Re: gdb

Eric Youngdale (eric@aib.com)
Wed, 8 May 1996 13:19:09 -0400


>Umm, while I agree that it might make it a lot easier to debug things on
the
>alpha, I hope nobody writes a patch like that for gdb. The reason I hope
>that is that I hope that it will be unnecessary. The _correct_ way to
handle >it is to never generate those d*mn instructions in the first place.
We need >an optimizing linker that gets rid of them and makes it into a
single "bsr"
>instruction instead.

This wouldn't be a hard optimization, and in fact it would make
shared libraries a bit easier for those cases where you can safely call
without going through a PLT (i.e. we are not using up a valuable .got
slot for a pointer). It would tend to be used for things like static
functions that we would want to do this in shared libs. For normal
executables, it is more just a question of whether the branch is in range or
not.

Remind me about this in about 4 months :-). I think the actual
code for this would be relatively simple. Note that we would leave behind
a NOP, since it would be a lot harder for the linker to compress this
out.

If someone else wants to attempt this they can, of course, and I
can give pointers as to what would be needed.

-Eric

-- 
"The woods are lovely, dark and deep.  But I have promises to keep,
And lines to code before I sleep, And lines to code before I sleep."