[oe] [meta-ti] ti81xx-psp-linux-2.6.37: broken kernel with gcc > 4.5

[Monk, Roger]

On Fri, May 18, 2012 at 15:40:25, Thilo Fromm wrote:
> Cc: meta-ti at yoctoproject.org;
> openembedded-devel at lists.openembedded.org
> Subject: Re: [meta-ti] [oe] ti81xx-psp-linux-2.6.37: broken kernel
> with gcc > 4.5
>
> Hello Koen,
>
> >>>>>> sorry for the cross-post (openembedded-devel, meta-ti), I think
> >>>>>> this affects both projects one way or the other.
> >>>>>>
> >>>>>> Recently, meta-openembedded/toolchain-layer removed gcc-4.5.
> This
> >>>>>> means with oe-core + meta-oe you now can choose between gcc-4.6
> and gcc-4.7.
> >>>>>> However, both compilers break the kernel we're using
> >>>>>> (ti81xx-psp-linux-2.6.37).
> >>>>>
> >>>>> What happens when you disable the mailbox? AIUI you shouldn't
> need
> >>>>> that on Ne/Ce.
> >>>>
> >>>> How do I disable the mailbox?
> >>>
> >>> In the .config
> >>
> >> Wait wait wait, according to drivers/staging/tidspbridge/Kconfig we
> >> need OMAP_MBOX_FWK (i.e. "the mailbox") in order to use the DSP
> >> bridge. Disabling it would mean cutting off our DSP. No way.
> >
> > Are you actually using dspbridge? Or are you using dsplink or
> syslink?
>
> No, I was wrong. OMAP_MBOX_FWK is forced to "Y" in our config and I
> thought DSPBRIDGE forced this. But DSPBRIDGE isn't even active in our
> configuration. I cannot turn OMAP_MBOX_FWK off, it's forced to Y, and
> I'm currently trying to figure out which part of ti81xx-psp-linux-
> 2.6.37's wicked configuration is causing this.

CONFIG_SYSLINK_NOTIFY turns this on.  Syslink uses the same mbox infrastructure.

>
> >>>> What do you mean by Ne/Ce?
> >>>
> >>> Netra/Centaurus, dm81xx, c6a81xx, am38xx, am39xx or whatever TI is
> calling it nowadays.
> >>
> >> We're on a "TI DaVinci DM8148", featuring a cortex-a8 combined with
> a
> >> C674x DSP.
> >
> > So a centaurus family SoC.
>
> Thanks for the clarification :) TI didn't approach us with any of the
> terms you mentioned. It's just DaVinci for them when they're talking
> with us.
>
> >
> >> I think the only ones left without DSPs are AM-...
> >> processors. So no, the DSP bridge (and therefore mailboxing) must
> >> stay.
> >
> > See what I said above, I doubt you're using dspbridge.
>
> You may be right; I didn't dig into DSP communication so far. Also,
> DSPBRIDGE depends on OMAP3, which we don't have (dm8148 is a
> "OMAP2PLUS"). So I was wrong with DSPBRIDGE forcing OMAP_MBOX_FWK.
> OMAP_MBOX_FWK is just active and I can't switch it off.
>
> Regards,
> Thilo
>
> --
> Dipl.-Ing (FH) Thilo Fromm, MSc., Embedded Systems Architect DResearch
> Fahrzeugelektronik GmbH Otto-Schmirgal-Str. 3, D-10319 Berlin, Germany
> Tel: +49 (30) 515 932 228   mailto:fromm at dresearch-fe.de
> Fax: +49 (30) 515 932 77    http://www.dresearch.de
> Amtsgericht: Berlin Charlottenburg, HRB 130120 B Ust.-IDNr.
> DE273952058
> Geschäftsführer: Dr. M. Weber, W. Mögle
>
Texas Instruments Limited, 800 Pavilion Drive, Northampton, NN4 7YL. Registered in England & Wales under company number 00574102

_______________________________________________
> meta-ti mailing list
> meta-ti at yoctoproject.org
> https://lists.yoctoproject.org/listinfo/meta-ti