chromium-dev — For developers of the Chromium framework or SPU's

[Chromium-dev] Adding dontclipToScreen to Chromium's dmx support...

[James S. <arr...@gm...>]

I would propose the following patch for Chromium's DMX module. I use
the following change for my dmxdireect SPU. Basically, if before
calling crDMXGetBackendWindowInfo I set the dontclipToScreen data
member to true in the CRDMXBackendWindowInfo (sense Chromium appears
to zero CRDMXBackendWindowInfo during allocation, it is set to false
by default).

Then, when crDMXGetBackendWindowInfo is called, it doesn't clip the
child window to the screen.

I chose this method because it appears to cause the least disruption
to existing code (Chromium was written in C, and I don't think C likes
function overloading etc.).

Index: include/cr_dmx.h
===================================================================
RCS file: /cvsroot/chromium/cr/include/cr_dmx.h,v
retrieving revision 1.1
diff -r1.1 cr_dmx.h
15a16
>   int dontclipToScreen;

Index: dmx/dmx.c
===================================================================
RCS file: /cvsroot/chromium/cr/dmx/dmx.c,v
retrieving revision 1.2
diff -r1.2 dmx.c
158,165c158,176
<               subwinX = backend->visrect.x1;
<               subwinY = backend->visrect.y1;
<               subwinW = backend->visrect.x2 - backend->visrect.x1;
<               subwinH = backend->visrect.y2 - backend->visrect.y1;
<               if (subwinW <= 0)
<                       subwinW = 1;
<               if (subwinH <= 0)
<                       subwinH = 1;
---
>               if(!backend->dontclipToScreen)
>                 {
>                   subwinX = backend->visrect.x1;
>                   subwinY = backend->visrect.y1;
>                   subwinW = backend->visrect.x2 - backend->visrect.x1;
>                   subwinH = backend->visrect.y2 - backend->visrect.y1;
>                   if (subwinW <= 0)
>                     subwinW = 1;
>                   if (subwinH <= 0)
>                     subwinH = 1;
>                 }
>               else
>                 {
>                   /*Used by dmxdirect */
>                   subwinX = 0;
>                   subwinY = 0;
>                   subwinW = dmxWinInfo[i].pos.width;
>                   subwinH = dmxWinInfo[i].pos.height;
>                 }

Thank you for your time,
James Steven Supancic III

Re: [Chromium-dev] Make chromium-checkin subscriber only?

[Daniel E. S. <dan...@gm...>]

Thank you. I really appreciate it.

-Dan

On 11/12/06, Brian Paul <bri...@tu...> wrote:
>
> Niraj Tolia wrote:
> > Daniel raised this issue on the chromium-checkin list but I am not
> > sure who is subscribed to it. As he had mentioned, there is a lot of
> > spam on the mailing list. Can we please make it subscriber only?
>
> Done.
>
> -Brian
>
>

Re: [Chromium-dev] Make chromium-checkin subscriber only?

[Brian P. <bri...@tu...>]

Niraj Tolia wrote:
> Daniel raised this issue on the chromium-checkin list but I am not
> sure who is subscribed to it. As he had mentioned, there is a lot of
> spam on the mailing list. Can we please make it subscriber only?

Done.

-Brian

[Chromium-dev] Make chromium-checkin subscriber only?

[Niraj T. <nt...@gm...>]

Daniel raised this issue on the chromium-checkin list but I am not
sure who is subscribed to it. As he had mentioned, there is a lot of
spam on the mailing list. Can we please make it subscriber only?

Thanks,
Niraj

-- 
http://www.cs.cmu.edu/~ntolia

Re: [Chromium-dev] Chromium hangs on Windows XP

[<dr...@bf...>]

>> I use Chromium on Windows XP to drive 6 servers from one client over a 
>> 1Gbps TCP/IP network. Every now and then Chromium locks up. The lock up 
>> occurres in __tcpip_read_exact()'s call to recv(). Even though select() 
>> did report the socket to be ready for reading the first call to recv() 
>> blocks.  If I kill the server of this connection recv() returns with the 
>> correct error code. See the stack trace below for more details.
>> Does anyone else experience similar problems? I suspect this might be a 
>> bug in some network driver (Intel) but I'm not sure.
> 
> If there's any bugs in Chromiums packer/unpacker code, a common 
> symptom is for the network layer to get stuck in recv() - waiting for 
> bytes that aren't coming.

