On Sun, Dec 5, 2010 at 12:21 PM, Andreas Färber andreas.faerber@web.de wrote:
Am 05.12.2010 um 03:08 schrieb Tarl Neustaedter:
On 2010-12-4 8:54 AM, Andreas Färber wrote:
[...] Just wondering, what does /virtual-memory@0,0 contain?
Actually the context here was sparc32/sun4m. Artyom reported no /chosen node on an SS-5 but apparently some /virtual-memory@0,0 node.
I thought he also reported the contents of the node here: http://lists.openbios.org/pipermail/openbios/2010-October/005611.html
For SS-5:
ok .attributes available 00000000 fff00000 00100000 00000000 fef00000 00e00000 00000000 00000000 fe400000 00000000 ffe15000 000cd000 00000000 ffd00000 00008000 00000000 fe400000 00b00000 reg 00000000 00000000 80000000 00000000 80000000 80000000 name virtual-memory
Btw, the contents of this node is the same regardless whether the ram size is 256M or 32M.
OpenBIOS/sparc32 on the other hand does have a /chosen node and an "mmu" property, but its value is 0 so that pretty-printing of /virtual-memory's "available" isn't triggered. So I'm thinking, if the virtual memory "available" property is there and we do have a /chosen node then "mmu" should point to /virtual-memory, even if there's no "translations" property. The alternative would be to [IFDEF] CONFIG_SPARC32 a special handling of "available" in the /virtual-memory node.
The question is how does the OS get this information. Can it be that the name of the node doesn't matter?
It Solaris wouldn't get the available memory from the firmware I'd expect some error message like "Can't deduct msgbuf from physical memory list". (had it when qemu had math and mmu problems).