SH patches for 5.12

SH patches for 5.12

[John Paul Adrian Glaubitz]

Hi!

Could we get the following patches picked up for 5.12?

> - [PATCH] [sh] fix trivial misannotations
> - https://marc.info/?l=linux-kernel&m=160945707001399&w=2
>
> - [PATCH] sh: check return code of request_irq
> - https://marc.info/?l=linux-kernel&m=160867050030140&w=2
>
> - [PATCH] sh: boards: Fix the cacography in irq.c
> - https://marc.info/?l=linux-sh&m=160578410511403&w=2
>
> - [PATCH 1/2] sh: boot: add intermediate vmlinux.bin* to targets instead of extra-y
> - https://marc.info/?l=linux-kernel&m=161088234517301&w=2
>
> - [PATCH 2/2] sh: boot: avoid unneeded rebuilds under arch/sh/boot/compressed/
> - https://marc.info/?l=linux-kernel&m=161088245817344&w=2
>
> - [PATCH] maple: fix wrong return value of maple_bus_init().
> - https://marc.info/?l=linux-kernel&m=160635878212678&w=2
>
> - [PATCH] sh: kdump: add some attribute to function
> - https://marc.info/?l=linux-kernel&m=160758311622653&w=2
>
> - [PATCH] sh: kernel: traps: remove unused variable
> - https://marc.info/?l=linux-kernel&m=160760435528709&w=2
>
> - [PATCH] scripts/recordmcount.pl: support big endian for ARCH sh
> - https://marc.info/?l=linux-sh&m=161296964604229&w=2

Adrian

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - glaubitz@debian.org
`. `'   Freie Universitaet Berlin - glaubitz@physik.fu-berlin.de
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913

Re: SH patches for 5.12

[Rob Landley]

On 2/21/21 3:42 AM, John Paul Adrian Glaubitz wrote:
> Hi!
> 
> Could we get the following patches picked up for 5.12?

I still have your previous email on this open in a window on my desktop, but the
weather in Austin's been a bit distracting recently. :)

>> - [PATCH] [sh] fix trivial misannotations
>> - https://marc.info/?l=linux-kernel&m=160945707001399&w=2

That is an _impressively_ broken archive. Even the 'raw' links screw up the
From: lines with despamming crap, and it corrupts them on top of that:

  $ wget  https://marc.info/?l=linux-kernel&m=160945707001399&q=mbox -O 1.patch
  $ git am ../sh/1.patch
  Applying: fix trivial misannotations
  error: corrupt patch at line 110

That's why I hadn't dealt with this faster, I have to track them down elsewhere
to get copies I can actually apply and test...

Rob

Re: SH patches for 5.12

[Rich Felker]

On Sun, Feb 21, 2021 at 10:42:52AM +0100, John Paul Adrian Glaubitz wrote:
> Hi!
> 
> Could we get the following patches picked up for 5.12?
> 
> > - [PATCH] [sh] fix trivial misannotations
> > - https://marc.info/?l=linux-kernel&m=160945707001399&w=2
> >
> > - [PATCH] sh: check return code of request_irq
> > - https://marc.info/?l=linux-kernel&m=160867050030140&w=2
> >
> > - [PATCH] sh: boards: Fix the cacography in irq.c
> > - https://marc.info/?l=linux-sh&m=160578410511403&w=2
> >
> > - [PATCH 1/2] sh: boot: add intermediate vmlinux.bin* to targets instead of extra-y
> > - https://marc.info/?l=linux-kernel&m=161088234517301&w=2
> >
> > - [PATCH 2/2] sh: boot: avoid unneeded rebuilds under arch/sh/boot/compressed/
> > - https://marc.info/?l=linux-kernel&m=161088245817344&w=2
> >
> > - [PATCH] maple: fix wrong return value of maple_bus_init().
> > - https://marc.info/?l=linux-kernel&m=160635878212678&w=2
> >
> > - [PATCH] sh: kdump: add some attribute to function
> > - https://marc.info/?l=linux-kernel&m=160758311622653&w=2
> >
> > - [PATCH] sh: kernel: traps: remove unused variable
> > - https://marc.info/?l=linux-kernel&m=160760435528709&w=2
> >
> > - [PATCH] scripts/recordmcount.pl: support big endian for ARCH sh
> > - https://marc.info/?l=linux-sh&m=161296964604229&w=2

Thanks! The last one seems to be upstream already and the rest LGTM.
I've applied them and will do some smoke checks now and then prepare
for -next.

