On 1/4/23 04:29, Andy Shevchenko wrote: > On Tue, Jan 03, 2023 at 09:08:20AM -0600, Pierre-Louis Bossart wrote: >> On 1/2/23 14:30, Andy Shevchenko wrote: >>> Instead of calling put_device(&adev->dev) where adev is a pointer >>> to an ACPI device, use specific call, i.e. acpi_dev_put(). >>> >>> Also move it out of the conditional to make it more visible in case >>> some other code will be added which may use that pointer. We need >>> to keep a reference as long as we use the pointer. >>> >>> Signed-off-by: Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx> >> >> Answering for the series: we should make the change across all Intel >> machine drivers. I see at least four cases that were missed >> >> bytcr_rt5640.c: put_device(&adev->dev); >> bytcr_rt5651.c: put_device(&adev->dev); >> bytcr_wm5102.c: put_device(&adev->dev); >> sof_es8336.c: put_device(&adev->dev); > > Aren't they (they all problematic, btw) covered by the fixes series > https://lore.kernel.org/r/20230102203014.16041-1-andriy.shevchenko@xxxxxxxxxxxxxxx? They are indeed, but if you group AMD-related patches with Intel ones, it's only human for reviewers to skip the thread entirely, even more so when catching up with email on January 3 :-) For this series Acked-by: Pierre-Louis Bossart <pierre-louis.bossart@xxxxxxxxxxxxxxx>
