[Devel] [PATCH RHEL7 COMMIT] fs/fuse kio_pcs: remove s_subtype check

2018-05-28 Thread Konstantin Khorenko
The commit is pushed to "branch-rh7-3.10.0-693.21.1.vz7.50.x-ovz" and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-693.21.1.vz7.50.2 --> commit 5d36d87d2ed05619a094bb362f88af0dc31e086e Author: Pavel Butsykin Date: Mon May 28 18:39:33

[Devel] [PATCH RHEL7 COMMIT] fuse kio: Declare pcs_sockio::parent type

2018-05-28 Thread Konstantin Khorenko
The commit is pushed to "branch-rh7-3.10.0-693.21.1.vz7.50.x-ovz" and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-693.21.1.vz7.50.2 --> commit 42767c7665b55dd5910424e4b53a927f951a7d70 Author: Kirill Tkhai Date: Mon May 28 18:38:57 2018

[Devel] [PATCH RHEL7 COMMIT] fuse kio: Do not requeue requests when connection is dead

2018-05-28 Thread Konstantin Khorenko
The commit is pushed to "branch-rh7-3.10.0-693.21.1.vz7.50.x-ovz" and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-693.21.1.vz7.50.2 --> commit 08a10aaca24b6df19e1ca5b571ded3e0d4ff4cac Author: Kirill Tkhai Date: Mon May 28 18:37:46 2018

Re: [Devel] [PATCH] fs/fuse kio_pcs: remove s_subtype check

2018-05-28 Thread Kirill Tkhai
On 28.05.2018 18:28, Pavel Butsykin wrote: > The s_subtype field is initialized after calling vfs_kern_mount(), so we can't > check s_subtype inside vfs_kern_mount(). Instead of s_subtype we can check > dev_name, but this leads to an unjustified modification of common fuse. > > Given that

[Devel] [PATCH RH7] fs/dcache: show fs_type in warning about unnamed root entry

2018-05-28 Thread Pavel Tikhomirov
We have a noizy warning from customer: Call Trace: [] dump_stack+0x19/0x1b [] __warn+0xd8/0x100 [] warn_slowpath_fmt+0x5f/0x80 [] prepend_path+0x25f/0x320 [] d_path+0xe5/0x150 [] proc_pid_readlink+0x9d/0x100 [] SyS_readlinkat+0x9b/0x130 [] SyS_readlink+0x1b/0x20 []

[Devel] [PATCH] fs/fuse kio_pcs: remove s_subtype check

2018-05-28 Thread Pavel Butsykin
The s_subtype field is initialized after calling vfs_kern_mount(), so we can't check s_subtype inside vfs_kern_mount(). Instead of s_subtype we can check dev_name, but this leads to an unjustified modification of common fuse. Given that s_subtype/dev_name check doesn't protect from the