<html><head> <meta http-equiv="Content-Type" content="text/html; charset=utf-8"> </head> <body> Hi Curry, Evan, It seems vcn1.0 power gate sequence are special. if the original solution is thread safe, then the following solution will be thread safe also. static int vcn_v1_0_hw_fini(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; cancel_delayed_work_sync(&adev->vcn.idle_work); if ((adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) || (adev->vcn.cur_state != AMD_PG_STATE_GATE && RREG32_SOC15(VCN, 0, mmUVD_STATUS))) { + if (adev->pm.dpm_enabled) + amdgpu_dpm_enable_uvd(adev, false); + else + vcn_v1_0_set_powergating_state(adev, AMD_PG_STATE_GATE); } Best Regards! James <div class="moz-cite-prefix">On 2021-12-13 3:55 a.m., Gong, Curry wrote: </div> <blockquote type="cite" cite="mid:BN7PR12MB2641CA6290B6CEFB4A5016EF9D749@BN7PR12MB2641.namprd12.prod.outlook.com"> <meta name="Generator" content="Microsoft Word 15 (filtered medium)"> <style>@font-face {font-family:SimSun; panose-1:2 1 6 0 3 1 1 1 1 1;}@font-face {font-family:"Cambria Math"; panose-1:2 4 5 3 5 4 6 3 2 4;}@font-face {font-family:DengXian; panose-1:2 1 6 0 3 1 1 1 1 1;}@font-face {font-family:Calibri; panose-1:2 15 5 2 2 2 4 3 2 4;}@font-face {font-family:"\@DengXian"; panose-1:2 1 6 0 3 1 1 1 1 1;}@font-face {font-family:Consolas; panose-1:2 11 6 9 2 2 4 3 2 4;}@font-face {font-family:"\@SimSun"; panose-1:2 1 6 0 3 1 1 1 1 1;}p.MsoNormal, li.MsoNormal, div.MsoNormal {margin:0in; font-size:11.0pt; font-family:"Calibri",sans-serif;}a:link, span.MsoHyperlink {mso-style-priority:99; color:blue; text-decoration:underline;}pre {mso-style-priority:99; mso-style-link:"HTML Preformatted Char"; margin:0in; font-size:10.0pt; font-family:"Courier New";}span.HTMLPreformattedChar {mso-style-name:"HTML Preformatted Char"; mso-style-priority:99; mso-style-link:"HTML Preformatted"; font-family:Consolas;}span.EmailStyle21 {mso-style-type:personal-reply; font-family:"Arial",sans-serif; color:windowtext;}.MsoChpDefault {mso-style-type:export-only; font-size:10.0pt;}div.WordSection1 {page:WordSection1;}</style> [AMD Official Use Only] <div> <div class="WordSection1"> Hi James:<o:p></o:p> <o:p> </o:p> With the following patch, an error will be reported when the driver is loaded<o:p></o:p> +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c<o:p></o:p> @@ -1202,6 +1204,9 @@ static int vcn_v1_0_stop(struct amdgpu_device *adev)<o:p></o:p> else<o:p></o:p> r = vcn_v1_0_stop_spg_mode(adev);<o:p></o:p> <o:p> </o:p> c<o:p></o:p> return r;<o:p></o:p> }<o:p></o:p> <o:p> </o:p> <o:p> </o:p> $ dmesg<o:p></o:p> [ 363.181081] INFO: task kworker/3:2:223 blocked for more than 120 seconds.<o:p></o:p> [ 363.181150] Tainted: G OE 5.11.0-41-generic #45~20.04.1-Ubuntu<o:p></o:p> [ 363.181208] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.<o:p></o:p> [ 363.181266] task:kworker/3:2 state:D stack: 0 pid: 223 ppid: 2 flags:0x00004000<o:p></o:p> [ 363.181276] Workqueue: events vcn_v1_0_idle_work_handler [amdgpu]<o:p></o:p> [ 363.181612] Call Trace:<o:p></o:p> [ 363.181618] __schedule+0x44c/0x8a0<o:p></o:p> [ 363.181627] schedule+0x4f/0xc0<o:p></o:p> [ 363.181631] schedule_preempt_disabled+0xe/0x10<o:p></o:p> [ 363.181636] __mutex_lock.isra.0+0x183/0x4d0<o:p></o:p> [ 363.181643] __mutex_lock_slowpath+0x13/0x20<o:p></o:p> [ 363.181648] mutex_lock+0x32/0x40<o:p></o:p> [ 363.181652] amdgpu_dpm_set_powergating_by_smu+0x9c/0x180 [amdgpu]<o:p></o:p> [ 363.182055] amdgpu_dpm_enable_uvd+0x38/0x110 [amdgpu]<o:p></o:p> [ 363.182454] vcn_v1_0_set_powergating_state+0x2e7e/0x3cf0 [amdgpu]<o:p></o:p> [ 363.182776] amdgpu_device_ip_set_powergating_state+0x6c/0xc0 [amdgpu]<o:p></o:p> [ 363.183028] smu10_powergate_vcn+0x2a/0x80 [amdgpu]<o:p></o:p> [ 363.183361] pp_set_powergating_by_smu+0xc5/0x2b0 [amdgpu]<o:p></o:p> [ 363.183699] amdgpu_dpm_set_powergating_by_smu+0xb6/0x180 [amdgpu]<o:p></o:p> [ 363.184040] amdgpu_dpm_enable_uvd+0x38/0x110 [amdgpu]<o:p></o:p> [ 363.184391] vcn_v1_0_idle_work_handler+0xe1/0x130 [amdgpu]<o:p></o:p> [ 363.184667] process_one_work+0x220/0x3c0<o:p></o:p> [ 363.184674] worker_thread+0x4d/0x3f0<o:p></o:p> [ 363.184677] ? process_one_work+0x3c0/0x3c0<o:p></o:p> [ 363.184680] kthread+0x12b/0x150<o:p></o:p> [ 363.184685] ? set_kthread_struct+0x40/0x40<o:p></o:p> [ 363.184690] ret_from_fork+0x22/0x30<o:p></o:p> [ 363.184699] INFO: task kworker/2:2:233 blocked for more than 120 seconds.<o:p></o:p> [ 363.184739] Tainted: G OE 5.11.0-41-generic #45~20.04.1-Ubuntu<o:p></o:p> [ 363.184782] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.<o:p></o:p> [ 363.184825] task:kworker/2:2 state:D stack: 0 pid: 233 ppid: 2 flags:0x00004000<o:p></o:p> [ 363.184831] Workqueue: events amdgpu_device_delayed_init_work_handler [amdgpu]<o:p></o:p> [ 363.185085] Call Trace:<o:p></o:p> [ 363.185087] __schedule+0x44c/0x8a0<o:p></o:p> [ 363.185092] schedule+0x4f/0xc0<o:p></o:p> [ 363.185095] schedule_timeout+0x202/0x290<o:p></o:p> [ 363.185099] ? sched_clock_cpu+0x11/0xb0<o:p></o:p> [ 363.185105] wait_for_completion+0x94/0x100<o:p></o:p> [ 363.185110] __flush_work+0x12a/0x1e0<o:p></o:p> [ 363.185113] ? worker_detach_from_pool+0xc0/0xc0<o:p></o:p> [ 363.185119] __cancel_work_timer+0x10e/0x190<o:p></o:p> [ 363.185123] cancel_delayed_work_sync+0x13/0x20<o:p></o:p> [ 363.185126] vcn_v1_0_ring_begin_use+0x20/0x70 [amdgpu]<o:p></o:p> [ 363.185401] amdgpu_ring_alloc+0x48/0x60 [amdgpu]<o:p></o:p> [ 363.185640] amdgpu_ib_schedule+0x493/0x600 [amdgpu]<o:p></o:p> [ 363.185884] amdgpu_job_submit_direct+0x3c/0xd0 [amdgpu]<o:p></o:p> [ 363.186186] amdgpu_vcn_dec_send_msg+0x105/0x210 [amdgpu]<o:p></o:p> [ 363.186460] amdgpu_vcn_dec_ring_test_ib+0x69/0x110 [amdgpu]<o:p></o:p> [ 363.186734] amdgpu_ib_ring_tests+0xf5/0x160 [amdgpu]<o:p></o:p> [ 363.186978] amdgpu_device_delayed_init_work_handler+0x15/0x30 [amdgpu]<o:p></o:p> [ 363.187206] process_one_work+0x220/0x3c0<o:p></o:p> [ 363.187210] worker_thread+0x4d/0x3f0<o:p></o:p> [ 363.187214] ? process_one_work+0x3c0/0x3c0<o:p></o:p> [ 363.187217] kthread+0x12b/0x150<o:p></o:p> [ 363.187221] ? set_kthread_struct+0x40/0x40<o:p></o:p> [ 363.187226] ret_from_fork+0x22/0x30<o:p></o:p> <o:p> </o:p> <o:p> </o:p> BR<o:p></o:p> Curry Gong<o:p></o:p> <div> <div style="border:none;border-top:solid #E1E1E1 1.0pt;padding:3.0pt 0in 0in 0in"> From: Zhu, James <a class="moz-txt-link-rfc2396E" href="mailto:James.Zhu@amd.com"><James.Zhu@amd.com></a> Sent: Saturday, December 11, 2021 5:07 AM To: Gong, Curry <a class="moz-txt-link-rfc2396E" href="mailto:Curry.Gong@amd.com"><Curry.Gong@amd.com></a>; <a class="moz-txt-link-abbreviated" href="mailto:amd-gfx@lists.freedesktop.org">amd-gfx@lists.freedesktop.org</a> Cc: Liu, Leo <a class="moz-txt-link-rfc2396E" href="mailto:Leo.Liu@amd.com"><Leo.Liu@amd.com></a>; Zhu, James <a class="moz-txt-link-rfc2396E" href="mailto:James.Zhu@amd.com"><James.Zhu@amd.com></a>; Quan, Evan <a class="moz-txt-link-rfc2396E" href="mailto:Evan.Quan@amd.com"><Evan.Quan@amd.com></a>; Deucher, Alexander <a class="moz-txt-link-rfc2396E" href="mailto:Alexander.Deucher@amd.com"><Alexander.Deucher@amd.com></a> Subject: Re: [PATCH] drm/amdgpu: When the VCN(1.0) block is suspended, powergating is explicitly enabled<o:p></o:p> </div> </div> <o:p> </o:p> <div> On 2021-12-10 6:41 