On Fri, Apr 5, 2013 at 3:13 AM, Brendan Dolan-Gavitt <brenda...@gatech.edu> wrote: > In target-i386 cpu_get_phys_page_debug, the CR4_PAE bit is checked > before CR0_PG. This means that if paging is disabled but the PAE bit has > been set in CR4, cpu_get_phys_page_debug will return the wrong result > (it will try to translate the address as virtual rather than using it as > a physical address). This patch fixes that by moving the CR0_PG check to > the beginning of the function. > > This shows up when booting the Linux kernel on amd64 with "-d in_asm". > The kernel turns on the PAE bit in CR4 before turning on paging, and so > QEMU's disassembler will fail because it will try to walk the page > tables to fetch code even though paging is disabled. The symptom is > incorrect disassembly and some "Disassembler disagrees with translator > over instruction decoding" messages. > > This was also reported as bug #1163065.
Hi, a while ago I sent similar patch: http://comments.gmane.org/gmane.comp.emulators.qemu/180776 and a suggestion for me was to unify cpu_get_phys_page_debug and cpu_x86_handle_mmu_fault implementations. -- Thanks. -- Max
