thr3ads.net - Xen users - [Xen-devel] VIRQ_CON

If this information is useful, please help other people find it:
Share via:

Jan Beulich

2009-Nov-12 14:19 UTC

[Xen-devel] VIRQ_CON_RING

Is there any real user for this vIRQ? I''m asking because, while
debugging
a problem that required to add some printk()s that could temporarily issue
at a high rate, I ran into the situation that this extra output prevented
guest/dom0 boot from making any progress. As I then realized this was
due to the tasklet_schedule() called from inside the console handling
code, which resulted in subsequent hypercall_preempt_check() to force
exit (after creating a continuation) from the respective hypercall handler
immediately.

While I realize that for compatibility reasons (even in the case of there
not being a current user) it may not be possible to drop this vIRQ
altogether, I wonder whether it would be possible to avoid scheduling
the tasklet when the vIRQ has no handler and/or is already pending.

Jan


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

Keir Fraser

2009-Nov-12 14:51 UTC

head link

Re: [Xen-devel] VIRQ_CON_RING

On 12/11/2009 14:19, "Jan Beulich" <JBeulich@novell.com> wrote:
> Is there any real user for this vIRQ?
xenconsoled when started with --log=hv|all
> While I realize that for compatibility reasons (even in the case of there
> not being a current user) it may not be possible to drop this vIRQ
> altogether, I wonder whether it would be possible to avoid scheduling
> the tasklet when the vIRQ has no handler and/or is already pending.
So this is due to you adding a really noisy printk in the middle of a
hypercall? I don''t think you should expect goodness to result from
that.
Seems to me the issue is as much the extreme load you put on printk as it is
printk''s overhead.

 -- Keir



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

Jan Beulich

2009-Nov-12 15:30 UTC

head link

Re: [Xen-devel] VIRQ_CON_RING

>>> Keir Fraser <keir.fraser@eu.citrix.com> 12.11.09 15:51
>>>
>On 12/11/2009 14:19, "Jan Beulich" <JBeulich@novell.com>
wrote:
>> While I realize that for compatibility reasons (even in the case of
there
>> not being a current user) it may not be possible to drop this vIRQ
>> altogether, I wonder whether it would be possible to avoid scheduling
>> the tasklet when the vIRQ has no handler and/or is already pending.
>
>So this is due to you adding a really noisy printk in the middle of a
>hypercall? I don''t think you should expect goodness to result from
that.
No, it wasn''t really meant to be that noisy (e.g., it was printing only
as
long as a guest didn''t have a page fault handler registered, and now
that I relocated the printing [without changing their amount] I know
that after an initial burst this printed just about a dozen lines).
>Seems to me the issue is as much the extreme load you put on printk as it is
>printk''s overhead.
I don''t think so: Since __putstr() calls tasklet_schedule() directly
and
(basically) unconditionally, it is clear that after every printk()
hypercall_preempt_check() will return true, and hence placing one
anywhere before such a check will make sure that preemption is going
to happen. Since the code path will be the same after the continuation
hypercall got invoked, it is impossible to make any progress if the
preemption check is placed at the beginning of a handler/loop (and
even if, like in alloc_l[34]_table(), it is placed after having done at
least one iteration, progress is possibly going to be unnoticeable).

Jan


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

Keir Fraser

2009-Nov-12 16:03 UTC

head link

Re: [Xen-devel] VIRQ_CON_RING

On 12/11/2009 15:30, "Jan Beulich" <JBeulich@novell.com> wrote:
>> Seems to me the issue is as much the extreme load you put on printk as
it is
>> printk''s overhead.
> 
> I don''t think so: Since __putstr() calls tasklet_schedule()
directly and
> (basically) unconditionally, it is clear that after every printk()
> hypercall_preempt_check() will return true, and hence placing one
> anywhere before such a check will make sure that preemption is going
> to happen. Since the code path will be the same after the continuation
> hypercall got invoked, it is impossible to make any progress if the
> preemption check is placed at the beginning of a handler/loop (and
> even if, like in alloc_l[34]_table(), it is placed after having done at
> least one iteration, progress is possibly going to be unnoticeable).
Well there is that, but who unconditionally prints at the start of a
hypercall? Only someone doing very specific and verbose debugging.

 -- Keir



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

Boris Derzhavets

2009-Nov-13 07:23 UTC

head link

[Xen-users] Failure to start X-Server for Xen on openSUSE 11.2

Regardless same bug was reported against RC2 :

https://bugzilla.novell.com/show_bug.cgi?id=552492

It''s present in final release :-

https://bugzilla.novell.com/show_bug.cgi?id=555172

Boris



      

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

Jan Beulich

2009-Nov-13 07:58 UTC

head link

[Xen-users] Re: Failure to start X-Server for Xen on openSUSE 11.2

Yes, the fix was not considered important enough to be taken for
the final release (and you may simply have missed that I set the target
milestone of the bug to "unspecified" just to express this).
You''ll have to
wait for a kernel maintenance update...

