Re: [PATCH v2 04/93] tcg: Manage splitwx in tc_ptr_to_region_tree by hand

[Alex Bennée]

Richard Henderson <richard.henderson@linaro.org> writes:

> On 2/4/21 5:01 AM, Alex Bennée wrote:
>> 
>> Richard Henderson <richard.henderson@linaro.org> writes:
>> 
>>> The use in tcg_tb_lookup is given a random pc that comes from the pc
>>> of a signal handler.  Do not assert that the pointer is already within
>>> the code gen buffer at all, much less the writable mirror of it.
>> 
>> Surely we are asserting that - or at least you can find a rt entry for
>> the pointer passed (which we always expect to work)?
>
> What?  No.  The pointer could be anything at all, depending on any other bug
> within qemu.

But you do assert it:

     struct tcg_region_tree *rt = tc_ptr_to_region_tree(tb->tc.ptr);
 
     g_assert(rt != NULL);

and rt is only NULL when it's !in_code_gen_buffer(p).

In tcg_tb_lookup you haven't removed an assert - you just ensure you
don't fail if it's not.

>
>
> r~


-- 
Alex Bennée