<html> <head> <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1"> <style type="text/css" style="display:none;"></style> </head> <body dir="ltr"> <div id="divtagdefaultwrapper" style="font-size:12pt;color:#000000;font-family:Calibri,Helvetica,sans-serif;" dir="ltr"> <p>I don't know if compiler is smart enough to optimize the following for statement out...</p> <p><br> </p> <p><font size="2"><span style="font-size:10pt;">+ for (i = 0; i < ttm->num_pages; i++) {<br> + trace_amdgpu_ttm_tt_populate(<br> + adev,<br> + gtt->ttm.dma_address[i],<br> + page_to_phys(ttm->pages[i]));<br> + }</span></font></p> <br> <p>-Alex Bin<br> </p> <br> <div style="color: rgb(0, 0, 0);"> <div> <hr tabindex="-1" style="display:inline-block; width:98%"> <div id="x_divRplyFwdMsg" dir="ltr"><font style="font-size:11pt" face="Calibri, sans-serif" color="#000000"><b>From:</b> Christian König <deathsimple@vodafone.de><br> <b>Sent:</b> Tuesday, August 1, 2017 11:41 AM<br> <b>To:</b> Xie, AlexBin; amd-gfx@lists.freedesktop.org<br> <b>Subject:</b> Re: [PATCH] drm/amd/amdgpu: Add tracepoint for DMA page mapping</font> <div> </div> </div> </div> <font size="2"><span style="font-size:10pt;"> <div class="PlainText">We can turn of the trace subsystem during compile time already.<br> <br> Additional to that the trace points use self modifying code to make sure <br> that they don't have any overhead as long as they are turned off. But I <br> don't think this works with the "if (trace_*_enabled()" as well.<br> <br> Anyway, not a performance critical code path so I don't really bother.<br> <br> Christian.<br> <br> Am 01.08.2017 um 17:00 schrieb axie:<br> > Because this is used by a debug tool, can we use a macro to <br> > conditionally compile this feature?<br> ><br> ><br> > On 2017-08-01 10:54 AM, Christian König wrote:<br> >> Am 01.08.2017 um 16:26 schrieb Tom St Denis:<br> >>> On 01/08/17 10:10 AM, Christian König wrote:<br> >>><br> >>>> You need to cover multiple code path here:<br> >>>> 1. The one you currently implemented which uses ttm_dma_populate() <br> >>>> and pci_map_page().<br> >>>> 2. The one using ttm_dma_populate().<br> >>><br> >>> I'll have to look into this one though it's my understanding that <br> >>> code path is only followed if there's no (hardware) IOMMU right?<br> >><br> >> Wrong, that one is used when "anything" in the system used the <br> >> SWIOTLB path once. So the detection doesn't always work reliable.<br> >><br> >>> This one seems rather straight forward but the only catch is I don't <br> >>> have access to "adev" inside that drm function.<br> >><br> >> When you have a ttm you can always do "amdgpu_ttm_adev(ttm->bdev)" to <br> >> get the adev.<br> >><br> >>> Would it be taboo to do something like<br> >>><br> >>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c <br> >>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c<br> >>> index 376078334f54..cd97ef2144c9 100644<br> >>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c<br> >>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c<br> >>> @@ -916,6 +916,12 @@ static int amdgpu_ttm_tt_populate(struct ttm_tt <br> >>> *ttm)<br> >>> drm_prime_sg_to_page_addr_arrays(ttm->sg, ttm->pages,<br> >>> gtt->ttm.dma_address, ttm->num_pages);<br> >>> ttm->state = tt_unbound;<br> >>> + for (i = 0; i < ttm->num_pages; i++) {<br> >>> + trace_amdgpu_ttm_tt_populate(<br> >>> + adev,<br> >>> + gtt->ttm.dma_address[i],<br> >>> + page_to_phys(ttm->pages[i]));<br> >>> + }<br> >>> return 0;<br> >>> }<br> >>><br> >>> This would normally result in a for loop around a nop which <br> >>> shouldn't be a huge performance hit but is one none the less.<br> >><br> >> Try using trace_trace_amdgpu_ttm_tt_populate_enabled(), those <br> >> functions are generated by the trace subsystem automatically when you <br> >> define a trace point.<br> >><br> >> It just doesn't use those nice tricks to modify the compiled binary <br> >> on the fly.