Subject: Re: New Convert/Simple Question
To: Christos Zoulas <christos@zoulas.com>
From: Christian Limpach <chris@Pin.LU>
List: port-next68k
Date: 07/11/2002 22:30:48
Quoting Christos Zoulas <christos@zoulas.com>:

> On the working scsi support, I just bit the bullet and committed
> your
> code. At least people should be able to use the SCSI support now.

good.  Thanks.

> The code needs some cleanup:
> 
> 	- indentation is inconsistent in places and lines wrap.

yes, I know, but the code that was there was already badly indented.  
I've re-indented most of the code where I made significant changes but 
left the rest alone.

> 	- some of the debugging stuff needs to be removed.

yes, I have cleanup the debugging code in my developement version 
already.  Most of it is still there but properly #ifdef'ed.

> 	- commented out lines of code should be eliminated.

definitely.  The last version I made available was very much a work in 
progress.  I have somewhat cleaned it up since.

> Do you want to work more on this?

yes.  I have some updates already.  Unfortunately I stopped keeping a 
tree which has only scsi specific changes a while ago since I made some 
bigger changes which made it too much work to keep such a tree.

(This is from looking through the diff between the last patch I made 
available and what I have now.  I haven't looked at this for a while...)
My current tree has:
- fix for machines where the network media code I added earlier didn't 
work
- using config to manage dma channels (will be needed to have an fd 
driver share a dma channel with the scsi driver)
- cleanup of the autoconfig code
- splitting the if_xe driver into MD (if_xe) and MI (mb8795) parts
- cleanup of the esp/ncr53c9x MD/MI split
- debug/trace code for the dma code
- using m68k's bus_dma instead of a private one (since they are the 
same now and it would be silly if the drifted apart)
- sync the interrupt code with the mvme68k port
- sync the pmap code with the mvme68k port
- some gratuitous changes to make code more legible
- support for keyboard leds
- lowering splclock to IPL3 (not sure about this one, but the clock 
interrupts shut out the network/scsi dma interrupts for too long)

cvs diff -bur chris-scsi-06-06 | grep '^[+-]' | wc -l says 6564


I'll try to update my tree with what you commited tomorrow or over the 
weekend.  I'll make a new patch and then we can go from there.

-- 
Christian Limpach <chris@Pin.LU>