diff options
author | Vivek Goyal <vgoyal@redhat.com> | 2020-05-04 14:33:15 -0400 |
---|---|---|
committer | Miklos Szeredi <mszeredi@redhat.com> | 2020-05-19 14:50:37 +0200 |
commit | 7fd3abfa8dd7c08ecacd25b2f9f9e1d3fb642440 (patch) | |
tree | 211d8f1b487e5be4eedd94c98bc6f67b9134b8d5 /ipc/namespace.c | |
parent | 5157da2ca42cbeca01709e29ce7b797cffed2432 (diff) | |
download | linux-stable-7fd3abfa8dd7c08ecacd25b2f9f9e1d3fb642440.tar.gz linux-stable-7fd3abfa8dd7c08ecacd25b2f9f9e1d3fb642440.tar.bz2 linux-stable-7fd3abfa8dd7c08ecacd25b2f9f9e1d3fb642440.zip |
virtiofs: do not use fuse_fill_super_common() for device installation
fuse_fill_super_common() allocates and installs one fuse_device. Hence
virtiofs allocates and install all fuse devices by itself except one.
This makes logic little twisted. There does not seem to be any real need
that why virtiofs can't allocate and install all fuse devices itself.
So opt out of fuse device allocation and installation while calling
fuse_fill_super_common().
Regular fuse still wants fuse_fill_super_common() to install fuse_device.
It needs to prevent against races where two mounters are trying to mount
fuse using same fd. In that case one will succeed while other will get
-EINVAL.
virtiofs does not have this issue because sget_fc() resolves the race
w.r.t multiple mounters and only one instance of virtio_fs_fill_super()
should be in progress for same filesystem.
Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Diffstat (limited to 'ipc/namespace.c')
0 files changed, 0 insertions, 0 deletions