On Wed, Dec 01 2021 at 13:19:03 +0100, Stefan Metzmacher quoth thus: > Hi Stefan, > > > On 11/29/21 5:08 PM, Clay Harris wrote: > >> On Mon, Nov 29 2021 at 14:12:52 -0800, Stefan Roesch quoth thus: > >> > >>> This adds the xattr support to io_uring. The intent is to have a more > >>> complete support for file operations in io_uring. > >>> > >>> This change adds support for the following functions to io_uring: > >>> - fgetxattr > >>> - fsetxattr > >>> - getxattr > >>> - setxattr > >> > >> You may wish to consider the following. > >> > >> Patching for these functions makes for an excellent opportunity > >> to provide a better interface. Rather than implement fXetattr > >> at all, you could enable io_uring to use functions like: > >> > >> int Xetxattr(int dfd, const char *path, const char *name, > >> [const] void *value, size_t size, int flags); > >> > >> Not only does this simplify the io_uring interface down to two > >> functions, but modernizes and fixes a deficit in usability. > >> In terms of io_uring, this is just changing internal interfaces. > >> > >> Although unnecessary for io_uring, it would be nice to at least > >> consider what parts of this code could be leveraged for future > >> Xetxattr2 syscalls. > > > > Clay, > > > > while we can reduce the number of calls to 2, providing 4 calls will > > ease the adoption of the interface. > > > > If you look at the userspace interface in liburing, you can see the > > following function signature: > > > > static inline void io_uring_prep_fgetxattr(struct io_uring_sqe *sqe, > > int fd, > > const char *name, > > const char *value, > > size_t len) > > > > This is very similar to what you proposed. > > What's with lsetxattr and lgetxattr, why are they missing. Do any filesystems even support xattrs on symbolic links? > I'd assume that even 6 helper functions in liburing would be able > to use just 2 low level iouring opcodes. > > *listxattr is also missing, are there plans for them? > > metze
