[PATCH] fix free of event channel in blkfront

[PATCH] fix free of event channel in blkfront

[Kasai Takanori]

[-- Attachment #1: Type: text/plain, Size: 459 bytes --]

Hi All,

We tested the xm block-attach/detach command.
It repeats block-attach/detach command for DomU and pv-on-hvm on HVM Domain.
(block-attach -> block-detach -> block-attach -> block-detach -> ...)

The block-attach command failed when repeating 256 times. 
It is because the channel had not been freed in blkfront. 
Therefore, it remain using the event channel. 

This patch is corrected to free the event channel in blkfront. 

Thanks

--
Takanori Kasai

[-- Attachment #2: blkfront-free.patch --]
[-- Type: application/octet-stream, Size: 819 bytes --]

# HG changeset patch
# User kasai.takanori@jp.fujisu.com
# Node ID 70a31395fe9ea036ac887818489d0d057f4f8d22
# Parent  159ae46d5e7fdf63a1cc21b3d0f634d85bf2703f
fix free of event channel in blkfront

Signed-off-by: Takanori Kasai <kasai.takanori@jp.fujitsu.com>
Signed-off-by: Tsunehisa Doi <Doi.Tsunehisa@jp.fujitsu.com>

diff -r 159ae46d5e7f -r 70a31395fe9e linux-2.6-xen-sparse/drivers/xen/blkfront/blkfront.c
--- a/linux-2.6-xen-sparse/drivers/xen/blkfront/blkfront.c	Mon Dec 25 17:02:16 2006 +0000
+++ b/linux-2.6-xen-sparse/drivers/xen/blkfront/blkfront.c	Tue Dec 26 19:09:19 2006 +0900
@@ -384,6 +384,8 @@ static int blkfront_remove(struct xenbus
 
 	DPRINTK("blkfront_remove: %s removed\n", dev->nodename);
 
+	xenbus_free_evtchn(dev, info->evtchn);
+
 	blkif_free(info, 0);
 
 	kfree(info);

[-- Attachment #3: Type: text/plain, Size: 138 bytes --]

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

Re: [PATCH] fix free of event channel in blkfront

[Steven Hand]

>We tested the xm block-attach/detach command.
>It repeats block-attach/detach command for DomU and pv-on-hvm on HVM Domain.
>(block-attach -> block-detach -> block-attach -> block-detach -> ...)
>
>The block-attach command failed when repeating 256 times. 

I can't reproduce this - just tested 500 or so attach/detach pairs in 
a tight loop and worked for me. However I was using a regular domU - 
is it pv-on-hvm only? Can you give a more detailed repro case (or a 
script)? 



cheers,

S.

Re: [PATCH] fix free of event channel in blkfront

[Kasai Takanori]

[-- Attachment #1: Type: text/plain, Size: 661 bytes --]

Hi Steven,

>>We tested the xm block-attach/detach command.
>>It repeats block-attach/detach command for DomU and pv-on-hvm on HVM Domain.
>>(block-attach -> block-detach -> block-attach -> block-detach -> ...)
>>
>>The block-attach command failed when repeating 256 times. 
> 
> I can't reproduce this - just tested 500 or so attach/detach pairs in 
> a tight loop and worked for me. However I was using a regular domU - 
> is it pv-on-hvm only? Can you give a more detailed repro case (or a 
> script)? 

We tested only pv-on-hvm. 
The script used when testing is appended. 

We will confirm it again with latest xen-unstable.hg. 

Thanks,

--
Takanori Kasai

[-- Attachment #2: test.sh --]
[-- Type: application/octet-stream, Size: 353 bytes --]

#!/bin/sh
DomainID=6
DevID=0
Count=0

while :
do
echo "##### network-attach ##### $Count"
xm network-attach $DomainID 
sleep 1
xm network-list $DomainID
sleep 1
echo "##### network-detach #####"
xm network-detach $DomainID $DevID
Count=`expr $Count + 1`
DevID=`expr $DevID + 1`
sleep 1
xm network-list $DomainID
sleep 1
done
exit 0

[-- Attachment #3: Type: text/plain, Size: 138 bytes --]

