[PATCH] [m68k] Atari: support for NetUSBee register access to isp116x.h

[PATCH] [m68k] Atari: support for NetUSBee register access to isp116x.h

[Michael Schmitz]

Geert,

the following patch allows selecting the correct register access
macros depending on whether the EtherNAT or NetUSBee variant of the
ISP1160 chip is used. With this patch applied on top of the previous
two, it should be possible to test the NetUSBee USB port.

Cheers,

  Michael


index 198b4bb..c8f0612 100644
--- a/drivers/usb/host/isp116x.h
+++ b/drivers/usb/host/isp116x.h
@@ -367,16 +367,16 @@ struct isp116x_ep {

 #ifdef CONFIG_ATARI
   /* 16 bit data bus byte swapped in hardware */
-#define isp_readw               __raw_readw
-#define isp_writew              __raw_writew
-#define isp_raw_readw           readw
-#define isp_raw_writew          writew
+#define isp_readw(p)           (__pa(p) < 1024 ? raw_rom_inw((p)) :
__raw_readw((p)))
+#define isp_writew(v,p)                (__pa(p) < 1024 ?
raw_rom_outw((v),(p)) : __raw_writew((v),(p)))
+#define isp_raw_readw(p)       (__pa(p) < 1024 ? rom_in_le16((p)) : readw((p)))
+#define isp_raw_writew(v,p)    (__pa(p) < 1024 ?
rom_out_le16((p),(v)) : writew((v),(p)))
 #else
   /* sane hardware */
-#define isp_readw               readw
-#define isp_writew              writew
-#define isp_raw_readw           __raw_readw
-#define isp_raw_writew          __raw_writew
+#define isp_readw              readw
+#define isp_writew             writew
+#define isp_raw_readw          __raw_readw
+#define isp_raw_writew         __raw_writew
 #endif

 static inline void isp116x_write_addr(struct isp116x *isp116x, unsigned reg)

Re: [PATCH] [m68k] Atari: support for NetUSBee register access to isp116x.h

[Geert Uytterhoeven]

Hi Michael,

On Tue, May 29, 2012 at 12:48 AM, Michael Schmitz
<schmitzmic@googlemail.com> wrote:
> the following patch allows selecting the correct register access
> macros depending on whether the EtherNAT or NetUSBee variant of the
> ISP1160 chip is used. With this patch applied on top of the previous
> two, it should be possible to test the NetUSBee USB port.

Thanks!

Just like a few of the previous patches, this one is severely
white-space damaged :-(
In addition, several of the patches showed checkpatch errors.
For bedtime reading, I can recommend linux/Documentation/email-clients.txt and
linux/Documentation/SubmittingPatches.

After some manual fixes, I did manage to apply all of them to my
private working tree.
Everything builds and boots (on ARAnyM) fine. I will comment on the
individual patches
when time permits.

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] Atari: support for NetUSBee register access to isp116x.h

[Michael Schmitz]

Hi Geert,

On Tue, May 29, 2012 at 10:46 PM, Geert Uytterhoeven
<geert@linux-m68k.org> wrote:
> Hi Michael,
>
> On Tue, May 29, 2012 at 12:48 AM, Michael Schmitz
> <schmitzmic@googlemail.com> wrote:
>> the following patch allows selecting the correct register access
>> macros depending on whether the EtherNAT or NetUSBee variant of the
>> ISP1160 chip is used. With this patch applied on top of the previous
>> two, it should be possible to test the NetUSBee USB port.
>
> Thanks!
>
> Just like a few of the previous patches, this one is severely
> white-space damaged :-(

Which ones were affected? (Some were piped straight to sendmail while
others used an e-mail client so I'd like to know what does actually
work)

> In addition, several of the patches showed checkpatch errors.

I need to teach git to run checkpatch for me :-(

> For bedtime reading, I can recommend linux/Documentation/email-clients.txt and
> linux/Documentation/SubmittingPatches.
>
> After some manual fixes, I did manage to apply all of them to my
> private working tree.

Thanks for making the effort!

> Everything builds and boots (on ARAnyM) fine. I will comment on the
> individual patches
> when time permits.

Thanks - this was all tested on the Falcon.

  Michael

Re: [PATCH] [m68k] Atari: support for NetUSBee register access to isp116x.h

[Geert Uytterhoeven]

On Fri, Jun 1, 2012 at 8:46 PM, Michael Schmitz
<schmitzmic@googlemail.com> wrote:
>> Just like a few of the previous patches, this one is severely
>> white-space damaged :-(
>
> Which ones were affected? (Some were piped straight to sendmail while
> others used an e-mail client so I'd like to know what does actually
> work)

The first large series (11 or so) was OK.
The replacement for the 11th patch was damaged.
The small series of 2 was also damaged.

