On Thu, 13 Oct 2011 13:20:18 +0400
Pavel Shilovsky <piastryyy@xxxxxxxxx> wrote:
> 2011/8/29 Jeff Layton <jlayton@xxxxxxxxxx>:
> > Eventually we'll want to allow cifsd to read data directly into the
> > pagecache. In order to do that we'll need a routine that can take a
> > kvec array and pass that directly to kernel_recvmsg.
> >
> > Unfortunately though, the kernel's recvmsg routines modify the kvec
> > array that gets passed in, so we need to use a copy of the kvec array
> > and refresh that copy on each pass through the loop.
> >
> > Signed-off-by: Jeff Layton <jlayton@xxxxxxxxxx>
> > ---
> > fs/cifs/connect.c | 72 +++++++++++++++++++++++++++++++++++++++++++++++-----
> > 1 files changed, 65 insertions(+), 7 deletions(-)
> >
> > diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c
> > index e6da201..08c6b0e 100644
> > --- a/fs/cifs/connect.c
> > +++ b/fs/cifs/connect.c
> > @@ -375,14 +375,58 @@ server_unresponsive(struct TCP_Server_Info *server)
> > return false;
> > }
> >
> > +/*
> > + * kvec_array_setup - clone a kvec array, and advance into it
> > + * @new: pointer to memory for cloned array
> > + * @iov: pointer to original array
> > + * @nr_segs: number of members in original array
> > + * @bytes: number of bytes to advance into the cloned array
> > + *
> > + * This function will copy the array provided in iov to a section of memory
> > + * and advance the specified number of bytes into the new array. It returns
> > + * the number of segments in the new array. "new" must be at least as big as
> > + * the original iov array.
> > + */
> > +static unsigned int
> > +kvec_array_init(struct kvec *new, struct kvec *iov, unsigned int nr_segs,
>
> I think it should be kvec_array_setup as it said in the description above.
>
Good catch...
Unless there is an objection, I'll also fix this in a follow-on patch
to the series.
> > + size_t bytes)
> > +{
> > + size_t base = 0;
> > +
> > + while (bytes || !iov->iov_len) {
> > + int copy = min(bytes, iov->iov_len);
> > +
> > + bytes -= copy;
> > + base += copy;
> > + if (iov->iov_len == base) {
> > + iov++;
> > + nr_segs--;
> > + base = 0;
> > + }
> > + }
> > + memcpy(new, iov, sizeof(*iov) * nr_segs);
> > + new->iov_base += base;
> > + new->iov_len -= base;
> > + return nr_segs;
> > +}
> > +
> > static int
> > -read_from_socket(struct TCP_Server_Info *server, char *buf,
> > - unsigned int to_read)
> > +readv_from_socket(struct TCP_Server_Info *server, struct kvec *iov_orig,
> > + unsigned int nr_segs, unsigned int to_read)
> > {
> > int length = 0;
> > int total_read;
> > + unsigned int segs;
> > struct msghdr smb_msg;
> > - struct kvec iov;
> > + struct kvec *iov;
> > +
> > + /*
> > + * FIXME: allocation here may cause deadlocks under memory pressure.
> > + * Switch this to use a fixed, per-socket buffer.
> > + */
> > + iov = kmalloc(sizeof(*iov_orig) * nr_segs, GFP_KERNEL);
> > + if (!iov)
> > + return -ENOMEM;
> >
> > smb_msg.msg_control = NULL;
> > smb_msg.msg_controllen = 0;
> > @@ -393,10 +437,11 @@ read_from_socket(struct TCP_Server_Info *server, char *buf,
> > break;
> > }
> >
> > - iov.iov_base = buf + total_read;
> > - iov.iov_len = to_read;
> > - length = kernel_recvmsg(server->ssocket, &smb_msg, &iov, 1,
> > - to_read, 0);
> > + segs = kvec_array_init(iov, iov_orig, nr_segs, total_read);
> > +
> > + length = kernel_recvmsg(server->ssocket, &smb_msg,
> > + iov, segs, to_read, 0);
> > +
> > if (server->tcpStatus == CifsExiting) {
> > total_read = -ESHUTDOWN;
> > break;
> > @@ -423,9 +468,22 @@ read_from_socket(struct TCP_Server_Info *server, char *buf,
> > break;
> > }
> > }
> > + kfree(iov);
> > return total_read;
> > }
> >
> > +static int
> > +read_from_socket(struct TCP_Server_Info *server, char *buf,
> > + unsigned int to_read)
> > +{
> > + struct kvec iov;
> > +
> > + iov.iov_base = buf;
> > + iov.iov_len = to_read;
> > +
> > + return readv_from_socket(server, &iov, 1, to_read);
> > +}
> > +
> > static bool
> > check_rfc1002_header(struct TCP_Server_Info *server, char *buf)
> > {
> > --
> > 1.7.6
> >
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-cifs" in
> > the body of a message to majordomo@xxxxxxxxxxxxxxx
> > More majordomo info at http://vger.kernel.org/majordomo-info.html
> >
>
>
>
> --
> Best regards,
> Pavel Shilovsky.
--
Jeff Layton <jlayton@xxxxxxxxxx>
--
To unsubscribe from this list: send the line "unsubscribe linux-cifs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
