On Thu, 21 Apr 2005, vocemanago wrote:
> > Here are my notes I compiled while kernel building......
> >
> >
> > Subject: ts-7200: compiling a kernel (fwd)
> >
> > uncompress and untar the latest linux24.tgz from the web site
> >
> > Then edit the linux24/Makefile....
> >
> > ....
> >
> > make modules
> > INSTALL_MOD_PATH=`pwd`/modules make modules_install
> >
> > installs kernel into linux24/modules
> >
>
> I can't seem to get the modules to do this "modules_install" action.
> Do I need to do this? I suppose , at this point, all I am doing is
> updating the kernel so it will have a more accurate clock. I imagine I
> wouldn't actually need to load in the recompilied modules for that.
Yes thats true.
>
> Of course, patching time.h is the goal, if I can just figure out how
> to form the "patch" that Jesse wrote in the one message and make it so
> that the patch program will accept it. It keeps throwing out my Hunks.
> I am having a tough time finding a website that lists what a "legit"
> patch looks like.
> This is my weakness right now. I realize that patches are generated
> from diff operations, but I would think that I could find an example
> "patch" with an explanation.
> I can't just drop the thing in the message into a file and make it work.
I didn't bother. I just hand edited the file! It's hardly a complicated
patch.
Also I found the value given by Jesse was sub-optimal. I posted an email
to this list sometime ago with a more accurate value...........
>From Fri Apr 22 21:11:45 2005
Date: Mon, 27 Dec 2004 20:12:44 +0000 (GMT)
From: Jim Jackson <>
To: Jesse Off <>
Cc: ts-7000 discussion group <>
Subject: Re: [ts-7000] ts7200 timing accuracy.....
Jesse et al.
I've found sometime to patch the kernel with this and try it out.
I've found that the value "5085" caused about 8 secs to be lost in just
over 10 hours (about 1 sec in 4500 lost) so I have used a value of 5084
which gives a better accuracy - at least for my crystal and board.
cheers
Jim
On Wed, 1 Dec 2004, Jesse Off wrote:
>
> It looks like the following patch to the kernel fixes (for the most part)
> the horrible time drift. Apply this patch to
> include/asm-arm/arch-ep93xx/time.h and recompile a kernel to fix.
>
> //Jesse Off
>
> Index: time.h
> ===================================================================
> RCS file: /cvsroot/ts-7200/dist/linux24/include/asm-arm/arch-ep93xx/time.h,v
> retrieving revision 1.1
> retrieving revision 1.2
> diff -u -r1.1 -r1.2
> --- time.h 22 Jul 2004 19:50:54 -0000 1.1
> +++ time.h 1 Dec 2004 20:03:07 -0000 1.2
> @@ -26,12 +26,6 @@
> /* First timer channel used for timer interrupt */
>
> /*
> - * We want the timer interval setup for 10,000 uSec.
> - */
> -#define TIMER_INTERVAL 10000 /* uSeconds */
> -
> -
> -/*
> * IRQ handler for the timer
> */
> static void ep93xx_timer_interrupt(int irq, void *dev_id, struct pt_regs
> *regs)
> @@ -66,8 +60,8 @@
> * the timer load value is the timer interval
> * divided by 2.
> */
> - outl( (TIMER_INTERVAL >> 1), TIMER1LOAD );
> - outl( (TIMER_INTERVAL >> 1), TIMER1VALUE); /* countdown */
> + outl( 5085, TIMER1LOAD );
> + outl( 5085, TIMER1VALUE); /* countdown */
>
> /*
> * Set up Timer 1 for 508 kHz clock and periodic mode.
>
>
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/ts-7000/
<*> To unsubscribe from this group, send an email to:
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
|