Re: i945G 1280x768 sync polarity bug
Tim Roberts wrote: Barry Scott wrote: Using the Xfree86 4.6.0 i810 driver I'm seeing a problem with sync polarity setting. This modeline is required for 1280x768 mode: Modeline 1280x768 79.30 1280 1335 1473 1665 768 769 772 793 -hsync +vsync But the syncs, as shown on an oscilloscope are +hsync +vsync. Does this matter any more? I thought the relevance of sync polarity ended in the middle of the Clinton administration. Good question. We note that the EDID data wants these sync polarities. Does the monitor use the pulse widths or the polarity to tell one mode from another? We think that the polarity is used, but we are far from certain. Looking at the VBE code it seems that as long as the Flags are defined correctly by the driver then its down to the INT10 BIOS to set the syncs on the hardware. Do you think this is a BIOS bug or a driver bug? It may be an expectations bug. As you note, the Intel driver, like the Savage driver, relies on the BIOS to set the mode. The BIOS has a limited set of video modes, with canned parameters for each timing. It is not infinitely variable. If the BIOS thinks 1280x768 should have positive syncs, then you are going to get positive syncs I'm using 915resolution to add 1280x768 and 1360x768 into the BIOS. It seems that if there is an entry in the BIOS for 1280x768 then the driver is happy to to call INT10 to pass in the timings and sync polarity data. If as you say the BIOS timings are the only ones used then what is the point of the INT10 call to pass in the timings? Barry ___ Devel mailing list Devel@XFree86.Org http://XFree86.Org/mailman/listinfo/devel
Re: i945G 1280x768 sync polarity bug
On Tue, Jun 13, 2006 at 11:44:05AM +0100, Barry Scott wrote: Good question. We note that the EDID data wants these sync polarities. Does the monitor use the pulse widths or the polarity to tell one mode from another? We think that the polarity is used, but we are far from certain. Talking CVT here? as sync polarity there is used to flag H:V proportion. What problem are you trying to solve? If any of your monitors does have a problem requiring tight control over sync polarity, please provide a log with the EDID block dumped. Luc Verhaegen. ___ Devel mailing list Devel@XFree86.Org http://XFree86.Org/mailman/listinfo/devel
Re: i945G 1280x768 sync polarity bug
On 6/13/06, Barry Scott [EMAIL PROTECTED] wrote: Tim Roberts wrote: Barry Scott wrote: Using the Xfree86 4.6.0 i810 driver I'm seeing a problem with sync polarity setting. This modeline is required for 1280x768 mode: Modeline 1280x768 79.30 1280 1335 1473 1665 768 769 772 793 -hsync +vsync But the syncs, as shown on an oscilloscope are +hsync +vsync. Does this matter any more? I thought the relevance of sync polarity ended in the middle of the Clinton administration. Good question. We note that the EDID data wants these sync polarities. Does the monitor use the pulse widths or the polarity to tell one mode from another? We think that the polarity is used, but we are far from certain. Looking at the VBE code it seems that as long as the Flags are defined correctly by the driver then its down to the INT10 BIOS to set the syncs on the hardware. Do you think this is a BIOS bug or a driver bug? It may be an expectations bug. As you note, the Intel driver, like the Savage driver, relies on the BIOS to set the mode. The BIOS has a limited set of video modes, with canned parameters for each timing. It is not infinitely variable. If the BIOS thinks 1280x768 should have positive syncs, then you are going to get positive syncs I'm using 915resolution to add 1280x768 and 1360x768 into the BIOS. It seems that if there is an entry in the BIOS for 1280x768 then the driver is happy to to call INT10 to pass in the timings and sync polarity data. If as you say the BIOS timings are the only ones used then what is the point of the INT10 call to pass in the timings? I'm not familiar with the i810's bios interface, but no other chip I've ever worked on has allowed me to specify exact timings via the BIOS. It's usually either a VESA mode number or a size and a refresh rate. I doubt the i810 is much different, especially since most i810 bioses don't even support modes like 1280x768 out of the box. Alex Barry ___ Devel mailing list Devel@XFree86.Org http://XFree86.Org/mailman/listinfo/devel