Rich

Re: SH patches for 5.12

[John Paul Adrian Glaubitz]

Hi!

On 2/23/21 1:56 AM, Rich Felker wrote:
> Thanks! The last one seems to be upstream already and the rest LGTM.
> I've applied them and will do some smoke checks now and then prepare
> for -next.

Any timeline for when the PR is sent to Linus?

Adrian

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - glaubitz@debian.org
`. `'   Freie Universitaet Berlin - glaubitz@physik.fu-berlin.de
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913

Re: SH patches for 5.12

[Rich Felker]

On Tue, Mar 02, 2021 at 09:51:23AM +0100, John Paul Adrian Glaubitz wrote:
> Hi!
> 
> On 2/23/21 1:56 AM, Rich Felker wrote:
> > Thanks! The last one seems to be upstream already and the rest LGTM.
> > I've applied them and will do some smoke checks now and then prepare
> > for -next.
> 
> Any timeline for when the PR is sent to Linus?

They're in next but I got a warning from the bot on one (7ec713058f21)
and haven't checked it in detail yet. In case you have a quick idea:

[linux-next:master 11837/12788] arch/sh/boot/compressed/../../lib/ashldi3.c:6:11: warning: no previous prototype for '__ashldi3'

Maybe it's just a duplicate of a warning that was already there and
that should be dealt with too.

Rich

Re: SH patches for 5.12

[Geert Uytterhoeven]

Hi Rich,

On Tue, Mar 2, 2021 at 3:03 PM Rich Felker <dalias@libc.org> wrote:
> On Tue, Mar 02, 2021 at 09:51:23AM +0100, John Paul Adrian Glaubitz wrote:
> > On 2/23/21 1:56 AM, Rich Felker wrote:
> > > Thanks! The last one seems to be upstream already and the rest LGTM.
> > > I've applied them and will do some smoke checks now and then prepare
> > > for -next.
> >
> > Any timeline for when the PR is sent to Linus?
>
> They're in next but I got a warning from the bot on one (7ec713058f21)
> and haven't checked it in detail yet. In case you have a quick idea:
>
> [linux-next:master 11837/12788] arch/sh/boot/compressed/../../lib/ashldi3.c:6:11: warning: no previous prototype for '__ashldi3'
>
> Maybe it's just a duplicate of a warning that was already there and
> that should be dealt with too.

I think that can be ignored safely: similar functions like __ashrdi3()
also lack forward declarations on SH.

Looks like only microblaze and powerpc do have forward declarations
for the gcc helpers:
arch/microblaze/lib/libgcc.h
arch/powerpc/include/asm/asm-prototypes.h

One day we may move them to asm-generic, and start using them
everywhere...

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: SH patches for 5.12

[Arnd Bergmann]

On Tue, Mar 2, 2021 at 3:17 PM Geert Uytterhoeven <geert@linux-m68k.org> wrote:
> On Tue, Mar 2, 2021 at 3:03 PM Rich Felker <dalias@libc.org> wrote:
> > On Tue, Mar 02, 2021 at 09:51:23AM +0100, John Paul Adrian Glaubitz wrote:
> > > On 2/23/21 1:56 AM, Rich Felker wrote:
> > > > Thanks! The last one seems to be upstream already and the rest LGTM.
> > > > I've applied them and will do some smoke checks now and then prepare
> > > > for -next.
> > >
> > > Any timeline for when the PR is sent to Linus?
> >
> > They're in next but I got a warning from the bot on one (7ec713058f21)
> > and haven't checked it in detail yet. In case you have a quick idea:
> >
> > [linux-next:master 11837/12788] arch/sh/boot/compressed/../../lib/ashldi3.c:6:11: warning: no previous prototype for '__ashldi3'
> >
> > Maybe it's just a duplicate of a warning that was already there and
> > that should be dealt with too.
>
> I think that can be ignored safely: similar functions like __ashrdi3()
> also lack forward declarations on SH.
>
> Looks like only microblaze and powerpc do have forward declarations
> for the gcc helpers:
> arch/microblaze/lib/libgcc.h
> arch/powerpc/include/asm/asm-prototypes.h
>
> One day we may move them to asm-generic, and start using them
> everywhere...

I thought these were fundamentally architecture specific. Is it even
possible to declare them properly?

Note that the '-Wmissing-prototype' warning is currently not enabled
by default, only if you use 'make W=1', or alternatively the corresponding
sparse warning with 'make C=1'. There is however ongoing work from
Lee Jones and others to address all W=1 warnings and then enable
these by default.

