From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ian Lance Taylor To: snowball3@usa.net Cc: gas2@cygnus.com Subject: Re: COFF weak symbol report... Date: Mon, 05 Oct 1998 08:00:00 -0000 Message-id: <199810051500.LAA22413@subrogation.cygnus.com> References: <199810051341.NAA16922@out4.ibm.net> X-SW-Source: 1998/msg00236.html From: "Mark E." Date: Mon, 5 Oct 1998 09:41:22 +0000 I've made some progress getting weak symbol support into COFF. First, I'd like to ask where do I send my diff patches and which format is preffered? You can send patches directly to me. You can also CC the list if you like. Send patches in diff -u or -c format, making the diff from the old file to the new file. Include ChangeLog entries, but include them as separate text rather than as a patch to the ChangeLog file (a patch to the ChangeLog file will almost certainly not apply cleanly). If the patches are more than a few lines, the FSF requires you to sign a copyright assignment. I'll send you the form separately. Below is is a sample asm program and the nm and objdump output. Please tell me if anything looks wrong. Also, it seems some UNIX systems using COFF (like SCO OS5) support C_WEAKEXT as 108, but I'm not familiar enough with UNIX to adjust C_WEAKEXT like I did for PE in pe.h. In the BFD context, we could only support C_WEAKEXT as a storage class for a specific target which did not already use 108 for anything else. For example, perhaps for coff-i386 (only) C_WEAKEXT could be supported as 108, while for any target C_WEAK could be defined as some other value. I see that you used 127, which seems OK to me; can anybody think of a reason to use a different value? Ian