<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <div class="moz-cite-prefix">Am 07.09.2017 um 12:14 schrieb Marek
      Olšák:<br>
    </div>
    <blockquote type="cite"
cite="mid:CAAxE2A7NoHzmO=0B1JO-Yis_kNZFabWWQbCmibTe5mwgRoOxGQ@mail.gmail.com">
      <div dir="auto">
        <div><br>
          <div class="gmail_extra"><br>
            <div class="gmail_quote">On Sep 7, 2017 12:08 PM, "Christian
              König" <<a href="mailto:deathsimple@vodafone.de"
                moz-do-not-send="true">deathsimple@vodafone.de</a>>
              wrote:<br type="attribution">
              <blockquote class="quote" style="margin:0 0 0
                .8ex;border-left:1px #ccc solid;padding-left:1ex">
                <div class="elided-text">Am 07.09.2017 um 11:23 schrieb
                  Michel Dänzer:<br>
                  <blockquote class="gmail_quote" style="margin:0 0 0
                    .8ex;border-left:1px #ccc solid;padding-left:1ex">
                    On 01/09/17 07:40 PM, Christian König wrote:<br>
                    <blockquote class="gmail_quote" style="margin:0 0 0
                      .8ex;border-left:1px #ccc solid;padding-left:1ex">
                      Am 01.09.2017 um 12:28 schrieb Michel Dänzer:<br>
                      <blockquote class="gmail_quote" style="margin:0 0
                        0 .8ex;border-left:1px #ccc
                        solid;padding-left:1ex">
                        On 01/09/17 07:23 PM, Nicolai Hähnle wrote:<br>
                        <blockquote class="gmail_quote" style="margin:0
                          0 0 .8ex;border-left:1px #ccc
                          solid;padding-left:1ex">
                          On 01.09.2017 11:58, Michel Dänzer wrote:<br>
                          <blockquote class="gmail_quote"
                            style="margin:0 0 0 .8ex;border-left:1px
                            #ccc solid;padding-left:1ex">
                            On 29/08/17 11:47 PM, Christian König wrote:<br>
                            <blockquote class="gmail_quote"
                              style="margin:0 0 0 .8ex;border-left:1px
                              #ccc solid;padding-left:1ex">
                              From: Marek Olšák <<a
                                href="mailto:marek.olsak@amd.com"
                                target="_blank" moz-do-not-send="true">marek.olsak@amd.com</a>><br>
                              <br>
                              For lower overhead in the CS ioctl.<br>
                              Winsys allocators are not used with
                              interprocess-sharable resources.<br>
                              <br>
                              v2: It shouldn't crash anymore, but the
                              kernel will reject the new<br>
                              flag.<br>
                              v3 (christian): Rename the flag, avoid
                              sending those buffers in the<br>
                              BO list.<br>
                              v4 (christian): Remove setting the kernel
                              flag for now<br>
                            </blockquote>
                            This change seems to have caused a GPU hang
                            when running piglit on my<br>
                            Kaveri with the radeon kernel driver.<br>
                          </blockquote>
                        </blockquote>
                      </blockquote>
                    </blockquote>
                    I think we can remove "seems to have". I'm still
                    reliably getting the<br>
                    GPUVM fault and hang with current master, but not if
                    I revert this<br>
                    commit (and the one after it).<br>
                    <br>
                    <blockquote class="gmail_quote" style="margin:0 0 0
                      .8ex;border-left:1px #ccc solid;padding-left:1ex">
                      <blockquote class="gmail_quote" style="margin:0 0
                        0 .8ex;border-left:1px #ccc
                        solid;padding-left:1ex">
                        <blockquote class="gmail_quote" style="margin:0
                          0 0 .8ex;border-left:1px #ccc
                          solid;padding-left:1ex">
                          <blockquote class="gmail_quote"
                            style="margin:0 0 0 .8ex;border-left:1px
                            #ccc solid;padding-left:1ex">
                            Haven't been able to isolate it to a
                            specific test, seems to only<br>
                            happen when running multiple tests
                            concurrently.<br>
