On Sun, Dec 27, 2009 at 7:02 PM, Blue Swirl blauwirbel@gmail.com wrote:
On Sun, Dec 27, 2009 at 11:23 AM, Igor Kovalenko igor.v.kovalenko@gmail.com wrote:
On Sun, Dec 27, 2009 at 12:30 PM, Blue Swirl blauwirbel@gmail.com wrote:
On Sat, Dec 26, 2009 at 10:07 PM, Igor V. Kovalenko igor.v.kovalenko@gmail.com wrote:
From: Igor V. Kovalenko igor.v.kovalenko@gmail.com
- claim both physical and virtual address ranges
while walking boot memory mappings
With this patch applied, SILO can't find enough memory. For example:
Easy to reproduce, and expected. Please try -m 256 in qemu command line, 128M default is not enough.
Why? Is this related to changed mapping of first 64MB? Perhaps that value should be lower, for example 16MB?
Relocating 16M is not enough, since we have devices at 0x1000000 and 0x2000000
Then, patched silo tries to allocate 64M (original one was 8M) Since we have to fit bss and stack into RAM as well, 128M is exhausted.