Han-Wen Nienhuys
Han-Wen Nienhuys
couldn't we provide stableAttr for the root node in the mount options?
Could you provide debug logs of what is happening? It's hard to tell from the bugreport.
(not that I don't have a mac, so just a repro scenario won't do).
the only suspect bit is this: 13:11:14.776379 rx 2: LISTXATTR i13850058 {sz 0} 13:11:14.776388 tx 2: 78=function not implemented These are currently not implemented, https://github.com/hanwen/go-fuse/blob/730713460d4fc41afdc2533bd37ff60c94c0c586/fs/loopback_darwin.go#L31 this should use functions from...
thanks for the report. That sounds simple to fix, but now I regret the default implementation here, as the sort can be expensive.
great to hear that you have solved the problem. Do you still need anything here?
it's hard to tell. You can probably tell from the getattr/open/read sequence what is going on. Normally, a program wouldn't just read 1 byte; is it really requesting 1 byte?
weird; what does it look like without DIRECT ? On Linux, I'd use strace to see the system calls executed; it seems strange for a program to read 1 byte,...
did this work with DIRECT in bazil.org FUSE ? Why do you need DIRECT ? the only difference I see is that your user-supplied inode number doesn't match the number...
``` 2020/09/20 19:41:37 rx 3: ACCESS i1 {u=501 g=20 w} 2020/09/20 19:41:37 tx 3: 13=permission denied ``` have you tried implementing the Access operation?