Just FYI: GeForce 7800 GT on a PowerMac Quad-core G5 with a 2560x1440 display: [mi] EQ overflowing

[Mark Millard]

On Mon Sep 8 12:41:12 UTC 2014 Jukka Ukkonen wrote:

> This may be a more generic problem and not really related to PowerPC 
> environments
> only. I have seen similar mi-complaints on a 12 core AMD64 system with a 
> Radeon GPU
> running 10-stable.
> These problems have only occurred after the introduction of VT-consoles 
> and KMS enabled
> X11.

The powerpc context that I reported on is still using scons and there is no NVIDIA KMS to my knowledge, much less a powerpc one spanning NVIDIA if I understand right.

So I'd guess that even if the time frame is similar for when you observed behavior changes that the actual issue is not VT nor KMS of themselves. If any more general/generic changes were also made that would span scons and NVIDIA then such could contribute.

But the same kind of NVIDIA card used on the same kind of PowerMac G5 with a 1920x1200 display does not generate those notices for the same SSD being used to boot. The notices that I reported seem to be tied to the card having more pixels to deal with (2560x1440), at least for the context I reported about. I have no evidence of it being tied to anything else but I've not tested an older context recently and do not remember the details from my prior use. (I had reasons to be testing basic Xorg/xfce4 display behavior recently and so noticed various notifications more than usual.)

I could imagine internal display-handling configuration tradeoffs in Xorg based on the number of pixels the display has. May be the 2560x1440 is just more extreme than the management of the display handling is designed to well span.

All guess work on my part.



===
Mark Millard
markmi at dsl-only.net

On Sep 8, 2014, at 1:47 AM, Mark Millard <markmi at dsl-only.net> wrote:

FYI: The GeForce 7800 GT on a PowerMac Quad-core G5 with a 2560x1440 display (ADC with converter to DVI) reports the following

[mi] EQ overflowing.  Additional events will be discarded until existing events are processed.
[mi] These backtraces from mieqEnqueue may point to a culprit higher up the stack.
[mi] mieq is *NOT* the cause.  It is a victim.
[mi] Increasing EQ size to 512 to prevent dropped events.
[mi] EQ processing has resumed after 51 dropped events.
[mi] This may be caused my a misbehaving driver monopolizing the server's resources.

Context:

I built a 10.1-PRERELEASE r271215 (to test Justin Hibbits' changes for powerpc). I also used portsnap and then rebuilt my ports from scratch.

FreeBSD FBSDG4S0 10.1-PRERELEASE FreeBSD 10.1-PRERELEASE #0 r271215: Sat Sep  6 23:56:15 PDT 2014     root at FBSDG4S0:/usr/obj/usr/src/sys/GENERIC  powerpc

(I'd be surprised if this was powerpc or PowerMac specific. But use of these old cards may be more likely for PCI-Express PowerMacs.)

===
Mark Millard
markmi at dsl-only.net