Jan
>>> Boris Derzhavets <bderzhavets@yahoo.com> 13.11.09 08:23
>>>
Regardless same bug was reported against RC2 :

https://bugzilla.novell.com/show_bug.cgi?id=552492 

It''s present in final release :-

https://bugzilla.novell.com/show_bug.cgi?id=555172 

Boris



      


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

Boris Derzhavets

2009-Nov-13 08:33 UTC

head link

Re: [Xen-users] Re: Failure to start X-Server for Xen on openSUSE 11.2

It sounds like Xen is not important enough for openSUSE 11.2 final release.
I don''t have any more questions.

Boris.

--- On Fri, 11/13/09, Jan Beulich <JBeulich@novell.com> wrote:

From: Jan Beulich <JBeulich@novell.com>
Subject: [Xen-users] Re: Failure to start X-Server for Xen on openSUSE 11.2
To: "Boris Derzhavets" <bderzhavets@yahoo.com>
Cc: xen-users@lists.xensource.com
Date: Friday, November 13, 2009, 2:58 AM

Yes, the fix was not considered important enough to be taken for
the final release (and you may simply have missed that I set the target
milestone of the bug to "unspecified" just to express this).
You''ll have to
wait for a kernel maintenance update...

Jan
>>> Boris Derzhavets <bderzhavets@yahoo.com> 13.11.09 08:23
>>>
Regardless same bug was reported against RC2 :

https://bugzilla.novell.com/show_bug.cgi?id=552492 

It''s present in final release :-

https://bugzilla.novell.com/show_bug.cgi?id=555172 

Boris



      


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



      

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

Andrew Lyon

2009-Nov-13 13:39 UTC

head link

Re: [Xen-users] Re: Failure to start X-Server for Xen on openSUSE 11.2

I don''t think thats a very fair comment Boris, Jan works extremely hard
to
forward port the Xen dom0 to new kernels, its just that this particular fix
was too late to make the final build as it had not been fully tested.

There are a lot of people who want to run Xen on workstations with X, but
even more who run Xen servers headless or with serial console so I
understand the decision, anyway you can easily get the patch from
https://bugzilla.novell.com/attachment.cgi?id=325743 and apply it yourself,
updated versions of the 11.2 2.6.31 kernel will be available in
ftp://ftp.suse.com/pub/projects/kernel/kotd/openSUSE-11.2/src/ for a long
time so thanks to Jan & openSUSE we have the choice of using a more up to
date and supported "classic" Xen dom0 kernel.

Andy

On Fri, Nov 13, 2009 at 8:33 AM, Boris Derzhavets
<bderzhavets@yahoo.com>wrote:
> It sounds like Xen is not important enough for openSUSE 11.2 final release.
> I don''t have any more questions.
>
> Boris.
>
> --- On *Fri, 11/13/09, Jan Beulich <JBeulich@novell.com>* wrote:
>
>
> From: Jan Beulich <JBeulich@novell.com>
> Subject: [Xen-users] Re: Failure to start X-Server for Xen on openSUSE 11.2
> To: "Boris Derzhavets" <bderzhavets@yahoo.com>
> Cc: xen-users@lists.xensource.com
> Date: Friday, November 13, 2009, 2:58 AM
>
>
> Yes, the fix was not considered important enough to be taken for
> the final release (and you may simply have missed that I set the target
> milestone of the bug to "unspecified" just to express this).
You''ll have to
> wait for a kernel maintenance update...
>
> Jan
>
> >>> Boris Derzhavets
<bderzhavets@yahoo.com<http://mc/compose?to=bderzhavets@yahoo.com>>
> 13.11.09 08:23 >>>
>
> Regardless same bug was reported against RC2 :
>
> https://bugzilla.novell.com/show_bug.cgi?id=552492
>
> It''s present in final release :-
>
> https://bugzilla.novell.com/show_bug.cgi?id=555172
>
> Boris
>
>
>
>
>
>
> _______________________________________________
> Xen-users mailing list
>
Xen-users@lists.xensource.com<http://mc/compose?to=Xen-users@lists.xensource.com>
> http://lists.xensource.com/xen-users
>
>
>
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
>

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

Boris Derzhavets

2010-Oct-09 20:10 UTC

head link

[Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

. . . . . . . . . . . . . . . 

  LD      drivers/built-in.o
  LD      vmlinux.o
  MODPOST vmlinux.o
  GEN     .version
  CHK     include/generated/compile.h
  UPD     include/generated/compile.h
  CC      init/version.o
  LD      init/built-in.o
  LD      .tmp_vmlinux1
drivers/built-in.o: In function `cpu_release_store'':
/root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:96: undefined reference to
`arch_cpu_release''
drivers/built-in.o: In function `cpu_probe_store'':
/root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:88: undefined reference to
`arch_cpu_probe''
drivers/built-in.o: In function `store_online'':
/root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:42: undefined reference to
`cpu_hotplug_driver_lock''
/root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:57: undefined reference to
`cpu_hotplug_driver_unlock''
make: *** [.tmp_vmlinux1] Error 1


