Bug 21820

Are you trying to fullscreen the video (eg exceed the maximum overlay image size)?

No. My test video is 480x270, and I am attempting to display it at is native size.  (The maximum XVideo size is listed as 1920x1088.)  In fact, my whole screen is only 1024x768.  I am running with a very wide Virtual screen so that I can plug in my 1440x900 external monitor when I go home.

Does this only happens when you set virtual>2048?

Does this only happen for overlay video? i.e. using textured video is fine?

Please attach your xorg.conf and Xorg.0.log.

(In reply to comment #3)
> Does this only happen for overlay video? i.e. using textured video is fine?

Ah, textured video is disabled for width>2048 (bug#16914)

Adding me to CC to track equiv mdv bug: https://qa.mandriva.com/show_bug.cgi?id=50857

FYI from the Mdv bug:
xvinfo: https://qa.mandriva.com/attachment.cgi?id=13919
Xorg.0.log: https://qa.mandriva.com/attachment.cgi?id=13846

Created attachment 26040 [details]
Xorg.0.log immediately after attempting to display a video

Created attachment 26041 [details]
xorg.conf in use at the time of video testing

Uploaded Xorg.0.log and xorg.conf.  The only thing to see in the logfile is lots of

(EE) intel(0): Failed to pin xv buffer

Aha!  I set the virtual screen back below 2048, and now xvinfo shows two adaptors: Textured Video and Overlay.  Using mplayer, I can confirm that the Textured Video adaptor works fine, and the Overlay adaptor does not work at all.  Specifically, attempting use the Overlay adaptor gives exactly the errors I've been describing.

So what's really happening is that Textured Video is adaptor number 0, the default, when width<2048.  For width>2048, Overlay becomes the default, but overlay doesn't work, regardless of screen size.

For the sake of completeness, I guess it's worth repeating that I am using intel driver 2.7.1, without KMS, on a vanilla linux kernel 2.6.27 and xorg-server 1.5.3.  I discovered this problem when running with Virtual 2464x1024, but I have since reproduced the essential bug (use of Overlay gives BadAlloc) even at Virtual sizes less than 2048.

I have a GMA950 chipset, in a Thinkpad x60s.

FWIW, in the Mandriva bug listed above, the user only has the one Overlay (no Textured Video Overlay) on his h/w. His virtual size is well below the 2048^2 limit. So due to this bug he is unable to play any Xv video :(

This is true. This is my bug at the Mandriva Bugzilla LOL

All the info is there in that bug listing. Same problem, I can watch flash and some plugins with mplayer work...but not calling up the Xv engine. Of the plugins that work (this is within browser and on website) they will play up to 20 seconds and then stop. Apps such as Mplayer Xine VLC and cheese have been tested. Xine and Cheese come up and then disappear. There has been some improvement of late. Used to be that the system would encounter these apps and would completely crash X to the point of the login page. 

Dunno why...Everything was good with the prior version with Mandriva 2009.0 then something updated and now, no joy.

This card is:

lspci -vvv
00:00.0 Host bridge: Intel Corporation 82845G/GL[Brookdale-G]/GE/PE DRAM Controller/Host-Hub Interface (rev 01)
	Subsystem: Dell Device 0160
	Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR+ FastB2B- DisINTx-
	Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort+ >SERR- <PERR- INTx-
	Latency: 0
	Region 0: Memory at f0000000 (32-bit, prefetchable) [size=128M]
	Capabilities: [e4] Vendor Specific Information <?>
	Kernel driver in use: agpgart-intel
	Kernel modules: intel-agp

00:02.0 VGA compatible controller: Intel Corporation 82845G/GL[Brookdale-G]/GE Chipset Integrated Graphics Device (rev 01) (prog-if 00 [VGA controller])
	Subsystem: Dell Device 0160
	Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
	Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
	Latency: 0
	Interrupt: pin A routed to IRQ 16
	Region 0: Memory at e8000000 (32-bit, prefetchable) [size=128M]
	Region 1: Memory at feb80000 (32-bit, non-prefetchable) [size=512K]
	Expansion ROM at <unassigned> [disabled]
	Capabilities: [d0] Power Management version 1
		Flags: PMEClk- DSI+ D1- D2- AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
		Status: D0 PME-Enable- DSel=0 DScale=0 PME-
	Kernel modules: intelfb


AKA the i810 integrated

This is also giving the same bad boots that the Ubuntu groups have noted. IE, system starts and then goes to black screen and full lock up during boot. Only fix there was to power off using power button. Went through several last night and then it came back. Granted, Ubuntu's solution was to disable compiz and 3D efx. I don't have either running. 

