On Nov. 10, 2008, 22:56 +0200, Benny Halevy <bhalevy@xxxxxxxxxxx> wrote:
> From: Andy Adamson <andros@xxxxxxxxxx>
>
> Cache all the result pages, including the rpc header in rq_respages[0],
> for a request in the slot table cache entry.
>
> Cache the statp pointer from nfsd_dispatch which points into rq_respages[0]
> just past the rpc header. When setting a cache entry, calculate and save the
> length of the nfs data minus the rpc header for rq_respages[0].
>
> When replaying a cache entry, replace the cached rpc header with the
> replayed request rpc result header, unless there is not enough room in the
> cached results first page. In that case, use the cached rpc header.
>
> Signed-off-by: Andy Adamson<andros@xxxxxxxxxx>
> Signed-off-by: Benny Halevy <bhalevy@xxxxxxxxxxx>
> ---
> fs/nfsd/nfs4state.c | 106 ++++++++++++++++++++++++++++++++++++++++++++
> fs/nfsd/nfssvc.c | 6 +++
> include/linux/nfsd/cache.h | 3 +
> include/linux/nfsd/state.h | 17 +++++--
> include/linux/nfsd/xdr4.h | 7 +++
> 5 files changed, 134 insertions(+), 5 deletions(-)
>
> diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
> index ba6ab19..230ce1d 100644
> --- a/fs/nfsd/nfs4state.c
> +++ b/fs/nfsd/nfs4state.c
> @@ -412,6 +412,7 @@ alloc_init_session(struct nfs4_client *clp, struct nfsd4_create_session *cses)
> for (i = 0; i < new->se_fnumslots; i++) {
> new->se_slots[i].sl_session = new;
> nfs41_set_slot_state(&new->se_slots[i], NFS4_SLOT_AVAILABLE);
> + spin_lock_init(&new->se_slots[i].sl_cache_entry.ce_lock);
> }
>
> new->se_client = clp;
> @@ -997,6 +998,111 @@ nfsd4_clear_respages(struct page **respages, short resused)
> }
>
> #if defined(CONFIG_NFSD_V4_1)
> +
> +void
> +nfsd41_set_statp(struct svc_rqst *rqstp, __be32 *statp)
> +{
> + struct nfsd4_compoundres *resp = rqstp->rq_resp;
> +
> + resp->statp = statp;
> +}
> +
> +/*
> + * Cache the reply pages, clearing the previous pages.
> + * Store the base and length of the rq_req.head[0] page
> + * of the NFSv4.1 data, just past the rpc header.
> + */
> +void
> +nfsd41_set_cache_entry(struct nfsd4_compoundres *resp)
> +{
> + struct nfs41_cache_entry *entry =
> + &resp->current_ses->cs_slot->sl_cache_entry;
> + struct svc_rqst *rqstp = resp->rqstp;
> + struct kvec *resv = &rqstp->rq_res.head[0];
> +
> + dprintk("--> %s entry %p\n", __func__, entry);
> +
> + /* Don't cache a failed OP_SEQUENCE */
> + if (resp->opcnt == 1 && resp->status)
> + return;
> + spin_lock(&entry->ce_lock);
review 11-13: we can probably get rid of the spinlock
and s/nfsd4_clear_respages/nfsd4_release_respages/
> + nfsd4_clear_respages(entry->ce_respages, entry->ce_resused);
> + nfsd4_cache_rqst_pages(rqstp, entry->ce_respages, &entry->ce_resused);
> + entry->ce_status = resp->status;
> + entry->ce_datav.iov_base = resp->statp;
> + entry->ce_datav.iov_len = resv->iov_len - ((char *)resp->statp -
> + (char *)page_address(rqstp->rq_respages[0]));
> + spin_unlock(&entry->ce_lock);
> +}
> +
> +/*
> + * Copy the cached NFSv4.1 reply skipping the cached rpc header into the
> + * replay result res.head[0] past the rpc header to end up with replay
> + * rpc header and cached NFSv4.1 reply.
> + */
> +static int
> +nfsd41_copy_replay_data(struct nfsd4_compoundres *resp,
> + struct nfs41_cache_entry *entry)
> +{
> + struct svc_rqst *rqstp = resp->rqstp;
> + struct kvec *resv = &resp->rqstp->rq_res.head[0];
> + int len;
> +
> + /* Current request rpc header length*/
> + len = (char *)resp->statp - (char *)page_address(rqstp->rq_respages[0]);
> + if (entry->ce_datav.iov_len + len > PAGE_SIZE) {
> + dprintk("%s v41 cached reply too large (%Zd).\n", __func__,
> + entry->ce_datav.iov_len);
> + return 0;
> + }
> + /* copy the cached reply nfsd data past the current rpc header */
> + memcpy((char *)resv->iov_base + len, entry->ce_datav.iov_base,
> + entry->ce_datav.iov_len);
> + resv->iov_len = len + entry->ce_datav.iov_len;
> + return 1;
> +}
> +
> +/*
> + * Keep the first page of the replay. Copy the NFSv4.1 data from the first
> + * cached page. Replace any futher replay pages from the cache.
> + */
> +__be32
> +nfsd41_replay_cache_entry(struct nfsd4_compoundres *resp)
> +{
> + struct nfs41_cache_entry *entry =
> + &resp->current_ses->cs_slot->sl_cache_entry;
> + __be32 status;
> +
> + dprintk("--> %s entry %p\n", __func__, entry);
> +
> + spin_lock(&entry->ce_lock);
review 11-13: no spin lock, yes state lock
do this in nfsd4_sequence under the state lock.
do nothing in nfs4svc_encode_compoundres
> +
> + if (!nfsd41_copy_replay_data(resp, entry)) {
> + /*
> + * Not enough room to use the replay rpc header, send the
> + * cached header. Release all the allocated result pages.
