Re: 2.6.14-rc2-ge484585e: kexec into same kernel: irq 11 nobody cared; but ehci_hcd should

From: Denis Vlasenko
Date: Sat Sep 24 2005 - 07:35:55 EST

On Saturday 24 September 2005 14:56, thomas.mey3r@xxxxxxxx wrote:
> Hi.
>
> I played a bit with the new kexec function:
>
> when i kexec into the same kernel i get this error message:
>
> [17179593.108000] ACPI: PCI Interrupt Link [LNKD] enabled at IRQ 11
> [17179593.108000] PCI: setting IRQ 11 as level-triggered
> [17179593.108000] ACPI: PCI Interrupt 0000:00:10.3[D] -> Link [LNKD] -> GSI 11 (level, low) -> IRQ 11
> [17179593.108000] ehci_hcd 0000:00:10.3: EHCI Host Controller
> [17179593.124000] ehci_hcd 0000:00:10.3: new USB bus registered, assigned bus number 4
> [17179593.936000] irq 11: nobody cared (try booting with the "irqpoll" option)
> [17179593.936000] [<c0103f9e>] dump_stack+0x1e/0x20
> [17179593.936000] [<c013dc6b>] __report_bad_irq+0x2b/0x90
> [17179593.936000] [<c013dd90>] note_interrupt+0x90/0xf0
> [17179593.936000] [<c013d69a>] __do_IRQ+0xca/0xe0
> [17179593.936000] [<c010530c>] do_IRQ+0x1c/0x30
> [17179593.936000] [<c0103b26>] common_interrupt+0x1a/0x20
> [17179593.936000] [<c0120ffa>] do_softirq+0x2a/0x30
> [17179593.936000] [<c01210a5>] irq_exit+0x35/0x40
> [17179593.936000] [<c0105311>] do_IRQ+0x21/0x30
> [17179593.936000] [<c0103b26>] common_interrupt+0x1a/0x20
> [17179593.936000] [<c013d931>] setup_irq+0xb1/0x110
> [17179593.936000] [<c013dae6>] request_irq+0x86/0xb0
> [17179593.936000] [<ec99c344>] usb_add_hcd+0x234/0x3a0 [usbcore]
> [17179593.936000] [<ec9a3a59>] usb_hcd_pci_probe+0x269/0x390 [usbcore]
> [17179593.936000] [<c01da239>] pci_call_probe+0x19/0x20
> [17179593.936000] [<c01da297>] __pci_device_probe+0x57/0x70
> [17179593.936000] [<c01da2df>] pci_device_probe+0x2f/0x60
> [17179593.936000] [<c021fa49>] driver_probe_device+0x39/0xc0
> [17179593.936000] [<c021fbaf>] __driver_attach+0x4f/0x60
> [17179593.936000] [<c021ef84>] bus_for_each_dev+0x54/0x80
> [17179593.936000] [<c021fbe8>] driver_attach+0x28/0x30
> [17179593.936000] [<c021f49d>] bus_add_driver+0x7d/0xe0
> [17179593.936000] [<c0220098>] driver_register+0x78/0x80
> [17179593.936000] [<c01da5d0>] pci_register_driver+0xb0/0xd0
> [17179593.936000] [<eca3d020>] init+0x20/0x26 [ehci_hcd]
> [17179593.936000] [<c0137e24>] sys_init_module+0x144/0x1c0
> [17179593.936000] [<c01030ff>] sysenter_past_esp+0x54/0x75
> [17179593.936000] handlers:
> [17179593.936000] [<ec99bf20>] (usb_hcd_irq+0x0/0x70 [usbcore])
> [17179593.936000] Disabling IRQ #11
> [17179593.972000] ehci_hcd 0000:00:10.3: irq 11, io mem 0xd0004800
> [17179593.972000] ehci_hcd 0000:00:10.3: USB 2.0 initialized, EHCI 1.00, driver 10 Dec 2004
> [17179593.976000] hub 4-0:1.0: USB hub found
> [17179593.976000] hub 4-0:1.0: 6 ports detected
>
> CPU0
> 0: 245067 XT-PIC timer
> 1: 2877 XT-PIC i8042
> 2: 0 XT-PIC cascade
> 4: 37766 XT-PIC uhci_hcd:usb1, eth0, via@pci:0000:01:00.0
> 5: 296 XT-PIC yenta, ohci1394, uhci_hcd:usb2
> 7: 2 XT-PIC parport0
> 8: 2 XT-PIC rtc
> 9: 51497 XT-PIC uhci_hcd:usb3, VIA8233
> 10: 9 XT-PIC acpi
> 11: 100000 XT-PIC ehci_hcd:usb4
> 12: 86427 XT-PIC i8042
> 14: 10409 XT-PIC ide0
> 15: 19973 XT-PIC ide1
> NMI: 0
> LOC: 0
> ERR: 0
> MIS: 0
>
> any ideas?

I suspect that interrupt happens before intr mast is written
ot corresponding register.

Add two printks, one before this writel:

printk("setting ehci->regs->intr_enable to %x\n", INTR_MASK);
writel (INTR_MASK, &ehci->regs->intr_enable); /* Turn On Interrupts */

and one here:

status &= INTR_MASK;
if (!status) { /* irq sharing? */
static int ratelimit = 20;
if(ratelimit) ratelimit--, printk("ehci_irq: IRQ_NONE\n");
spin_unlock(&ehci->lock);
return IRQ_NONE;
}

and see whether first printk happens before second.
--
vda
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/