Commenting undefined references gives an option to build the 2.6.34.7 xenified
aka Suse :-
--- cpu.c.orig    2010-10-09 23:31:26.641580071 +0400
+++ cpu.c    2010-10-10 00:02:40.913581011 +0400
@@ -39,7 +39,7 @@
     struct cpu *cpu = container_of(dev, struct cpu, sysdev);
     ssize_t ret;
 
-    cpu_hotplug_driver_lock();
+    /* cpu_hotplug_driver_lock(); */
     switch (buf[0]) {
     case ''0'':
         ret = cpu_down(cpu->sysdev.id);
@@ -54,7 +54,7 @@
     default:
         ret = -EINVAL;
     }
-    cpu_hotplug_driver_unlock();
+       /* cpu_hotplug_driver_unlock(); */
 
     if (ret >= 0)
         ret = count;
@@ -85,7 +85,8 @@
                    const char *buf,
                    size_t count)
 {
-    return arch_cpu_probe(buf, count);
+      /* return arch_cpu_probe(buf, count); */
+      return 0;
 }
 
 static ssize_t cpu_release_store(struct sysdev_class *class,
@@ -93,7 +94,8 @@
                  const char *buf,
                  size_t count)
 {
-    return arch_cpu_release(buf, count);
+      /* return arch_cpu_release(buf, count); */
+      return 0;
 }

Boris.






      

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

Boris Derzhavets

2010-Oct-10 14:35 UTC

head link

Re: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

Another way to disable this calls via include/linux/cpu.h , having
CONFIG_HOTPLUG_CPU=y is to  change arch/x86/Kconfig :-

--- arch/x86/Kconfig.orig    2010-10-10 18:29:24.490068330 +0400
+++ arch/x86/Kconfig    2010-10-10 17:50:22.383715068 +0400
@@ -254,8 +254,8 @@
     def_bool X86_32
 
 config ARCH_CPU_PROBE_RELEASE
-    def_bool y
-    depends on HOTPLUG_CPU
+    def_bool n
+#    depends on HOTPLUG_CPU

Boris

--- On Sat, 10/9/10, Boris Derzhavets <bderzhavets@yahoo.com> wrote:

From: Boris Derzhavets <bderzhavets@yahoo.com>
Subject: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7
vanilla
To: "Andrew Lyon" <andrew.lyon@gmail.com>
Cc: xen-users@lists.xensource.com
Date: Saturday, October 9, 2010, 4:10 PM

. . . . . . . . . . . . . . . 

  LD      drivers/built-in.o
  LD      vmlinux.o
  MODPOST vmlinux.o
  GEN     .version
  CHK     include/generated/compile.h
  UPD     include/generated/compile.h
  CC      init/version.o
  LD      init/built-in.o
  LD      .tmp_vmlinux1
drivers/built-in.o: In function `cpu_release_store'':
/root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:96: undefined reference to
`arch_cpu_release''
drivers/built-in.o: In function `cpu_probe_store'':
/root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:88: undefined reference to
`arch_cpu_probe''
drivers/built-in.o: In function
 `store_online'':
/root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:42: undefined reference to
`cpu_hotplug_driver_lock''
/root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:57: undefined reference to
`cpu_hotplug_driver_unlock''
make: *** [.tmp_vmlinux1] Error 1


Commenting undefined references gives an option to build the 2.6.34.7 xenified
aka Suse :-
--- cpu.c.orig    2010-10-09 23:31:26.641580071 +0400
+++ cpu.c    2010-10-10 00:02:40.913581011 +0400
@@ -39,7 +39,7 @@
     struct cpu *cpu = container_of(dev, struct cpu, sysdev);
     ssize_t ret;
 
-    cpu_hotplug_driver_lock();
+    /* cpu_hotplug_driver_lock(); */
     switch (buf[0]) {
     case ''0'':
         ret = cpu_down(cpu->sysdev.id);
@@ -54,7 +54,7
 @@
     default:
         ret = -EINVAL;
     }
-    cpu_hotplug_driver_unlock();
+       /* cpu_hotplug_driver_unlock(); */
 
     if (ret >= 0)
         ret = count;
@@ -85,7 +85,8 @@
                    const char *buf,
                    size_t count)
 {
-    return arch_cpu_probe(buf, count);
+      /* return arch_cpu_probe(buf, count); */
+      return 0;
 }
 
 static ssize_t cpu_release_store(struct sysdev_class *class,
@@ -93,7 +94,8 @@
    
              const char *buf,
                  size_t count)
 {
-    return arch_cpu_release(buf, count);
+      /* return arch_cpu_release(buf, count); */
+      return 0;
 }

Boris.






