On Tue, May 31, 2011 at 7:17 AM, Dmitry Eremin-Solenikov <dbarysh...@gmail.com> wrote: > Hello, > > Long time ago (around July of 2010), I've reported a problem of binding > pci error reporting driver to the hardware (See > http://thread.gmane.org/gmane.linux.kernel.edac/181). Grant Likely has > come with a patch to create platform devices (of_platform at that time), > for device_nodes which have "compatible" properties > (http://article.gmane.org/gmane.linux.ports.ppc.embedded/36981). > > The patch was not applied at that time and the problem still arises at > 3.0-rc1. Should the patch in question be applied, or is there any new > solution? > Current version of that patch which I have in my tree is attached > to this mail (no original signoff by Grant).
Hi Dmitry, I've since looked at the problem more, and I've been working on refactoring both of_platform_bus_probe(), and creating a new function, of_platform_populate(), which implements the symantics that I think are needed for your use case. The work isn't done yet, but it has gone out for at least one round of review, and I hope to get it into the 3.1 merge window. g. _______________________________________________ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
