[dpdk-dev] [PATCH] mk: fix the combined library problems by replacing it with a linker script

[Neil Horman]

On Mon, Nov 30, 2015 at 08:41:02AM -0800, Stephen Hemminger wrote:
> On Mon, 30 Nov 2015 10:03:43 -0500
> Neil Horman <nhorman at tuxdriver.com> wrote:
> 
> > On Wed, Nov 25, 2015 at 08:08:37AM -0800, Stephen Hemminger wrote:
> > > On Wed, 25 Nov 2015 10:38:48 +0200
> > > Panu Matilainen <pmatilai at redhat.com> wrote:
> > > 
> > > > On 11/25/2015 12:46 AM, Stephen Hemminger wrote:
> > > > > On Tue, 24 Nov 2015 16:31:17 +0200
> > > > > Panu Matilainen <pmatilai at redhat.com> wrote:
> > > > >
> > > > >> The physically linked-together combined library has been an increasing
> > > > >> source of problems, as was predicted when library and symbol versioning
> > > > >> was introduced. Replace the complex and fragile construction with a
> > > > >> simple linker script which achieves the same without all the problems,
> > > > >> remove the related kludges from eg mlx drivers.
> > > > >>
> > > > >> Since creating the linker script is practically zero cost, remove the
> > > > >> config option and just create it always.
> > > > >>
> > > > >> Based on a patch by Sergio Gonzales Monroy, linker script approach
> > > > >> initially suggested by Neil Horman.
> > > > >>
> > > > >> Suggested-by: Sergio Gonzalez Monroy <sergio.gonzalez.monroy at intel.com>
> > > > >> Suggested-by: Neil Horman <nhorman at tuxdriver.com>
> > > > >> Signed-off-by: Panu Matilainen <pmatilai at redhat.com>
> > > > >
> > > > > But it now means distros have to ship 20 libraries which seems like
> > > > > a step back.
> > > > 
> > > > That's how Fedora and RHEL are shipping it already and nobody has so 
> > > > much as noticed anything strange, much less complained about it. 20 
> > > > libraries is but a drop in the ocean on a average distro. But more to 
> > > > the point, distros will prefer 50 working libraries over one that doesn't.
> > > > 
> > > > The combined library as it is simply is no longer a viable option. 
> > > > Besides just being broken (witness the strange hacks people are coming 
> > > > up with to work around issues in it) its ugly because it basically gives 
> > > > the middle finger to all the effort going into version compatibility, 
> > > > and its also big. Few projects will use every library in DPDK, but with 
> > > > the combined library they're forced to lug the 800 pound gorilla along 
> > > > needlessly.
> > > > 
> > > > 	- Panu -
> > > > 
> > > 
> > > Fixing the combined library took less than an hour for us.
> > How did you fix the versioning issue?
> > 
> > Neil
> 
> This is what I did. 
> Also decided to keep shared library version == major DPDK version
> to avoid confusion.
> 
Ok, thats the library versioning, which is great, but I was more concerned about
the symbol versioning - i.e. the collection of version linker scripts that get
applied when you build individual libraries, but completely ignored when you
build the combined library

Neil

> 
> mk: fix when building combined shared library
> 
> The DPDK mk file does not set shared object name or version
> information as required by Debian.
> 
> Signed-off-by: Stephen Hemminger <stephen at networkplumber.org>
> 
> --- a/mk/rte.sharelib.mk
> +++ b/mk/rte.sharelib.mk
> @@ -51,10 +51,10 @@ ifeq ($(LINK_USING_CC),1)
>  # Override the definition of LD here, since we're linking with CC
>  LD := $(CC) $(CPU_CFLAGS)
>  O_TO_S = $(LD) $(call linkerprefix,$(CPU_LDFLAGS)) \
> -	-shared $(OBJS) -o $(RTE_OUTPUT)/lib/$(LIB_ONE)
> +	-shared $(OBJS) -Wl,-soname,$(LIB_ONE).$(RTE_LIBVERS) -o $(RTE_OUTPUT)/lib/$(LIB_ONE)
>  else
>  O_TO_S = $(LD) $(CPU_LDFLAGS) \
> -	-shared $(OBJS) -o $(RTE_OUTPUT)/lib/$(LIB_ONE)
> +	-shared $(OBJS) -soname $(LIB_ONE).$(RTE_LIBVERS) -o $(RTE_OUTPUT)/lib/$(LIB_ONE)
>  endif
>  
>  O_TO_S_STR = $(subst ','\'',$(O_TO_S)) #'# fix syntax highlight
> --- a/mk/rte.vars.mk
> +++ b/mk/rte.vars.mk
> @@ -74,8 +74,10 @@ ifneq ($(BUILDING_RTE_SDK),)
>  endif
>  
>  RTE_LIBNAME := $(CONFIG_RTE_LIBNAME:"%"=%)
> +RTE_LIBVERS := $(CONFIG_RTE_LIBVERS:"%"=%)
>  ifeq ($(RTE_LIBNAME),)
>  RTE_LIBNAME := intel_dpdk
> +RTE_LIBVERS := 2
>  endif
>  
>  # RTE_TARGET is deducted from config when we are building the SDK.
>