Ian Lynagh wrote:
cc1: warnings being treated as errors
LibFFI.hsc: In function 'main':
LibFFI.hsc:81: warning: format '%d' expects type 'int', but argument 3
has type 'long unsigned int'
By x86_64, do you mean x86_64/Linux? If so, it works for me.
Are all your repos up-to-date - in particular, your hsc2hs repo?
("./darcs-all pull" or "./darcs-all pull -a" to update them). The
hsc_type macro in template-hsc.h should have "%s%lu" as the format
string, not "%s%d".
Yeah, my x86_64 repo wasn't completely up to date, sorry for the hassle.
The PPC build is actually broken though.
On Tue, May 26, 2009 at 09:28:18PM +1000, Ben Lippmeier wrote:
src/powerpc/ffi_darwin.c:454: warning: passing argument 6 of
'ffi_call_AIX' from incompatible pointer type
This looks like a libffi bug to me, so I'm CCing
[email protected]
The 6th argument of ffi_call_AIX and ffi_call_DARWIN is given as
void (*fn2)(void)
in the prototype, but they are being passed ffi_prep_args, which has
type
void ffi_prep_args(extended_cif *ecif, unsigned *const stack)
The latest release (3.0.8) looks like it has the same problem.
When I hack around this problem by casting ffi_prep_args to ( void
(*)(void) ) the build compiles all the source files but fails to link
with the following message:
ld warning: atom sorting error for
_ghczm6zi11zi20090527_LibFFI_Czuffizucif_closure_tbl and
_ghczm6zi11zi20090527_LibFFI_Czuffizutype_closure_tbl in
compiler/stage2/build/LibFFI.o
ld warning: atom sorting error for
_ghczm6zi11zi20090527_LibFFI_Czuffizucif_closure_tbl and
_ghczm6zi11zi20090527_LibFFI_Czuffizutype_closure_tbl in
compiler/stage2/build/LibFFI.o
ld: scattered reloc r_address too large for inferred architecture ppc
I don't fully understand what the linker error message means, but the
build bot logs say it's been happening since the 4th of April, so I'm
pretty sure it's not due to the casts I added.
Thorkil's tnaur PPC OSX head2 build was failing with that message from
the 4th April until the 27th. At that point it started failing with
"unknown language flags -XCPP -XExistentialQuantification ... " instead.
Thorkil's got it setup to compile with 6.6.1.
I tried building it with a fresh copy of 6.10.3, but it wen't back to
failing with the above linker error.
Seems to be related to tickets #1958 and #1843
Ben.
_______________________________________________
Cvs-ghc mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/cvs-ghc
