[U-Boot-Users] [PATCH-1/2] LAN91C111

[Woodruff, Richard]

Hi Stephen,

Yes I suppose.  When I snoop packets I have gotten what I expected.  Could
be luck I suppose, I'll check the address block we use.  The reset of the
changes looked reasonable....switching the one word access to a byte access
might need looking at...I seem to recall reading threads where some boards
weren't wired up to do 8 byte accesses correctly.  I'd cross check with the
kernel level driver 91c1111 or 91c9x.c, before making that change.  Might be
ok.

Regards,

Richard W.

-----Original Message-----
From: Stephan Linz [mailto:linz at mazet.de] 
Sent: Monday, November 03, 2003 7:02 AM
To: Woodruff, Richard; u-boot-users at lists.sourceforge.net
Subject: Re: [U-Boot-Users] [PATCH-1/2] LAN91C111

Am Montag,  3. November 2003 13:13 schrieb Woodruff, Richard:
> I've been using the 91111 for a while now.  I've not had any real problems
> with it.  As the environment I'm in has some 30 odd boards, I've never
used
> the hard coded mac address.

Hi Richard

It is not a problem of hard coded mac addresses because smc_mac_addr[] will 
be used as buffer storage too. Furthermore it's a problem of type casting 
from signed char to unsigned short (type of word) when writing the mac 
address into LAN91C111 (near line smc91111.c:smc_open():700 the USE_32_BIT 
branch).

Note: it could be _only_ a NIOS related problem, but I think the type 
unsigned char for smc_mac_addr[] is more exact in common case, or not?


Regards,
Stephan

>
> Regards,
>
> Richard W.
>
> -----Original Message-----
> From: u-boot-users-admin at lists.sourceforge.net
> [mailto:u-boot-users-admin at lists.sourceforge.net] On Behalf Of Stephan
Linz
> Sent: Monday, November 03, 2003 5:02 AM
> To: u-boot-users at lists.sourceforge.net
> Subject: [U-Boot-Users] [PATCH-1/2] LAN91C111
>
> Hi all,
>
> in context of one of my emails last week:
>
http://sourceforge.net/mailarchive/forum.php?thread_id=3364771&forum_id=128
>9 8
>
>
> I'm wondering about how many people are using the LAN91C111 driver without
> significant problems. There is (just was :-) a fat signed/unsigned bug in
> drivers/smc91111.c which is significant at NIOS plattforms when you are
> using
> MAC addresses with address elements above 0x7f. Example: 00:07:ed:0a:a4:7b
> goes over to 00:07:ed:ff:a4:ff inside the chip. Simple solution:
>
> 	static char smc_mac_addr[] = { . . . };
>
> have to be:
>
> 	static char unsigned smc_mac_addr[6] = { . . . };
>
>
> In attatchment you will find my smc91111_fix patch. It includes some other
> minor fixes I've made. This patch has been working successful at NIOS
> Stratix
> and Cyclone boards.
>
> Please try out and patch CVS tree.
>
>
> Best Regards,