Re: [Qemu-devel] [PATCH 0/5] trace: Add events for vCPU memory accesses

[Lluís Vilanova]

Stefan Hajnoczi writes:

> On Wed, Mar 16, 2016 at 03:10:01PM +0000, Peter Maydell wrote:
>> On 2 March 2016 at 10:55, Stefan Hajnoczi <address@hidden> wrote:
>> > On Tue, Feb 23, 2016 at 07:22:07PM +0100, Lluís Vilanova wrote:
>> >> NOTE: This series applies on top of "trace: Show vCPU info in guest code 
>> >> events"
>> >>
>> >> This series adds to new events:
>> >>
>> >> * guest_vmem: memory accesses performed by vCPUs (guest code)
>> >>
>> >> * guest_vmem_user_syscall: memory accesses performed by syscall emulation 
>> >> when
>> >>   running QEMU in user-mode.
>> >>
>> >> Signed-off-by: Lluís Vilanova <address@hidden>
>> >> ---
>> >>
>> >> Lluís Vilanova (5):
>> >>       exec: [tcg] Track which vCPU is performing translation and execution
>> >>       trace: [all] Add "guest_vmem" event
>> >>       user: Refactor lock_user body into do_lock_user
>> >>       user: Set current vCPU during syscall execution
>> >>       trace: [all] Add "guest_vmem_user_syscall" event
>> 
>> > Any comments from TCG folks?
>> 
>> The first two patches which add TCG guest data access tracing look
>> OK to me, but I'm much less sure about the last three which are
>> adding tracing into linux-user syscall emulation. I'm not sure
>> that lock_user is the right place to put that tracepoint.

> Any thoughts on this, Lluís?

Mmmm, I was struggling to find a place to easily add the tracing events whenever
the syscall emulation code accesses guest memory.

The lock_user function is used precisely for that, but it can be a bit
heavy-handed as to what memory is actually read/written, since it only marks the
"potential" ability of doing so (it's a sort of acquire/release interface that
differentiates between read and write acquires).

So, while it's not the perfect place to do it, I think it's the best one we have
now. Having calls to both lock_user and trace_guest_vmem_user_syscall in syscall
emulation code should be avoided to keep the code cleaner.


Thanks,
  Lluis