I dont think its a bug in Chromium since the block occurs on the first 
call to recv() after select() reported the socket to be readable. That 
is crTCPIPRecv() calls select(), finds the socket to be readable and 
calls crTCPIPReceiveMessage() which blocks right away on the statement

if ( __tcpip_read_exact( sock, &len, sizeof(len)) <= 0 ).

So this really shouldn't block or should it?

> Is the lock-up only happening with certain apps?  Do those apps work 
> ok on other Chromium systems?

It does happen with at least a couple of apps I tried, one of which is 
atlantis. I didn't test on other systems though.

Michael

Re: [Chromium-dev] Chromium hangs on Windows XP

[Brian P. <bri...@tu...>]

Michael D=FCrig wrote:
> Hi
>=20
> I use Chromium on Windows XP to drive 6 servers from one client over a=20
> 1Gbps TCP/IP network. Every now and then Chromium locks up. The lock up=
=20
> occurres in __tcpip_read_exact()'s call to recv(). Even though select()=
=20
> did report the socket to be ready for reading the first call to recv()=20
> blocks.  If I kill the server of this connection recv() returns with th=
e=20
> correct error code. See the stack trace below for more details.
> Does anyone else experience similar problems? I suspect this might be a=
=20
> bug in some network driver (Intel) but I'm not sure.

If there's any bugs in Chromiums packer/unpacker code, a common=20
symptom is for the network layer to get stuck in recv() - waiting for=20
bytes that aren't coming.

Is the lock-up only happening with certain apps?  Do those apps work=20
ok on other Chromium systems?

-Brian

Re: [Chromium-dev] proper OpenGL version support with local configuration?

[Brian P. <bri...@tu...>]

James Supancic wrote:
> proper OpenGL version with local configuration?
> 
> I wrote a new SPU (I named it dmxdirect), it basically allows me to
> use Chromium in a local configuration on a single DMX backend display
> (rendering is restricted to one of my four monitors). The
> configuration is as follows
> AppNode:
> 	dmxdirectspu
> 	renderspu
> 
> The dmxdirectspu basicly does what is needed to get the renderspu to
> render to backend window (like the tilesortSPU but for a single
> machine).
> 
> This is much faster than using the tilesort spu, and as I only need 3d
> support on one of the four monitors it is better for my configuration.
> 
> Overall things run much better, but I still have some problems, a
> number of applications don't render correctly. I suspect this is
> because Chromium is returning the OpenGL capabilities of the card, not
> the intersection of it's capabilities and the cards. For example,
> glxinfo says that my OpenGL version string is 2.0.2.

First, you can test your hypothesis by changing renderspuGetString() 
in spu/render/renderspu.c so that it returns "1.5" or thereabouts.

If that doesn't fix things, the problem is elsewhere.


> My question is, is there an easy way to compile Chromium with support
> for OpenGL 2 in a local configuration? I shouldn't have to do that
> much to get OpenGL 2 working in this configuration as all Chromium
> must do is forward the calls as they are, there is no need for any
> packing/tilesorting etc.

You'd basically have to add the OpenGL 2.0 functions to the API 
dispatch layer.  The relevant directories are opengl_stub/, 
glapi_parser/ and spu_loader/.  Not easy, but not too hard either.

-Brian

[Chromium-dev] Chromium hangs on Windows XP

[<dr...@bf...>]

Hi

I use Chromium on Windows XP to drive 6 servers from one client over a 
1Gbps TCP/IP network. Every now and then Chromium locks up. The lock up 
occurres in __tcpip_read_exact()'s call to recv(). Even though select() 
did report the socket to be ready for reading the first call to recv() 
blocks.  If I kill the server of this connection recv() returns with the 
correct error code. See the stack trace below for more details.
Does anyone else experience similar problems? I suspect this might be a 
bug in some network driver (Intel) but I'm not sure.

