On Tuesday 14 July 2009 14:31:50 Eibach, Dirk wrote:
Hmm, not sure yet what causes the problem here, but this changes in cpu/ppc4xx/start.S look fishy...
I found that the old code does not clear TLB entry #0. So this helps in my case:
Yes, I already generated exactly this patch here as well. :)
diff --git a/cpu/ppc4xx/start.S b/cpu/ppc4xx/start.S index ac96fc2..7be73a7 100644 --- a/cpu/ppc4xx/start.S +++ b/cpu/ppc4xx/start.S @@ -452,7 +452,7 @@ skip_debug_init: addis r0,0,0x0000 li r1,0x003f /* 64 TLB entries */ mtctr r1
li r4,0 /* Start with TLB #0 */
li r4,1 /* Start with TLB #1 */rsttlb: #ifdef CONFIG_SYS_RAMBOOT tlbre r3,r4,0 /* Read contents from TLB word #0 to get EPN */
But I am not sure if it's also correct for the RAMBOOT case. What's your opinion, Stefan?
Unfortunately, RAM-booting on Sequoia doesn't work in all cases (started via BDI) on Sequoia with this patch. I'm still investigating...
Best regards, Stefan
===================================================================== DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: +49-8142-66989-0 Fax: +49-8142-66989-80 Email: office@denx.de =====================================================================