[OpenAFS-devel] Re: Problem with IP address based ACL?

[Derrick J Brashear]

On Thu, 29 Jun 2006, Derrick J Brashear wrote:

> On Wed, 21 Jun 2006, Joe Buehler wrote:
>
>> Jeffrey Altman wrote:
>> 
>>> ubik_Call() does provide a prototype because it is used as a
>>> pseudo-vararg function.
>>> 
>>> afs_int32
>>> ubik_Call(int (*aproc) (), struct ubik_client *aclient,
>>>      afs_int32 aflags, long p1, long p2, long p3, long p4,
>>>      long p5, long p6, long p7, long p8, long p9,
>>>      long p10, long p11, long p12, long p13, long p14,
>>>      long p15, long p16)
>>> 
>>> But what happens when the parameters that are being passed into
>>> ubik_Call for later use by 'aproc' do not have the same size as
>>> a 'long'?  In that case the alignment of parameters on the stack
>> 
>> This is perilous for other reasons.  On AIX, for example, my recollection
>> is that the first 8 parameters of a given type (float or int) are passed
>> in registers -- and floats and ints go into different registers.  If there
>> are floats involved the above method of doing varargs may fail badly 
>> because
>> the registers will get all confused.
>> 
>> I would say that the faster this sort of thing gets removed, the better --
>> it is based on ABI assumptions that are not valid for all machines.
>
> I have a patch to kill it in testing now.

/afs/andrew.cmu.edu/usr/shadow/ubik-all.diff-2

Look at it, try it, etc, etc. It doesn't kill every use of ubik_Call, yet, 
but it's good progress in that direction.