Re: [PATCH v14 08/10] Adding info [tb-list|tb] commands to HMP (WIP)

[Peter Maydell]

On Thu, 8 Jun 2023 at 13:06, Dr. David Alan Gilbert <dave@treblig.org> wrote:
>
> * Peter Maydell (peter.maydell@linaro.org) wrote:
> > On Thu, 8 Jun 2023 at 08:44, Wu, Fei <fei2.wu@intel.com> wrote:
> > >
> > > On 6/7/2023 8:49 PM, Wu, Fei wrote:
> > > > On 6/1/2023 10:40 AM, Richard Henderson wrote:
> > > >> Did you really need something different than monitor_disas?  You almost
> > > >> certainly want to read physical memory and not virtual anyway.
> > > >>
> > > > Yes, it's usually okay for kernel address, but cannot re-gen the code
> > > > for userspace virtual address (guest kernel panic on riscv guest). I
> > > > tried monitor_disas() but it failed to disas too:
> > > >     monitor_disas(mon, mon_get_cpu(mon), tbs->phys_pc, num_inst, true);
> > > >
> > > > How to use this function correctly?
> > > >
> > > 'phys_pc' in tbs is returned by get_page_addr_code_hostp(), which is not
> > > guest phys address actually, but ram_addr_t instead, so it's always
> > > wrong for monitor_disas. After some dirty changes, tbs can record the
> > > guest pa. Now we can disas both kernel and user space code. But still,
> > > no code is regenerated, disas in 'info tb' is just a convenient way to 
> > > 'xp'.
> > >
> > > Is there any existing function to convert ram_addr_t to guest pa?
> >
> > Such a function would not be well-defined, because a block of RAM
> > as specified by a ram_addr_t could be present at (aliased to) multiple
> > guest physical addresses, or even currently not mapped to any guest
> > physical address at all. And it could be present at different physical
> > addresses for different vCPUs.
>
> True, but isn't there a similar mechanism for when an MCE happens
> in the host memory?

That goes via kvm_physical_memory_addr_from_host(), which looks
up the host address in the KVMSlot array (and just picks the
first hit). It won't work for TCG.

thanks
-- PMM