This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: PATCH: Reserve memory for ia64 ld.so (Bug in IA64 runtimelinker)
- From: Jim Wilson <wilson at specifixinc dot com>
- To: "H. J. Lu" <hjl at lucon dot org>
- Cc: Jakub Jelinek <jakub at redhat dot com>, binutils at sources dot redhat dot com,Ed Connell <Ed dot Connell at sas dot com>, libc-alpha at sources dot redhat dot com
- Date: 05 Apr 2004 15:01:01 -0700
- Subject: Re: PATCH: Reserve memory for ia64 ld.so (Bug in IA64 runtimelinker)
- References: <E0EA1095200B074CA58322B446C8D53904F3FA86@merc17.na.sas.com><20040401225901.GA12488@lucon.org><20040401211212.GH514@sunsite.ms.mff.cuni.cz><20040402161441.GA29990@lucon.org> <20040403192951.GA23446@lucon.org>
On Sat, 2004-04-03 at 11:29, H. J. Lu wrote:
> * elfxx-ia64.c (elfNN_ia64_size_dynamic_sections): Always
> reserve the memory for dynamic linker
The comment mentions only one problem, but there are two.
There is an internal inconsistency in bfd. It allocates the PLT_RESERVE
space in elfNN_ia64_size_dynamic_sections only if there are PLT entries,
but it defines the DT_IA_64_PLT_RESERVE reloc in the same function if
any dynamic section has been created. Thus we can get
DT_IA_64_PLT_RESERVE even when the space has not been allocated.
There is also the inconsistency between bfd and the glibc dynamic
linker. The dynamic linker always uses DT_IA_64_PLT_RESERVE, but the
linker is trying to allocate it only if there are plt entries.
Even if we fix glibc, we still need backwards compatibility with
existing systems for a long while, and thus bfd should always create the
DT_IA_64_PLT_RESERVE reloc, and allocate space for it, which is what
your patch does.
Your patch is OK.
--
Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com