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

Re: [PATCH] fix free of event channel in blkfront

[Kasai Takanori]

Hi Steven,

>>>We tested the xm block-attach/detach command.
>>>It repeats block-attach/detach command for DomU and pv-on-hvm on HVM Domain.
>>>(block-attach -> block-detach -> block-attach -> block-detach -> ...)
>>>
>>>The block-attach command failed when repeating 256 times.
>>
>> I can't reproduce this - just tested 500 or so attach/detach pairs in
>> a tight loop and worked for me. However I was using a regular domU -
>> is it pv-on-hvm only? Can you give a more detailed repro case (or a
>> script)?
>
> We tested only pv-on-hvm.
> The script used when testing is appended.

Sorry, the appended script was for network.
It tested in the following scripts.

-------------------------------------------------
 #!/bin/sh
DomainID=25
DevID=8448
Count=0

while :
do
echo "##### block-attach ##### $Count"
xm block-attach $DomainID phy:/dev/sdc1 hde w
sleep 1
xm block-list $DomainID
sleep 1
echo "##### block-detach #####"
xm block-detach $DomainID $DevID
Count=`expr $Count + 1`
#DevID=`expr $DevID + 1`
sleep 1
xm block-list $DomainID
sleep 1
done
exit 0
-------------------------------------------------


> We will confirm it again with latest xen-unstable.hg.

I confirmed it with up-to-date xen-unstable.hg. (cs 13282)

It did not reproduce it in DomU.
Because VBD was not able to be used, pv-on-hvm was not able to confirm it.
The following errors occur.

# insmod xen-vbd.ko
xen-vbd: Unknown symbol irq_to_evtchn_port
insmod: error inserting 'xen-vbd.ko': -1 Unknown symbol in module


When reproducing it with pv-on-hvm, evtchn had not been liberated as follows.

# xm block-attach <DomID> file:/test.img hde w
# xm block-list <DomID>
Vdev  BE handle state evt-ch ring-ref BE-path
2049    0    0     4      6      8    /local/domain/0/backend/vbd/6/2049
8448    0    0     4      7      306  /local/domain/0/backend/vbd/6/8448
# xm block-detach <DomID> 8448
# xm block-list <DomID>
Vdev  BE handle state evt-ch ring-ref BE-path
2049    0    0     4      6      8    /local/domain/0/backend/vbd/6/2049
# xm block-attach <DomID> file:/test.img hde w
# xm block-list <DomID>
Vdev  BE handle state evt-ch ring-ref BE-path
2049    0    0     4      6      8    /local/domain/0/backend/vbd/6/2049
8448    0    0     4      8      306  /local/domain/0/backend/vbd/6/8448
                         ^^^<= not 7

If the evtchn number is liberated, it is sure to become the same number.
The evtchn number keeps increase if it repeats.
I will confirm whether to be liberated as follows and correctly.

・xen-unstable.hg : cs 13192
       Revert cset 13181:f087c013cf12 since the event channel is already unbound 
via
        blkif_free() -> unbind_from_irqhandler() -> unbind_from_irq()


Thanks,

--
Takanori Kasai 

Re: [PATCH] fix free of event channel in blkfront

[Kasai Takanori]

Hi Steven,

>>>>We tested the xm block-attach/detach command.
>>>>It repeats block-attach/detach command for DomU and pv-on-hvm on HVM Domain.
>>>>(block-attach -> block-detach -> block-attach -> block-detach -> ...)
>>>>
>>>>The block-attach command failed when repeating 256 times.
>>>
>>> I can't reproduce this - just tested 500 or so attach/detach pairs in
>>> a tight loop and worked for me. However I was using a regular domU -
>>> is it pv-on-hvm only? Can you give a more detailed repro case (or a
>>> script)?

It already seems to have been corrected.

・xen-unstable.hg : cs 13198
  [PV-ON-HVM] Update evtchn interface to match new PV Linux interfaces.