       Arnd

Re: SH patches for 5.12

[John Paul Adrian Glaubitz]

On 3/2/21 3:17 PM, Geert Uytterhoeven wrote:
>> Maybe it's just a duplicate of a warning that was already there and
>> that should be dealt with too.
> 
> I think that can be ignored safely: similar functions like __ashrdi3()
> also lack forward declarations on SH.
> 
> Looks like only microblaze and powerpc do have forward declarations
> for the gcc helpers:
> arch/microblaze/lib/libgcc.h
> arch/powerpc/include/asm/asm-prototypes.h
> 
> One day we may move them to asm-generic, and start using them
> everywhere...

OK. Then let's send the PR for 5.12 and tackle this for 5.13?

Adrian

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - glaubitz@debian.org
`. `'   Freie Universitaet Berlin - glaubitz@physik.fu-berlin.de
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913

Re: SH patches for 5.12

[Geert Uytterhoeven]

Hi Arnd,

On Tue, Mar 2, 2021 at 3:27 PM Arnd Bergmann <arnd@arndb.de> wrote:
> On Tue, Mar 2, 2021 at 3:17 PM Geert Uytterhoeven <geert@linux-m68k.org> wrote:
> > On Tue, Mar 2, 2021 at 3:03 PM Rich Felker <dalias@libc.org> wrote:
> > > On Tue, Mar 02, 2021 at 09:51:23AM +0100, John Paul Adrian Glaubitz wrote:
> > > > On 2/23/21 1:56 AM, Rich Felker wrote:
> > > > > Thanks! The last one seems to be upstream already and the rest LGTM.
> > > > > I've applied them and will do some smoke checks now and then prepare
> > > > > for -next.
> > > >
> > > > Any timeline for when the PR is sent to Linus?
> > >
> > > They're in next but I got a warning from the bot on one (7ec713058f21)
> > > and haven't checked it in detail yet. In case you have a quick idea:
> > >
> > > [linux-next:master 11837/12788] arch/sh/boot/compressed/../../lib/ashldi3.c:6:11: warning: no previous prototype for '__ashldi3'
> > >
> > > Maybe it's just a duplicate of a warning that was already there and
> > > that should be dealt with too.
> >
> > I think that can be ignored safely: similar functions like __ashrdi3()
> > also lack forward declarations on SH.
> >
> > Looks like only microblaze and powerpc do have forward declarations
> > for the gcc helpers:
> > arch/microblaze/lib/libgcc.h
> > arch/powerpc/include/asm/asm-prototypes.h
> >
> > One day we may move them to asm-generic, and start using them
> > everywhere...
>
> I thought these were fundamentally architecture specific. Is it even
> possible to declare them properly?

AFAIK their asm implementation, and if the compiler generates calls
to them, is architecture-specific.  The C versions are identical.

In fact we do have shared C implementations in lib/.
And at least __ashldi3() and __ashrdi3() are identical for SH and lib/,
so the latter should be used.

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: SH patches for 5.12

[Arnd Bergmann]

On Tue, Mar 2, 2021 at 3:33 PM Geert Uytterhoeven <geert@linux-m68k.org> wrote:
> On Tue, Mar 2, 2021 at 3:27 PM Arnd Bergmann <arnd@arndb.de> wrote:
> > On Tue, Mar 2, 2021 at 3:17 PM Geert Uytterhoeven <geert@linux-m68k.org> wrote:
> > > On Tue, Mar 2, 2021 at 3:03 PM Rich Felker <dalias@libc.org> wrote:
> > > > On Tue, Mar 02, 2021 at 09:51:23AM +0100, John Paul Adrian Glaubitz wrote:
> > > > > On 2/23/21 1:56 AM, Rich Felker wrote:
> > > > > > Thanks! The last one seems to be upstream already and the rest LGTM.
> > > > > > I've applied them and will do some smoke checks now and then prepare
> > > > > > for -next.
> > > > >
> > > > > Any timeline for when the PR is sent to Linus?
> > > >
> > > > They're in next but I got a warning from the bot on one (7ec713058f21)
> > > > and haven't checked it in detail yet. In case you have a quick idea:
> > > >
> > > > [linux-next:master 11837/12788] arch/sh/boot/compressed/../../lib/ashldi3.c:6:11: warning: no previous prototype for '__ashldi3'
> > > >
> > > > Maybe it's just a duplicate of a warning that was already there and
> > > > that should be dealt with too.
> > >
> > > I think that can be ignored safely: similar functions like __ashrdi3()
> > > also lack forward declarations on SH.
> > >
> > > Looks like only microblaze and powerpc do have forward declarations
> > > for the gcc helpers:
> > > arch/microblaze/lib/libgcc.h
> > > arch/powerpc/include/asm/asm-prototypes.h
> > >
> > > One day we may move them to asm-generic, and start using them
> > > everywhere...
> >
> > I thought these were fundamentally architecture specific. Is it even
> > possible to declare them properly?
>
> AFAIK their asm implementation, and if the compiler generates calls
> to them, is architecture-specific.  The C versions are identical.
>
> In fact we do have shared C implementations in lib/.
> And at least __ashldi3() and __ashrdi3() are identical for SH and lib/,
> so the latter should be used.

