Matt Schalit writes: 

> Mike Noyes wrote:
>> 
>> At 2002-02-12 16:33 +0000, Johan Ugander wrote:
>> >Charles Steinkuehler writes:
>> >>Sounds like interrupts aren't getting routed properly once linux takes
>> >>over the hardware...
>> >
>> >Yes, this seems to be the problem. I've altered every bios setting
>> >imaginable and I still can't get it to work. This is the sole problem
>> >remaining in my attempts. I've tried turning off the on-board IDE, I've
>> >tried pretty much every setting. I disabled floppy and FDC all together.
>> >It's booting right now as we speak...trudging through unnumerable
>> >'interrupt lost' listings as it's trying to get the packages up. hdk is
>> >there, and it sees it, and it accesses it, it just loses the interrupt.
>> >ANY ideas, and any experience with the problem under ANY sort of
>> >circumstances would be greatly appreciated. If you've ever recieved an
>> >'interrupt lost' output, and fixed it somehow, please tell so that I may
>> >try and learn from your solution. 
>> 
> 
> So we have several things going on here.  Let's see if we
> can summarize them, quoting Johan from his threads: 
> 
>    > I am looking to boot from a pcmcia flash card on an embedded
>    > pc (pc/104) with bios level pcmcia boot support. 
> 
>    > Oh, and by the way, my pc/104 pcmcia bridge is the tri-m aaeon pcm3115b
>    >      http://www.tri-m.com/products/aaeon/manual/pcm3115b.pdf 
> 
> 
> Ok, so he's not using a regular IBM PC/clone 486 or Pentium
> sort of mainboard, but rather a special micro mainboard called
> an embedded pc/104.  He never tells us his LEAF version.  Alrighty
> then.  Let's figure it's a 2.2 kernel, maybe Dachstein.

Yes, I am using 2.2.16, sorry about that. The system is a pentium 133 
embedded pc/104 with 20MB of RAM. I can shed more light on that if needed. 

> Later we have: 
>
>> it is a compact flash device... 8MB... a fullsize pcmcia card.
>> I also have a 16MB smaller camera-format compact flash card, 
>> which fits into a pcmcia adapter
>  
> 
> Much later we have the description of the 16 MB one
> he's trying to boot these days: 
> 
> 
>> hdk: Hitachi CV 7.1.1, ATA Disk drive
>> hdk: IRQ probe failed (0)
>>      ide5 at 0x160-0x167, 0x366 on irq 12
>> hdk: Hitachi CV 7.1.1, 15MB w/ 1kB Cache, CHS 246/4/32
>  
> 
> So he has two PC Cards.   
> 
> 
> Let's start with a definition of PCMCIA, as the folks at 
> the PCMCIA would appreciate it.  Every device of this nature 
> is a PC Card and plugs into PC Card slots. Devices and slots 
> should no longer be called PCMCIA cards or PCMCIA slots.  
> The PCMCIA requests this. 
> 
> 
> What does 'PCMCIA' mean and who is the PCMCIA? 
> 
>    Personal Computer Memory Card International Association 
> 
> and it was established in 1991 to standardize flash memory
> addin cards.  Just memory cards back then, no I/O.
>    The standards were enhanced in 1994 and to include the 
> PC CardATA specification for dealing with PC Card disk devices 
> and PC Card Flash disk devices.  That 1994 2.1 specification 
> included improvements for the Card Information Structure, too.  
> The CIS is the layer that interfaces to the mainboard bios so 
> that you can hotplug PC Card devices and get things recognized.
>    All the time these were 16-bit devices.  Then in 1995 they
> released the CardBus specification for PC Cards giving them
> a 32-bit bus interfacing directly to the PCI bus via the CardBus
> bridge.  More on that in another post.

I apologize for misusing the lingo. =[ 


> He said his card was a: 
> 
>     Hitachi CV 7.1.1, 15MB w/ 1kB Cache, CHS 246/4/32
>     
> I wonder what his feelings are on this paragraph out of the pcm3115b
> manual pdf, page 6-6, 
>
> He mentions this A: vs. D: issue later, though not in detail.
> Ok, let's wait on discussing it.

Well that "15MB" card is actually 16..and appears as C: under windows...but 
lets hold on that topic. 


