On Sat, May 12, 2007 at 06:06:57PM +0100, Steve Gane wrote: > Hi all, > > I think Apex has a problem with 64M in one bank (two chips), because when > the Slug hardware is configured for more than 64M, as it is when Apex scans, > the 64M appears (at least on my slug!) as four separate 16M regions with > 16M > in between, which fools Apex's scanner. > > I've sent Marc an email with more technical detail, and I may try to > mend it myself. > > Meanwhile, I reiterate my first question: > Has anyone used Apex 1.4.17 or 1.4.18 to boot a 64MB, two chip, FatSlug? > I suspect the answer is no. I think Apex will boot a 128MB FatSlug but > not a 64MB.
Interesting. I'll look more closely at your dump. Yes, I know that 2 chips and 128MiB works. I would have guessed that 2 cips and 64MiB would work as well, but there may be something wrong in the order of the scan. > > Steve Gane wrote: > >Hi everyone, > > > >I'm trying to use Apex 1.4.18 to boot my 64M FatSlug, but I'm not > >having much luck. > >I'm using the "proper" Debian/NSLU2 Etch. > > > >I can build Apex on a PC, swap it, prepend a header, pad it with FF, > >and write it to mtdblock2. > >(http://www.nslu2-linux.org/wiki/Debian/CompileApex and > >http://wiki.buici.com/wiki/NSLU2_Memory_Expansion_--_Fat_Slug helped a > >lot!) > >I have a serial port to observe the boot. > > > >But whatever I do, it always reports 32 MiB: > >- Whether I build Apex with 0x2000000 or 0x4000000 memory bank size > >- Whether or not I do "sdram-init; memscan -u 0+256m" or "sdram-init; > >memscan -u 0+64m" > >sdram-init always detects 1 pair of 128 Mib chips = 32 MiB > >memscan always says there's only 32MB. > > > >I used to boot with a hacked Redboot (that set the SDRAM config for > >64M) with the old > >bootstrapped Debian; now I can't install stuff because I don't have > >enough memory :-( > > > >Steve > > > > > > > -- > To UNSUBSCRIBE, email to [EMAIL PROTECTED] > with a subject of "unsubscribe". Trouble? Contact > [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