----------------------------------------------------------------------------------------------------
--- a/unmodified_drivers/linux-2.6/platform-pci/evtchn.c	Sat Dec 30 18:23:27 
2006 +0000
+++ b/unmodified_drivers/linux-2.6/platform-pci/evtchn.c	Sun Dec 31 11:59:47 
2006 +0000
@@ -118,8 +144,12 @@ void unbind_from_irqhandler(unsigned int

 	mask_evtchn(evtchn);
 	evtchns[evtchn].handler = NULL;
+
+	if (evtchns[evtchn].close) {
+		struct evtchn_close close = { .port = evtchn };
+		HYPERVISOR_event_channel_op(EVTCHNOP_close, &close);
+	}
 }
-
 EXPORT_SYMBOL(unbind_from_irqhandler);

 void notify_remote_via_irq(int irq)
----------------------------------------------------------------------------------------------------


However, the following problems remain.
I think that it operates if the following problems can be solved.

It is necessary to correct pv-on-hvm according to the following corrections.

・xen-unstable.hg : cs 13197
  [LINUX] Extend the event-channel interfaces to provide helper methods


> I confirmed it with up-to-date xen-unstable.hg. (cs 13282)
>
> It did not reproduce it in DomU.
> Because VBD was not able to be used, pv-on-hvm was not able to confirm it.
> The following errors occur.
>
> # insmod xen-vbd.ko
> xen-vbd: Unknown symbol irq_to_evtchn_port
> insmod: error inserting 'xen-vbd.ko': -1 Unknown symbol in module

Thanks,

--
Takanori Kasai 

Re: [PATCH] fix free of event channel in blkfront

[tgh]

Hi
I am using the xen-3.0.3 on paravirtualVM
and I try to use xm block-attach dom6 /vm1/tp.img sda2 w
It seems to be ok, no error ,
but in the dom6 there is no sda2
and xm block-detach dom6 2050,it does not work
and xm block-list dom6 as follows ,

[root@= dev]# xm block-list dom6
Vdev BE handle state evt-ch ring-ref BE-path
2049 0 0 4 7 8 /local/domain/0/backend/vbd/127/2049
2050 0 0 3 8 665 /local/domain/0/backend/vbd/127/2050

I am confused about the function of these commands and how to use these
commands correctly

could you give me a help
Thanks in advance







Kasai Takanori 写道:
> Hi Steven,
>
>>>> We tested the xm block-attach/detach command.
>>>> It repeats block-attach/detach command for DomU and pv-on-hvm on
>>>> HVM Domain.
>>>> (block-attach -> block-detach -> block-attach -> block-detach -> ...)
>>>>
>>>> The block-attach command failed when repeating 256 times.
>>>
>>> I can't reproduce this - just tested 500 or so attach/detach pairs in
>>> a tight loop and worked for me. However I was using a regular domU -
>>> is it pv-on-hvm only? Can you give a more detailed repro case (or a
>>> script)?
>>
>> We tested only pv-on-hvm.
>> The script used when testing is appended.
>
> Sorry, the appended script was for network.
> It tested in the following scripts.
>
> -------------------------------------------------
> #!/bin/sh
> DomainID=25
> DevID=8448
> Count=0
>
> while :
> do
> echo "##### block-attach ##### $Count"
> xm block-attach $DomainID phy:/dev/sdc1 hde w
> sleep 1
> xm block-list $DomainID
> sleep 1
> echo "##### block-detach #####"
> xm block-detach $DomainID $DevID
> Count=`expr $Count + 1`
> #DevID=`expr $DevID + 1`
> sleep 1
> xm block-list $DomainID
> sleep 1
> done
> exit 0
> -------------------------------------------------
>
>
>> We will confirm it again with latest xen-unstable.hg.
>
> I confirmed it with up-to-date xen-unstable.hg. (cs 13282)
>
> It did not reproduce it in DomU.
> Because VBD was not able to be used, pv-on-hvm was not able to confirm
> it.
> The following errors occur.
>
> # insmod xen-vbd.ko
> xen-vbd: Unknown symbol irq_to_evtchn_port
> insmod: error inserting 'xen-vbd.ko': -1 Unknown symbol in module
>
>
> When reproducing it with pv-on-hvm, evtchn had not been liberated as
> follows.
>
> # xm block-attach <DomID> file:/test.img hde w
> # xm block-list <DomID>
> Vdev BE handle state evt-ch ring-ref BE-path
> 2049 0 0 4 6 8 /local/domain/0/backend/vbd/6/2049
> 8448 0 0 4 7 306 /local/domain/0/backend/vbd/6/8448
> # xm block-detach <DomID> 8448
> # xm block-list <DomID>
> Vdev BE handle state evt-ch ring-ref BE-path
> 2049 0 0 4 6 8 /local/domain/0/backend/vbd/6/2049
> # xm block-attach <DomID> file:/test.img hde w
> # xm block-list <DomID>
> Vdev BE handle state evt-ch ring-ref BE-path
> 2049 0 0 4 6 8 /local/domain/0/backend/vbd/6/2049
> 8448 0 0 4 8 306 /local/domain/0/backend/vbd/6/8448
> ^^^<= not 7
>
> If the evtchn number is liberated, it is sure to become the same number.
> The evtchn number keeps increase if it repeats.
> I will confirm whether to be liberated as follows and correctly.
>
> ・xen-unstable.hg : cs 13192
> Revert cset 13181:f087c013cf12 since the event channel is already
> unbound via
> blkif_free() -> unbind_from_irqhandler() -> unbind_from_irq()
>
>
> Thanks,
>
> -- 
> Takanori Kasai
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
>
>

