> On July 14, 2014, 4:42 p.m., Andreas Sandberg wrote: > > I have two high-level comments that need to be fixed: > > > > * This is should be four different commits (configuration, CPUID updates, > > segment initialization, kvm, process). > > * Write a proper commit message (see http://www.m5sim.org/Commit_Access). > > Specifically, include a short one-line summary and a longer description of > > what the patch does and why. > > > > I'm also a bit unsure about the design of this patch. From RB2312, it > > /seems/ like the useArchPT flag is supposed to enable page tables in SE > > emulation mode, however in this patch you seem to overload it to mean "use > > magic syscall redirection". Could you clean that up a bit? > > > > Could you redesign the syscall & fault handler to use the memory mapped > > m5ops interface instead? That way, the code would in the kvm CPU would be > > much cleaner and you'd get rid of the magic ports. As a side effect, we > > could potentially scrap the SE-specific bits of the ISA (this is definitely > > something I'd like to see long term). > > > > Alexandru Dutu wrote: > Thank you for your review! If I break this in four different commits, the > simulator might not compile after each commit just after all of them are > applied. That does not sound acceptable to me.
My understanding of the memory mapped m5ops in KVM is that certain portion of /dev/mem (0xFFFF0000 - 0xFFFFFFFF) has to be mapped in the virtual space. I will have to figure out how to do this without a kernel, one might consider just mapping that particular chunk of physical memory in the page table. Currently, in gem5 there is no physical memory allocator in SE mode, so any new virtual to physical mapping will just use the next physical page available. - Alexandru ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: http://reviews.gem5.org/r/2313/#review5187 ----------------------------------------------------------- On July 11, 2014, 4:01 p.m., Alexandru Dutu wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > http://reviews.gem5.org/r/2313/ > ----------------------------------------------------------- > > (Updated July 11, 2014, 4:01 p.m.) > > > Review request for Default. > > > Repository: gem5 > > > Description > ------- > > Changeset 10254:661c482f7d58 > --------------------------- > Making KvmCPU model usable in SE mode. > > > Diffs > ----- > > configs/example/se.py c625a3c51bac050879457e666dd83299a36d761b > src/arch/x86/cpuid.cc c625a3c51bac050879457e666dd83299a36d761b > src/arch/x86/process.cc c625a3c51bac050879457e666dd83299a36d761b > src/arch/x86/regs/misc.hh c625a3c51bac050879457e666dd83299a36d761b > src/arch/x86/system.hh c625a3c51bac050879457e666dd83299a36d761b > src/arch/x86/system.cc c625a3c51bac050879457e666dd83299a36d761b > src/cpu/kvm/base.hh c625a3c51bac050879457e666dd83299a36d761b > src/cpu/kvm/base.cc c625a3c51bac050879457e666dd83299a36d761b > src/cpu/kvm/x86_cpu.hh c625a3c51bac050879457e666dd83299a36d761b > src/cpu/kvm/x86_cpu.cc c625a3c51bac050879457e666dd83299a36d761b > > Diff: http://reviews.gem5.org/r/2313/diff/ > > > Testing > ------- > > Quick regressions passed. > > > Thanks, > > Alexandru Dutu > > _______________________________________________ gem5-dev mailing list gem5-dev@gem5.org http://m5sim.org/mailman/listinfo/gem5-dev
