9 Sep
2009
9 Sep
'09
6:28 a.m.
James Cameron wrote:
On Tue, Sep 08, 2009 at 07:17:30PM -1000, david
wrote:
The usb port I'm using is on INT 21 (uhci_hcd:usb2), which doesn't share
with anything.
andy
USB not sharing interrupts with anything?
Andy's original post contained /proc/interrupts which showed some
sharing was happening. Here's a cut:
16: 380396 233180 IO-APIC-fasteoi uhci_hcd:usb4, yenta
17: 829711 532117 IO-APIC-fasteoi uhci_hcd:usb5, ohci1394, HDA Intel
18: 0 0 IO-APIC-fasteoi mmc0
19: 2 0 IO-APIC-fasteoi ehci_hcd:usb6
20: 18 17 IO-APIC-fasteoi uhci_hcd:usb1
21: 0 0 IO-APIC-fasteoi uhci_hcd:usb2
22: 305246 302929 IO-APIC-fasteoi ehci_hcd:usb3
Sharing interrupts adds a tiny delay to interrupt handling; if this has
truly become significant, there's a design issue somewhere else.
(Shared interrupts with modern kernels are not a huge problem, since the
drivers check to see if the interrupt they have is legitimate before the
kernel chains on to the next driver. It's the waste of that additional
check that people want to avoid.)