linuxconsole-dev

Synced to 2.4.10

[James S. <jsi...@tr...>]

After alot of work I synced our stuff up with 2.4.10. I tested it and it
worked. Please test and let me know if you have any problems. Thank you.

P.S
   In 2.4.10 their where issues (in dmi_scan.c) with PS/2 and suspend on
certain Dell devices. It might be time to think about more in depth
thinking about power management.

Re: Synced to 2.4.10

[Johann D. <jo...@Do...>]

On Sat, 6 Oct 2001, James Simmons wrote:

> 
> After alot of work I synced our stuff up with 2.4.10. I tested it and it
> worked. Please test and let me know if you have any problems. Thank you.
> 

I still have this error in include/asm/hw_irq.h:
                 from vgacon.c:44:
/home/johann/linuxconsole/linux/include/asm/hw_irq.h: In function
`x86_do_profile':
/home/johann/linuxconsole/linux/include/asm/hw_irq.h:201: `current'
undeclared (first use in this function)
/home/johann/linuxconsole/linux/include/asm/hw_irq.h:201: (Each undeclared
identifier is reported only once
/home/johann/linuxconsole/linux/include/asm/hw_irq.h:201: for each
function it appears in.)

The best solution would probably be that hw_irq.h included sched.h. That's
not our job. Or a quick fix would be to include sched.h in vgacon.c. That
we can do.

-- 
Johann Deneux

Re: Synced to 2.4.10

[M. R. B. <mr...@0x...>]

* Johann Deneux <jo...@Do...> on Sun, Oct 07, 2001:

> 
> The best solution would probably be that hw_irq.h included sched.h. That's
> not our job. Or a quick fix would be to include sched.h in vgacon.c. That
> we can do.
> 

No, including sched.h in vgacon.c is better, if you try to include it in
hw_irq.h, you may end up with circular references, etc.  asm/ header files
should never include anything out of linux/, it should always be the other
way around.

M. R.

Re: Synced to 2.4.10

[James S. <jsi...@tr...>]

> Or a quick fix would be to include sched.h in vgacon.c. That we can do.

Done!

Re: Synced to 2.4.10

[Johann D. <jo...@Do...>]

On Sat, 6 Oct 2001, James Simmons wrote:

> 
> After alot of work I synced our stuff up with 2.4.10. I tested it and it
> worked. Please test and let me know if you have any problems. Thank you.
> 
I encountered other problems:
1.
in drivers/video/vga.c, line 318:
                spin_unlock_irqrestore(&vga_lock, flags);
vga_lock is not known. I guessed a fix, but I don't know if it works. At
least, it compiles. The fix is to replace &vga_lock by &state->vga_lock.

2.
In lib/bust_spinlocks.c, line 25:
#ifdef CONFIG_SMP
                atomic_set(&global_irq_lock,0);
#endif
global_irq_lock isn't known.

-- 
Johann Deneux

Re: Synced to 2.4.10

[James S. <jsi...@tr...>]

> I encountered other problems:

Fixed. Thanks for picking these bugs up. 

Re: Synced to 2.4.10

[Johann D. <jo...@Do...>]

On Mon, 8 Oct 2001, James Simmons wrote:

> 
> > I encountered other problems:
> 
> Fixed. Thanks for picking these bugs up. 
> 
We are not finished, yet ;)
I had to add a few includes to linux/lib/bust_spinlocks.c (asm/smp.h and
linux/sched.h). I also had to replace the atomic_set(&global_irq_lock,
0) by global_irq_lock=0 (hmm, is that safe ?). Indeed, atomic_set needs an
atomic_t, while global_irq_lock is an integer.
Once this was done, I managed to get the bzImage file compiled.
Unfortunately, trying to boot it gave nothing at all. It did not even seem
to crash, as it hardly started.
After choosing the version to boot on in grub, the screen went blank. The
keyboard did not seem to respond. I do not even remember seeing the
"Uncompressing kernel image ..........." message.
The logs show nothing at all, as if the kernel was never booted.

My linux box is a bi-PII-333 on an ASUS P2BD. My keyboard is plugged on a
PS2 port, and my graphic card is a Matrox G200. I guess the rest of the
configuration does not really matter for the moment.
The original 2.4.10 kernel works fairly well.

Btw, enabling the debug options in "Kernel Hacking" leads to multiple
definitions of do_BUG. But I guess I can live with that for now.

-- 
Johann Deneux

Re: Synced to 2.4.10

[James S. <jsi...@tr...>]

> > Fixed. Thanks for picking these bugs up. 
> > 
> We are not finished, yet ;)
> I had to add a few includes to linux/lib/bust_spinlocks.c (asm/smp.h and
> linux/sched.h). 

Ah!! Okay. I put a new file into CVS. Lets see if that works. 

I also had to replace the atomic_set(&global_irq_lock,
> 0) by global_irq_lock=0 (hmm, is that safe ?). Indeed, atomic_set needs an
> atomic_t, while global_irq_lock is an integer.

You got a good point. Let me look...

Ug! The problem is global_irq_lock is atomic on some platforms and a
spinlock on others. 

> Once this was done, I managed to get the bzImage file compiled.
> Unfortunately, trying to boot it gave nothing at all. It did not even seem
> to crash, as it hardly started.

I will test it tonight.

> After choosing the version to boot on in grub, the screen went blank. The
> keyboard did not seem to respond. I do not even remember seeing the
> "Uncompressing kernel image ..........." message.

Really strange. 

> bi-PII-333 

Try it without SMP support.

> Btw, enabling the debug options in "Kernel Hacking" leads to multiple
> definitions of do_BUG. But I guess I can live with that for now.

Will fix tonight.

Re: Synced to 2.4.10

[Johann D. <jo...@Do...>]

On Fri, 12 Oct 2001, James Simmons wrote:

> > Once this was done, I managed to get the bzImage file compiled.
> > Unfortunately, trying to boot it gave nothing at all. It did not even seem
> > to crash, as it hardly started.
> 
> I will test it tonight.
> 
> > After choosing the version to boot on in grub, the screen went blank. The
> > keyboard did not seem to respond. I do not even remember seeing the
> > "Uncompressing kernel image ..........." message.
> 
> Really strange. 

Depending on my .config, I sometimes manage to get this
message. Sometimes, the system even works "properly" except that I cannot
interact with it (keyboard useless, kernel unable to open initial
consoles)

> > > bi-PII-333 
> 
> Try it without SMP support.

I did, and the same behaviour was observed.

I will try to find out what configurations lead to what errors. This may
take some time however, as the all make mrproper; make xconfig; make
dep; make bzImage... takes quite some time

-- 
Johann Deneux