[PATCH] Parallel/Serial/Monitor/pyGRUB ReiserFS/others support/fixes

[PATCH] Parallel/Serial/Monitor/pyGRUB ReiserFS/others support/fixes

[Daniel Kiper]

[-- Attachment #1: Type: text/plain, Size: 2278 bytes --]

Hello,

Parallel/Serial/Monitor/pyGRUB ReiserFS/others support/fixes
for xen-3.4-testing (tested with Xen Ver. 3.4.2) and xen-unstable
(not tested; Sorry however I do not have spare machine to compile
and test them).

Details are below:

xen-3.4-testing-hvmloader-strip_20091222.patch:
  Strip hvmloader after compile. dh_strip from debhelper
  does not strip binaries without execute attribute.

xen-3.4-testing-parallel-serial-monitor_20091222.patch:
  Add option for parallel port configuration in PV/HVM mode
  (PV is prepared for further development;
  HVM is fully functional).
  Add support for serial port (console)
  configuration in PV mode.
  Add support for QEMU monitor configuration in PV/HVM mode
  (previous code did not allowed disabling of QEMU monitor
  or redirection QEMU monitor to other device).
  This patch does not change default behavior of Xen domU.

xen-3.4-testing-pygrub-reiserfs_20091222.patch:
  Small but important correction for pyGRUB ReiserFS support.
  Without this patch pyGRUB could not read ReiserFS.

xen-3.4-testing-videoram_20091222.patch:
  Remove videoram option from vfb config.
  This option is only valid in main config.

xen-unstable-hvmloader-strip_20091222.patch:
  Strip hvmloader after compile. dh_strip from debhelper
  does not strip binaries without execute attribute.

xen-unstable-parallel-serial-monitor_20091223.patch:
  Add option for parallel port configuration in PV/HVM mode
  (PV is prepared for further development;
  HVM is fully functional).
  Add support for serial port (console)
  configuration in PV mode.
  Add support for QEMU monitor configuration in PV/HVM mode
  (previous code did not allowed disabling of QEMU monitor
  or redirection QEMU monitor to other device).
  This patch does not change default behavior of Xen domU.

xen-unstable-pygrub-reiserfs_20091222.patch:
  Small but important correction for pyGRUB ReiserFS support.
  Without this patch pyGRUB could not read ReiserFS.

xen-unstable-videoram_20091222.patch:
  Remove videoram option from vfb config.
  This option is only valid in main config.

Those patches listed above are also available for Xen Ver. 3.4.2:
ftp://ftp.local.net-space.pl/pub/net-space/xen/xen-3.4.2/

If you have a questions please drop me a line.

Daniel Kiper

[-- Attachment #2: xen-3.4-testing-patch_20091223.tar.bz2 --]
[-- Type: application/octet-stream, Size: 3963 bytes --]

[-- Attachment #3: xen-unstable-patch_20091223.tar.bz2 --]
[-- Type: application/octet-stream, Size: 3821 bytes --]

[-- Attachment #4: Type: text/plain, Size: 138 bytes --]

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

Re: [PATCH] Parallel/Serial/Monitor/pyGRUB ReiserFS/others support/fixes

[Keir Fraser]

On 23/12/2009 00:54, "Daniel Kiper" <dkiper@net-space.pl> wrote:

> xen-unstable-hvmloader-strip_20091222.patch:
>   Strip hvmloader after compile. dh_strip from debhelper
>   does not strip binaries without execute attribute.

I don't see the point of doing this. We don't in general strip stuff we
build under tools/, the extra disc space taken is trivial, and symbols can
be convenient to have around. I just don't see the win.
 - Not applying to xen-unstable or xen-3.4

> xen-unstable-parallel-serial-monitor_20091223.patch:

Xend parts look good. The hvmloader parts look unnecessary (since we
currently always declare LPT1 present) and dangerous (Windows guests which
were installed on old Xen iirc get upset if they later boot and see LTP1
not-present in ACPI tables). There'd need to be a *good* reason for hiding
LTP1 in DSDT therefore.
 - Partially applied to unstable; not applied to 3.4

> xen-unstable-pygrub-reiserfs_20091222.patch:
>   Small but important correction for pyGRUB ReiserFS support.
>   Without this patch pyGRUB could not read ReiserFS.

