1082929 - [OMTC] 100% kernel CPU usage while dragging OpenStreetMap/GoogleMapsClassic

Status: RESOLVED FIXED

(Core :: Graphics, defect)

Description

[Piotr Jurkiewicz]

User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:33.0) Gecko/20100101 Firefox/33.0
Build ID: 20141011015303

Steps to reproduce:

Go to http://www.openstreetmap.org or https://www.google.com/maps?output=classic.

Drag a map.


Actual results:

Starting with version 33 on Windows 7, the CPU usage is very high while dragging a map. On my i5-3570k it fully uses 2/4 CPU cores! 95% of this load is kernel time.

Moreover, dragging is VERY sluggish and feels unresponsive, comparing to Firefox 32.

This is my graphics support information:

"graphics": {
    "numTotalWindows": 1,
    "numAcceleratedWindows": 1,
    "windowLayerManagerType": "Direct3D 11",
    "windowLayerManagerRemote": true,
    "adapterDescription": "Intel(R) HD Graphics 4000",
    "adapterVendorID": "0x8086",
    "adapterDeviceID": "0x0162",
    "adapterRAM": "Unknown",
    "adapterDrivers": "igdumdim64 igd10iumd64 igd10iumd64 igdumdim32 igd10iumd32 igd10iumd32",
    "driverVersion": "10.18.10.3412",
    "driverDate": "1-29-2014",
    "adapterDescription2": "",
    "adapterVendorID2": "",
    "adapterDeviceID2": "",
    "adapterRAM2": "",
    "adapterDrivers2": "",
    "driverVersion2": "",
    "driverDate2": "",
    "isGPU2Active": false,
    "direct2DEnabled": true,
    "directWriteEnabled": true,
    "directWriteVersion": "6.2.9200.16571",
    "webglRenderer": "Google Inc. -- ANGLE (Intel(R) HD Graphics 4000 Direct3D9Ex vs_3_0 ps_3_0)",
    "info": {
      "AzureCanvasBackend": "direct2d",
      "AzureSkiaAccelerated": 0,
      "AzureFallbackCanvasBackend": "cairo",
      "AzureContentBackend": "direct2d"
    }
}

Comment 1

[Piotr Jurkiewicz]

Attachment: Hide left panel on GoogleMaps to speed up dragging

Moreover, I have discovered a very interesting thing. If you hide the left panel on Google Maps Classic (see attached screenshot), kernel CPU overload disappear and dragging is blazingly fast now. I think that it can be an useful trace for hunting this bug.

I have not discovered such a workaround for OpenStreetMap yet.

Comment 2

[Alice0775 White]

Could you test with  Nightly36.0a1 ( https://nightly.mozilla.org/ ) ?

Comment 3

[Piotr Jurkiewicz]

On Nightly 36.0a1 the problem is gone.

Comment 4

[Alice0775 White]

Regression window(m-i)
Good:
https://hg.mozilla.org/integration/mozilla-inbound/rev/bb8fc1380b00
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:32.0) Gecko/20100101 Firefox/32.0 ID:20140519202628
Bad:
https://hg.mozilla.org/integration/mozilla-inbound/rev/46d9ffb97fe3
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:32.0) Gecko/20100101 Firefox/32.0 ID:20140519224628
Pushlog:
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=bb8fc1380b00&tochange=46d9ffb97fe3

Regressed by: 
  46d9ffb97fe3	Bas Schouten — Bug 899785: Switch on OMTC and async video on windows. r=BenWa


Progression window(m-i)
Bad:
https://hg.mozilla.org/integration/mozilla-inbound/rev/2badc0f1f829
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:35.0) Gecko/20100101 Firefox/35.0 ID:20140914114608
Fixed:
https://hg.mozilla.org/integration/mozilla-inbound/rev/8e28464849fa
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:35.0) Gecko/20100101 Firefox/35.0 ID:20140914145408
Progression pushlog:
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=2badc0f1f829&tochange=8e28464849fa

Fixed by:
	7861be6bf039	Bas Schouten — Bug 1060588: Use PushClipRect when possible in ClipToRegionInternal. r=jrmuizel

Comment 5

[Lawrence Mandel [:lmandel] (use needinfo)]

First attempt to fix this is by uplifting the patch in bug 1060588, which now has beta approval.

Comment 6

[Lawrence Mandel [:lmandel] (use needinfo)]

Does this issue still reproduce with Firefox 34 beta2 or later?

Comment 7

[Piotr Jurkiewicz]

I have tested 34 beta 4. This issue does not occur.

Comment 8

[Arthur K. (he/him)]

(In reply to Piotr Jurkiewicz from comment #0)
> User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:33.0) Gecko/20100101
> Firefox/33.0
> Build ID: 20141011015303
> 
> Steps to reproduce:
> 
> Go to http://www.openstreetmap.org or
> https://www.google.com/maps?output=classic.
> 
> Drag a map.
> 
> 
> Actual results:
> 
> Starting with version 33 on Windows 7, the CPU usage is very high while
> dragging a map. On my i5-3570k it fully uses 2/4 CPU cores! 95% of this load
> is kernel time.
> 
> Moreover, dragging is VERY sluggish and feels unresponsive, comparing to
> Firefox 32.
> 
> This is my graphics support information:
> 
> "graphics": {
>     "numTotalWindows": 1,
>     "numAcceleratedWindows": 1,
>     "windowLayerManagerType": "Direct3D 11",
>     "windowLayerManagerRemote": true,
>     "adapterDescription": "Intel(R) HD Graphics 4000",
>     "adapterVendorID": "0x8086",
>     "adapterDeviceID": "0x0162",
>     "adapterRAM": "Unknown",
>     "adapterDrivers": "igdumdim64 igd10iumd64 igd10iumd64 igdumdim32
> igd10iumd32 igd10iumd32",
>     "driverVersion": "10.18.10.3412",
>     "driverDate": "1-29-2014",
>     "adapterDescription2": "",
>     "adapterVendorID2": "",
>     "adapterDeviceID2": "",
>     "adapterRAM2": "",
>     "adapterDrivers2": "",
>     "driverVersion2": "",
>     "driverDate2": "",
>     "isGPU2Active": false,
>     "direct2DEnabled": true,
>     "directWriteEnabled": true,
>     "directWriteVersion": "6.2.9200.16571",
>     "webglRenderer": "Google Inc. -- ANGLE (Intel(R) HD Graphics 4000
> Direct3D9Ex vs_3_0 ps_3_0)",
>     "info": {
>       "AzureCanvasBackend": "direct2d",
>       "AzureSkiaAccelerated": 0,
>       "AzureFallbackCanvasBackend": "cairo",
>       "AzureContentBackend": "direct2d"
>     }
> }

Try bumping driver up to 10.18.10.3958.

Comment 9

[Lawrence Mandel [:lmandel] (use needinfo)]

Thank you for checking Piotr.

Resolving as fixed based on comment 7.