Here comes a brief summary of a discussion I had with Denys -CS
> > There are two different ways how the number n in alignment
> > specifications are treated: one represents the number as is
> (so align
> > to n*wordsize), the other (more common form) is that it
> means align to
> > 2^n. The test tries to find out which effect the align directive
> > has...
>
> Aaahh! That's what this does :-)
Wasn't that absolutely obvious? ;-)
> > But you are clear that this is only valid if you are using gas?
> No, I had no clue. So we need to check that the assembler IS gas?
No, why? If it is not gas the test will fail anyway. If it fails, there
is a fallback (should at least).
> > And the only effect it has is that the mapping of instructions for
> > threaded code to opcode numbers is fast (not tha important).
>
> Briefly, how?
The idea is quite simple, the instruction in threaded-code is an address
to the instruction's code. Now some information such as the opcode
number (and some other stuff I luckily forgot about but also indexed by
instruction) is stored at a fixed offset from that address before the
actually code starts. That's also why I need to enforce alignment, so
that I actually know how much the data is stored away from the
instructions actual code start. The data is then emitted by properly
aligned assembly code.
Pretty bizarre, eh? Just sketching it makes me feel ashamed of myself...
"Denys Duchier" <duchier@ps.uni-sb.de> wrote in message
news:<86vg28a85o.fsf@speedy.ps.uni-sb.de>...
> Hi Christian,
>
> In platform/emulator/configure.in there is one test that seems to have
> been written by you: "Find out which alignment assembler uses".
>
> I am unable to understand this test. It seems to always fail and on
> hppa I even get an ICE. Can you explain this test a bit? Is there
> another way to implement it?
>
> Thanks
>
> --Denys
> -
> Please send submissions to hackers@mozart-oz.org
> and administriva mail to hackers-request@mozart-oz.org.
> The Mozart Oz web site is at http://www.mozart-oz.org/.
-
Please send submissions to hackers@mozart-oz.org
and administriva mail to hackers-request@mozart-oz.org.
The Mozart Oz web site is at http://www.mozart-oz.org/.