[OE-core] [PATCH] qemuwrapper-cross: fix postinst failed warning

cli10 changqing.li at windriver.com
Wed May 30 09:37:25 UTC 2018


Hi, Alex

I have test with your patch,  test passed,  all warnings are fixed:-).

But I have a  problem about this, could you help to give some comments? 
Thanks.


if in the future,  some recipe inherit allarch,  but use 
@qemu_run_binary directly,  not call qemuwrapper,

I think maybe still will met problem of get wrong base_libdir. so I 
think maybe we should change qemu_wrapper_cmdline in

qemu.bbclass, like this:

-    return "PSEUDO_UNLOAD=1 " + qemu_binary + " " + qemu_options + " -L " + rootfs_path\
+    return "PSEUDO_UNLOAD=1 " + "qemuwrapper" + " " + qemu_options + " -L " + rootfs_path\
              + " -E LD_LIBRARY_PATH=" + ":".join(library_paths) + " "

but some recipe or class like: gtk-doc.bbclass and 
gobject-introspection_1.56.1.bb,  they use

"@qemu_wrapper_cmdline" directly,  if we change qemu_wrapper_cmdline 
like above way,  we will met

another problem that  qemuwrapper have set LD_LIBRARY_PATH,  if user 
need to set extra lib path,  user

setting will override setting in qemuwrapper,  and also this is why in 
my patch I add a python script qemuwrapperhelper to this.



On 05/30/2018 03:05 PM, cli10 wrote:
> OK, I will test with your patch:-)
>
> On 05/30/2018 02:41 PM, Alexander Kanavin wrote:
>> 2018-05-30 7:27 GMT+03:00<changqing.li at windriver.com>:
>>> From: Changqing Li<changqing.li at windriver.com>
>>>
>>> 1. some binary like udev-hwdb/pixbufcache/gio-module-cache/fontcache
>>> uses qemu usermode by default,  but some architecture such as Intel
>>> skylake does not support qemu usermode, this can lead to a build warning as below:
>>> 2. font is not arch related, but receipe like liberation-fonts inherit fontcache,
>>> fontcache called update_font_cache, update_font_cache use qemuwrapper,
>>> qemuwrapper is arch related, so update_font_cache will core dumped as below:
>> I need to apologize. I went ahead and fixed all of these issues myself
>> over the weekend, and improved a few other things in how postinsts
>> (and postinst intercepts) are handled. Should've told you immediately.
>> The seven patches have been sent, so if you can test and confirm that
>> they indeed fix these issues, I'd appreciate.
>> http://lists.openembedded.org/pipermail/openembedded-core/2018-May/151196.html
>>
>> Alex
>>
>
> -- 
> BRs
>
> Sandy(Li Changqing)
> +861084778653
>
>

-- 
BRs

Sandy(Li Changqing)
+861084778653

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openembedded.org/pipermail/openembedded-core/attachments/20180530/f7efe850/attachment-0002.html>


More information about the Openembedded-core mailing list