by Valdis Klētnieks

[permalink] [raw]

Subject: Re: 2.5.69-mm6

On Fri, 16 May 2003 16:37:43 BST, Alan Cox said:
> All of this stuff should be disablable and far more. It probably all
> wants hiding under a single "Shrink feature set" type option most people
> can skip over as they do with kernel debugging.

No, this sort of thing should be in the .config file, but NOT accessible
via the point-and-drool interface. Make them vi it and do it the hard way...

The difference being that if you turn on kernel debugging, there's less chance
of random program XYZ you download from someplace throwing an ENOSYS because
some syscall is missing/broken, with the ensuing hilarity of debugging (since
the actual problem might be in some library rather than in the code you
downloaded).

Of course, I may just be jaded - I've gotten too many reports from bozos with
firewalls that read "ntp-1.vt.edu is portscanning from port 123, please stop it
right now or I'm calling the FBI". But although most people that build their
own kernels can reason through "maybe my FooBar2003 would work better if I
included the driver for it", but figuring that Frobozz isn't working because
libgronk is broken because somebody removed futexes because they thought they
sounded futile.....

Attachments:

(No filename) (226.00 B)

2003-05-16 17:18:31

On Fri, May 16, 2003 at 06:55:39PM +0100, Dave Jones wrote:
> On Fri, May 16, 2003 at 07:28:34PM +0200, Andreas Henriksson wrote:
>
> > I had to remove "static" from the agp_init-function in
> > drivers/char/agp/backend.c to get the kernel to link (when building
> > Intel 810 Framebuffer into the kernel).
>
> wrong fix. nuke the agp_init() call from i810fb
> note, it still won't actually work. i810fb still fails to init
> the agpgart for some reason.
>

Ok.. thanks for the quick reply .. I just booted the kernel and noticed
that the console got stuck (but the it booted fine except from that).

Though I find this weird:
The last thing I saw on the console was the detection of hdb and it's
partitions.... I tried unplugging hdb and it booted just fine...
coinsidence?

And by the way.... the framebuffer flickers (is that the right word?)
for me.... It looks like an old TV... (Has done with all the (2.5)
kernels I've tried).. Is this a known problem and if so is there a
solution?
I'm using a TFT monitor and this is my append-line..
append="video=i810fb:xres:1280,yres:1024,bpp:16,hsync1:30,hsync2:82, \
vsync1:50,vsync2:75,accel"
(... if it matters.)

> > I also got unresolved symbols for two modules.
> > arch/i386/kernel/suspend.ko: enable_sep_cpu, default_ldt, init_tss
> > arch/i386/kernel/apm.ko: save_processor_state, restore_processor_state
>
> Mikael's patch for these has been posted several times already in the
> last few days.
>
Ok.. thanks again...

> Dave

Regards,
Andreas Henriksson

Attachments:

(No filename) (1.49 kB)
(No filename) (189.00 B)
Download all attachments

2003-05-16 18:16:03

by Dave Jones

[permalink] [raw]

Subject: Re: 2.5.69-mm6

On Fri, May 16, 2003 at 08:10:42PM +0200, Andreas Henriksson wrote:

> Ok.. thanks for the quick reply .. I just booted the kernel and noticed
> that the console got stuck (but the it booted fine except from that).

The reports I've seen have shown that i810fb outputs some warnings
during startup about being unable to allocate using agpgart.
As i810fb depends on agp, this would imply its functionality would break.
Are you sure it's working, and not falling back to non-fb mode ?

> And by the way.... the framebuffer flickers (is that the right word?)
> for me.... It looks like an old TV... (Has done with all the (2.5)
> kernels I've tried).. Is this a known problem and if so is there a
> solution?
> I'm using a TFT monitor and this is my append-line..
> append="video=i810fb:xres:1280,yres:1024,bpp:16,hsync1:30,hsync2:82, \
> vsync1:50,vsync2:75,accel"
> (... if it matters.)

That's probably one for Antonio to figure out, unfortunatly he's busy
right now.

Dave

2003-05-16 18:42:21

by Dave Jones

[permalink] [raw]

Subject: Re: [OOPS] 2.5.69-mm6

On Fri, May 16, 2003 at 11:08:48AM -0700, William Lee Irwin III wrote:
> On Fri, May 16, 2003 at 01:26:20PM +0200, Alexander Hoogerhuis wrote:
> > This one goes in -mm5 as well, machine runs fine for a while in X, but
> > trying to switch to a vty send the machine into the tall weeds...
>
> Could you run with the radeon driver non-modular and kernel debugging
> on? Then when it oopses could you use addr2line(1) to resolve this to
> a line number?
>
> I'm at something of a loss with respect to dealing with DRM in general.

Not that I'm pointing fingers, but it could be that
reslabify-pgds-and-pmds.patch again ? Maybe it's still not quite right?
Might be worth backing out and retesting, just to rule it out.

Dave

2003-05-16 18:54:00

Hello!

On Sat, May 17, 2003 at 01:51:29PM +0100, Alan Cox wrote:
> > What else should be disablable?
[...]
> swapping

This one is alreay disablable, I think.
At least my 2.5.69 from current bk have this
CONFIG_SWAP in "General setup" section of config ;)

Bya,
Oleg

2003-05-17 21:54:49

Hello,

Has anyone figured out the radeon_cp_init problem? I am still suffering of
this with 2.5.70-bk14. This is with Gentoo 1.4 and gcc 3.2.3 (on a notebook
with Radeon 9000 Mobility and SiS AGP chipset).

Regards, Sami

P.S. Please CC me any reply as I am not subscribed to the list.

> On Mon, 19 May 2003, Alexander Hoogerhuis wrote:
> > --[PinePGP]--------------------------------------------------[begin]--
> > The oops is gone, and I'm now left with this one:
>
> Ultra Cool
>
> > Linux agpgart interface v0.100 (c) Dave Jones
> > [drm] Initialized radeon 1.8.0 20020828 on minor 0
> > [drm:radeon_cp_init] *ERROR* radeon_cp_init called without lock held
> > [drm:radeon_unlock] *ERROR* Process 4421 using kernel context 0
> >
> > This one only seems to appear when I'm compiling it modular.
>
> Wading through that code isn't something to undertake at this hour, i'll
> have a look a bit later.
>
> Zwane

--
Linux 2.5.70-bk14