Re: LOR: sys/kern/sys_generic.c:902 sys/fs/specfs/spec_vnops.c:387

Daniel Lang wrote:
> Hi John,
>
> Jon Noack wrote on Sat, Apr 10, 2004 at 12:54:14PM -0500:
> [..]
>> I got this same one today.  Seems like this one even has a PR:
>> http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/55175
>>
>> Like I said, I'm not going to put any work into the web page until I
>> find
>> someone to commit a pointer to it when you get a lock order reversal.
>> Otherwise, the page is of limited value (we will still get peppered with
>> questions on the list).
>
> Why do you insist on the pointer to be present before you start your
> webpage? You are making a chicken-egg-problem out of it, while there is
> no need. IMHO a webpage with information, but no prominent pointer is
> preferred over a pointer in the LOR message to a nonexisting webpage.
>
> I am confident, once your webpage is set up and contains some
> useful information, there is someone, who will add the hint to
> the error-message. But not the other way round, no one will
> commit a hint to something, which doesn't exist.

This may not have been clear: I wasn't asking anyone to commit it *before*
the page goes up.  I was asking for someone to make a *commitment* to
commit a pointer after the page goes up.  I'm not going to put the time
into making the page and still have people asking the same question for
the millionth time because few know about the page.  Heck, part of the
reason I'm making the page is to save myself the annoyance of seeing the
same questions every day.  It's best for all involved for the page to be
prominent (as in, mentioned in the LOR warning).

Brian Feldman has now made the commitment.  Once I put up a decent page,
he'll commit the pointer.  If you'd like to help, let me know.

Jon Noack