<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <div class="moz-cite-prefix">
      <blockquote type="cite">The PCI Express controller as instantiated
        on this chip does not
        support hardware coherency. All incoming PCI Express
        transactions are made non IO-coherent.
        <div><br>
        </div>
        <div>Would AMDGPU still work with that PCI Express controller,
          or is this a show-stopper?</div>
      </blockquote>
      <br>
      I'm really wondering what this comment in the documentation means.<br>
      <br>
      As far as I know PCIe doesn't support cache coherency in the
      downstream and supporting it in the up stream is a must have.<br>
      <br>
      So what exactly is meant here with IO-coherent?<br>
      <br>
      Regards,<br>
      Christian.<br>
      <br>
      Am 10.01.19 um 11:55 schrieb Bas Vermeulen:<br>
    </div>
    <blockquote type="cite"
cite="mid:CAGtoiRiPHW8a3afkO3ui5FtLP-OztBjVz6CgMgLifgLvnk4QOg@mail.gmail.com">
      <meta http-equiv="content-type" content="text/html; charset=UTF-8">
      <div dir="ltr">
        <div dir="ltr">
          <div dir="ltr">
            <div dir="ltr">
              <div dir="ltr">Hi Alex,
                <div><br>
                </div>
                <div>I've managed to get a little further. I am
                  currently running mainline (5.0.0-rc1) and am getting
                  the errors below.</div>
                <div>Looking at the datasheet for the LS1012A, it
                  mentions in the PCI Express section that:<br>
                  <br>
                  The PCI Express controller as instantiated on this
                  chip does not
                  support hardware coherency. All incoming PCI Express
                  transactions are made non IO-coherent.</div>
                <div><br>
                </div>
                <div>Would AMDGPU still work with that PCI Express
                  controller, or is this a show-stopper?</div>
                <div><br>
                  <div>[    5.727691] [drm] initializing kernel
                    modesetting (POLARIS11 0x1002:0x67E8 0x1DA2:0xE362
                    0x80).</div>
                  <div>[    5.761767] [drm] register mmio base:
                    0x58000000</div>
                  <div>[    5.777272] [drm] register mmio size: 262144</div>
                  <div>[    5.825973] [drm] add ip block number 0
                    <vi_common></div>
                  <div>[    5.832242] [drm] add ip block number 1
                    <gmc_v8_0></div>
                  <div>[    5.837767] [drm] add ip block number 2
                    <tonga_ih></div>
                  <div>[    5.843121] [drm] add ip block number 3
                    <gfx_v8_0></div>
                  <div>[    5.848480] [drm] add ip block number 4
                    <sdma_v3_0></div>
                  <div>[    5.853969] [drm] add ip block number 5
                    <powerplay></div>
                  <div>[    5.859413] [drm] add ip block number 6
                    <dm></div>
                  <div>[    5.864238] [drm] add ip block number 7
                    <uvd_v6_0></div>
                  <div>[    5.869690] [drm] add ip block number 8
                    <vce_v3_0></div>
                  <div>[    5.875067] [drm] UVD is enabled in VM mode</div>
                  <div>[    5.879858] [drm] UVD ENC is enabled in VM
                    mode</div>
                  <div>[    5.884985] [drm] VCE enabled in VM mode</div>
                  <div>[    6.114020] ATOM BIOS: 113-C98511-U01</div>
                  <div>[    6.117757] [drm] GPU posting now...</div>
                  <div>[    6.238112] [drm] vm size is 64 GB, 2 levels,
                    block size is 10-bit, fragment size is 9-bit</div>
                  <div>[    6.247290] amdgpu 0000:01:00.0: BAR 2:
                    releasing [mem 0x4050000000-0x40501fffff 64bit pref]</div>
                  <div>[    6.257354] amdgpu 0000:01:00.0: BAR 0:
                    releasing [mem 0x4040000000-0x404fffffff 64bit pref]</div>
