From: Mauro Carvalho Chehab <mchehab@s-opensource.com>
To: Sakari Ailus <sakari.ailus@iki.fi>
Cc: Sakari Ailus <sakari.ailus@linux.intel.com>,
linux-media@vger.kernel.org, gjasny@googlemail.com
Subject: Re: [v4l-utils PATCH 1/1] Fix static linking of v4l2-compliance and v4l2-ctl
Date: Mon, 26 Sep 2016 13:59:45 -0300 [thread overview]
Message-ID: <20160926135945.351384ca@vento.lan> (raw)
In-Reply-To: <20160926154640.GA3225@valkosipuli.retiisi.org.uk>
Em Mon, 26 Sep 2016 18:46:40 +0300
Sakari Ailus <sakari.ailus@iki.fi> escreveu:
> Hi Mauro,
>
> On Mon, Sep 19, 2016 at 11:19:12AM -0300, Mauro Carvalho Chehab wrote:
> > Em Mon, 19 Sep 2016 16:21:30 +0300
> > Sakari Ailus <sakari.ailus@linux.intel.com> escreveu:
> >
> > > Hi Mauro,
> > >
> > > On 09/19/16 14:22, Mauro Carvalho Chehab wrote:
> > > > Em Mon, 19 Sep 2016 13:50:25 +0300
> > > > Sakari Ailus <sakari.ailus@linux.intel.com> escreveu:
> > > >
> > > >> v4l2-compliance and v4l2-ctl depend on librt and libpthread. The symbols
> > > >> are found by the linker only if these libraries are specified after the
> > > >> objects that depend on them.
> > > >>
> > > >> As LDFLAGS variable end up expanded on libtool command line before LDADD,
> > > >> move the libraries to LDADD after local objects. -lpthread is added as on
> > > >> some systems librt depends on libpthread. This is the case on Ubuntu 16.04
> > > >> for instance.
> > > >>
> > > >> After this patch, creating a static build using the command
> > > >>
> > > >> LDFLAGS="--static -static" ./configure --disable-shared --enable-static
> > > >
> > > > It sounds weird to use LDFLAGS="--static -static" here, as the
> > > > configure options are already asking for static.
> > > >
> > > > IMHO, the right way would be to change configure.ac to add those LDFLAGS
> > > > when --disable-shared is used.
> > >
> > > That's one option, but then shared libraries won't be built at all.
> >
> > Well, my understanding is that --disable-shared is meant to disable
> > building the shared library build :)
> >
> > > I'm
> > > not sure what would be the use cases for that, though: static linking
> > > isn't very commonly needed except when you need to run the binaries
> > > elsewhere (for whatever reason) where you don't have the libraries you
> > > linked against available.
> >
> > Yeah, that's the common usage. It is also interesting if someone
> > wants to build 2 versions of the same utility, each using a
> > different library, for testing purposes.
> >
> > The usecase I can't see is to use --disable-shared but keeping
> > using the dynamic library for the exec files.
>
> There are three primary options here,
>
> 1. build an entirely static binary,
> 2. build a binary that relies on dynamic libraries as well and
> 3. build a binary that relies on dynamic libraries outside v4l-utils package
> but that links v4l-utils originating libraries statically.
>
> If you say 3. is not needed then we could just use --disable-shared also to
> tell that static binaries are to be built.
>
> 3. is always used for libv4l2subdev and libmediactl as the libraries do not
> have stable APIs.
Sakari,
I can't see what you mean by scenario (2). I mean, if
--disable-shared is called, it *should not* use dynamic libraries
for any library provided by v4l-utils, as the generated binaries will
either:
a) don't work, because those libraries weren't built;
b) will do the wrong thing, as they'll be dynamically linked
to an older version of the library.
So, there are only 3 possible scenarios, IMHO:
1) dynamic libraries, dynamic execs
2) static v4l-utils libraries, static execs
3) static v4l-utils libraries, static links for v4l-utils libs, dyn for the rest.
In practice, I don't see any reason for keeping support for both (2)
and (3), as all usecases for (3) can be covered by a fully static
exec. It is also very confusing for one to understand that.
For example, right now, we have those static/shared options:
--enable-static[=PKGS] build static libraries [default=yes]
--enable-shared[=PKGS] build shared libraries [default=yes]
with, IMHO, sounds confusing, as those options don't seem to be
orthogonal. I mean, what happens someone calls ./configure with:
./configure --disable-static --disable-shared
Thanks,
Mauro
next prev parent reply other threads:[~2016-09-26 16:59 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-09-19 10:50 [v4l-utils PATCH 1/1] Fix static linking of v4l2-compliance and v4l2-ctl Sakari Ailus
2016-09-19 11:22 ` Mauro Carvalho Chehab
2016-09-19 13:21 ` Sakari Ailus
2016-09-19 14:19 ` Mauro Carvalho Chehab
2016-09-26 15:46 ` Sakari Ailus
2016-09-26 16:59 ` Mauro Carvalho Chehab [this message]
2016-09-26 21:40 ` Sakari Ailus
2016-09-27 16:04 ` Mauro Carvalho Chehab
2016-09-27 17:35 ` Mauro Carvalho Chehab
2016-09-19 13:22 ` [v4l-utils PATCH 2/2] Add --with-static-binaries option to link binaries statically Sakari Ailus
2016-09-19 14:21 ` Mauro Carvalho Chehab
2016-09-19 14:34 ` Sakari Ailus
2016-09-26 17:41 ` Gregor Jasny
2016-09-26 18:02 ` Mauro Carvalho Chehab
2018-02-22 9:02 ` Sakari Ailus
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20160926135945.351384ca@vento.lan \
--to=mchehab@s-opensource.com \
--cc=gjasny@googlemail.com \
--cc=linux-media@vger.kernel.org \
--cc=sakari.ailus@iki.fi \
--cc=sakari.ailus@linux.intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).