Michael


NTDLL.DLL!_KiFastSystemCallRet@0() 	
NTDLL.DLL!_ZwWaitForSingleObject@12()  + 0xc bytes	
MSWSOCK.DLL!_SockWaitForSingleObject@16()  + 0x3c8 bytes	
MSWSOCK.DLL!_WSPRecv@36()  + 0x1487 bytes	
WS2_32.DLL!_recv@16()  + 0x6f bytes	
crutil.dll!__tcpip_read_exact(unsigned int sock=1892, void * 
buf=0x0012fb98, unsigned int len=4)  Line 212 + 0x14 bytes	C
crutil.dll!crTCPIPReceiveMessage(CRConnection * conn=0x00add6c0)  Line 
868 + 0xf bytes	C
crutil.dll!crTCPIPRecv()  Line 1089 + 0xc bytes	C
crutil.dll!crNetRecv()  Line 1183 + 0x5 bytes	C
tilesortspu.dll!tilesortspu_SwapBuffers(int window=1, int flags=0)  Line 
103 + 0x5 bytes	C
opengl32.dll!stubSwapBuffers(const window_info_t * window=0x00edfe98, 
int flags=0)  Line 987	C
opengl32.dll!wglSwapBuffers_prox(HDC__ * hdc=0xb5010b1e)  Line 275 + 
0xb bytes	C
gdi32.dll!_SwapBuffers@4()  + 0x25 bytes	
glut32.dll!10009e55() 	
[Frames below may be incorrect and/or missing, no symbols loaded for 
glut32.dll]	
atlantis.exe!Display()  Line 349	C
glut32.dll!100050c2() 	
atlantis.exe!WhalePilot(_fishRec * fish=0x00411c60)  Line 61 + 0x26 bytes	C
atlantis.exe!Animate()  Line 169 + 0xa bytes	C
0012ff58()	
glut32.dll!100048d6() 	
atlantis.exe!main(int argc=3, char * * argv=0x00ad5d68)  Line 426	C
atlantis.exe!mainCRTStartup()  Line 398 + 0x11 bytes	C
kernel32.dll!_BaseProcessStart@4()  + 0x23 bytes	

[Chromium-dev] proper OpenGL version support with local configuration?

[James S. <arr...@gm...>]

proper OpenGL version with local configuration?

I wrote a new SPU (I named it dmxdirect), it basically allows me to
use Chromium in a local configuration on a single DMX backend display
(rendering is restricted to one of my four monitors). The
configuration is as follows
AppNode:
	dmxdirectspu
	renderspu

The dmxdirectspu basicly does what is needed to get the renderspu to
render to backend window (like the tilesortSPU but for a single
machine).

This is much faster than using the tilesort spu, and as I only need 3d
support on one of the four monitors it is better for my configuration.

Overall things run much better, but I still have some problems, a
number of applications don't render correctly. I suspect this is
because Chromium is returning the OpenGL capabilities of the card, not
the intersection of it's capabilities and the cards. For example,
glxinfo says that my OpenGL version string is 2.0.2.

My question is, is there an easy way to compile Chromium with support
for OpenGL 2 in a local configuration? I shouldn't have to do that
much to get OpenGL 2 working in this configuration as all Chromium
must do is forward the calls as they are, there is no need for any
packing/tilesorting etc.

Thank you for your time,
James Steven Supancic III

Re: [Chromium-dev] Patch for compiling with VC8

[Brian P. <bri...@tu...>]

Michael D=FCrig wrote:
>=20
> The patch attached adds support for Visual Studio 2005 (i.e. VC8).
>=20
> - When VC8 is detected in WIN_NT.MK a different set of flags is set.
> There are a lot of deprecation warnings. I choose to ignore them (/wd49=
96).
>=20
> - By default the linker produces an external manifest file. This is
> problematic for appfaker since an application fails to start when the
> manifest is not in the applications directory. I therefore added a
> further link step which embeds the manifest into the application's bina=
ry.
>=20
> - Some .def files used the DESCRIPTION keyword which is deprecated now.
> Since the description was always an empty string I removed it entirely.
>=20
> - I think it is wrong to declare cr_packer_globals DLLDATA
> (__declspec(dllimport)) in cr_pack.h/pack_init.c. After all the variabl=
e
> is exported.

