[PATCH] lib/oe/image.py: fix working directory

[PATCH] lib/oe/image.py: fix working directory

[Jonathan Liu]

The working directory needs to be changed before the image creation
commands instead of afterwards.

Signed-off-by: Jonathan Liu <net147@gmail.com>
---
 meta/lib/oe/image.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/meta/lib/oe/image.py b/meta/lib/oe/image.py
index c15296f..31193f2 100644
--- a/meta/lib/oe/image.py
+++ b/meta/lib/oe/image.py
@@ -176,8 +176,8 @@ class Image(object):
             bb.data.update_data(localdata)
             localdata.setVar('type', type)
 
-            cmds.append("\t" + localdata.getVar("IMAGE_CMD", True))
             cmds.append(localdata.expand("\tcd ${DEPLOY_DIR_IMAGE}"))
+            cmds.append("\t" + localdata.getVar("IMAGE_CMD", True))
 
             if type in cimages:
                 for ctype in cimages[type]:
-- 
1.8.5.4

Re: [PATCH] lib/oe/image.py: fix working directory

[Richard Purdie]

On Mon, 2014-02-17 at 21:46 +1100, Jonathan Liu wrote:
> The working directory needs to be changed before the image creation
> commands instead of afterwards.
> 
> Signed-off-by: Jonathan Liu <net147@gmail.com>
> ---
>  meta/lib/oe/image.py | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/meta/lib/oe/image.py b/meta/lib/oe/image.py
> index c15296f..31193f2 100644
> --- a/meta/lib/oe/image.py
> +++ b/meta/lib/oe/image.py
> @@ -176,8 +176,8 @@ class Image(object):
>              bb.data.update_data(localdata)
>              localdata.setVar('type', type)
>  
> -            cmds.append("\t" + localdata.getVar("IMAGE_CMD", True))
>              cmds.append(localdata.expand("\tcd ${DEPLOY_DIR_IMAGE}"))
> +            cmds.append("\t" + localdata.getVar("IMAGE_CMD", True))
>  
>              if type in cimages:
>                  for ctype in cimages[type]:

I applied this but ended up reverting it since its clear the image
commands assume a different working directory and this causes failures.

I'd like to see more explanation about the problem this is trying to
solve.

Cheers,

Richard

Re: [PATCH] lib/oe/image.py: fix working directory

[Jonathan Liu]

On 18/02/2014 7:41 PM, Richard Purdie wrote:
> On Mon, 2014-02-17 at 21:46 +1100, Jonathan Liu wrote:
>> The working directory needs to be changed before the image creation
>> commands instead of afterwards.
>>
>> Signed-off-by: Jonathan Liu <net147@gmail.com>
>> ---
>>   meta/lib/oe/image.py | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/meta/lib/oe/image.py b/meta/lib/oe/image.py
>> index c15296f..31193f2 100644
>> --- a/meta/lib/oe/image.py
>> +++ b/meta/lib/oe/image.py
>> @@ -176,8 +176,8 @@ class Image(object):
>>               bb.data.update_data(localdata)
>>               localdata.setVar('type', type)
>>   
>> -            cmds.append("\t" + localdata.getVar("IMAGE_CMD", True))
>>               cmds.append(localdata.expand("\tcd ${DEPLOY_DIR_IMAGE}"))
>> +            cmds.append("\t" + localdata.getVar("IMAGE_CMD", True))
>>   
>>               if type in cimages:
>>                   for ctype in cimages[type]:
> I applied this but ended up reverting it since its clear the image
> commands assume a different working directory and this causes failures.
>
> I'd like to see more explanation about the problem this is trying to
> solve.
>
> Cheers,
>
> Richard
>
The sdimg image classes in meta-raspberrypi and meta-sunxi broke when I 
switched from dora to master branch.
See: 
https://github.com/djwillis/meta-raspberrypi/blob/master/classes/sdcard_image-rpi.bbclass#L43
It is using relative path to rootfs which assumes the working directory 
is DEPLOY_DIR_IMAGE.
I was able to build the SD card images fine after the fix.

Regards,
Jonathan

Re: [PATCH] lib/oe/image.py: fix working directory

[Jonathan Liu]

On 18/02/2014 7:55 PM, Jonathan Liu wrote:
> On 18/02/2014 7:41 PM, Richard Purdie wrote:
>> On Mon, 2014-02-17 at 21:46 +1100, Jonathan Liu wrote:
>>> The working directory needs to be changed before the image creation
>>> commands instead of afterwards.
>>>
>>> Signed-off-by: Jonathan Liu <net147@gmail.com>
>>> ---
>>>   meta/lib/oe/image.py | 2 +-
>>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/meta/lib/oe/image.py b/meta/lib/oe/image.py
>>> index c15296f..31193f2 100644
>>> --- a/meta/lib/oe/image.py
>>> +++ b/meta/lib/oe/image.py
>>> @@ -176,8 +176,8 @@ class Image(object):
>>>               bb.data.update_data(localdata)
>>>               localdata.setVar('type', type)
>>>   -            cmds.append("\t" + localdata.getVar("IMAGE_CMD", True))
>>>               cmds.append(localdata.expand("\tcd ${DEPLOY_DIR_IMAGE}"))
>>> +            cmds.append("\t" + localdata.getVar("IMAGE_CMD", True))
>>>                 if type in cimages:
>>>                   for ctype in cimages[type]:
>> I applied this but ended up reverting it since its clear the image
>> commands assume a different working directory and this causes failures.
>>
>> I'd like to see more explanation about the problem this is trying to
>> solve.
>>
>> Cheers,
>>
>> Richard
>>
> The sdimg image classes in meta-raspberrypi and meta-sunxi broke when 
> I switched from dora to master branch.
> See: 
> https://github.com/djwillis/meta-raspberrypi/blob/master/classes/sdcard_image-rpi.bbclass#L43
> It is using relative path to rootfs which assumes the working 
> directory is DEPLOY_DIR_IMAGE.
> I was able to build the SD card images fine after the fix.
Also, it makes absolutely no sense whatsoever to change working 
directory at the end of the function:
img_creation_func() {
     oe_mkext234fs ext4 -i 8192
     cd /home/build/poky/build/tmp/deploy/images/qemux86
}

...

img_creation_func

This would mean the last command executed in create_image.ext4 (see temp 
folder for image recipe) would be changing the directory.

>
> Regards,
> Jonathan

Re: [PATCH] lib/oe/image.py: fix working directory

[Richard Purdie]

On Tue, 2014-02-18 at 20:11 +1100, Jonathan Liu wrote:
> On 18/02/2014 7:55 PM, Jonathan Liu wrote:
> > On 18/02/2014 7:41 PM, Richard Purdie wrote:
> >> On Mon, 2014-02-17 at 21:46 +1100, Jonathan Liu wrote:
> >>> The working directory needs to be changed before the image creation
> >>> commands instead of afterwards.
> >>>
> >>> Signed-off-by: Jonathan Liu <net147@gmail.com>
> >>> ---
> >>>   meta/lib/oe/image.py | 2 +-
> >>>   1 file changed, 1 insertion(+), 1 deletion(-)
> >>>
> >>> diff --git a/meta/lib/oe/image.py b/meta/lib/oe/image.py
> >>> index c15296f..31193f2 100644
> >>> --- a/meta/lib/oe/image.py
> >>> +++ b/meta/lib/oe/image.py
> >>> @@ -176,8 +176,8 @@ class Image(object):
> >>>               bb.data.update_data(localdata)
> >>>               localdata.setVar('type', type)
> >>>   -            cmds.append("\t" + localdata.getVar("IMAGE_CMD", True))
> >>>               cmds.append(localdata.expand("\tcd ${DEPLOY_DIR_IMAGE}"))
> >>> +            cmds.append("\t" + localdata.getVar("IMAGE_CMD", True))
> >>>                 if type in cimages:
> >>>                   for ctype in cimages[type]:
> >> I applied this but ended up reverting it since its clear the image
> >> commands assume a different working directory and this causes failures.
> >>
> >> I'd like to see more explanation about the problem this is trying to
> >> solve.
> >>
> >> Cheers,
> >>
> >> Richard
> >>
> > The sdimg image classes in meta-raspberrypi and meta-sunxi broke when 
> > I switched from dora to master branch.
> > See: 
> > https://github.com/djwillis/meta-raspberrypi/blob/master/classes/sdcard_image-rpi.bbclass#L43
> > It is using relative path to rootfs which assumes the working 
> > directory is DEPLOY_DIR_IMAGE.
> > I was able to build the SD card images fine after the fix.
> Also, it makes absolutely no sense whatsoever to change working 
> directory at the end of the function:
> img_creation_func() {
>      oe_mkext234fs ext4 -i 8192
>      cd /home/build/poky/build/tmp/deploy/images/qemux86
> }
> 
> ...
> 
> img_creation_func
> 
> This would mean the last command executed in create_image.ext4 (see temp 
> folder for image recipe) would be changing the directory.

Its being done for the compression code that immediately follows it. So
it does make sense and its why changing it broke things.

Cheers,

Richard