1 |
Worse, ideally you wouldn't be using SWIOTLB but I don't know how to |
2 |
disable this without re-compiling the kernel. |
3 |
|
4 |
|
5 |
On 05/21/2017 07:12 PM, Adam Carter wrote: |
6 |
> <snip> |
7 |
> |
8 |
>>> [ 0.991863] iommu: Adding device 0000:06:00.0 to group 12 |
9 |
>>> [ 0.991982] iommu: Adding device 0000:07:04.0 to group 12 |
10 |
>>> [ 1.063849] AMD-Vi: Found IOMMU at 0000:00:00.2 cap 0x40 |
11 |
>>> [ 1.063962] AMD-Vi: Interrupt remapping enabled |
12 |
>>> [ 1.064145] AMD-Vi: Lazy IO/TLB flushing enabled |
13 |
>>> [ 1.065331] perf: AMD NB counters detected |
14 |
> q |
15 |
> I'm similar, but have a couple of extra entries. I've read a little bit |
16 |
> about them, but so far am unable to determine if their existence indicates |
17 |
> a better or worse kernel config. |
18 |
> |
19 |
> [ 1.036309] AMD-Vi: Lazy IO/TLB flushing enabled |
20 |
> [ 1.036419] PCI-DMA: Using software bounce buffering for IO (SWIOTLB) |
21 |
> [ 1.036529] software IO TLB [mem 0xba61a000-0xbe61a000] (64MB) mapped at |
22 |
> [ffffa3b87a61a000-ffffa3b87e619fff] |
23 |
> [ 1.036744] perf: AMD NB counters detected |
24 |
> |
25 |
> And the Linux AGP Driver ( in-kernel ) is working now. |
26 |
>> Now this is showing properly with lspci : |
27 |
>> 00:00.2 IOMMU: Advanced Micro Devices, Inc. [AMD/ATI] RD890S/RD990 I/O |
28 |
>> Memory Management Unit (IOMMU) |
29 |
>> |
30 |
> Same. |
31 |
> |