This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: Re: [committed] TILE-Gx GOLD: better offset ranges for large got.
- From: Jiong Wang <jiwang at tilera dot com>
- To: "amodra at gmail dot com" <amodra at gmail dot com>
- Cc: Walter Lee <walt at tilera dot com>, "binutils at sourceware dot org" <binutils at sourceware dot org>
- Date: Wed, 3 Oct 2012 06:23:09 +0000
- Subject: Re: Re: [committed] TILE-Gx GOLD: better offset ranges for large got.
- References: <20121003025950.GA25219@bubble.grove.modra.org> <506BC553.4090603@tilera.com>
Hi Alan,
Thanks for review.
for tilegx, we initially point _GLOBAL_OFFSET_TABLE_ to the start of .got, all calculation is based on this. Now if we change the position where _GLOBAL_OFFSET_TABLE_ point to, those relocation aginst _GLOBAL_OFFSET_TABLE_ will be affected, so we also change the offset with the same distance, then it should be OK.
And tilegx do not have x86_64 style got header, we only reserve the first entry in .got which kept the address of .dynamic, all .plt stub will not be affected by the _GLOBAL_OFFSET_TABLE_
Thanks
---
Jiong. Wang
Tilera Corporation
-------- Original Message --------
Subject: Re: [committed] TILE-Gx GOLD: better offset ranges for large got.
Date: Wed, 3 Oct 2012 12:29:50 +0930
From: Alan Modra <amodra@gmail.com>
To: Walter Lee <walt@tilera.com>
CC: <binutils@sourceware.org>
On Tue, Oct 02, 2012 at 08:59:39PM -0400, Walter Lee wrote:
>
> This patch adjust the value of _GLOBAL_OFFSET_TABLE_ for got bigger
> than 0x8000 bytes, so that the 16 bit relocations have a greater
> chance of working.
Don't you need to put the got header at _GLOBAL_OFFSET_TABLE_?
--
Alan Modra
Australia Development Lab, IBM