* [IA64] Weekly benchmark results [2010ww36]
@ 2010-09-10 8:19 KUWAMURA Shin'ya
2010-09-10 17:51 ` Ian Jackson
0 siblings, 1 reply; 7+ messages in thread
From: KUWAMURA Shin'ya @ 2010-09-10 8:19 UTC (permalink / raw)
To: xen-ia64-devel; +Cc: xen-devel
Hi,
I report a benchmark result of this week on IPF against xen-unstable.
- The test environment went back to RHEL4U2.
- Linux-2.6.18-xen cannot be built:
In file included from /linux-2.6.18-xen.hg/arch/ia64/xen/xcom_privcmd.c:27:
/linux-2.6.18-xen.hg/include/xen/interface/domctl.h:284: error: field `cpumap' has incomplete type
This issue was fixed ad hoc.
- Tools cannot be built:
ia64/acpi/platform/aclinux.h:50:24: error: xen/config.h: No such file or directory
I have already sent the patch.
- On RHEL4U2, libxl cannot be built:
In file included from xl_cmdtable.c:17:
libxl.h:188: error: syntax error before "libxl_version_info"
Because gentypes.py uses Python 2.4 feature. This issue was fixed ad hoc.
# Notice: RHEL4U2 uses Python 2.3.
- On RHEL4U2, starting xend failed, since it uses @staticmethod and
@staticmethod is not supported by Python 2.3.
- Creating domains failed with the following error:
Error: Moving domain to target pool failed
I will send the patch.
TEST ENVIRONMENT
Machine : Tiger4
Kernel : 2.6.18.8-xen
Changeset : 22101:1831912d4109 (xen-unstable)
1029:9b1adfb8b0b3 (linux-2.6.18-xen)
137:3ad73b4314e3 (efi-vfirmware)
6abea57 (qemu-xen-unstable)
Dom0 OS : RHEL4 U2 (2P)
DomU OS : RHEL4 U2 (8P, using tap:aio)
DomVTi OS : RHEL4 U2 (8P, with PV-on-HVM drivers)
TEST RESULTS
DomU:
unixbench4.1.0 : Pass
bonnie++-1.03 : Pass
ltp-full-20070930 : Pass
iozone3_191 : Pass
lmbench-3.0-a5 : Pass
DomVTi:
unixbench4.1.0 : Pass
bonnie++-1.03 : Pass
ltp-full-20070930 : Pass
iozone3_191 : Pass
lmbench-3.0-a5 : Pass
Best regards,
KUWAMURA Shin'ya and FUJITSU members
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [IA64] Weekly benchmark results [2010ww36]
2010-09-10 8:19 [IA64] Weekly benchmark results [2010ww36] KUWAMURA Shin'ya
@ 2010-09-10 17:51 ` Ian Jackson
2010-09-13 8:07 ` [PATCH] [Linux] ia64, xencomm: fix 1028:6f7bda25a4de (Re: [Xen-devel] [IA64] Weekly benchmark results [2010ww36]) KUWAMURA Shin'ya
0 siblings, 1 reply; 7+ messages in thread
From: Ian Jackson @ 2010-09-10 17:51 UTC (permalink / raw)
To: KUWAMURA Shin'ya; +Cc: xen-devel, xen-ia64-devel
KUWAMURA Shin'ya writes ("[Xen-devel] [IA64] Weekly benchmark results [2010ww36]"):
> - Linux-2.6.18-xen cannot be built:
> In file included from /linux-2.6.18-xen.hg/arch/ia64/xen/xcom_privcmd.c:27:
> /linux-2.6.18-xen.hg/include/xen/interface/domctl.h:284: error: field `cpumap' has incomplete type
> This issue was fixed ad hoc.
If you'd like to send the ad-hoc fixes you used, that would be very
helpful. If they're suitable for inclusion we'll commit them, and if
not they'll be good explanations of the bugs.
Thanks,
Ian.