                          </blockquote>
                        </blockquote>
                      </blockquote>
                    </blockquote>
                    I reproduced the problem with piglit process
                    separation enabled as well,<br>
                    and all four tests running when it hung were
                    textureGather tests.<br>
                    Before, reproducing the problem twice with piglit
                    process separation<br>
                    disabled, three textureGather tests were running
                    when it hung both times<br>
                    as well. I've been unable to reproduce the problem
                    by manually running<br>
                    the same textureGather tests in parallel though.<br>
                    <br>
                    <br>
                    <blockquote class="gmail_quote" style="margin:0 0 0
                      .8ex;border-left:1px #ccc solid;padding-left:1ex">
                      <blockquote class="gmail_quote" style="margin:0 0
                        0 .8ex;border-left:1px #ccc
                        solid;padding-left:1ex">
                        <blockquote class="gmail_quote" style="margin:0
                          0 0 .8ex;border-left:1px #ccc
                          solid;padding-left:1ex">
                          <blockquote class="gmail_quote"
                            style="margin:0 0 0 .8ex;border-left:1px
                            #ccc solid;padding-left:1ex">
                            There's a GPUVM fault before the hang, I
                            suspect it's related:<br>
                            <br>
                            radeon 0000:00:01.0: GPU fault detected: 146
                            0x0ae6760c<br>
                            radeon 0000:00:01.0: 
                             VM_CONTEXT1_PROTECTION_FAULT_<wbr>ADDR 
                             0x000001D7<br>
                            radeon 0000:00:01.0: 
                             VM_CONTEXT1_PROTECTION_FAULT_<wbr>STATUS
                            0x0607600C<br>
                            VM fault (0x0c, vmid 3) at page 471, read
                            from 'CPF' (0x43504600) (118)<br>
                            <br>
                            <br>
                            Any ideas?<br>
                          </blockquote>
                        </blockquote>
                      </blockquote>
                      Not the slightest, but I'm still investigating
                      problems with that on<br>
                      amdgpu.<br>
                      <br>
                      If we can't find the root cause till Monday it
                      might be a good idea to<br>
                      revert the patches for now.<br>
                    </blockquote>
                    What's the status on that?<br>
                  </blockquote>
                  <br>
                  <br>
                </div>
                I've found and fixed the remaining kernel bugs over the
                last weekend/beginning of this week.<br>
                <br>
                Still need to commit the fix for UVD/VCE, but that one
                shouldn't affect GFX at all.</blockquote>
            </div>
          </div>
        </div>
        <div dir="auto"><br>
        </div>
        <div dir="auto">Michel is seeing hangs on the radeon KMD, which
          should be unaffected by you kernel work I think.</div>
        <div dir="auto"><br>
        </div>
        <div dir="auto">We could revert this to unbreak Michel's Kaveri,
          but I think it shouldn't be so difficult to find the culprit
          in this patch if there is one.</div>
      </div>
    </blockquote>
    <br>
    The only crux is that the userspace patch shouldn't affect radeon at
    all. So the real question is what the heck is going on here?<br>
    <br>
    Christian.<br>
    <br>
    <blockquote type="cite"
cite="mid:CAAxE2A7NoHzmO=0B1JO-Yis_kNZFabWWQbCmibTe5mwgRoOxGQ@mail.gmail.com">
      <div dir="auto">
        <div dir="auto"><br>
        </div>
        <div dir="auto">Marek</div>
        <div dir="auto"><br>
        </div>
        <div dir="auto">
          <div class="gmail_extra">
            <div class="gmail_quote">
              <blockquote class="quote" style="margin:0 0 0
                .8ex;border-left:1px #ccc solid;padding-left:1ex"><font
                  color="#888888"><br>
                  <br>
                  Christian.<br>
                  <br>
                </font></blockquote>
            </div>
            <br>
          </div>
        </div>
      </div>
    </blockquote>
    <p><br>
    </p>
  </body>
</html>