[prev in list] [next in list] [prev in thread] [next in thread] 

List:       acpi4linux
Subject:    [ACPI] ACPI problems with Samsung A10XTC
From:       Marcel Schleusner <schleusner () web ! de>
Date:       2002-06-24 19:50:11
[Download RAW message or body]

Hi everybody,

>It should be easy to decide which one it is... One printk before
>ioremap and one phys_to_virt. Compared to number of printk's you must
>have already done....

I had not done this one (I got lazy after numerous reboots), but I tried it 
out now:

ioremap((unsigned long) phys, size);

is the culprit. 

>phys = 0xdef * 64K ~= 256 * 64K * 16 = 4K * 64K = 256M. Do you have at
>least 256MB ram? If not I can see why phys_to_virt would fail...
I have 256 MB precisely. But the upper 32 MB are used by the vga chipset. It's 
a VIA Twister-K which uses the normal memory for his grafic memory. So I have 
to pass mem=224M.

But this brought me an idea and I tried several options:

2.4.18+acpi-patch, no mem=, grub without --disable-auto-linux-mem-opt
							-> AE_NO_MEMORY
now everything with grub --disable-auto-linux-mem-opt:
2.4.18+acpi-patch, mem=224M -> AE_NO_MEMORY
2.4.18+acpi-patch, mem=256M -> kernel panic
2.4.18+acpi-patch, no mem= -> ACPI detected, system hangs
2.5.24, mem=224M -> ACPI detected, system hangs
2.5.21, no mem= -> ACPI detected, system hangs
2.4.18, mem=224M -> AE_NO_MEMORY
2.4.18, no mem= -> ACPI detected, system hangs

So this AE_NO_MEMORY comes whenever I pass an mem=224M or when grub autopasses 
it (how nasty in combination with this error!). As it's gone in 2.5.21-24, 
this is a severe bug in the 2.4.18 kernel, that does something wrong together 
with mem=224M, however not an error of the acpi-system.

The system hangs in 2.5.21/24 and 2.4.18 without mem=224M:
Unfortunately they do not have something to do with the unstable state of the 
2.5. kernels (as I thought before), but seem to be a generic problem of the 
ACPI subsystem, because 2.4.18 with or without the acpi-patch have the same 
problem, if the mem= problem is curcumvented.

Here is what I get as debug messages from 2.4.18+patch:
tbxface-0100 [03] Acpi_load_tables: ACPI Tables successfully loaded
Parsing Methods: ..................... [lots of dots]................
Table [DSDT] - 376 Objects with 42 Devices 105 Methods 14 Regions
Parsing Methods: [no dots]
Tables [SSDT] - 0 Objects with 0 Devices 0 Methods 0 Regions
ACPI Namespace successfully loaded at root c0316e7c

[STOP]
Then nothing happens and the system reacts only to the power button (no soft 
reboot, no SYSRQ, nothing). I even had to manually copy the printout, as 
nothing gets saved to kern.log. Linux 2.4.18 (without acpi-patch) is similar, 
it stops with "ACPI: Core subsystem version [20011018]".
I tried recompiling with [x] CPU enumeration only. That does work (!), but as 
soon as I turn on the full ACPI again, even without any options like AC 
Adapter, CPU or Thermal, I get the hang again. So there is something wrong 
with the ACPI system, just that I don't have a clear error message this time.
As other laptops with VIA Twister-K chipset were reported to work, I don't 
think it's a general incompatibility with this chipset, but what else can 
this be? As soon as the first problem (mem=224M) is solved, we have the next 
now.
Any ideas? I won't give up as long as I have something to try out...

Regards,
Marcel


-------------------------------------------------------
Sponsored by:
ThinkGeek at http://www.ThinkGeek.com/
_______________________________________________
Acpi-devel mailing list
Acpi-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/acpi-devel
[prev in list] [next in list] [prev in thread] [next in thread]