[meta-ti] [PATCH] ipc: IPC for communication between multiple processors

[Nelson, Sam]

> -----Original Message-----
> From: Dmytriyenko, Denys
> Sent: Thursday, November 21, 2013 6:41 PM
> To: Ring, Chris
> Cc: Maupin, Chase; Hingolikar, Mrinmayee; meta-ti at yoctoproject.org;
> Nelson, Sam
> Subject: Re: [meta-ti] [PATCH] ipc: IPC for communication between multiple
> processors
> 
> On Thu, Nov 21, 2013 at 06:38:55PM -0500, Ring, Chris wrote:
> > Below...
> >
> > > -----Original Message-----
> > > From: Dmytriyenko, Denys
> > > Sent: Wednesday, November 20, 2013 7:19 AM
> > > To: Maupin, Chase
> > > Cc: Hingolikar, Mrinmayee; meta-ti at yoctoproject.org; Ring, Chris;
> Nelson,
> > > Sam
> > > Subject: Re: [meta-ti] [PATCH] ipc: IPC for communication between
> multiple
> > > processors
> > >
> > > See my comments below.
> > >
> > > <chop>
> > >
> > > > >+
> > > > >+Signed-off-by: Mrinmayee Hingolikar <mrinmayee at ti.com>
> > > > >+---
> > > > >+ ipc-linux.mak |    1 +
> > > > >+ products.mak  |    1 +
> > > > >+ 2 files changed, 2 insertions(+)
> > > > >+
> > > > >+diff --git a/ipc-linux.mak b/ipc-linux.mak
> > > > >+index 788a5a2..01579e4 100644
> > > > >+--- a/ipc-linux.mak
> > > > >++++ b/ipc-linux.mak
> > > > >+@@ -38,6 +38,7 @@ include products.mak
> > > > >+ config:
> > > > >+ 	@echo "Configuring Linux Ipc ..."
> > > > >+ 	./configure --host=$(TOOLCHAIN_LONGNAME) \
> > > > >++            --prefix=$(PREFIX) \
> > > > >+             CC=$(TOOLCHAIN_PREFIX)gcc \
> > > > >+             PLATFORM=$(PLATFORM) \
> > > > >+             CMEM_INSTALL_DIR=$(CMEM_INSTALL_DIR) \
> > > >
> > > > This line makes me think there is a dependency missing.  In the past
> > > > something like ti-paths.inc would pass this value along.  I'm not saying
> > > > that should be replicated as is, but likely you want to se this to
> > > > STAGING_DIR_TARGET or something similar and the cmem recipe
> should stage
> > > > appropriately.
> > > >
> > > > I think it would be good if you walked through your software stack to
> > > make
> > > > sure you are sending your patches in order.  Likewise, there was some
> > > > discussion with Sam about cmem recipes and basically about aligning
> > > recipes
> > > > between your groups.  I thin the approach of breaking linux-utils up
> > > into a
> > > > recipe per component is going to be the cleanest approach.  I'll ping on
> > > > Sam's patches as well to get an update posted to meta-ti instead of
> > > > meta-arago list.
> > > >
> > > > Any pointers you can provide to documentation about the configuration
> > > and
> > > > building of these components you are pushing up would also be useful
> to
> > > help
> > > > in reviewing so we can understand what you are trying to accomplish.
> > >
> > > Nice catch. The first question - does it depend and require cmem?
> >
> > The IPC build, including required variables, is described here:
> >     http://processors.wiki.ti.com/index.php/IPC_Install_Guide_Linux
> >
> > CMEM_INSTALL_DIR is only required if you want to build the examples
> which
> > leverage CMEM.  It shouldn't be needed if you only want to build the IPC
> > libraries.
> 
> Thanks, Chris!
> 
> Sam, Mrinmayee, can you comment on whether you need IPC examples
> and/or pass
> CMEM_INSTALL_DIR. Thanks.
[Sam] My general thought on this, is that  the test or examples should go into a separate recipe may be ipc-test or ipc-examples. Which in turn can depend on multiple components.  This I guess need changes to build the examples separately.  
> 
> --
> Denys