Open Bug 1889489 Opened 2 years ago Updated 2 years ago

Prime video not playing movies without them constantly buffering

Categories

(Core :: Audio/Video: Playback, defect, P3)

Firefox 124
defect

Tracking

()

UNCONFIRMED

People

(Reporter: grantrussell34, Unassigned)

Details

Attachments

(1 file)

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:124.0) Gecko/20100101 Firefox/124.0

Steps to reproduce:

Prime video not playing movies without them constantly buffering. It works fine on other browsers and netflix, youtube and twitch all work fine.

Shows on Prime also work fine its just the movies.

Things I have tried: Clear cache/browsing data, fresh install, completly deleting program files, driver updates

Actual results:

Prime video not playing movies without them constantly buffering. It works fine on other browsers and netflix, youtube and twitch all work fine.

Shows on Prime also work fine its just the movies.

Things I have tried: Clear cache/browsing data, fresh install, completly deleting program files, driver updates

Expected results:

Prime video not playing movies without them constantly buffering. It works fine on other browsers and netflix, youtube and twitch all work fine.

Shows on Prime also work fine its just the movies.

Things I have tried: Clear cache/browsing data, fresh install, completly deleting program files, driver updates

Component: Untriaged → Audio/Video: Playback
Product: Firefox → Core

Hi, thanks for the report! Does this behavior happen on nightly as well? Would you be able to share the contents of your about:support page and some logging information? To collect and share logging information, please follow the directions in this video. Please make sure that you start logging before recreating the bug and stop logging immediately after you've recreated the bug. If you can paste the permanent link it gives you at the end of the process here, it would be helpful. Thank you!

Flags: needinfo?(grantrussell34)
Flags: needinfo?(grantrussell34)
Blocks: media-triage
Severity: -- → S3
Priority: -- → P3
Flags: needinfo?(alwu)

For a preliminary investigation, I will share how I diagnose the issue here.


tl;dr, the issue would probably be related with CDM operations, either ChromiumCDMChild or ChromiumCDMParent, or Amazon operates EME API incorrectly. But that needs a further investigation.

  1. roughly check the overall markers to understand what happened

Go to MediaDecoderStateMachine (MDSM) thread's marker chart, we can see MDSM switched betweeb "decoding" to "buffering" multiple times, which matches what the reporter described. Okay, then we can narrow the time range to just one transition to take a closer look to first 3.5s.

From PushAudio and PushVideo markers (there is an attribute queue length), we can know that video seems not enough comparing with audio, which could be the reason of entering buffering. Also, the last video MDSM received is [4462,791, 450,4499], which seems wrong, we shouldn't have a data which end time is smaller than its start time. But we didn't know what exact problem it's yet, so just keep this in mind and debug this frame later.

  1. check the flow of decoded video frames

Video is decoded by GMP process, so we start checking GMP process. Now we can see the last video frame GMP process decoded has timestamp 4838166, but it didn't show up in MDSM's video queue. There must be something wrong. Now check the content process GMP thread to see if we receive those data, we can see the last received video frame is 4713041, so there were some frames already missing (because we can see more frames decoded in the GMP process).

But actually more frames were missing. When we check MediaFormatReader (MFR), the last output we received was [4462,791, 450,4499], which was the last frame pushed into MDSM's video queue. So now we need to figure out why those frames were missing.

  1. Keep tracing the missing data

So now we want to check full logs for the GMP process and GMP thread in the content process in order to see if there are any suspicious logs. We want to focus the time around where GMP decoded the last frame, which is around 2.460s.

In the GMP process side, we can see ChromiumCDMChild::TimerExpired(context=0x21), so it seems the certification was expired and Widevine was asking website to update the credential again. Was this possibly affect the video decoding?


Now we can keep investigating the issue by above directions. Is this time-expiration unexpected causing losing frame? maybe some status reset? or abandon some decoding promises? Is the time expire expected?

Let me know if you would like to keep debugging this issue. If not, feel free to NI me and then I can do another round of investigation.

Flags: needinfo?(alwu) → needinfo?(azebrowski)

Are you still seeing this? Also, can you post your about:support text? Thanks.

Flags: needinfo?(grantrussell34)

Hi yes the bug is still persisting

How do I post the about support text?

Flags: needinfo?(grantrussell34)

(In reply to Jim Mathies [:jimm] from comment #4)

Are you still seeing this? Also, can you post your about:support text? Thanks.

Hi yes the bug is still persisting

How do I post the about support text?

Try running in safe mode to temporary disable extensions and see if you have the same problem.

https://www.wikihow.com/Start-Firefox-in-Safe-Mode

Flags: needinfo?(azebrowski) → needinfo?(grantrussell34)

(In reply to Jim Mathies [:jimm] from comment #8)

Try running in safe mode to temporary disable extensions and see if you have the same problem.

https://www.wikihow.com/Start-Firefox-in-Safe-Mode

doesnt work still

Flags: needinfo?(grantrussell34)

Hmm, we're unsure of the cause. Might be something Malwarebytes is doing. Hard to say. At this point we only have one report, so not actionable on our end at this point.

No longer blocks: media-triage
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: