Re: [PATCH] NFS: Use wait_on_atomic_t() for unlock after readahead

[Trond Myklebust <trond.myklebust@primarydata.com>]

On Wed, Jan 6, 2016 at 10:40 AM, Benjamin Coddington
<bcodding@redhat.com> wrote:
> The use of wait_on_atomic_t() for waiting on I/O to complete before
> unlocking allows us to git rid of the NFS_IO_INPROGRESS flag, and thus the
> nfs_iocounter's flags member, and finally the nfs_iocounter altogether.
> The count of I/O is moved to the lock context, and the counter
> increment/decrement functions become simple enough to open-code.
>
> Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
> ---
>  fs/nfs/file.c          |    2 +-
>  fs/nfs/inode.c         |   18 ++++++++++++------
>  fs/nfs/internal.h      |    9 ++-------
>  fs/nfs/pagelist.c      |   48 +++++++-----------------------------------------
>  include/linux/nfs_fs.h |    8 +-------
>  5 files changed, 23 insertions(+), 62 deletions(-)
>
> diff --git a/fs/nfs/file.c b/fs/nfs/file.c
> index 93e2364..37ded36 100644
> --- a/fs/nfs/file.c
> +++ b/fs/nfs/file.c
> @@ -756,7 +756,7 @@ do_unlk(struct file *filp, int cmd, struct file_lock *fl, int is_local)
>
>         l_ctx = nfs_get_lock_context(nfs_file_open_context(filp));
>         if (!IS_ERR(l_ctx)) {
> -               status = nfs_iocounter_wait(&l_ctx->io_count);
> +               status = nfs_iocounter_wait(l_ctx);
>                 nfs_put_lock_context(l_ctx);
>                 if (status < 0)
>                         return status;
> diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
> index c7e8b87..c1bfee4 100644
> --- a/fs/nfs/inode.c
> +++ b/fs/nfs/inode.c
> @@ -71,19 +71,25 @@ nfs_fattr_to_ino_t(struct nfs_fattr *fattr)
>         return nfs_fileid_to_ino_t(fattr->fileid);
>  }
>
> -/**
> - * nfs_wait_bit_killable - helper for functions that are sleeping on bit locks
> - * @word: long word containing the bit lock
> - */
> -int nfs_wait_bit_killable(struct wait_bit_key *key, int mode)
> +static int nfs_wait_killable(int mode)
>  {
>         freezable_schedule_unsafe();
>         if (signal_pending_state(mode, current))
>                 return -ERESTARTSYS;
>         return 0;
>  }
> +
> +int nfs_wait_bit_killable(struct wait_bit_key *key, int mode)
> +{
> +       return nfs_wait_killable(mode);
> +}
>  EXPORT_SYMBOL_GPL(nfs_wait_bit_killable);
>
> +int nfs_wait_atomic_killable(atomic_t *p)
> +{
> +       return nfs_wait_killable(TASK_KILLABLE);
> +}
> +
>  /**
>   * nfs_compat_user_ino64 - returns the user-visible inode number
>   * @fileid: 64-bit fileid
> @@ -699,7 +705,7 @@ static void nfs_init_lock_context(struct nfs_lock_context *l_ctx)
>         l_ctx->lockowner.l_owner = current->files;
>         l_ctx->lockowner.l_pid = current->tgid;
>         INIT_LIST_HEAD(&l_ctx->list);
> -       nfs_iocounter_init(&l_ctx->io_count);
> +       atomic_set(&l_ctx->io_count, 0);
>  }
>
>  static struct nfs_lock_context *__nfs_find_lock_context(struct nfs_open_context *ctx)
> diff --git a/fs/nfs/internal.h b/fs/nfs/internal.h
> index 9dea85f..6c9b72f 100644
> --- a/fs/nfs/internal.h
> +++ b/fs/nfs/internal.h
> @@ -238,7 +238,7 @@ extern void nfs_pgheader_init(struct nfs_pageio_descriptor *desc,
>                               struct nfs_pgio_header *hdr,
>                               void (*release)(struct nfs_pgio_header *hdr));
>  void nfs_set_pgio_error(struct nfs_pgio_header *hdr, int error, loff_t pos);
> -int nfs_iocounter_wait(struct nfs_io_counter *c);
> +int nfs_iocounter_wait(struct nfs_lock_context *l_ctx);
>
>  extern const struct nfs_pageio_ops nfs_pgio_rw_ops;
>  struct nfs_pgio_header *nfs_pgio_header_alloc(const struct nfs_rw_ops *);
> @@ -252,12 +252,6 @@ void nfs_free_request(struct nfs_page *req);
>  struct nfs_pgio_mirror *
>  nfs_pgio_current_mirror(struct nfs_pageio_descriptor *desc);
>
> -static inline void nfs_iocounter_init(struct nfs_io_counter *c)
> -{
> -       c->flags = 0;
> -       atomic_set(&c->io_count, 0);
> -}
