On Thu, 2007-11-08 at 19:22 +0100, Christoph Hellwig wrote: > Any chance we can get this whole patch series into 2.6.24? It's just > a blind cleanup of the symbols and having two new drivers that can't > be built into the kernel or loaded into the kernel is rather > embarassing. No. It's not a that "simple" cleanup. We actually found a endless recursive bug in your previous patch (already fixed in this one). There is a static iwl_init_hw_rates() in iwl4965-base.c and there is also a public iwl4965_init_hw_rates() in iwl-4965.c. When you do the string replacement, it became a endless recursive for iwl4965_init_hw_rates(). Furthermore, although both of the drivers can be built into kernel at the same time, it is useless if you do so now. Because if you make a network driver build in the kernel, it is initialized in a very early stage (before hd, sata). Currently the iwlwifi drivers (I think also other wireless drivers) load the firmware in pci_probe(), we have to move it to interface open() time to resolve this problem. But this is not so easy as it looks like: iwlwifi drivers cannot get MAC address before firmware is loaded and running, but mac80211 requires MAC address from the driver when ieee80211_register_hw() is called. ieee80211_register_hw() creates the wlan0 interface. You have no interface if you don't call ieee80211_register_hw() at pci_probe()... I'll come up a patch to delay the device MAC address get until wlan0->start() is called for mac80211. So that I can delay the firmware loading from pci probe to interface open time for iwlwifi drivers. In conclusion, the patch only makes iwlwifi drivers compile but not work for the kernel build in case. More patches will come. And we have to target it for 2.6.25. Thanks, -yi - To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