Re: [PATCH] fix free of event channel in blkfront

[Kasai Takanori]

Hi

> I am using the xen-3.0.3 on paravirtualVM
> and I try to use xm block-attach dom6 /vm1/tp.img sda2 w
> It seems to be ok, no error ,
> but in the dom6 there is no sda2
> and xm block-detach dom6 2050,it does not work
> and xm block-list dom6 as follows ,

The block-attach command seems to have failed because your use is wrong.
# xm block-attach dom6 /vm1/tp.img sda2 w
                       ^^^^^^^^^^^
When backend device is a file, it is necessary to specify file.
Moreover, when it is a physical device, it is necessary to specify phy.

The example is as follows.
# xm block-attach dom6 file:/vm1/tp.img sda2 w
# xm block-attach dom6 phy:/dev/sdb sda2 w

The block-detach command works correctly, if the block-attach command succeeds.

Best Regards,

--
Takanori Kasai 

Re: [PATCH] fix free of event channel in blkfront

[tgh]

Thank you for reply
I try to do it under your guide as follows
# xm block-attach dom6 file:/vm1/template/tp.img sdb2 w
[root@genode012 vmcserver]# xm block-list dom1
Vdev BE handle state evt-ch ring-ref BE-path
2049 0 0 4 7 8 /local/domain/0/backend/vbd/102/2049
2050 0 0 4 8 815 /local/domain/0/backend/vbd/102/2050

and xm block-detach can work
but xm console dom1, then there is not the new disk in dom1
what is wrong with it?



that is the log in /var/log/xen/xend.log as follows:
[2007-02-01 14:08:47 xend 6460] DEBUG (DevController:110) DevController:
writing {'backend-id': '0', 'virtual-device': '2050''
, 'device-type': 'disk', 'state': '1', 'backend':
'/local/domain/0/backend/vbd/102/2050'} to /local/domain/102/device/vbd/2055
0.
[2007-02-01 14:08:47 xend 6460] DEBUG (DevController:112) DevController:
writing {'domain': 'dom1', 'frontend': '/local/domaii
n/102/device/vbd/2050', 'dev': 'sda2', 'state': '1', 'params':
'/vm1/template/tp1.img', 'mode': 'w', 'online': '1', 'frontendd
-id': '102', 'type': 'file'} to /local/domain/0/backend/vbd/102/2050.
[2007-02-01 14:08:47 xend 6460] DEBUG (DevController:149) Waiting for 2050.
[2007-02-01 14:08:47 xend 6460] DEBUG (DevController:464)
hotplugStatusCallback /local/domain/0/backend/vbd/102/2050/hotplug--
status.
[2007-02-01 14:08:47 xend 6460] DEBUG (DevController:464)
hotplugStatusCallback /local/domain/0/backend/vbd/102/2050/hotplug--
status.
[2007-02-01 14:08:47 xend 6460] DEBUG (DevController:478)
hotplugStatusCallback 1.


