CFT: vr(4)

[Pyun YongHyeon]

On Sun, Feb 24, 2008 at 10:29:31AM +0100, Milan Obuch wrote:
 > On Friday 22 February 2008, Pyun YongHyeon wrote:
 > 
 > [ snip ]
 > 
 > >  >
 > >  > I've put up a new version under the old URL.
 > >  > It's not tested in sparc64 but it seems to work on i386.
 > >  > Would you give it spin?
 > >
 > > Any progress here? Does the updated one works on your box?
 > 
 > Well, I arranged a test today and the result is no change.
 > After some time box hard locks. It occurs only with if_vr kldload'ed and after 
 > some traffic occured. No ability to enter debugger though. So while it seems 
 > related to vr ethernet card, I can't prove it, I can't dig any usefull data 
 > from the box.
 > Still, if there is some debug possibility built into driver you can point me 
 > at, I will try. But now I have no idea what's wrong here.

Hmm, this really make me mad. :-(
Please apply attached patch and let me know how it goes.

 > Regards,
 > Milan
 > 

-- 
Regards,
Pyun YongHyeon
-------------- next part --------------
--- if_vr.c.orig	2008-02-18 10:46:10.000000000 +0900
+++ if_vr.c	2008-02-25 13:30:56.000000000 +0900
@@ -638,6 +638,10 @@
 	pci_enable_busmaster(dev);
 	sc->vr_revid = pci_get_revid(dev);
 	device_printf(dev, "Revision: 0x%x\n", sc->vr_revid);
+#if 1
+	if (sc->vr_revid == 0x96)
+		sc->vr_quirks |= VR_Q_NEEDALIGN;
+#endif
 
 	/*
 	 * Prefer memory-mapped register accesses over IO-mapped one.