> + */
> + svc_free_res_pages(resp->rqstp);
> + nfsd4_move_pages(resp->rqstp->rq_respages, entry->ce_respages,
> + entry->ce_resused);
> + } else {
> + /* Release all but the first allocatek result page */
review 11-13: s/allocatek/allocated/
> +
> + resp->rqstp->rq_resused--;
> + svc_free_res_pages(resp->rqstp);
> +
> + nfsd4_move_pages(&resp->rqstp->rq_respages[1],
> + &entry->ce_respages[1],
> + entry->ce_resused - 1);
> + }
> +
> + resp->rqstp->rq_resused = entry->ce_resused;
> + status = entry->ce_status;
> + spin_unlock(&entry->ce_lock);
> +
> + return status;
> +}
> +
> /*
> * Set the exchange_id flags returned by the server.
> */
> diff --git a/fs/nfsd/nfssvc.c b/fs/nfsd/nfssvc.c
> index 07e4f5d..a9f90b6 100644
> --- a/fs/nfsd/nfssvc.c
> +++ b/fs/nfsd/nfssvc.c
> @@ -543,6 +543,12 @@ nfsd_dispatch(struct svc_rqst *rqstp, __be32 *statp)
> + rqstp->rq_res.head[0].iov_len;
> rqstp->rq_res.head[0].iov_len += sizeof(__be32);
>
> +#ifdef CONFIG_NFSD_V4_1
> + /* NFSv4.1 DRC requires statp */
> + if (rqstp->rq_vers == 4)
> + nfsd41_set_statp(rqstp, statp);
> +#endif /* CONFIG_NFSD_V4_1 */
> +
> /* Now call the procedure handler, and encode NFS status. */
> nfserr = proc->pc_func(rqstp, rqstp->rq_argp, rqstp->rq_resp);
> nfserr = map_new_errors(rqstp->rq_vers, nfserr);
> diff --git a/include/linux/nfsd/cache.h b/include/linux/nfsd/cache.h
> index 04b355c..9315699 100644
> --- a/include/linux/nfsd/cache.h
> +++ b/include/linux/nfsd/cache.h
> @@ -75,5 +75,8 @@ int nfsd_reply_cache_init(void);
> void nfsd_reply_cache_shutdown(void);
> int nfsd_cache_lookup(struct svc_rqst *, int);
> void nfsd_cache_update(struct svc_rqst *, int, __be32 *);
> +#ifdef CONFIG_NFSD_V4_1
> +void nfsd41_set_statp(struct svc_rqst *, __be32*);
> +#endif /* CONFIG_NFSD_V4_1 */
>
> #endif /* NFSCACHE_H */
> diff --git a/include/linux/nfsd/state.h b/include/linux/nfsd/state.h
> index 4f37e2f..e4dc32f 100644
> --- a/include/linux/nfsd/state.h
> +++ b/include/linux/nfsd/state.h
> @@ -125,15 +125,22 @@ enum {
> NFS4_SLOT_INPROGRESS
> };
>
> +struct nfs41_cache_entry {
> + spinlock_t ce_lock;
> + __be32 ce_status;
> + struct kvec ce_datav; /* encoded NFSv4.1 data in rq_res.head[0] */
> + struct page *ce_respages[RPCSVC_MAXPAGES];
> + short ce_resused;
> +};
> +
> /*
> * nfs41_slot
> - *
> - * for now, just slot sequence number - will hold DRC for this slot.
> */
> struct nfs41_slot {
> - atomic_t sl_state;
> - struct nfs41_session *sl_session;
> - u32 sl_seqid;
> + atomic_t sl_state;
> + struct nfs41_session *sl_session;
> + u32 sl_seqid;
> + struct nfs41_cache_entry sl_cache_entry;
> };
>
> /*
> diff --git a/include/linux/nfsd/xdr4.h b/include/linux/nfsd/xdr4.h
> index fc42fba..554f2cd 100644
> --- a/include/linux/nfsd/xdr4.h
> +++ b/include/linux/nfsd/xdr4.h
> @@ -491,6 +491,11 @@ struct nfsd4_compoundres {
> u32 opcnt;
> __be32 * tagp; /* where to encode tag and opcount */
> u32 minorversion;
> +#if defined(CONFIG_NFSD_V4_1)
> + __be32 *statp;
> + u32 status;
> + struct current_session *current_ses;
> +#endif /* CONFIG_NFSD_V4_1 */
> };
>
> #define NFS4_SVC_XDRSIZE sizeof(struct nfsd4_compoundargs)
> @@ -534,6 +539,8 @@ extern __be32 nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
> #if defined(CONFIG_NFSD_V4_1)
> extern void nfsd41_current_session_free(struct current_session *cses);
> extern struct current_session *nfsd41_current_session_alloc(void);
> +extern void nfsd41_set_cache_entry(struct nfsd4_compoundres *resp);
> +extern __be32 nfsd41_replay_cache_entry(struct nfsd4_compoundres *resp);
> extern __be32 nfsd4_exchange_id(struct svc_rqst *rqstp,
> struct nfsd4_compound_state *,
> struct nfsd4_exchange_id *);
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