Please do take a look at https://qa.mandriva.com/show_bug.cgi?id=50857 and see if there is something that I am missing or that should be implimented

Thanks
Rich

Getting the same problem here... blue xvideo window and BadAlloc from mplayer.  System here is FreeBSD-7.2 with xorg-7.4_2 and xf86-video-intel-2.7.1.  Video card reports as:

May 21 08:36:44 linwhf kernel: agp0: <Intel 82865G (865G GMCH) SVGA controller> on vga
pci0
May 21 08:36:44 linwhf kernel: agp0: detected 8060k stolen memory
May 21 08:36:44 linwhf kernel: agp0: aperture size is 128M
May 21 08:38:05 linwhf kernel: drm0: <Intel i865G GMCH> on vgapci0
May 21 08:38:05 linwhf kernel: vgapci0: child drm0 requested pci_enable_busmaster
May 21 08:38:05 linwhf kernel: info: [drm] AGP at 0xf0000000 128MB
May 21 08:38:05 linwhf kernel: info: [drm] Initialized i915 1.6.0 20080730
May 21 08:38:06 linwhf kernel: drm0: [ITHREAD]

xvinfo:
X-Video Extension version 2.2
screen #0
  Adaptor #0: "Intel(R) Video Overlay"
    number of ports: 1
    port base: 76
    operations supported: PutImage 
    supported visuals:
      depth 24, visualID 0x21
    number of attributes: 6

I'm guessing from reading this thread that the cause is known, so I won't post configs and logs, but if they're needed, just ask.

*** Bug 21925 has been marked as a duplicate of this bug. ***

Anything else I can do to help with this?

Rich

video overlay works well on my 945GM, with upstream driver (xf86-video-intel master branch tip).

Could any of you try the master branch?

I have just tried compiling the xf86-video-intel code from git.  Not good.

Not only does the xvideo BadAlloc/video blue screen problem remain, but with this version, I now also have serious loss of pixels in xterm to the point where about 60% of xterm lines are totally blanked out and it is therefore pretty much unusable.  No problem like this at all with the 2.7.1 version where I only had the BadAlloc problem.

System is FreeBSD-7.2 (well, 7-STABLE from 2009/05/15).

Video hardware is:
May 28 11:37:00 linwhf kernel: vga0: <Generic ISA VGA> at port 0x3c0-0x3df iomem 0xa0000-0xbffff on isa0

May 28 11:37:00 linwhf kernel: agp0: <Intel 82865G (865G GMCH) SVGA controller> on vgapci0
May 28 11:37:00 linwhf kernel: agp0: detected 8060k stolen memory
May 28 11:37:00 linwhf kernel: agp0: aperture size is 128M
May 28 11:37:07 linwhf kernel: drm0: <Intel i865G GMCH> on vgapci0
May 28 11:37:07 linwhf kernel: vgapci0: child drm0 requested pci_enable_busmaster
May 28 11:37:07 linwhf kernel: info: [drm] AGP at 0xf0000000 128MB
May 28 11:37:07 linwhf kernel: info: [drm] Initialized i915 1.6.0 20080730

libdrm is 2.4.11.

I'm more intested in Linux test result.

I tried compiling the master branch, but it failed with the error

Requested 'xorg-server >= 1.6' but version of xorg-server is 1.5.3

Version 2.7.1 compiles fine against xorg-server 1.5.3

Mine still follows the same crash. Even with later updates from Mandriva. Not sure what those updates were to fix....so the Intel Corporation 82845G/GL[Brookdale-G] is still in a bad place.

(In reply to comment #20)
> Mine still follows the same crash. Even with later updates from Mandriva.

Rich, Keith was talking about git master, not the versions Mandriva ship as updates. Have you compiled your own version from git master, (I suspect you are probably not :)).

I should be able to to a Linux test with the git master build in a few days, but someone is welcome to do the test first. (And apologies if you have done the test already Rich :))

(In reply to comment #21)
> (In reply to comment #20)
> > Mine still follows the same crash. Even with later updates from Mandriva.
> 
> Rich, Keith was talking about git master, not the versions Mandriva ship as
> updates. Have you compiled your own version from git master, (I suspect you are
> probably not :)).
> 
> I should be able to to a Linux test with the git master build in a few days,
> but someone is welcome to do the test first. (And apologies if you have done
> the test already Rich :))
> 

Thanks for the reply and you are right I have not done the git master. Would there be an advantage to doing this?

