Re: [PATCH] block: Return the real error code in bdrv_getlength

[Eric Blake]

On 11/4/20 11:41 PM, Tuguoyi wrote:
> Sorry, please ignore this patch, it's not a right fix

What's not right about it?


>> +++ b/block.c
>> @@ -5082,6 +5082,10 @@ int64_t bdrv_getlength(BlockDriverState *bs)
>>  {
>>      int64_t ret = bdrv_nb_sectors(bs);
>>
>> +    if (ret < 0) {
>> +        return ret;
>> +    }
>> +
>>      ret = ret > INT64_MAX / BDRV_SECTOR_SIZE ? -EFBIG : ret;

You are correct that the old code was broken (since
INT64_MAX/BDRV_SECTOR_SIZE is unsigned, any negative ret is likewise
promoted to unsigned and we mistakenly return -EFBIG instead of the
original error).  But your new code works just fine: because we
guarantee with the early return above that ret is a positive value, it
doesn't matter that the rest of this ?: is performed in unsigned math.

>>      return ret < 0 ? ret : ret * BDRV_SECTOR_SIZE;

Of course, having two ?: in a row is odd; we could instead write:

if (ret < 0){
    return ret;
}
if (ret > INT64_MAX / BDRV_SECTOR_SIZE) {
    return -EFBIG;
}
return ret * BDRV_SECTOR_SIZE;

for the same result.

-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3226
Virtualization:  qemu.org | libvirt.org