      
-----Inline Attachment Follows-----

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


      

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

Mike Viau

2010-Oct-10 14:51 UTC

head link

RE: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

> On Sun, 10 Oct 2010 07:35:55 -0700 <bderzhavets@yahoo.com> wrote:
>
> Another way to disable this calls via include/linux/cpu.h , having
> CONFIG_HOTPLUG_CPU=y is to change arch/x86/Kconfig :-
>
> --- arch/x86/Kconfig.orig 2010-10-10 18:29:24.490068330 +0400
> +++ arch/x86/Kconfig 2010-10-10 17:50:22.383715068 +0400
> @@ -254,8 +254,8 @@
> def_bool X86_32
>
> config ARCH_CPU_PROBE_RELEASE
> - def_bool y
> - depends on HOTPLUG_CPU
> + def_bool n
> +# depends on HOTPLUG_CPU
>
> Boris
What are the reciprocation of turning off the ability to HOTPLUG_CPU? Is this
feature ever used by XEN?

Thanks.

-M 
 		 	   		  
_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

listmail

2010-Oct-10 15:44 UTC

head link

Re: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

On 10/10/2010 10:51 AM, Mike Viau wrote:>> On Sun, 10 Oct 2010 07:35:55 -0700<bderzhavets@yahoo.com>  wrote:
>>
>> Another way to disable this calls via include/linux/cpu.h , having
>> CONFIG_HOTPLUG_CPU=y is to change arch/x86/Kconfig :-
>>
>> --- arch/x86/Kconfig.orig 2010-10-10 18:29:24.490068330 +0400
>> +++ arch/x86/Kconfig 2010-10-10 17:50:22.383715068 +0400
>> @@ -254,8 +254,8 @@
>> def_bool X86_32
>>
>> config ARCH_CPU_PROBE_RELEASE
>> - def_bool y
>> - depends on HOTPLUG_CPU
>> + def_bool n
>> +# depends on HOTPLUG_CPU
>>
>> Boris
> What are the reciprocation of turning off the ability to HOTPLUG_CPU? Is
this feature ever used by XEN?
>
> Thanks.
>
> -M
>   		 	   		
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
Does it compile as is with CONFIG_HOTPLUG_CPU=n ?

I am curious about this myself. I run the base 2.6.34 kernel (xenified) 
right now.    
http://kerneltrap.org/mailarchive/git-commits-head/2008/10/12/3624204 
was created for being able to build without it set. 
http://lists.xensource.com/archives/cgi-bin/mesg.cgi?a=xen-devel&i=E2263E4A5B2284449EEBD0AAB751098418E5557549%40PDSMSX501.ccr.corp.intel.com
shows work for physical CPU hot-add on the hypervisor.   This makes me 
think that you could not do any physical hot adding of CPU''s but that 
other functionality would not be impacted.  Is it that simple?

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

Boris Derzhavets

2010-Oct-10 16:41 UTC

head link

RE: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

You cannot dynamically take CPU on-line  or off-line. The step bellow will break
HOTPLUG_CPU anyway ( it''s hack).
But , i believe that native Suse''s 11.3 xenified 2.6.34.7-0.3-xen
doesn''t do that.

Boris.

--- On Sun, 10/10/10, Mike Viau <viaum@sheridanc.on.ca> wrote:

From: Mike Viau <viaum@sheridanc.on.ca>
Subject: RE: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to
2.6.34.7 vanilla
To: xen-users@lists.xensource.com
Date: Sunday, October 10, 2010, 10:51 AM

> On Sun, 10 Oct 2010 07:35:55 -0700 <bderzhavets@yahoo.com> wrote:
>
> Another way to disable this calls via include/linux/cpu.h , having
> CONFIG_HOTPLUG_CPU=y is to change arch/x86/Kconfig :-
>
> --- arch/x86/Kconfig.orig 2010-10-10 18:29:24.490068330 +0400
> +++ arch/x86/Kconfig 2010-10-10 17:50:22.383715068 +0400
> @@ -254,8 +254,8 @@
> def_bool X86_32
>
> config ARCH_CPU_PROBE_RELEASE
> - def_bool y
> - depends on HOTPLUG_CPU
> + def_bool n
> +# depends on HOTPLUG_CPU
>
> Boris
What are the reciprocation of turning off the ability to HOTPLUG_CPU? Is this
feature ever used by XEN?

Thanks.

-M 
                           
_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users



      

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

Boris Derzhavets

2010-Oct-10 16:59 UTC

head link

Re: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

> Does it compile as is with CONFIG_HOTPLUG_CPU=n ?
Not exactly.  To set  CONFIG_HOTPLUG_CPU=n  I had to do  
( not sure steps are really required ) :-

1) kernel/power/Kconfig.
Step 1
config PM_SLEEP_SMP
        bool
        depends on SMP
        depends on ARCH_SUSPEND_POSSIBLE || ARCH_HIBERNATION_POSSIBLE
        depends on PM_SLEEP
#       select HOTPLUG_CPU
        default y

2) arch/x86/Kconfig
Step 2
config HOTPLUG_CPU
#       bool "Support for hot-pluggable CPUs"
        def_bool n
#       depends on SMP && HOTPLUG
        ---help---
          Say Y here to allow turning CPUs off and on. CPUs can be
          controlled through /sys/devices/system/cpu.
          ( Note: power management support will enable this option
            automatically on SMP systems. )
          Say N if you want to disable CPU hotplug.

