libfuse
libfuse copied to clipboard
libfuse
The reference implementation of the Linux FUSE (Filesystem in Userspace) interface
In addition to specifying the maximum number of idle threads, we should support a maximum total number of threads. This would probably allow us to get rid of the "forget"...
In the passthrough example lo_open uses the proc self trick rather then openat with O_EMPTY on the file descriptor opened with O_PATH: https://github.com/libfuse/libfuse/blob/ee0b29552ae92a6d991b02c778377fda6bfbfac5/example/passthrough_ll.c#L414 I wanted to know if this is...
The way readdir works in the high-level API is not documented well. Explaining how the internal buffer is allocated and better explaining the return value of the filler function in...
Michael Kerrisk feels favorably about adding a *fuse(7)* manpage. It would document the mount options that can be passed to the kernel, so that *mount.fuse(8)* can focus on just the...
At the moment, this is a mount option. This is rather odd given that max_write, max_readahead etc can all be specified in the init() handler.
These days, most FUSE settings are negotiated in the `init()` handler after opening the FUSE device, rather than passed as mount options. For consistency, it would be nice to extend...
These days, most FUSE settings are negotiated in the `init()` handler after opening the FUSE device, rather than passed as mount options. For consistency, it would be nice to extend...
We need to keep track of changes to the FUSE kernel module because sometimes revelant commits aren't Cc'ed to the fuse-devel list, and then the corresponding userspace changes never make...
The filler function is convenient but the fact the internal buffer may be realloc'ed multiple times could be costly as well as the multiple copying of data. The code is...
The kernel protocol does not seem to be well documented. It would ease development of alternative libraries and features if there was a proper document describing the /dev/fuse protocol and...