This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: binutils-2.13.91: ARM_26D Relocs
- From: Shaun Jackman <sjackman at pathwayconnect dot com>
- To: Rainer Keuchel <rkeuchel at allgeier dot com>
- Cc: "Craig A. Vanderborgh" <craigv at voxware dot com>, binutils at sources dot redhat dot com
- Date: 18 Aug 2003 17:02:21 -0600
- Subject: Re: binutils-2.13.91: ARM_26D Relocs
- Organization: Pathway Connectivity
Hello Rainer,
I'm also using Craig's arm-wince-pe tool-chain. I've been happily using
binutils 2.13.2. When I upgraded to binutils 2.14, it stopped working. I
discovered that it's specifically gas that's causing my problem. I think
it might be related to the new ARM_26D relocation, since that's the only
difference I can see between the two object files. Have you learned
anything else about this problem?
Thanks,
Shaun
> I am working with Craig on the gcj port for WinCE (arm-wince-pe).
>
> I seem to have strange problems with ARM_26D relocs.
> pe-dll.c:generate_reloc() says "Error: 24-Bit reloc in dll".
>
> The problematic functions all seem to have callers
> from the file they are in and also external calls,
> (which should be no problem). The external calls might
> be "far away". They are also in .stab entries.
>
> I do not have ARM_26D relocs when assembling with
> an older gas.
>
> Craig said it would be fixed in the latest snapshot,
> but I cannot see any fix in the diffs I have run,
> only a new option $enable_26bit in a configure file,
> which seems to be dead for compilng...
>
> Could someone briefly explain how ARM_26D relocs are
> supposed to work and why they are used?
>
> Is there a performance penalty when not using them?