[Soekris] Cold boot failures on Net5501?

[Chris Babcock]

K K wrote:
> On Sat, Jun 7, 2008 at 9:46 AM, Wim Vandeputte wrote:
>> I would blame it on the 1.33b soekris BIOS, not OpenBSD, due to the initial
>> timeout I see when detecting the Pri Mas
> 
> Soekris explains (in the URL posted previously) why they blame it on
> the flash card manufacturers.
> 
> I've been testing with a Sandisk card this afternoon, haven't had the
> problem occur in several (4+) power cycles.
> 
> This "SanDisk Extreme III model SDCFX3-002G-A21" card comes up as:
> Pri Mas  SanDisk SDCFX3-2048             LBA Xlt 992-64-63  2001 Mbyte
> 
> OpenBSD shows the following:
> $ dmesg |egrep "^(wd0|pciide)"
> pciide0 at pci0 dev 20 function 2 "AMD CS5536 IDE" rev 0x01: DMA,
> channel 0 wired
> to compatibility, channel 1 wired to compatibility
> wd0 at pciide0 channel 0 drive 0: <SanDisk SDCFX3-2048>
> wd0: 4-sector PIO, LBA, 1953MB, 4001760 sectors
> wd0(pciide0:0:0): using PIO mode 4, DMA mode 2
> pciide0: channel 1 ignored (disabled)
> $ exit
> 
> Interesting that this card shows up as "4-sector PIO" while my
> Kingston and A-Data cards were all "1-sector".
> 

I have tried many different models and brands of non-industrial flash, 
and there seems to be significant differences in what ATA capabilities 
different cards have.  In practice, the vendors don't seem to include 
the ATA interface differences in their benchmarks, and the ATA 
capabilities can make a significant difference in performance.  I have a 
Lexar card here that is supposed to be a 300x card and a SanDisk that is 
a 200x card, and the SanDisk performs about 20% faster despite the 
difference in "official" speed.  (Note: Lexar cards _SUCK_ on Soekris 
because they don't honor CS in any sane way)