[linux-audio-dev] Re: [linux-audio-user] Re: [a-users] Re: [a-devel] File system and Voluntary Kernel Preemption

New Message Reply About this list Date view Thread view Subject view Author view Other groups

Subject: [linux-audio-dev] Re: [linux-audio-user] Re: [a-users] Re: [a-devel] File system and Voluntary Kernel Preemption
From: Lee Revell (rlrevell_AT_joe-job.com)
Date: Fri Jul 16 2004 - 22:24:23 EEST


On Fri, 2004-07-16 at 04:33, Malcolm Baldridge wrote:
> > Some of these fixes were to
> > extremely important areas (get_user_pages, a fix for
> > framebuffer-scrolling issue, and more).
>
> It goes against common sense (to anyone with kernel knowledge) to use the
> console framebuffer anyway.
>
> The duration of kernel locks and the CPU expense of blowing bitmaps around
> like that are shocking.
>
> Seriously. Use text-mode consoles, or a native AGP-driver a la nvidia, ati,
> or whatever. But using a "stupid" framebuffer driver is asking for hardcore
> latency blow-outs.
>

I am using text-mode consoles, and switching back and forth between
those and X using ctrl-alt-F[1-7]. I sent Andrew Morton some XRUN
traces and he said one of them was caused by framebuffer scolling inside
lock_kernel(), and the fix was to remove (un)lock_kernel() from
do_tty_write(). This particular fix may not be in -mm1, because he
seemed to think it would cause problems (display corruption, etc). It
has not caused any problems for me.

Maybe I am confused about the 'framebuffer' vs. a text-mode console, I
thought they were the same. Anyway, the point is, switching from X to a
text-mode console and back caused XRUNS before, and now it doesn't.

Lee


New Message Reply About this list Date view Thread view Subject view Author view Other groups

This archive was generated by hypermail 2b28 : Fri Jul 16 2004 - 22:29:24 EEST