cvs commit: src/sys/emulation/linux/i386/linprocfs linprocfs.h

YONETANI Tomokazu qhwt+dfly at les.ath.cx
Thu Feb 8 07:41:49 PST 2007


On Sat, Feb 03, 2007 at 11:39:08AM +0100, Sascha Wildner wrote:
> YONETANI Tomokazu wrote:
> >y0netan1    2007/02/03 01:50:49 PST
> >
> >DragonFly src repository
> >
> >  Modified files:
> >    sys/emulation/linux/i386/linprocfs linprocfs.h 
> >  Log:
> >  add prototype for linprocfs_init()
> 
> Is this function used at all (I remember vaguely that it isn't)?

(I didn't notice that my reply didn't make it to commits@ ...)
No.  The driver happens to work without calling linprocfs_init() because
all it does is to call lwkt_token_init(), which only zero'es the token,
which is already zero'ed when it was loaded.  Besides that,
lwkt_token_uninit() is called from the bottom of linprocfs_exit(), which is
wrong, because it's at_exit()'s callback function and can be called
more than once.  Yeah, I know lwkt_token_uninit() is a no-op anyway, so
yanking lwkt_token_{,un}init() practically solves the problem, but it
doesn't look right...

Cheers.





More information about the Commits mailing list