Re: [Qemu-devel] [RFC 01/10] target-arm: protect cpu_exclusive_*.

[Peter Maydell]

On 27 February 2015 at 16:54, Mark Burton <address@hidden> wrote:
>
>> On 26 Feb 2015, at 23:56, Peter Maydell <address@hidden> wrote:
>> cpu_physical_memory_rw would bypass the TLB and so be much slower.
>> Make sure you use the functions which go via the TLB if you do
>> this in a helper (and remember that they will longjmp out on a
>> tlb miss!)
>
> At this point speed isn’t our main concern - it’s simplicity of 
> implementation - we want it to work, then we can worry about a better 
> implementation (which likely should not go this path at all - as discussed 
> above).
> Given that - isn’t it reasonable to pass through cpu_physical_memory_rw - and 
> hence not have to worry about the long jump ? Or am I missing something?

If you use cpu_physical_memory_rw you need to do the
virt-to-phys translation by hand (preferably via the TLB).
That might be something you needed to do anyway if we want
to have architecturally correct monitors that work on
physaddrs rather than vaddrs, but if not then the two
step process is a bit awkward.

>> Pretty sure we've already discussed how the current ldrex/strex
>> implementation is not architecturally correct. I think this is
>> another of those areas.
>
> We have indeed discussed this - but this is a surprise.

You're right that I didn't specifically realise this exact
part of our incorrectness earlier.

-- PMM