^ permalink raw reply [flat|nested] 7+ messages in thread
* [PATCH] [Linux] ia64, xencomm: fix 1028:6f7bda25a4de (Re: [Xen-devel] [IA64] Weekly benchmark results [2010ww36])
2010-09-10 17:51 ` Ian Jackson
@ 2010-09-13 8:07 ` KUWAMURA Shin'ya
2010-09-13 8:38 ` [PATCH] [Linux] ia64, xencomm: fix 1028:6f7bda25a4de (Re: " Jan Beulich
0 siblings, 1 reply; 7+ messages in thread
From: KUWAMURA Shin'ya @ 2010-09-13 8:07 UTC (permalink / raw)
To: Ian.Jackson; +Cc: xen-devel, xen-ia64-devel
[-- Attachment #1: Type: Text/Plain, Size: 1870 bytes --]
Hi Ian,
>>>>> On Fri, 10 Sep 2010 18:51:23 +0100
>>>>> Ian.Jackson@eu.citrix.com(Ian Jackson) said:
>
> KUWAMURA Shin'ya writes ("[Xen-devel] [IA64] Weekly benchmark results [2010ww36]"):
> > - Linux-2.6.18-xen cannot be built:
> > In file included from /linux-2.6.18-xen.hg/arch/ia64/xen/xcom_privcmd.c:27:
> > /linux-2.6.18-xen.hg/include/xen/interface/domctl.h:284: error: field `cpumap' has incomplete type
> > This issue was fixed ad hoc.
>
> If you'd like to send the ad-hoc fixes you used, that would be very
> helpful. If they're suitable for inclusion we'll commit them, and if
> not they'll be good explanations of the bugs.
Thank you for your advice.
I attached the patch. The cause is as follows:
1028:6f7bda25a4de includes xen-unstable 21568:05bfc5a472b which
moved struct xenctl_cpumap from domctl.h to xen.h.
However, xcom_privcmd.c includes xen.h without __XEN__ or
__XEN_TOOLS__, so struct xenctl_cpumap is invisible from
xcom_privcmd.c.
This patch fixes by defining __XEN_TOOLS__ at the top of xcom_privcmd.c.
But this causes a warning:
In file included from /linux-2.6.18-xen.hg/include/xen/interface/xen.h:30,
from /linux-2.6.18-xen.hg/include/xen/interface/arch-ia64.h:26,
from include2/asm/xen/privop.h:16,
from include2/asm/privop.h:14,
from include2/asm/intrinsics.h:189,
from include2/asm/bitops.h:14,
from /linux-2.6.18-xen.hg/include/linux/bitops.h:9,
from /linux-2.6.18-xen.hg/include/linux/kernel.h:15,
from /linux-2.6.18-xen.hg/arch/ia64/xen/xcom_privcmd.c:22:
/linux-2.6.18-xen.hg/include/xen/interface/xen-compat.h:34:1: warning: "__XEN_INTERFACE_VERSION__" redefined
<command line>:5:1: warning: this is the location of the previous definition
Best regards,
--
KUWAMURA Shin'ya
[-- Attachment #2: fix-linux-xenctl.patch --]
[-- Type: Text/X-Patch, Size: 1240 bytes --]
# HG changeset patch
# User KUWAMURA Shin'ya <kuwa@jp.fujitsu.com>
# Date 1284363596 -32400
# Node ID c92a45a06e995eadd874957fc9e857c7c169fc0a
# Parent 9b1adfb8b0b3b37c13f06c0adb8dd17b2a0a077d
ia64, xencomm: fix 1028:6f7bda25a4de
This patch fixes the following errors in linux-2.6.18-xen.hg:
In file included from /linux-2.6.18-xen.hg/arch/ia64/xen/xcom_privcmd.c:27:
/linux-2.6.18-xen.hg/include/xen/interface/domctl.h:284: error: field 'cpumap' has incomplete type
In file included from /linux-2.6.18-xen.hg/arch/ia64/xen/xcom_privcmd.c:28:
/linux-2.6.18-xen.hg/include/xen/interface/sysctl.h:74: error: field 'cpu_mask' has incomplete type
/linux-2.6.18-xen.hg/include/xen/interface/sysctl.h:540: error: field 'cpumap' has incomplete type
Signed-off-by: KUWAMURA Shin'ya <kuwa@jp.fujitsu.com>
diff -r 9b1adfb8b0b3 -r c92a45a06e99 arch/ia64/xen/xcom_privcmd.c
--- a/arch/ia64/xen/xcom_privcmd.c Thu Aug 26 11:27:25 2010 +0100
