On Thu, Jan 27, 2022 at 1:44 PM Martin Faltesek <mfaltesek@xxxxxxxxxx> wrote: +stable@ > > +gregkh@xxxxxxxxxxxxxxxxxxx > > ddbcd0c58a6a is upstream fix for this and needs to be applied to affected stable releases. > My little tool tells me that this is only needed in v5.10.y. Guenter > On Thu, Jan 27, 2022 at 3:21 PM Martin Faltesek <mfaltesek@xxxxxxxxxx> wrote: >> >> Hi Bryan, >> >> I was merging this patch into our code base that was in 5.10.94: >> >> commit eeefa2eae8fc82ad757a2241b9f82ac33e99e6b4 >> Author: Bryan O'Donoghue <bryan.odonoghue@xxxxxxxxxx> >> AuthorDate: Fri Feb 5 19:11:49 2021 +0100 >> Commit: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> >> CommitDate: Thu Jan 27 10:53:51 2022 +0100 >> >> media: venus: core, venc, vdec: Fix probe dependency error >> >> >> >> @@ -364,11 +365,14 @@ static int venus_remove(struct platform_device *pdev) >> pm_runtime_disable(dev); >> >> if (pm_ops->core_put) >> - pm_ops->core_put(dev); >> + pm_ops->core_put(core); >> + >> + v4l2_device_unregister(&core->v4l2_dev); >> >> hfi_destroy(core); >> >> v4l2_device_unregister(&core->v4l2_dev); >> + >> mutex_destroy(&core->pm_lock); >> mutex_destroy(&core->lock); >> venus_dbgfs_deinit(core); >> >> This section introduces unregistering twice, before and after destroy. >> I think there must be a patch that fixed this since this error does >> not exist in linus's tree. >> I'm guessing there might be a missing patch you'd want to grab. >> >> Can you comment on this? >> >> Thanks, >> >> Martin