Looks good.
 - Applied to unstable and 3.4.

> xen-unstable-videoram_20091222.patch:
>   Remove videoram option from vfb config.
>   This option is only valid in main config.

Looks good but not critical for 3.4.
 - Applied to unstable; not applied to 3.4.

 -- Keir

Re: [PATCH] Parallel/Serial/Monitor/pyGRUB ReiserFS/others support/fixes

[Ian Campbell]

On Wed, 2009-12-23 at 07:50 +0000, Keir Fraser wrote:
> 
> > xen-unstable-hvmloader-strip_20091222.patch:
> >   Strip hvmloader after compile. dh_strip from debhelper
> >   does not strip binaries without execute attribute.
> 
> I don't see the point of doing this. We don't in general strip stuff
> we build under tools/, the extra disc space taken is trivial, and
> symbols can be convenient to have around. I just don't see the win.
>  - Not applying to xen-unstable or xen-3.4 

Yes, it is the packager's responsibility to take care of this if they
deem it necessary, either manually or by adding +x so that dh_strip
takes care of it.

Ian.

Re: [PATCH] Parallel/Serial/Monitor/pyGRUB ReiserFS/others support/fixes

[Keir Fraser]

On 23/12/2009 08:39, "Ian Campbell" <Ian.Campbell@eu.citrix.com> wrote:

>> I don't see the point of doing this. We don't in general strip stuff
>> we build under tools/, the extra disc space taken is trivial, and
>> symbols can be convenient to have around. I just don't see the win.
>>  - Not applying to xen-unstable or xen-3.4
> 
> Yes, it is the packager's responsibility to take care of this if they
> deem it necessary, either manually or by adding +x so that dh_strip
> takes care of it.

Actually hvmloader seems to be a+x already, in all my build trees.

 -- Keir

Re: [PATCH] Parallel/Serial/Monitor/pyGRUB ReiserFS/others support/fixes

[Daniel Kiper]

Hello,

>>> > xen-unstable-hvmloader-strip_20091222.patch:
>>> >   Strip hvmloader after compile. dh_strip from debhelper
>>> >   does not strip binaries without execute attribute.
>>>
>>> I don't see the point of doing this. We don't in general strip stuff
>>> we build under tools/, the extra disc space taken is trivial, and
>>> symbols can be convenient to have around. I just don't see the win.
>>>  - Not applying to xen-unstable or xen-3.4
>>
>> Yes, it is the packager's responsibility to take care of this if they
>> deem it necessary, either manually or by adding +x so that dh_strip
>> takes care of it.
>
> Actually hvmloader seems to be a+x already, in all my build trees.

FYI:
xen-3.4.2/tools/firmware/hvmloader/hvmloader: 0755
xen-3.4.2/dist/install/usr/lib/xen/boot/hvmloader: 0644

OK, I agree that this patch break general
rule and should not be applied.

> > xen-unstable-parallel-serial-monitor_20091223.patch:
>
> Xend parts look good. The hvmloader parts look unnecessary (since we
> currently always declare LPT1 present) and dangerous (Windows guests which
> were installed on old Xen iirc get upset if they later boot and see LTP1
> not-present in ACPI tables). There'd need to be a *good* reason for hiding
> LTP1 in DSDT therefore.
>  - Partially applied to unstable; not applied to 3.4

1) I think that parallel port configuration should be similar to
   serial port configuration. With that idea in mind this patch was
   created (and some code were copied from serial config).
2) I tested this patch with Windows Vista Business SP2 and
   Finnix Ver. 92.1 as a guests and everything worked without
   any problem.
3) I do not like that broken/disabled devices appear in config/device manager
   without any warning. In this situation disabled LPT1 port appear in
   Windows Device Manager without any warning. I think it is confusing
   for users when they see working devices in WDM and they could not use
   them.
4) As I mentioned earlier, all patches (including this one) do not
   change default behavior of Xen. It means that default QEMU configuration
   is not changed when parallel option is not specified in config file.
   If somebody would like to disable LPT1, he/she should have possibility
   to do that in proper way. I think it is another question that he/she
   is aware what is he/she doing.

That is why I think xen-unstable-parallel-serial-monitor_20091223.patch
patch should be fully applied.

