Re: OCaml 3.10.2 FTBFS on arm with gcc 4.3
Stefano Zacchiroli wrote:
> ... and here we go with the first transition problem (quite early :( ).
> http://buildd.debian.org/fetch.cgi?pkg=ocaml;ver=3.10.2-2;arch=arm;stamp=1211163665
> Here is the error:
>
> boot/ocamlrun boot/ocamlc -nostdlib -I boot -warn-error A -I utils -I parsing -I typing -I bytecomp -I asmcomp -I driver -I toplevel -c parsing/parser.ml
> *** glibc detected *** boot/ocamlrun: free(): invalid next size (normal): 0x000801c8 ***
> make[1]: *** [parsing/parser.cmo] Aborted
>
> which looks like a badly handled pointer in ocamlrun, though it is
> strange to me that it happens only on arm.
>
> OCaml FTBFS on arm. As Stéphane noticed in the past it has been built
> properly on the very same arch (this is probably why we didn't notice
> this before now):
> http://experimental.debian.net/build.php?pkg=ocaml&arch=arm&ver=3.10.2-1,
> the main difference on the two builds are gcc-4.2 vs gcc-4.3.
Note that recently netwinder and non-netwinder arm machines did also
differ enough to FTBFS or not...
If hedges is a Netwinder, I'll probably take care of scheduling ocaml on
toffee which is a CATS...
> Note that, if all goes badly, we can't fix this by just not supporting
> ocamlopt on arm, as the problem is with ocamlrun (bytecode interpreter),
> so our only alternative to fix the bug would be not having at all ocaml
> on arm.
Lets first look at what made it FTBFS before thinking about the worst...
Btw, uploading Arch: all packages (build) depending on the new ocaml is
fine now...
Cheers
Luk
Reply to: