<html> <head> <meta http-equiv="Content-Type" content="text/html; charset=windows-1252"> </head> <body text="#000000" bgcolor="#FFFFFF"> <div class="moz-cite-prefix">Checking more code and documentation and thinking about it over my vacation I think I have some new conclusions here.<br> <br> Currently we are using get_user_pages() together with an MMU notifier to guarantee coherent address space view, because get_user_pages() works by grabbing a reference to the pages and ignoring concurrent page table updates.<br> <br> But HMM uses a different approach by checking the address space for modifications using hmm_vma_range_done() and re-trying when the address space has changed.<br> <br> Now what you are trying to do is to change that into get_user_pages() and HMM callback and this is what won't work. We can either use get_user_pages() with MMU notifier or we can use HMM for the work, but we can't mix and match.<br> <br> So my initial guess was correct that we just need to change both sides of the implementation at the same time.<br> <br> Regards,<br> Christian.<br> <br> Am 28.09.2018 um 17:13 schrieb Koenig, Christian:<br> </div> <blockquote type="cite" cite="mid:b8686e6b-0c3e-4feb-afbd-80397aac31a0@email.android.com"> <meta http-equiv="Content-Type" content="text/html; charset=windows-1252"> <meta content="text/html; charset=Windows-1252"> <div dir="auto">No it definitely isn't. <div dir="auto"><br> </div> <div dir="auto">We have literally worked month on this with the core MM developers.</div> <div dir="auto"><br> </div> <div dir="auto">Making sure that we have a consistent page array is absolutely vital for correct operation.</div> <div dir="auto"><br> </div> <div dir="auto">Please also check Jerome's presentation from XDC it also perfectly explains why this approach won't work correctly.</div> <div dir="auto"><br> </div> <div dir="auto">Christian.</div> </div> <div class="gmail_extra"><br> <div class="gmail_quote">Am 28.09.2018 17:07 schrieb "Yang, Philip" <a class="moz-txt-link-rfc2396E" href="mailto:Philip.Yang@amd.com"><Philip.Yang@amd.com></a>:<br type="attribution"> </div> </div> <div> <div class="moz-cite-prefix">For B path, we take mm->mmap_sem, then call hmm_vma_get_pfns() or get_user_pages(). This is obvious.<br> <br> For A path, mmu notifier mmu_notifier_invalidate_range_start()/mmu_notifier_invalidate_range_end() is called in many places, and the calling path is quit complicated inside mm, it's not obvious. I checked many of the them, for example:<br> <br> do_munmap() <br> down_write(&mm->mmap_sem)<br> arch_unmap()<br> mpx_notify_unmap()...<br> zap_bt_entries_mapping()<br> zap_page_range()<br> up_write(&mm->mmap_sem)<br> <br> void zap_page_range(struct vm_area_struct *vma, unsigned long start,<br> unsigned long size)<br> {<br> struct mm_struct *mm = vma->vm_mm;<br> struct mmu_gather tlb;<br> unsigned long end = start + size;<br> <br> lru_add_drain();<br> tlb_gather_mmu(&tlb, mm, start, end);<br> update_hiwater_rss(mm);<br> mmu_notifier_invalidate_range_start(mm, start, end);<br> for ( ; vma && vma->vm_start < end; vma = vma->vm_next)<br> unmap_single_vma(&tlb, vma, start, end, NULL);<br> mmu_notifier_invalidate_range_end(mm, start, end);<br> tlb_finish_mmu(&tlb, start, end);<br> }<br> <br> So AFAIK it's okay without invalidate_range_end() callback. <br> <br> Regards,<br> Philip<br> <br> On 2018-09-28 01:25 AM, Koenig, Christian wrote:<br> </div> <blockquote type="cite"> <meta content="text/html; charset=utf-8"> <div dir="auto">No, that is incorrect as well :) <div dir="auto"><br> </div> <div dir="auto">The mmap_sem isn't necessary taken during page table updates.</div> <div dir="auto"><br> </div> <div dir="auto">What you could do is replace get_user_pages() directly with HMM. If I'm not completely mistaken that should work as expected.</div> <div dir="auto"><br> </div> <div dir="auto">Christian.</div> </div> <div class="gmail_extra"><br> <div class="gmail_quote">Am 27.09.2018 22:18 schrieb "Yang, Philip" <a class="moz-txt-link-rfc2396E" href="mailto:Philip.Yang@amd.com" moz-do-not-send="true"> <Philip.Yang@amd.com></a>:<br type="attribution"> </div> </div> <div> <div class="moz-cite-prefix">I was trying to understand the way how HMM handle this concurrent issue and how we handle it in amdgpu_ttm_tt_userptr_needs_pages() and amdgpu_ttm_tt_affect_userptr(). HMM uses range->valid flag, we use gtt->mmu_invalidations and gtt->last_set_pages. Both use the same lock plus flag idea actually.<br> <br> Thanks for the information, now I understand fence ttm_eu_fence_buffer_objects() put to BOs will block CPU page table update. This is another side of this concurrent issue I didn't know.<br> <br> I had same worry that it has issue without invalidate_range_end() callback as the calling sequence Felix lists. Now I think it's fine after taking a look again today because of mm->mmap_sem usage, this is my understanding:<br> <br> A path:<br> <br> down_write(&mm->mmap_sem);<br> mmu_notifier_invalidate_range_start()<br> take_lock()<br> release_lock()<br> CPU page table update<br> mmu_notifier_invalidate_range_end()<br> up_write(&mm->mmap_sem);<br> <br> B path:<br> <br> again: <br> down_read(&mm->mmap_sem);<br> hmm_vma_get_pfns()<br> up_read(&mm->mmap_sem);<br> ....<br> ....<br> take_lock()<br> if (!hmm_vma_range_done()) {<br> release_lock()<br> goto again<br> }<br> submit command job...<br> release_lock()<br> <br> If you agree, I will submit patch v5 with some minor changes, and submit another patch to replace get_user_page() with HMM.<br> <br> Regards,<br> Philip <br> <br> On 2018-09-27 11:36 AM, Christian König wrote:<br> </div> <blockquote type="cite"> <div class="moz-cite-prefix">Yeah, I've read that as well.<br> <br> My best guess is that we just need to add a call to hmm_vma_range_done() after taking the lock and also replace get_user_pages() with hmm_vma_get_pfns().<br> <br> But I'm still not 100% sure how all of that is supposed to work together.<br> <br> Regards,<br> Christian.<br> <br> Am 27.09.2018 um 16:50 schrieb Kuehling, Felix:<br> </div> <blockquote type="cite"> <meta name="Generator" content="Microsoft Word 15 (filtered medium)"> <style>  </style> <div class="WordSection1"> <p class="MsoNormal"><span style="color:windowtext">I think the answer is here: <a href="https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/vm/hmm.rst#n216" moz-do-not-send="true"> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/vm/hmm.rst#n216</a></span></p> <p class="MsoNormal"><span style="color:windowtext"> </span></p> <p class="MsoNormal"><span style="color:windowtext">Regards,</span></p> <p class="MsoNormal"><span style="color:windowtext"> Felix</span></p> <p class="MsoNormal"><span style="color:windowtext"> </span></p> <div> <div style="border:none; border-top:solid #E1E1E1 1.0pt; padding:3.0pt 0in 0in 0in"> <p class="MsoNormal"><b><span style="color:windowtext">From:</span></b><span style="color:windowtext"> Koenig, Christian <br> <b>Sent:</b> Thursday, September 27, 2018 10:30 AM<br> <b>To:</b> Kuehling, Felix <a class="moz-txt-link-rfc2396E" href="mailto:Felix.Kuehling@amd.com" moz-do-not-send="true"> <Felix.Kuehling@amd.com></a><br> <b>Cc:</b> <a class="moz-txt-link-abbreviated" href="mailto:j.glisse@gmail.com" moz-do-not-send="true">j.glisse@gmail.com</a>; Yang, Philip <a class="moz-txt-link-rfc2396E" href="mailto:Philip.Yang@amd.com" moz-do-not-send="true"><Philip.Yang@amd.com></a>; <a class="moz-txt-link-abbreviated" href="mailto:amd-gfx@lists.freedesktop.org" moz-do-not-send="true">amd-gfx@lists.freedesktop.org</a><br> <b>Subject:</b> RE: [PATCH] drm/amdgpu: use HMM mirror callback to replace mmu notifier v4</span></p> </div> </div> <p class="MsoNormal"> </p> <div> <p class="MsoNormal"><span style="color:windowtext">At least with get_user_pages() that is perfectly possible. </span></p> <div> <p class="MsoNormal"><span style="color:windowtext"> </span></p> </div> <div> <p class="MsoNormal"><span style="color:windowtext">For HMM it could be that this is prevented somehow.</span></p> </div> <div> <p class="MsoNormal"><span style="color:windowtext"> </span></p> </div> <div> <p class="MsoNormal"><span style="color:windowtext">Christian.</span></p> </div> </div> <div> <p class="MsoNormal"><span style="color:windowtext"> </span></p> <div> <p class="MsoNormal"><span style="color:windowtext">Am 27.09.2018 16:27 schrieb "Kuehling, Felix" <<a href="mailto:Felix.Kuehling@amd.com" moz-do-not-send="true">Felix.Kuehling@amd.com</a>>:</span></p> </div> </div> <div> <div> <p class="MsoNormal"><span style="color:windowtext">> In this case you can end up accessing pages which are invalidated while get_user_pages is in process.</span></p> <p class="MsoNormal"><span style="color:windowtext"> </span></p> <p class="MsoNormal"><span style="color:windowtext">What’s the sequence of events you have in mind? Something like this?</span></p> <p class="MsoNormal"><span style="color:windowtext"> </span></p> <ul style="margin-top:0in" type="disc"> <li class="MsoListParagraph" style="color:windowtext; margin-left:0in">Page table is updated and triggers MMU notifier </li> <li class="MsoListParagraph" style="color:windowtext; margin-left:0in">amdgpu MMU notifier runs and waits for pending CS to finish while holding the read lock </li> <li class="MsoListParagraph" style="color:windowtext; margin-left:0in">New CS starts just after invalidate_range_start MMU notifier finishes but before the page table update is done </li> <li class="MsoListParagraph" style="color:windowtext; margin-left:0in">get_user_pages returns outdated physical addresses </li> </ul> <p class="MsoNormal"><span style="color:windowtext"> </span></p> <p class="MsoNormal"><span style="color:windowtext">I hope that’s not actually possible and that get_user_pages or hmm_vma_fault would block until the page table update is done. That is, invalidate_range_start marks the start of a page table update, and while that update is in progress, get_user_pages or hmm_vma_fault block. Jerome, can you comment on that?</span></p> <p class="MsoNormal"><span style="color:windowtext"> </span></p> <p class="MsoNormal"><span style="color:windowtext">Thanks,<br> Felix</span></p> <p class="MsoNormal"><span style="color:windowtext"> </span></p> <div> <div style="border:none; border-top:solid #E1E1E1 1.0pt; padding:3.0pt 0in 0in 0in"> <p class="MsoNormal"><b><span style="color:windowtext">From:</span></b><span style="color:windowtext"> Koenig, Christian <br> <b>Sent:</b> Thursday, September 27, 2018 9:59 AM<br> <b>To:</b> Kuehling, Felix <<a href="mailto:Felix.Kuehling@amd.com" moz-do-not-send="true">Felix.Kuehling@amd.com</a>><br> <b>Cc:</b> Yang, Philip <<a href="mailto:Philip.Yang@amd.com" moz-do-not-send="true">Philip.Yang@amd.com</a>>; <a href="mailto:amd-gfx@lists.freedesktop.org" moz-do-not-send="true">amd-gfx@lists.freedesktop.org</a>; Jerome Glisse <<a href="mailto:j.glisse@gmail.com" moz-do-not-send="true">j.glisse@gmail.com</a>><br> <b>Subject:</b> RE: [PATCH] drm/amdgpu: use HMM mirror callback to replace mmu notifier v4</span></p> </div> </div> <p class="MsoNormal"> </p> <div> <p class="MsoNormal"><span style="color:windowtext">Yeah I understand that, but again that won't work. </span></p> <div> <p class="MsoNormal"><span style="color:windowtext"> </span></p> </div> <div> <p class="MsoNormal"><span style="color:windowtext">In this case you can end up accessing pages which are invalidated while get_user_pages is in process.</span></p> </div> <div> <p class="MsoNormal"><span style="color:windowtext"> </span></p> </div> <div> <p class="MsoNormal"><span style="color:windowtext">Christian.</span></p> </div> </div> <div> <p class="MsoNormal"><span style="color:windowtext"> </span></p> <div> <p class="MsoNormal"><span style="color:windowtext">Am 27.09.2018 15:41 schrieb "Kuehling, Felix" <<a href="mailto:Felix.Kuehling@amd.com" moz-do-not-send="true">Felix.Kuehling@amd.com</a>>:</span></p> </div> </div> <div> <div> <p class="MsoNormal"><span style="color:windowtext">> I’m not planning to change that. I don’t think there is any need to change it.</span></p> <p class="MsoNormal">><br> > Yeah, but when HMM doesn't provide both the start and the end hock of the invalidation this way won't work any more.<br> ><br> > So we need to find a solution for this,<br> > Christian.</p> <p class="MsoNormal"> </p> <p class="MsoNormal">My whole argument is that you don’t need to hold the read lock until the invalidate_range_end. Just read_lock and read_unlock in the invalidate_range_start function.</p> <p class="MsoNormal"> </p> <p class="MsoNormal">Regards,</p> <p class="MsoNormal" style="margin-bottom:12.0pt"> Felix</p> <p class="MsoNormal"><span style="color:windowtext"> </span></p> <div> <div style="border:none; border-top:solid #E1E1E1 1.0pt; padding:3.0pt 0in 0in 0in"> <p class="MsoNormal"><b><span style="color:windowtext">From:</span></b><span style="color:windowtext"> Koenig, Christian <br> <b>Sent:</b> Thursday, September 27, 2018 9:22 AM<br> <b>To:</b> Kuehling, Felix <<a href="mailto:Felix.Kuehling@amd.com" moz-do-not-send="true">Felix.Kuehling@amd.com</a>><br> <b>Cc:</b> Yang, Philip <<a href="mailto:Philip.Yang@amd.com" moz-do-not-send="true">Philip.Yang@amd.com</a>>; <a href="mailto:amd-gfx@lists.freedesktop.org" moz-do-not-send="true">amd-gfx@lists.freedesktop.org</a>; Jerome Glisse <<a href="mailto:j.glisse@gmail.com" moz-do-not-send="true">j.glisse@gmail.com</a>><br> <b>Subject:</b> Re: [PATCH] drm/amdgpu: use HMM mirror callback to replace mmu notifier v4</span></p> </div> </div> <p class="MsoNormal"> </p> <div> <p class="MsoNormal">Am 27.09.2018 um 15:18 schrieb Kuehling, Felix:</p> </div> <blockquote style="margin-top:5.0pt; margin-bottom:5.0pt"> <p class="MsoNormal">> The problem is here:<br> ></p> <p class="MsoNormal">> ttm_eu_fence_buffer_objects(&p->ticket, &p->validated, p->fence);</p> <p class="MsoNormal">> amdgpu_mn_unlock(p->mn); </p> <p class="MsoNormal" style="margin-bottom:12.0pt">><br> > We need to hold the lock until the fence is added to the reservation object.<br> ><br> > Otherwise somebody could have changed the page tables just in the moment between the check of amdgpu_ttm_tt_userptr_needs_pages() and adding the fence to the reservation object.</p> <p class="MsoNormal"><span style="color:windowtext">I’m not planning to change that. I don’t think there is any need to change it.</span></p> </blockquote> <p class="MsoNormal" style="margin-bottom:12.0pt"><br> Yeah, but when HMM doesn't provide both the start and the end hock of the invalidation this way won't work any more.<br> <br> So we need to find a solution for this,<br> Christian.</p> <blockquote style="margin-top:5.0pt; margin-bottom:5.0pt"> <p class="MsoNormal"><span style="color:windowtext"> </span></p> <p class="MsoNormal"><span style="color:windowtext">Regards,</span></p> <p class="MsoNormal"><span style="color:windowtext"> Felix</span></p> <p class="MsoNormal"><span style="color:windowtext"> </span></p> <div> <div style="border:none; border-top:solid #E1E1E1 1.0pt; padding:3.0pt 0in 0in 0in"> <p class="MsoNormal"><b><span style="color:windowtext">From:</span></b><span style="color:windowtext"> Koenig, Christian <br> <b>Sent:</b> Thursday, September 27, 2018 7:24 AM<br> <b>To:</b> Kuehling, Felix <a href="mailto:Felix.Kuehling@amd.com" moz-do-not-send="true"><Felix.Kuehling@amd.com></a><br> <b>Cc:</b> Yang, Philip <a href="mailto:Philip.Yang@amd.com" moz-do-not-send="true"><Philip.Yang@amd.com></a>; <a href="mailto:amd-gfx@lists.freedesktop.org" moz-do-not-send="true">amd-gfx@lists.freedesktop.org</a>; Jerome Glisse <a href="mailto:j.glisse@gmail.com" moz-do-not-send="true"><j.glisse@gmail.com></a><br> <b>Subject:</b> Re: [PATCH] drm/amdgpu: use HMM mirror callback to replace mmu notifier v4</span></p> </div> </div> <p class="MsoNormal"> </p> <div> <p class="MsoNormal">Am 27.09.2018 um 13:08 schrieb Kuehling, Felix:</p> </div> <blockquote style="margin-top:5.0pt; margin-bottom:5.0pt"> <p class="MsoNormal">> We double check that there wasn't any page table modification while we prepared the submission and restart the whole process when there actually was some update.<br> ><br> > The reason why we need to do this is here:<br> ></p> <p class="MsoNormal">> ttm_eu_fence_buffer_objects(&p->ticket, &p->validated, p->fence);<br> > amdgpu_mn_unlock(p->mn);</p> <p class="MsoNormal" style="margin-bottom:12.0pt">><br> > Only after the new fence is added to the buffer object we can release the lock so that any invalidation will now block on our command submission to finish before it modifies the page table.<br> <br> </p> <p class="MsoNormal"><span style="color:windowtext">I don’t see why this requires holding the read-lock until invalidate_range_end. amdgpu_ttm_tt_affect_userptr gets called while the mn read-lock is held in invalidate_range_start notifier.</span></p> </blockquote> <p class="MsoNormal" style="margin-bottom:12.0pt"><br> That's not related to amdgpu_ttm_tt_affect_userptr(), this function could actually be called outside the lock.<br> <br> The problem is here:</p> <blockquote style="margin-top:5.0pt; margin-bottom:5.0pt"> <p class="MsoNormal">ttm_eu_fence_buffer_objects(&p->ticket, &p->validated, p->fence);</p> </blockquote> <blockquote style="margin-top:5.0pt; margin-bottom:5.0pt"> <p class="MsoNormal">amdgpu_mn_unlock(p->mn); </p> </blockquote> <p class="MsoNormal" style="margin-bottom:12.0pt"><br> We need to hold the lock until the fence is added to the reservation object.<br> <br> Otherwise somebody could have changed the page tables just in the moment between the check of amdgpu_ttm_tt_userptr_needs_pages() and adding the fence to the reservation object.<br> <br> Regards,<br> Christian.<br> <br> <br> </p> <blockquote style="margin-top:5.0pt; margin-bottom:5.0pt"> <p class="MsoNormal"><span style="color:windowtext"> </span></p> <p class="MsoNormal"><span style="color:windowtext">Regards,</span></p> <p class="MsoNormal"><span style="color:windowtext"> Felix</span></p> <p class="MsoNormal"><span style="color:windowtext"> </span></p> <div> <div style="border:none; border-top:solid #E1E1E1 1.0pt; padding:3.0pt 0in 0in 0in"> <p class="MsoNormal"><b><span style="color:windowtext">From:</span></b><span style="color:windowtext"> Koenig, Christian <br> <b>Sent:</b> Thursday, September 27, 2018 5:27 AM<br> <b>To:</b> Kuehling, Felix <a href="mailto:Felix.Kuehling@amd.com" moz-do-not-send="true"><Felix.Kuehling@amd.com></a><br> <b>Cc:</b> Yang, Philip <a href="mailto:Philip.Yang@amd.com" moz-do-not-send="true"><Philip.Yang@amd.com></a>; <a href="mailto:amd-gfx@lists.freedesktop.org" moz-do-not-send="true">amd-gfx@lists.freedesktop.org</a>; Jerome Glisse <a href="mailto:j.glisse@gmail.com" moz-do-not-send="true"><j.glisse@gmail.com></a><br> <b>Subject:</b> Re: [PATCH] drm/amdgpu: use HMM mirror callback to replace mmu notifier v4</span></p> </div> </div> <p class="MsoNormal"> </p> <div> <p class="MsoNormal" style="margin-bottom:12.0pt">That is correct, but take a look what we do when after calling the amdgpu_mn_read_lock():<br> <br> <br> </p> <blockquote style="margin-top:5.0pt; margin-bottom:5.0pt"> <p class="MsoNormal"> /* No memory allocation is allowed while holding the mn lock */<br> amdgpu_mn_lock(p->mn);<br> amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {<br> struct amdgpu_bo *bo = ttm_to_amdgpu_bo(e->tv.bo);<br> <br> if (amdgpu_ttm_tt_userptr_needs_pages(bo->tbo.ttm)) {<br> r = -ERESTARTSYS;<br> goto error_abort;<br> }<br> }</p> </blockquote> <p class="MsoNormal" style="margin-bottom:12.0pt"><br> We double check that there wasn't any page table modification while we prepared the submission and restart the whole process when there actually was some update.<br> <br> The reason why we need to do this is here:<br> <br> </p> <blockquote style="margin-top:5.0pt; margin-bottom:5.0pt"> <p class="MsoNormal"> ttm_eu_fence_buffer_objects(&p->ticket, &p->validated, p->fence);<br> amdgpu_mn_unlock(p->mn);</p> </blockquote> <p class="MsoNormal"><br> Only after the new fence is added to the buffer object we can release the lock so that any invalidation will now block on our command submission to finish before it modifies the page table.<br> <br> The only other option would be to add the fence first and then check if there was any update to the page tables.<br> <br> The issue with that approach is that adding a fence can't be made undone, so if we find that there actually was an update to the page tables we would need to somehow turn the CS into a dummy (e.g. overwrite all IBs with NOPs or something like that) and still submit it.<br> <br> Not sure if that is actually possible.<br> <br> Regards,<br> Christian.<br> <br> Am 27.09.2018 um 10:47 schrieb Kuehling, Felix:</p> </div> <blockquote style="margin-top:5.0pt; margin-bottom:5.0pt"> <p class="MsoNormal">So back to my previous question:</p> <p class="MsoPlainText"> </p> <p class="MsoPlainText">>> But do we really need another lock for this? Wouldn't the </p> <p class="MsoPlainText">>> re-validation of userptr BOs (currently calling get_user_pages) force </p> <p class="MsoPlainText">>> synchronization with the ongoing page table invalidation through the </p> <p class="MsoPlainText">>> mmap_sem or other MM locks?</p> <p class="MsoPlainText">> </p> <p class="MsoPlainText">> No and yes. We don't hold any other locks while doing command submission, but I expect that HMM has its own mechanism to prevent that.</p> <p class="MsoPlainText">> </p> <p class="MsoPlainText">> Since we don't modify amdgpu_mn_lock()/amdgpu_mn_unlock() we are certainly not using this mechanism correctly.</p> <p class="MsoNormal"> </p> <p class="MsoNormal">The existing amdgpu_mn_lock/unlock should block the MMU notifier while a command submission is in progress. It should also block command submission while an MMU notifier is in progress.</p> <p class="MsoNormal"> </p> <p class="MsoNormal">What we lose with HMM is the ability to hold a read-lock for the entire duration of the invalidate_range_start until invalidate_range_end. As I understand it, that lock is meant to prevent new command submissions while the page tables are being updated by the kernel. But my point is, that get_user_pages or hmm_vma_fault should do the same kind of thing. Before the command submission can go ahead, it needs to update the userptr addresses. If the page tables are still being updated, it will block there even without holding the amdgpu_mn_read_lock.</p> <p class="MsoNormal"> </p> <p class="MsoNormal">Regards,</p> <p class="MsoNormal"> Felix</p> <p class="MsoNormal"> </p> <div> <div style="border:none; border-top:solid #E1E1E1 1.0pt; padding:3.0pt 0in 0in 0in"> <p class="MsoNormal"><b>From:</b> Koenig, Christian <br> <b>Sent:</b> Thursday, September 27, 2018 3:00 AM<br> <b>To:</b> Kuehling, Felix <a href="mailto:Felix.Kuehling@amd.com" moz-do-not-send="true"><Felix.Kuehling@amd.com></a><br> <b>Cc:</b> Yang, Philip <a href="mailto:Philip.Yang@amd.com" moz-do-not-send="true"><Philip.Yang@amd.com></a>; <a href="mailto:amd-gfx@lists.freedesktop.org" moz-do-not-send="true">amd-gfx@lists.freedesktop.org</a>; Jerome Glisse <a href="mailto:j.glisse@gmail.com" moz-do-not-send="true"><j.glisse@gmail.com></a><br> <b>Subject:</b> RE: [PATCH] drm/amdgpu: use HMM mirror callback to replace mmu notifier v4</p> </div> </div> <p class="MsoNormal"> </p> <div> <div> <p class="MsoNormal">No, that won't work. We would still run into lock inversion problems.</p> <div> <p class="MsoNormal"> </p> </div> <div> <p class="MsoNormal">What we could do with the scheduler is to turn submissions into dummies if we find that the page tables are now outdated.</p> </div> <div> <p class="MsoNormal"> </p> </div> <div> <p class="MsoNormal">But that would be really hacky and I'm not sure if that would really work in all cases.</p> </div> <div> <p class="MsoNormal"> </p> </div> <div> <p class="MsoNormal">Christian.</p> </div> </div> <div> <p class="MsoNormal"> </p> <div> <p class="MsoNormal">Am 27.09.2018 08:53 schrieb "Kuehling, Felix" <<a href="mailto:Felix.Kuehling@amd.com" moz-do-not-send="true">Felix.Kuehling@amd.com</a>>:</p> </div> </div> </div> <div> <p class="MsoNormal">I had a chat with Jerome yesterday. He pointed out that the new blockable parameter can be used to infer whether the MMU notifier is being called in a reclaim operation. So if blockable==true, it should even be safe to take the BO reservation lock without problems. I think with that we should be able to remove the read-write locking completely and go back to locking (or try-locking for blockable==false) the reservation locks in the MMU notifier?<br> <br> Regards,<br> Felix<br> <br> -----Original Message-----<br> From: amd-gfx <<a href="mailto:amd-gfx-bounces@lists.freedesktop.org" moz-do-not-send="true">amd-gfx-bounces@lists.freedesktop.org</a>> On Behalf Of Christian König<br> Sent: Saturday, September 15, 2018 3:47 AM<br> To: Kuehling, Felix <<a href="mailto:Felix.Kuehling@amd.com" moz-do-not-send="true">Felix.Kuehling@amd.com</a>>; Yang, Philip <<a href="mailto:Philip.Yang@amd.com" moz-do-not-send="true">Philip.Yang@amd.com</a>>; <a href="mailto:amd-gfx@lists.freedesktop.org" moz-do-not-send="true">amd-gfx@lists.freedesktop.org</a>; Jerome Glisse <<a href="mailto:j.glisse@gmail.com" moz-do-not-send="true">j.glisse@gmail.com</a>><br> Subject: Re: [PATCH] drm/amdgpu: use HMM mirror callback to replace mmu notifier v4<br> <br> Am 14.09.2018 um 22:21 schrieb Felix Kuehling:<br> > On 2018-09-14 01:52 PM, Christian König wrote:<br> >> Am 14.09.2018 um 19:47 schrieb Philip Yang:<br> >>> On 2018-09-14 03:51 AM, Christian König wrote:<br> >>>> Am 13.09.2018 um 23:51 schrieb Felix Kuehling:<br> >>>>> On 2018-09-13 04:52 PM, Philip Yang wrote:<br> >>>>> [SNIP]<br> >>>>>> + amdgpu_mn_read_unlock(amn);<br> >>>>>> +<br> >>>>> amdgpu_mn_read_lock/unlock support recursive locking for multiple <br> >>>>> overlapping or nested invalidation ranges. But if you'r locking <br> >>>>> and unlocking in the same function. Is that still a concern?