+++ b/arch/ia64/xen/xcom_privcmd.c Mon Sep 13 16:39:56 2010 +0900
@@ -16,6 +16,7 @@
* Authors: Hollis Blanchard <hollisb@us.ibm.com>
* Tristan Gingold <tristan.gingold@bull.net>
*/
+#define __XEN_TOOLS__
#include <linux/types.h>
#include <linux/errno.h>
#include <linux/kernel.h>
[-- Attachment #3: Type: text/plain, Size: 152 bytes --]
_______________________________________________
Xen-ia64-devel mailing list
Xen-ia64-devel@lists.xensource.com
http://lists.xensource.com/xen-ia64-devel
^ permalink raw reply [flat|nested] 7+ messages in thread
* [PATCH] [Linux] ia64, xencomm: fix 1028:6f7bda25a4de (Re: [IA64] Weekly benchmark results [2010ww36])
2010-09-13 8:07 ` [PATCH] [Linux] ia64, xencomm: fix 1028:6f7bda25a4de (Re: [Xen-devel] [IA64] Weekly benchmark results [2010ww36]) KUWAMURA Shin'ya
@ 2010-09-13 8:38 ` Jan Beulich
2010-09-14 7:39 ` [PATCH] [Linux] ia64, xencomm: fix 1028:6f7bda25a4de KUWAMURA Shin'ya
[not found] ` <m2n.s.1OvQ9b-0017xQ@chiark.greenend.org.uk>
0 siblings, 2 replies; 7+ messages in thread
From: Jan Beulich @ 2010-09-13 8:38 UTC (permalink / raw)
To: KUWAMURA Shin'ya; +Cc: xen-devel, Ian.Jackson, xen-ia64-devel
>>> On 13.09.10 at 10:07, "KUWAMURA Shin'ya" <kuwa@jp.fujitsu.com> wrote:
> Hi Ian,
>
>>>>>> On Fri, 10 Sep 2010 18:51:23 +0100
>>>>>> Ian.Jackson@eu.citrix.com(Ian Jackson) said:
>>
>> KUWAMURA Shin'ya writes ("[Xen-devel] [IA64] Weekly benchmark results
> [2010ww36]"):
>> > - Linux-2.6.18-xen cannot be built:
>> > In file included from
> /linux-2.6.18-xen.hg/arch/ia64/xen/xcom_privcmd.c:27:
>> > /linux-2.6.18-xen.hg/include/xen/interface/domctl.h:284: error: field
> `cpumap' has incomplete type
>> > This issue was fixed ad hoc.
>>
>> If you'd like to send the ad-hoc fixes you used, that would be very
>> helpful. If they're suitable for inclusion we'll commit them, and if
>> not they'll be good explanations of the bugs.
>
> Thank you for your advice.
> I attached the patch. The cause is as follows:
>
> 1028:6f7bda25a4de includes xen-unstable 21568:05bfc5a472b which
> moved struct xenctl_cpumap from domctl.h to xen.h.
> However, xcom_privcmd.c includes xen.h without __XEN__ or
> __XEN_TOOLS__, so struct xenctl_cpumap is invisible from
> xcom_privcmd.c.
>
> This patch fixes by defining __XEN_TOOLS__ at the top of xcom_privcmd.c.
> But this causes a warning:
>
> In file included from /linux-2.6.18-xen.hg/include/xen/interface/xen.h:30,
> from
> /linux-2.6.18-xen.hg/include/xen/interface/arch-ia64.h:26,
> from include2/asm/xen/privop.h:16,
> from include2/asm/privop.h:14,
> from include2/asm/intrinsics.h:189,
> from include2/asm/bitops.h:14,
> from /linux-2.6.18-xen.hg/include/linux/bitops.h:9,
> from /linux-2.6.18-xen.hg/include/linux/kernel.h:15,
> from /linux-2.6.18-xen.hg/arch/ia64/xen/xcom_privcmd.c:22:
> /linux-2.6.18-xen.hg/include/xen/interface/xen-compat.h:34:1: warning:
> "__XEN_INTERFACE_VERSION__" redefined
> <command line>:5:1: warning: this is the location of the previous definition
This doesn't seem to be the right (or a sufficient) fix then.
Even more, a few lines down in the same source file __XEN__
already gets #define-d, so the first choice imo would be to simply
move that definition up. Or does this cause any *more* problems
than the warning above (which I think needs to be dealt with
regardless)?
