From: Willy Smith (numatico@racsa.co.cr)
Date: Wed Jan 17 2001 - 18:05:59 EST
John, Are you sure it's a software problem? Is this a new board design that hasn't been (software) tested before? Is there a hardware handshake implemented between the 8019 and the 32C50100? This could be the problem. In my experience, having a processor hang with just a store instruction is not a software problem. Or maybe I just am not understanding the exact sequence of events? Regards, Willy Smith Costa Rica At 01:59 PM 1/17/01 -0800, you wrote: >I'm working on a board containing an ARM7TDMI based Samsung part (the >KS32C50100-01). Hooked to the nECS2 pin is an Realtek RT8019AS ethernet >controller. > >I've been trying to get the processor to talk to the 8019, but so far all I >get are hung systems. This is the first external device I've played with -- >so I suspect I'm just doing something stoopid. > >Here's the math I've used: > >My REFEXTCON register's address is 0x3ff303c. > 0x9c218360 REFEXTCON's contents >& 0x000003ff Need 10 bits for Ext IO Bnk0 base ptr >= 0x00000360 ><< 16 From page 4-41 of KS32C50100 manual >= 0x03600000 >+ 0x00008000 My 8019 is on nECS2, so 2*0x4000 >= 0x03608000 >| 0x04000000 Noncachable area >= 0x07608000 >+ 0x00000300 Base address of NIC >= 0x07608300 This should be the cmd addr of the NIC > >I tried storing an 0x21 (E8019_NODMA+E8019_STOP) to 0x7608300. As soon as I >do the machine hangs. For grins, I tried writing the same to 0x43608300, >0x3608300 -- same results. > >I imagine I'm just blowing the math or am mis-understanding how this stuff >works. Can anyone set me straight (or point me to a document that will?) > >Thanks, >John Lombardo > >This message resent by the uclinux-dev@uclinux.org list server http://www.uClinux.org/ This message resent by the uclinux-dev@uclinux.org list server http://www.uClinux.org/
This archive was generated by hypermail 2.1.4 : Thu Sep 19 2002 - 13:19:25 EDT