<html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40"> <head> <meta http-equiv="Content-Type" content="text/html; charset=utf-8"> <meta name="Generator" content="Microsoft Word 15 (filtered medium)"> <style></style> </head> <body bgcolor="white" lang="ZH-CN" link="blue" vlink="purple"> <div class="WordSection1"> <p class="MsoNormal"><span lang="EN-US">>That sounds sane, but unfortunately might not be possible with the existing IOCTL. Keep in mind that we need to keep backward compatibility here.<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:等线;color:windowtext"><o:p> </o:p></span></p> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:等线;color:windowtext">unfortunately the current scheme on amdgpu_ctx_query() won’t work with TDR feature, which is aim to support vulkan/mesa/close-ogl/radv …<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:等线;color:windowtext">It’s enumeration is too limited to MESA implement …<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:等线;color:windowtext"><o:p> </o:p></span></p> <p class="MsoNormal"><b><span lang="EN-US" style="font-size:10.5pt;font-family:等线;color:windowtext">Do you have good idea ? both keep the compatibility here and give flexible ?<o:p></o:p></span></b></p> <p class="MsoNormal"><b><span lang="EN-US" style="font-size:10.5pt;font-family:等线;color:windowtext">looks like we need to add a new amdgpu_ctx_query_2() INTERFACE ….<o:p></o:p></span></b></p> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:等线;color:windowtext"><o:p> </o:p></span></p> <p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto;margin-left:0cm;text-indent:-18.0pt;mso-list:l1 level1 lfo2"> <![if !supportLists]><span lang="EN-US" style="font-size:10.0pt;font-family:Symbol"><span style="mso-list:Ignore">·<span style="font:7.0pt "Times New Roman""> </span></span></span><![endif]><b><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">A new IOCTL added for context:</span></b><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">Void amdgpu_ctx_reinit(){<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"> Ctx</span><span lang="EN-US" style="font-size:10.5pt;font-family:等线">→</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">vram_lost_counter = adev</span><span lang="EN-US" style="font-size:10.5pt;font-family:等线">→</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">vram_lost_counter;<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"> Ctx</span><span lang="EN-US" style="font-size:10.5pt;font-family:等线">→</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">reset_counter = adev</span><span lang="EN-US" style="font-size:10.5pt;font-family:等线">→</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">reset_counter;<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Courier New"">}</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US"><br> Mhm, is there any advantage to just creating a new context?<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US"><o:p> </o:p></span></p> <p class="MsoNormal"><span lang="EN-US">[ML] sorry, this function is wrong, here is my original idea:<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US"><o:p> </o:p></span></p> <p class="MsoNormal"><span lang="EN-US">MESA can create a new ctx based on an old one,like:<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US"><o:p> </o:p></span></p> <p class="MsoNormal"><span lang="EN-US">Create gl-ctx1,<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US">Create BO-A under gl-ctx1 …<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US">VRAM LOST<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US">Create gl-ctx2 from gl-ctx1 (share list, I’m not familiar with UMD, David Mao an Nicolai can input)<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US">Create BO-b UNDER gl-ctx2 …<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US">Submit job upon gl-ctx2, but it can refer to BO-A,<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US"><o:p> </o:p></span></p> <p class="MsoNormal"><span lang="EN-US">With our design, kernel won’t block submit from context2 (from gl-ctx2) since its vram_lost_counter equals to latest adev copy<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US">But gl-ctx2 is a clone from gl-ctx1, the only difference is the vram_lost/gpu_reset counter is updated to latest one<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US">So logically we should also block submit from gl-ctx2 (mapping to kernel context2), and we failed do so …<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US"><o:p> </o:p></span></p> <p class="MsoNormal"><span lang="EN-US">That’s why I want to add a new “amdgpu_ctx_clone”, which should work like:<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US"><o:p> </o:p></span></p> <p class="MsoNormal"><span lang="EN-US">Int amdgpu_ctx_clone(struct context *original, struct context *new) {<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US"> New->vram_lost_counter = original->vram_lost_counter;<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US"> New->gpu_reset_counter = original->gpu_reset_counter;<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US">}<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:等线;color:windowtext"><o:p> </o:p></span></p> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:等线;color:windowtext"><o:p> </o:p></span></p> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:等线;color:windowtext">BR Monk<o:p></o:p></span></p> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:等线;color:windowtext"><o:p> </o:p></span></p> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:等线;color:windowtext"><o:p> </o:p></span></p> <div> <div style="border:none;border-top:solid #E1E1E1 1.0pt;padding:3.0pt 0cm 0cm 0cm"> <p class="MsoNormal"><b><span lang="EN-US" style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:windowtext">From:</span></b><span lang="EN-US" style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:windowtext"> Koenig, Christian <br> <b>Sent:</b> 2017</span><span style="font-size:11.0pt;color:windowtext">年</span><span lang="EN-US" style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:windowtext">10</span><span style="font-size:11.0pt;color:windowtext">月</span><span lang="EN-US" style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:windowtext">12</span><span style="font-size:11.0pt;color:windowtext">日</span><span lang="EN-US" style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:windowtext"> 19:50<br> <b>To:</b> Liu, Monk <Monk.Liu@amd.com>; Haehnle, Nicolai <Nicolai.Haehnle@amd.com>; Michel Dänzer <michel@daenzer.net>; Olsak, Marek <Marek.Olsak@amd.com>; Deucher, Alexander <Alexander.Deucher@amd.com>; Zhou, David(ChunMing) <David1.Zhou@amd.com>; Mao, David <David.Mao@amd.com><br> <b>Cc:</b> Ramirez, Alejandro <Alejandro.Ramirez@amd.com>; amd-gfx@lists.freedesktop.org; Filipas, Mario <Mario.Filipas@amd.com>; Ding, Pixel <Pixel.Ding@amd.com>; Li, Bingley <Bingley.Li@amd.com>; Jiang, Jerry (SW) <Jerry.Jiang@amd.com><br> <b>Subject:</b> Re: TDR and VRAM lost handling in KMD (v2)<o:p></o:p></span></p> </div> </div> <p class="MsoNormal"><span lang="EN-US"><o:p> </o:p></span></p> <div> <p class="MsoNormal"><span lang="EN-US">Am 12.10.2017 um 13:37 schrieb Liu, Monk:<o:p></o:p></span></p> </div> <blockquote style="margin-top:5.0pt;margin-bottom:5.0pt"> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">Hi team<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">Very good, many policy and implement are agreed, looks we only have some arguments in amdgpu_ctx_query(), well I also confused with the current implement of it, </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Segoe UI Emoji",sans-serif">☹</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">First, I want to know if you guys agree that we<b> don't update ctx->reset_counter in amdgpu_ctx_query() </b>? because I want to make the query result always consistent upon a given context</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Courier New"">,</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> </blockquote> <p class="MsoNormal"><span lang="EN-US"><br> That sounds like a good idea to me, but I'm not sure if it won't break userspace (I don't think so). Nicolai or Marek need to comment.<br> <br> <br> <o:p></o:p></span></p> <blockquote style="margin-top:5.0pt;margin-bottom:5.0pt"> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">Second, I want to say that for KERNEL, it shouldn't use the term from MESA or OGL or VULKAN, e.g. kernel shouldn't use AMDGPU_INNOCENT_RESET to map to GL_INNOCENT_RESET_ARB, etc.</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Courier New"">..</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">Because that way kernel will be very limited to certain UMD, so I suggest we totally re-name the context status, and each UMD has its own way to map the kernel context's result to gl-context/vk-context/etc</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Courier New"">…</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> </blockquote> <p class="MsoNormal"><span lang="EN-US"><br> Yes, completely agree.<br> <br> <br> <o:p></o:p></span></p> <blockquote style="margin-top:5.0pt;margin-bottom:5.0pt"> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">Kernel should only provide below *<b>FLAG bits</b>* on a given context:<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto;margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> <![if !supportLists]><span lang="EN-US" style="font-size:10.0pt;font-family:Symbol"><span style="mso-list:Ignore">·<span style="font:7.0pt "Times New Roman""> </span></span></span><![endif]><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">Define AMDGPU_CTX_STATE_GUILTY 0x1 //as long as TDR detects a job hang, KMD set the context behind this context as "guilty"<o:p></o:p></span></p> <p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto;margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> <![if !supportLists]><span lang="EN-US" style="font-size:10.0pt;font-family:Symbol"><span style="mso-list:Ignore">·<span style="font:7.0pt "Times New Roman""> </span></span></span><![endif]><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">Define AMDGPU_CTX_STATE_VRAMLOST 0x2 //as long as there is a VRAM lost event hit after this context created, we mark this context "VRAM_LOST", so UMD can say that all BO under this context may lost their content, since kernel have no relationship between context and BO so this is UMD's call to judge if a BO considered "VRAM lost" or not.<o:p></o:p></span></p> <p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto;margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> <![if !supportLists]><span lang="EN-US" style="font-size:10.0pt;font-family:Symbol"><span style="mso-list:Ignore">·<span style="font:7.0pt "Times New Roman""> </span></span></span><![endif]><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">Define AMDGPU_CTX_STATE_RESET 0x3 //as long as there is a gpu reset occurred after context creation, this flag shall be set<o:p></o:p></span></p> </blockquote> <p class="MsoNormal"><span lang="EN-US"><br> That sounds sane, but unfortunately might not be possible with the existing IOCTL. Keep in mind that we need to keep backward compatibility here. <br> <br> <br> <o:p></o:p></span></p> <blockquote style="margin-top:5.0pt;margin-bottom:5.0pt"> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">Sample code:<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"> <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">Int amdgpu_ctx_query(struct amdgpu_ctx_query_parm * out, </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Courier New"">…..</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">) {<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"> if (ctx- >vram_lost_counter != adev->vram_lost_counter)<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"> out- >status |= AMDGPU_CTX_STATE_VRAM_LOST;<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">if (ctx- >reset_counter != adev</span><span lang="EN-US" style="font-size:10.5pt;font-family:等线">→</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">reset_counter) {<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">out- >status |= AMDGPU_CTX_STATE_RESET;<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">if (ctx- >guilty == TRUE)<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"> out- >status |= AMDGPU_CTX_STATE_GUILTY;<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Courier New",serif">}</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">return 0;<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">}<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">For UMD if it found "out.status == 0" means there is no gpu reset even occurred, the context is totally regular</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Courier New"">,</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto;margin-left:0cm;text-indent:-18.0pt;mso-list:l1 level1 lfo2"> <![if !supportLists]><span lang="EN-US" style="font-size:10.0pt;font-family:Symbol"><span style="mso-list:Ignore">·<span style="font:7.0pt "Times New Roman""> </span></span></span><![endif]><b><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">A new IOCTL added for context:</span></b><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">Void amdgpu_ctx_reinit(){<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"> Ctx</span><span lang="EN-US" style="font-size:10.5pt;font-family:等线">→</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">vram_lost_counter = adev</span><span lang="EN-US" style="font-size:10.5pt;font-family:等线">→</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">vram_lost_counter;<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"> Ctx</span><span lang="EN-US" style="font-size:10.5pt;font-family:等线">→</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">reset_counter = adev</span><span lang="EN-US" style="font-size:10.5pt;font-family:等线">→</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">reset_counter;<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Courier New"">}</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> </blockquote> <p class="MsoNormal"><span lang="EN-US"><br> Mhm, is there any advantage to just creating a new context?<br> <br> Regards,<br> Christian.<br> <br> <br> <o:p></o:p></span></p> <blockquote style="margin-top:5.0pt;margin-bottom:5.0pt"> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><b><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">if UMD decide *not* to release the "guilty" context but continue using it after UMD acknowledged GPU hang on certain job/context, I suggest UMD call "amdgpu_ctx_reinit()":</span></b><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">That way after you re</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Courier New"">-</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">init() this context, you can get updated result from "amdgpu_ctx_query", which will probably give you "out.status == 0" as long as no gpu reset/vram lost hit after re-init()</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Courier New"">.</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">BR Monk<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">-----Original Message-----<br> From: Koenig, Christian <br> Sent: 2017</span><span style="font-size:10.5pt;font-family:等线">年</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">10</span><span style="font-size:10.5pt;font-family:等线">月</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">12</span><span style="font-size:10.5pt;font-family:等线">日</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"> 18:13<br> To: Haehnle, Nicolai <a href="mailto:Nicolai.Haehnle@amd.com"><Nicolai.Haehnle@amd.com></a>; Michel D</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Courier New"">ä</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">nzer <a href="mailto:michel@daenzer.net"><michel@daenzer.net></a>; Liu, Monk <a href="mailto:Monk.Liu@amd.com"> <Monk.Liu@amd.com></a>; Olsak, Marek <a href="mailto:Marek.Olsak@amd.com"><Marek.Olsak@amd.com></a>; Deucher, Alexander <a href="mailto:Alexander.Deucher@amd.com"><Alexander.Deucher@amd.com></a>; Zhou, David(ChunMing) <a href="mailto:David1.Zhou@amd.com"><David1.Zhou@amd.com></a>; Mao, David <a href="mailto:David.Mao@amd.com"> <David.Mao@amd.com></a><br> Cc: Ramirez, Alejandro <a href="mailto:Alejandro.Ramirez@amd.com"><Alejandro.Ramirez@amd.com></a>; <a href="mailto:amd-gfx@lists.freedesktop.org">amd-gfx@lists.freedesktop.org</a>; Filipas, Mario <a href="mailto:Mario.Filipas@amd.com"><Mario.Filipas@amd.com></a>; Ding, Pixel <a href="mailto:Pixel.Ding@amd.com"> <Pixel.Ding@amd.com></a>; Li, Bingley <a href="mailto:Bingley.Li@amd.com"><Bingley.Li@amd.com></a>; Jiang, Jerry (SW) <a href="mailto:Jerry.Jiang@amd.com"><Jerry.Jiang@amd.com></a><br> Subject: Re: TDR and VRAM lost handling in KMD (v2)<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">Am 12.10.2017 um 11:44 schrieb Nicolai H</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Courier New"">ä</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">hnle:<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> On 12.10.2017 11:35, Michel D</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Courier New",serif">ä</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">nzer wrote:<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>> On 12/10/17 11:23 AM, Christian K</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Courier New"">ö</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">nig wrote:<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>> Am 12.10.2017 um 11:10 schrieb Nicolai H</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Courier New"">ä</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">hnle:<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>> On 12.10.2017 10:49, Christian K</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Courier New"">ö</span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">nig wrote:<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>>> However, !guilty && ctx->reset_counter != adev->reset_counter <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>>> does not imply that the context was lost.<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>>> The way I understand it, we should return <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>>> AMDGPU_CTX_INNOCENT_RESET if !guilty && ctx->vram_lost_counter != adev->vram_lost_counter.<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>>> As far as I understand it, the case of !guilty && <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>>> ctx->reset_counter != adev->reset_counter && <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>>> ctx->vram_lost_counter ==<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>>> adev->vram_lost_counter should return AMDGPU_CTX_NO_RESET, <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>>> adev->because a<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>>> GPU reset occurred, but it didn't affect our context.<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>> I disagree on that.<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>> AMDGPU_CTX_INNOCENT_RESET just means what it does currently, there <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>> was a reset but we haven't been causing it.<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>> That the OpenGL extension is specified otherwise is unfortunate, <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>>> but I think we shouldn't use that for the kernel interface here.