site  contact  subhomenews

KMS versus UMS

February 26, 2013 — BarryK
I was thinking about Kernel Mode Setting and the Xorg Wizard. It is still rather messy.

I was reminded about this when I tried Wary with a 3.8 kernel -- just playing, but I got a weird screen with the Xorg 'vesa' driver, and had to disable KMS for the 'i915' kernel driver.

What I was thinking, what if I built Racy with KMS disabled for i915, radeon and nouveau kernel drivers. Then we would have a UMS (User Mode Setting) environment at first boot, and can see what driver Xorg uses and works with.

Then, from that working base, the Xorg Wizard could offer to turn on KMS to test the intel, radeon or nouveau video hardware, if one of those hardwares is detected.

...that would require a reboot I think.


Video selection

Username: Sage
"Difficult call. Folks migrating from you-know-what enjoy luxuriating in not needing to reboot every time they install or change some parameter! Let's hope your superior skills and intellect can find a novel and innovative solution?

NM10 Chipset
Username: Raffy
"In newer Puppies, the Intel NM10 Express Chipset in Asus X101CH laptop can't be handled by Xorg (915resolution fails, too). It's the low-power chipset of Intel for the 1024x600 resolution netbooks (However, the desktop version works without problem). Wary 5.5rc gets 800x600 in Xorg-Vesa. This is OK, but video can't be played full-screen.

Displaying the desktop at boot-time
Username: GCMartin
"I'm not sure if this will help, but it may. Let's back up for a second and ask, Can the desktop when launched provide feedback of whether there is information on the monitor? If not, is there a way to add that support by bringing it to the attention of development of the X-environment or the X-Org utility. The problem because of the vastness of the video cards-drivers and the monitors for display is one where there should be some test methods or tools that can give feedback such that desktop can be presented visually to allow further tweaking should it be necessary. If there are no methods-tools, then for a reasonable experience from a developer's point of view is to make educated guesses (as we do using the methods available currently) or to dump it into the laps of the users (which is really NOT a good idea). Development must/does understand that they cannot expect dumb users (and most users are really dumb when it comes to knowing all the things that affect a video experience) to offer reasonable guidance in the initial video/desktop experience. To revisit history, as I remember it, the issues of screening display issues are no more and no less that what we had when booting users were force thru text based setup. What has happen, though, is that hitting display issues is much faster now, because, the booting technology does a reasonable job taking us to desktop. Maybe we should over-think the problem, rather, look for additional system's information that can assist which driver-monitor sequence will deliver a useful visual to the user. Is there any system feedback that can assist boot decision process for a comfortable experience that we may not have looked for?

Intel drivers
Username: scsijon
"The only problem I have found with intel drivers is that they now only use one name for all the modern ones. I recently had problems with an i965 until I upgraded the driver and it's 'partner file'. You need ensure that the AND is a 'matched pair' and both are copied across when you upgrade. One by itself is a recipe for disaster.

OT strings
Username: 01micko
"I notice that the "strings" entry in util-linux template is removed. Why? Does it bork something in deb based distros? No problem to me, I'll just make a pet out of it and add it to the devx.

Re strings
Username: BarryK
"01micko, Is the busybox 'strings' not good enough?

Tags: woof