[Buildroot] [PATCH 1/1] package/libglib2: fix util-linux-libs dependency

Arnout Vandecappelle arnout at mind.be
Tue Dec 1 21:12:00 UTC 2020


 [Replying to this one, because it's the one with most of the discussion, but my
comments really pertain to [1]]
 [1]
http://patchwork.ozlabs.org/project/buildroot/patch/20201103205745.123950-1-fontaine.fabrice@gmail.com/



On 03/11/2020 21:38, Yann E. MORIN wrote:
> Fabrice, All,
> 
> On 2020-11-01 16:03 +0100, Fabrice Fontaine spake thusly:
>> Le dim. 1 nov. 2020 à 14:04, Yann E. MORIN <yann.morin.1998 at free.fr> a écrit :
>>> On 2020-11-01 13:15 +0100, Fabrice Fontaine spake thusly:
>>>> Commit 2acaa86ce630f616b12aa92493e0269421c5bc97 replaced util-linux by
>>>> util-linux-libs but this raises the following build failure:
>>>>
>>>> Makefile:585: *** util-linux-libs is in the dependency chain of libglib2 that has added it to its _DEPENDENCIES variable without selecting it or depending on it from Config.in.  Stop.
>>>>
>>>> To fix this build failure, use util-linux or util-linux-libs depending
>>>> on BR2_PACKAGE_UTIL_LINUX_LIBS
>>>
>>> OK, so this is exactly what I did not like in this util-linux vs
>>> util-linus-libs split, and the fact that util-linux-libs is optional...
>>>
>>> IMHO:
>>>
>>>  1. util-linux-libs must be enabled as soon as util-linux is enabled,
>>>
>>>  2. packages should build-depend on util-linux-libs, and never on
>>>     util-linux, and should 'select BR2_PACKAGE_UTIL_LINUX_LIBS' if
>>>     all they need is a library.
>> We can't select BR2_PACKAGE_UTIL_LINUX_LIBS without selecting
>> BR2_PACKAGE_UTIL_LINUX as BR2_PACKAGE_UTIL_LINUX_LIBS is a hidden
>> option of BR2_PACKAGE_UTIL_LINUX.
> 
> No, we really need to make util-linux-libs a real package, not a hidden
> one, like
> 
>     config BR2_PACAKGE_UTIL_LINUX_LIBS
>         bool "util-linux-libs"
>         help
>           Only the libraries provided bby util-linux; if you need the
>           tools too, select util-linux.

 There's no need to make this user-selectable AFAICS.

> 
> and:
> 
>     config BR2_PACAKGE_UTIL_LINUX
>         bool "util-linux"
>         select BR2_PACAKGE_UTIL_LINUX_LIBS
> 
>     (or:)
> 
>     config BR2_PACAKGE_UTIL_LINUX_LIBMOUNT
>         bool "libmount"
>         select BR2_PACAKGE_UTIL_LINUX_LIBS

 This was one of the possibilities that was considered/tried when Carlos and I
worked on the util-linux split. I don't remember the details of why it didn't
work, but basically it didn't :-) It *might* be just avoiding building
util-linux twice if it's not needed though.


> 
>> Moreover, we'll select BR2_PACKAGE_UTIL_LINUX_LIBS as soon as
>> BR2_PACKAGE_UTIL_LINUX is enabled (as per your first suggestion).
>> As a result, util-linux will always be selected even if packages (such
>> as abootimg) only build-depend on util-linux-libs.
> 
> Now that util-linux-libs is a full-gfledged package, it is selctable by
> packages. See for example eudev os systemd, which already both select
> BR2_PACKAGE_UTIL_LINUX_LIBS and biuld-depend on util-linux-libs.

 And that's exactly the approach that should have been taken by libglib2 as well:

	select BR2_PACKAGE_UTIL_LINUX_LIBS if BR2_PACKAGE_UTIL_LINUX_LIBMOUNT

and unconditionally depend on util-linux-libs in libglib2.mk.


> 
>> So, in the end, we'll fix/clean the build order but util-linux will
>> still be built but this is unavoidable unless util-linux-libs is moved
>> to its own package.

 Yes, util-linux will definitely always be built.

 Regards,
 Arnout

> 
> It is already its own package; just it is not "user-visible".
> 
>> Did I misunderstood something?
> 
> The above, I guess. ;-)
> 
>> I'll send a 2-patch serie that implements your first suggestion and
>> your second suggestion on abootimg. If it pleases you, I'll update the
>> other packages.
> 
> Well, we can do it multi-pass, of course; fixing packages to build-depend
> on util-linux-libs is not a lost effort anyway.
> 
> Thanks for your work! :-)
> 
> Regards,
> Yann E. MORIN.
> 
>>> So, for now I'll apply this patch, because we do need a fix.
>>>
>>> But ultimately, we wnat to sanitise this situation.
>>>
>>> Regards,
>>> Yann E. MORIN.
>>>
>>>> Fixes:
>>>>  - http://autobuild.buildroot.org/results/8bf6360a343edd10c73a756ab86cdba727f9ea43
>>>>
>>>> Signed-off-by: Fabrice Fontaine <fontaine.fabrice at gmail.com>
>>>> ---
>>>>  package/libglib2/libglib2.mk | 4 ++++
>>>>  1 file changed, 4 insertions(+)
>>>>
>>>> diff --git a/package/libglib2/libglib2.mk b/package/libglib2/libglib2.mk
>>>> index 43fcb27fae..6e9dbd7b26 100644
>>>> --- a/package/libglib2/libglib2.mk
>>>> +++ b/package/libglib2/libglib2.mk
>>>> @@ -86,8 +86,12 @@ endif
>>>>
>>>>  ifeq ($(BR2_PACKAGE_UTIL_LINUX_LIBMOUNT),y)
>>>>  LIBGLIB2_CONF_OPTS += -Dlibmount=enabled
>>>> +ifeq ($(BR2_PACKAGE_UTIL_LINUX_LIBS),y)
>>>>  LIBGLIB2_DEPENDENCIES += util-linux-libs
>>>>  else
>>>> +LIBGLIB2_DEPENDENCIES += util-linux
>>>> +endif
>>>> +else
>>>>  LIBGLIB2_CONF_OPTS += -Dlibmount=disabled
>>>>  endif
>>>>
>>>> --
>>>> 2.28.0
>>>>
>>>> _______________________________________________
>>>> buildroot mailing list
>>>> buildroot at busybox.net
>>>> http://lists.busybox.net/mailman/listinfo/buildroot
>>>
>>> --
>>> .-----------------.--------------------.------------------.--------------------.
>>> |  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
>>> | +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
>>> | +33 561 099 427 `------------.-------:  X  AGAINST      |  \e/  There is no  |
>>> | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
>>> '------------------------------^-------^------------------^--------------------'
>> Best Regards,
>>
>> Fabrice
> 


More information about the buildroot mailing list