But , the result should be the same. HOTPLUG_CPU feature should be broken.
It''s second kernel been built. I have not tested it yet.
  First one with HOTPLUG_CPU=y and ARCH_CPU_PROBE_RELEASE=n seems to work OK.

Boris.

Boris.

--- On Sun, 10/10/10, listmail <listmail@triad.rr.com> wrote:

From: listmail <listmail@triad.rr.com>
Subject: Re: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to
2.6.34.7 vanilla
To: "Boris Derzhavets" <bderzhavets@yahoo.com>
Cc: xen-users@lists.xensource.com
Date: Sunday, October 10, 2010, 11:44 AM

 On 10/10/2010 10:51 AM, Mike Viau wrote:>> On Sun, 10 Oct 2010 07:35:55 -0700<bderzhavets@yahoo.com>  wrote:
>> 
>> Another way to disable this calls via include/linux/cpu.h , having
>> CONFIG_HOTPLUG_CPU=y is to change arch/x86/Kconfig :-
>> 
>> --- arch/x86/Kconfig.orig 2010-10-10 18:29:24.490068330 +0400
>> +++ arch/x86/Kconfig 2010-10-10 17:50:22.383715068 +0400
>> @@ -254,8 +254,8 @@
>> def_bool X86_32
>> 
>> config ARCH_CPU_PROBE_RELEASE
>> - def_bool y
>> - depends on HOTPLUG_CPU
>> + def_bool n
>> +# depends on HOTPLUG_CPU
>> 
>> Boris
> What are the reciprocation of turning off the ability to HOTPLUG_CPU? Is
this feature ever used by XEN?
> 
> Thanks.
> 
> -M
>                           
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
Does it compile as is with CONFIG_HOTPLUG_CPU=n ?

I am curious about this myself. I run the base 2.6.34 kernel (xenified) right
now.    http://kerneltrap.org/mailarchive/git-commits-head/2008/10/12/3624204
was created for being able to build without it set.
http://lists.xensource.com/archives/cgi-bin/mesg.cgi?a=xen-devel&i=E2263E4A5B2284449EEBD0AAB751098418E5557549%40PDSMSX501.ccr.corp.intel.com
shows work for physical CPU hot-add on the hypervisor.   This makes me think
that you could not do any physical hot adding of CPU''s but that other
functionality would not be impacted.  Is it that simple?

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

listmail

2010-Oct-10 17:37 UTC

head link

Re: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

On 10/10/2010 12:59 PM, Boris Derzhavets wrote:> > Does it compile as is with CONFIG_HOTPLUG_CPU=n ?
>
> Not exactly.  To set  CONFIG_HOTPLUG_CPU=n  I had to do
> ( not sure steps are really required ) :-
>
> 1) kernel/power/Kconfig.
> Step 1
> config PM_SLEEP_SMP
>         bool
>         depends on SMP
>         depends on ARCH_SUSPEND_POSSIBLE || ARCH_HIBERNATION_POSSIBLE
>         depends on PM_SLEEP
> #       select HOTPLUG_CPU
>         default y
>
> 2) arch/x86/Kconfig
> Step 2
> config HOTPLUG_CPU
> #       bool "Support for hot-pluggable CPUs"
>         def_bool n
> #       depends on SMP && HOTPLUG
>         ---help---
>           Say Y here to allow turning CPUs off and on. CPUs can be
>           controlled through /sys/devices/system/cpu.
>           ( Note: power management support will enable this option
>             automatically on SMP systems. )
>           Say N if you want to disable CPU hotplug.
>
>
>those functions (undefined references) are defined here 
http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.34.y.git;a=blob_plain;f=arch/x86/kernel/smpboot.c;hb=f42346985c7721798c577606cb4d0f719940d33e

I wonder if this file is not getting used because of some other setting 
is not set or perhaps its because arch is xen and there are no such file 
smpboot.c in arch 
<http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.34.y.git;a=tree;f=arch;hb=f42346985c7721798c577606cb4d0f719940d33e>
/ x86 
<http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.34.y.git;a=tree;f=arch/x86;hb=f42346985c7721798c577606cb4d0f719940d33e>
/ xen 
<http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.34.y.git;a=tree;f=arch/x86/xen;hb=f42346985c7721798c577606cb4d0f719940d33e>
?

