Re: [RFC] migration/dirtyrate: check malloc() return

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

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

> On 4/9/22 20:58, jianchunfu wrote:
>> Handling potential memory allocation failures in dirtyrate.
>> Signed-off-by: jianchunfu <jianchunfu@cmss.chinamobile.com>
>> ---
>>   migration/dirtyrate.c | 8 ++++++++
>>   1 file changed, 8 insertions(+)
>> diff --git a/migration/dirtyrate.c b/migration/dirtyrate.c
>> index aace12a787..5dd40f32c8 100644
>> --- a/migration/dirtyrate.c
>> +++ b/migration/dirtyrate.c
>> @@ -523,9 +523,17 @@ static void calculate_dirtyrate_dirty_ring(struct DirtyRateConfig config)
>>       }
>>         dirty_pages = malloc(sizeof(*dirty_pages) * nvcpu);
>> +    if (!dirty_pages) {
>> +        error_report("malloc dirty pages for vcpus failed.");
>> +        exit(1);
>> +    }
>>         DirtyStat.dirty_ring.nvcpu = nvcpu;
>>       DirtyStat.dirty_ring.rates = malloc(sizeof(DirtyRateVcpu) * nvcpu);
>> +    if (!DirtyStat.dirty_ring.rates) {
>> +        error_report("malloc dirty rates for vcpu ring failed.");
>> +        exit(1);
>> +    }
>
> You might as well use g_new(), which handles the sizeof and
> multiplication, and error reporting.

It will also assert if the alloc fails. If this is an allocation QEMU
can recover from then you need to use the try_new variants of the
g_malloc/new functions. However here we are exiting so no actual check
is needed as the g_malloc will exit for us.

-- 
Alex Bennée