Re: [patch/rfa] fix call-dummies for hppa

[Randolph Chung <randolph at tausq dot org>]

> Which compilers?  I'm suspicious of GCC - it too often gets struct 
> parameters and return values internally consistent but wrong :-(

gcc only; I don't have access to the hp compilers. However, Dave 
(the hppa gcc maintainer) is quite careful about these things, so I 
think they are correct ;-)

> Be careful of white space change, this shouldn't be included.  If you 
> want to fix some indention just do it separatly.

ok, there were some stray tabs in the file so i was cleaning them up
along the way, but i'll remove that from this diff.

> (I've now got a copy of the 32-bit ABI but it doesn't help much)

this is the som runtime doc?  it's not particularly clear about small 
structs.....

> the comment doesn't match the assignment.
> 
> >+      /* The first parameter goes into sp-36, each stack slot is 4-bytes. 
> >*/
> >+      CORE_ADDR param_ptr = 32;

it does, actually, because the param_ptr is incremented by 4 for each
argument, so the first one goes to 36.

> >+	  else if (TYPE_CODE (type) == TYPE_CODE_FLT)
> >+            {
> 
> more comments (the rest is well commented), ``&& TYPE_LENGTH () == 4'' 
> test needed?
> 
> >+	      param_len = align_up (TYPE_LENGTH (type), 4);
> >+	      memcpy (param_val, VALUE_CONTENTS (arg), param_len);

yes, this bit is wrong. i found some more bugs in this function. will
send a new version with the whitespace changes removed and comments
added.

thanks,
randolph
-- 
Randolph Chung
Debian GNU/Linux Developer, hppa/ia64 ports
http://www.tausq.org/