cross-endian GHC generates invalid .hi files
I've reproduced the bug in the following setup:
./configure \
--enable-unregisterised \
--host=x86_64-pc-linux-gnu \
--target=m68k-unknown-linux-gnu
m68k is a 32-bit big-endian machine.
$ LD_BIND_NOW=1 inplace/bin/ghc-stage2 --show-iface libraries/base/dist-install/build/System/IO.hi
Magic: Wanted 129742,
got 3472490752
magic number mismatch: old/corrupt interface file? (wanted 129742, got 3472490752)
$ printf "%08X - %08X\n" 129742 3472490752
0001FACE - CEFA0100
Reverting changeset:fbcef83a restores .hi loading and ghc-stage2.
I'm still not sure what exactly breaks.
WORDS_BIGENDIAN looks suspicious.
I guess it breaks ghc-stage1 endianness understanding.
Trac metadata
Trac field | Value |
---|---|
Version | 8.1 |
Type | Bug |
TypeOfFailure | OtherFailure |
Priority | normal |
Resolution | Unresolved |
Component | Compiler |
Test case | |
Differential revisions | |
BlockedBy | |
Related | |
Blocking | |
CC | |
Operating system | |
Architecture |