From: Jouni Malinen (jkmaline_at_cc.hut.fi)
Date: 2002-07-12 06:10:44 UTC
On Thu, Jul 11, 2002 at 02:30:44PM +0200, Andreas Kotes wrote:
> no, you didn't. the driver assumes an incompatible card when it can't
> set the Vcc parameters as it expects to be able to. some ISA<->PCMCIA /
> ISA<->CardBus bridges don't support setting the voltage the same way
> notebook or PCI adapters do, so ..
> The attached patch (against current CVS) fixes the problem and makes
> hostap compatible to i82365-based (e.g. Vadem, like you can get from
> (ex-)Lucent) ISA<->PCMCIA bridges.
> P.S: I'd be happy to see this change go into the next release, along
> with a short update of the manual :)
Is this really a bug in the driver and not in the CIS data? The loop tries to find a suitable CFTABLE entry and verifies that Vcc matches. I won't be applying this kind of patch unless someone can convince me of this really being a bug in the driver code and that it is OK to accept incorrect CFTABLE entry. If the CIS were correct, there should be another similar CFTABLE entry with the correct Vcc..
There is already a module parameter ignore_cis_vcc that can be used to get the same behavior from the driver than with this patch. However, it is checked also for the default CFTABLE entry if the current CFTABLE entry does not include Vcc argument (this was missing from the patch for some reason).
-- Jouni Malinen PGP id EFC895FA