Re: [flashrom] [coreboot] Dualbios on GA-MA770-UD3

[Stefan Tauner]

On Fri, 09 Apr 2010 19:08:21 +0400
vadimgirlin at gmail.com (Vadim Girlin) wrote:
> On 04/09/2010 04:18 PM, Carl-Daniel Hailfinger wrote:
> > Hi Vadim,
> >
> > thanks for your mail. I have added the flashrom mailing list in CC:
> > because we should develop a generic way to handle DualBIOS and similar
> > techniques.
> >
> > On 09.04.2010 08:12, Vadim Girlin wrote:
> > > I'm going to try coreboot on Gigabyte GA-MA770-UD3.
> > > It's AMD 770 (RX780 / SB700).
> > >
> > > My motherboard supports hardware dual bios - with two chips on it.
> > > I'm going to try flashing backup chip and boot from it. It seems to be
> > > possible - I've tested it (flashing at least). Chips on this board could
> > > be switched by changing bit 0 at undocumented register EF on LDN 7 of
> > > IT8720. I can use slightly patched flashrom for accessing any chip I
> > > want without any problems. And this is tested many times.
> > >
> > > My idea is to use backup chip for debugging - that always keeps my
> > > chance to reboot from main bios chip. And removes the need for
> > > desoldering etc.
> > >
> > > And second problem is that original bios is checking second chip - and
> > > trying to recover it by flashing the bios from main chip on reboot?
> > > rewriting coreboot. AFAICS this could be solved by including some
> > > signatures etc. It seems to be easy to find out. May be someone is
> > > working on this?
> > >
> > > BTW I can send the patch for flashrom - but I think that with
> > > information I mentioned above somebody could make it much better than my
> > > ugly hack. I hope the regs should be the same for all latest Gigabyte
> > > MBs using IT8720/18
> >
> > It would be great if you could send that patch. It will help us make a
> > flashrom design decision that works for all boards with multiple flash
> > chips.
>
> OK, I'm sending the patch for flashrom - but it is based on RE and needs
> careful testing.
> It works fine on GA-MA770-UD3 (rev 1.0) but should be tested with other MBs.
> Anyway I hope it should work for all latest Gigabyte MBs with dual bios
> chips connected through IT8720/18.
> I think it's not ready for inclusion in flashrom - it probably should be
> done in more safe and generic way.
> Probably it should check for default values in regs etc.
>
> It may be used as following:
>
> flashrom -p it87spi:gbdualindex=0 ...
> flashrom -p it87spi:gbdualindex=1 ...
>
> This value needs to be set only once and further ops on selected chip
> can be performed without parameters.
> Index: it87spi.c
>
> --- it87spi.c	(revision 992)
> +++ it87spi.c	(working copy)
> @@ -155,6 +155,28 @@
>    			sio_write(port, 0x65, (flashport & 0xff));
>    			free(portpos);
>    		}
> 	portpos = extract_param(&programmer_param,
> 				"gbdualindex=", ",:");
> 	if (portpos) {
> 		int chip_index = strtol(portpos, (char **)NULL, 0);
> 		if ((chip_index!=0) && (chip_index!=1))	{
> 			msg_perr("Dual bios: Invalid chip index requested: %d\n",chip_index);
> 			flashport=0;
> 		} else {
> 			tmp=sio_read(port,0xEF);
> 			msg_pinfo("Dual bios: Current chip : %d\n",tmp&1);
> 			if (chip_index!=(tmp&1)) {
> 				sio_write(port,0xEF,(tmp&0xFE)|chip_index);
> 				tmp=sio_read(port,0xEF)&1;
> 				if (tmp!=chip_index) {
> 					msg_perr("Dual bios: Chip selection failed.\n");
> 					flashport=0;
> 				} else msg_pinfo("Dual bios: Selected chip: %d\n",tmp&1);
> 			}
> 		}
> 		free(portpos);
> 	}
> }
> exit_conf_mode_ite(port);
> break;

Hello Vadim!

Someone has tested a refinement of your patch with an ITE 8728 and it
(still) worked fine. I'd like to request a sign-off from you in case we
want to eventually integrate it. (see
http://www.coreboot.org/Development_Guidelines#Sign-off_Procedure for details).
Thanks!