[PATCH 0/7] fuse: Improve disconnect scheme and avoid taking fpq->lock on hot paths

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



There is no a reason to set individual FR_ABORTED state
for every request, since fuse_abort_conn() aborts all
unlocked requests at once. FR_ABORTED bit and fpq->io
list just allow fuse_copy_aborted() to end some of requests,
which are in the middle of fuse_dev_do_read() and
fuse_dev_do_write(). But this is not a big deal, since
these functions abort the requests themselves.

This patchset introduces a better scheme for fuse_abort_conn(),
which allows to remove excess flags and fpq->io, and optimizes
hot paths fuse_dev_do_read() and fuse_dev_do_write() by avoiding
taking fpq->lock there.

---

Kirill Tkhai (7):
      fuse: Check for fc->connected in fuse_dev_alloc()
      fuse: Move flush_bg_queue() up in fuse_abort_conn()
      fuse: Drop and reacquire fc->lock in middle of fuse_abort_conn()
      fuse: Add fud pointer to struct fuse_copy_state
      fuse: Introduce generic fuse_copy_aborted()
      fuse: Kill unused FR_ABORTED, FR_LOCKED and FR_PRIVATE flags
      fuse: Kill fuse_pqueue::io list and avoid taking fpq->lock on hot paths


 fs/fuse/dev.c    |  131 ++++++++++++++++--------------------------------------
 fs/fuse/fuse_i.h |   20 ++------
 fs/fuse/inode.c  |   10 ++++
 3 files changed, 52 insertions(+), 109 deletions(-)

--
Signed-off-by: Kirill Tkhai <ktkhai@xxxxxxxxxxxxx>



[Index of Archives]     [Linux Ext4 Filesystem]     [Union Filesystem]     [Filesystem Testing]     [Ceph Users]     [Ecryptfs]     [AutoFS]     [Kernel Newbies]     [Share Photos]     [Security]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux Cachefs]     [Reiser Filesystem]     [Linux RAID]     [Samba]     [Device Mapper]     [CEPH Development]

  Powered by Linux