qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [PATCH 0/8] tcg: tidy the type of code_ptr


From: Peter Maydell
Subject: Re: [Qemu-devel] [PATCH 0/8] tcg: tidy the type of code_ptr
Date: Sat, 29 Mar 2014 20:26:57 +0000

On 29 March 2014 00:27, Richard Henderson <address@hidden> wrote:
> Here's where I think we should go with the cleanup that Peter started.
>
> I've only converted a couple of the backends as examples.  It's not 100%
> mechanical, since one has to be aware of the change in the semantics of
> pointer arithmetic (e.g. s->code_ptr - s->code_buf).
>
> Taking Sparc as an example, the before code size of tcg.o is 0xad60, and
> the after code size is 0xa200.  A savings of a bit less than 3k.
>
> Thoughts?

Looks good to me -- I like the way a lot of the casts go away.
It seems like it might be worth abstracting out "give me the byte
difference between these two code pointers" rather than having
inline (uintptr_t)codeptr1 - (uintptr_t)codeptr2, but I dunno.

Is there a better name than 'tcg_itype' ? Putting 'type' in the
name of a type is a bit redundant, and suggests it contains
a type rather than an insn.

thanks
-- PMM



reply via email to

[Prev in Thread] Current Thread [Next in Thread]