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

listmail

2010-Oct-10 17:51 UTC

head link

Re: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

On 10/10/2010 1:37 PM, listmail wrote:> On 10/10/2010 12:59 PM, Boris Derzhavets wrote:
>> > Does it compile as is with CONFIG_HOTPLUG_CPU=n ?
>>
>> Not exactly.  To set  CONFIG_HOTPLUG_CPU=n  I had to do
>> ( not sure steps are really required ) :-
>>
>> 1) kernel/power/Kconfig.
>> Step 1
>> config PM_SLEEP_SMP
>>         bool
>>         depends on SMP
>>         depends on ARCH_SUSPEND_POSSIBLE || ARCH_HIBERNATION_POSSIBLE
>>         depends on PM_SLEEP
>> #       select HOTPLUG_CPU
>>         default y
>>
>> 2) arch/x86/Kconfig
>> Step 2
>> config HOTPLUG_CPU
>> #       bool "Support for hot-pluggable CPUs"
>>         def_bool n
>> #       depends on SMP && HOTPLUG
>>         ---help---
>>           Say Y here to allow turning CPUs off and on. CPUs can be
>>           controlled through /sys/devices/system/cpu.
>>           ( Note: power management support will enable this option
>>             automatically on SMP systems. )
>>           Say N if you want to disable CPU hotplug.
>>
>>
>>
> those functions (undefined references) are defined here 
>
http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.34.y.git;a=blob_plain;f=arch/x86/kernel/smpboot.c;hb=f42346985c7721798c577606cb4d0f719940d33e
>
> I wonder if this file is not getting used because of some other 
> setting is not set or perhaps its because arch is xen and there are no 
> such file smpboot.c in arch 
>
<http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.34.y.git;a=tree;f=arch;hb=f42346985c7721798c577606cb4d0f719940d33e>
> / x86 
>
<http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.34.y.git;a=tree;f=arch/x86;hb=f42346985c7721798c577606cb4d0f719940d33e>
> / xen 
>
<http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.34.y.git;a=tree;f=arch/x86/xen;hb=f42346985c7721798c577606cb4d0f719940d33e>
> ?
>
Err, or perhaps that the xen patches provide xen/core/smpboot.c but they 
do not have those particular defines.  I don''t have access to a patched
tree right now otherwise I''d probably diff the 2.6.34.4, pvops and 
Xenified versions to see if some missing code could be pasted in.  Might 
not be worth the effort though.



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

Boris Derzhavets

2010-Oct-10 18:31 UTC

head link

Re: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

Virt-manager ( 0.8.4) & libvirt 0.8.3 working fine with 2.6.32.24 pvops,
cannot properly obtain information via xend connection , when i attempted to
create PV guest
with 2.6.34.7-xen kernel ( xm create f14.install works OK via VFB). Base OS
Ubuntu 10.04.1 Server.

Boris.


  



      

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

Andrew Lyon

2010-Oct-10 19:05 UTC

head link

[Xen-users] Re: Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

I will upload a updated patchset tomorrow, -4 was made before .7 was
out and I''m not surprised that it fails to apply cleanly.

I''ve been too busy to update the patches much recently but I will try
to do them all this week.

Andy

On 09/10/2010, Boris Derzhavets <bderzhavets@yahoo.com>
wrote:> . . . . . . . . . . . . . . .
>
>   LD      drivers/built-in.o
>   LD      vmlinux.o
>   MODPOST vmlinux.o
>   GEN     .version
>   CHK     include/generated/compile.h
>   UPD     include/generated/compile.h
>   CC      init/version.o
>   LD      init/built-in.o
>   LD      .tmp_vmlinux1
> drivers/built-in.o: In function `cpu_release_store'':
> /root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:96: undefined reference
to
> `arch_cpu_release''
> drivers/built-in.o: In function `cpu_probe_store'':
> /root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:88: undefined reference
to
> `arch_cpu_probe''
> drivers/built-in.o: In function `store_online'':
> /root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:42: undefined reference
to
> `cpu_hotplug_driver_lock''
> /root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:57: undefined reference
to
> `cpu_hotplug_driver_unlock''
> make: *** [.tmp_vmlinux1] Error 1
>
>
> Commenting undefined references gives an option to build the 2.6.34.7
> xenified aka Suse :-
> --- cpu.c.orig    2010-10-09 23:31:26.641580071 +0400
> +++ cpu.c    2010-10-10 00:02:40.913581011 +0400
> @@ -39,7 +39,7 @@
>      struct cpu *cpu = container_of(dev, struct cpu, sysdev);
>      ssize_t ret;
>
> -    cpu_hotplug_driver_lock();
> +    /* cpu_hotplug_driver_lock(); */
>      switch (buf[0]) {
>      case ''0'':
>          ret = cpu_down(cpu->sysdev.id);
> @@ -54,7 +54,7 @@
>      default:
>          ret = -EINVAL;
>      }
> -    cpu_hotplug_driver_unlock();
> +       /* cpu_hotplug_driver_unlock(); */
>
>      if (ret >= 0)
>          ret = count;
> @@ -85,7 +85,8 @@
>                     const char *buf,
>                     size_t count)
>  {
> -    return arch_cpu_probe(buf, count);
> +      /* return arch_cpu_probe(buf, count); */
> +      return 0;
>  }
>
>  static ssize_t cpu_release_store(struct sysdev_class *class,
> @@ -93,7 +94,8 @@
>                   const char *buf,
>                   size_t count)
>  {
> -    return arch_cpu_release(buf, count);
> +      /* return arch_cpu_release(buf, count); */
> +      return 0;
>  }
>
> Boris.
>
>
>
>
>
>
>
_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Boris Derzhavets

2010-Oct-11 08:48 UTC

head link

Re: [Xen-users] Re: Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

> I''m not surprised that it fails to apply cleanly.
It applies cleanly, but gives errors on kernel build in drivers/base/cpu.c
related with HOTPLUG_CPU=y

Boris.

--- On Sun, 10/10/10, Andrew Lyon <andrew.lyon@gmail.com> wrote:

From: Andrew Lyon <andrew.lyon@gmail.com>
Subject: [Xen-users] Re: Attempt to apply xen-patches-2.6.34-4.tar.bz2 to
2.6.34.7 vanilla
To: "Boris Derzhavets" <bderzhavets@yahoo.com>,
xen-users@lists.xensource.com
Date: Sunday, October 10, 2010, 3:05 PM

I will upload a updated patchset tomorrow, -4 was made before .7 was
out and I''m not surprised that it fails to apply cleanly.

I''ve been too busy to update the patches much recently but I will try
to do them all this week.

Andy

On 09/10/2010, Boris Derzhavets <bderzhavets@yahoo.com>
wrote:> . . . . . . . . . . . . . . .
>
>   LD      drivers/built-in.o
>   LD      vmlinux.o
>   MODPOST vmlinux.o
>   GEN     .version
>   CHK     include/generated/compile.h
>   UPD     include/generated/compile.h
>   CC      init/version.o
>   LD      init/built-in.o
>   LD      .tmp_vmlinux1
> drivers/built-in.o: In function `cpu_release_store'':
> /root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:96: undefined reference
to
> `arch_cpu_release''
> drivers/built-in.o: In function `cpu_probe_store'':
> /root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:88: undefined reference
to
> `arch_cpu_probe''
> drivers/built-in.o: In function `store_online'':
> /root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:42: undefined reference
to
> `cpu_hotplug_driver_lock''
> /root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:57: undefined reference
to
> `cpu_hotplug_driver_unlock''
> make: *** [.tmp_vmlinux1] Error 1
>
>
> Commenting undefined references gives an option to build the 2.6.34.7
> xenified aka Suse :-
> --- cpu.c.orig    2010-10-09 23:31:26.641580071 +0400
> +++ cpu.c    2010-10-10 00:02:40.913581011 +0400
> @@ -39,7 +39,7 @@
>      struct cpu *cpu = container_of(dev, struct cpu, sysdev);
>      ssize_t ret;
>
> -    cpu_hotplug_driver_lock();
> +    /* cpu_hotplug_driver_lock(); */
>      switch (buf[0]) {
>      case ''0'':
>          ret = cpu_down(cpu->sysdev.id);
> @@ -54,7 +54,7 @@
>      default:
>          ret = -EINVAL;
>      }
> -    cpu_hotplug_driver_unlock();
> +       /* cpu_hotplug_driver_unlock(); */
>
>      if (ret >= 0)
>          ret = count;
> @@ -85,7 +85,8 @@
>                     const char *buf,
>                     size_t count)
>  {
> -    return arch_cpu_probe(buf, count);
> +      /* return arch_cpu_probe(buf, count); */
> +      return 0;
>  }
>
>  static ssize_t cpu_release_store(struct sysdev_class *class,
> @@ -93,7 +94,8 @@
>                   const char *buf,
>                   size_t count)
>  {
> -    return arch_cpu_release(buf, count);
> +      /* return arch_cpu_release(buf, count); */
> +      return 0;
>  }
>
> Boris.
>
>
>
>
>
>
>
_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users



      

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

Andrew Lyon

2010-Oct-14 08:14 UTC

head link

Re: [Xen-users] Re: Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

Boris,

The best way to check which kernel version the patches are for is to check
the .ebuild, for example:

xen-sources-2.6.34-r3.ebuild

K_GENPATCHES_VER="9"
XENPATCHES_VER="4"

Check
http://sourcestest.gentoo.org/cgi-bin/viewvc.cgi/linux-patches/genpatches-2.6/tags/2.6.34-9and
you can see that 2.6.34.5 is the appropriate version.

New patches for .7:

http://gentoo-xen-kernel.googlecode.com/files/xen-sources-2.6.34-r4.ebuild
http://gentoo-xen-kernel.googlecode.com/files/xen-patches-2.6.34-5.tar.bz2

Andy




On Mon, Oct 11, 2010 at 9:48 AM, Boris Derzhavets
<bderzhavets@yahoo.com>wrote:
> > I''m not surprised that it fails to apply cleanly.
>
> It applies cleanly, but gives errors on kernel build in drivers/base/cpu.c
> related with HOTPLUG_CPU=y
>
> Boris.
>
> --- On *Sun, 10/10/10, Andrew Lyon <andrew.lyon@gmail.com>* wrote:
>
>
> From: Andrew Lyon <andrew.lyon@gmail.com>
> Subject: [Xen-users] Re: Attempt to apply xen-patches-2.6.34-4.tar.bz2 to
> 2.6.34.7 vanilla
> To: "Boris Derzhavets" <bderzhavets@yahoo.com>,
> xen-users@lists.xensource.com
> Date: Sunday, October 10, 2010, 3:05 PM
>
> I will upload a updated patchset tomorrow, -4 was made before .7 was
> out and I''m not surprised that it fails to apply cleanly.
>
> I''ve been too busy to update the patches much recently but I will
try
> to do them all this week.
>
> Andy
>
> On 09/10/2010, Boris Derzhavets
<bderzhavets@yahoo.com<http://mc/compose?to=bderzhavets@yahoo.com>>
> wrote:
> > . . . . . . . . . . . . . . .
> >
> >   LD      drivers/built-in.o
> >   LD      vmlinux.o
> >   MODPOST vmlinux.o
> >   GEN     .version
> >   CHK     include/generated/compile.h
> >   UPD     include/generated/compile.h
> >   CC      init/version.o
> >   LD      init/built-in.o
> >   LD      .tmp_vmlinux1
> > drivers/built-in.o: In function `cpu_release_store'':
> > /root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:96: undefined
reference
> to
> > `arch_cpu_release''
> > drivers/built-in.o: In function `cpu_probe_store'':
> > /root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:88: undefined
reference
> to
> > `arch_cpu_probe''
> > drivers/built-in.o: In function `store_online'':
> > /root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:42: undefined
reference
> to
> > `cpu_hotplug_driver_lock''
> > /root/xenkernel/linux-2.6.34.7/drivers/base/cpu.c:57: undefined
reference
> to
> > `cpu_hotplug_driver_unlock''
> > make: *** [.tmp_vmlinux1] Error 1
> >
> >
> > Commenting undefined references gives an option to build the 2.6.34.7
> > xenified aka Suse :-
> > --- cpu.c.orig    2010-10-09 23:31:26.641580071 +0400
> > +++ cpu.c    2010-10-10 00:02:40.913581011 +0400
> > @@ -39,7 +39,7 @@
> >      struct cpu *cpu = container_of(dev, struct cpu, sysdev);
> >      ssize_t ret;
> >
> > -    cpu_hotplug_driver_lock();
> > +    /* cpu_hotplug_driver_lock(); */
> >      switch (buf[0]) {
> >      case ''0'':
> >          ret = cpu_down(cpu->sysdev.id);
> > @@ -54,7 +54,7 @@
> >      default:
> >          ret = -EINVAL;
> >      }
> > -    cpu_hotplug_driver_unlock();
> > +       /* cpu_hotplug_driver_unlock(); */
> >
> >      if (ret >= 0)
> >          ret = count;
> > @@ -85,7 +85,8 @@
> >                     const char *buf,
> >                     size_t count)
> >  {
> > -    return arch_cpu_probe(buf, count);
> > +      /* return arch_cpu_probe(buf, count); */
> > +      return 0;
> >  }
> >
> >  static ssize_t cpu_release_store(struct sysdev_class *class,
> > @@ -93,7 +94,8 @@
> >                   const char *buf,
> >                   size_t count)
> >  {
> > -    return arch_cpu_release(buf, count);
> > +      /* return arch_cpu_release(buf, count); */
> > +      return 0;
> >  }
> >
> > Boris.
> >
> >
> >
> >
> >
> >
> >
>
> _______________________________________________
> Xen-users mailing list
>
Xen-users@lists.xensource.com<http://mc/compose?to=Xen-users@lists.xensource.com>
> http://lists.xensource.com/xen-users
>
>
>

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

Boris Derzhavets

2010-Oct-20 11:04 UTC

head link

[Xen-users] Build linux-image.deb via make-kpkg for 2.6.34.7 xenified aka Suse

Follow up the thread "make-kpkg (Debian based tool) does not build xen
patched".
Not sure it''s still important, however "kernel.hack"
suggested at
http://lists.xensource.com/archives/html/xen-users/2010-09/msg00394.html  by
''listmail''
worked for me after patching vanilla kernel 2.6.34.7 via 
xen-patches-2.6.34-5.tar.bz2.

Boris.






      

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

Apparently Analagous Threads

Search for more possibly parallel threads

Xen users - Nov 2009 - VIRQ_CON_RING

[Xen-devel] VIRQ_CON_RING

Re: [Xen-devel] VIRQ_CON_RING

Re: [Xen-devel] VIRQ_CON_RING

Re: [Xen-devel] VIRQ_CON_RING

[Xen-users] Failure to start X-Server for Xen on openSUSE 11.2

[Xen-users] Re: Failure to start X-Server for Xen on openSUSE 11.2

Re: [Xen-users] Re: Failure to start X-Server for Xen on openSUSE 11.2

Re: [Xen-users] Re: Failure to start X-Server for Xen on openSUSE 11.2

[Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

Re: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

RE: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

Re: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

RE: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

Re: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

Re: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

Re: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

Re: [Xen-users] Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

[Xen-users] Re: Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

Re: [Xen-users] Re: Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

Re: [Xen-users] Re: Attempt to apply xen-patches-2.6.34-4.tar.bz2 to 2.6.34.7 vanilla

[Xen-users] Build linux-image.deb via make-kpkg for 2.6.34.7 xenified aka Suse

Apparently Analagous Threads