On Tue, Sep 16, 2025 at 07:34:42PM +0200, Marek Vasut wrote: >On 9/16/25 12:06 PM, Rain Yang wrote: > >Hello everyone, > >> you're right. >> *0x4d810008=1, this register is a write-once register, so it was moved into >> SM >> since imx 6.12.3 release, and latest 6.6.52 release. some document work is >> still >> needed in the future. >> >> Hi Marek, >> thanks for your effort to make the i.MX95 GPU upstreamed. >> I created one PR to support i.MX95 GPU expcept the multi power domain[1], >> but it seemed to be placed in the wrong location. No one responded to it, >> so I closed it. >> >> for the suspend/resume issue, the panthor driver works well with 6.12.34 >> patched with pm_domains operation like you did. >> run vkmark->suspend 10min in the 4th console->wakeup in the 4th console. >> >> can you show me your SM version when uboot start, > >Build 470, Commit a07928b4 Sep 03 2025 23:43:49 > >> and the G310 firmware version which can be found by search git_sha string. > >That would be 50.2 . I now tried 53.0 and I suspect something might have been >fixed there ? > >As for the reset via block controller at 0x4d810000, without it, I get >SError, so it seems the reset is still needed. > >[ 2.788317] panthor 4d900000.gpu: [drm] clock rate = 800000000 >[ 2.795246] panthor 4d900000.gpu: EM: created perf domain >[ 2.801029] Internal error: synchronous external abort: 0000000096000010 >[#1] SMP >[ 2.808597] Modules linked in: >[ 2.811658] CPU: 3 UID: 0 PID: 52 Comm: kworker/u24:1 Tainted: G M >6.17.0-rc6-next-20250916-00076-ga73c9babac81-dirty #496 PREEMPT >[ 2.824939] Tainted: [M]=MACHINE_CHECK >[ 2.828684] Hardware name: Board (DT) >[ 2.834770] Workqueue: events_unbound deferred_probe_work_func >[ 2.840601] pstate: 60400009 (nZCv daif +PAN -UAO -TCO -DIT -SSBS >BTYPE=--) >[ 2.847545] pc : panthor_hw_init+0x18/0x620 >[ 2.851728] lr : panthor_device_init+0x388/0x5e0 >[ 2.856333] sp : ffff800082383aa0 >[ 2.859643] x29: ffff800082383aa0 x28: 0000000000000000 x27: >0000000000000000 >[ 2.866782] x26: ffff000080034c28 x25: 0000000000000000 x24: >0000000000000000 >[ 2.873909] x23: ffff00008036f010 x22: ffff000081082638 x21: >ffff00008036f010 >[ 2.881033] x20: ffff000081082000 x19: ffff000081082000 x18: >0000000000000801 >[ 2.888165] x17: 0000000000000000 x16: 0000000000000100 x15: >0000000000000001 >[ 2.895298] x14: 0140000000000000 x13: ffff800083800000 x12: >ffff800083c7ffff >[ 2.902422] x11: 000000004dd80000 x10: 0000000040000000 x9 : >ffff8000809696b8 >[ 2.909555] x8 : ffff8000823839a0 x7 : ffff000081082000 x6 : >ffff800082383a50 >[ 2.916679] x5 : ffff8000823839d0 x4 : ffff800081beb4d0 x3 : >ffff800081beb000 >[ 2.923803] x2 : 0000000000000000 x1 : ffff0000809a90c0 x0 : >ffff800083800000 >[ 2.930934] Call trace: >[ 2.933378] panthor_hw_init+0x18/0x620 (P) >[ 2.937555] panthor_device_init+0x388/0x5e0 >[ 2.941812] panthor_probe+0x40/0x68 >[ 2.945383] platform_probe+0x60/0xa8 >[ 2.949040] really_probe+0xc0/0x2b8 >[ 2.952611] __driver_probe_device+0x7c/0x120 >[ 2.956962] driver_probe_device+0x40/0x180 >[ 2.961130] __device_attach_driver+0xb4/0x140 >[ 2.965567] bus_for_each_drv+0x88/0xf8 >[ 2.969398] __device_attach+0xa0/0x1a0 >[ 2.973220] device_initial_probe+0x18/0x30 >[ 2.977397] bus_probe_device+0x9c/0xa8 >[ 2.981228] deferred_probe_work_func+0x88/0xc8 >[ 2.985743] process_one_work+0x158/0x3a8 >[ 2.989754] worker_thread+0x2d8/0x410 >[ 2.993498] kthread+0x144/0x200 >[ 2.996722] ret_from_fork+0x10/0x20 >[ 3.000296] Code: a9bb7bfd aa0003e7 910003fd f9430800 (b9400000) >[ 3.006378] ---[ end trace 0000000000000000 ]---
Hi Marek, sorry for the inconvienence. this uboot version don't include SM-184's change. I heard you're using i.MX95 A1 Chip, so you can extract the uboot in below link[1], or build from source, or raise one ticket in this website[2]. [1] https://www.nxp.com/webapp/Download?colCode=L6.6.52_2.2.1_MX95&appType=license&location=null [2] https://community.nxp.com/
