Re: Should pointers to PGPROC be volatile-qualified? - Mailing list pgsql-hackers

From Tom Lane
Subject Re: Should pointers to PGPROC be volatile-qualified?
Date
Msg-id 21312.1189019807@sss.pgh.pa.us
Whole thread Raw
In response to Re: Should pointers to PGPROC be volatile-qualified?  (Brian Hurt <bhurt@janestcapital.com>)
List pgsql-hackers
Brian Hurt <bhurt@janestcapital.com> writes:
> Note that all sorts of weirdnesses are possible when you have shared 
> mutable state between multiple different threads.

Yeah.  In the majority of places this isn't a big problem because
access to shared memory looks like
LWLockAcquire(some_lock);... mess with shared state ...LWLockRelease(some_lock);

and as long as the LWLock functions are extern and not inlineable
all is well.  But we do need to be more careful in places where
we're violating that coding rule, and the various stuff that looks
at or changes xid and xmin is violating it.

What I'm inclined to do for now is put "volatile" into those places in
procarray.c where there seems to be a risk.
        regards, tom lane


pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: loose ends in lazy-XID-assigment patch
Next
From: Tom Lane
Date:
Subject: Re: Oddity with psql \d and pg_table_is_visible