18:10:15 | prpplague | howdy russ |
18:10:21 | Russ | morning |
18:10:32 | prpplague | hows it going? |
18:10:48 | Russ | ok |
18:15:12 | prpplague | got a couple of minutes for me to pick your brain? |
18:16:00 | Russ | I might disappear here and there |
18:17:27 | prpplague | ok here's the deal, i'm working on a sa-1100 board that implemented a cl-pd6722 pcmcia controler instead of the sa-1100 onboard controler |
18:17:49 | prpplague | the cl-pd6722 is supported under 2.4.x kernel, but... |
18:18:09 | prpplague | its setup up with lots of ifdef CONFIG_ISA blocks |
18:18:23 | Russ | ouch |
18:19:35 | prpplague | i don't know what to do..., should i break the code out into a seperate function, or should i write a platform specific fixup to define CONFIG_ISA, or should i replace the CONFIG_ISA with something like CONFIG_CLPD67XX ? |
18:20:32 | prpplague | any words of wisdom? |
18:20:34 | Russ | I would fix it |
18:21:16 | prpplague | use a fixup? |
18:21:51 | Russ | maybe any time it does outb, inb, chance it to an inline function, write_byte, read_byte or something |
18:21:57 | Russ | then ifdef the inline function |
18:24:09 | prpplague | so leave the the ifdef CONFIG_ISA alone, and just and a seperate ifdef for the platform to define the write/read functions? |
18:25:01 | prpplague | s/and/add |
18:39:12 | Russ | is it just inb, outb? |
18:39:31 | Russ | also, you'll probably have to change addresses from int to long |
18:39:35 | Russ | (unsigned) |
18:46:15 | prpplague | no the enitire cl-pd67xx series is only available when CONFIG_ISA is defined |
18:46:31 | Russ | thats easy to fix |
18:46:39 | Russ | its just the x86ism in the driver |
18:46:53 | Russ | like assuming the address is < 65536 |
18:47:45 | prpplague | ya thats what i thought, i just don't what to change it too, should i make it something independent like CONFIG_CLPD67XX ? |
18:48:09 | prpplague | s/don't/didn't know |
18:48:20 | prpplague | i can't type today |