Re: [GIT PULL] sh: remove sh5 support

[Andrew Morton]

On Fri, 5 Jun 2020 20:23:07 +0200 Geert Uytterhoeven <geert@xxxxxxxxxxxxxx> wrote:

> Hi Adrian,
> 
> On Fri, Jun 5, 2020 at 7:58 PM John Paul Adrian Glaubitz
> <glaubitz@xxxxxxxxxxxxxxxxxxx> wrote:
> > On 6/5/20 5:59 PM, Rich Felker wrote:
> > >> Ah, sorry, I missed that. You're right, it should probably go through
> > >> someone else's tree then.
> > >
> > > Do you know if it's needed to un-break sh4? If so we should push to
> > > get whoever has jurisdiction over it to include it; otherwise I'm
> > > indifferent.
> 
> I think the above is about the asm-generic/vmlinux.lds.h patch?
> https://marc.info/?l=linux-sh&m=158429470221261&w=2
> 
> That one falls under Arnd's jurisdiction.
> And AFAIUC, it matters for recent binutils only? So Arnd should know.
> 

Aligning RO_AFTER_INIT_DATA is simply a no-brainer:

--- a/include/asm-generic/vmlinux.lds.h~include-asm-generic-vmlinuxldsh-align-ro_after_init
+++ a/include/asm-generic/vmlinux.lds.h
@@ -374,6 +374,7 @@
  */
 #ifndef RO_AFTER_INIT_DATA
 #define RO_AFTER_INIT_DATA						\
+	. = ALIGN(8);							\
 	__start_ro_after_init = .;					\
 	*(.data..ro_after_init)						\
 	JUMP_TABLE_DATA							\

So I'll send this in to Linus for 5.8-rc1.  And I think I'll add a cc:stable to
it, so people can use older kernels with newer toolchains.


I still have a few other sh patches sitting here:

https://ozlabs.org/~akpm/mmotm/broken-out/sh-clkfwk-remove-r8-r16-r32.patch
https://ozlabs.org/~akpm/mmotm/broken-out/sh-remove-call-to-memset-after-dma_alloc_coherent.patch
https://ozlabs.org/~akpm/mmotm/broken-out/sh-use-generic-strncpy.patch
https://ozlabs.org/~akpm/mmotm/broken-out/sh-add-missing-export_symbol-for-__delay.patch

What should I do with those?