                  <div>[    6.267261] pcieport 0000:00:00.0: BAR 15:
                    releasing [mem 0x4040000000-0x4057ffffff 64bit pref]</div>
                  <div>[    6.277288] pcieport 0000:00:00.0: BAR 15: no
                    space for [mem size 0x300000000 64bit pref]</div>
                  <div>[    6.286843] pcieport 0000:00:00.0: BAR 15:
                    failed to assign [mem size 0x300000000 64bit pref]</div>
                  <div>[    6.296708] amdgpu 0000:01:00.0: BAR 0: no
                    space for [mem size 0x200000000 64bit pref]</div>
                  <div>[    6.306009] amdgpu 0000:01:00.0: BAR 0: failed
                    to assign [mem size 0x200000000 64bit pref]</div>
                  <div>[    6.315635] amdgpu 0000:01:00.0: BAR 2: no
                    space for [mem size 0x00200000 64bit pref]</div>
                  <div>[    6.324804] amdgpu 0000:01:00.0: BAR 2: failed
                    to assign [mem size 0x00200000 64bit pref]</div>
                  <div>[    6.334366] pcieport 0000:00:00.0: PCI bridge
                    to [bus 01-ff]</div>
                  <div>[    6.341323] pcieport 0000:00:00.0:   bridge
                    window [io  0x1000-0x1fff]</div>
                  <div>[    6.349158] pcieport 0000:00:00.0:   bridge
                    window [mem 0x4058000000-0x40580fffff]</div>
                  <div>[    6.358095] pcieport 0000:00:00.0: PCI bridge
                    to [bus 01-ff]</div>
                  <div>[    6.365054] pcieport 0000:00:00.0:   bridge
                    window [io  0x1000-0x1fff]</div>
                  <div>[    6.372981] pcieport 0000:00:00.0:   bridge
                    window [mem 0x4058000000-0x40580fffff]</div>
                  <div>[    6.381917] pcieport 0000:00:00.0:   bridge
                    window [mem 0x4040000000-0x4057ffffff 64bit pref]</div>
                  <div>[    6.391789] [drm] Not enough PCI address space
                    for a large BAR.</div>
                  <div>[    6.391820] amdgpu 0000:01:00.0: BAR 0:
                    assigned [mem 0x4040000000-0x404fffffff 64bit pref]</div>
                  <div>[    6.407776] amdgpu 0000:01:00.0: BAR 2:
                    assigned [mem 0x4050000000-0x40501fffff 64bit pref]</div>
                  <div>[    6.417672] amdgpu 0000:01:00.0: VRAM: 8192M
                    0x000000F400000000 - 0x000000F5FFFFFFFF (8192M used)</div>
                  <div>[    6.426586] amdgpu 0000:01:00.0: GART: 256M
                    0x000000FF00000000 - 0x000000FF0FFFFFFF</div>
                  <div>[    6.436171] [drm] Detected VRAM RAM=8192M,
                    BAR=256M</div>
                  <div>[    6.478952] [drm] GART: num cpu pages 65536,
                    num gpu pages 65536</div>
                  <div>
                    <div>[    6.487871] [drm] PCIE GART of 256M enabled
                      (table at 0x000000F400000000).</div>
                    <div>[    6.496316] amdgpu 0000:01:00.0: Direct
                      firmware load for amdgpu/polaris11_pfp_2.bin
                      failed with error -2</div>
                    <div>[    6.508078] amdgpu 0000:01:00.0: Direct
                      firmware load for amdgpu/polaris11_me_2.bin failed
                      with error -2</div>
                    <div>[    6.519538] amdgpu 0000:01:00.0: Direct
                      firmware load for amdgpu/polaris11_ce_2.bin failed
                      with error -2</div>
                    <div>[    6.531496] amdgpu 0000:01:00.0: Direct
                      firmware load for amdgpu/polaris11_mec_2.bin
                      failed with error -2</div>
                    <div>[    6.544214] amdgpu 0000:01:00.0: Direct
                      firmware load for amdgpu/polaris11_mec2_2.bin
                      failed with error -2</div>
                    <div>[    6.565121] [drm] Found UVD firmware
                      Version: 1.79 Family ID: 16</div>
                    <div>[    6.571241] [drm] UVD ENC is disabled</div>