could you give me help
Thanks in advance










Kasai Takanori 写道:
> Hi
>
>> I am using the xen-3.0.3 on paravirtualVM
>> and I try to use xm block-attach dom6 /vm1/tp.img sda2 w
>> It seems to be ok, no error ,
>> but in the dom6 there is no sda2
>> and xm block-detach dom6 2050,it does not work
>> and xm block-list dom6 as follows ,
>
> The block-attach command seems to have failed because your use is wrong.
> # xm block-attach dom6 /vm1/tp.img sda2 w
> ^^^^^^^^^^^
> When backend device is a file, it is necessary to specify file.
> Moreover, when it is a physical device, it is necessary to specify phy.
>
> The example is as follows.
> # xm block-attach dom6 file:/vm1/tp.img sda2 w
> # xm block-attach dom6 phy:/dev/sdb sda2 w
>
> The block-detach command works correctly, if the block-attach command
> succeeds.
>
> Best Regards,
>
> -- 
> Takanori Kasai
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
>
>

Re: [PATCH] fix free of event channel in blkfront

[Kasai Takanori]

Hi

> I try to do it under your guide as follows
> # xm block-attach dom6 file:/vm1/template/tp.img sdb2 w
> [root@genode012 vmcserver]# xm block-list dom1
> Vdev BE handle state evt-ch ring-ref BE-path
> 2049 0 0 4 7 8 /local/domain/0/backend/vbd/102/2049
> 2050 0 0 4 8 815 /local/domain/0/backend/vbd/102/2050
> 
> and xm block-detach can work
> but xm console dom1, then there is not the new disk in dom1
> what is wrong with it?

Perhaps, it seems that xm block-attach/detach can work. 
Does the specified file have the partition? (/vm1/template/tp.img)

Please attach to sdb if there is a partition.
dom0# xm block-attach dom6 file:/vm1/template/tp.img sdb w

It is recognized with dom1 as follows.(3 partitions)
dom1# cat /proc/diskstats
   8   16 sdb 78 353 927 632 8 1 18 4 0 360 636
   8   17 sdb1 173 346 9 18
   8   18 sdb2 185 185 0 0
   8   19 sdb3 47 188 0 0
dom1# mount /dev/sdb3 /mnt
dom1# ls /mnt
bin   dev  home  lost+found  misc  net  proc  sbin     srv  tmp  var
boot  etc  lib   media       mnt   opt  root  selinux  sys  usr      


However, if SCSI driver can work on dom1, it is not possible to attach to sdx.
It is because VBD overlaps with SCSI driver's majors number. 
In that case, please attach to xvda.
dom0# xm block-attach dom6 file:/vm1/template/tp.img xvda w
 
Best Regards,

--
Takanori Kasai

Re: [PATCH] fix free of event channel in blkfront

[tgh]

Thank you for reply

I boot the VM with a img file and the conf file is as follows

kernel = "/boot/vmlinuz-2.6.16-xenU"
memory = 256
name = "dom3"
vcpus = 1
vif = [ '' ]
disk = [ 'file:/vm/dom3.img,sda1,w' ]
netmask = "255.255.0.0"
gateway = "192.168.150.1"
hostname = "office2"
ip = "192.168.104.5"
root = "/dev/sda1 ro"
extra = "4"

and if I want to attach another img file to the VM ,then
the VM boot img must have many partitions such as sda ,sda1, sda2 when
the img file is made and
xm block-attach file:/vm1/template/another.img sda1 w
then in the VM ,mount the sda1,and access it
I am not for sure whether I understand your mean
or do you mean that the another img file with I want to attach to the VM
needs have many partitions?


