xorg X11 using i81x drivers is working now

[Jim Shepherd]

Just upgraded to the 6.7.99.903-1 release of xorg-X11 that became
available this morning and X is finally working on my PC with i810.
Previously, X would not start when using the i810 driver and I received
the same errors as many others on this list with similar Intel chipsets.

-Jim

[Jim Cornette]

Jim Shepherd wrote:
> Just upgraded to the 6.7.99.903-1 release of xorg-X11 that became
> available this morning and X is finally working on my PC with i810.
> Previously, X would not start when using the i810 driver and I received
> the same errors as many others on this list with similar Intel chipsets.
>
> -Jim

This makes things much easier! Thanks for the info. X now works with no 
needed tricks w/ the 815 card type also.

Jim

xorg X11 using i81x drivers is very slow now! It does come up though!

[Jim Cornette]

Jim Cornette wrote:
> Jim Shepherd wrote:
> > Just upgraded to the 6.7.99.903-1 release of xorg-X11 that became
> > available this morning and X is finally working on my PC with i810.
> > Previously, X would not start when using the i810 driver and I received
> > the same errors as many others on this list with similar Intel chipsets.
> >
> > -Jim

Never mind! Things started to work very slowly w/ X (virtually locked up 
activity, after a very short time.)

I copied the i810 driver from the earlier installation to see if things 
seemed to work better.

If someone posted a bug report yet. What number bug report it is?

Attached is my first launched xorg session report. This with everything 
at rpm default.

Jim

xorg X11 using i81x drivers is very slow now! It does come up though!

[Jim Shepherd]

On Sun, 2004-08-29 at 16:25 -0400, Jim Cornette wrote:
> Jim Cornette wrote:
> Never mind! Things started to work very slowly w/ X (virtually locked up 
> activity, after a very short time.)

I've been running the 6.7.99.903-1 release for a day now and there is no
noticeable slow down and definitely not any lock-ups.  This is using the
i810 driver on an i810 chipset.

-Jim

xorg X11 using i81x drivers is very slow now! It does come up though!

[Jim Cornette]

Jim Shepherd wrote:
> On Sun, 2004-08-29 at 16:25 -0400, Jim Cornette wrote:
> > Jim Cornette wrote:
> > Never mind! Things started to work very slowly w/ X (virtually locked up 
> > activity, after a very short time.)
>
> I've been running the 6.7.99.903-1 release for a day now and there is no
> noticeable slow down and definitely not any lock-ups.  This is using the
> i810 driver on an i810 chipset.
>
> -Jim

Running with the i686 compiled binaries that I installed after the 
problem surfaced on my 815 chipset, X is working fine.

If I load the binary i810 driver from the latest xorg-x11 binary rpm, 
even with the i686 compiled rpms installed, it slows down *tremendously*

If it wouldn't be of too much trouble for you to do. Please file your 
working Xorg.0.log for comparison on the below bug report. This might be 
useful for debugging the error.

Thanks,
Jim

https://freedesktop.org/bugzilla/show_bug.cgi?id=1232

xorg X11 using i81x drivers is very slow now! It does come up though!

[Alan Cox]

On Mon, Aug 30, 2004 at 07:46:59AM -0400, Jim Cornette wrote:
> If I load the binary i810 driver from the latest xorg-x11 binary rpm, 
> even with the i686 compiled rpms installed, it slows down *tremendously*

Do you see vast numbers of interrupts of some form ?

xorg X11 using i81x drivers is very slow now! It does come up though!

[Jim Cornette]

Alan Cox wrote:
> On Mon, Aug 30, 2004 at 07:46:59AM -0400, Jim Cornette wrote:
> > If I load the binary i810 driver from the latest xorg-x11 binary rpm, 
> > even with the i686 compiled rpms installed, it slows down *tremendously*
>
> Do you see vast numbers of interrupts of some form ?

It seems to be that the major fallout is with the screen not refreshing 
properly. If you go to the hat (or foot, with my theme), you only see 
the menu items if you hover the mouse over w/ the mouse button depressed.

There is a new patch submitted which is recorded from the first bug 
submitted by Mike. The patch is below. Why the i686 compilation worked 
and the binary rpm version on cvs did not is beyond me.

My interrupts seem to be pretty much ordinary. I hope the new patch 
works better. mtrr usually adds the last line to the file. With the 
rawhide version binary driver, the last line is never added (reg02)

cat /proc/mtrr
reg00: base=0x00000000 (   0MB), size= 512MB: write-back, count=1
reg01: base=0x1ff00000 ( 511MB), size=   1MB: uncachable, count=1
reg02: base=0xf8000000 (3968MB), size=  64MB: write-combining, count=1

I have a screenshot of the resulting screen attached to the bug report.

https://freedesktop.org/bugzilla/attachment.cgi?id=782&action=view

The patch is in the below bug report.

https://freedesktop.org/bugzilla/show_bug.cgi?id=1084

Jim

Index: i810_driver.c
===================================================================
RCS file: 
/cvs/xorg/xc/programs/Xserver/hw/xfree86/drivers/i810/i810_driver.c,v
retrieving revision 1.8
diff -u -r1.8 i810_driver.c
--- i810_driver.c	30 Jul 2004 21:36:40 -0000	1.8
+++ i810_driver.c	23 Aug 2004 04:56:12 -0000
@@ -274,6 +274,7 @@
     "drmAgpRelease",
     "drmAgpUnbind",
     "drmAuthMagic",
+   "drmCommandNone",
     "drmCommandWrite",
     "drmCreateContext",
     "drmCtlInstHandler",
@@ -1123,6 +1124,14 @@
     xf86SetOperatingState(resVgaIo, pI810->pEnt->index, ResUnusedOpr);
     xf86SetOperatingState(resVgaMem, pI810->pEnt->index, ResDisableOpr);

+   pI810->LpRing = xalloc(sizeof(I810RingBuffer));
+   if (!pI810->LpRing) {
+     xf86DrvMsg(pScrn->scrnIndex, X_ERROR,
+		"Could not allocate lpring data structure.\n");
+     I810FreeRec(pScrn);
+     return FALSE;
+   }
+
     return TRUE;
  }

@@ -2495,6 +2504,9 @@
      */
     xf86GARTCloseScreen(scrnIndex);

+   xfree(pI810->LpRing);
+   pI810->LpRing = NULL;
+
     pScrn->vtSema = FALSE;
     pScreen->CloseScreen = pI810->CloseScreen;
     return (*pScreen->CloseScreen) (scrnIndex, pScreen);