Re: FUSE Call for Testing

On Fri, Aug 9, 2019 at 7:38 AM Gary Jennejohn <gljennjohn_at_gmail.com> wrote:
>
> On Fri, 9 Aug 2019 06:49:36 -0600
> Alan Somers <asomers_at_freebsd.org> wrote:
>
> [snip test results]
>
> > Thanks for the report, Gary.  BTW, fusefs mounts are only
> > interruptible when mounted with "-o intr".  If you didn't use that
> > option, then the signal would only interrupt cp after the write to
> > fusefs was done.  Also, not every fuse file system supports
> > interrupts.  Looking through its sources, I don't think that
> > fusefs-ntfs does.
> >
>
> I'm not convinced of that.  I interrupted the 5.2GiB transfer
> several times to try out different UBLIO environment settings
> and it stopped pretty much immediately.  I also did not use
> mount but rather a direct invocation of ntfs-3g.
>
> I suspect that ntfs-3g stops after the interrupt once it has
> flushed the UBLIO buffers.
>
> --
> Gary Jennejohn (gj_at_)

Well, I can guarantee that the write(2) into fusefs was not
interrupted if you did not specify "-o intr".  I know, because I wrote
that code.  You would need to use "-o intr" whether you mount by
invoking mount or ntfs-3g.  But cp probably only copies a few KB to a
few hundred KB at a time.  So even without "-o intr", a signal would
probably feel fairly responsive.
-Alan