Jan
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH] [Linux] ia64, xencomm: fix 1028:6f7bda25a4de
2010-09-13 8:38 ` [PATCH] [Linux] ia64, xencomm: fix 1028:6f7bda25a4de (Re: " Jan Beulich
@ 2010-09-14 7:39 ` KUWAMURA Shin'ya
[not found] ` <m2n.s.1OvQ9b-0017xQ@chiark.greenend.org.uk>
1 sibling, 0 replies; 7+ messages in thread
From: KUWAMURA Shin'ya @ 2010-09-14 7:39 UTC (permalink / raw)
To: JBeulich; +Cc: xen-devel, Ian.Jackson, xen-ia64-devel
[-- Attachment #1: Type: Text/Plain, Size: 1889 bytes --]
Hi Jan,
>>>>> On Mon, 13 Sep 2010 09:38:50 +0100
>>>>> JBeulich@novell.com("Jan Beulich") said:
>
> >>> On 13.09.10 at 10:07, "KUWAMURA Shin'ya" <kuwa@jp.fujitsu.com> wrote:
> > This patch fixes by defining __XEN_TOOLS__ at the top of xcom_privcmd.c.
> > But this causes a warning:
> >
> > In file included from /linux-2.6.18-xen.hg/include/xen/interface/xen.h:30,
> > from
> > /linux-2.6.18-xen.hg/include/xen/interface/arch-ia64.h:26,
> > from include2/asm/xen/privop.h:16,
> > from include2/asm/privop.h:14,
> > from include2/asm/intrinsics.h:189,
> > from include2/asm/bitops.h:14,
> > from /linux-2.6.18-xen.hg/include/linux/bitops.h:9,
> > from /linux-2.6.18-xen.hg/include/linux/kernel.h:15,
> > from /linux-2.6.18-xen.hg/arch/ia64/xen/xcom_privcmd.c:22:
> > /linux-2.6.18-xen.hg/include/xen/interface/xen-compat.h:34:1: warning:
> > "__XEN_INTERFACE_VERSION__" redefined
> > <command line>:5:1: warning: this is the location of the previous definition
> This doesn't seem to be the right (or a sufficient) fix then.
>
> Even more, a few lines down in the same source file __XEN__
> already gets #define-d, so the first choice imo would be to simply
> move that definition up. Or does this cause any *more* problems
> than the warning above (which I think needs to be dealt with
> regardless)?
Thank you for your advice.
I fixed this warning by adding an #include line. I attached the patch.
Because xen-compat.h defines __XEN_INTERFACE_VERSION__ only if __XEN__
or __XEN_TOOLS__ is defined, I added #include xen-compat.h before
#define __XEN_TOOLS__.
I confirmed that dom0 kernel could be built without warnings and
guests could be created.
Signed-off-by: KUWAMURA Shin'ya <kuwa@jp.fujitsu.com>
Best regards,
--
KUWAMURA Shin'ya
[-- Attachment #2: fix-linux-xenctl-v2.patch --]
[-- Type: Text/X-Patch, Size: 453 bytes --]
diff -r 9b1adfb8b0b3 arch/ia64/xen/xcom_privcmd.c
--- a/arch/ia64/xen/xcom_privcmd.c Thu Aug 26 11:27:25 2010 +0100
+++ b/arch/ia64/xen/xcom_privcmd.c Tue Sep 14 16:22:34 2010 +0900
@@ -16,6 +16,8 @@
* Authors: Hollis Blanchard <hollisb@us.ibm.com>
* Tristan Gingold <tristan.gingold@bull.net>
*/
+#include <xen/interface/xen-compat.h>
+#define __XEN_TOOLS__
#include <linux/types.h>
#include <linux/errno.h>
#include <linux/kernel.h>
[-- Attachment #3: Type: text/plain, Size: 152 bytes --]
_______________________________________________
Xen-ia64-devel mailing list
Xen-ia64-devel@lists.xensource.com
http://lists.xensource.com/xen-ia64-devel
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH] [Linux] ia64, xencomm: fix 1028:6f7bda25a4de
[not found] ` <m2n.s.1OvQ9b-0017xQ@chiark.greenend.org.uk>
@ 2010-09-16 17:08 ` Ian Jackson
2010-09-17 1:17 ` KUWAMURA Shin'ya
0 siblings, 1 reply; 7+ messages in thread
From: Ian Jackson @ 2010-09-16 17:08 UTC (permalink / raw)
To: KUWAMURA Shin'ya; +Cc: xen-devel, JBeulich, xen-ia64-devel
KUWAMURA Shin'ya writes ("[Xen-devel] Re: [PATCH] [Linux] ia64, xencomm: fix 1028:6f7bda25a4de"):
> On Mon, 13 Sep 2010 09:38:50 +0100
> JBeulich@novell.com("Jan Beulich") said:
> > Even more, a few lines down in the same source file __XEN__
> > already gets #define-d, so the first choice imo would be to simply
> > move that definition up. Or does this cause any *more* problems
> > than the warning above (which I think needs to be dealt with
> > regardless)?
