Re: [PATCH 2/6] contrib/plugins/cache: fix warning when compiling on 32bits host

["Alex Bennée" <alex.bennee@linaro.org>]

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

> On 8/15/24 21:46, Alex Bennée wrote:
>> Thomas Huth <thuth@redhat.com> writes:
>> 
>>> On 15/08/2024 01.36, Pierrick Bouvier wrote:
>>>> Found on debian stable (i386).
>>>> ../contrib/plugins/cache.c: In function 'vcpu_tb_trans':
>>>> ../contrib/plugins/cache.c:477:30: error: cast from pointer to integer of different size [-Werror=pointer-to-int-cast]
>>>>     477 |             effective_addr = (uint64_t) qemu_plugin_insn_haddr(insn);
>>>>         |
>>>> Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
>>>> ---
>>>>    contrib/plugins/cache.c | 6 +++---
>>>>    1 file changed, 3 insertions(+), 3 deletions(-)
>>>> diff --git a/contrib/plugins/cache.c b/contrib/plugins/cache.c
>>>> index 512ef6776b7..82ed734d6d4 100644
>>>> --- a/contrib/plugins/cache.c
>>>> +++ b/contrib/plugins/cache.c
>>>> @@ -471,12 +471,12 @@ static void vcpu_tb_trans(qemu_plugin_id_t id, struct qemu_plugin_tb *tb)
>>>>        n_insns = qemu_plugin_tb_n_insns(tb);
>>>>        for (i = 0; i < n_insns; i++) {
>>>>            struct qemu_plugin_insn *insn = qemu_plugin_tb_get_insn(tb, i);
>>>> -        uint64_t effective_addr;
>>>> +        uintptr_t effective_addr;
>>>>              if (sys) {
>>>> -            effective_addr = (uint64_t) qemu_plugin_insn_haddr(insn);
>>>> +            effective_addr = (uintptr_t) qemu_plugin_insn_haddr(insn);
>>>>            } else {
>>>> -            effective_addr = (uint64_t) qemu_plugin_insn_vaddr(insn);
>>>> +            effective_addr = (uintptr_t)
>>>> qemu_plugin_insn_vaddr(insn);
>>>>            }
>>>
>>> Is this the right fix? I assume effective_addr stores an address of
>>> the guest, so if the guest is 64-bit and the host is 32-bit, you now
>>> lose the upper bits of the address...?
>> I think the problem is higher up, it was a mistake to have:
>>    void *qemu_plugin_insn_haddr(const struct qemu_plugin_insn
>> *insn);
>> return *void.
>
> No, not a bug.  This is a host addr, right there in the name.
> Returning uint64_t would be a bug.

No it's:

 * Returns: hardware (physical) target address of instruction

I was kinda assuming that was what the underlying host_addr[] fields in
DisasContextDB are. Are we just saying its QEMU's vaddr of where the
guest physical address is mapped into QEMU?

>
>
> r~

-- 
Alex Bennée
Virtualisation Tech Lead @ Linaro