I checked in your patch.  Not sure about the DLLDATA thing.  Maybe=20
another Windows user can comment.

-Brian

[Chromium-dev] Patch for compiling with VC8

[<dr...@bf...>]

The patch attached adds support for Visual Studio 2005 (i.e. VC8).

- When VC8 is detected in WIN_NT.MK a different set of flags is set.
There are a lot of deprecation warnings. I choose to ignore them (/wd4996).

- By default the linker produces an external manifest file. This is
problematic for appfaker since an application fails to start when the
manifest is not in the applications directory. I therefore added a
further link step which embeds the manifest into the application's binary.

- Some .def files used the DESCRIPTION keyword which is deprecated now.
Since the description was always an empty string I removed it entirely.

- I think it is wrong to declare cr_packer_globals DLLDATA
(__declspec(dllimport)) in cr_pack.h/pack_init.c. After all the variable
is exported.

Michael

Re: [Chromium-dev] warning in dmx.c

[Brian P. <bri...@tu...>]

James Supancic wrote:
>> Alternately, does changing "unsigned int" to "int" on line 97 silence
>> the warnings?
> 
> 
> Yea, that gets rid of them, but now I get
> dmx.c:103: warning: comparison between signed and unsigned
> for
> CRASSERT(numScreens == numBackendWindows);

OK, fixed in CVS.


>> Which compiler is this?
> 
> gcc 4.1.1, apparently it doesn't like implicit conversions from signed
> to unsigned values...

g++ has complained about those in earlier versions.  I guess gcc is 
catching up.

-Brian

Re: [Chromium-dev] warning in dmx.c

[James S. <arr...@gm...>]

> Alternately, does changing "unsigned int" to "int" on line 97 silence
> the warnings?

Yea, that gets rid of them, but now I get
dmx.c:103: warning: comparison between signed and unsigned
for
CRASSERT(numScreens == numBackendWindows);

> Which compiler is this?
gcc 4.1.1, apparently it doesn't like implicit conversions from signed
 to unsigned values...

Thank you for your time,
James Steven Supancic III

Re: [Chromium-dev] warning in dmx.c

[Brian P. <bri...@tu...>]

James Supancic wrote:
> warning in dmx.c
> 
> I am getting a warning in /opt/cr/dmx/dmx.c, due to a type differnce.
> I have attached a patch which fixes the problem.
> 
> Thank you for your time,
> James Steven Supancic III
> 
> 
> ------------------------------------------------------------------------
> 
> ? dmx.c.diff
> Index: dmx.c
> ===================================================================
> RCS file: /cvsroot/chromium/cr/dmx/dmx.c,v
> retrieving revision 1.1
> diff -r1.1 dmx.c
> 102c102
> < 	DMXGetScreenCount(dpy, &numScreens);
> ---
> 
>>	DMXGetScreenCount(dpy, (int*)&numScreens);
> 
> 127c127
> < 	if (!DMXGetWindowAttributes(dpy, xwin, &count, numScreens, dmxWinInfo))
> ---
> 
>>	if (!DMXGetWindowAttributes(dpy, xwin, (int*)&count, numScreens, dmxWinInfo))

Alternately, does changing "unsigned int" to "int" on line 97 silence 
the warnings?

Which compiler is this?

-Brian

[Chromium-dev] warning in dmx.c

[James S. <arr...@gm...>]

warning in dmx.c

I am getting a warning in /opt/cr/dmx/dmx.c, due to a type differnce.
I have attached a patch which fixes the problem.