could you make me clear about it

Thank you in advance







Kasai Takanori 写道:
> Hi
>
>> I try to do it under your guide as follows
>> # xm block-attach dom6 file:/vm1/template/tp.img sdb2 w
>> [root@genode012 vmcserver]# xm block-list dom1
>> Vdev BE handle state evt-ch ring-ref BE-path
>> 2049 0 0 4 7 8 /local/domain/0/backend/vbd/102/2049
>> 2050 0 0 4 8 815 /local/domain/0/backend/vbd/102/2050
>>
>> and xm block-detach can work
>> but xm console dom1, then there is not the new disk in dom1
>> what is wrong with it?
>
> Perhaps, it seems that xm block-attach/detach can work. Does the
> specified file have the partition? (/vm1/template/tp.img)
>
> Please attach to sdb if there is a partition.
> dom0# xm block-attach dom6 file:/vm1/template/tp.img sdb w
>
> It is recognized with dom1 as follows.(3 partitions)
> dom1# cat /proc/diskstats
> 8 16 sdb 78 353 927 632 8 1 18 4 0 360 636
> 8 17 sdb1 173 346 9 18
> 8 18 sdb2 185 185 0 0
> 8 19 sdb3 47 188 0 0
> dom1# mount /dev/sdb3 /mnt
> dom1# ls /mnt
> bin dev home lost+found misc net proc sbin srv tmp var
> boot etc lib media mnt opt root selinux sys usr
>
> However, if SCSI driver can work on dom1, it is not possible to attach
> to sdx.
> It is because VBD overlaps with SCSI driver's majors number. In that
> case, please attach to xvda.
> dom0# xm block-attach dom6 file:/vm1/template/tp.img xvda w
>
> Best Regards,
>
> -- 
> Takanori Kasai
>
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
>
>

Re: [PATCH] fix free of event channel in blkfront

[Kasai Takanori]

Hi,


> I boot the VM with a img file and the conf file is as follows
> ...
> disk = [ 'file:/vm/dom3.img,sda1,w' ]
>
> and if I want to attach another img file to the VM ,then
> the VM boot img must have many partitions such as sda ,sda1, sda2 when
> the img file is made and
> xm block-attach file:/vm1/template/another.img sda1 w
> then in the VM ,mount the sda1,and access it
> I am not for sure whether I understand your mean
> or do you mean that the another img file with I want to attach to the VM
> needs have many partitions?

The boot image when starting has already been allocated to sda1.
(BackDev: file:/vm/dom3.img)
Therefore, it is not possible to attach to sda1 by block-attach command.
Please do not overlap with FrontEnd device that has already been used.

It is recognized as follows if there is a partition in the attached file.
# xm block-attach file:/vm1/template/another.img sdb w

  another.img file (BackDev)  |   GuestOS(FrontDev)
-------------------------------------------------------
  3 partitions(1,2,3)         |   sdb, sdb1, sdb2, sdb3
  1 partitions(1)             |   sdb, sdb1
※ 1 -> sdb1, 2 -> sdb2, 3->sdb3

It is possible to recognized as follows if there is no partition in attached 
file.
# xm block-attach file:/vm1/template/another.img sdb1 w

Best Regards,

--
Takanori Kasai

Re: [PATCH] fix free of event channel in blkfront

[tgh]

Thank you for your detail reply
Now I boot the VM with the conf file
...
disk = [ 'file:/vm/dom3.img,sda1,w' ]
...
and then commit the commands as follows

[root@genode012 vmcserver]# xm block-attach dom3 
file:/vm1/template/tp1.img sda2 w
[root@genode012 vmcserver]# xm block-list dom3
Vdev BE handle state evt-ch ring-ref BE-path
2049 0 0 4 6 8 /local/domain/0/backend/vbd/107/2049
2050 0 0 4 8 534 /local/domain/0/backend/vbd/107/2050
[root@genode012 vmcserver]# xm console dom3