                    <div>[    6.580927] [drm] Found VCE firmware
                      Version: 52.4 Binary ID: 3</div>
                    <div>[    6.656854] amdgpu 0000:01:00.0: GPU fault
                      detected: 147 0x00004802 for process  pid 0
                      thread  pid 0</div>
                    <div>[    6.666013] amdgpu 0000:01:00.0: 
                       VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x0017F000</div>
                    <div>[    6.673508] amdgpu 0000:01:00.0: 
                       VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x02048002</div>
                    <div>[    6.681006] amdgpu 0000:01:00.0: VM fault
                      (0x02, vmid 1, pasid 0) at page 1568768, read from
                      'TC0' (0x54433000) (72)</div>
                    <div>[    6.691557] amdgpu 0000:01:00.0: GPU fault
                      detected: 147 0x00004402 for process  pid 0
                      thread  pid 0</div>
                    <div>[    6.700706] amdgpu 0000:01:00.0: 
                       VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x0017F000</div>
                    <div>[    6.708200] amdgpu 0000:01:00.0: 
                       VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x02048002</div>
                    <div>[    6.715697] amdgpu 0000:01:00.0: VM fault
                      (0x02, vmid 1, pasid 0) at page 1568768, read from
                      'TC0' (0x54433000) (72)</div>
                    <div>[    6.726246] amdgpu 0000:01:00.0: GPU fault
                      detected: 147 0x00000402 for process  pid 0
                      thread  pid 0</div>
                    <div>[    6.735395] amdgpu 0000:01:00.0: 
                       VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x0017F000</div>
                    <div>[    6.742889] amdgpu 0000:01:00.0: 
                       VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x02048002</div>
                    <div>[    6.750386] amdgpu 0000:01:00.0: VM fault
                      (0x02, vmid 1, pasid 0) at page 1568768, read from
                      'TC0' (0x54433000) (72)</div>
                    <div>[    6.984485] amdgpu 0000:01:00.0:
                      [drm:amdgpu_ring_test_helper [amdgpu]] *ERROR*
                      ring gfx test failed (-110)</div>
                    <div>[    6.994562] [drm:amdgpu_device_init
                      [amdgpu]] *ERROR* hw_init of IP block
                      <gfx_v8_0> failed -110</div>
                    <div>[    7.004942] amdgpu 0000:01:00.0:
                      amdgpu_device_ip_init failed</div>
                    <div>[    7.011963] amdgpu 0000:01:00.0: Fatal error
                      during GPU init</div>
                    <div>[    7.018890] [drm] amdgpu: finishing device.</div>
                  </div>
                </div>
                <div>
                  <div>[    7.308898] WARNING: CPU: 0 PID: 2084 at
                    /home/bas/linux/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c:915
                    amdgpu_bo_unpin+0xe4/0x110 [amdgpu]</div>
                  <div>[    7.321451] Modules linked in: amdgpu(+)
                    realtek chash gpu_sched ttm drm_kms_helper drm
                    crct10dif_ce drm_panel_orientation_quirks pfe(C)
                    ip_tables x_tables ipv6</div>
                  <div>[    7.335850] CPU: 0 PID: 2084 Comm:
                    systemd-udevd Tainted: G         C       
                    5.0.0-rc1-00001-g3bd6e94bec12-dirty #1</div>
                  <div>[    7.346303] Hardware name: LS1012A RDB Board
                    (DT)</div>
                  <div>[    7.351014] pstate: 40000005 (nZcv daif -PAN
                    -UAO)</div>
                  <div>[    7.356090] pc : amdgpu_bo_unpin+0xe4/0x110
                    [amdgpu]</div>
                  <div>[    7.361307] lr :
                    amdgpu_bo_free_kernel+0x7c/0x148 [amdgpu]</div>
                  <div>[    7.366799] sp : ffff0000118c3730</div>
                  <div>[    7.370114] x29: ffff0000118c3730 x28:
                    ffff000008e63710 </div>
                  <div>[    7.375434] x27: ffff0000118c3df0 x26:
                    0000000000000100 </div>
                  <div>[    7.380754] x25: ffff80003599cb60 x24:
                    ffff8000359927d0 </div>
                  <div>[    7.386073] x23: ffff800035994750 x22:
                    ffff8000359927d0 </div>
                  <div>[    7.391393] x21: ffff8000358cb800 x20:
                    ffff0000111fd000 </div>
                  <div>[    7.396713] x19: ffff8000358cb800 x18:
                    000000000000001e </div>
                  <div>[    7.402033] x17: 0000000000000000 x16:
                    0000000000000002 </div>
                  <div>[    7.407351] x15: 0000000000000400 x14:
                    0000000000000400 </div>
                  <div>[    7.412670] x13: 000000000000cb72 x12:
                    000000000000b308 </div>
                  <div>[    7.417989] x11: ffff7e0000d02488 x10:
                    ffff800037798030 </div>
                  <div>[    7.423308] x9 : ffff0000118c36e4 x8 :
                    0000000000000000 </div>
                  <div>[    7.428627] x7 : 0000000000210d00 x6 :
                    0000000000000000 </div>
                  <div>[    7.433946] x5 : 0000000000000001 x4 :
                    0000000000000001 </div>
                  <div>[    7.439265] x3 : ffff8000358cb87c x2 :
                    0000000000000000 </div>
                  <div>[    7.444583] x1 : 0000000000000000 x0 :
                    ffff8000358cb800 </div>
                  <div>[    7.449903] Call trace:</div>
                  <div>[    7.452583]  amdgpu_bo_unpin+0xe4/0x110
                    [amdgpu]</div>
                  <div>[    7.457437]  amdgpu_bo_free_kernel+0x7c/0x148
                    [amdgpu]</div>
                  <div>[    7.462812]  amdgpu_gfx_rlc_fini+0x50/0x78
                    [amdgpu]</div>
                  <div>[    7.467926]  gfx_v8_0_sw_fini+0xfc/0x1c0
                    [amdgpu]</div>
                  <div>[    7.472864]  amdgpu_device_fini+0x1e0/0x480
                    [amdgpu]</div>
                  <div>[    7.478058] 
                    amdgpu_driver_unload_kms+0xa0/0x150 [amdgpu]</div>
                  <div>[    7.483686] 
                    amdgpu_driver_load_kms+0x144/0x1f8 [amdgpu]</div>
                  <div>[    7.489110]  drm_dev_register+0x14c/0x1e0
                    [drm]</div>
                  <div>[    7.493906]  amdgpu_pci_probe+0xcc/0x188
                    [amdgpu]</div>
                  <div>[    7.498622]  local_pci_probe+0x3c/0xb0</div>
                  <div>[    7.502375]  pci_device_probe+0x150/0x1b8</div>
                  <div>[    7.506390]  really_probe+0x1f0/0x298</div>
                  <div>[    7.510055]  driver_probe_device+0x58/0x100</div>
                  <div>[    7.514242]  __driver_attach+0xd4/0xd8</div>
                  <div>[    7.517993]  bus_for_each_dev+0x74/0xc8</div>
                  <div>[    7.521832]  driver_attach+0x20/0x28</div>
                  <div>[    7.525410]  bus_add_driver+0x1ac/0x218</div>
                  <div>[    7.529250]  driver_register+0x60/0x110</div>
                  <div>[    7.533089]  __pci_register_driver+0x40/0x48</div>
                  <div>[    7.537599]  amdgpu_init+0x58/0x1000 [amdgpu]</div>
                  <div>[    7.541964]  do_one_initcall+0x5c/0x178</div>
                  <div>[    7.545804]  do_init_module+0x58/0x1b0</div>
                  <div>[    7.549556]  load_module+0x1dc8/0x2178</div>
                  <div>[    7.553308]  __se_sys_finit_module+0xbc/0xd0</div>
                  <div>[    7.557582] 
                    __arm64_sys_finit_module+0x18/0x20</div>
                  <div>[    7.562119]  el0_svc_common+0x60/0x100</div>
                  <div>[    7.565872]  el0_svc_handler+0x2c/0x80</div>
                </div>
                <div>
                  <div>[    7.581715] amdgpu 0000:01:00.0:
                    (____ptrval____) unpin not necessary</div>
