RE: [PATCH] exclude runtime-pseudo-reloc symbols from auto-export

2002-12-22 Thread Ralf Habacker
Hi Chris, 1. I don't see how you could do that since the symbol is associated with an existing place in memory. We could put the whole function in a different segment but that's not the kind of solution I was thinking of. 2. I was thinking that there might be an unused attribute that could be

RE: [PATCH] exclude runtime-pseudo-reloc symbols from auto-export

2002-12-21 Thread Ralf Habacker
Maybe the horse has left the barn already but it would have been nice (tm) if these type of symbols were marked in some generic way so that we wouldn't have to keep remembering to extend this table. I recall commenting on this aspect in a recent binutils thread in the cygwin lists, and

Re: [PATCH] exclude runtime-pseudo-reloc symbols from auto-export

2002-12-21 Thread Charles Wilson
Robert Collins wrote: On Sat, 2002-12-21 at 15:39, Christopher Faylor wrote: Maybe the horse has left the barn already but it would have been nice (tm) if these type of symbols were marked in some generic way so that we wouldn't have to keep remembering to extend this table. I recall

Re: [PATCH] exclude runtime-pseudo-reloc symbols from auto-export

2002-12-21 Thread Christopher Faylor
On Sat, Dec 21, 2002 at 02:10:18PM +0100, Ralf Habacker wrote: Maybe the horse has left the barn already but it would have been nice (tm) if these type of symbols were marked in some generic way so that we wouldn't have to keep remembering to extend this table. I recall commenting on this aspect

Re: [PATCH] exclude runtime-pseudo-reloc symbols from auto-export

2002-12-21 Thread Christopher Faylor
On Sat, Dec 21, 2002 at 06:48:09PM +1100, Robert Collins wrote: On Sat, 2002-12-21 at 15:39, Christopher Faylor wrote: Maybe the horse has left the barn already but it would have been nice (tm) if these type of symbols were marked in some generic way so that we wouldn't have to keep

Re: [PATCH] exclude runtime-pseudo-reloc symbols from auto-export

2002-12-21 Thread Robert Collins
On Sun, 2002-12-22 at 04:55, Christopher Faylor wrote: Didn't you suggest that anything in a library residing in /usr/lib or /usr/local/lib be excluded? That's not quite the same thing as what I was musing about. I was talking about marking individual symbols. Ah, ok. oops. (goes and

Re: [PATCH] exclude runtime-pseudo-reloc symbols from auto-export

2002-12-20 Thread Christopher Faylor
On Fri, Dec 20, 2002 at 11:23:36PM -0500, Charles Wilson wrote: I don't think these symbols should be auto-exported by DLLs; they are internal, and are added (if the platform supports it) to every DLL and EXE. Since everybody has their own copy, there's no need for any DLL to export them.

Re: [PATCH] exclude runtime-pseudo-reloc symbols from auto-export

2002-12-20 Thread Robert Collins
On Sat, 2002-12-21 at 15:39, Christopher Faylor wrote: Maybe the horse has left the barn already but it would have been nice (tm) if these type of symbols were marked in some generic way so that we wouldn't have to keep remembering to extend this table. I recall commenting on this aspect in a