Gentlemen, this is not strictly speaking an openwrt development problem but I figured this is the best place to raise it (feel free to hush me away if it is not).
I was trying to use ipw2200 module in an Asus WL-500GPv1 and I instantly ran
into the insmod Oops described in ticket #711. As some comment in the ticket
suggested that the problem may be fixed in a later kernel, I pulled from trunk
and tried it (it gave me the kernel 2.6.32, and the driver module from a side
source, apparently no less fresh). The problem is still there.
As per some other comment in the ticket #711, the line that causes the oops is
ipw_write32(priv, IPW_READ_INT_REGISTER,
IPW_BIT_INT_HOST_SRAM_READ_INT_REGISTER);
From a cursory look at the source, this is the only place where write is done at
the address so close to the end of the mapped region (if I interpret things
right, I am no kernel hacker myself):
pci 0000:00:02.0: reg 10 32bit mmio: [0x000000-0x000fff]
and
#define IPW_READ_INT_REGISTER 0xFF4
(all other writes are below 256 byte offset). Now, I can speculate that this may
be some architecture-specific bug in boundary checking, or something like that,
but that's just a blind shot.
The point is, that it may be the Right Thing to report the problem to the
developers of the driver, they are probably just not aware that the problem
exists. Does anyone know if ipw2200 works on ARM for *someone*? I guess not many
people try it because the driver does not support master mode anyway.
Now, I don't even know what "low-level PLL activation" means, but if I remove
the offending operation, the device actually works normally for me!
So, there are two questions:
1. Shall I file a bug report with the developers of the ipw2200 driver?
2. Should openwrt (in the meanwhile) include a patch to remove the operation?
Opinions, comments?
Thanks,
Eugene
signature.asc
Description: OpenPGP digital signature
_______________________________________________ openwrt-devel mailing list [email protected] https://lists.openwrt.org/mailman/listinfo/openwrt-devel
