On Tue, Apr 5, 2016 at 9:51 AM, Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
>
> We had already been through that discussion, IIRC, with that exact file.
> And the same question remains - why not have that flag cleared by xenbus
> ->open()? You are using very odd heuristics to catch files that have
> unusual locking requirements;
So I actually much prefer Jan's patch and don't think his heuristics
are very unusual at all.
Instead of special-casing something lkike xenbus, Jan's patch says
"if position isn't something meaningful, let's not waste time and
effort on locking that makes no sense".
So to me, Jan's patch is the generic and clean solution, and the fact
that if fixes xenbus may be the reason the patch got written, but the
patch makes sense to me on its own.
But if you hate it, I guess a xenbus-specific hack would be fine, but
I think it smells hackier than just saying "nonseekable also implies
that you don't care about position locking". That statement not only
describes the patch fairly well, it simply makes sense to me.
Linus
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
