This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
Re: [patch 1/2] [BUG]kallsyms_lookup_name should return the text addres
- From: Keith Owens <kaos at sgi dot com>
- To: Keshavamurthy Anil S <anil dot s dot keshavamurthy at intel dot com>
- Cc: Paulo Marques <pmarques at grupopie dot com>, Linux Kernel <linux-kernel at vger dot kernel dot org>, akpm at osdl dot org, tony dot luck at intel dot com, Systemtap <systemtap at sources dot redhat dot com>, Jim Keniston <jkenisto at us dot ibm dot com>
- Date: Wed, 11 Jan 2006 11:02:55 +1100
- Subject: Re: [patch 1/2] [BUG]kallsyms_lookup_name should return the text addres
Keshavamurthy Anil S (on Tue, 10 Jan 2006 15:29:05 -0800) wrote:
>On Wed, Jan 11, 2006 at 10:11:26AM +1100, Keith Owens wrote:
>> Keshavamurthy Anil S (on Tue, 10 Jan 2006 13:07:37 -0800) wrote:
>> >On Tue, Jan 10, 2006 at 08:45:02PM +0000, Paulo Marques wrote:
>> >But my [patch 2/2] speeds up the lookup and that can go in, I think.
>> >Please ack that patch if you think so.
>>
>> Your second patch changes the behaviour of kallsyms lookup w.r.t
>> duplicate symbols.
>With this send patch, kallsyms lookup first finds
>the real text address which is what we want. If you consider
>this as the change in behaviour, what is the negetive effect of this
>I am unable to get it.
Local symbols can be (and are) duplicated in the kernel code, and these
duplicate symbols can appear in modules. Changing the list order of
loaded modules also changes which version of a duplicated symbol is
returned by the kallsyms code. Not a big deal, but annoying enough to
say "don't change the module list order".
Changing the thread slightly, kallsyms_lookup_name() has never coped
with duplicate local symbols and it cannot do so without changing its
API, and all its callers. For debugging purposes, it would be nicer if
the kernel did not have any duplicate symbols. Perhaps some kernel
janitor would like to take that task on.