Re: [PATCH 3/3] s390x/cpu: Use timer_free() in the finalize function to avoid memleaks

[David Hildenbrand]

On 04.12.20 09:36, Cornelia Huck wrote:
> On Fri, 4 Dec 2020 16:12:09 +0800
> Gan Qixin <ganqixin@huawei.com> wrote:
> 
>> When running device-introspect-test, a memory leak occurred in the 
>> s390_cpu_initfn
>> function, this patch use timer_free() in the finalize function to fix it.
>>
>> ASAN shows memory leak stack:
>>
>> Direct leak of 3552 byte(s) in 74 object(s) allocated from:
>>     #0 0xfffeb3d4e1f0 in __interceptor_calloc (/lib64/libasan.so.5+0xee1f0)
>>     #1 0xfffeb36e6800 in g_malloc0 (/lib64/libglib-2.0.so.0+0x56800)
>>     #2 0xaaad51a8f9c4 in timer_new_full qemu/include/qemu/timer.h:523
>>     #3 0xaaad51a8f9c4 in timer_new qemu/include/qemu/timer.h:544
>>     #4 0xaaad51a8f9c4 in timer_new_ns qemu/include/qemu/timer.h:562
>>     #5 0xaaad51a8f9c4 in s390_cpu_initfn qemu/target/s390x/cpu.c:304
>>     #6 0xaaad51e00f58 in object_init_with_type qemu/qom/object.c:371
>>     #7 0xaaad51e0406c in object_initialize_with_type qemu/qom/object.c:515
>>     #8 0xaaad51e042e0 in object_new_with_type qemu/qom/object.c:729
>>     #9 0xaaad51e3ff40 in qmp_device_list_properties 
>> qemu/qom/qom-qmp-cmds.c:153
>>     #10 0xaaad51910518 in qdev_device_help qemu/softmmu/qdev-monitor.c:283
>>     #11 0xaaad51911918 in qmp_device_add qemu/softmmu/qdev-monitor.c:801
>>     #12 0xaaad51911e48 in hmp_device_add qemu/softmmu/qdev-monitor.c:916
>>
>> Reported-by: Euler Robot <euler.robot@huawei.com>
>> Signed-off-by: Gan Qixin <ganqixin@huawei.com>
>> ---
>> Cc: Thomas Huth <thuth@redhat.com>
> 
> [Adding missing maintainers. How did you build the cc: list?]
> 
>> ---
>>  target/s390x/cpu.c | 5 +++++
>>  1 file changed, 5 insertions(+)
>>
>> diff --git a/target/s390x/cpu.c b/target/s390x/cpu.c
>> index 7b66718c44..8a734c2f8c 100644
>> --- a/target/s390x/cpu.c
>> +++ b/target/s390x/cpu.c
>> @@ -313,6 +313,11 @@ static void s390_cpu_finalize(Object *obj)
>>  #if !defined(CONFIG_USER_ONLY)
>>      S390CPU *cpu = S390_CPU(obj);
>>  
>> +    timer_del(cpu->env.tod_timer);
>> +    timer_free(cpu->env.tod_timer);
>> +    timer_del(cpu->env.cpu_timer);
>> +    timer_free(cpu->env.cpu_timer);
>> +
>>      qemu_unregister_reset(s390_cpu_machine_reset_cb, cpu);
>>      g_free(cpu->irqstate);
>>  #endif
> 
> Looks sane at first glance.
> 

Could have sworn we had these in the code at one point - but I don't
find anything in the git history.

Reviewed-by: David Hildenbrand <david@redhat.com>

-- 
Thanks,

David / dhildenb