When the memory allocation for a response msg in cs_get_hdr() fails, it falls back to rpc_get_hdr() to try to allocate. However the response msg allocated by rpc_get_hdr() will cause the data being received into the newly allocated memory area instead of expected destination, and then complete the request without error, thus returning incorrect data back. This bug only affect small sized (<4k) read.
To fix, we set the request on a recoverable err, PCS_ERR_INTERRUPTED, if the allocation in cs_get_hdr fail. This will trigger a retry by client. Related to $VSTOR-107235 https://virtuozzo.atlassian.net/browse/VSTOR-107235 Signed-off-by: Liu Kui <kui....@virtuozzo.com> --- fs/fuse/kio/pcs/pcs_cs.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/fs/fuse/kio/pcs/pcs_cs.c b/fs/fuse/kio/pcs/pcs_cs.c index 6eef58994fef..10d6c860c3f8 100644 --- a/fs/fuse/kio/pcs/pcs_cs.c +++ b/fs/fuse/kio/pcs/pcs_cs.c @@ -468,8 +468,10 @@ struct pcs_msg *cs_get_hdr(struct pcs_rpc *ep, struct pcs_rpc_hdr *h) return NULL; resp = pcs_rpc_alloc_input_msg(ep, sizeof(struct pcs_cs_iohdr)); - if (!resp) + if (!resp) { + pcs_set_local_error(&msg->error, PCS_ERR_INTERRUPTED); return NULL; + } memcpy(resp->_inline_buffer, h, sizeof(struct pcs_rpc_hdr)); resp->size = h->len; -- 2.39.5 (Apple Git-154) _______________________________________________ Devel mailing list Devel@openvz.org https://lists.openvz.org/mailman/listinfo/devel
