

This makes the OFFLOAD_STATUS operation ineffective during the
window between the completion of an asynchronous COPY and the
server's receipt of the corresponding CB_OFFLOAD response. This is
important if, for example, the client responds with NFS4ERR_DELAY,
or the transport is lost before the server receives the response. A
client might use OFFLOAD_STATUS to query the server about the still
pending asynchronous COPY, but NFSD will respond to OFFLOAD_STATUS
as if it had never heard of the presented copy stateid.

This patch starts to address this issue by extending the lifetime of
struct nfsd4_copy at least until the server has seen the client's
CB_OFFLOAD response, or the CB_OFFLOAD has timed out.

Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

...

This commit is contained in:

Chuck Lever 2024-10-31 09:40:04 -04:00

parent 62a8642ba0

commit 409d6f52bd

2 changed files with 13 additions and 7 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download patch file Download diff file Expand all files Collapse all files

17

fs/nfsd/nfs4proc.c

View file

3

fs/nfsd/xdr4.h

View file