[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Unexpected behavior of 'tail --follow=name' on special file via syml
From: |
Glenn Golden |
Subject: |
Re: Unexpected behavior of 'tail --follow=name' on special file via symlink |
Date: |
Mon, 6 Feb 2023 07:06:53 -0700 |
Pádraig Brady <P@draigBrady.com> [2023-02-05 19:13:15 +0000]:
>
> On 05/02/2023 16:56, Glenn Golden wrote:
> > Your two patches (the one mentioned above and the earlier one avoiding the
> > xlseek() on a non-regular file) do almost fix the problem, but not quite.
> > I added one additional mod, and with that, it now seems to address the
> > issue fully. (And doesn't seem to break any "make check" tests.)
>
> Oh right I see. Sorry for missing that.
> After an fd change we need to go around the loop again
> to reinit the local fd and also guard against fd==-1 etc.
> Previously we always went around the loop again after a recheck()
> so any previous use of fd was fine.
>
Aha, yes, I see what you did there by just adding fd != f[i].fd as an
additional condition to force the 'continue' after the recheck. Much
cleaner and more general than the way I had done it.
>
> Attaching the final two patches I hope to apply for this.
> They should apply to coreutils-9.1
>
Yes, confirming those patches do apply cleanly to 9.1, and do fix the issue.
Thanks Pádraig,
Glenn