Re: [meta-oe][PATCHv2] libsoc: new recipe

[Jack Mitchell <ml@communistcode.co.uk>]

On 13/09/13 16:17, Jack Mitchell wrote:
> On 13/09/13 15:56, Koen Kooi wrote:
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> Op 13-09-13 16:53, Paul Eggleton schreef:
>>> On Friday 13 September 2013 15:23:36 Jack Mitchell wrote:
>>>> On 13/09/13 15:18, Koen Kooi wrote:
>>>>> Op 13-09-13 15:22, Jack Mitchell schreef:
>>>>>> From: Jack Mitchell <jack@embed.me.uk>
>>>>>>
>>>>>> Signed-off-by: Jack Mitchell <jack@embed.me.uk> ---
>>>>>>
>>>>>> Change from v1: added signed-off-by
>>>>>>
>>>>>> meta-oe/recipes-support/libsoc/libsoc_0.3.bb | 18
>>>>>> ++++++++++++++++++ 1 file changed, 18 insertions(+) create mode
>>>>>> 100644 meta-oe/recipes-support/libsoc/libsoc_0.3.bb
>>>>>>
>>>>>> diff --git a/meta-oe/recipes-support/libsoc/libsoc_0.3.bb
>>>>>> b/meta-oe/recipes-support/libsoc/libsoc_0.3.bb new file mode 100644
>>>>>> index 0000000..e91e152 --- /dev/null +++
>>>>>> b/meta-oe/recipes-support/libsoc/libsoc_0.3.bb @@ -0,0 +1,18 @@
>>>>>> +SUMMARY = "libsoc is a C library for interfacing with common SoC
>>>>>> peripherals through generic kernel interfaces" +HOMEPAGE =
>>>>>> "https://github.com/jackmitch/libsoc" + +LICENSE = "LGPLv2.1"
>>>>>> +LIC_FILES_CHKSUM =
>>>>>> "file://COPYING;md5=e0bfebea12a718922225ba987b2126a5" + +inherit
>>>>>> autotools + +DEPENDS = "libgcc"
>>>>>
>>>>> Isn't that automatically implied with gcc-cross?
>>>>
>>>> I wasn't sure, so I thought I would stick it in and see if anyone
>>>> piped up, I'll remove it in v3.
>>>>
>>>> Also on that note I noticed that Apache2 also DEPENDS on libgcc, is
>>>> that also wrong?
>>>
>>> I don't know anything about this particular recipe, but IIRC, apache
>>> complains if libgcc is not installed at runtime, and at least when I
>>> added that to the recipe it was not being installed otherwise.
>>
>> But that's RDEPENDS, no?
>>
>> regards,
>>
>> Koen
>
> My apologies, I just checked and it is as an RDEPENDS. The reason that I
> put libgcc in the DEPENDS is that when I was developing libsoc I ran
> into a function (pthread_cancel) which required libgcc.so.1 to be on the
> rootfs, which it wasn't. Having just checked a new build now, with only
> core-image-minimal there doesn't appear to be any libgcc.so.1 in the
> rootfs again, so should I have libgcc in RDEPENDS? I thought OE put all
> the libraries it linked against in the rootfs automatically... or have I
> misunderstood?
>
> Cheers,
>
>

Ah-hah, some context:

http://comments.gmane.org/gmane.comp.handhelds.openembedded.core/37086

So, based on that it seems as though it was never really resolved and 
having just checked eglibc,inc, the line mentioned in the above thread 
is still commented.

So, as tackling this eglibc problem is way above my league (I'd probably 
hinder more than help!) I'm going to do a v4 with RDEPENDS on libgcc.

I'll search for a bug report, and if there isn't one I'll do my best to 
submit a suitably informative one.

Cheers,

-- 
   Jack Mitchell (jack@embed.me.uk)
   Embedded Systems Engineer
   Cambrideshire, UK
   http://www.embed.me.uk
--