Ah, makes sense. Anything that has an implementation in lib/ is
probably generic anyway. I got a little confused because on ARM
there are separate names for these based on the ABI (oabi/eabi),
with the default version using non-generic names.

        Arnd

Re: SH patches for 5.12

[Rich Felker]

On Tue, Mar 02, 2021 at 03:33:58PM +0100, Geert Uytterhoeven wrote:
> Hi Arnd,
> 
> On Tue, Mar 2, 2021 at 3:27 PM Arnd Bergmann <arnd@arndb.de> wrote:
> > On Tue, Mar 2, 2021 at 3:17 PM Geert Uytterhoeven <geert@linux-m68k.org> wrote:
> > > On Tue, Mar 2, 2021 at 3:03 PM Rich Felker <dalias@libc.org> wrote:
> > > > On Tue, Mar 02, 2021 at 09:51:23AM +0100, John Paul Adrian Glaubitz wrote:
> > > > > On 2/23/21 1:56 AM, Rich Felker wrote:
> > > > > > Thanks! The last one seems to be upstream already and the rest LGTM.
> > > > > > I've applied them and will do some smoke checks now and then prepare
> > > > > > for -next.
> > > > >
> > > > > Any timeline for when the PR is sent to Linus?
> > > >
> > > > They're in next but I got a warning from the bot on one (7ec713058f21)
> > > > and haven't checked it in detail yet. In case you have a quick idea:
> > > >
> > > > [linux-next:master 11837/12788] arch/sh/boot/compressed/../../lib/ashldi3.c:6:11: warning: no previous prototype for '__ashldi3'
> > > >
> > > > Maybe it's just a duplicate of a warning that was already there and
> > > > that should be dealt with too.
> > >
> > > I think that can be ignored safely: similar functions like __ashrdi3()
> > > also lack forward declarations on SH.
> > >
> > > Looks like only microblaze and powerpc do have forward declarations
> > > for the gcc helpers:
> > > arch/microblaze/lib/libgcc.h
> > > arch/powerpc/include/asm/asm-prototypes.h
> > >
> > > One day we may move them to asm-generic, and start using them
> > > everywhere...
> >
> > I thought these were fundamentally architecture specific. Is it even
> > possible to declare them properly?
> 
> AFAIK their asm implementation, and if the compiler generates calls
> to them, is architecture-specific.  The C versions are identical.
> 
> In fact we do have shared C implementations in lib/.
> And at least __ashldi3() and __ashrdi3() are identical for SH and lib/,
> so the latter should be used.

I haven't looked at the code, but are you sure this is okay? C
implementations of variable bitshift seem subject to the compiler
auto-replacing them with <<n and >>n, which would then create a
recursive definition on archs (pre-sh3 only) that don't have a barrel
shift instruction and need the library calls.

Rich

Re: SH patches for 5.12

[Geert Uytterhoeven]

