[Caml-list] Ocaml and 64 bit address spaces?

[Caml-list] Ocaml and 64 bit address spaces?

[Niall Dalton]

Hello,

I haven't had a chance to check, but does the Ocaml compiler
allow one to use 64 bit address spaces on 64 bit machines? (modulo
the requisite OS support etc., limits with address lines and so on)
I'm interested in particular 64-bit ppc and x86-64 based machines.
If it does, I'd also be interested in any gotchas that anyone
has run across when working with very large heap sizes (several
GB).

Thanks,
niall

-------------------
To unsubscribe, mail caml-list-request@inria.fr Archives: http://caml.inria.fr
Bug reports: http://caml.inria.fr/bin/caml-bugs FAQ: http://caml.inria.fr/FAQ/
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners

Re: [Caml-list] Ocaml and 64 bit address spaces?

[Xavier Leroy]

> I haven't had a chance to check, but does the Ocaml compiler
> allow one to use 64 bit address spaces on 64 bit machines? (modulo
> the requisite OS support etc., limits with address lines and so on)

Yes, on Alpha and Itanium machines.

On Sparc, Mips and PPC, it should be possible to compile the bytecode
interpreter in 64-bit mode, but the native-code compiler still
generates 32-bit code.  That might change if there is enough demand,
e.g. for the new G5 machines.

The x86-64 port of OCaml will be fully 64-bit once it is completed,
that is, shortly after I get access to one of these machines.  
However, given the lack of "testdrive" programs for the x86-64, I have
no idea when I'll get a shell account on one of these toys.

- Xavier Leroy

-------------------
To unsubscribe, mail caml-list-request@inria.fr Archives: http://caml.inria.fr
Bug reports: http://caml.inria.fr/bin/caml-bugs FAQ: http://caml.inria.fr/FAQ/
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners

Re: [Caml-list] Ocaml and 64 bit address spaces?

[Niall Dalton]

On 25/6/03 4:32 PM, "Xavier Leroy" <xavier.leroy@inria.fr> wrote:

> Yes, on Alpha and Itanium machines.
> 
> On Sparc, Mips and PPC, it should be possible to compile the bytecode
> interpreter in 64-bit mode, but the native-code compiler still
> generates 32-bit code.  That might change if there is enough demand,
> e.g. for the new G5 machines.

Afaik, even though you can use some 64-bit instructions on the G5,
the OS (including "Panther", the next release) does not support 64 bit
address spaces (yet).

Thanks for the helpful response!

niall

-------------------
To unsubscribe, mail caml-list-request@inria.fr Archives: http://caml.inria.fr
Bug reports: http://caml.inria.fr/bin/caml-bugs FAQ: http://caml.inria.fr/FAQ/
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners

[Caml-list] Ocaml program crashes on Opteron

[Aleksey Nogin]

I've been trying to use OCaml 3.07 on an Opteron machine running the 
latest Fedora Core development (AKA Red Hat Rawhide) with a binary 
kernel from Mandrake 9.2beta3. I've successfully compiled Ocaml 3.07, 
but when I compile our "omake" system with it, omake segfaults in a 
reproducible fashion (it reliably works for some things and reliably 
crashes for others).

After compiling omake with "ocamlopt.opt -inline 0 -ccopt -g -cc gcc32 
-S -verbose", and then re-running the last linking step with .s files 
instead of .o files (to get debugging information tied to the assembly 
files), I get from gdb:

> ...
> Program received signal SIGSEGV, Segmentation fault.
> Omake_wild__wild_subst_429 () at omake_wild.s:419
> 419             movzbq  -8(%rbx), %rdi
> Current language:  auto; currently asm
> (gdb) bt
> #0  Omake_wild__wild_subst_429 () at omake_wild.s:419
> #1  0x0000002a9557ae00 in ?? ()
> #2  0x000000000044fa02 in List__map_86 ()
> #3  0x0000002a95c9b698 in ?? ()
> #4  0x0000002a9557a508 in ?? ()
> #5  0x0000002a95cdc780 in ?? ()
> #6  0x000000000041df9a in Omake_env__collect_3575 () at omake_env.s:1780
> ...

The relevant part of omake_wild.s is:
> ...
>    .text
>    .align   16
>    .globl   Omake_wild__wild_subst_429
> Omake_wild__wild_subst_429:
>    subq  $8, %rsp
> .L145:
>    movzbq   -8(%rbx), %rdi    <<< it crashes here <<<
>    testq %rdi, %rdi
>    je .L144
> ...

(note that there are no references to L145 in that file) and the Ocaml
code for Omake_wild.wild_subst is:

> let wild_subst (len, s) wild =
>    match wild with
>       WildLiteral node ->
>          WildNode node
>     | WildSplit (plen, prefix, slen, suffix) ->
>          WildString (prefix ^ s ^ suffix)
> 

and it is being called from

>                match search patterns with
>                   Some subst ->
>                      let source_args = List.map (wild_subst subst) sources in

Some additional information (full back trace, register contents, etc) is
available at http://cvs.metaprl.org:12000/bugzilla/show_bug.cgi?id=111

Any guesses as to what might be causing this? Any hints on what I could 
do to invertigate this further?

-- 
Aleksey Nogin

Home Page: http://nogin.org/
E-Mail: nogin@cs.caltech.edu (office), aleksey@nogin.org (personal)
Office: Jorgensen 70, tel: (626) 395-2907

-------------------
To unsubscribe, mail caml-list-request@inria.fr Archives: http://caml.inria.fr
Bug reports: http://caml.inria.fr/bin/caml-bugs FAQ: http://caml.inria.fr/FAQ/
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners