OK, great. If we could do something in PCI to make debugging problems like that easier, please propose it. On Mon, Sep 5, 2011 at 9:29 AM, Sagar Borikar <sagar.borikar@xxxxxxxxx> wrote: > Hi Bjorn, > > I solved the problem after further debugging. The issue was in bios as > it was programming one extra bus number when there was no bridge > present beyond the bridge. > > Thanks > Sagar > > On Sat, Sep 3, 2011 at 10:18 AM, Bjorn Helgaas <bhelgaas@xxxxxxxxxx> wrote: >> I'm sorry I haven't been able to follow up on this issue yet. Is >> there still something that isn't working right? If so, please open a >> report at http://bugzilla.kernel.org (Drivers/PCI component) and CC >> me. >> >> I'm still hoping to take a look at your dmesg log and see if we can >> improve the kernel output, but I haven't had a chance to work on it >> yet. >> >> Bjorn >> >> On Fri, Jul 1, 2011 at 7:48 PM, Sagar Borikar <sagar.borikar@xxxxxxxxx> wrote: >>> Attaching the dmesg output. Kindly pardon me for flooding the log with >>> my prints which I used for debugging. >>> >>> Thanks >>> Sagar >>> >>> On Sat, Jul 2, 2011 at 1:05 AM, Bjorn Helgaas <bhelgaas@xxxxxxxxxx> wrote: >>>> On Fri, Jul 1, 2011 at 10:24 AM, Sagar Borikar <sagar.borikar@xxxxxxxxx> wrote: >>>>> Actually I don't get any prints related to non-detected buses. But I >>>>> see that the bus number is skipped for those devices. Tried to debug >>>>> in the kernel by inserting prints in the pci scan path but I don't >>>>> even see that bus is getting created. Is it dependent on the bios >>>>> order? Looks like bios skips the device which is immediately after the >>>>> ari capable device. I also don't see any failure message in the >>>>> dmesg. The problem is parented_bus scan as well individual single >>>>> device bus scan / pci slot scan is not failing. This is x86 based >>>>> server >>>>> >>>>> lspci output is: >>>>> 08:00.0 >>>>> 0a:00.0 >>>>> 0c:00.0 >>>>> 0e:00.0 >>>>> >>>>> I can post the dmesg output with the explicitly added prints. >>>> >>>> Please do. Maybe we can use this to improve what we put in dmesg. >>>> >>>>> Is the scanning of the bus depends upon what bios exposes to the OS? >>>> >>>> The architected way is for the OS to use ACPI to discover PCI host >>>> bridges, then use native PCI enumeration below each host bridge. >>>> >>>> Bjorn >>>> >>> >> > -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