> Again going back to: 
> 
>> hdk: IRQ probe failed (0)
>>      ide5 at 0x160-0x167, 0x366 on irq 12
>> hdk: Hitachi CV 7.1.1, 15MB w/ 1kB Cache, CHS 246/4/32
> 
> Using the 16MB card, he gets irq 12.  Sounds like the mouse to me.
> That can't be good. 
> 
> 
> Then he says: 
> 
>> I then rebooted on the ide'd compact flash... 8MB card on the outer
>> pcmcia slot. I got into the box, and trying to mount hdk, it gave an 
>> almost enless amount of 'hdk: lost interrupt' before working. 
>  
> 
> But he doesn't show us what interrupt it was assigned.  If it
> takes a known and spoken-for interrupt, that set's off bells and
> whistles.  12, 13, 14, 8, 6, 2, 1, are often taken or not ok to 
> use, even if the usual device is disabled.

Sorry, it was taking 12. The ide0 [actual ide, not pc card ide] was taking 
14. 

>> Conclusions
>>         -----------
>>         The bios complication of <15 being A: and >15 being C: seems to be 
>>         nonexistant. The problem was the top and bottom slot.
>  
> 
> Ok.  He noticed what I pointed out.  It's the slot that matters.
> Still, given two PC Cards he got hdk irrespective of the size.
> I don't quite get the details of this, but it may be an extension
> of his IRQ problem.  As I mentioned in another post, what happened
> to hda, hdb, hdc, hdd, hde, hdf, hdg, hdh, and hdi?  

Sorry for not responding to your other post, I missed the comments further 
down the mail. I just saw the comment on my lack-of-driving-license (which I 
now have I'm 17) and figured I'd be social as soon as I got things working. 
That sort of still holds..this problem has very high priority right now. I 
can tell all sorts of fun stories later. 

I have no idea why it starts with hdk. My guess is that it is the same 
reason it starts with ide5. 


> Moving on: 
> 
>>         The compact flash boots fine on ide as hda. 
> 
> Attaching to flash to the IDE connector he said he got: 
> 
>>   ide0 at 0x1f0-0x1f7, 0x3f6 on irq 14
> 
> That's what it's supposed to get.  
>    Ide base usually starts at 0x01f0 and is 1 byte long
>    Ide ctl is usually base+0x206
>    Ide irq is usually 12. 
> 
> Here's what gets assigned by oxygen (from syslog)
>    Feb 10 15:51:24 schalit kernel: ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
>    Feb 10 15:51:24 schalit kernel: ide1 at 0x170-0x177,0x376 on irq 15

I'm using..well actually its eigerstien at heart. Oxygen doesn't apply. 


>>         The 'lost interrupt' repeats seem to be caused by some sort of 
>>         slowness on the pcmcia bus and/or an irq problem. How can this be 
>>         addressed?
> 
> The PC Card is getting an IRQ that clobbers with the mouse IRQ12. 
> 
>    Solution 1:
>    -----------
>       Force it to use the irq you want with append parameters
>       that are used with syslinux and sent to boot via the
>       syslinux.cfg.  Something like: 
> 
>             append ide0=0x1f0,0x3f6,14

When I tried this, the pc card dissappeared all together. The pc card comes 
up on 0x160-0x167, 0x366 when it errors out. I used this to make an append 
line: 

ide5=0x160,0x3f6,14 

This had no affect -- It still went to 12. I tried with 7, it went to 12. I 
don't think this is having any affect. 

I tried 

ide0=0x160,0x3f6,14 

That didn't do anything either. It still loads on ide5, irq 12. 

The only time the ide append line did anything was when I changed the base 
to your 0x1f0. 

Any clue here?

>       (which may even be valid for 2.2.x kernels :-)
>       Check out his link for IDE boot: append parameters:
>          http://www.ibiblio.org/mdw/HOWTO/BootPrompt-HOWTO-7.html#ss7.1

On that page it mentioned only working on idex (x=0-3), which may be a 
problem. or? 

>    Solution 2:
>   ------------

I'm not using 2.4.x for other reasons. I'd really like to get this to work 
under 2.2... 


Matt, THANK YOU! 

This cleared up a lot. I feel reeeally close to a solution. So close, yet so 
far. Any ideas? 

/johan

_______________________________________________
Leaf-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/leaf-devel

Reply via email to