Bug 208725

Summary: IOMMU timeouts on AMD Radeon Pro W5700
Product: Drivers Reporter: Kai-Heng Feng (kai.heng.feng)
Component: PCIAssignee: drivers_pci (drivers_pci)
Status: RESOLVED CODE_FIX    
Severity: normal CC: bjorn, commandline, nutodafozo
Priority: P1    
Hardware: All   
OS: Linux   
Kernel Version: mainline Subsystem:
Regression: No Bisected commit-id:
Attachments: dmesg
lspci -vvnn

Description Kai-Heng Feng 2020-07-28 10:35:07 UTC 
ATS quirk is required otherwise the device doesn't work properly:
[    3.375841] iommu ivhd0: AMD-Vi: Event logged [IOTLB_INV_TIMEOUT device=63:00.0 address=0x42b5b01a0]
[    3.375845] iommu ivhd0: AMD-Vi: Event logged [IOTLB_INV_TIMEOUT device=63:00.0 address=0x42b5b01c0]

It's the same bug as [1] on a different model.

[1] https://gitlab.freedesktop.org/drm/amd/-/issues/1015
Comment 1 Kai-Heng Feng 2020-07-28 10:37:26 UTC 
Created attachment 290645 [details]
dmesg
Comment 2 Kai-Heng Feng 2020-07-28 10:37:48 UTC 
Created attachment 290647 [details]
lspci -vvnn
Comment 3 nutodafozo 2020-08-26 18:44:54 UTC 
I see the same on my RX 550 [1002:67ff], pci=noats also helps.
Could you also send another patch for rx550?
Comment 4 Kai-Heng Feng 2020-08-28 05:12:47 UTC 
(In reply to nutodafozo from comment #3)
> I see the same on my RX 550 [1002:67ff], pci=noats also helps.
> Could you also send another patch for rx550?

Yea, please file a new bug and attach dmesg and `sudo lspci -vvnn`.
Comment 5 Joris L. 2020-10-22 12:26:36 UTC 
also on AMD RADEON VII (VEGA20)

 iommu ivhd0: AMD-Vi: Event logged [IOTLB_INV_TIMEOUT device=30:00.0 address=0x7fb59fb70]

X370 chipset with recent bios including AGESA 1.0.0.6
Comment 6 Bjorn Helgaas 2020-12-07 20:56:46 UTC 
Kai-Heng, when you mark things resolved/fixed/etc, can you please include some reference to the change that fixed it?  E.g., attach the patch here, include a URL to a git commit, etc?  Otherwise it's really hard to figure out exactly what the fix is and which kernels have it.
Comment 7 Kai-Heng Feng 2020-12-08 01:19:01 UTC 
Sorry. It was fixed by commit 45beb31d3afb ("PCI: Mark AMD Navi10 GPU rev 0x00 ATS as broken")