call_pnp_bios() okay
Quoted 6 lines
Given that the args are u16s and u16s are unsigned shorts, it looks to me as if this is going to zero out all the odd-numbered args. But if that's what's happening then I'm amazed this driver works at all. I see that in some cases the odd-numbered args are zero anyway, but in others not. Result #1: The driver isn't getting a real value for the maximum node size.
But a random value will sometimes not oops the kernel.
Quoted 8 lines
Result #2: PnP BIOS is sometimes getting 0 as its DS selector Result #3: The get_dev_node config selector is always 0 (should be 1 or 2) Result #4: The set_dev_node handle is 0; but this is duplicated in thenode info structure, so the function may still work. However, the selector number of the node data is wrong
I'm off to patch this bug and see if it fixes my problem. It may fix the Sony and Dell problems too.
Well, on closer look the call_pnp_bios code is okay after all. The variables get promoted to 32 bits prior to the 16 bit shifts, despite the shifts being in parentheses. I.e, Never Mind. :)
Stelian: Sorry, I put your e-mail address in the previous subject heading by mistake.
Signature · 2 lines
Thomas Hood (Don't reply to the From: address but to jdthood_AT_yahoo.co.uk)