Thanks again
Rich

> Thanks for the reply and you are right I have not done the git master. Would
> there be an advantage to doing this?

It's the information Keith was looking for in his comment further up.

Actually, in addition to my last comment, Keith, I have tried both ports 
on my system with the same 2.7.1 driver that fails for Rich and both my
   Adaptor #0: "Intel(R) Textured Video"
and
   Adaptor #1: "Intel(R) Video Overlay"

work fine, so I suspect if your test is similar from git master, it 
could give a false positive in the same way mine does.

I'll try and build a git master driver for Rich to test, but I suspect 
the problem will remain.

Col

Gah! In my last comment I mentioned "Keith" (meaning Keith Packard) but I actually meant to refer to Gordon Jin's comment. Sorry for the confusion!

/me was trying to help coordinate the testing on the bug but just adds to the confusion instead. I'll get more sleep!

FWIW, Gordon, I can use my both XV overlays fine on the same driver that fails for Rich (same Mandriva package), so I suspect your test on an 945GM (as I have) will probably succeed too with a false positive from git master. But I will get a git master package made for Rich so he can test properly and report back with more concrete results.

(In reply to comment #25)
> Gah! In my last comment I mentioned "Keith" (meaning Keith Packard) but I
> actually meant to refer to Gordon Jin's comment. Sorry for the confusion!
> 
> /me was trying to help coordinate the testing on the bug but just adds to the
> confusion instead. I'll get more sleep!
> 
> FWIW, Gordon, I can use my both XV overlays fine on the same driver that fails
> for Rich (same Mandriva package), so I suspect your test on an 945GM (as I
> have) will probably succeed too with a false positive from git master. But I
> will get a git master package made for Rich so he can test properly and report
> back with more concrete results.
> 

Colin you still rock no matter what you say about yourself...LOL...Thanks guys for the work you do. This is why I stay with Linux and not try to run off to some other OS.

Rich

XVideo overlay also works fine on 945GM for me with the following components:
xf86-video-intel:4afbe3b75e53cca1254883f2e847cf93cf9f0d71
drm: 2cb4c64d7310904b354365c2cbc263211e9eb4a1
xserver: server-1.6-branch(5cd5a01259ba349f1868ca4af04207cf120d69e4)
Linux OS: 2.6.30-rc8.

So I mark it as fixed. If anyone still experiences the same problem, feel free to reopen it.

(In reply to comment #27)
> XVideo overlay also works fine on 945GM for me with the following components:
> xf86-video-intel:4afbe3b75e53cca1254883f2e847cf93cf9f0d71
> drm: 2cb4c64d7310904b354365c2cbc263211e9eb4a1
> xserver: server-1.6-branch(5cd5a01259ba349f1868ca4af04207cf120d69e4)
> Linux OS: 2.6.30-rc8.
> 
> So I mark it as fixed. If anyone still experiences the same problem, feel free
> to reopen it.
> 

My problem is with the 845GM. Can I modify this bug in the summary to say:

[845GM] XVideo Overlay gets BadAlloc error and then reopen it? I think that the two problems are related so even though the 945 is ok, the 845 is not, with the same error. 

Direction??

The problem also persists for me on the i865G, which I have reported twice earlier in this thread.  To close this ticket, ignoring this information, is mind-blowingly belligerent, if you ask me.

I am re-opening it (and adjusting the Summary, as the previous commenter suggested).

(In reply to comment #29)
> The problem also persists for me on the i865G, which I have reported twice
> earlier in this thread.  To close this ticket, ignoring this information, is
> mind-blowingly belligerent, if you ask me.
> 
> I am re-opening it (and adjusting the Summary, as the previous commenter
> suggested).
> 

Very cool. Neat to see that was all that needed to be done

Thanks

Rich

(In reply to comment #30)
> (In reply to comment #29)
> > The problem also persists for me on the i865G, which I have reported twice
> > earlier in this thread.  To close this ticket, ignoring this information, is
> > mind-blowingly belligerent, if you ask me.
> > 
> > I am re-opening it (and adjusting the Summary, as the previous commenter
> > suggested).
> > 
> 
> Very cool. Neat to see that was all that needed to be done
> 
> Thanks
> 
> Rich
> 


UPDATE
colin working on a solution, new driver in testing from git does not work in mandriva. Still working on a solution.

