This is the mail archive of the binutils@sources.redhat.com mailing list for the binutils project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] MIPS gas: Fix macro expansions for .set noat


On Thu, 17 Feb 2005, Thiemo Seufer wrote:

> Ok, this patch converts $at uses in .set noat from warnings to errors.

 How about extracting these bits of test cases that used to be expected to 
trigger warnings into separate cases to test if errors are generated 
correctly?

> I haven't changed the .set nomacro behaviour yet, partially because
> this patch is already big enough, and partially because nomacro isn't
> fully equivalent to noat. Nomacro probably shouldn't fail if the number
> of instructions stays the same.

 That's the whole point IMHO.  A macro that expands into a single 
instruction can be seen as a mere alias and it doesn't violate code length 
or delay slot restrictions.  That is I think it should be allowed, for 
readability if nothing else, to call an instruction "li $2,1" instead of 
"addiu $2,$0,1" even with ".set nomacro" in effect.

  Maciej


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]