On 2017-11-14 16:02, Alberto Garcia wrote:
> On Fri 10 Nov 2017 09:31:10 PM CET, Max Reitz wrote:
>> +static int64_t get_refblock_offset(BlockDriverState *bs, uint64_t offset)
>> +{
>> + BDRVQcow2State *s = bs->opaque;
>> + uint32_t index = offset_to_reftable_index(s, offset);
>> + int64_t covering_refblock_offset = 0;
>> +
>> + if (index < s->refcount_table_size) {
>> + covering_refblock_offset = s->refcount_table[index] &
>> REFT_OFFSET_MASK;
>> + }
>> + if (!covering_refblock_offset) {
>> + qcow2_signal_corruption(bs, true, -1, -1, "Refblock at %#" PRIx64 "
>> is "
>> + "not covered by the refcount structures",
>> + offset);
>> + return -EIO;
>> + }
>> +
>> + return covering_refblock_offset;
>> +}
>
> Isn't it simpler to do something like this instead?
>
> if (index >= s->refcount_table_size) {
> qcow2_signal_corruption(...);
> return -EIO;
> }
> return s->refcount_table[index] & REFT_OFFSET_MASK;But that doesn't cover the case were s->refcount_table[index] & REFT_OFFSET_MASK is 0. Max
signature.asc
Description: OpenPGP digital signature