a.m., chen gong wrote:<o:p></o:p> </div> <blockquote style="margin-top:5.0pt;margin-bottom:5.0pt"> <pre>Play a video on the raven (or PCO, raven2) platform, and then do the S3<o:p></o:p></pre> <pre>test. When resume, the following error will be reported:<o:p></o:p></pre> <pre><o:p> </o:p></pre> <pre>amdgpu 0000:02:00.0: [drm:amdgpu_ring_test_helper [amdgpu]] *ERROR* ring<o:p></o:p></pre> <pre>vcn_dec test failed (-110)<o:p></o:p></pre> <pre>[drm:amdgpu_device_ip_resume_phase2 [amdgpu]] *ERROR* resume of IP block<o:p></o:p></pre> <pre><vcn_v1_0> failed -110<o:p></o:p></pre> <pre>amdgpu 0000:02:00.0: amdgpu: amdgpu_device_ip_resume failed (-110).<o:p></o:p></pre> <pre>PM: dpm_run_callback(): pci_pm_resume+0x0/0x90 returns -110<o:p></o:p></pre> <pre><o:p> </o:p></pre> <pre>[why]<o:p></o:p></pre> <pre>When playing the video: The power state flag of the vcn block is set to<o:p></o:p></pre> <pre>POWER_STATE_ON.<o:p></o:p></pre> <pre><o:p> </o:p></pre> <pre>When doing suspend: There is no change to the power state flag of the<o:p></o:p></pre> <pre>vcn block, it is still POWER_STATE_ON.<o:p></o:p></pre> <pre><o:p> </o:p></pre> <pre>When doing resume: Need to open the power gate of the vcn block and set<o:p></o:p></pre> <pre>the power state flag of the VCN block to POWER_STATE_ON.<o:p></o:p></pre> <pre>But at this time, the power state flag of the vcn block is already<o:p></o:p></pre> <pre>POWER_STATE_ON. The power status flag check in the "8f2cdef drm/amd/pm:<o:p></o:p></pre> <pre>avoid duplicate powergate/ungate setting" patch will return the<o:p></o:p></pre> <pre>amdgpu_dpm_set_powergating_by_smu function directly.<o:p></o:p></pre> <pre>As a result, the gate of the power was not opened, causing the<o:p></o:p></pre> <pre>subsequent ring test to fail.<o:p></o:p></pre> <pre><o:p> </o:p></pre> <pre>[how]<o:p></o:p></pre> <pre>In the suspend function of the vcn block, explicitly change the power<o:p></o:p></pre> <pre>state flag of the vcn block to POWER_STATE_OFF.<o:p></o:p></pre> <pre><o:p> </o:p></pre> <pre>Signed-off-by: chen gong <a href="mailto:curry.gong@amd.com" moz-do-not-send="true"><curry.gong@amd.com></a><o:p></o:p></pre> <pre>---<o:p></o:p></pre> <pre> drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c | 7 +++++++<o:p></o:p></pre> <pre> 1 file changed, 7 insertions(+)<o:p></o:p></pre> <pre><o:p> </o:p></pre> <pre>diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c<o:p></o:p></pre> <pre>index d54d720..d73676b 100644<o:p></o:p></pre> <pre>--- a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c<o:p></o:p></pre> <pre>+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c<o:p></o:p></pre> <pre>@@ -246,6 +246,13 @@ static int vcn_v1_0_suspend(void *handle)<o:p></o:p></pre> <pre> {<o:p></o:p></pre> <pre> int r;<o:p></o:p></pre> <pre> struct amdgpu_device *adev = (struct amdgpu_device *)handle;<o:p></o:p></pre> <pre>+ bool cancel_success;<o:p></o:p></pre> <pre>+<o:p></o:p></pre> <pre>+ cancel_success = cancel_delayed_work_sync(&adev->vcn.idle_work);<o:p></o:p></pre> </blockquote> [JZ] Can you refer to vcn_v3_0_stop , and add amdgpu_dpm_enable_uvd(adev, false); to the end of vcn_v1_0_stop?<o:p></o:p> See if it also can help. <o:p></o:p> <blockquote style="margin-top:5.0pt;margin-bottom:5.0pt"> <pre><o:p> </o:p></pre> <pre>+ if (cancel_success) {<o:p></o:p></pre> <pre>+ if (adev->pm.dpm_enabled)<o:p></o:p></pre> <pre>+ amdgpu_dpm_enable_uvd(adev, false);<o:p></o:p></pre> <pre>+ }<o:p></o:p></pre> <pre> <o:p></o:p></pre> <pre> r = vcn_v1_0_hw_fini(adev);<o:p></o:p></pre> <pre> if (r)<o:p></o:p></pre> </blockquote> </div> </div> </blockquote> </body> </html>