1233495 - OpenH264 plugin release v1.6

Status: RESOLVED FIXED

(Core :: Audio/Video: GMP, defect)

Description

[Hank Peng]

Opening as a meta bug to track all of the things we should get into the next release (>1.5.3) of OpenH264.

Comment 1

[Daniel Veditz [:dveditz]]

This fixes at least two identified sec-high bugs: we need to get this into the hands of Firefox users. The patches were committed two months ago, how much longer will it take to get a release? If the patches have been made to a branch doing feature work can we have an update off a branch from the current 1.5.3 release that just contains the security and stability fixes?

Comment 2

[Maire Reavy [:mreavy]]

We can absolutely do a release.  I think my open question is whether Cisco is ready to stabilize a full release of OpenH264 or prefers to cherry-pick the security fixes we need to do an update of the 1.5.3 release ASAP.

Hank -- I think you are best qualified to answer my open question (above).  What do you think?  Thanks.

Comment 3

[Hank Peng]

Maire, yeah, it is a time for a new release. Let me sync up with the team and see what is the best way to make a release. Thanks.

Comment 4

[Hank Peng]

Maire, we'd like to ship a full release of OpenH264. It will include all the sec-bugs found and some enhancements in rate control and simulcast support. The team needs 2 to 3 weeks to do a comprehensive regression and performance test. After that, we'll start to release it out. What do you think? Thanks.

Comment 5

[Maire Reavy [:mreavy]]

(In reply to Hank Peng from comment #4)
> Maire, we'd like to ship a full release of OpenH264. It will include all the
> sec-bugs found and some enhancements in rate control and simulcast support.
> The team needs 2 to 3 weeks to do a comprehensive regression and performance
> test. After that, we'll start to release it out. What do you think? Thanks.

I believe that proposed schedule is fine.  The only potential "fly in the ointment" is that my vacation is happening from June 18th through July 4th.  However, I can ask someone on my team to shepherd this release in my absence if we feel we're ready to test and release the update while I'm on PTO.

Dan -- Just needinfo'ing you so you see the proposed timeline.  To be clear, after Hank's team stabilized the release, my team would do some targeted testing for a few days and then we'd roll it out to Nightly, Aurora, Beta and Release over a week's time (or two weeks).  Obviously, if we found problems, we'd pause to investigate the problem which would extend the timeline.  We'd likely roll out Desktop and Mobile plugin updates at the same time.

Comment 6

[Liz Henry (:lizzard) (relman/hg->git project)]

Sounds good, that will be in the Firefox 47 timeframe. Note that the Mozilla all hands meeting will be between June 13-18 also. So if you are working with someone in releng to do the rollout let's let them know in advance too.

Comment 7

[Daniel Veditz [:dveditz]]

That timeline sounds fine.

Comment 8

[Hank Peng]

The gmp-openh264 v1.6 is ready to pick up. The code branch is v1.6-Firefox39: https://github.com/cisco/openh264/tree/v1.6-Firefox39.

The current tip commit is: https://github.com/cisco/openh264/commit/4fa3a4ec0367e11e779c202a09a94c3d18df3802. 

Please note that due to the fix of https://bugzilla.mozilla.org/show_bug.cgi?id=1225488 "Fix build process to automatically put openh264 pdb file into FF symbol server to help debug crash", now when build with option "BUILDTYPE=Release DEBUGSYMBOLS=True", the output binary libgmpopenh264.a/so is the stripped version and libgmpopenh264_debug_symbols.a/so contains the symbols for debugging. 
Please update your build script when integrating the new version of gmp-openh264 plugin.

Comment 9

[Hank Peng]

The current tip commit for "v1.6-Firefox39" branch is 97276f5c4da1eb450d26e666e473a32ef45a0bc3: Add an build option to disable the AVX2 assembly code for the lower version of nasm that doesn't support it.

Comment 10

[Sylvestre Ledru [:Sylvestre]]

Too late for 48.

Comment 11

[Liz Henry (:lizzard) (relman/hg->git project)]

This is live on nightly now according to email from Maire and Callek. 
Should we mark this fixed?

Comment 12

[Justin Wood (:Callek)]

My gut feeling is yes, but I'd defer to Maire on that for certain.

Comment 13

[Maire Reavy [:mreavy]]

We typically wait to close meta bugs like this until the update has been successfully deployed to all channels (Nightly through Release/ESR).  So once v1.6 is stable on Release/ESR (I'm targeting the end of next week), we can close this.

Comment 14

[Daniel Veditz [:dveditz]]

Release (48) and ESR-45.3 have 1.6