[root@office2 ~]# cat /proc/diskstats
1 0 ram0 0 0 0 0 0 0 0 0 0 0 0
1 1 ram1 0 0 0 0 0 0 0 0 0 0 0
1 2 ram2 0 0 0 0 0 0 0 0 0 0 0
1 3 ram3 0 0 0 0 0 0 0 0 0 0 0
1 4 ram4 0 0 0 0 0 0 0 0 0 0 0
1 5 ram5 0 0 0 0 0 0 0 0 0 0 0
1 6 ram6 0 0 0 0 0 0 0 0 0 0 0
1 7 ram7 0 0 0 0 0 0 0 0 0 0 0
1 8 ram8 0 0 0 0 0 0 0 0 0 0 0
1 9 ram9 0 0 0 0 0 0 0 0 0 0 0
1 10 ram10 0 0 0 0 0 0 0 0 0 0 0
1 11 ram11 0 0 0 0 0 0 0 0 0 0 0
1 12 ram12 0 0 0 0 0 0 0 0 0 0 0
1 13 ram13 0 0 0 0 0 0 0 0 0 0 0
1 14 ram14 0 0 0 0 0 0 0 0 0 0 0
1 15 ram15 0 0 0 0 0 0 0 0 0 0 0
8 1 sda1 1837 184 60546 3330 1034 1155 17512 16510 0 2290 19840
9 0 md0 0 0 0 0 0 0 0 0 0 0 0
8 2 sda2 0 0 0 0 0 0 0 0 0 0 0
[root@office2 ~]# df -h
Filesystem 瀹归噺 宸茬敤 鍙敤 宸茬敤% 鎸傝浇鐐?
/dev/sda1 5.8G 4.0G 1.5G 73% /
none 128M 0 128M 0% /dev/shm
[root@office2 ~]#

what do I do wrong? How should I make it ?
I am really confused with it







Kasai Takanori 写道:
> Hi,
>
>
>> I boot the VM with a img file and the conf file is as follows
>> ...
>> disk = [ 'file:/vm/dom3.img,sda1,w' ]
>>
>> and if I want to attach another img file to the VM ,then
>> the VM boot img must have many partitions such as sda ,sda1, sda2 when
>> the img file is made and
>> xm block-attach file:/vm1/template/another.img sda1 w
>> then in the VM ,mount the sda1,and access it
>> I am not for sure whether I understand your mean
>> or do you mean that the another img file with I want to attach to the VM
>> needs have many partitions?
>
> The boot image when starting has already been allocated to sda1.
> (BackDev: file:/vm/dom3.img)
> Therefore, it is not possible to attach to sda1 by block-attach command.
> Please do not overlap with FrontEnd device that has already been used.
>
> It is recognized as follows if there is a partition in the attached file.
> # xm block-attach file:/vm1/template/another.img sdb w
>
> another.img file (BackDev) | GuestOS(FrontDev)
> -------------------------------------------------------
> 3 partitions(1,2,3) | sdb, sdb1, sdb2, sdb3
> 1 partitions(1) | sdb, sdb1
> ※ 1 -> sdb1, 2 -> sdb2, 3->sdb3
>
> It is possible to recognized as follows if there is no partition in 
> attached file.
> # xm block-attach file:/vm1/template/another.img sdb1 w
>
> Best Regards,
>
> -- 
> Takanori Kasai
>
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
>
>

Re: [PATCH] fix free of event channel in blkfront

[Kasai Takanori]

Hi,

> Now I boot the VM with the conf file
> ...
> disk = [ 'file:/vm/dom3.img,sda1,w' ]
> ...
> and then commit the commands as follows
>
> [root@genode012 vmcserver]# xm block-attach dom3 file:/vm1/template/tp1.img 
> sda2 w
> [root@genode012 vmcserver]# xm block-list dom3
> Vdev BE handle state evt-ch ring-ref BE-path
> 2049 0 0 4 6 8 /local/domain/0/backend/vbd/107/2049
> 2050 0 0 4 8 534 /local/domain/0/backend/vbd/107/2050
> [root@genode012 vmcserver]# xm console dom3
>
> [root@office2 ~]# cat /proc/diskstats
> ...
>
> 8 2 sda2 0 0 0 0 0 0 0 0 0 0 0

The block-attach command seems to have succeeded.
It can be confirmed that sda2 is attached.

It can confirm the content if the mount is done.
# mount /dev/sda2 /mnt
# ls /mnt


> [root@office2 ~]# df -h
> Filesystem /dev/sda1 5.8G 4.0G 1.5G 73% /
> none 128M 0 128M 0% /dev/shm
>
> what do I do wrong? How should I make it ?
> I am really confused with it

If the mount is done, it comes to see it.
If it is recognized when starting, the block-attach command should not be used.
Then, it is necessary to define two disks by configuration file.
When starting, automatic mount will be able to be done if it set to /etc/fstab
as well as Native Linux.

