Re: [PATCH 3/4] atl1: Main C file for Attansic L1 driver

[Jay Cliburn <jacliburn@bellsouth.net>]

Arjan, thank you very much for reviewing the driver.

Arjan van de Ven wrote:
> On Sun, 2007-01-21 at 15:06 -0600, Jay Cliburn wrote:
[snip]
>> +void atl1_irq_disable(struct atl1_adapter *adapter)
>> +{
>> +	atomic_inc(&adapter->irq_sem);
>> +	iowrite32(0, adapter->hw.hw_addr + REG_IMR);
>> +	synchronize_irq(adapter->pdev->irq);
>> +}
> 
> doesn't this want a PCI posting flush?
> I'm also a bit sceptical about irq_sem ...

PCI posting flush will be added.  Would you mind elaborating on your 
skepticism about irq_sem?

>> +/**
>> + * When ACPI resume on some VIA MotherBoard, the Interrupt Disable bit/0x400
>> + * on PCI Command register is disable.
>> + * The function enable this bit.
>> + * Brackett, 2006/03/15
>> + */
>> +static void atl1_via_workaround(struct atl1_adapter *adapter)
>> +{
>> +	unsigned long value;
>> +
>> +	value = ioread16(adapter->hw.hw_addr + PCI_COMMAND);
>> +	if (value & PCI_COMMAND_INTX_DISABLE)
>> +		value &= ~PCI_COMMAND_INTX_DISABLE;
>> +	iowrite32(value, adapter->hw.hw_addr + PCI_COMMAND);
>> +}
> 
> hmm I wonder if this shouldn't be a more generic PCI level quirk, not so
> much a driver level quirk...

The vendor put this code here, and we're loathe to remove it unless 
absolutely necessary.  Is it okay with you if we leave it?

Thanks,
Jay