Re: [PATCH v2] dma-pool: Fix too large DMA pools on medium systems

[Christoph Hellwig]

On Mon, Jun 22, 2020 at 05:07:55PM +0100, Robin Murphy wrote:
> Another angle, though, is to question why this driver is making such a 
> large allocation with GFP_ATOMIC in the first place. At a glance it looks 
> like there's no reason at all other than that it's still using the legacy 
> pci_alloc_consistent() API, since every path to that appears to have 
> CAN_SLEEP passed as its flag - modernising that would arguably be an even 
> better long-term win.

Maybe we can just try that for now?  If other problems show up we
can still increase the initial pool size later in this cycle.

I'll try to cook up a patch.