Best Regards,

--
Takanori Kasai 

Re: [PATCH] fix free of event channel in blkfront

[tgh]

Thank you for your patient and detail guide
It works as you said
there is something wrong with the img file which I want to attach to the VM
Now it works well

Thanks again






Kasai Takanori 写道:
> Hi,
>
>> Now I boot the VM with the conf file
>> ...
>> disk = [ 'file:/vm/dom3.img,sda1,w' ]
>> ...
>> and then commit the commands as follows
>>
>> [root@genode012 vmcserver]# xm block-attach dom3
>> file:/vm1/template/tp1.img sda2 w
>> [root@genode012 vmcserver]# xm block-list dom3
>> Vdev BE handle state evt-ch ring-ref BE-path
>> 2049 0 0 4 6 8 /local/domain/0/backend/vbd/107/2049
>> 2050 0 0 4 8 534 /local/domain/0/backend/vbd/107/2050
>> [root@genode012 vmcserver]# xm console dom3
>>
>> [root@office2 ~]# cat /proc/diskstats
>> ...
>>
>> 8 2 sda2 0 0 0 0 0 0 0 0 0 0 0
>
> The block-attach command seems to have succeeded.
> It can be confirmed that sda2 is attached.
>
> It can confirm the content if the mount is done.
> # mount /dev/sda2 /mnt
> # ls /mnt
>
>
>> [root@office2 ~]# df -h
>> Filesystem /dev/sda1 5.8G 4.0G 1.5G 73% /
>> none 128M 0 128M 0% /dev/shm
>>
>> what do I do wrong? How should I make it ?
>> I am really confused with it
>
> If the mount is done, it comes to see it.
> If it is recognized when starting, the block-attach command should not
> be used.
> Then, it is necessary to define two disks by configuration file.
> When starting, automatic mount will be able to be done if it set to
> /etc/fstab
> as well as Native Linux.
>
> Best Regards,
>
> -- 
> Takanori Kasai
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
>
>

Re: [PATCH] fix free of event channel in blkfront

[tgh]

hi
I try to understand the code relative to blk-attach
it seems like that device_creat()->createdevice()->addStoreEntries()
,add store entry into xenstore
then backend and frontend will watch and do several related work to real
add-device in backend and frontend ,is it right? and xenstore have
watches to check the related entry ,is it right?
I am confused ,which code or function to the real adding work
I guess there is somekind watches which do it ,but I cannot find that
fucntions

I am confused about it

could you help me
Thanks in advance

Kasai Takanori 写道:
> Hi,
>
>> Now I boot the VM with the conf file
>> ...
>> disk = [ 'file:/vm/dom3.img,sda1,w' ]
>> ...
>> and then commit the commands as follows
>>
>> [root@genode012 vmcserver]# xm block-attach dom3
>> file:/vm1/template/tp1.img sda2 w
>> [root@genode012 vmcserver]# xm block-list dom3
>> Vdev BE handle state evt-ch ring-ref BE-path
>> 2049 0 0 4 6 8 /local/domain/0/backend/vbd/107/2049
>> 2050 0 0 4 8 534 /local/domain/0/backend/vbd/107/2050
>> [root@genode012 vmcserver]# xm console dom3
>>
>> [root@office2 ~]# cat /proc/diskstats
>> ...
>>
>> 8 2 sda2 0 0 0 0 0 0 0 0 0 0 0
>
> The block-attach command seems to have succeeded.
> It can be confirmed that sda2 is attached.
>
> It can confirm the content if the mount is done.
> # mount /dev/sda2 /mnt
> # ls /mnt
>
>
>> [root@office2 ~]# df -h
>> Filesystem /dev/sda1 5.8G 4.0G 1.5G 73% /
>> none 128M 0 128M 0% /dev/shm
>>
>> what do I do wrong? How should I make it ?
>> I am really confused with it
>
> If the mount is done, it comes to see it.
> If it is recognized when starting, the block-attach command should not
> be used.
> Then, it is necessary to define two disks by configuration file.
> When starting, automatic mount will be able to be done if it set to
> /etc/fstab
> as well as Native Linux.
>
> Best Regards,
>
> -- 
> Takanori Kasai
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
>
>