                  <div>[    7.594743] [TTM] Finalizing pool allocator</div>
                  <div>[    7.599867] [TTM] Finalizing DMA pool
                    allocator</div>
                  <div>[    7.605265] [TTM] Zone  kernel: Used memory at
                    exit: 0 kiB</div>
                  <div>[    7.611482] [drm] amdgpu: ttm finalized</div>
                  <div>[    7.618028] amdgpu: probe of 0000:01:00.0
                    failed with error -110</div>
                </div>
                <div><br>
                </div>
                <div>Bas Vermeulen</div>
                <div><br>
                </div>
              </div>
            </div>
          </div>
        </div>
      </div>
      <br>
      <div class="gmail_quote">
        <div dir="ltr">On Thu, Dec 20, 2018 at 5:11 PM Bas Vermeulen
          <<a href="mailto:bas@daedalean.ai" moz-do-not-send="true">bas@daedalean.ai</a>>
          wrote:<br>
        </div>
        <blockquote class="gmail_quote" style="margin:0px 0px 0px
          0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
          <div dir="ltr">Hi Alex,
            <div><br>
            </div>
            <div>I already have a similar patch in, that doesn't fix it
              just yet. I'll investigate some more.</div>
            <div><br>
            </div>
            <div>Bas Vermeulen</div>
          </div>
          <br>
          <div class="gmail_quote">
            <div dir="ltr">On Thu, Dec 20, 2018 at 4:27 PM Alex Deucher
              <<a href="mailto:alexdeucher@gmail.com" target="_blank"
                moz-do-not-send="true">alexdeucher@gmail.com</a>>
              wrote:<br>
            </div>
            <blockquote class="gmail_quote" style="margin:0px 0px 0px
              0.8ex;border-left:1px solid
              rgb(204,204,204);padding-left:1ex">On Thu, Dec 20, 2018 at
              9:06 AM Bas Vermeulen <<a
                href="mailto:bas@daedalean.ai" target="_blank"
                moz-do-not-send="true">bas@daedalean.ai</a>> wrote:<br>
              ><br>
              > Hi all,<br>
              ><br>
              > I have connected an E9260 (Polaris 11 based) to a
              mini-PCIe slot on my NXP LS1012ARDB.<br>
              > The GPU is seen, all the BARs are correctly assigned
              (but there's not enough PCIe memory space for the big
              BARs).<br>
              ><br>
              > When I try to load the amdgpu module, I can't get the
              driver to enable the acceleration (the scratch register
              check fails).<br>
              ><br>
              > The E9260 is connected to a PCIe x1 (Gen2) slot.<br>
              ><br>
              > Anyone have an idea on where to look or how to fix
              this? This is a test-bed before we get an LS1046ARDB with
              quad core A72's.<br>
              ><br>
              > Any help would be appreciated,<br>
              <br>
              Something like this patch should fix it assuming this is
              an ARM based platform:<br>
              <a href="https://patchwork.freedesktop.org/patch/269367/"
                rel="noreferrer" target="_blank" moz-do-not-send="true">https://patchwork.freedesktop.org/patch/269367/</a><br>
              <br>
              Alex<br>
              <br>
              ><br>
              > Bas Vermeulen<br>
              > _______________________________________________<br>
              > amd-gfx mailing list<br>
              > <a href="mailto:amd-gfx@lists.freedesktop.org"
                target="_blank" moz-do-not-send="true">amd-gfx@lists.freedesktop.org</a><br>
              > <a
                href="https://lists.freedesktop.org/mailman/listinfo/amd-gfx"
                rel="noreferrer" target="_blank" moz-do-not-send="true">https://lists.freedesktop.org/mailman/listinfo/amd-gfx</a><br>
            </blockquote>
          </div>
        </blockquote>
      </div>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
      <pre class="moz-quote-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>