On 1/20/25 4:54 AM, Huang, Honglei1 wrote: > On 2024/12/27 10:02, Huang, Honglei1 wrote: >> On 2024/12/22 9:59, Demi Marie Obenour wrote: >>> On 12/20/24 10:35 AM, Simona Vetter wrote: >>>> On Fri, Dec 20, 2024 at 06:04:09PM +0800, Honglei Huang wrote: >>>>> From: Honglei Huang<Honglei1.Huang@xxxxxxx> >>>>> >>>>> A virtio-gpu userptr is based on HMM notifier. >>>>> Used for let host access guest userspace memory and >>>>> notice the change of userspace memory. >>>>> This series patches are in very beginning state, >>>>> User space are pinned currently to ensure the host >>>>> device memory operations are correct. >>>>> The free and unmap operations for userspace can be >>>>> handled by MMU notifier this is a simple and basice >>>>> SVM feature for this series patches. >>>>> The physical PFNS update operations is splited into >>>>> two OPs in here. The evicted memories won't be used >>>>> anymore but remap into host again to achieve same >>>>> effect with hmm_rang_fault. >>>> So in my opinion there are two ways to implement userptr that make sense: >>>> >>>> - pinned userptr with pin_user_pages(FOLL_LONGTERM). there is not mmu >>>> notifier >>>> >>>> - unpinnned userptr where you entirely rely on userptr and do not hold any >>>> page references or page pins at all, for full SVM integration. This >>>> should use hmm_range_fault ideally, since that's the version that >>>> doesn't ever grab any page reference pins. >>>> >>>> All the in-between variants are imo really bad hacks, whether they hold a >>>> page reference or a temporary page pin (which seems to be what you're >>>> doing here). In much older kernels there was some justification for them, >>>> because strange stuff happened over fork(), but with FOLL_LONGTERM this is >>>> now all sorted out. So there's really only fully pinned, or true svm left >>>> as clean design choices imo. >>>> >>>> With that background, why does pin_user_pages(FOLL_LONGTERM) not work for >>>> you? >>> +1 on using FOLL_LONGTERM. Fully dynamic memory management has a huge cost >>> in complexity that pinning everything avoids. Furthermore, this avoids the >>> host having to take action in response to guest memory reclaim requests. >>> This avoids additional complexity (and thus attack surface) on the host side. >>> Furthermore, since this is for ROCm and not for graphics, I am less concerned >>> about supporting systems that require swappable GPU VRAM. >> Hi Sima and Demi, I totally agree the flag FOLL_LONGTERM is needed, I >> will add it in next version. And for the first pin variants >> implementation, the MMU notifier is also needed I think. >> Cause the userptr feature in UMD generally used like this: the >> registering of userptr >> always is explicitly invoked by user code like >> "registerMemoryToGPU(userptrAddr, ...)", >> but for the userptr release/free, there is no explicit API for it, at >> least in hsakmt/KFD stack. >> User just need call system call "free(userptrAddr)", thenkernel driver >> will release the userptr >> by MMU notifier callback.Virtio-GPU has no other way to know if user has >> been free the userptr >> except for MMU notifior.And in UMD theres is no way to get the free() >> operation is invoked by user. >> the only way is use MMU notifierin virtio-GPU driver and free the >> corresponding data in host >> by some virtio CMDs as far as I can see. >> And for the second way that is use hmm_range_fault, there is a >> predictable issues as far as I can see, at least in hsakmt/KFD stack. >> That is the memory may migrate when GPU/device is working. In bare >> metal, when memory is migrating KFD driver will pause the compute work >> of the device in mmap_wirte_lock then use hmm_range_fault to remap the >> migrated/evicted memories to GPU then restore the compute work of device >> to ensure the correction of the data. But in virtio-GPU driver the >> migration happen in guest kernel, the evict mmu notifier callback >> happens in guest, a virtio CMD can be used for notify host but as lack >> of mmap_write_lock protection in host kernel, host will hold invalid >> data for a short period of time, this may lead to some issues. And it is >> hard to fix as far as I can see. >> Finally I will extract some APIs into helper according to your request, >> and I will refactor the whole userptr >> implementation, use some callbacks in page getting path, let the pin >> method and hmm_range_fault can be choiced >> in this series patches. >> >> Regards, >> Honglei > > Hi Sima, > > I modified the code, remove all the MMU nitifior and use > pin_user_pages_fast only. Under this implementation userptr fully > managed by UMD. We did a performance test, it decreased by 30% in > OpenCL stack in Geekbench6 benmark. > We use AMD V2000 for test: > use MMU notifior + pin_user_pages: > near 13000 score: https://browser.geekbench.com/v6/compute/3257793 > > use pin_user_pages only: > near 10000 socre: https://browser.geekbench.com/v6/compute/3496228 > > The code is under clean up, I will send out later. > > And I found a another thing, it seems like in intel i915 userptr > implementation, the pin_user_pages is also used in MMU notifior. > Code path is: > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/i915/gem/i915_gem_userptr.c?h=v6.13#:~:text=ret%20%3D%20pin_user_pages_fast(obj%2D%3Euserptr.ptr%20%2B%20pinned%20*%20PAGE_SIZE%2C > > Patch set: > https://lore.kernel.org/all/159353552439.22701.14005121342739071590@xxxxxxxxxxxxxxxxxxxxxx/T/ > https://patchwork.kernel.org/project/intel-gfx/patch/20210323155059.628690-17-maarten.lankhorst@xxxxxxxxxxxxxxx/#24064663 > > And I didn't find the hmm_range_fault code path, maybe I missed it? A 30% performance penalty is consistent with the GPU being forced to use 4K pages instead of its preferred 2M huge pages. AMD GPUs have TLBs that are optimized for 2M pages, so using 4K pages will cause lots of TLB misses. CC Christian König who pointed out that a highly fragmented physical address space is a bad idea if you care about GPU performance. -- Sincerely, Demi Marie Obenour (she/her/hers)
