Submitter | xdrudis |
---|---|
Date | 2010-09-27 20:56:59 |
Message ID | <20100927205659.GA4358@ideafix.casa.ct> |
Download | mbox | patch |
Permalink | /patch/1995/ |
State | Accepted |
Headers | show |
Comments
On 9/27/10 10:56 PM, xdrudis wrote: > Hello. > > While trying the latest svn code I've seen curious results. It mostly > works better than before but still hangs at various places (it seems > random which one of the places reaches at each boot attempt). Since > I haven't finished fixes in fidvid.c it may well be that. But if > I turn on post to serial port it seems to always hang at memory training. > I'll disregard all that atcontinue with fidvid fornow > > In any case, I've come across this code that looks wrong. > This patch makes no difference for me at runtime, but I still > feel it must fix something. I'm not sure whether the hardcoded 8 for the > number of nodes in pDCTstat is correct. I think it might have more > elements, but maybe it's enough to deal with the first 8 for some > reason ? > > Signed-of-by: Xavi Drudis Ferran <xdrudis@tinet.cat> > Index: mct_d.c > =================================================================== > --- mct_d.c (revision 5864) > +++ mct_d.c (working copy) Thanks a lot! r5867.
Patch
Index: mct_d.c =================================================================== --- mct_d.c (revision 5864) +++ mct_d.c (working copy) @@ -3516,10 +3516,11 @@ for (Node = 0; Node < 8; Node++) { pDCTstat = pDCTstatA + Node; - if (pDCTstat->NodePresent) + if (pDCTstat->NodePresent) { mct_BeforeDQSTrainSamp(pDCTstat); /* only Bx */ mct_ResetDLL_D(pMCTstat, pDCTstat, 0); mct_ResetDLL_D(pMCTstat, pDCTstat, 1); + } } }