[oe] chromium detected as stripped

Andreas Müller schnitzeltony at googlemail.com
Mon Aug 14 14:10:25 UTC 2017


On Mon, Aug 14, 2017 at 12:22 PM, Jussi Kukkonen
<jussi.kukkonen at intel.com> wrote:
>
> On 14 August 2017 at 10:04, Jussi Kukkonen <jussi.kukkonen at intel.com> wrote:
>>
>> On 13 August 2017 at 02:10, Andreas Müller <schnitzeltony at googlemail.com>
>> wrote:
>>>
>>> Hi,
>>>
>>> in current OE environment 'file' says:
>>>
>>> <...>image/usr/bin/chromium/chrome: ELF 32-bit LSB shared object, ARM,
>>> EABI5 version 1 (SYSV), dynamically linked, interpreter
>>> /lib/ld-linux-armhf.so.3, stripped
>>>
>>> This file is 4.3G of size and this does not fir into most of my sdcards!
>>>
>>> One note: I have an elder environment: meta-browser
>>> c79cf6c7811afd2cd167e8967e3c29b1e797371e. There chrome binary is
>>> detected as not stripped and properly stripped on do_package.
>>>
>>> Any idea why it is detected as stripped?
>>
>>
>> I have a vague recollection that 'file' and other tools can get very
>> confused when given 32 bit executables that are larger than 4 GB. It may
>> make sense to build a non-debug version to begin with to get around this.
>
>
> I chatted with rakuco who's done nice work with up to date chromium in
> meta-crosswalk, he's got some relevant commits related to this:
>
> https://github.com/rakuco/meta-crosswalk/commit/5aae1539c3d62b295dba40f494397c0797140b0f
> https://github.com/rakuco/meta-crosswalk/commit/c56aa2bf047d3a01e14f4f9bfe5f88664b333f36
>
> On a related note, this good work is currently a bit hidden in
> meta-crosswalk -- is there a chance we could get this combined with
> meta-browser (or whatever place makes sense)? I know next to nothing about
> chromium packaging so CCing Raphael and the meta-browser maintainers for
> comments.
>
That is interesting. If your suspicion is correct, we are very close
to the edge of file working properly or not. With meta-browser
c79cf6c7811afd2cd167e8967e3c29b1e797371e (16 commits behind master) I
get a proper detected and then split chromium. With meta-browser
master head file incorrectly reports that chrome is split already. I
did not check the exact size numbers but my file browser reports 4.3G
for both cases.

Just to gather further information I think about bisecting meta-browser.

Thanks

Andreas



More information about the Openembedded-devel mailing list