On 8/7/24 06:44, Dmitry Baryshkov wrote:> Could you please clarify, I was under the impression that currently whole suspend/resume is broken, so it's more than a dmesg message.
71174f362d67 specifically, or v6.9 more broadly regress in that we get "[dpu error]connector not connected 3" and "[drm:drm_mode_config_helper_resume] *ERROR* Failed to resume (-22)" if suspending and resuming the system while external display is connected over USB-C DP. Suspend and resume itself still works, and the external display also works after the resume, albeit perhaps with a small delay due to the dpu error. This is also mentioned in the issue description of https://gitlab.freedesktop.org/drm/msm/-/issues/57. So while suspend/resume isn't fully broken, the error is still unexpected and I thus bisected and identified 71174f362d67 as the first commit to trigger it. While your patch avoids the dpu/drm error, it triggers issue with the CRTC state, breaking the CRTC functionality after resume.
Might we be facing a race condition here, which is accidentally exposed by 71174f362d67 but requires a separate fix?