On Mon, Jan 13, 2014 at 06:55:50PM +0200, Michael S. Tsirkin wrote:
On Mon, Jan 13, 2014 at 11:45:40AM -0500, Kevin O'Connor wrote:
On Mon, Jan 13, 2014 at 06:36:56PM +0200, Michael S. Tsirkin wrote:
On Mon, Jan 13, 2014 at 11:31:20AM -0500, Kevin O'Connor wrote:
+void +pci_resume(void) +{
- if (!CONFIG_QEMU)
return;- if (PiixPMBDF >= 0)
so this does nothing unless piix4_pm_setup run but if it did why do we need to resume anything?
PiixPMBDF is set during the SeaBIOS init phase. The proposed pci_resume() function is called during the resume phase. The goal of the proposed patch is to store the BDF during init so that it's available during resume without requiring a bus scan.
-Kevin
Confused. I thought pci_init_device is the one calling piix4_pm_setup.
It is. Call chain on init is:
handle_post()->...->qemu_platform_setup()->pci_setup()->pci_bios_init_devices()->pci_bios_init_device()->pci_init_device()->piix4_pm_setup()
Call chain on proposed resume is:
handle_resume()->handle_resume32()->s3_resume()->pci_resume().
What am I missing?
You have me confused now. Perhaps I've misunderstood your question or misunderstood the intent of your/Marcel's original patch.
-Kevin