<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>> Two counterpoints:<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>> 1. Why should any application care that there was a reset while it <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>> was idle? The OpenGL behavior is what makes sense.<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>> The application is certainly not interest if a reset happened or <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>> not, but I though that the driver stack might be.<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>> 2. AMDGPU_CTX_INNOCENT_RESET doesn't actually mean anything today <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>> because we never return it :)<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>> Good point.<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>> amdgpu_ctx_query only ever returns AMDGPU_CTX_UNKNOWN_RESET, which <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>> is in line with the OpenGL spec: we're conservatively returning <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>> that a reset happened because we don't know whether the context was <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>> affected, and we return UNKNOWN because we also don't know whether <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>> the context was guilty or not.<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>> Returning AMDGPU_CTX_NO_RESET in the case of !guilty &&<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>> ctx->vram_lost_counter == adev->vram_lost_counter is simply a<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>> refinement and improvement of the current, overly conservative <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>>> behavior.<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>> Ok let's reenumerate what I think the different return values should<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>> mean:<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>> * AMDGPU_CTX_GUILTY_RESET<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>> guilty is set to true for this context.<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>> * AMDGPU_CTX_INNOCENT_RESET<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>> guilty is not set and vram_lost_counter has changed.<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>> * AMDGPU_CTX_UNKNOWN_RESET<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>> guilty is not set and vram_lost_counter has not changed, but <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>>> gpu_reset_counter has changed.<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>> I don't understand the distinction you're proposing between <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>> AMDGPU_CTX_INNOCENT_RESET and AMDGPU_CTX_UNKNOWN_RESET. I think both <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>> cases you're describing should return either <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>> AMDGPU_CTX_INNOCENT_RESET, if the value of guilty is reliable, or <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">>> AMDGPU_CTX_UNKNOWN_RESET if it's not.<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> I think it'd make more sense if it was called <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> "AMDGPU_CTX_UNAFFECTED_RESET".<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> So:<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> - AMDGPU_CTX_GUILTY_RESET --> the context was affected by a reset, and <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> we know that it's the context's fault<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> - AMDGPU_CTX_INNOCENT_RESET --> the context was affected by a reset, <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> and we know that it *wasn't* the context's fault (no context job <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> active)<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> - AMDGPU_CTX_UNKNOWN_RESET --> the context was affected by a reset, <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> and we don't know who's responsible (this could be returned in the <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> unlikely case where context A's gfx job has not yet finished, but <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> context B's gfx job has already started; it could be the fault of A, <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> it could be the fault of B -- which somehow manages to hang a part of <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> the hardware that then prevents A's job from finishing -- or it could <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> be both; but it's a bit academic)<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> - AMDGPU_CTX_UNAFFECTED_RESET --> there was a reset, but this context <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> wasn't affected<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> This last value would currently just be discarded by Mesa (because we <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> should only disturb applications when we have to), but perhaps <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> somebody else could find it useful?<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"> <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">Yes, that's what I had in mind as well.<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"> <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">Cause otherwise we would return AMDGPU_CTX_NO_RESET while there actually was a reset and that certainly doesn't sound correct to me.<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"> <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">Regards,<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">Christian.<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"> <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">><o:p> </o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> Cheers,<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif">> Nicolai<o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"> <o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> <div> <p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt;font-family:"Times New Roman",serif"> </span><span lang="EN-US" style="font-size:10.5pt;font-family:"Calibri",sans-serif"><o:p></o:p></span></p> </div> </blockquote> <p><span lang="EN-US"><o:p> </o:p></span></p> </div> </body> </html>