[PATCH] m68k/mac: Make SCC reset work more reliably

[PATCH] m68k/mac: Make SCC reset work more reliably

[Finn Thain]

For SCC initialization we cannot assume that the control register is in 
the correct state to accept a register pointer. So first read from the 
control register in order to "sync" up.

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>

---

This patch fixes a boot failure (presumably caused by unhandled SCC 
interrupt) on PowerBook 180 with CONFIG_EARLY_PRINTK and with the 
bootloader not configured to wake up the serial ports. AFAIK this happens 
only on PowerBooks. Even with this fix, Penguin still has to wake the 
serial ports or else they don't function on PowerBooks like this one.

diff --git a/arch/m68k/kernel/head.S b/arch/m68k/kernel/head.S
--- a/arch/m68k/kernel/head.S
+++ b/arch/m68k/kernel/head.S
@@ -2909,7 +2909,9 @@ func_start	serial_init,%d0/%d1/%a0/%a1
 
 #if defined(MAC_USE_SCC_A) || defined(MAC_USE_SCC_B)
 	movel	%pc@(L(mac_sccbase)),%a0
-	/* Reset SCC device */
+	/* Reset SCC register pointer */
+	moveb	%a0@(mac_scc_cha_a_ctrl_offset),%d0
+	/* Reset SCC device: write register pointer then register value */
 	moveb	#9,%a0@(mac_scc_cha_a_ctrl_offset)
 	moveb	#0xc0,%a0@(mac_scc_cha_a_ctrl_offset)
 	/* Wait for 5 PCLK cycles, which is about 68 CPU cycles */

Re: [PATCH] m68k/mac: Make SCC reset work more reliably

[Geert Uytterhoeven]

Hi Finn,

On Sun, Jan 12, 2014 at 2:56 PM, Finn Thain <fthain@telegraphics.com.au> wrote:
> For SCC initialization we cannot assume that the control register is in
> the correct state to accept a register pointer. So first read from the
> control register in order to "sync" up.
>
> Signed-off-by: Finn Thain <fthain@telegraphics.com.au>

Thaks for your patch!

> ---
>
> This patch fixes a boot failure (presumably caused by unhandled SCC
> interrupt) on PowerBook 180 with CONFIG_EARLY_PRINTK and with the
> bootloader not configured to wake up the serial ports. AFAIK this happens
> only on PowerBooks. Even with this fix, Penguin still has to wake the
> serial ports or else they don't function on PowerBooks like this one.

IMHO the above should be part of the patch description (i.e. above the
three dashes), as it contains valuable information. Is that OK for you?

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: [PATCH] m68k/mac: Make SCC reset work more reliably

[Finn Thain]

On Mon, 13 Jan 2014, Geert Uytterhoeven wrote:

> Hi Finn,
> 
> On Sun, Jan 12, 2014 at 2:56 PM, Finn Thain <fthain@telegraphics.com.au> 
> wrote:
> > For SCC initialization we cannot assume that the control register is 
> > in the correct state to accept a register pointer. So first read from 
> > the control register in order to "sync" up.
> >
> > Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
> 
> Thaks for your patch!
> 
> > ---
> >

> > This patch fixes a boot failure (presumably caused by unhandled SCC 
> > interrupt) on PowerBook 180 with CONFIG_EARLY_PRINTK and with the 
> > bootloader not configured to wake up the serial ports.

> > AFAIK this happens only on PowerBooks. Even with this fix, Penguin 
> > still has to wake the serial ports or else they don't function on 
> > PowerBooks like this one.

> 
> IMHO the above should be part of the patch description (i.e. above the 
> three dashes), as it contains valuable information. Is that OK for you?

The first part is good information, especially if this were submitted to 
-stable.

The rest, beginning with "AFAIK", is really guesswork based on anecdote 
(e.g. reports from users not affected). Even the mailing list archives are 
perhaps too permanent (e.g. the need for the wake-up workaround may go 
away one day).

I'm unsure about submitting this to -stable. I don't like the idea that 
the first debian user to boot a powerbook might see a crash (depending on 
bootloader etc). But not all stable trees carry df66834a which is a 
prerequisite. This patch is only applicable to 3.4.y and >= 3.9.y. Can you 
nominate particular stable trees?

