1 |
On 05/21/2017 06:12 PM, Adam Carter wrote: |
2 |
> <snip> |
3 |
> |
4 |
> > [ 0.991863] iommu: Adding device 0000:06:00.0 to group 12 |
5 |
> > [ 0.991982] iommu: Adding device 0000:07:04.0 to group 12 |
6 |
> > [ 1.063849] AMD-Vi: Found IOMMU at 0000:00:00.2 cap 0x40 |
7 |
> > [ 1.063962] AMD-Vi: Interrupt remapping enabled |
8 |
> > [ 1.064145] AMD-Vi: Lazy IO/TLB flushing enabled |
9 |
> > [ 1.065331] perf: AMD NB counters detected |
10 |
> |
11 |
> |
12 |
> I'm similar, but have a couple of extra entries. I've read a little bit |
13 |
> about them, but so far am unable to determine if their existence |
14 |
> indicates a better or worse kernel config. |
15 |
> |
16 |
> [ 1.036309] AMD-Vi: Lazy IO/TLB flushing enabled |
17 |
> [ 1.036419] PCI-DMA: Using software bounce buffering for IO (SWIOTLB) |
18 |
> [ 1.036529] software IO TLB [mem 0xba61a000-0xbe61a000] (64MB) mapped |
19 |
> at [ffffa3b87a61a000-ffffa3b87e619fff] |
20 |
> [ 1.036744] perf: AMD NB counters detected |
21 |
> |
22 |
> And the Linux AGP Driver ( in-kernel ) is working now. |
23 |
> |
24 |
> Now this is showing properly with lspci : |
25 |
> 00:00.2 IOMMU: Advanced Micro Devices, Inc. [AMD/ATI] RD890S/RD990 I/O |
26 |
> Memory Management Unit (IOMMU) |
27 |
> |
28 |
> |
29 |
> Same. |
30 |
|
31 |
The (SWIOTLB) should not be the default. |
32 |
|
33 |
What kernel parameters for IOMMUs are you using now? |
34 |
|
35 |
The listed result I posted is with nothing related to IOMMUs in the |
36 |
kernel parameters, and NO GART IOMMU support compiled. |
37 |
|
38 |
|
39 |
Corbin |