-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hello, On Sep 11, 2008, at 7:21 AM, Marco Trillo wrote:
src/sys/arch/macppc/dev/if_bm.c function bmac_reset_chip() uses extern symbol 'heathrow_FCR' to toggle some bits in the FCR (EnetEnable, ResetEnetCell, etc.). However such 'heathrow_FCR' symbol does not seem to be initialized anywhere in arch/macppc/ nor in arch/powerpc/: $ grep -r heathrow_FCR macppc macppc/dev/if_bm.c:extern volatile uint32_t *heathrow_FCR; macppc/dev/if_bm.c: v = in32rb(heathrow_FCR); macppc/dev/if_bm.c: out32rb(heathrow_FCR, v); macppc/dev/if_bm.c: out32rb(heathrow_FCR, v); macppc/dev/if_bm.c: out32rb(heathrow_FCR, v); macppc/dev/if_bm.c: out32rb(heathrow_FCR, v); macppc/macppc/machdep.c:volatile uint32_t *heathrow_FCR = NULL; $ grep -r heathrow_FCR powerpc $ ^D Previously to ppcoea-renovation, the symbol was initialized on arch/macppc/macppc/extintr.c. A simple fix (rather ugly) would be to use 'ca->ca_baseaddr + 0x38', where 'ca' is the struct confargs passed by obio, and drop the heathrow_FCR variable.
Hmm, I'm pretty sure I at least tried to preserve it - guess it fell under the table. Also, at least my beige G3 doesn't need its bm enabled by hand - the firmware already does that. ( and hardly anybody really wants to use bm anyway if there's any alternative, at least the 10MBit/s variant sucks pretty badly )
If obio exported accessor functions for the FCRs, this would be anothercandidate to use them.
Absolutely. have fun Michael -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (Darwin) iQEVAwUBSMkanspnzkX8Yg2nAQIOzwf/YpCrWwpnGoE22lFs7+NSWLzoV7/2QEaO Di3l9XhqTJkZVvrp5PlMhjtIRpgs1aUszVkAYUfBIMx0JHxnWOH3g1yXYZXAuI7E xocwrw0y7hUb4yoi7GD42Y2BJ7LPkyll+egpNTWqZDXhOC002HTAsGM2YivbIzlk trBd+O5KQtOIF8BGLTjzkEMB+nDCfoGP+NLIk8+/LSe5zsO0Zs8OYq6Akhr8bO0T BwwZBk2bzz5Bui87hTzakaSVUU8MjvFaQZV0pRK0YeuMC7fpROgLKTU7p66KYYAH 8QSDIMgKIBGvi7YiINmCBcALK74oGPYelTgjn1HL949Qdp7AUOSEJQ== =XbLX -----END PGP SIGNATURE-----