596144 - [optimus] Hardware acceleration makes select all operation slow (paints line by line)

Status: RESOLVED WORKSFORME

(Core :: Graphics, defect)

Description

[Mike Beltzner [:beltzner, not reading bugmail]]

Using a Sept 13 nightly build of Minefield, with following graphics infrastructure:

Graphics
Adapter Description     Intel(R) HD Graphics
Vendor ID               8086
Device ID               0046
Adapter RAM             Unknown
Adapter Drivers         igdumd64 igd10umd64 igdumdx32 igd10umd32
Driver Version          8.15.10.2119
Driver Date             4-21-2010
Direct2D Enabled        false
DirectWrite Enabled     false
GPU Accelerated Windows 1/1 Direct3D 9

STR:
1. Go to a wiki page or something else with a lot of text in a <textarea>
2. Select all using Ctrl+A

Actual: you can watch the selection paint, line-by-line
Expected: everything turns blue in the blink of an eye

Comment 1

[Mike Beltzner [:beltzner, not reading bugmail]]

Should have mentioned that when I turn off hw accel and restart, I see the expected behaviour.

Comment 2

[Mike Beltzner [:beltzner, not reading bugmail]]

Can reproduce in a clean profile, here's an AVI of it happening:

http://people.mozilla.org/~beltzner/bug596144.avi

Comment 3

[Timothy Nikkel (:tnikkel)]

The layer tree has nothing unexpected and I couldn't reproduce this on Windows 7 with ATI graphics card. Maybe it is specific to Intel?

Comment 4

[Mike Beltzner [:beltzner, not reading bugmail]]

I think it might actually be specific to an NVIDIA card. Even though about:support shows that I'm using an Intel card, I'm on a system that has two graphics cards. When I launch Firefox using the integrated graphics, things work fine. When I launch it using the NVIDIA card, I see the issue.

I'll bring the laptop to Joe tomorrow for further diagnosis.

Comment 5

[Mike Beltzner [:beltzner, not reading bugmail]]

Resolved by updating my NVIDIA card to the latest drivers; fwiw, the previous version (175 or something?) should probably be blocklisted. GeForce GT 335M

Comment 6

[Joe Drew (not getting mail)]

We need the previous version number of your drivers so we can actually blocklist it.

Comment 7

[Mike Beltzner [:beltzner, not reading bugmail]]

Any ideas on how I can get at those? Can you not just blocklist 100-199? I think Vlad was saying that everything pre-200 is pants.

Comment 8

[Mike Beltzner [:beltzner, not reading bugmail]]

Current version is 257.30 according to NVIDIA (http://grab.by/6BXd), or 8.17.12.5730 according to Windows (http://grab.by/6BXb)

Comment 9

[Joe Drew (not getting mail)]

We should block some older drivers on NVIDIA Optimus.

Comment 10

[Benoit Jacob [:bjacob] (mostly away)]

Need to blacklist old NVIDIA drivers with Optimus. Prerequisite: write code that detects Optimus.

Comment 11

[Asa Dotzler [:asa]]

should the summary of this bug be updated to reflect the real issue "blocklisting old NVIDIA Optimus drivers"?

Comment 12

[Benoit Jacob [:bjacob] (mostly away)]

This is going to be solved by the patch in bug 623338: we're going to be blocking all nvidia drivers older than 257.21  (june 2010).

Comment 13

[Bas Schouten (:bas.schouten)]

(In reply to comment #12)
> This is going to be solved by the patch in bug 623338: we're going to be
> blocking all nvidia drivers older than 257.21  (june 2010).

Except Optimus reports the intel device and driver version ....

Comment 14

[(no longer active)]

This seems like a hardblocker to me, unless we're willing to ship code which can break users running old Optimus cards...

Comment 15

[Benoit Jacob [:bjacob] (mostly away)]

(In reply to comment #13)
> (In reply to comment #12)
> > This is going to be solved by the patch in bug 623338: we're going to be
> > blocking all nvidia drivers older than 257.21  (june 2010).
> 
> Except Optimus reports the intel device and driver version ....

Ouch. That would mean that we need to implement bug 628129.

Comment 16

[Bas Schouten (:bas.schouten)]

(In reply to comment #15)
> (In reply to comment #13)
> > (In reply to comment #12)
> > > This is going to be solved by the patch in bug 623338: we're going to be
> > > blocking all nvidia drivers older than 257.21  (june 2010).
> > 
> > Except Optimus reports the intel device and driver version ....
> 
> Ouch. That would mean that we need to implement bug 628129.

This might be even trickier with Optimus, since it presents itself as a single GPU :(.

Comment 17

[Jeff Muizelaar [:jrmuizel]]

This doesn't happen on our optimus machine and the problem isn't that critical so I don't think this should hardblock. Further, adding more adding more blacklisting code (for detecting optimus) and blacklisting this situation is not risky enough to require a beta.

Comment 18

[Mike Beltzner [:beltzner, not reading bugmail]]

** PRODUCT DRIVERS PLEASE NOTE **

This bug is one of 7 automatically changed from blocking2.0:final+ to blocking2.0:.x during the endgame of Firefox 4 for the following reasons:

 - it was marked as a soft blocking issue without a requirement for beta coverage

Comment 19

[Joe Drew (not getting mail)]

This was fixed by blocking all old drivers in bug 623338.