Added command line parameter to specify OEM content location.

Added command line parameter to specify OEM content location.
Added populating OEM partner partition with content.

BUG=http://crosbug.com/3828
TEST=Install Chrome OS on the device from USB stick. /mnt/partner_partition contains etc/... directory with customization manifests on first boot.

[jrbarnette, 2010-07-09 16:42:00]

http://codereview.chromium.org/2938003/diff/1/2
File build_image (right):

http://codereview.chromium.org/2938003/diff/1/2#newcode573
build_image:573: sudo touch "${STATEFUL_FS_DIR}/.consider_oobe"
I think touching consider_oobe isn't the right thing to do here.
The "consider_oobe" flag file is a temporary workaround to prevent
a performance regression until a better fix is in place.  If there's a
better fix to the original performance regression, the code in
chromeos_startup should be ripped out rather than creating one more
file here.

However, I don't see anything in this change that would address the
original performance regression; my reading is that this change will just
re-introduce the problem, with one more layer of complexity thrown in.
Is there some change somewhere else that guarantees that the
"oobe_completed" flag will get created?

[adlr, 2010-07-09 18:18:16]

http://codereview.chromium.org/2938003/diff/1/2
File build_image (right):

http://codereview.chromium.org/2938003/diff/1/2#newcode547
build_image:547: sudo find "${OEM_FS_DIR}" -type f -exec chmod -R 644 "{}" \;
should you also set owner/group to 0/0?

http://codereview.chromium.org/2938003/diff/1/2#newcode573
build_image:573: sudo touch "${STATEFUL_FS_DIR}/.consider_oobe"
On 2010/07/09 16:42:00, jrbarnette wrote:
> I think touching consider_oobe isn't the right thing to do here.
> The "consider_oobe" flag file is a temporary workaround to prevent
> a performance regression until a better fix is in place.  If there's a
> better fix to the original performance regression, the code in
> chromeos_startup should be ripped out rather than creating one more
> file here.
> 
> However, I don't see anything in this change that would address the
> original performance regression; my reading is that this change will just
> re-introduce the problem, with one more layer of complexity thrown in.
> Is there some change somewhere else that guarantees that the
> "oobe_completed" flag will get created?
> 

Yes, for now please remove this line for now, or put it behind a flag that
defaults to not touching the file. The point of .consider_oobe was that nobody
would get OEM partition behavior except for those working on the OEM stuff, who
would manually touch that file themselves.

Once OEM stuff is all working correctly, we'll remove the check for this file in
the startup script.

[dr, 2010-07-09 19:12:22]

http://codereview.chromium.org/2938003/diff/1/2
File build_image (right):

http://codereview.chromium.org/2938003/diff/1/2#newcode547
build_image:547: sudo find "${OEM_FS_DIR}" -type f -exec chmod -R 644 "{}" \;
On 2010/07/09 18:18:17, adlr wrote:
> should you also set owner/group to 0/0?

Done.

http://codereview.chromium.org/2938003/diff/1/2#newcode573
build_image:573: sudo touch "${STATEFUL_FS_DIR}/.consider_oobe"
Ahhh, I finally got the idea about .consider_oobe file. Removed this line from
here. Yes, I have a CL under review in Chrome that creates the .oobe_complete
file. I will submit them together with this one.

On 2010/07/09 16:42:00, jrbarnette wrote:
> I think touching consider_oobe isn't the right thing to do here.
> The "consider_oobe" flag file is a temporary workaround to prevent
> a performance regression until a better fix is in place.  If there's a
> better fix to the original performance regression, the code in
> chromeos_startup should be ripped out rather than creating one more
> file here.
> 
> However, I don't see anything in this change that would address the
> original performance regression; my reading is that this change will just
> re-introduce the problem, with one more layer of complexity thrown in.
> Is there some change somewhere else that guarantees that the
> "oobe_completed" flag will get created?
>