[2316] in Release_7.7_team
Re: problems with knfs
daemon@ATHENA.MIT.EDU (Garry Zacheiss)
Mon Jun 19 17:19:13 2000
Message-Id: <200006192119.RAA25405@sweet-transvestite.mit.edu>
To: Michal N Lusztig <miki@MIT.EDU>
Cc: release-team@MIT.EDU
In-Reply-To: Your message of "Mon, 19 Jun 2000 14:44:23 EDT."
<200006191844.OAA10884@joust.MIT.EDU>
Date: Mon, 19 Jun 2000 17:19:04 -0400
From: Garry Zacheiss <zacheiss@MIT.EDU>
I'm not going to be able to be at the next meeting, but I did
have some time to look into the problem this afternoon. It looks like
it's another case case of our solaris 7 sources not matching the actual
binaries they ship. The function in question that seems to be causing
problems is nd_whirlygig(), which is defined in our sources although the
symbol doesn't appear in the nfs module from the binary distribution.
If you remove any reference to the function from nfs_dump.c and then
build the kernel module from the source tree, you end up with fs/nfs and
misc/nfssrv both loading happily at boot time.
The comment for nd_whirlygig() is:
/*
* we don't want the little whirling line to fill up the message buffer, so
* we turn msgbufinit off while we print it.
*/
so while I haven't really had time to examine the ramifications
of punting the function entirely, I think it should be safe.
We should, of course, yell at Sun about this.
Garry