Re: xfree86 4.2.0-0pre1v1 available at the X Strike Force
On Sun, Jun 16, 2002 at 08:15:43PM +0100, John Gay wrote:
> I'm fetching this as I type, but since the connection seems slow at teh
> moment, probably everyone and their grannie fetching the same thing, and I
> have to pay for Internet access here in Ireland, It might be a while before I
> complete the fetch.
>
> My system:
>
> Dual PII 350 with GVX1 32M AGP card.
> At the moment I'm running xfree86 4.2.0 compiled from source and installed
> over my old xfree86 4.1.0 .deb's. Not the brightest thing I've ever done, but
> it worked for me.
>
> I am running mostly testing updated 15/06/02 with the X packages on hold due
> to my GVX1 card needing 4.2.0
>
> I also have the Accelerated X server from xig for 3-D acceleration and stereo
> features.
>
> I use gdm since it seems to self-configure for the different destops/window
> mangers than xdm or kdm.
>
> So far I only had to update libfreetype6 to unstable to apt-get install
> x-window-system. If I have any problems once the update is finished, I've
> still got the compiled xfree86 4.2.0 os I should be able to just make install
> it again to get me back where I am now.
>
> One problem I am currently having with my existing set-up is the display
> occasionally disappears? <ctrl><alt><F1> to a terminal and then <alt><F7>
> back usually fixes it, though occasionally X dies and I'm back at gdm. I have
> not been able to determine what causes this, it seems completely random? I am
> hoping that your packages will fix this as well.
Well, this is a known bug in the glint driver with the pm3 chips, we did
search a long time for it, and lately Alan Hourihane seems to have fixed
this bug, i have not yet tested it though.
It is already in 4.2.99.1, fix 128
128. Fix RGB/BGR inversion problem and screen blanking in the glint
driver (Alan Hourihane).
And i attach the extracted diff here, so that Branden can include it in
the forthcoming 4.2.0 packages (well, i just hope it don't depend on
other post 4.2.0 stuff, i did not test it, but basically it is just a
call to xf86BlockSIGIO/xf86UnblockSIGIO, so it should be easy to check
if those exist already in 4.2.0.
Anyway, in the meantime, you could disable the hardware cursor in your
XF86Config file or try to start X with the -nosilk option as a
workaround.
Friendly,
Sven Luther
===================================================================
RCS file: /home/anoncvs/cvs/xc/programs/Xserver/hw/xfree86/drivers/glint/glint_driver.c,v
retrieving revision 1.146
retrieving revision 1.147
diff -u -p -r1.146 -r1.147
--- xc/programs/Xserver/hw/xfree86/drivers/glint/glint_driver.c 2002/02/27 18:41:04 1.146
+++ xc/programs/Xserver/hw/xfree86/drivers/glint/glint_driver.c 2002/05/07 23:15:58 1.147
@@ -28,7 +28,7 @@
* this work is sponsored by S.u.S.E. GmbH, Fuerth, Elsa GmbH, Aachen,
* Siemens Nixdorf Informationssysteme and Appian Graphics.
