Re: [PATCH] NFS: remove trailing semicolon in macro definition

[Tom Rix]

On 11/29/20 8:50 AM, Trond Myklebust wrote:
> On Sun, 2020-11-29 at 16:42 +0000, Trond Myklebust wrote:
>> Hi Tom,
>>
>> On Fri, 2020-11-27 at 11:43 -0800, trix@xxxxxxxxxx wrote:
>>> From: Tom Rix <trix@xxxxxxxxxx>
>>>
>>> The macro use will already have a semicolon.
>>>
>>> Signed-off-by: Tom Rix <trix@xxxxxxxxxx>
>>> ---
>>>  net/sunrpc/auth_gss/gss_generic_token.c | 2 +-
>>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/net/sunrpc/auth_gss/gss_generic_token.c
>>> b/net/sunrpc/auth_gss/gss_generic_token.c
>>> index fe97f3106536..9ae22d797390 100644
>>> --- a/net/sunrpc/auth_gss/gss_generic_token.c
>>> +++ b/net/sunrpc/auth_gss/gss_generic_token.c
>>> @@ -46,7 +46,7 @@
>>>  /* TWRITE_STR from gssapiP_generic.h */
>>>  #define TWRITE_STR(ptr, str, len) \
>>>         memcpy((ptr), (char *) (str), (len)); \
>>> -       (ptr) += (len);
>>> +       (ptr) += (len)
>>>  
>>>  /* XXXX this code currently makes the assumption that a mech oid
>>> will
>>>     never be longer than 127 bytes.  This assumption is not
>>> inherent
>>> in
>> There is exactly 1 use of this macro in the code AFAICS. Can we
>> please
>> just get rid of it, and make the code trivially easier to read?
>>
>
> BTW: To illustrate just how obfuscating this kind of macro can be, note
> that the line you are changing above will be completely optimised away
> in the 1 use case we're talking about. It is bumping a pointer value
> that immediately gets discarded.

Yes, I agree.

I was wondering about expanding treewide, all the single shot macros defined/used in c files.

other fixers that cleanup unused variables would remove the unneeded expansions like this one.