Allocating resources to isab children

[John Baldwin]

On 5/12/11 5:45 PM, Henrik Brix Andersen wrote:
> Hi John,
>
> On Mar 14, 2011, at 15:28, John Baldwin wrote:
>> On Sunday, March 13, 2011 12:12:59 pm Henrik Brix Andersen wrote:
>>> On Mar 12, 2011, at 20:47, Henrik Brix Andersen wrote:
>>> I have a patch ready for allowing the PCI-ISA bridge to proxy SYS_RES_MEMORY
>> and SYS_RES_IOPORT resource allocations from direct children to the parent PCI
>> bus.
>>> The patch is available from here: http://people.freebsd.org/~brix/src-
>> patches/isa_pci.c.diff
>>
>> Yes, I think this looks good.
>
> Having completed the driver for the I2C host bus controller, I have revised the patch a little.
> The updated patch is available at http://people.freebsd.org/~brix/src-patches/isa_pci.c.diff

A few cosmetic nits:

- I would use 'isab_pci_*' rather than 'isa_pci_*' as it is an ISA 
bridge rather than an ISA bus device.
- You can probably leave out the blank line in isa[b]_pci_attach() 
between bus_generic_probe() and isab_attach().

Other than that, you are fine to commit this with a reviewed by from me.

> I would appreciate it if you could find the time to review this and - if possible - approve a commit of this patch.
>
> The patch for the glxiic(4) I2C host bus driver is available at http://people.freebsd.org/~brix/src-patches/glxiic.diff - if somebody could find the time to review this as well, I would be very grateful. Since I do not have a commit bit for src, I will eventually need an approval for this to enter the tree as well.

This mostly looks good to me.  I can't evaluate the actual driver logic, 
but the new-bus, etc. bits look fine.  One thing is that since you use 
callout_init_mtx(), I don't think you need any of the callout stuff at 
the start of your timeout routine (callout_pending/active/deactivate). 
With callout_init_mtx() your callout will only be called if it is 
active.  That is the only nit that I saw however.

-- 
John Baldwin