Re: strange stuff in arch/i386/kernel/head.S

Cyrille Chepelov (chepelov@rip.ens-cachan.fr)
Wed, 11 Nov 1998 06:48:08 +0100 (MET)


On Tue, 10 Nov 1998, Philip Gladstone wrote:

> My ASUS P2B motherboard has the RSDT, FACP and DSDT. In fact I've
> spent the afternoon writing an AML to ASL converter to decode
> the junk that appears in the DSDT. This stuff is *truly* ghastly.

I finally found the beta BIOS for TX97-XE with ACPI (it seems ASUS' German
offices are a bit ahead of the others). While it seems some people have
reliability problems under win98 (which I simply don't care about), it has
the dreaded tables, so I can at least work :

zagadka%cat /proc/sysmemmap
System memory map entries: 9
Addr=0x0000000000000000 Length=0x000000000009fc00 Type=1 AddressRangeMemory
Addr=0x000000000009fc00 Length=0x0000000000000400 Type=2 AddressRangeReserved
Addr=0x00000000000f0000 Length=0x0000000000010000 Type=2 AddressRangeReserved
Addr=0x0000000000100000 Length=0x0000000000d00000 Type=1 AddressRangeMemory
Addr=0x0000000000e00000 Length=0x0000000000200000 Type=2 AddressRangeReserved
Addr=0x0000000001000000 Length=0x0000000004ffd000 Type=1 AddressRangeMemory
Addr=0x0000000005ffd000 Length=0x0000000000002000 Type=3 AddressRangeACPI
Addr=0x0000000005fff000 Length=0x0000000000001000 Type=4 AddressRangeNVS
Addr=0x00000000ffff0000 Length=0x0000000000010000 Type=2 AddressRangeReserved

(this is using my QSAM patch, and with an artificial memory hole at 14-16
MB for testing and shaking purposes)

What I think would be really cool would be to add a /proc/acpi, with the
whole damn tree mapped to /proc/acpi/tree, some information and
kernel/userland communication files in /proc/acpi (in particular to let
the events flow up to the userland support processes)
The kernel half would be responsible for managing the tree, interpreting
AML code and packaging the system events.
The userland half would be comprised of an AML-to-ASL decoder (an ASL
compiler would be great too, to fix broken BIOSes : once the tree
management works, having a capability to replace a node method on the fly
shouldn't be difficult to do), a facility to tell the kernel to run a
particular AML node method, and of course finally an OSPM daemon.

Whew. Too bad I'm out today and can't code. A big task list for this
evening !

-- Cyrille

------------------------------
Je me demande pourquoi les gens me regardent comme ēa, dans le RER...
les cheveux bleus, sūrement !

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/