[U-Boot-Users] Re: [patch] add support for AMD Alchemy Pb1x00 boards to u-boot-1.1.2

[U-Boot-Users] Re: [patch] add support for AMD Alchemy Pb1x00 boards to u-boot-1.1.2

[Wolfgang Denk]

In message <1106244186.8838.23.camel@kronenbourg.scs.ch> you wrote:
> 
> Here's a patch to support AMD's Pb1x00 eval board, based on the Dbau1x00
> support. I've tested it by loading it into the board SDRAM.

Added, thanks.

> I've also fixed the mii and usb commands which can dereference a null
> pointer when called without arguments.

No, this is not possible. The command definition requires  always  at
least one argument. This part rejected.

> Furthermore, I added mii routines to the au1x00 ethernet driver.

Added, thanks.

> And I had to add the objects as explicit target in the examples
> makefile, to avoid the use of a builtin make rule that goes directly
> from source to executable without going through the object files.

Fixed in the meantime.

> Note that the usb ohci driver doesn't yet work for me. Root hub seems to
> work, but any data transfer from a device fails. On the other hand,
> current linux mips 2.6 hangs when I load the ohci driver, so it doesn't
> work there too.

Added as is. I understand you will fix this eventually?

> Btw, it seems to me that the Dbau1x00 port sets up the auxilliary PLL
> way too fast.

?

Please read the Conding Style guidelines. Your patch violates several
of the rules (indentation, trailing white space, // comments).

Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Here is an Appalachian version of management's answer  to  those  who
are  concerned  with  the fate of the project: "Don't worry about the
mule. Just load the wagon."         - Mike Dennison's hillbilly uncle

[U-Boot-Users] Re: [patch] add support for AMD Alchemy Pb1x00 boards to u-boot-1.1.2

[Andrew Dyer]

On 9/24/05, Wolfgang Denk <wd@denx.de> wrote:
> > I've also fixed the mii and usb commands which can dereference a null
> > pointer when called without arguments.
>
> No, this is not possible. The command definition requires  always  at
> least one argument. This part rejected.
>

I beg to differ - I've seen this problem also.  The code in
main.c:run_command() checks for more than more than a
maximum but never checks a minimum number of arguments,
so entering mii<CR> makes cmd_mii get called with argv[0] = 'mii'
and argc = 1.  main.c:parse_line() sets argv[1] to NULL.

The code in cmd_mii.c assumes there will be a valid argv[1],
but never checks argc before using argv[1].  This leads to an
exception on machines where 0 is not a valid address (like MIPS).

--
Hardware, n.:
        The parts of a computer system that can be kicked.