1299757 - Only use build id for Intel blocklist entry 'FEATURE_FAILURE_BUG_594877'

Status: RESOLVED FIXED

(Core :: Graphics, defect)

Description

[Michael Leu[:Lenzak](UTC+8)]

IMPLEMENT_INTEL_DRIVER_BLOCKLIST_D2D(OperatingSystem::WindowsVista, IntelGMA500,   V(7,14,10,1006), "FEATURE_FAILURE_594877_1");
    IMPLEMENT_INTEL_DRIVER_BLOCKLIST_D2D(OperatingSystem::WindowsVista, IntelGMA900,   GfxDriverInfo::allDriverVersions, "FEATURE_FAILURE_594877_2");
    IMPLEMENT_INTEL_DRIVER_BLOCKLIST_D2D(OperatingSystem::WindowsVista, IntelGMA950,   V(7,14,10,1504), "FEATURE_FAILURE_594877_3");
    IMPLEMENT_INTEL_DRIVER_BLOCKLIST_D2D(OperatingSystem::WindowsVista, IntelGMA3150,  V(7,14,10,2124), "FEATURE_FAILURE_594877_4");
    IMPLEMENT_INTEL_DRIVER_BLOCKLIST_D2D(OperatingSystem::WindowsVista, IntelGMAX3000, V(7,15,10,1666), "FEATURE_FAILURE_594877_5");
    IMPLEMENT_INTEL_DRIVER_BLOCKLIST_D2D(OperatingSystem::WindowsVista, IntelHDGraphicsToSandyBridge, V(8,15,10,2202), "FEATURE_FAILURE_594877_6");

    IMPLEMENT_INTEL_DRIVER_BLOCKLIST_D2D(OperatingSystem::Windows7, IntelGMA500,   V(5,0,0,2026), "FEATURE_FAILURE_594877_7");
    IMPLEMENT_INTEL_DRIVER_BLOCKLIST_D2D(OperatingSystem::Windows7, IntelGMA900,   GfxDriverInfo::allDriverVersions, "FEATURE_FAILURE_594877_8");
    IMPLEMENT_INTEL_DRIVER_BLOCKLIST_D2D(OperatingSystem::Windows7, IntelGMA950,   V(8,15,10,1930), "FEATURE_FAILURE_594877_9");
    IMPLEMENT_INTEL_DRIVER_BLOCKLIST_D2D(OperatingSystem::Windows7, IntelGMA3150,  V(8,14,10,2117), "FEATURE_FAILURE_594877_10");
    IMPLEMENT_INTEL_DRIVER_BLOCKLIST_D2D(OperatingSystem::Windows7, IntelGMAX3000, V(8,15,10,1930), "FEATURE_FAILURE_594877_11");
    IMPLEMENT_INTEL_DRIVER_BLOCKLIST_D2D(OperatingSystem::Windows7, IntelHDGraphicsToSandyBridge, V(8,15,10,2202), "FEATURE_FAILURE_594877_12");

Per discussion in bug1294771, we should investigate what we are trying to block and see if we should use build id for this.

[1]https://dxr.mozilla.org/mozilla-central/source/widget/windows/GfxInfo.cpp#989

Comment 1

[Michael Leu[:Lenzak](UTC+8)]

Attachment: bug1299757 - Only use build id for Intel blocklist

Comment 2

[Michael Leu[:Lenzak](UTC+8)]

Comment on attachment 8788085 [details] [diff] [review]
bug1299757 - Only use build id for Intel blocklist

The latest Vista driver for Intel Graphics is 2869 which is quite old comparing to current driver (4425), so I think we should not unify Windows7 and Windows Vista's condition to prevent over-block.

Comment 3

[Milan Sreckovic [:milan] (needinfo for best results)]

To understand better - are there Intel driver versions on Windows 7 that are not 8.x.x.x, and on Windows Vista that are not 7.x.x.x?

Comment 4

[Michael Leu[:Lenzak](UTC+8)]

(In reply to Milan Sreckovic [:milan] from comment #3)
> To understand better - are there Intel driver versions on Windows 7 that are
> not 8.x.x.x, and on Windows Vista that are not 7.x.x.x?

It's possible that somebody uses Windows Vista and updates to Windows 7 without graphics driver updated.
In that case, the user runs Windows 7, but the driver would stay in 7.x.x.x.
There are similar situation in Windows7, 8, 10, too.

So I think the first version number is not that accurate to describe users' Windows version.

Comment 5

[Milan Sreckovic [:milan] (needinfo for best results)]

Anthony, let's keep an eye out on these graphics cards, and see if we get a change in crash frequency:
0x8108, 0x8109, 0x2582, 0x2782, 0x2592, 0x2792, 0x2772, 0x2776
0x27a2, 0x27a6, 0x27ae, 0xa00, 0xa00, 0xa01, 0xa012, 0x2972, 0x010a
0x2973, 0x2982, 0x2983, 0x2992, 0x2993, 0x29a2, 0x29a3, 0x29b
0x29b3, 0x29c2, 0x29c3, 0x29d2, 0x29d3, 0x2a02, 0x2a03, 0x2a12
0x2a13, 0x0042, 0x0102, 0x0106, 0x0112, 0x0116, 0x0122, 0x0126

Comment 6

[u279076]

(In reply to Milan Sreckovic [:milan] from comment #5)
> Anthony, let's keep an eye out on these graphics cards

Due to the sheer number of device IDs it's not possible for me to track these individually. However I have set up a chart that will track Nightly crashes for these devices in aggregate.

http://ashughes1.github.io/metrics-graphics-gfx/#v2-bug-1299757

Comment 7

[Milan Sreckovic [:milan] (needinfo for best results)]

Thanks - I should have been clearer - in the aggregate is perfect.

Comment 8

[Milan Sreckovic [:milan] (needinfo for best results)]

Michael, please this graph over the next couple of weeks and see whether we can see changes (good or bad) based on landing this.

Comment 9

[Michael Leu[:Lenzak](UTC+8)]

OK, I will keep track of the crash rate after this bug landed.

Comment 10

[Michael Leu[:Lenzak](UTC+8)]

Here is tryserver result
https://treeherder.mozilla.org/#/jobs?repo=try&revision=ce52df760bff

Comment 11

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/39135ac0e961
Only use build id for Intel blocklist entry 'FEATURE_FAILURE_BUG_594877'. r=ethlin, r=milan

Comment 12

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/39135ac0e961