Hardware overlays in JB4.3 are broken with 4GiB of RAM

[EricNelson]

Hi all,

With KitKat just released, this may be a little late, but we just discovered a bug in the binary library hwcomposer_viv.imx6.so.

We've been testing our new Nitrogen6 Max board that has dual LVDS and 4GiB of RAM, and noticed that when using dual displays, the second display doesn't refresh properly. Instead, it often displays something a few frames behind, and sometimes appears to

Either checking the box in Settings|Disable HW Overlays or renaming /system/lib/hw/hwcomposer_viv.imx6.so so that it won't be found fixes the issue, and both displays operate correctly.

Similarly, passing a kernel command-line argument of mem=2G allows things to operate properly.

It's just an educated guess, but this appears to be another form of the bug addressed here in the GStreamer plugins, where an allocation test is checking for <0 to mean failure instead of testing for -1.

We haven't yet tested KitKat on this board, but will soon.

[CommunityBot · Accepted solution]

This an automatic process.

We are marking this post as solved, due to the either low activity or any reply marked as correct.

If you have additional questions, please create a new post and reference to this closed post.

NXP Community!

[HugoOsornio]

Hello Eric,

Does the issue appear also with a combination of HDMI and LVDS?

I am acquiring HW in order to reproduce the issue and start the root cause identification process.

Cheers,

Hugo

[EricNelson]

Yes. That's how we first saw it (Hannstar LVDS+HDMI).

We've also seen it with an RGB panel+HDMI and RGB+Hannstar, but those are not-often-used scenarios for us.

[karina_valencia]

Our team is  checking this case and  will provide an update ASAP.

[EricNelson]

Thanks Karina.

Note that in addition to the dual-display case, we also saw trouble with dual-channel LVDS (a 1080P display).

I think that can be tested on a SABRE Auto board.

[karina_valencia]

Hi Eric,

Ok, thanks   for the information.