Tiny Core Base > Alpha Releases

Tiny Core 3.0 Alpha 1 Testing

<< < (8/14) > >>

bmarkus:
Is there any benefit to recompile an extension in 3.0 which otherwise works perfectly from 2.x?

curaga:
Possible speed gains and size losses, nothing guaranteed though.

Juanito:
..and I guess also the extension could make use of additional functionality in the updated glibc and/or kernel.

I did notice some significant size differences in some cases - perhaps as a result of cloog, etc

yoshi314:

--- Quote from: curaga on May 10, 2010, 05:38:41 AM ---Short: Intel graphics users wanting Xorg, you'll need to change the default to UMS mode for Xorg 7.4. 7.5 will support KMS.

Long:

What you're seeing is Intel's new graphics stack.
They only support KMS now (in the Xorg driver, the kernel driver still has the UMS mode for backwards compatibility); if you wish to use UMS, add "options i915 modeset=0" to /etc/modprobe.conf before loading the extension/module.

As you've seen, the KMS stack conflicts with both the vesa framebuffer and the vesa driver (both tinyX and Xorg), because it sets the resolution itself. It gives less flicker, and the optimal resolution from the start (full-res console).

The module in KMS mode should be loaded in pure text mode; inside X or with a framebuffer it's likely to crash and burn.


So far Intel is the only one going this far; radeon and nouveau also support KMS, but theirs defaults off.

The intel driver in Xorg 7.4 doesn't support KMS IIRC, so to run with it turn modesetting off.

--- End quote ---
i think ati already does KMS by default.

worth noting - if your display gives corrupt EDID (old [worn down] CRTs tend to do it), you'd better steer clear from ati KMS. it can lockup your system, display garbage and lockup your system, or give you a white screen of death and lockup your system ;) (all of this on radeon module init, and it's pretty random).

there is work going into fixing such lockups, but it's still not commited nor finished.

recently a safe bet was to use "nomodeset" parameter which would disable modesetting on any driver, but it seems it might not work on all KMS aware drivers anymore.

curaga:

--- Quote from: Juanito on May 10, 2010, 06:50:53 AM ---..and I guess also the extension could make use of additional functionality in the updated glibc and/or kernel.

--- End quote ---
Indeed, for example an updated Qemu would now take advantage of KSM. Not high on my todo list though, the userbase running a TC host with hundreds of similar VMs in KVM is quite small ;)

@yoshi314: it wasn't enabled by default, I left it that way. Ati users will need to enable it explicitly, via doing the same but with 1 of course.

Navigation

[0] Message Index

[#] Next page

[*] Previous page

Go to full version