<br> >><br> >> Christian.<br> >><br> >>><br> >>>> 4. The in amdgpu_ttm_tt_pin_userptr() which uses dma_map_sg() for <br> >>>> userptrs.<br> >>>><br> >>>> Basically all just take gtt->ttm.ttm.pages and fill <br> >>>> gtt->ttm.dma_address.<br> >>><br> >>> Same comment as #3. I can tackle this with a for loop around the <br> >>> trace if that is ok.<br> >>><br> >>>> I suggest to add a helper you can call to trace all <br> >>>> pages->dma_address mappings inside a ttm.<br> >>><br> >>> One thing at a time :-). That would probably be a bit better since <br> >>> the trace log gets filled with remappings (which is why my <br> >>> proof-of-concept umr patch only grabs the latest mapping as it reads <br> >>> the trace).<br> >>><br> >>> Is there a handy place we can grab a list of ttm_tt's bound to our <br> >>> device? If so then in theory I could write a debugfs interface <br> >>> instead.<br> >>><br> >>> Thanks for your patience as I'm definitely less familiar with the VM <br> >>> side of things :-)<br> >>><br> >>> Cheers,<br> >>> Tom<br> >>> _______________________________________________<br> >>> amd-gfx mailing list<br> >>> amd-gfx@lists.freedesktop.org<br> >>> <a href="https://lists.freedesktop.org/mailman/listinfo/amd-gfx" id="LPlnk518445" previewremoved="true"> https://lists.freedesktop.org/mailman/listinfo/amd-gfx</a> <div id="LPBorder_GT_15016027613140.08295823990357198" style="margin-bottom: 20px; overflow: auto; width: 100%; text-indent: 0px;"> <table id="LPContainer_15016027613110.1813202064689361" style="width: 90%; background-color: rgb(255, 255, 255); position: relative; overflow: auto; padding-top: 20px; padding-bottom: 20px; margin-top: 20px; border-top: 1px dotted rgb(200, 200, 200); border-bottom: 1px dotted rgb(200, 200, 200);" role="presentation" cellspacing="0"> <tbody> <tr style="border-spacing: 0px;" valign="top"> <td id="TextCell_15016027613120.5421751103538535" style="vertical-align: top; position: relative; padding: 0px; display: table-cell;" colspan="2"> <div id="LPRemovePreviewContainer_15016027613120.2548737792421861"></div> <div id="LPTitle_15016027613120.23212871138685842" style="top: 0px; color: rgb(0, 120, 215); font-weight: 400; font-size: 21px; font-family: "wf_segoe-ui_light","Segoe UI Light","Segoe WP Light","Segoe UI","Segoe WP",Tahoma,Arial,sans-serif; line-height: 21px;"> <a id="LPUrlAnchor_15016027613130.2533724279446574" style="text-decoration: none;" href="https://lists.freedesktop.org/mailman/listinfo/amd-gfx" target="_blank">amd-gfx Info Page - freedesktop.org</a></div> <div id="LPMetadata_15016027613130.7874404914990427" style="margin: 10px 0px 16px; color: rgb(102, 102, 102); font-weight: 400; font-family: "wf_segoe-ui_normal","Segoe UI","Segoe WP",Tahoma,Arial,sans-serif; font-size: 14px; line-height: 14px;"> lists.freedesktop.org</div> <div id="LPDescription_15016027613130.5169047949058398" style="display: block; color: rgb(102, 102, 102); font-weight: 400; font-family: "wf_segoe-ui_normal","Segoe UI","Segoe WP",Tahoma,Arial,sans-serif; font-size: 14px; line-height: 20px; max-height: 100px; overflow: hidden;"> Subscribing to amd-gfx: Subscribe to amd-gfx by filling out the following form. Use of all freedesktop.org lists is subject to our Code of ...</div> </td> </tr> </tbody> </table> </div> <br> >><br> >><br> >> _______________________________________________<br> >> amd-gfx mailing list<br> >> amd-gfx@lists.freedesktop.org<br> >> <a href="https://lists.freedesktop.org/mailman/listinfo/amd-gfx" id="LPlnk434060" previewremoved="true"> https://lists.freedesktop.org/mailman/listinfo/amd-gfx</a> <div id="LPBorder_GT_15016027613220.7871711856219648" style="margin-bottom: 20px; overflow: auto; width: 100%; text-indent: 0px;"> <table id="LPContainer_15016027613190.2615781169078024" style="width: 90%; background-color: rgb(255, 255, 255); position: relative; overflow: auto; padding-top: 20px; padding-bottom: 20px; margin-top: 