Re: [PATCH for-8.0 v2 1/3] async: Suppress GCC13 false positive in aio_bh_poll()

["Philippe Mathieu-Daudé" <philmd@linaro.org>]

On 22/3/23 14:27, Stefan Hajnoczi wrote:
> On Wed, Mar 22, 2023 at 08:11:37AM +0100, Thomas Huth wrote:
>> On 21/03/2023 17.16, Cédric Le Goater wrote:
>>> From: Cédric Le Goater <clg@redhat.com>
>>>
>>> GCC13 reports an error :
>>>
>>> ../util/async.c: In function ‘aio_bh_poll’:
>>> include/qemu/queue.h:303:22: error: storing the address of local variable ‘slice’ in ‘*ctx.bh_slice_list.sqh_last’ [-Werror=dangling-pointer=]
>>>     303 |     (head)->sqh_last = &(elm)->field.sqe_next;                          \
>>>         |     ~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~
>>> ../util/async.c:169:5: note: in expansion of macro ‘QSIMPLEQ_INSERT_TAIL’
>>>     169 |     QSIMPLEQ_INSERT_TAIL(&ctx->bh_slice_list, &slice, next);
>>>         |     ^~~~~~~~~~~~~~~~~~~~
>>> ../util/async.c:161:17: note: ‘slice’ declared here
>>>     161 |     BHListSlice slice;
>>>         |                 ^~~~~
>>> ../util/async.c:161:17: note: ‘ctx’ declared here
>>>
>>> But the local variable 'slice' is removed from the global context list
>>> in following loop of the same routine. Add a pragma to silent GCC.
>>>
>>> Cc: Stefan Hajnoczi <stefanha@redhat.com>
>>> Cc: Paolo Bonzini <pbonzini@redhat.com>
>>> Cc: Daniel P. Berrangé <berrange@redhat.com>
>>> Signed-off-by: Cédric Le Goater <clg@redhat.com>
>>> ---
>>>    util/async.c | 13 +++++++++++++
>>>    1 file changed, 13 insertions(+)
>>>
>>> diff --git a/util/async.c b/util/async.c
>>> index 21016a1ac7..de9b431236 100644
>>> --- a/util/async.c
>>> +++ b/util/async.c
>>> @@ -164,7 +164,20 @@ int aio_bh_poll(AioContext *ctx)
>>>        /* Synchronizes with QSLIST_INSERT_HEAD_ATOMIC in aio_bh_enqueue().  */
>>>        QSLIST_MOVE_ATOMIC(&slice.bh_list, &ctx->bh_list);
>>> +
>>> +    /*
>>> +     * GCC13 [-Werror=dangling-pointer=] complains that the local variable
>>> +     * 'slice' is being stored in the global 'ctx->bh_slice_list' but the
>>> +     * list is emptied before this function returns.
>>> +     */
>>> +#if !defined(__clang__)
>>> +#pragma GCC diagnostic push
>>> +#pragma GCC diagnostic ignored "-Wdangling-pointer="
>>
>> That warning parameter looks like a new one in GCC 13 ?
>> ... so you have to check whether it's available before disabling
>> it, otherwise this will fail with older versions of GCC. I just
>> gave it a try with my GCC 8.5 and got this:
>>
>> ../../devel/qemu/util/async.c: In function ‘aio_bh_poll’:
>> ../../devel/qemu/util/async.c:175:32: error: unknown option after ‘#pragma GCC diagnostic’ kind [-Werror=pragmas]
>>   #pragma GCC diagnostic ignored "-Wdangling-pointer="
>>                                  ^~~~~~~~~~~~~~~~~~~~~
>> cc1: all warnings being treated as errors
>>
>>   Thomas
>>
>> What about reworking the code like this:
>>
>> diff --git a/util/async.c b/util/async.c
>> index 21016a1ac7..b236bdfbd8 100644
>> --- a/util/async.c
>> +++ b/util/async.c
>> @@ -156,15 +156,14 @@ void aio_bh_call(QEMUBH *bh)
>>   }
>>   /* Multiple occurrences of aio_bh_poll cannot be called concurrently. */
>> -int aio_bh_poll(AioContext *ctx)
>> +static int aio_bh_poll_slice(AioContext *ctx, BHListSlice *slice)
>>   {
>> -    BHListSlice slice;
>>       BHListSlice *s;
>>       int ret = 0;
>>       /* Synchronizes with QSLIST_INSERT_HEAD_ATOMIC in aio_bh_enqueue().  */
>> -    QSLIST_MOVE_ATOMIC(&slice.bh_list, &ctx->bh_list);
>> -    QSIMPLEQ_INSERT_TAIL(&ctx->bh_slice_list, &slice, next);
>> +    QSLIST_MOVE_ATOMIC(&slice->bh_list, &ctx->bh_list);
>> +    QSIMPLEQ_INSERT_TAIL(&ctx->bh_slice_list, slice, next);
>>       while ((s = QSIMPLEQ_FIRST(&ctx->bh_slice_list))) {
>>           QEMUBH *bh;
>> @@ -191,6 +190,13 @@ int aio_bh_poll(AioContext *ctx)
>>       return ret;
>>   }
>> +int aio_bh_poll(AioContext *ctx)
>> +{
>> +    BHListSlice slice;
>> +
>> +    return aio_bh_poll_slice(ctx, &slice);
>> +}
>> +
>>   void qemu_bh_schedule_idle(QEMUBH *bh)
>>   {
>>       aio_bh_enqueue(bh, BH_SCHEDULED | BH_IDLE);
>>
>> Would that work with GCC 13 and be acceptable?
> 
> Fine by me. Please add a comment into aio_bh_poll() explaining that this
> wrapper function exists to silence the gcc dangling-pointer warning.
> Otherwise someone may be tempted to remove the function.

IMO by using #pragmas it is clearer this is a kludge. Also we can
revert this commit adding the pragmas/comment once the compiler
are fixed.