Daniel Kiper

RE: [PATCH] Parallel/Serial/Monitor/pyGRUB ReiserFS/others support/fixes

[Xu, Dongxiao]

Hi, Daniel, 
	Could you help to check your patch which is checked in as C/S 20717? 
It seems that VM save/restore/migration and reboot is broken since this changeset.

Thanks!
Dongxiao

Daniel Kiper wrote:
> Hello,
> 
>>>>> xen-unstable-hvmloader-strip_20091222.patch:
>>>>>   Strip hvmloader after compile. dh_strip from debhelper
>>>>>   does not strip binaries without execute attribute.
>>>> 
>>>> I don't see the point of doing this. We don't in general strip
>>>> stuff we build under tools/, the extra disc space taken is
>>>> trivial, and symbols can be convenient to have around. I just
>>>>  don't see the win. - Not applying to xen-unstable or xen-3.4
>>> 
>>> Yes, it is the packager's responsibility to take care of this if
>>> they deem it necessary, either manually or by adding +x so that
>>> dh_strip takes care of it.
>> 
>> Actually hvmloader seems to be a+x already, in all my build trees.
> 
> FYI:
> xen-3.4.2/tools/firmware/hvmloader/hvmloader: 0755
> xen-3.4.2/dist/install/usr/lib/xen/boot/hvmloader: 0644
> 
> OK, I agree that this patch break general
> rule and should not be applied.
> 
>>> xen-unstable-parallel-serial-monitor_20091223.patch:
>> 
>> Xend parts look good. The hvmloader parts look unnecessary (since we
>> currently always declare LPT1 present) and dangerous (Windows guests
>> which were installed on old Xen iirc get upset if they later boot
>> and see LTP1 not-present in ACPI tables). There'd need to be a
>>  *good* reason for hiding LTP1 in DSDT therefore. - Partially
>> applied to unstable; not applied to 3.4 
> 
> 1) I think that parallel port configuration should be similar to
>    serial port configuration. With that idea in mind this patch was
>    created (and some code were copied from serial config).
> 2) I tested this patch with Windows Vista Business SP2 and
>    Finnix Ver. 92.1 as a guests and everything worked without
>    any problem.
> 3) I do not like that broken/disabled devices appear in config/device
>    manager without any warning. In this situation disabled LPT1 port
>    appear in Windows Device Manager without any warning. I think it
>    is confusing for users when they see working devices in WDM and
>    they could not use them.
> 4) As I mentioned earlier, all patches (including this one) do not
>    change default behavior of Xen. It means that default QEMU
>    configuration is not changed when parallel option is not specified
>    in config file. If somebody would like to disable LPT1, he/she
>    should have possibility to do that in proper way. I think it is
>    another question that he/she is aware what is he/she doing.
> 
> That is why I think
> xen-unstable-parallel-serial-monitor_20091223.patch patch should be
> fully applied. 
> 
> Daniel Kiper
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel

Re: [PATCH] Parallel/Serial/Monitor/pyGRUB ReiserFS/others support/fixes

[Keir Fraser]

Hopefully I've fixed this with c/s 20719.

 -- Keir

On 24/12/2009 07:43, "Xu, Dongxiao" <dongxiao.xu@intel.com> wrote:

