Messages in this thread | | | Date | Tue, 4 Nov 2008 15:57:22 +0000 | From | Matthew Garrett <> | Subject | Re: Add option to passively listen for PCIE hotplug events |
| |
On Tue, Nov 04, 2008 at 03:44:23PM +0000, Alan Jenkins wrote:
> I instrumented all the delays I could find. hpc_check_lnk_status() was > the only one that came up. 1000ms delay per call, and I get 6 such delays.
Ok, which is getting called from pciehp_enable_slot(). I'm not quite clear on why this is being called 6 times - any chance you can put a dump_stack() in there? Some are probably directly from the resume code. This confuses me a little, since there's a check for whether there's a card in the slot first. If that fails, pciehp_enable_slot() isn't called. If it succeeds, pciehp_enable_slot checks whether the slot is already powered up. If it is, it bails before doing the link setup. So something funny is going on here, since on the Eee I'd expect those slots to either be empty or already powered up.
-- Matthew Garrett | mjg59@srcf.ucam.org
| |