On 03.07.2008 19:53, Peter Stuge wrote:
> On Thu, Jul 03, 2008 at 04:47:57PM +0200, Carl-Daniel Hailfinger wrote:
>
>> New iteration.
>>
>> - Clean up Geode companion chip CS5536 code.
>> - Eliminate a few redundant dev_find_pci_device() calls.
>> - Fix a compile warning intruduced in r689.
>>
>> This should be an equivalence transformation.
>>
>> Build tested on norwich, db800, alix.1c, alix.2c3, dbe62, dbe61.
>>
>> Signed-off-by: Carl-Daniel Hailfinger <[EMAIL PROTECTED]>
>>
>
> Acked-by: Peter Stuge <[EMAIL PROTECTED]>
>
Thanks!
>> @@ -659,7 +649,7 @@
>>
>> setup_i8259();
>> lpc_init(sb);
>> - uarts_init(sb);
>> + uarts_init(sb, dev);
>>
>> if (sb->enable_gpio_int_route) {
>> printk(BIOS_SPEW, "cs5536: call vr_write\n");
>>
>
> More generally to the v3 design - should we try to create finer
> granularity hooks?
>
> I'm thinking per PCI-device in each component rather than per
> component?
>
Absolutely. The ultimate goal is to kill dev_find_pci_device completely.
We have a really nice device model and we should not work around it with
v2-style code. The cleanup I did is structured in a way that allows easy
migration to per-PCI-device hooks.
Regards,
Carl-Daniel
--
http://www.hailfinger.org/
--
coreboot mailing list
[email protected]
http://www.coreboot.org/mailman/listinfo/coreboot