> -
>  static inline bool nfs_pgio_has_mirroring(struct nfs_pageio_descriptor *desc)
>  {
>         WARN_ON_ONCE(desc->pg_mirror_count < 1);
> @@ -380,6 +374,7 @@ extern void nfs_clear_inode(struct inode *);
>  extern void nfs_evict_inode(struct inode *);
>  void nfs_zap_acl_cache(struct inode *inode);
>  extern int nfs_wait_bit_killable(struct wait_bit_key *key, int mode);
> +extern int nfs_wait_atomic_killable(atomic_t *p);
>
>  /* super.c */
>  extern const struct super_operations nfs_sops;
> diff --git a/fs/nfs/pagelist.c b/fs/nfs/pagelist.c
> index 452a011..abf1dad 100644
> --- a/fs/nfs/pagelist.c
> +++ b/fs/nfs/pagelist.c
> @@ -101,53 +101,18 @@ nfs_page_free(struct nfs_page *p)
>         kmem_cache_free(nfs_page_cachep, p);
>  }
>
> -static void
> -nfs_iocounter_inc(struct nfs_io_counter *c)
> -{
> -       atomic_inc(&c->io_count);
> -}
> -
> -static void
> -nfs_iocounter_dec(struct nfs_io_counter *c)
> -{
> -       if (atomic_dec_and_test(&c->io_count)) {
> -               clear_bit(NFS_IO_INPROGRESS, &c->flags);
> -               smp_mb__after_atomic();
> -               wake_up_bit(&c->flags, NFS_IO_INPROGRESS);
> -       }
> -}
> -
> -static int
> -__nfs_iocounter_wait(struct nfs_io_counter *c)
> -{
> -       wait_queue_head_t *wq = bit_waitqueue(&c->flags, NFS_IO_INPROGRESS);
> -       DEFINE_WAIT_BIT(q, &c->flags, NFS_IO_INPROGRESS);
> -       int ret = 0;
> -
> -       do {
> -               prepare_to_wait(wq, &q.wait, TASK_KILLABLE);
> -               set_bit(NFS_IO_INPROGRESS, &c->flags);
> -               if (atomic_read(&c->io_count) == 0)
> -                       break;
> -               ret = nfs_wait_bit_killable(&q.key, TASK_KILLABLE);
> -       } while (atomic_read(&c->io_count) != 0 && !ret);
> -       finish_wait(wq, &q.wait);
> -       return ret;
> -}
> -
>  /**
>   * nfs_iocounter_wait - wait for i/o to complete
> - * @c: nfs_io_counter to use
> + * @l_ctx: nfs_lock_context with io_counter to use
>   *
>   * returns -ERESTARTSYS if interrupted by a fatal signal.
>   * Otherwise returns 0 once the io_count hits 0.
>   */
>  int
> -nfs_iocounter_wait(struct nfs_io_counter *c)
> +nfs_iocounter_wait(struct nfs_lock_context *l_ctx)
>  {
> -       if (atomic_read(&c->io_count) == 0)
> -               return 0;
> -       return __nfs_iocounter_wait(c);
> +       return wait_on_atomic_t(&l_ctx->io_count, nfs_wait_atomic_killable,
> +                       TASK_KILLABLE);
>  }
>
>  /*
> @@ -370,7 +335,7 @@ nfs_create_request(struct nfs_open_context *ctx, struct page *page,
>                 return ERR_CAST(l_ctx);
>         }
>         req->wb_lock_context = l_ctx;
> -       nfs_iocounter_inc(&l_ctx->io_count);
> +       atomic_inc(&l_ctx->io_count);
>
>         /* Initialize the request struct. Initially, we assume a
>          * long write-back delay. This will be adjusted in
> @@ -431,7 +396,8 @@ static void nfs_clear_request(struct nfs_page *req)
>                 req->wb_page = NULL;
>         }
>         if (l_ctx != NULL) {
> -               nfs_iocounter_dec(&l_ctx->io_count);
> +               if (atomic_dec_and_test(&l_ctx->io_count))
> +                       wake_up_atomic_t(&l_ctx->io_count);
>                 nfs_put_lock_context(l_ctx);
>                 req->wb_lock_context = NULL;
>         }
> diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h
> index c0e9614..de6ba5a 100644
> --- a/include/linux/nfs_fs.h
> +++ b/include/linux/nfs_fs.h
> @@ -60,18 +60,12 @@ struct nfs_lockowner {
>         pid_t l_pid;
>  };
>
> -#define NFS_IO_INPROGRESS 0
> -struct nfs_io_counter {
> -       unsigned long flags;
> -       atomic_t io_count;
> -};
> -
>  struct nfs_lock_context {
>         atomic_t count;
>         struct list_head list;
>         struct nfs_open_context *open_context;
>         struct nfs_lockowner lockowner;
> -       struct nfs_io_counter io_count;
> +       atomic_t io_count;
>  };
>
>  struct nfs4_state;

Thanks Ben! This looks good.