Re: [PATCH] Fix i386 HVF compilation failures

["Nguyen Dinh Phi [SG]" <phind.uet@gmail.com>]

On 28/11/25 9:02 pm, Philippe Mathieu-Daudé wrote:
> On 28/11/25 09:46, Nguyen Dinh Phi [SG] wrote:
>> On 26/11/25 9:39 pm, Philippe Mathieu-Daudé wrote:
>>> On 26/11/25 10:46, phind.uet@gmail.com wrote:
>>>> From: Nguyen Dinh Phi <phind.uet@gmail.com>
>>>>
>>>> Recent changes introduced build errors in the i386 HVF backend:
>>>>
>>>> - ../accel/hvf/hvf-accel-ops.c:163:17: error: no member named 
>>>> 'guest_debug_enabled' in 'struct AccelCPUState'
>>>>    163 |     cpu->accel->guest_debug_enabled = false;
>>>>
>>>> - ../accel/hvf/hvf-accel-ops.c:151:51
>>>>    error: no member named 'unblock_ipi_mask' in 'struct AccelCPUState'
>>>>
>>>> - ../target/i386/hvf/hvf.c:736:5
>>>>    error: use of undeclared identifier 'rip'
>>>>
>>>> - ../target/i386/hvf/hvf.c:737:5
>>>>    error: use of undeclared identifier 'env'
>>>>
>>>> This patch corrects the field usage and move identifier to correct
>>>> function ensuring successful compilation of the i386 HVF backend.
>>>>
>>>> These issues were caused by:
>>>>
>>>> Fixes: 2ad756383e1b (“accel/hvf: Restrict ARM-specific fields of 
>>>> AccelCPUState”)
>>>> Fixes: 2a21c9244740 (“target/i386/hvf: Factor hvf_handle_vmexit() out”)
>>>
>>> Oops.
>>>
>>>>
>>>> Signed-off-by: Nguyen Dinh Phi <phind.uet@gmail.com>
>>>> ---
>>>>   accel/hvf/hvf-accel-ops.c | 5 +++--
>>>>   target/i386/hvf/hvf.c     | 6 ++----
>>>>   2 files changed, 5 insertions(+), 6 deletions(-)
>>>>
>>>> diff --git a/accel/hvf/hvf-accel-ops.c b/accel/hvf/hvf-accel-ops.c
>>>> index 3e5feecd8a..e2cb8f202b 100644
>>>> --- a/accel/hvf/hvf-accel-ops.c
>>>> +++ b/accel/hvf/hvf-accel-ops.c
>>>> @@ -148,19 +148,20 @@ static int hvf_init_vcpu(CPUState *cpu)
>>>>       sigact.sa_handler = dummy_signal;
>>>>       sigaction(SIG_IPI, &sigact, NULL);
>>>> +#ifdef __aarch64__
>>>>       pthread_sigmask(SIG_BLOCK, NULL, &cpu->accel->unblock_ipi_mask);
>>>>       sigdelset(&cpu->accel->unblock_ipi_mask, SIG_IPI);
>>>> -#ifdef __aarch64__
>>>>       r = hv_vcpu_create(&cpu->accel->fd,
>>>>                          (hv_vcpu_exit_t **)&cpu->accel->exit, NULL);
>>>>   #else
>>>>       r = hv_vcpu_create(&cpu->accel->fd, HV_VCPU_DEFAULT);
>>>>   #endif
>>>>       assert_hvf_ok(r);
>>>> +#ifdef __aarch64__
>>>>       cpu->vcpu_dirty = true;
>>>
>>> Don't we want the ifdef *after* this line?
>> Oops, that was acutally that I meant to do, I think I made mistake 
>> when formatting the code. I will send a new version to move the #ifdef 
>> line. Somehow the vm still work normally on my intel Macbook.
> 
> Does that mean you tested this patch and it works for you? I posted v2:
> https://lore.kernel.org/qemu-devel/20251126134434.14355-1- 
> philmd@linaro.org/
Aw, sorry, I didn't aware of this.
> What do you plan to change with it?
Generally,it is them same as your post.
Yes, I tested my patch and it works fine, however, my original plan was to
make the change similar to what you posted. I tried your patch too, and it
works as expected, I think your patch does it better.

Thanks,
Phi