On 11/04/19 02:57, Li Qiang wrote: > > > Paolo Bonzini <pbonz...@redhat.com <mailto:pbonz...@redhat.com>> 于2019 > 年4月10日周三 下午11:55写道： > > On 10/04/19 16:33, Li Qiang wrote: > > Hi all, > > > > > > > > I see the link device ‘_PRS’ uses irq line 5, 10, 11 in > > ‘build_link_dev’ function. > > > > But I never see the 5 lines uses in the guest, just uses 10 and 11. > > > > Why this happen? Maybe related with the guest? > > Because the MADT table tells the guest to only use lines 10 and 11. The > BIOS configures the chipset that way. > > > Hi Paolo, > > I read the MADT spec, and found that it may related with 'Entry Type 2 : > Interrupt Source Override'. > However, in build_madt function, I found following code when fill > interrupt source overide. > > #define ACPI_BUILD_PCI_IRQS ((1<<5) | (1<<9) | (1<<10) | (1<<11)) > if (!(ACPI_BUILD_PCI_IRQS & (1 << i))) { > /* No need for a INT source override structure. */ > continue; > } > > Seems the madt doesn't play a role? > > Also in the seabios code, I see the pci interrupt linking device is > routed hard-coded by pci_irqs. > So this means the seabios doesn't use the madt/dsdt table to configure > PCI interrupt routing?
Yeah, the MADT doesn't matter, I got confused. It's just the firmware's definition of pci_irqs that affects the behavior you're seeing. Paolo
