cvs commit: fptools/ghc/rts Linker.c

[Sigbjorn Finne]

Further around-pokage in BFD & ld has convinced me that
something's amiss with how 'ld -r' is handled for the pe-i386
backend. The overflow handling code that does the right thing
is in there, but isn't getting called.

I'm hoping to get some feedback from the binutils mailing list
as to what might be the cause of this.

--sigbjorn

----- Original Message ----- 
From: "Julian Seward (Intl Vendor)" <v-julsew@microsoft.com>
To: "Sigbjorn Finne" <sof@glass.cse.ogi.edu>; <cvs-ghc@haskell.org>
Sent: Tuesday, January 29, 2002 02:26
Subject: RE: cvs commit: fptools/ghc/rts Linker.c


> 
> |   This is the Right Way to compute the relocation count, but
> |   unfortunately libbfd / GNU ld is generating bogus output
> |   when the reloc field overflows (causing objdump/nm etc. to
> |   crash when trying to read the generated output!) Looking into it.
> 
> Just as a comment, I've seen some incarnations of libbfd/ld emit
> a warning message on overflow, and then continue anyway.  And others
> which just produce wrong output but no warning.  
> 
> |   Once this has been cleared up/fixed, the splitting up of HSstd.o
> |   (and HSwin32.o) should be a thing of the past.
> 
> That would be cool.
> 
> J