> Hi, Daniel, 
> Could you help to check your patch which is checked in as C/S 20717?
> It seems that VM save/restore/migration and reboot is broken since this
> changeset.
> 
> Thanks!
> Dongxiao
> 
> Daniel Kiper wrote:
>> Hello,
>> 
>>>>>> xen-unstable-hvmloader-strip_20091222.patch:
>>>>>>   Strip hvmloader after compile. dh_strip from debhelper
>>>>>>   does not strip binaries without execute attribute.
>>>>> 
>>>>> I don't see the point of doing this. We don't in general strip
>>>>> stuff we build under tools/, the extra disc space taken is
>>>>> trivial, and symbols can be convenient to have around. I just
>>>>>  don't see the win. - Not applying to xen-unstable or xen-3.4
>>>> 
>>>> Yes, it is the packager's responsibility to take care of this if
>>>> they deem it necessary, either manually or by adding +x so that
>>>> dh_strip takes care of it.
>>> 
>>> Actually hvmloader seems to be a+x already, in all my build trees.
>> 
>> FYI:
>> xen-3.4.2/tools/firmware/hvmloader/hvmloader: 0755
>> xen-3.4.2/dist/install/usr/lib/xen/boot/hvmloader: 0644
>> 
>> OK, I agree that this patch break general
>> rule and should not be applied.
>> 
>>>> xen-unstable-parallel-serial-monitor_20091223.patch:
>>> 
>>> Xend parts look good. The hvmloader parts look unnecessary (since we
>>> currently always declare LPT1 present) and dangerous (Windows guests
>>> which were installed on old Xen iirc get upset if they later boot
>>> and see LTP1 not-present in ACPI tables). There'd need to be a
>>>  *good* reason for hiding LTP1 in DSDT therefore. - Partially
>>> applied to unstable; not applied to 3.4
>> 
>> 1) I think that parallel port configuration should be similar to
>>    serial port configuration. With that idea in mind this patch was
>>    created (and some code were copied from serial config).
>> 2) I tested this patch with Windows Vista Business SP2 and
>>    Finnix Ver. 92.1 as a guests and everything worked without
>>    any problem.
>> 3) I do not like that broken/disabled devices appear in config/device
>>    manager without any warning. In this situation disabled LPT1 port
>>    appear in Windows Device Manager without any warning. I think it
>>    is confusing for users when they see working devices in WDM and
>>    they could not use them.
>> 4) As I mentioned earlier, all patches (including this one) do not
>>    change default behavior of Xen. It means that default QEMU
>>    configuration is not changed when parallel option is not specified
>>    in config file. If somebody would like to disable LPT1, he/she
>>    should have possibility to do that in proper way. I think it is
>>    another question that he/she is aware what is he/she doing.
>> 
>> That is why I think
>> xen-unstable-parallel-serial-monitor_20091223.patch patch should be
>> fully applied. 
>> 
>> Daniel Kiper
>> 
>> _______________________________________________
>> Xen-devel mailing list
>> Xen-devel@lists.xensource.com
>> http://lists.xensource.com/xen-devel

Re: [PATCH] Parallel/Serial/Monitor/pyGRUB ReiserFS/others support/fixes

[Keir Fraser]

On 24/12/2009 00:57, "Daniel Kiper" <dkiper@net-space.pl> wrote:

> That is why I think xen-unstable-parallel-serial-monitor_20091223.patch
> patch should be fully applied.

Well, let's give it a go.

 -- Keir

RE: [PATCH] Parallel/Serial/Monitor/pyGRUB ReiserFS/others support/fixes

[Xu, Dongxiao]

Yeah, it's fixed. 

Thanks!
Dongxiao

Keir Fraser wrote:
> Hopefully I've fixed this with c/s 20719.
> 
>  -- Keir
> 
> On 24/12/2009 07:43, "Xu, Dongxiao" <dongxiao.xu@intel.com> wrote:
> 
>> Hi, Daniel,
>> Could you help to check your patch which is checked in as C/S 20717?
>> It seems that VM save/restore/migration and reboot is broken since
>> this changeset. 
>> 
>> Thanks!
>> Dongxiao
>> 
>> Daniel Kiper wrote:
>>> Hello,
>>> 
>>>>>>> xen-unstable-hvmloader-strip_20091222.patch:
>>>>>>>   Strip hvmloader after compile. dh_strip from debhelper
>>>>>>>   does not strip binaries without execute attribute.
>>>>>> 
>>>>>> I don't see the point of doing this. We don't in general strip
>>>>>> stuff we build under tools/, the extra disc space taken is
>>>>>> trivial, and symbols can be convenient to have around. I just
>>>>>>  don't see the win. - Not applying to xen-unstable or xen-3.4
>>>>> 
>>>>> Yes, it is the packager's responsibility to take care of this if
>>>>> they deem it necessary, either manually or by adding +x so that
>>>>> dh_strip takes care of it.
>>>> 
>>>> Actually hvmloader seems to be a+x already, in all my build trees.
>>> 
>>> FYI:
>>> xen-3.4.2/tools/firmware/hvmloader/hvmloader: 0755
>>> xen-3.4.2/dist/install/usr/lib/xen/boot/hvmloader: 0644
>>> 
>>> OK, I agree that this patch break general
>>> rule and should not be applied.
>>> 
>>>>> xen-unstable-parallel-serial-monitor_20091223.patch:
>>>> 
>>>> Xend parts look good. The hvmloader parts look unnecessary (since
>>>> we currently always declare LPT1 present) and dangerous (Windows
>>>> guests which were installed on old Xen iirc get upset if they
>>>> later boot and see LTP1 not-present in ACPI tables). There'd need
>>>>  to be a *good* reason for hiding LTP1 in DSDT therefore. -
>>>> Partially applied to unstable; not applied to 3.4
>>> 
>>> 1) I think that parallel port configuration should be similar to
>>>    serial port configuration. With that idea in mind this patch was
>>>    created (and some code were copied from serial config).
>>> 2) I tested this patch with Windows Vista Business SP2 and
>>>    Finnix Ver. 92.1 as a guests and everything worked without   
>>> any problem. 3) I do not like that broken/disabled devices appear
>>>    in config/device manager without any warning. In this situation
>>>    disabled LPT1 port appear in Windows Device Manager without any
>>>    warning. I think it is confusing for users when they see working
>>> devices in WDM and    they could not use them. 4) As I mentioned
>>>    earlier, all patches (including this one) do not change default
>>>    behavior of Xen. It means that default QEMU configuration is not
>>>    changed when parallel option is not specified in config file. If
>>>    somebody would like to disable LPT1, he/she should have
>>>    possibility to do that in proper way. I think it is another
>>> question that he/she is aware what is he/she doing. 
>>> 
>>> That is why I think
>>> xen-unstable-parallel-serial-monitor_20091223.patch patch should be
>>> fully applied. 
>>> 
>>> Daniel Kiper
>>> 
>>> _______________________________________________
>>> Xen-devel mailing list
>>> Xen-devel@lists.xensource.com
>>> http://lists.xensource.com/xen-devel

Re: [PATCH] Parallel/Serial/Monitor/pyGRUB ReiserFS/others support/fixes

[Daniel Kiper]

[-- Attachment #1: Type: text/plain, Size: 793 bytes --]

Hello,

Sorry for late reply however I was very busy.

On Thu, Dec 24, 2009 at 09:00:52AM +0000, Keir Fraser wrote:
> Hopefully I've fixed this with c/s 20719.

Small fix. Without this patch it is not possible to
disable QEMU monitor by setting monitor
option to 0 in config file.

diff -r bb34d8e578d2 -r 4dded69f097e tools/python/xen/xend/image.py
--- a/tools/python/xen/xend/image.py	Wed Dec 30 13:10:03 2009 +0000
+++ b/tools/python/xen/xend/image.py	Wed Dec 30 23:40:17 2009 +0100
@@ -376,6 +376,8 @@ class ImageHandler:
                 ret = ret + ['-monitor', vmConfig['platform'].get('monitor_path')]
             else:
                 ret = ret + ['-monitor', 'vc']
+        else:
+            ret = ret + ['-monitor', 'null']
 
         return ret
 

Happy New Year,

Daniel Kiper

[-- Attachment #2: xen-unstable-monitor_20091230.patch --]
[-- Type: text/plain, Size: 835 bytes --]

# HG changeset patch
# User root@dte40r0-um-i386.local.net-space.pl
# Date 1262212817 -3600
# Node ID 4dded69f097e9eb2e3bf174b3ced17c012fde4f0
# Parent  bb34d8e578d204d4fd59f4c6a3a20ad16e862a15
Without this patch it is not possible to
disable QEMU monitor by setting monitor
option to 0 in config file.

Signed-off-by: Daniel Kiper <dkiper@net-space.pl>

diff -r bb34d8e578d2 -r 4dded69f097e tools/python/xen/xend/image.py
--- a/tools/python/xen/xend/image.py	Wed Dec 30 13:10:03 2009 +0000
+++ b/tools/python/xen/xend/image.py	Wed Dec 30 23:40:17 2009 +0100
@@ -376,6 +376,8 @@ class ImageHandler:
                 ret = ret + ['-monitor', vmConfig['platform'].get('monitor_path')]
             else:
                 ret = ret + ['-monitor', 'vc']
+        else:
+            ret = ret + ['-monitor', 'null']
 
         return ret
 

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

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