Re: [coreboot] Problem with implementing workaround from errata

[Kyösti Mälkki]

Hi

Thanks, we can work for a clean workaround implementation once x120e
port is otherwise accepted and a board-status entry has been submitted
for it. Looks like error is not specific to x120e although it has not
been reported on other similar platforms.

A reminder here, in its current state *ALL* agesa boards will see
obsoletion in October 2017 and will be removed from master branch.
Owners of such boards should wake up and start to participate in the
review process to keep them alive... I guess I should send a separate
mail about that once again.

[1] https://www.mail-archive.com/[e-mail]/msg49234.html

Kyösti

On Sun, Jul 23, 2017 at 2:43 PM, Łukasz Dobrowolski
 wrote:
> On 07/23/2017 01:29 PM, Kyösti Mälkki wrote:
> > Hi
> >
> > Please mention the errata document number, looks like the attachment
> > was dropped.
>
> It's 47792 (AMD A45/A50M/A55E Errata).
> > Also you can just post your work-in-progress source to gerrit,
> > including the workaround you have implemented.
>
> I don't know where to start with the workaround.
> I'll clean up the code and submit it later today.
> > Kyösti
> >
> > On Sun, Jul 23, 2017 at 10:10 AM, Łukasz Dobrowolski
> >  wrote:
> > > Hi!
> > > I've resumed my work non ThinkPad X120e.
> > >
> > > I have a problem with random hangs during boot.
> > > The hang occurs in vendorcode/amd/cimx/sb800/SBPort.c:296.
> > >
> > > After a long fight I was able to diagnose the issue. It's described in
> > > errata (attached) on page 13. There is a suggested workaround, however
> > > the description is not verbose enough for me.
> > >
> > > Could somebody give me some hints?
> > >
> > > The functions are entered in this order:
> > > (?)
> > > cache_as_ram_setup: //cpu/amd/agesa/cache_as_ram.inc
> > > void * asmlinkage romstage_main(unsigned long bist)
> > > //cpu/amd/agesa/romstage.c
> > > void platform_once(struct sysinfo *cb) //cpu/amd/agesa/family14/romstage.c
> > > void sb_Poweron_Init(void) //southbridge/amd/cimx/sb800/early.c
> > > VOID sbPowerOnInit(IN AMDSBCFG* pConfig)
> > > //vendorcode/amd/cimx/sb800/SBPort.c (hang!)
> > >
> > > Where should I put the workaround if I can't pollute the shared code?
> > >
> > > The code for the port is based on asrock/e350m1. I can upload it to
> > > gerrit if needed but not much was changed.