Re: [Xen-devel] pvops0 kernel 2.6.32.21 does not boot

[Jeremy Fitzhardinge <jeremy@xxxxxxxx>]

 On 10/01/2010 03:53 AM, Andreas Kinzler wrote:
> On 22.09.2010 19:23, Jeremy Fitzhardinge wrote:
>>>>>>>>>> current git stable 2.6.32.x (2968b25) crashes on boot with the
>>>>>>>>>> attached debug output. The same config file works perfectly on
>>>>>>>>>> 2.6.32.18 (e6b9b2c).
>>>>>>>>> Thanks for the report. I've pushed an update to
>>>>>>>>> xen/stable-2.6.32.x
>>>>>>>>> which should fix this.
>>>>>>>> Sorry, still does not boot. I attached the serial console log and
>>>>>>>> manually created a traceback from the stack hex dump.
>>>>>>> Does adding "mem=1g" to the kernel command line help?
>>>>>> No (log attached). But the stacktrace changed back to the first
>>>>>> "version":
>>> Are you still checking this? Any news?
>> Haven't forgotten about it, but I was waiting for some inspiration to
>> strike.  Could you try some earlier versions and try to isolate when the
>> problem appeared?
>
> Actually I found something very strange. Even the vanilla kernels from
> kernel.org stopped working as a domU with the transition of 2.6.32.21
> to 2.6.32.22. No output, no log, no nothing, just eating CPU time. Are
> there any debug options how to debug domUs that don't output anything?
>
> I am actually starting to wonder if the problems are related to
> kernels at all. I tried Xen 3.4.4 with the same result. Perhaps the
> whole problem is a binutils/gcc problem. What is your version of
> binutils / gcc / glibc?

Vanilla upstream 2.6.32.22 and 23 are broken due to a typo that crept
into a Xen patch between mainline and stable-2.6.32; I'm still waiting
for that fix to hit (stable-2.6.35 was also affected, but it has since
been fixed).

Change b0dd220aefe1a is the fix, included below.

And on the subject of the original bug, I committed some fixes which may
fix it for you (certainly fixed *some* memory init stuff), so if you
could test out xen/next-2.6.32 I'd appreciate it.

Thanks,
    J

From: James Dingwall <james.dingwall@xxxxxxxxxx>
Date: Mon, 27 Sep 2010 09:37:17 +0100
Subject: [PATCH] Xen: fix typo in previous patch

Correctly name the irq_chip structure to fix an immediate failure when booting
as a xen pv_ops guest with a NULL pointer exception. The missing 'x' was
introduced in commit [fb412a178502dc498430723b082a932f797e4763] applied to
2.6.3[25]-stable trees.  The commit to mainline was
[aaca49642b92c8a57d3ca5029a5a94019c7af69f] which did not have the problem.

Signed-off-by: James Dingwall <james@xxxxxxxxxxxxxx>
Reported-by: Pawel Zuzelski <pawelz@xxxxxxxxxxxxx>
Tested-by: Pawel Zuzelski <pawelz@xxxxxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxx>

diff --git a/drivers/xen/events.c b/drivers/xen/events.c
index 14ed358..6830a4d 100644
--- a/drivers/xen/events.c
+++ b/drivers/xen/events.c
@@ -935,7 +935,7 @@ static struct irq_chip xen_dynamic_chip __read_mostly = {
        .retrigger      = retrigger_dynirq,
 };
 
-static struct irq_chip en_percpu_chip __read_mostly = {
+static struct irq_chip xen_percpu_chip __read_mostly = {
        .name           = "xen-percpu",
 
        .disable        = disable_dynirq,




_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel