[AMD Official Use Only - AMD Internal Distribution Only]
> -----Original Message-----
> From: Koenig, Christian <Christian.Koenig@xxxxxxx>
> Sent: October 16, 2024 9:16 AM
> To: Lazar, Lijo <Lijo.Lazar@xxxxxxx>; amd-gfx@xxxxxxxxxxxxxxxxxxxxx; Liu, Leo
> <Leo.Liu@xxxxxxx>
> Cc: Zhang, Hawking <Hawking.Zhang@xxxxxxx>; Deucher, Alexander
> <Alexander.Deucher@xxxxxxx>; Sundararaju, Sathishkumar
> <Sathishkumar.Sundararaju@xxxxxxx>; Jiang, Sonny
> <Sonny.Jiang@xxxxxxx>; Zhou, Hao (Claire) <Hao.Zhou@xxxxxxx>
> Subject: Re: [PATCH] drm/amdgpu: Save VCN shared memory with init reset
>
> Am 15.10.24 um 08:23 schrieb Lijo Lazar:
> > VCN shared memory is in framebuffer and there are some flags
> > initialized during sw_init. Ideally, such programming should be during
> hw_init.
>
> IIRC that was intentionally not done during hw_init for some reason.
> @Leo do you remember why?
>
We need to keep some of the status from share memory(driver and FW), since some of them are changed by FW, that is why the init cannot be in the hw_init stage with suspend/resume case.
+int amdgpu_vcn_suspend(struct amdgpu_device *adev) {
> + bool in_ras_intr = amdgpu_ras_intr_triggered();
> +
> + cancel_delayed_work_sync(&adev->vcn.idle_work);
> +
> + /* err_event_athub will corrupt VCPU buffer, so we need to
> + * restore fw data and clear buffer in amdgpu_vcn_resume() */
> + if (in_ras_intr)
> + return 0;
> +
> + return amdgpu_vcn_save_fw_shared_region(adev);
The saved bo here is not only for fw shared memory, also for FW runtime stack/heap as well.
Regards,
Leo
> +}
> +