Hopefully he will be along in a bit to explain...

Rich

I have tested Colin's git packages, I still have no Xv working with my 82830:

[goetz@goetz ~]$ xvinfo
X-Video Extension version 2.2
screen #0
  Adaptor #0: "Intel(R) Video Overlay"
    number of ports: 1
    port base: 65
    operations supported: PutImage 
    supported visuals:
      depth 16, visualID 0x21
    number of attributes: 6
      "XV_COLORKEY" (range 0 to 16777215)
              client settable attribute
              client gettable attribute (current value is 2110)
      "XV_BRIGHTNESS" (range -128 to 127)
              client settable attribute
              client gettable attribute (current value is -19)
      "XV_CONTRAST" (range 0 to 255)
              client settable attribute
              client gettable attribute (current value is 75)
      "XV_SATURATION" (range 0 to 1023)
              client settable attribute
              client gettable attribute (current value is 146)
      "XV_DOUBLE_BUFFER" (range 0 to 1)
              client settable attribute
              client gettable attribute (current value is 1)
      "XV_PIPE" (range -1 to 1)
              client settable attribute
              client gettable attribute (current value is -1)
    maximum XvImage size: 1024 x 1088
    Number of image formats: 5
      id: 0x32595559 (YUY2)
        guid: 59555932-0000-0010-8000-00aa00389b71
        bits per pixel: 16
        number of planes: 1
        type: YUV (packed)
      id: 0x32315659 (YV12)
        guid: 59563132-0000-0010-8000-00aa00389b71
        bits per pixel: 12
        number of planes: 3
        type: YUV (planar)
      id: 0x30323449 (I420)
        guid: 49343230-0000-0010-8000-00aa00389b71
        bits per pixel: 12
        number of planes: 3
        type: YUV (planar)
      id: 0x59565955 (UYVY)
        guid: 55595659-0000-0010-8000-00aa00389b71
        bits per pixel: 16
        number of planes: 1
        type: YUV (packed)
      id: 0x434d5658 (XVMC)
        guid: 58564d43-0000-0010-8000-00aa00389b71
        bits per pixel: 12
        number of planes: 3
        type: YUV (planar)
[goetz@goetz ~]$ xrandr 
Screen 0: minimum 320 x 200, current 1024 x 768, maximum 2048 x 2048
VGA disconnected (normal left inverted right x axis y axis)
LVDS connected 1024x768+0+0 (normal left inverted right x axis y axis) 0mm x 0mm
   1024x768       60.0*+   60.0  
   800x600        60.3  
   640x480        59.9  


This was working fine in Mandriva 2009 with x11-driver-video-intel-2.4.2.

Colin's intel driver snapshot from git bears the 20090606 timestamp.

Thanks very much for the test Götz, it's very much appreciated.

This testing platform was from intel git master a couple days ago as Götz mentioned, libdrm 2.4.11, and mesa 7.3.

I suspect the problem is at the driver level rather than mesa related but I'm a good way short of being an expert in this area :)

Gordon, Haihao: Do you guys have a 82830 to test this on? It's affecting a few of our users now :(

(In reply to comment #34)
> Gordon, Haihao: Do you guys have a 82830 to test this on? It's affecting a few
> of our users now :(

No, we don't have 82830.

Haien, is xvideo working on your 855GM?

If we can reproduce it, I'll suggest to make this as high priority bug. 

(In reply to comment #35)
> (In reply to comment #34)
> > Gordon, Haihao: Do you guys have a 82830 to test this on? It's affecting a few
> > of our users now :(
> 
> No, we don't have 82830.
> 
> Haien, is xvideo working on your 855GM?
> 
> If we can reproduce it, I'll suggest to make this as high priority bug. 
> 
Is the 845 and 855 that much different??

Mine is the 845GM that started the Mandriva Bug. Lots of info there regarding it.

