Patchwork [V2] iputils: Fix postinst to run on cross rootfs generation

rpm shows error messages like /bin/traceroute6 not found
when building cross rootfs

Yocto # 2892

Signed-off-by: Khem Raj <raj.khem@gmail.com>
---
 meta/recipes-extended/iputils/iputils_s20101006.bb |    8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

On 08/23/2012 09:13 PM, Khem Raj wrote:
> rpm shows error messages like /bin/traceroute6 not found
> when building cross rootfs
>
Richard pointed out that with Pseudo the pkg_postinsts are un-needed, so 
I offered a different patch to that effect.

Sau!

> Yocto # 2892
>
> Signed-off-by: Khem Raj <raj.khem@gmail.com>
> ---
>   meta/recipes-extended/iputils/iputils_s20101006.bb |    8 ++++----
>   1 file changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/meta/recipes-extended/iputils/iputils_s20101006.bb b/meta/recipes-extended/iputils/iputils_s20101006.bb
> index 62e2b5e..bbaff5e 100644
> --- a/meta/recipes-extended/iputils/iputils_s20101006.bb
> +++ b/meta/recipes-extended/iputils/iputils_s20101006.bb
> @@ -13,7 +13,7 @@ LIC_FILES_CHKSUM = "file://ping.c;beginline=1;endline=35;md5=f9ceb201733e9a6cf8f
>
>   DEPENDS = "openssl docbook-utils-native sgmlspl-native"
>
> -PR = "r5"
> +PR = "r6"
>
>   SRC_URI = "http://www.skbuff.net/iputils/${BPN}-${PV}.tar.bz2 \
>              file://debian/fix-dead-host-ping-stats.diff \
> @@ -59,15 +59,15 @@ ALTERNATIVE_${PN}-ping6 = "ping6"
>   ALTERNATIVE_LINK_NAME[ping6] = "${base_bindir}/ping6"
>
>   pkg_postinst_${PN}-ping () {
> -       chmod 4555 ${base_bindir}/ping
> +       chmod 4555 $D${base_bindir}/ping
>   }
>
>   pkg_postinst_${PN}-ping6 () {
> -       chmod 4555 ${base_bindir}/ping6
> +       chmod 4555 $D${base_bindir}/ping6
>   }
>
>   pkg_postinst_${PN}-traceroute6 () {
> -	chmod 4555 ${base_bindir}/traceroute6
> +	chmod 4555 $D${base_bindir}/traceroute6
>   }
>
>   PACKAGES += "${PN}-ping ${PN}-ping6 ${PN}-arping ${PN}-tracepath ${PN}-tracepath6 ${PN}-traceroute6"
>

On Thu, Aug 23, 2012 at 9:59 PM, Saul Wold <sgw@linux.intel.com> wrote:
>>
> Richard pointed out that with Pseudo the pkg_postinsts are un-needed, so I
> offered a different patch to that effect.

yes I see

http://git.openembedded.org/openembedded-core/commit/?id=a2bc81032c85548f7c72dc76e6020ab4df9af6de

how does it work on device installs isnt clear to me though

with existing patch atleast I understood that with Online Package
management the postinst would run

On Fri, 2012-08-24 at 09:55 -0700, Khem Raj wrote:
> On Thu, Aug 23, 2012 at 9:59 PM, Saul Wold <sgw@linux.intel.com> wrote:
> >>
> > Richard pointed out that with Pseudo the pkg_postinsts are un-needed, so I
> > offered a different patch to that effect.
> 
> yes I see
> 
> http://git.openembedded.org/openembedded-core/commit/?id=a2bc81032c85548f7c72dc76e6020ab4df9af6de
> 
> how does it work on device installs isnt clear to me though
> 
> with existing patch atleast I understood that with Online Package
> management the postinst would run

The file has the right permissions in the package file now. This means
we don't need any postinst.

We used to have problems preserving permissions/owners in packages but
with pseudo, we don't, that is its big win.

So this was leftover stuff from a previous age :)

Cheers,

Richard

On Fri, Aug 24, 2012 at 10:06 AM, Richard Purdie
<richard.purdie@linuxfoundation.org> wrote:
> On Fri, 2012-08-24 at 09:55 -0700, Khem Raj wrote:
>> On Thu, Aug 23, 2012 at 9:59 PM, Saul Wold <sgw@linux.intel.com> wrote:
>> >>
>> > Richard pointed out that with Pseudo the pkg_postinsts are un-needed, so I
>> > offered a different patch to that effect.
>>
>> yes I see
>>
>> http://git.openembedded.org/openembedded-core/commit/?id=a2bc81032c85548f7c72dc76e6020ab4df9af6de
>>
>> how does it work on device installs isnt clear to me though
>>
>> with existing patch atleast I understood that with Online Package
>> management the postinst would run
>
> The file has the right permissions in the package file now. This means
> we don't need any postinst.
>
> We used to have problems preserving permissions/owners in packages but
> with pseudo, we don't, that is its big win.
>
> So this was leftover stuff from a previous age :)
>

hmmm interesting. There are more cases of postinst I see.
may be we should look through all of them

> Cheers,
>
> Richard
>