https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=261434
Alan Somers changed:
What|Removed |Added
Resolution|--- |Works As Intended
Status
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=261434
--- Comment #12 from Ivan Rozhuk ---
(In reply to Alan Somers from comment #11)
> Looking at the code you sent, sshfs_utimens is ignoring the tv_nsec field,
> which is clearly a bug.
I agree that libfuse+sshfs code is not clean, but sshf
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=261434
--- Comment #11 from Alan Somers ---
(In reply to Ivan Rozhuk from comment #10)
Always sending mtime along with atime would be exactly the wrong thing for the
kernel to do. Only sending atime is what it's supposed to do.
Looking at the co
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=261434
--- Comment #10 from Ivan Rozhuk ---
(In reply to Alan Somers from comment #9)
> All the kernel does it send upcalls like FUSE_SETATTR to the fuse server
It come here:
https://github.com/libfuse/libfuse/blob/b08e275083d0316fa4d735d457869d
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=261434
--- Comment #9 from Alan Somers ---
(In reply to Ivan Rozhuk from comment #8)
What is utimens? FreeBSD has no such thing. Do you mean utimensat? utimensat
is a syscall. The kernel can't call it. All the kernel does it send upcalls
like
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=261434
--- Comment #8 from Ivan Rozhuk ---
(In reply to Alan Somers from comment #7)
> The FUSE_SETATTR command allows the kernel to set any combination of
> timestamps
It may allow whatever, but it call utimens() with limited interface.
Even m
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=261434
--- Comment #7 from Alan Somers ---
I meant, "the kernel only tries to change the file's atime". Sorry for the
typo. The FUSE_SETATTR command allows the kernel to set any combination of
timestamps, or none. For example, here is the debug
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=261434
--- Comment #6 from Ivan Rozhuk ---
(In reply to Alan Somers from comment #5)
> The kernel only tries to change the file's mtime.
_M_time?
> I think this must be a bug in sshfs. Have you tried reporting it upstream?
No, I do not try t
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=261434
--- Comment #5 from Alan Somers ---
I can reproduce the problem with sshfs, but not with fuse-ext2 or bfffs. The
kernel only tries to change the file's mtime. I think this must be a bug in
sshfs. Have you tried reporting it upstream?
--
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=261434
--- Comment #4 from Ivan Rozhuk ---
That debug show that on file opened as RO some attributes changed before close.
root@rimwks:~# ls /home/rim/mnt/tmp/root/minipro/dist.sh
-rwxr-xr-x 1 root wheel 748B Jan 24 23:45:46 2022
/home/rim/mn
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=261434
--- Comment #3 from Alan Somers ---
Are you sure that mtime is changing? The debug traffic you pasted below
doesn't show that. Please show the output of "stat -x SOME_FILE" before and
after "cat". It might be interesting to check its out
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=261434
--- Comment #2 from Ivan Rozhuk ---
(In reply to Alan Somers from comment #1)
atime changed with mtime and ctime - this is bug.
--
You are receiving this mail because:
You are the assignee for the bug.
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=261434
Ivan Rozhuk changed:
What|Removed |Added
Summary|[fusefs] change mtime and |[fusefs] mtime and ctime
13 matches
Mail list logo
