On Mon, Jan 21, 2013 at 5:11 PM, Markus Armbruster <[email protected]> wrote:
> Blue Swirl <[email protected]> writes:
>
>> Recent Clang compilers have preliminary support for finding
>> unannotated fallthrough cases in switch statements with
>> compiler flag -Wimplicit-fallthrough. The support is incomplete,
>> it's only possible to annotate the case in C++ but not in C, so it
>> wouldn't be useful to enable the flag for QEMU yet.
>>
>> Mark cases which don't have a comment about fall through with
>> a comment. In legitimate fall through cases the comment can be
>> edited later to mark the case for future readers.
>
> Let's clean this up properly instead, as far as we can. Details inline.
> Maintainers, please check out the parts that apply to your code.
>
>> Signed-off-by: Blue Swirl <[email protected]>
[...]
>> diff --git a/target-xtensa/op_helper.c b/target-xtensa/op_helper.c
>> index 3813a72..d829702 100644
>> --- a/target-xtensa/op_helper.c
>> +++ b/target-xtensa/op_helper.c
>> @@ -443,8 +443,10 @@ void HELPER(check_atomctl)(CPUXtensaState *env,
>> uint32_t pc, uint32_t vaddr)
>> switch (access & PAGE_CACHE_MASK) {
>> case PAGE_CACHE_WB:
>> atomctl >>= 2;
>> + /* XXX: questionable fallthrough */
>> case PAGE_CACHE_WT:
>> atomctl >>= 2;
>> + /* XXX: questionable fallthrough */
>> case PAGE_CACHE_BYPASS:
>> if ((atomctl & 0x3) == 0) {
>> HELPER(exception_cause_vaddr)(env, pc,
>
> Looks intentional. Max Filippov?
Correct, these are intentional.
--
Thanks.
-- Max