Thank you for your time,
James Steven Supancic III

[Chromium-dev] Fwd: Custom crServer sync question

[Daniel E. S. <dan...@gm...>]

[snip]

> >
> >     I have a custom crServer that is embedded into a frame loop. Every
> >     frame the server checks for client data which is drawn along with my
> >     custom drawing updates...  This results in choppy frames because the
> >     frames aren't synchronized that are coming over the wire on a
> >     glFlush/SwapBuffers/etc(spu that no-ops them).  I am trying to find
> >     a way to find out when a client(in our case there will always just
> >     be one) is done with the current frame(buffer swapping for my app is
>
> >     done for me so its out of my control).  I see where I can check to
> >     see if the server is in between Begin/End calls..but not between
> >     frames.
>
> OpenGL doesn't have any explicit begin/end-frame commands.  The
> closest approximation is glClear and SwapBuffers/glFlush.


This I know.  I am not using Chromium like most though. I am trying to
insert crServer into a frame loop that I don't have much control over(VR
Juggler).

If your app calls [glX/wgl]SwapBuffers() you can intercept that in
> Chromium and do end-of-frame processing.


This is what I was asking. I want to intercept these in the server stream,
not in an SPU. Is anything like this supported/can I make it happen?

-Dan

Re: [Chromium-dev] Custom crServer sync question

[Brian P. <bri...@tu...>]

Daniel E. Shipton wrote:
> Any advice from this list? Since I am digging deep into the chromium 
> code this list may be more appropriate.
> 
> On 9/11/06, *Daniel E. Shipton* < dan...@gm... 
> <mailto:dan...@gm...>> wrote:
> 
>     I have a custom crServer that is embedded into a frame loop. Every
>     frame the server checks for client data which is drawn along with my
>     custom drawing updates...  This results in choppy frames because the
>     frames aren't synchronized that are coming over the wire on a
>     glFlush/SwapBuffers/etc(spu that no-ops them).  I am trying to find
>     a way to find out when a client(in our case there will always just
>     be one) is done with the current frame(buffer swapping for my app is
>     done for me so its out of my control).  I see where I can check to
>     see if the server is in between Begin/End calls..but not between
>     frames.

OpenGL doesn't have any explicit begin/end-frame commands.  The 
closest approximation is glClear and SwapBuffers/glFlush.

If your app calls [glX/wgl]SwapBuffers() you can intercept that in 
Chromium and do end-of-frame processing.

-Brian

Re: [Chromium-dev] Solution for bug in tilesortspu_DrawRangeElements.

[Brian P. <bri...@tu...>]

James Supancic wrote:
> Solution for bug in tilesortspu_DrawRangeElements.
> 
> All the for loops in this function should use i=0 not i=start. Start
> and End have nothing to do with which parts of the index buffer should
> be sent, they refer to the maximum and minimum array index contained
> in the index buffer. If you like you could add an if statement in the
> for loops to filter out array indexes that are not bound by start and
> end.
> 
> for (i = start ; i < count ; i++)
> {
> //pack array element
> }
> 
> should be replaced with:
> 
> for (i = 0 ; i < count ; i++)
> {
> //pack array element
> }

Thanks.  Fixed in CVS.

-Brian

[Chromium-dev] Solution for bug in tilesortspu_DrawRangeElements.

[James S. <arr...@gm...>]

Solution for bug in tilesortspu_DrawRangeElements.

All the for loops in this function should use i=0 not i=start. Start
and End have nothing to do with which parts of the index buffer should
be sent, they refer to the maximum and minimum array index contained
in the index buffer. If you like you could add an if statement in the
for loops to filter out array indexes that are not bound by start and
end.

for (i = start ; i < count ; i++)
{
//pack array element
}

should be replaced with:

for (i = 0 ; i < count ; i++)
{
//pack array element
}


Thank you for your time,
James Steven Supancic III

Re: [Chromium-dev] Custom crServer sync question

[Daniel E. S. <dan...@gm...>]

Any advice from this list? Since I am digging deep into the chromium code
this list may be more appropriate.