20px; border-top: 1px dotted rgb(200, 200, 200); border-bottom: 1px dotted rgb(200, 200, 200);" role="presentation" cellspacing="0"> <tbody> <tr style="border-spacing: 0px;" valign="top"> <td id="TextCell_15016027613200.08716742963535162" style="vertical-align: top; position: relative; padding: 0px; display: table-cell;" colspan="2"> <div id="LPRemovePreviewContainer_15016027613200.6016969126942272"></div> <div id="LPTitle_15016027613200.1704710881698085" style="top: 0px; color: rgb(0, 120, 215); font-weight: 400; font-size: 21px; font-family: "wf_segoe-ui_light","Segoe UI Light","Segoe WP Light","Segoe UI","Segoe WP",Tahoma,Arial,sans-serif; line-height: 21px;"> <a id="LPUrlAnchor_15016027613210.1055556852355326" style="text-decoration: none;" href="https://lists.freedesktop.org/mailman/listinfo/amd-gfx" target="_blank">amd-gfx Info Page - freedesktop.org</a></div> <div id="LPMetadata_15016027613210.07358644878449205" style="margin: 10px 0px 16px; color: rgb(102, 102, 102); font-weight: 400; font-family: "wf_segoe-ui_normal","Segoe UI","Segoe WP",Tahoma,Arial,sans-serif; font-size: 14px; line-height: 14px;"> lists.freedesktop.org</div> <div id="LPDescription_15016027613210.74097043846146" style="display: block; color: rgb(102, 102, 102); font-weight: 400; font-family: "wf_segoe-ui_normal","Segoe UI","Segoe WP",Tahoma,Arial,sans-serif; font-size: 14px; line-height: 20px; max-height: 100px; overflow: hidden;"> Subscribing to amd-gfx: Subscribe to amd-gfx by filling out the following form. Use of all freedesktop.org lists is subject to our Code of ...</div> </td> </tr> </tbody> </table> </div> <br> ><br> > _______________________________________________<br> > amd-gfx mailing list<br> > amd-gfx@lists.freedesktop.org<br> > <a href="https://lists.freedesktop.org/mailman/listinfo/amd-gfx" id="LPlnk374752" previewremoved="true"> https://lists.freedesktop.org/mailman/listinfo/amd-gfx</a> <div id="LPBorder_GT_15016028208780.5931683559589285" style="margin-bottom: 20px; overflow: auto; width: 100%; text-indent: 0px;"> <table id="LPContainer_15016028208740.92491337649414" style="width: 90%; background-color: rgb(255, 255, 255); position: relative; overflow: auto; padding-top: 20px; padding-bottom: 20px; margin-top: 20px; border-top: 1px dotted rgb(200, 200, 200); border-bottom: 1px dotted rgb(200, 200, 200);" role="presentation" cellspacing="0"> <tbody> <tr style="border-spacing: 0px;" valign="top"> <td id="TextCell_15016028208750.5039491761540736" style="vertical-align: top; position: relative; padding: 0px; display: table-cell;" colspan="2"> <div id="LPRemovePreviewContainer_15016028208750.8435103435108974"></div> <div id="LPTitle_15016028208750.3076890768275301" style="top: 0px; color: rgb(0, 120, 215); font-weight: 400; font-size: 21px; font-family: "wf_segoe-ui_light","Segoe UI Light","Segoe WP Light","Segoe UI","Segoe WP",Tahoma,Arial,sans-serif; line-height: 21px;"> <a id="LPUrlAnchor_15016028208760.5932119976653863" style="text-decoration: none;" href="https://lists.freedesktop.org/mailman/listinfo/amd-gfx" target="_blank">amd-gfx Info Page - freedesktop.org</a></div> <div id="LPMetadata_15016028208760.3609779223258601" style="margin: 10px 0px 16px; color: rgb(102, 102, 102); font-weight: 400; font-family: "wf_segoe-ui_normal","Segoe UI","Segoe WP",Tahoma,Arial,sans-serif; font-size: 14px; line-height: 14px;"> lists.freedesktop.org</div> <div id="LPDescription_15016028208770.6822461316553391" style="display: block; color: rgb(102, 102, 102); font-weight: 400; font-family: "wf_segoe-ui_normal","Segoe UI","Segoe WP",Tahoma,Arial,sans-serif; font-size: 14px; line-height: 20px; max-height: 100px; overflow: hidden;"> Subscribing to amd-gfx: Subscribe to amd-gfx by filling out the following form. Use of all freedesktop.org lists is subject to our Code of ...</div> </td> </tr> </tbody> </table> </div> <br> <br> <br> </div> </span></font></div> </div> </body> </html>