Finn

Re: [PATCH] m68k/mac: Make SCC reset work more reliably

[Geert Uytterhoeven]

On Tue, Jan 14, 2014 at 9:31 AM, Finn Thain <fthain@telegraphics.com.au> wrote:
> On Mon, 13 Jan 2014, Geert Uytterhoeven wrote:
>> On Sun, Jan 12, 2014 at 2:56 PM, Finn Thain <fthain@telegraphics.com.au>
>> wrote:
>> > For SCC initialization we cannot assume that the control register is
>> > in the correct state to accept a register pointer. So first read from
>> > the control register in order to "sync" up.
>> >
>> > Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
>>
>> Thaks for your patch!
>>
>> > ---
>> >
>
>> > This patch fixes a boot failure (presumably caused by unhandled SCC
>> > interrupt) on PowerBook 180 with CONFIG_EARLY_PRINTK and with the
>> > bootloader not configured to wake up the serial ports.
>
>> > AFAIK this happens only on PowerBooks. Even with this fix, Penguin
>> > still has to wake the serial ports or else they don't function on
>> > PowerBooks like this one.
>
>> IMHO the above should be part of the patch description (i.e. above the
>> three dashes), as it contains valuable information. Is that OK for you?
>
> The first part is good information, especially if this were submitted to
> -stable.
>
> The rest, beginning with "AFAIK", is really guesswork based on anecdote
> (e.g. reports from users not affected). Even the mailing list archives are
> perhaps too permanent (e.g. the need for the wake-up workaround may go
> away one day).

OK.

> I'm unsure about submitting this to -stable. I don't like the idea that
> the first debian user to boot a powerbook might see a crash (depending on
> bootloader etc). But not all stable trees carry df66834a which is a
> prerequisite. This patch is only applicable to 3.4.y and >= 3.9.y. Can you
> nominate particular stable trees?

Sure, just add it to the comment, cfr. Documentation/stable_kernel_rules.txt:

 - If the patch requires other patches as prerequisites which can be
   cherry-picked than this can be specified in the following format in
   the sign-off area:

     Cc: <stable@vger.kernel.org> # 3.3.x: a1f84a3: sched: Check for idle
     Cc: <stable@vger.kernel.org> # 3.3.x: 1b9508f: sched: Rate-limit newidle
     Cc: <stable@vger.kernel.org> # 3.3.x: fd21073: sched: Fix affinity logic
     Cc: <stable@vger.kernel.org> # 3.3.x

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: [PATCH] m68k/mac: Make SCC reset work more reliably

[Finn Thain]

On Tue, 14 Jan 2014, Geert Uytterhoeven wrote:

> 
> > I'm unsure about submitting this to -stable. I don't like the idea 
> > that the first debian user to boot a powerbook might see a crash 
> > (depending on bootloader etc). But not all stable trees carry df66834a 
> > which is a prerequisite. This patch is only applicable to 3.4.y and >= 
> > 3.9.y. Can you nominate particular stable trees?
> 
> Sure, just add it to the comment, cfr. 
> Documentation/stable_kernel_rules.txt:

Yes. I read the docs before I wrote.

> 
>  - If the patch requires other patches as prerequisites which can be 
>    cherry-picked than this can be specified in the following format in 
>    the sign-off area:
> 
>      Cc: <stable@vger.kernel.org> # 3.3.x: a1f84a3: sched: Check for idle
>      Cc: <stable@vger.kernel.org> # 3.3.x: 1b9508f: sched: Rate-limit newidle
>      Cc: <stable@vger.kernel.org> # 3.3.x: fd21073: sched: Fix affinity logic
>      Cc: <stable@vger.kernel.org> # 3.3.x

I didn't want the commit to be cherry-picked as it's already there (under 
different commit IDs) in some stable trees.

The prerequisite commit is df66834a in mainline (merged in 3.10). It is 
known as f64747c9 in 3.9.y and 8cfd67a3 in 3.4.y.

I only checked the trees in linux-stable.git which is not exhaustive but 
there are some stable trees that don't have it at all.

Perhaps it would be best to request that the new patch only gets applied 
where the prerequisite patch was already in mainline, i.e. >= 3.10.

Finn

> 
> Gr{oetje,eeting}s,
> 
>                         Geert
> 

Re: [PATCH] m68k/mac: Make SCC reset work more reliably

[Thorsten Glaser]

Finn Thain <fthain <at> telegraphics.com.au> writes:

> Perhaps it would be best to request that the new patch only gets applied 
> where the prerequisite patch was already in mainline, i.e. >= 3.10.

FWIW Debian has 3.12, with 3.13 coming Soon™.

//mirabilos

Re: [PATCH] m68k/mac: Make SCC reset work more reliably

[Geert Uytterhoeven]

On Sun, Jan 12, 2014 at 2:56 PM, Finn Thain <fthain@telegraphics.com.au> wrote:
> For SCC initialization we cannot assume that the control register is in
> the correct state to accept a register pointer. So first read from the
> control register in order to "sync" up.
>
> Signed-off-by: Finn Thain <fthain@telegraphics.com.au>

Thanks, applied and queued for 3.14.

We can decide about the stableness after it has been merged into
upstream.

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