Re: [Qemu-devel] VFIO VGA test branches

[Alex Williamson]

On Tue, 2013-05-28 at 07:33 +0200, Knut Omang wrote:
> 
> I noticed this warning in the host log - I suppose it is unrelated but
> thought I'd mention it just in case there is some side effect I do not
> understand here:
> 
> [    0.538124] IOMMU: Setting identity map for device 0000:00:1f.0 [0x0 - 
> 0xffffff]
> [    0.538619] PCI-DMA: Intel(R) Virtualization Technology for Directed I/O
> [    0.538676] ------------[ cut here ]------------
> [    0.538681] WARNING: at drivers/pci/search.c:46 
> pci_find_upstream_pcie_bridge+0x58/0x80()
> [    0.538683] Hardware name: To be filled by O.E.M.
> [    0.538685] Modules linked in:
> [    0.538687] Pid: 1, comm: swapper/0 Not tainted 3.9.0+ #1
> [    0.538689] Call Trace:
> [    0.538694]  [<ffffffff8105ed2f>] warn_slowpath_common+0x7f/0xc0
> [    0.538697]  [<ffffffff8105ed8a>] warn_slowpath_null+0x1a/0x20
> [    0.538699]  [<ffffffff8132dc28>] pci_find_upstream_pcie_bridge+0x58/0x80
> [    0.538703]  [<ffffffff8152e26b>] intel_iommu_add_device+0x4b/0x1f0
> [    0.538706]  [<ffffffff81525b30>] ? bus_set_iommu+0x60/0x60
> [    0.538708]  [<ffffffff81525b63>] add_iommu_group+0x33/0x60
> [    0.538712]  [<ffffffff813f38fd>] bus_for_each_dev+0x5d/0xa0
> [    0.538714]  [<ffffffff81525b1b>] bus_set_iommu+0x4b/0x60
> [    0.538718]  [<ffffffff81d47d61>] intel_iommu_init+0xa72/0xb9a
> [    0.538722]  [<ffffffff81d0db94>] ? memblock_find_dma_reserve+0x13d/0x13d
> [    0.538724]  [<ffffffff81d0dba7>] pci_iommu_init+0x13/0x3e
> [    0.538727]  [<ffffffff8100215a>] do_one_initcall+0x12a/0x180
> [    0.538730]  [<ffffffff81d0603b>] kernel_init_freeable+0x150/0x1df
> [    0.538732]  [<ffffffff81d0588d>] ? do_early_param+0x8c/0x8c
> [    0.538736]  [<ffffffff81646580>] ? rest_init+0x80/0x80
> [    0.538738]  [<ffffffff8164658e>] kernel_init+0xe/0xf0
> [    0.538742]  [<ffffffff8166af6c>] ret_from_fork+0x7c/0xb0
> [    0.538744]  [<ffffffff81646580>] ? rest_init+0x80/0x80
> [    0.538749] ---[ end trace f4e8b5168095f9c1 ]---


There's a bug for this:

https://bugzilla.kernel.org/show_bug.cgi?id=44881


Chances are your system includes one of the non-compliant PCIe-to-PCI
bridges that doesn't include a PCIe capability.  So long as you're not
assigning anything behind that bridge, it shouldn't matter, but I think
we'll setup the wrong grouping and use the wrong source ID for devices
behind it.  Thanks,

Alex