*/
-/* $XFree86: xc/programs/Xserver/hw/xfree86/drivers/glint/glint_driver.c,v 1.145 2002/02/27 18:27:36 alanh Exp $ */
+/* $XFree86: xc/programs/Xserver/hw/xfree86/drivers/glint/glint_driver.c,v 1.146 2002/02/27 18:41:04 alanh Exp $ */
#include "fb.h"
#include "cfb8_32.h"
@@ -3631,12 +3631,15 @@ GLINTBlockHandler (
ScreenPtr pScreen = screenInfo.screens[i];
ScrnInfoPtr pScrn = xf86Screens[i];
GLINTPtr pGlint = GLINTPTR(pScrn);
+ int sigstate = xf86BlockSIGIO();
if(pGlint->CursorColorCallback)
(*pGlint->CursorColorCallback)(pScrn);
if(pGlint->LoadCursorCallback)
(*pGlint->LoadCursorCallback)(pScrn);
+
+ xf86UnblockSIGIO(sigstate);
pScreen->BlockHandler = pGlint->BlockHandler;
(*pScreen->BlockHandler) (i, blockData, pTimeout, pReadmask);
===================================================================
RCS file: /home/anoncvs/cvs/xc/programs/Xserver/hw/xfree86/drivers/glint/pm2v_dac.c,v
retrieving revision 1.28
retrieving revision 1.29
diff -u -p -r1.28 -r1.29
--- xc/programs/Xserver/hw/xfree86/drivers/glint/pm2v_dac.c 2002/03/20 23:14:51 1.28
+++ xc/programs/Xserver/hw/xfree86/drivers/glint/pm2v_dac.c 2002/05/07 23:15:59 1.29
@@ -27,7 +27,7 @@
* this work is sponsored by S.u.S.E. GmbH, Fuerth, Elsa GmbH, Aachen and
* Siemens Nixdorf Informationssysteme
*/
-/* $XFree86: xc/programs/Xserver/hw/xfree86/drivers/glint/pm2v_dac.c,v 1.27 2001/11/28 21:53:01 alanh Exp $ */
+/* $XFree86: xc/programs/Xserver/hw/xfree86/drivers/glint/pm2v_dac.c,v 1.28 2002/03/20 23:14:51 alanh Exp $ */
#include "xf86.h"
#include "xf86_OSproc.h"
@@ -447,16 +447,6 @@ Permedia2vLoadCursorCallback(
Permedia2vOutIndReg(pScrn, PM2VDACRDCursorPattern+i, 0x00,
pGlint->HardwareCursorPattern[i]);
- if (pGlint->Chipset == PCI_VENDOR_3DLABS_CHIP_PERMEDIA3) {
- /*
- * Seems like a bug in the PM3, this flips back the RGB/BGR
- * triggered by a cursor update. Strange, but true....
- */
- GLINTRegPtr glintReg = &pGlint->ModeReg[0];
- Permedia2vOutIndReg(pScrn, PM2VDACRDColorFormat, 0x00,
- glintReg->DacRegs[PM2VDACRDColorFormat]);
- }
-
pGlint->LoadCursorCallback = NULL;
}
===================================================================
RCS file: /home/anoncvs/cvs/xc/programs/Xserver/hw/xfree86/drivers/glint/README.pm3,v
retrieving revision 1.12
retrieving revision 1.13
diff -u -p -r1.12 -r1.13
--- xc/programs/Xserver/hw/xfree86/drivers/glint/README.pm3 2001/08/18 11:37:30 1.12
+++ xc/programs/Xserver/hw/xfree86/drivers/glint/README.pm3 2002/05/07 23:15:58 1.13
@@ -1,4 +1,4 @@
-/* $XFree86: xc/programs/Xserver/hw/xfree86/drivers/glint/README.pm3,v 1.11 2001/05/08 19:31:22 alanh Exp $ */
+/* $XFree86: xc/programs/Xserver/hw/xfree86/drivers/glint/README.pm3,v 1.12 2001/08/18 11:37:30 alanh Exp $ */
STATUS as of Tue, 8 May 2001 19:01:39 +0200
@@ -41,7 +41,6 @@ Working :
- 1 -> 25% Video, 75% FB
- 2 -> 75% Video, 25% FB
- 3 -> Video Only
- * DRI : Work is underway.
Not Working :
@@ -53,19 +52,6 @@ Not Working :
- Render extension initialization.
- CPUToScreenTexture.
- CPUToScreenAlphaTexture.
- * [WORK IS UNDERWAY] DRI/OpenGL (only together with a gamma chip) :
- - DRI support should be ok, but accelerated openGL is not yet supported.
- - The accelerated OpenGL library supposes we are using a gamma together
- with a MX rasterizer, not a Permedia3 one.
-
-Known problems :
-
- * [FIXED] R-B inversion. Sometimes, all of a sudden, it seems as the
- RGB order gets changed. going to console and back solves this.
- Well, this is partly fixed, but still appears under very heavy load.
- => [NOTE] If this still happens, try disabling the Hardware cursor, with
- the "SWCursor" option to your device XF86Config section.
- => [NOTE2] Or, try starting the Xserver with the -nosilk option.
Sven Luther <luther@dpt-info.u-strasbg.fr>
Alan Hourihane <alanh@fairlite.demon.co.uk>
Reply to: