Re: Kernel Summit presentation

Re: Kernel Summit presentation

[Alexander E. Patrakov]

Jon Smirl wrote:

> I attached the slides from my kernel summit presentation. The major
> change is that Linus does not want a new VT state. Instead we should
> open a new VT, switch it to VT_GRAPHICS, and then refuse to
> acknowledge VT swap requests. Dave Airlie did another presentation on
> mode setting and suspend/resume.
> 
> The simple overview: goal is to get rid of the current VT swap system
> but replace it with something that looks just like it. To do this the
> console is split into two halves.
> 
> 1) kernel console - it is as simple as possible and always guaranteed
> to work. It is simpler than current fbconsole. It automatically uses
> whatever mode is set on the monitor. There is only one kernel console
> - no VTs.
> 
> 2) user console - implements your normal VTs. Fully accelerated,
> supports Unicode, etc.
> 
> No one was really against restructuring the consoles. There was
> support for a console that always works even when something like X is
> running.

I am really interested in this, and want to help (with code, once 
architectural issues are solved). But there is a question:

How should applications that currently rely upon a kernel VT system being 
present (e.g. Xorg, MPlayer, SDL) be modified? (I mean, is there any proposed 
API for hunting for a free userspace VC?)

-- 
Alexander E. Patrakov


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO September
19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf

Re: Re: Kernel Summit presentation

[Jon Smirl]

On 7/28/05, Alexander E. Patrakov <patrakov@ums.usu.ru> wrote:
> Jon Smirl wrote:
> 
> > I attached the slides from my kernel summit presentation. The major
> > change is that Linus does not want a new VT state. Instead we should
> > open a new VT, switch it to VT_GRAPHICS, and then refuse to
> > acknowledge VT swap requests. Dave Airlie did another presentation on
> > mode setting and suspend/resume.
> >
> > The simple overview: goal is to get rid of the current VT swap system
> > but replace it with something that looks just like it. To do this the
> > console is split into two halves.
> >
> > 1) kernel console - it is as simple as possible and always guaranteed
> > to work. It is simpler than current fbconsole. It automatically uses
> > whatever mode is set on the monitor. There is only one kernel console
> > - no VTs.
> >
> > 2) user console - implements your normal VTs. Fully accelerated,
> > supports Unicode, etc.
> >
> > No one was really against restructuring the consoles. There was
> > support for a console that always works even when something like X is
> > running.
> 
> I am really interested in this, and want to help (with code, once
> architectural issues are solved). But there is a question:
> 
> How should applications that currently rely upon a kernel VT system being
> present (e.g. Xorg, MPlayer, SDL) be modified? (I mean, is there any proposed
> API for hunting for a free userspace VC?)

It depends on how the app is implemented. If the app directly messes
with the hardware it will have to be significantly changed. For
example there is a new Xserver under development that plays by the new
rules. The old Xorg server will not work once you enter the new state.
This is probably true for SDL too.

The main goal of the new system is to weed out old apps that directly
mess with the hardware. These apps will need to be rewritten to work
in the new model. The fbdev, DRM and OpenGL/EGL APIs will be available
in the new system.

Standalone SDL would easily port to OpenGL/EGL. EGL is an OpenGL
extension that allows mode setting and screen control. Using
OpenGL/EGL will hide the changes from you.

The new system allows two users, each logged on to a separate head of
a single card. You can't running an old app that messes with the
hardware in this case. The old app would kill both users.

This is a complicated change in the VT system. It will probably be
several years before everything is transitioned.


> 
> --
> Alexander E. Patrakov
> 
> 
> -------------------------------------------------------
> SF.Net email is Sponsored by the Better Software Conference & EXPO September
> 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
> Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
> Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
> _______________________________________________
> Linux-fbdev-devel mailing list
> Linux-fbdev-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/linux-fbdev-devel
> 


-- 
Jon Smirl
jonsmirl@gmail.com


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO September
19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf

Re: Re: Kernel Summit presentation

[Jon Smirl]

On 7/28/05, Alexander E. Patrakov <patrakov@ums.usu.ru> wrote:
> Jon Smirl wrote:
> 
> > I attached the slides from my kernel summit presentation. The major
> > change is that Linus does not want a new VT state. Instead we should
> > open a new VT, switch it to VT_GRAPHICS, and then refuse to
> > acknowledge VT swap requests. Dave Airlie did another presentation on
> > mode setting and suspend/resume.
> >
> > The simple overview: goal is to get rid of the current VT swap system
> > but replace it with something that looks just like it. To do this the
> > console is split into two halves.
> >
> > 1) kernel console - it is as simple as possible and always guaranteed
> > to work. It is simpler than current fbconsole. It automatically uses
> > whatever mode is set on the monitor. There is only one kernel console
> > - no VTs.
> >
> > 2) user console - implements your normal VTs. Fully accelerated,
> > supports Unicode, etc.
> >
> > No one was really against restructuring the consoles. There was
> > support for a console that always works even when something like X is
> > running.
> 
> I am really interested in this, and want to help (with code, once
> architectural issues are solved). But there is a question:
> 
> How should applications that currently rely upon a kernel VT system being
> present (e.g. Xorg, MPlayer, SDL) be modified? (I mean, is there any proposed
> API for hunting for a free userspace VC?)

If you are really interested the first place to help is getting the
new Xegl Xserver working. It is discussed on the xorg, dri-devel and
dri-egl mailing lists at freedesktop.org.

We probably won't start working on the other console pieces until the
Xserver is ready. Without the Xserver it is hard to do anything in the
new environment.


> 
> --
> Alexander E. Patrakov
> 
> 
> -------------------------------------------------------
> SF.Net email is Sponsored by the Better Software Conference & EXPO September
> 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
> Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
> Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
> _______________________________________________
> Linux-fbdev-devel mailing list
> Linux-fbdev-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/linux-fbdev-devel
> 


-- 
Jon Smirl
jonsmirl@gmail.com


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO September
19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf