Re: __register_frame_info: A complete scan
- To: Jason Gunthorpe <firstname.lastname@example.org>
- Cc: Richard Braakman <email@example.com>, firstname.lastname@example.org
- Subject: Re: __register_frame_info: A complete scan
- From: Matthias Klose <email@example.com>
- Date: Mon, 30 Nov 1998 00:55:57 +0100 (MET)
- Message-id: <13921.56612.646858.589158@bolero>
- In-reply-to: <Pine.LNX.3.96.981129134648.30365G-100000@wakko>
- References: <E0zk9R7-0005Bv-00@night> <Pine.LNX.3.96.981129134648.30365G-100000@wakko>
Jason Gunthorpe writes:
> On Sun, 29 Nov 1998, Richard Braakman wrote:
> > All these binaries need to be recompiled for slink; they will break with
> > the new libc6. I excluded C++ binaries from the scan, since they don't
> > need to be recompiled -- they just need an updated libstdc++.
> Is this actually right? AFAIK NO library or aplication should define this
> as a dynamic symbol, it should always be statically linked. Has anyone
> asked the egcs people about it?
AFAIK the libc6/libm defining the __register_frame_info symbol was
compiled by an egcs-1.0.x or a self installed egcs. Before asking, I
want to verify, that this is still the case with the current
egcs-1.1.1 prerelease #4.
PS: The first egcs-126.96.36.199.59 upload was corrupted :-( Is there any
way that I can verify that the files are correctly built?