Jan's suggestion seems sensible.
> Thank you for your advice.
>
> I fixed this warning by adding an #include line. I attached the patch.
...
> +#include <xen/interface/xen-compat.h>
> +#define __XEN_TOOLS__
Why do you choose to #define __XEN_TOOLS__ rather than __XEN__ (which
latter is already defined in this file, as Jan notes) ?
Does moving the definition of __XEN__ further up cause other problems ?
Ian.
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: Re: [PATCH] [Linux] ia64, xencomm: fix 1028:6f7bda25a4de
2010-09-16 17:08 ` Ian Jackson
@ 2010-09-17 1:17 ` KUWAMURA Shin'ya
0 siblings, 0 replies; 7+ messages in thread
From: KUWAMURA Shin'ya @ 2010-09-17 1:17 UTC (permalink / raw)
To: Ian.Jackson; +Cc: xen-devel, JBeulich, xen-ia64-devel
Hi Ian,
>>>>> On Thu, 16 Sep 2010 18:08:12 +0100
>>>>> Ian.Jackson@eu.citrix.com(Ian Jackson) said:
>
> KUWAMURA Shin'ya writes ("[Xen-devel] Re: [PATCH] [Linux] ia64, xencomm: fix 1028:6f7bda25a4de"):
> > I fixed this warning by adding an #include line. I attached the patch.
> ...
> > +#include <xen/interface/xen-compat.h>
> > +#define __XEN_TOOLS__
>
> Why do you choose to #define __XEN_TOOLS__ rather than __XEN__ (which
> latter is already defined in this file, as Jan notes) ?
> Does moving the definition of __XEN__ further up cause other problems ?
__XEN__ causes the following compilation error:
In file included from include2/asm/page.h:236,
from include2/asm/ptrace.h:234,
from include2/asm/processor.h:19,
from include2/asm/thread_info.h:11,
from /linux-2.6.18-xen.hg/include/linux/thread_info.h:21,
from /linux-2.6.18-xen.hg/include/linux/preempt.h:9,
from /linux-2.6.18-xen.hg/include/linux/spinlock.h:49,
from /linux-2.6.18-xen.hg/include/linux/mmzone.h:7,
from /linux-2.6.18-xen.hg/include/linux/gfp.h:4,
from /linux-2.6.18-xen.hg/arch/ia64/xen/xcom_privcmd.c:23:
include2/asm/hypervisor.h:63: error: expected '=', ',', ';', 'asm' or '__attribute__' before '*' token
The line of the problem is as follows:
extern shared_info_t *HYPERVISOR_shared_info;
The identifier `shared_info_t' is defined in include/xen/interface/xen.h
if __XEN__ is NOT defined:
include/xen/interface/xen.h:
#ifndef __XEN__
typedef struct shared_info shared_info_t;
#endif
Best regards,
--
KUWAMURA Shin'ya
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2010-09-17 1:17 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2010-09-10 8:19 [IA64] Weekly benchmark results [2010ww36] KUWAMURA Shin'ya
2010-09-10 17:51 ` Ian Jackson
2010-09-13 8:07 ` [PATCH] [Linux] ia64, xencomm: fix 1028:6f7bda25a4de (Re: [Xen-devel] [IA64] Weekly benchmark results [2010ww36]) KUWAMURA Shin'ya
2010-09-13 8:38 ` [PATCH] [Linux] ia64, xencomm: fix 1028:6f7bda25a4de (Re: " Jan Beulich
2010-09-14 7:39 ` [PATCH] [Linux] ia64, xencomm: fix 1028:6f7bda25a4de KUWAMURA Shin'ya
[not found] ` <m2n.s.1OvQ9b-0017xQ@chiark.greenend.org.uk>
2010-09-16 17:08 ` Ian Jackson
2010-09-17 1:17 ` KUWAMURA Shin'ya
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.