[v2,0/7] net/9p: remove msize limit in virtio transport

Message

Christian Schoenebeck Sept. 20, 2021, 4:43 p.m. UTC

This series aims to get get rid of the current 500k 'msize' limitation in
the 9p virtio transport, which is currently a bottleneck for performance
of 9p mounts.

This is a follow-up of the following series and discussion:
https://lore.kernel.org/all/2[e-mail]/T/#eb647d0c013616cee3eb8ba9d87da7d8b1f476f37

Known limitation: With this series applied I can run

  QEMU host <-> 9P virtio <-> Linux guest

with up to 3 MB msize. If I try to run it with 4 MB it seems to hit some
limitation on QEMU side:

  qemu-system-x86_64: virtio: too many write descriptors in indirect table

I haven't looked into this issue yet.

Testing and feedback appreciated!

v1 -> v2:

  * Fix memory leak [patch 6].

  * Use pr_info() instead of p9_debug(P9_DEBUG_ERROR, ...) for user info
    [patch 1].

  * Show user info message if chained SG lists are not supported on
    architecture [patch 6].

  * Kernel doc fixes [patch 4], [patch 6].

Christian Schoenebeck (7):
  net/9p: show error message if user 'msize' cannot be satisfied
  9p/trans_virtio: separate allocation of scatter gather list
  9p/trans_virtio: turn amount of sg lists into runtime info
  9p/trans_virtio: introduce struct virtqueue_sg
  net/9p: add trans_maxsize to struct p9_client
  9p/trans_virtio: support larger msize values
  9p/trans_virtio: resize sg lists to whatever is possible

 include/net/9p/client.h |   2 +
 net/9p/client.c         |  17 ++-
 net/9p/trans_virtio.c   | 290 ++++++++++++++++++++++++++++++++++------
 3 files changed, 269 insertions(+), 40 deletions(-)