On Tue, Mar 2, 2021 at 3:33 PM Geert Uytterhoeven <geert@linux-m68k.org> wrote:
> On Tue, Mar 2, 2021 at 3:27 PM Arnd Bergmann <arnd@arndb.de> wrote:
> > On Tue, Mar 2, 2021 at 3:17 PM Geert Uytterhoeven <geert@linux-m68k.org> wrote:
> > > On Tue, Mar 2, 2021 at 3:03 PM Rich Felker <dalias@libc.org> wrote:
> > > > On Tue, Mar 02, 2021 at 09:51:23AM +0100, John Paul Adrian Glaubitz wrote:
> > > > > On 2/23/21 1:56 AM, Rich Felker wrote:
> > > > > > Thanks! The last one seems to be upstream already and the rest LGTM.
> > > > > > I've applied them and will do some smoke checks now and then prepare
> > > > > > for -next.
> > > > >
> > > > > Any timeline for when the PR is sent to Linus?
> > > >
> > > > They're in next but I got a warning from the bot on one (7ec713058f21)
> > > > and haven't checked it in detail yet. In case you have a quick idea:
> > > >
> > > > [linux-next:master 11837/12788] arch/sh/boot/compressed/../../lib/ashldi3.c:6:11: warning: no previous prototype for '__ashldi3'
> > > >
> > > > Maybe it's just a duplicate of a warning that was already there and
> > > > that should be dealt with too.
> > >
> > > I think that can be ignored safely: similar functions like __ashrdi3()
> > > also lack forward declarations on SH.
> > >
> > > Looks like only microblaze and powerpc do have forward declarations
> > > for the gcc helpers:
> > > arch/microblaze/lib/libgcc.h
> > > arch/powerpc/include/asm/asm-prototypes.h
> > >
> > > One day we may move them to asm-generic, and start using them
> > > everywhere...
> >
> > I thought these were fundamentally architecture specific. Is it even
> > possible to declare them properly?
>
> AFAIK their asm implementation, and if the compiler generates calls
> to them, is architecture-specific.  The C versions are identical.
>
> In fact we do have shared C implementations in lib/.
> And at least __ashldi3() and __ashrdi3() are identical for SH and lib/,
> so the latter should be used.

SH might never have been converted as the files are also used for the
boot loader.  Seems to work, though, will send a patch shortly.

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: SH patches for 5.12

[Geert Uytterhoeven]

Hi Rich,

On Tue, Mar 2, 2021 at 3:48 PM Rich Felker <dalias@libc.org> wrote:
> On Tue, Mar 02, 2021 at 03:33:58PM +0100, Geert Uytterhoeven wrote:
> > On Tue, Mar 2, 2021 at 3:27 PM Arnd Bergmann <arnd@arndb.de> wrote:
> > > On Tue, Mar 2, 2021 at 3:17 PM Geert Uytterhoeven <geert@linux-m68k.org> wrote:
> > > > On Tue, Mar 2, 2021 at 3:03 PM Rich Felker <dalias@libc.org> wrote:
> > > > > On Tue, Mar 02, 2021 at 09:51:23AM +0100, John Paul Adrian Glaubitz wrote:
> > > > > > On 2/23/21 1:56 AM, Rich Felker wrote:
> > > > > > > Thanks! The last one seems to be upstream already and the rest LGTM.
> > > > > > > I've applied them and will do some smoke checks now and then prepare
> > > > > > > for -next.
> > > > > >
> > > > > > Any timeline for when the PR is sent to Linus?
> > > > >
> > > > > They're in next but I got a warning from the bot on one (7ec713058f21)
> > > > > and haven't checked it in detail yet. In case you have a quick idea:
> > > > >
> > > > > [linux-next:master 11837/12788] arch/sh/boot/compressed/../../lib/ashldi3.c:6:11: warning: no previous prototype for '__ashldi3'
> > > > >
> > > > > Maybe it's just a duplicate of a warning that was already there and
> > > > > that should be dealt with too.
> > > >
> > > > I think that can be ignored safely: similar functions like __ashrdi3()
> > > > also lack forward declarations on SH.
> > > >
> > > > Looks like only microblaze and powerpc do have forward declarations
> > > > for the gcc helpers:
> > > > arch/microblaze/lib/libgcc.h
> > > > arch/powerpc/include/asm/asm-prototypes.h
> > > >
> > > > One day we may move them to asm-generic, and start using them
> > > > everywhere...
> > >
> > > I thought these were fundamentally architecture specific. Is it even
> > > possible to declare them properly?
> >
> > AFAIK their asm implementation, and if the compiler generates calls
> > to them, is architecture-specific.  The C versions are identical.
> >
> > In fact we do have shared C implementations in lib/.
> > And at least __ashldi3() and __ashrdi3() are identical for SH and lib/,
> > so the latter should be used.
>
> I haven't looked at the code, but are you sure this is okay? C
> implementations of variable bitshift seem subject to the compiler
> auto-replacing them with <<n and >>n, which would then create a
> recursive definition on archs (pre-sh3 only) that don't have a barrel
> shift instruction and need the library calls.

The SH version is a C implementation, too.

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds