Re: [Nmh-workers] things still to fix for this release?

[pmaydell]

Oliver Kiddle wrote:
>address@hidden wrote:
>> http://savannah.nongnu.org/bugs/?func=detailitem&item_id=7917
>> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=226825
>> http://savannah.nongnu.org/bugs/?func=detailitem&item_id=14738
>>  -- post is trying to lock /dev/null so blocks for 30secs+
>>  (the last of these includes a patch -- is it sane?)
>
>The patch doesn't look sane to me. I don't know enough about it but I
>expect that it shouldn't be trying to lock /dev/null in the first place
>and if it is doing locking, it is doing it for a reason so removing the
>locking may cause another problem.

I've investigated a bit, and I think the problem is that context_foil()
sets the global variable context to "/dev/null". This subsequently
gets used in eg context_read(). This used to be fine because those
routines did no locking -- but a change to make them use lkfopen()
rather than plain fopen() means that we sometimes try to lock /dev/null.

A clean fix is probably going to be to have context_foil() work 
differently (eg by setting context to NULL and handling this where
we use it?).

Does anybody have an easy reproduce case for this? (I did the above
by pure code inspection.)

-- PMM