Re: [PATCH 1/2] pci: Block config access during BIST

[Brian King <brking@us.ibm.com>]

Benjamin Herrenschmidt wrote:
>>It still doesn't address Greg's issue about making this apply to the
>>pci_bus_* functions as well, but I'm not sure of a good way to do that
>>due to the reasons given earlier.
> 
> 
> Looks good to me, I don't sure we actually have to deal with pci_bus_*
> functions, do we ? When are they called ?

For what we are trying to solve, which is blocking userspace config
accesses, I don't think we do. Greg - are you ok with this?

>>+void pci_block_config_access(struct pci_dev *dev)
>>+{
>>+	unsigned long flags;
>>+
>>+	spin_lock_irqsave(&pci_lock, flags);
>>+	dev->block_cfg_access = 1;
>>+	spin_unlock_irqrestore(&pci_lock, flags);
>>+}
> 
> 
> Shouldn't we save the config space here ?

I thought about that when coding this up and thought it would
be better to simply have the function do what it advertises and no
more. Seems strange that a function called pci_block_config_access
would go and do a bunch of pci config accesses, but we can
certainly add it if you like.

-Brian

-- 
Brian King
eServer Storage I/O
IBM Linux Technology Center