[coreboot] HTX (FPGA) device needs more time for initialization butHOW?

Myles Watson mylesgw at gmail.com
Thu Jul 1 17:13:30 CEST 2010

> -----Original Message-----
> From: Knut Kujat [mailto:knuku at gap.upv.es]
> Sent: Thursday, July 01, 2010 9:00 AM
> To: Myles Watson
> Cc: 'coreboot'
> Subject: Re: [coreboot] HTX (FPGA) device needs more time for
> initialization butHOW?
> Myles Watson escribió:
> >> my version of coreboot (on Supermicro H8QME-2+) is working just
> fabulous
> >> but now I have a little issue. I have a HTX board with a FPGA on it
> >> which needs to boot with coreboot on the motherboard. The problem is
> >> that it fails to boot most of the time because it doesn't had time to
> >> initialize itself properly. I know that because at HT non coherent
> >> device initialization it marks bit 1(InitComplete) in the link type
> >> register of the device as 0. What I tried to do is to hard_reset()
> every
> >> time it fails but with no luck. I did that because after a power cycle
> >> the card seems to have enough time to initialize and the boot process
> >> completes successfully.
> >>
> >
> > Doing a hard reset works for me here.  I'm surprised that power cycling
> the
> > machine works.  That doesn't work for me.  I always figured that it was
> > because on a warm hard reset the power and clocks were already stable.
> >
> By power cycling I mean one of those where you plug the cable in again
> and the board becomes alive immediately  without any need of pushing the
> power button. After cycling you need to push the button it fails also.

That's surprising.  Maybe it has something to do with the configuration of
the power button?

> Don't know much about that FPGA because I'm not working on/with it. But
> the main idea is to modify coreboot to not to modify the FPGA.

Even if you don't modify it, knowing why it's failing from the FPGA's
perspective might be helpful.

Good luck,

More information about the coreboot mailing list