Rich

(In reply to comment #35)
> (In reply to comment #34)
> > Gordon, Haihao: Do you guys have a 82830 to test this on? It's affecting a few
> > of our users now :(
> 
> No, we don't have 82830.
> 
> Haien, is xvideo working on your 855GM?
> 
> If we can reproduce it, I'll suggest to make this as high priority bug. 
> 

hi, sorry for response late.
xvideo doesn't work on our 855gm now. zhenyu tell me overlay video isn't ready in KMS. I will try UMS soon.
following is the output of xvinfo:
[root@x-855gm ~]# xvinfo
X-Video Extension version 2.2
screen #0
 no adaptors present

(In reply to comment #37)
> (In reply to comment #35)
> > (In reply to comment #34)
> > > Gordon, Haihao: Do you guys have a 82830 to test this on? It's affecting a few
> > > of our users now :(
> > 
> > No, we don't have 82830.
> > 
> > Haien, is xvideo working on your 855GM?
> > 
> > If we can reproduce it, I'll suggest to make this as high priority bug. 
> > 
> 
> hi, sorry for response late.
> xvideo doesn't work on our 855gm now. zhenyu tell me overlay video isn't ready
> in KMS. I will try UMS soon.
> following is the output of xvinfo:
> [root@x-855gm ~]# xvinfo
> X-Video Extension version 2.2
> screen #0
>  no adaptors present
> 

overlay video works well with UMS on our 855gm.

(In reply to comment #37)
> hi, sorry for response late.
> xvideo doesn't work on our 855gm now. zhenyu tell me overlay video isn't ready
> in KMS. I will try UMS soon.

FWIW, the people who are having problems in Mandriva are not using KMS, just a normal boot and the the default video accelmethod, which for that case is EXA. But as Götz tried the git master, it must have been using UXA in that case (albeit without GEM).

> following is the output of xvinfo:
> [root@x-855gm ~]# xvinfo
> X-Video Extension version 2.2
> screen #0
>  no adaptors present

As can be seen from previous debug logs the xvinfo program always reported the overlay as being present, it just couldn't be used without a badalloc error.

Hope this helps. 

(In reply to comment #39)
> (In reply to comment #37)
> > hi, sorry for response late.
> > xvideo doesn't work on our 855gm now. zhenyu tell me overlay video isn't ready
> > in KMS. I will try UMS soon.
> 
> FWIW, the people who are having problems in Mandriva are not using KMS, just a
> normal boot and the the default video accelmethod, which for that case is EXA.
> But as Götz tried the git master, it must have been using UXA in that case
> (albeit without GEM).
> 
> > following is the output of xvinfo:
> > [root@x-855gm ~]# xvinfo
> > X-Video Extension version 2.2
> > screen #0
> >  no adaptors present
> 
> As can be seen from previous debug logs the xvinfo program always reported the
> overlay as being present, it just couldn't be used without a badalloc error.
> 
> Hope this helps. 
> 
Overlay works here (see commit #38, 855 and 845 have the same code path in intel driver with little difference). When you tested the git master, did you also upgrade your DRM module? 

As mentioned by Colin in comment #34, he gave me  the git master of the intel driver, libdrm 2.4.11 and mesa 7.3. That's what I had tested.

I finally could make Xv work by simply upgrading the kernel to an unofficial Mandriva build of 2.6.30.

(In reply to comment #42)
> I finally could make Xv work by simply upgrading the kernel to an unofficial
> Mandriva build of 2.6.30.
> 

Is this still working for you and if so, how do I get the kernel? Can I go thru add/remove and get it? Or do I have to roll one?

I think considering that things are confirmed working with an updated kernel that this bug can be closed. If people disagree, then it can be reopened (again!).

Rich, I'll comment on https://qa.mandriva.com/show_bug.cgi?id=50857 about the kernel as it's not really relevant to other folks here and we should keep the FDO bug reports as clean and specific to the problem as possible :)

I am using FreeBSD.  The fact that, for Linux users, a kernel update helps, does not make any difference here where the problem persists.

If you believe this is due to changes external to the driver code, could you document them here, so that I can see if similar changes can be made to the FreeBSD kernel.

That said, don't you think it is strange that this driver stopped working for both Linux and FreeBSD users when the driver was updated?

(In reply to comment #44)
> I think considering that things are confirmed working with an updated kernel
> that this bug can be closed. If people disagree, then it can be reopened
> (again!).
> 
> Rich, I'll comment on https://qa.mandriva.com/show_bug.cgi?id=50857 about the
> kernel as it's not really relevant to other folks here and we should keep the
> FDO bug reports as clean and specific to the problem as possible :)
> 

OK Colin

It is unhelpful to keep these bugs open on known buggy versions of the drivers simply because the infrastructure to test fixes is lacking. Hopefully the situation will improve in the near future and we will be able to start work again on an updated userspace.