[PATCH] Fix lguest w/ lockdep

[PATCH] Fix lguest w/ lockdep

[Rusty Russell]

On Wed, 2007-06-27 at 19:14 -0500, Matt Mackall wrote:
> lguest: unhandled trap 14 at 0xc01362f7 (0x0)
> Dump of assembler code for function __lock_acquire:

Thanks for the bug report Matt!  I wonder if other
paravirt_disable_iospace users have the same issue...
===

Fix "lguest: unhandled trap 14 at 0xc013630f (0x0)" with CONFIG_LOCKDEP=y

paravirt_disable_iospace -> request_resource -> write_lock -> __lock_acquire()

lockdep_init() is already idempotent: simply call it before
paravirt_disable_iospace().

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

diff -r bea2b8147985 drivers/lguest/lguest.c
--- a/drivers/lguest/lguest.c	Thu Jun 07 22:50:36 2007 +1000
+++ b/drivers/lguest/lguest.c	Thu Jun 28 23:21:26 2007 +1000
@@ -593,6 +593,8 @@ __init void lguest_init(void *boot)
 
 	reserve_top_address(lguest_data.reserve_mem);
 
+	lockdep_init();
+
 	paravirt_disable_iospace();
 
 	cpu_detect(&new_cpu_data);

Re: [PATCH] Fix lguest w/ lockdep

[Jeremy Fitzhardinge]

Rusty Russell wrote:
> On Wed, 2007-06-27 at 19:14 -0500, Matt Mackall wrote:
>   
>> lguest: unhandled trap 14 at 0xc01362f7 (0x0)
>> Dump of assembler code for function __lock_acquire:
>>     
>
> Thanks for the bug report Matt!  I wonder if other
> paravirt_disable_iospace users have the same issue...
>   

I call it a little later (from arch_setup), and I think lockdep_init 
will have already been called at that point.  Hm, yes, it gets called at 
the top of start_kernel.

    J