<br> >>> I don't understand the possible recursive case, but<br> >>> amdgpu_mn_read_lock() still support recursive locking.<br> >>>> Well the real problem is that unlocking them here won't work.<br> >>>><br> >>>> We need to hold the lock until we are sure that the operation which <br> >>>> updates the page tables is completed.<br> >>>><br> >>> The reason for this change is because hmm mirror has <br> >>> invalidate_start callback, no invalidate_end callback<br> >>><br> >>> Check mmu_notifier.c and hmm.c again, below is entire logic to <br> >>> update CPU page tables and callback:<br> >>><br> >>> mn lock amn->lock is used to protect interval tree access because <br> >>> user may submit/register new userptr anytime.<br> >>> This is same for old and new way.<br> >>><br> >>> step 2 guarantee the GPU operation is done before updating CPU page <br> >>> table.<br> >>><br> >>> So I think the change is safe. We don't need hold mn lock until the <br> >>> CPU page tables update is completed.<br> >> No, that isn't even remotely correct. The lock doesn't protects the <br> >> interval tree.<br> >><br> >>> Old:<br> >>> 1. down_read_non_owner(&amn->lock)<br> >>> 2. loop to handle BOs from node->bos through interval tree<br> >>> amn->object nodes<br> >>> gfx: wait for pending BOs fence operation done, mark user <br> >>> pages dirty<br> >>> kfd: evict user queues of the process, wait for queue <br> >>> unmap/map operation done<br> >>> 3. update CPU page tables<br> >>> 4. up_read(&amn->lock)<br> >>><br> >>> New, switch step 3 and 4<br> >>> 1. down_read_non_owner(&amn->lock)<br> >>> 2. loop to handle BOs from node->bos through interval tree<br> >>> amn->object nodes<br> >>> gfx: wait for pending BOs fence operation done, mark user <br> >>> pages dirty<br> >>> kfd: evict user queues of the process, wait for queue <br> >>> unmap/map operation done<br> >>> 3. up_read(&amn->lock)<br> >>> 4. update CPU page tables<br> >> The lock is there to make sure that we serialize page table updates <br> >> with command submission.<br> > As I understand it, the idea is to prevent command submission (adding <br> > new fences to BOs) while a page table invalidation is in progress.<br> <br> Yes, exactly.<br> <br> > But do we really need another lock for this? Wouldn't the <br> > re-validation of userptr BOs (currently calling get_user_pages) force <br> > synchronization with the ongoing page table invalidation through the <br> > mmap_sem or other MM locks?<br> <br> No and yes. We don't hold any other locks while doing command submission, but I expect that HMM has its own mechanism to prevent that.<br> <br> Since we don't modify amdgpu_mn_lock()/amdgpu_mn_unlock() we are certainly not using this mechanism correctly.<br> <br> Regards,<br> Christian.<br> _______________________________________________<br> amd-gfx mailing list<br> <a href="mailto:amd-gfx@lists.freedesktop.org" moz-do-not-send="true">amd-gfx@lists.freedesktop.org</a><br> <a href="https://lists.freedesktop.org/mailman/listinfo/amd-gfx" moz-do-not-send="true">https://lists.freedesktop.org/mailman/listinfo/amd-gfx</a></p> </div> </blockquote> <p class="MsoNormal"> </p> </blockquote> <p class="MsoNormal"> </p> </blockquote> <p class="MsoNormal"> </p> </div> </div> </div> </div> </div> </blockquote> <br> </blockquote> <br> </div> </blockquote> <br> </div> <br> <fieldset class="mimeAttachmentHeader"></fieldset> <br> <pre wrap="">_______________________________________________ amd-gfx mailing list <a class="moz-txt-link-abbreviated" href="mailto:amd-gfx@lists.freedesktop.org">amd-gfx@lists.freedesktop.org</a> <a class="moz-txt-link-freetext" href="https://lists.freedesktop.org/mailman/listinfo/amd-gfx">https://lists.freedesktop.org/mailman/listinfo/amd-gfx</a> </pre> </blockquote> <br> </body> </html>