Up to this patch, tests did the same as without the patches,
hang after setAMDMSR , but still I think they are useful,
since they get closer to documentation and don't make things worse.
Signed off by: Xavi Drudis Ferran <xdru...@tinet.cat>
Complete code for errata 343. Revision Guide for AMD Family10h
processors (#41322) rev 3.74 June 2010 says to set the register
to 1 before CAR and to 0 after. We were setting it to 0 after CAR,
but not to 1 before.
apply after patch.rbc3inErr346
Index: src/cpu/amd/model_10xxx/defaults.h
===================================================================
--- src/cpu/amd/model_10xxx/defaults.h (revision 5692)
+++ src/cpu/amd/model_10xxx/defaults.h (working copy)
@@ -88,6 +88,10 @@
{ CPUIDFEATURES, AMD_FAM10_ALL, AMD_PTYPE_DC,
0x00000000, 1 << (33-32),
0x00000000, 1 << (33-32) }, /* [ExtendedFeatEn]=1 */
+
+ { BU_CFG2, AMD_DRBH_Cx , AMD_PTYPE_ALL,
+ 0x00000000, 1 << (35-32),
+ 0x00000000, 1 << (35-32) }, /* Erratum 343 (set to 0 after CAR, in
post_cache_as_ram() ) */
};
--
coreboot mailing list: coreboot@coreboot.org
http://www.coreboot.org/mailman/listinfo/coreboot