On 9/11/06, Daniel E. Shipton <dan...@gm...> wrote:
>
> I have a custom crServer that is embedded into a frame loop. Every frame
> the server checks for client data which is drawn along with my custom
> drawing updates...  This results in choppy frames because the frames aren't
> synchronized that are coming over the wire on a glFlush/SwapBuffers/etc(spu
> that no-ops them).  I am trying to find a way to find out when a client(in
> our case there will always just be one) is done with the current
> frame(buffer swapping for my app is done for me so its out of my control).
> I see where I can check to see if the server is in between Begin/End
> calls..but not between frames.
>
> Within my draw function I have:
>
> // Bunch of custom drawing above...
>    glPushMatrix();
>          bool ran_queue = false;
>          while(!ran_queue)
>          {
>             if(cr_server.run_queue != NULL)
>             {
>            crServerServiceClients();
>                ran_queue = true;
>             }
>          }
>    glPopMatrix();
>
>
> Ideas?
>
> -Dan
>

[Chromium-dev] socket clean-up, crserver exit

[Brian P. <bri...@tu...>]

A problem that Chromium has had for a while is it doesn't always close 
down cleanly.  When app or crserver process exits, the other processes 
may linger until they're manually killed.

I've checked in some changes to address this.

The network layer had some provision for connection-close callback 
functions but it wasn't fully utilized.  I've updated that code.  The 
connection-close callback that's registered with crNetInit() is now 
called whenever a connection is closed (it was hit or miss before). 
I've updated the Pack SPU to use this properly and will update some 
other SPUs soon.

The other new thing is a crserver config option called 
'exit_if_no_clients'.  It's enabled by default.  Basically, when the 
crserver detects that there are no longer any client connections, the 
crserver exits.  If you'd prefer the crserver to stay alive, waiting 
for future client connections, set exit_if_no_clients=0.

Hopefully I haven't broke anything.  Let me know if I did.

-Brian

Re: [Chromium-dev] QT 4.0 (Windows) compatibility patch

[Brian P. <bri...@tu...>]

Michael D=FCrig wrote:
>=20
> The attached patch enhances compatibility with QT 4.0 on Windows. QT=20
> implements an abstraction of an OpenGL context in its QGLContext class.
> To create a context QGLContext always calls wglCreateLayerContext() eve=
n=20
> though a call to wglCreateContext() would do. To account for this I=20
> changed the implementation of wglCreateLayirContext_prox() in=20
> opengl_stub/wgl.c such that it forwards the call to=20
> wglCreateContext_prox() if layerPlane =3D=3D 0.
>=20

Thanks.  I checked in your patch.

-Brian

[Chromium-dev] QT 4.0 (Windows) compatibility patch

[<dr...@bf...>]

The attached patch enhances compatibility with QT 4.0 on Windows. QT 
implements an abstraction of an OpenGL context in its QGLContext class.
To create a context QGLContext always calls wglCreateLayerContext() even 
though a call to wglCreateContext() would do. To account for this I 
changed the implementation of wglCreateLayirContext_prox() in 
opengl_stub/wgl.c such that it forwards the call to 
wglCreateContext_prox() if layerPlane == 0.

Michael

Re: [Chromium-dev] Query: SPU development and Bucket Information?

[Brian P. <bri...@tu...>]

James Supancic wrote:
> I have a few questions.
> 
> What do the fields in the TileSortBucketInfo structure mean? I
> understand the meaning of the hits data member, but what are the
> pixelBounds, objectMin and objectMax data members for?

Those are the bounding boxes for the object we're bucketing in object 
coords and in window coords.


> My second question is, can I SPU change it's "super" at run time? For
> example, lets create a SPU with a NOP SPU as it's super, but later I
> want to change it to a passthough SPU. Can I do this?
> What would be the technique?

I don't think that's possible.

However, the 'apichange' SPU shows how to change GL functions on the 
fly.  Maybe that'll help.

-Brian

[Chromium-dev] Query: SPU development and Bucket Information?

[James S. <arr...@gm...>]

I have a few questions.

What do the fields in the TileSortBucketInfo structure mean? I
understand the meaning of the hits data member, but what are the
pixelBounds, objectMin and objectMax data members for?

My second question is, can I SPU change it's "super" at run time? For
example, lets create a SPU with a NOP SPU as it's super, but later I
want to change it to a passthough SPU. Can I do this?
What would be the technique?

Thank you for your time,
James Steven Supancic III