Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 01 Aug 2011 17:22:41 +0300
From:      Andriy Gapon <avg@FreeBSD.org>
To:        John Baldwin <jhb@FreeBSD.org>
Cc:        freebsd-current@FreeBSD.org, Andrew Thompson <andy@fud.org.nz>
Subject:   Re: variable init
Message-ID:  <4E36B6B1.60804@FreeBSD.org>
In-Reply-To: <201108010852.37607.jhb@freebsd.org>
References:  <CAFAOGNRN3e_9SyHU5yfRf-EhEWbWhpepCVQT-=n4LL-V44J1EQ@mail.gmail.com> <201108010852.37607.jhb@freebsd.org>

next in thread | previous in thread | raw e-mail | index | archive | help
on 01/08/2011 15:52 John Baldwin said the following:
> On Monday, August 01, 2011 3:09:27 am Andrew Thompson wrote:
>> Hi,
>>
>> Looking at,
>>
>> http://www.freebsd.org/cgi/query-pr.cgi?pr=159345
>>
>> The lock is a global variable, declared as
>>
>> static struct mtx       lagg_list_mtx;
>>
>> I would expect this to be zeroed memory, is this guaranteed?
> 
> Yes, but when if_lagg.ko is loaded, the runtime linker resolves references to 
> 'lagg_list_mtx' to point to the lock that is in the kernel's .bss from the 
> compiled-in lagg, not the one in the .ko file.  However, both initializer 
> functions run (the one from the kernel and the one from the .ko) which causes 
> the panic.

Tangentially related: kib has/had a patch for proper hiding of static symbols from
kernel runtime linker symbol resolution.  It would be nice to get that into the
tree some day.

-- 
Andriy Gapon



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4E36B6B1.60804>