Well, I've figured out my crashing problem, but I'm not certain how to fix it... I've got a couple of choices, and my first choice doesn't seem to work too well, tho I'm not sure if that's my fault or another bug... Basically, what happens is this: The SDRAM is in two banks (0,2) in equal parts. So, on our 128MByte boards, it's two banks of 64MBytes. On our 512MByte boards, it's 2 banks of 256MBytes each. The bridge is programmed to place the first bank at 0x0, and the other bank at 256MBytes. add_memory_region() is called earlier in the boot sequence to set up the first 64MBytes at address 0. Now we call add_memory_region() with incorrect parameters. So, I thought to myself, let's just change the 'start' address of the add_memory_region() call. For good luck, I even threw in some calls with BOOT_MEM_RESERVED, so we now have (printed on bootup): Determined physical RAM map: memory: 04000000 @ 00000000 (usable) memory: 0c000000 @ 04000000 (reserved) memory: 04000000 @ 10000000 (usable) memory: 0c000000 @ 14000000 (reserved) Which looks okay to me. The problem is, my ethernet driver has gone to the dogs. It works, but it's _really_ slow and the console is printing out messages like: -- Matthew D. Dharm Senior Software Designer Momentum Computer Inc. 1815 Aston Ave. Suite 107 (760) 431-8663 X-115 Carlsbad, CA 92008-7310 Momentum Works For You www.momenco.com > -----Original Message----- > From: Pete Popov [mailto:ppopov@pacbell.net] > Sent: Wednesday, January 30, 2002 5:35 PM > To: Matthew Dharm > Cc: linux-mips > Subject: RE: More data: I've made a CVS build that doesn't crash! > > > On Wed, 2002-01-30 at 17:23, Matthew Dharm wrote: > > Well, I'm closer... and more confused. > > > > I've managed to make a 2.4.3 build which does not exhibit > any of the > > instability or crashing... but I did it by disabling half of the > > memory! > > > > In linux/arch/mips/gt64120/momenco_ocelot/setup.c is some > code to read > > a PLD and add a memory region. 64MByte is already added > much earlier, > > and now we're adding the rest. The board I'm testing on > is 128MByte, > > so it tries to add another 64MByte region which is physically > > contiguous to the first region. > > > > As far as I can tell, all of my memory works perfectly. > I'm going to > > do some more tests, but both vxWorks and OpenBSD run on this board > > without any problems. > > > > So, can anyone think of some likely culprits for what is > wrong here? > > Some piece of code which only works with addresses under 64MByte, > > perhaps? > > And 2.4.2 works with all of the memory? > > Pete >
