Re: kern/40384 (64 bit time_t broke wscons)

[yamt%mwd.biglobe.ne.jp@localhost (YAMAMOTO Takashi)]

The following reply was made to PR kern/40384; it has been noted by GNATS.

From: yamt%mwd.biglobe.ne.jp@localhost (YAMAMOTO Takashi)
To: christos%zoulas.com@localhost
Cc: gnats-bugs%NetBSD.org@localhost, kern-bug-people%netbsd.org@localhost, 
netbsd-bugs%netbsd.org@localhost,
        gnats-admin%netbsd.org@localhost
Subject: Re: kern/40384 (64 bit time_t broke wscons)
Date: Wed, 14 Jan 2009 23:57:49 +0900 (JST)

 hi,
 
 > On Jan 14, 11:48am, yamt%mwd.biglobe.ne.jp@localhost (YAMAMOTO Takashi) 
 > wrote:
 > -- Subject: Re: kern/40384 (64 bit time_t broke wscons)
 > 
 > | hi,
 > | 
 > | > Synopsis: 64 bit time_t broke wscons
 > | > 
 > | > State-Changed-From-To: open->closed
 > | > State-Changed-By: christos%NetBSD.org@localhost
 > | > State-Changed-When: Tue, 13 Jan 2009 14:01:19 -0500
 > | > State-Changed-Why:
 > | > fixed, thanks
 > | 
 > | thanks for a quick fix.
 > | won't it be confused by concurrent operations?
 > | eg. concurrent SETVERSION ioctls
 > | i think that it's better to always use the new version for
 > | in-kernel structures and do conversion when doing copyin/out.
 > 
 > Each setversion flushes the queue. I thought about doing it the way you
 > suggest, but it makes the code more complicated. I think that for the
 > most part there will be one setversion per open or none.
 > 
 > christos
 
 it flushes the queue, yes.  but it seems unsafe to me.
 eg. if malloc() sleeps.
 eg. if someone is in the middle of wsevent_read().
 necessary locking complicates the code more, i guess.
 
 YAMAMOTO Takashi