Description of problem: On Sprucewood machine running RHEL 5.2 x86 xen kernel with one network adapter(Intel PRO/1000 PT Dual Port 39y6126) installed. If we bring up the second interface of Intel NIC like this: "ifconfig eth3 up", then the error message appears in the log file: irq 20: nobody cared (try booting with the "irqpoll" option) [<c04478aa>] __report_bad_irq+0x2b/0x69 [<c0447a9b>] note_interrupt+0x1b3/0x1ec [<c04470f5>] handle_IRQ_event+0x49/0x51 [<c0447198>] __do_IRQ+0x9b/0xd6 [<c0406e6e>] do_IRQ+0x93/0xae [<c0549913>] evtchn_do_upcall+0x64/0x9b [<c04055d9>] hypervisor_callback+0x3d/0x48 [<c0408632>] raw_safe_halt+0x8c/0xaf [<c040321a>] xen_idle+0x22/0x2e [<c0403339>] cpu_idle+0x91/0xab [<c06eb9f5>] start_kernel+0x37a/0x381 ======================= handlers: [<ee2393b7>] (e1000_intr+0x0/0xe7 [e1000e]) Disabling IRQ #20 Version-Release number of selected component (if applicable): 2.6.18-92.el5xen #1 SMP i686 GNU/Linux Machine Type = x3500 sprucewood ref.5, 7977-42Z How reproducible: Always Steps to Reproduce: 1. Install RHEL 5.2 x86 xen kernel on Sprucewood advanced ref.5 machine with Intel Network card (PRO/1000 PT Dual Port 39y6126) installed. 2. Booting into xen kernel, type command "ifconfig eth3 up" (eth3 is the second port of Intel NIC), then the error message shows up in the log file. Additional info: Passing 'noirqdebug' to dom0 kernel workaround this issue indicating that this appears to be fixed by the patch posted at: http://lists.xensource.com/archives/html/xen-devel/2006-06/msg00066.html
Looks like a problem, but I don't agree with the last statement; the patch there was never accepted upstream, but it's replacement was. The thing is, we already have the replacement in 5.2, so it must be something else. Chris Lalancette
------- Comment From anoop.vijayan.com 2009-02-07 22:33 EDT------- Redhat, FYI, This issue is also noticed in 5.3 GA. Thanks! This event sent from IssueTracker by jkachuck issue 192002
It appears this issue was resolved in RHEL 5.4