LP-PHY and DMA errors: first success

LP-PHY and DMA errors: first success

[Rafał Miłecki]

Hey,

I wanted to share some great (however still early) news. For last 2
weeks I've been working on Juan's Lenovo machine to resolve DMA errors
on LP-PHY. I produced some patches, compare wl and b43 dumps, made
init in them almost identical... and nothing.

So I decided to check how much wl has to perform to make hardware
stable for b43. We disabled NetworkManager to don't up interface
automatically and just performed "modprobe wl; rmmod wl;". Thanksfully
I was enough to correctly init hardware! We got even more luck as wl
init appeared to be quite short, only about 200 lines of operations.

So I dumped wl init ops to wl.short.txt.b43 and decided to write
separate driver cloning that ops.

This way highly hackish "jc43" driver was created. I don't include
whole driver, just single file of it that's interesting for us. It's
attached scan.c.

Driver jc43 implements only about a half of wl init ops, but it
appears to be enough! Loading and unloading jc43 before using b43
seems to init hardware correctly :)

Now we just need to see what part of jc43 ops is important one and
implement it in ssb/b43 :)

-- 
Rafa?
-------------- next part --------------
A non-text attachment was scrubbed...
Name: wl.short.txt.b43
Type: application/octet-stream
Size: 7499 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/b43-dev/attachments/20110503/d926eb05/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: scan.c
Type: text/x-csrc
Size: 7330 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/b43-dev/attachments/20110503/d926eb05/attachment.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: jc43.txt.b43
Type: application/octet-stream
Size: 2625 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/b43-dev/attachments/20110503/d926eb05/attachment-0001.obj>

LP-PHY and DMA errors: first success

[Larry Finger]

On 05/03/2011 02:03 PM, Rafa? Mi?ecki wrote:
> Hey,
>
> I wanted to share some great (however still early) news. For last 2
> weeks I've been working on Juan's Lenovo machine to resolve DMA errors
> on LP-PHY. I produced some patches, compare wl and b43 dumps, made
> init in them almost identical... and nothing.
>
> So I decided to check how much wl has to perform to make hardware
> stable for b43. We disabled NetworkManager to don't up interface
> automatically and just performed "modprobe wl; rmmod wl;". Thanksfully
> I was enough to correctly init hardware! We got even more luck as wl
> init appeared to be quite short, only about 200 lines of operations.
>
> So I dumped wl init ops to wl.short.txt.b43 and decided to write
> separate driver cloning that ops.
>
> This way highly hackish "jc43" driver was created. I don't include
> whole driver, just single file of it that's interesting for us. It's
> attached scan.c.
>
> Driver jc43 implements only about a half of wl init ops, but it
> appears to be enough! Loading and unloading jc43 before using b43
> seems to init hardware correctly :)
>
> Now we just need to see what part of jc43 ops is important one and
> implement it in ssb/b43 :)

Good news indeed. :)

Larry