[PATCH] drm/[amdgpu|radeon]: fix memset on io mem

Thu Dec 17 13:45:06 UTC 2020

On 2020-12-17 10:25, Christian König wrote:
> Am 17.12.20 um 02:07 schrieb Chen Li:
>> On Wed, 16 Dec 2020 22:19:11 +0800,
>> Christian König wrote:
>>> Am 16.12.20 um 14:48 schrieb Chen Li:
>>>> On Wed, 16 Dec 2020 15:59:37 +0800,
>>>> Christian König wrote:
>>>>> [SNIP]
>>>> Hi, Christian. I'm not sure why this change is a hack here. I cannot 
>>>> see the problem and wll be grateful if you give more explainations.
>>> __memset is supposed to work on those addresses, otherwise you can't 
>>> use the
>>> e8860 on your arm64 system.
>> If __memset is supposed to work on those adresses, why this 
>> commit(https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Ftorvalds%2Flinux%2Fcommit%2Fba0b2275a6781b2f3919d931d63329b5548f6d5f&data=04%7C01%7Cchristian.koenig%40amd.com%7C4ed3c075888746b7f41408d8a22811c5%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637437640274023350%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=HhWxUaLo3WpzoV6hjV%2BG1HICaIOXwsoNpzv5tNMNg8A%3D&reserved=0) 
>> is needed? (I also notice drm/radeon didn't take this change though) 
>> just out of curiosity.
> 
> We generally accept those patches as cleanup in the kernel with the hope 
> that we can find a way to work around the userspace restrictions.
> 
> But when you also have this issue in userspace then there isn't much we 
> can do for you.
> 
>>> Replacing the the direct write in the kernel with calls to writel() or
>>> memset_io() will fix that temporary, but you have a more general 
>>> problem here.
>> I cannot see what's the more general problem here :( u mean performance?
> 
> No, not performance. See standards like OpenGL, Vulkan as well as VA-API 
> and VDPAU require that you can mmap() device memory and execute 
> memset/memcpy on the memory from userspace.
> 
> If your ARM base board can't do that for some then you can't use the 
> hardware with that board.

If the VRAM lives in a prefetchable PCI bar then on most sane Arm-based 
systems I believe it should be able to mmap() to userspace with the 
Normal memory type, where unaligned accesses and such are allowed, as 
opposed to the Device memory type intended for MMIO mappings, which has 
more restrictions but stricter ordering guarantees.

Regardless of what happens elsewhere though, if something is mapped 
*into the kernel* with ioremap(), then it is fundamentally wrong per the 
kernel memory model to reference that mapping directly without using I/O 
accessors. That is not specific to any individual architecture, and 
Sparse should be screaming about it already. I guess in this case the 
UVD code needs to pay more attention to whether radeon_bo_kmap() ends up 
going via ttm_bo_ioremap() or not.

(I'm assuming the initial fault was memset() with 0 trying to perform 
"DC ZVA" on a Device-type mapping from ioremap() - FYI a stacktrace on 
its own without the rest of the error dump showing what actually 
triggered it isn't overly useful)

Robin.

>>>>> For amdgpu I suggest that we allocate the UVD message in GTT 
>>>>> instead of VRAM
>>>>> since we don't have the hardware restriction for that on the new 
>>>>> generations.
>>>>>
>>>> Thanks, I will try to dig into deeper. But what's the "hardware 
>>>> restriction" meaning here? I'm not familiar with video driver stack 
>>>> and amd gpu, sorry.
>>> On older hardware (AGP days) the buffer had to be in VRAM (MMIO) 
>>> memory, but on
>>> modern system GTT (system memory) works as well.
>> IIUC, e8860 can use amdgpu(I use radeon now) beause its device id 6822 
>> is in amdgpu's table. But I cannot tell whether e8860 has iommu, and I 
>> cannot find iommu from lspci, so graphics translation table may not 
>> work here?
> 
> That is not related to IOMMU. IOMMU is a feature of the CPU/motherboard. 
> This is implemented using GTT, e.g. the VM page tables inside the GPU.
> 
> And yes it should work I will prepare a patch for it.
> 
>>>>> BTW: How does userspace work on arm64 then? The driver stack 
>>>>> usually only works
>>>>> if mmio can be mapped directly.
>>>> I also post two usespace issue on mesa, and you may be interested 
>>>> with them:
>>>>    
>>>> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgitlab.freedesktop.org%2Fmesa%2Fmesa%2F-%2Fissues%2F3954&data=04%7C01%7Cchristian.koenig%40amd.com%7C4ed3c075888746b7f41408d8a22811c5%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637437640274023350%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=ZR7pDS%2BCLUuMjCeKcMAXfHtbczt8WdUwSeLZCuHfCHw%3D&reserved=0 
>>>>
>>>>    
>>>> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgitlab.freedesktop.org%2Fmesa%2Fmesa%2F-%2Fissues%2F3951&data=04%7C01%7Cchristian.koenig%40amd.com%7C4ed3c075888746b7f41408d8a22811c5%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637437640274033344%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=jAJo3aG2I1oIDTZXWhNgcKoKbd6tTdiAtc7vE4hJJPY%3D&reserved=0 
>>>>
>>>> I paste some virtual memory map in userspace there. (and the two 
>>>> problems do bother me quite a long time.)
>>> I don't really see a solution for those problems.
>>>
>>> See it is perfectly valid for an application to memset/memcpy on 
>>> mmaped MMIO
>>> space which comes from OpenGL or Vulkan.
>>>
>>> So your CPU simply won't work with the hardware. We could work around 
>>> that with
>>> a couple of hacks, but this is a pretty much general problem.
>>>
>>> Regards,
>>> Christian.
>> Thanks! Can you provid some details about these hacks? Should I post 
>> another issue on the mail list?
> 
> Adjust the kernel and/or user space to never map VRAM to the CPU.
> 
> This violates the OpenGL/Vulkan specification in some ways. So not sure 
> if that will work or not.
> 
> Regards,
> Christian.
> _______________________________________________
> dri-devel mailing list
> dri-devel at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/dri-devel