Re: [Libguestfs] [PATCH nbdkit v5 FINAL 15/19] file: Implement extents.

On Wed, Apr 24, 2019 at 02:39:03PM -0500, Eric Blake wrote: > On 3/28/19 11:18 AM, Richard W.M. Jones wrote: >> This uses lseek SEEK_DATA/SEEK_HOLE to search for allocated data and >> holes in the underlying file.

> Should we also return 0 if the lseek() returned the offset of EOF?

> reason against: if we expect NBD_CMD_TRIM/NBD_CMD_WRITE_ZEROES or even a > third-party to be punching holes in the meantime, then opting out of > future lseek() won't see those later additions of holes. This last point seems crucial - the file could become sparse, even if it starts off as non-sparse. The test is designed really to eliminate the case where SEEK_HOLE isn't supported and returns an error.

However you rightly point out that there's another case we don't cover: what if we're using tmpfs where SEEK_HOLE is supported but has poor performance? I would say the best solution is to fix tmpfs! But if that's not possible, perhaps we can do {f,}statfs(2) on the file and blacklist certain combinations of f_type and kernel version?