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] Atari: support for NetUSBee register access to isp116x.h

[Thorsten Glaser]

Michael Schmitz <schmitzmic <at> googlemail.com> writes:

> Which ones were affected? (Some were piped straight to sendmail while
> others used an e-mail client so I'd like to know what does actually
> work)

I found out that the following works reliably:

① scp/rsync files to a box I can mail from
② $EDITOR 00*
③ Add a To: line, and possibly a In-Reply-To: line referencing
  the Message-ID from the introductory mail (to get them threaded),
  to each file; save and exit https://www.mirbsd.org/jupp.htm ^Wthe editor
④ for a in 00*; do sendmail -oem -oi -t -f sender@mydomain.com <"$a"; done

!BUT!

This does only work if the patchfile does NOT contain any non-7bit
characters. (This has affected me in the past when writing commit
messages with “typogra’phic” marks, as I’m used to, nowadays.) There
is *no* reliable way to transport inline patches through SMTP when
either the commit message or the patches/the patch context contain
such characters, as the SMTP transport path is probably not 8-bit
transparent (may convert to QP or Base64, or strip bit7).

So, in case I ever need to send such mails because the nōn-7bit
part is in the patch (as opposed to the commit message), please
bear with me when I need to attach them to a regular eMail.

bye,
//mirabilos – via GMane

Re: [PATCH] [m68k] Atari: support for NetUSBee register access to isp116x.h

[Geert Uytterhoeven]

On Tue, Jun 5, 2012 at 5:51 PM, Thorsten Glaser <tg@mirbsd.org> wrote:
> ② $EDITOR 00*
> ③ Add a To: line, and possibly a In-Reply-To: line referencing
>  the Message-ID from the introductory mail (to get them threaded),
>  to each file; save and exit https://www.mirbsd.org/jupp.htm ^Wthe editor
> ④ for a in 00*; do sendmail -oem -oi -t -f sender@mydomain.com <"$a"; done

"git send-email --to user@domain 00*"?
That can also take care of the threading.

> !BUT!
>
> This does only work if the patchfile does NOT contain any non-7bit
> characters. (This has affected me in the past when writing commit
> messages with “typogra’phic” marks, as I’m used to, nowadays.) There
> is *no* reliable way to transport inline patches through SMTP when
> either the commit message or the patches/the patch context contain
> such characters, as the SMTP transport path is probably not 8-bit
> transparent (may convert to QP or Base64, or strip bit7).
>
> So, in case I ever need to send such mails because the nōn-7bit
> part is in the patch (as opposed to the commit message), please
> bear with me when I need to attach them to a regular eMail.

Does git send-email handle that correctly?
It will not use base64, as that's rejected by vger.kernel.org.

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] Atari: support for NetUSBee register access to isp116x.h

[Thorsten Glaser]

Geert Uytterhoeven dixit:

>On Tue, Jun 5, 2012 at 5:51 PM, Thorsten Glaser <tg@mirbsd.org> wrote:
>> ② $EDITOR 00*
>> ③ Add a To: line, and possibly a In-Reply-To: line referencing
>>  the Message-ID from the introductory mail (to get them threaded),
>>  to each file; save and exit https://www.mirbsd.org/jupp.htm ^Wthe editor
>> ④ for a in 00*; do sendmail -oem -oi -t -f sender@mydomain.com <"$a"; done
>
>"git send-email --to user@domain 00*"?

You conveniently cut off Item #1 above, which involves copying
the patches to a machine that can mail out.

>Does git send-email handle that correctly?

There is no way to handle 8bit correctly in SMTP, that does not
involve converting them to QP or Base64.

>It will not use base64, as that's rejected by vger.kernel.org.

Interesting. How are QP encoded patches handled (other than,
what I assume, spit from Linux developers)?

Honestly, the requirement to mail patches inline and not MIME
encoded reminds me of my time trying to contribute to OpenBSD
(before that ended due to personal, not technical, issues),
where the lead developers used non-MIME mailx(1) to read mails…
I think accepting an encoded mail (not necessarily multipart,
i.e. “attachment”, but to put it through the SMTP path) and
decoding it in the “git am” part should be possible.

bye,
//mirabilos
-- 
13:37⎜«Natureshadow» Deep inside, I hate mirabilos. I mean, he's a good
guy. But he's always right! In every fsckin' situation, he's right. Even
with his deeply perverted taste in software and borked ambition towards
broken OSes - in the end, he's damn right about it :(! […] works in mksh

Re: [PATCH] [m68k] Atari: support for NetUSBee register access to isp116x.h

[Andreas Schwab]

How about using formail and mailx instead?

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: [PATCH] [m68k] Atari: support for NetUSBee register access to isp116x.h

[Geert Uytterhoeven]

On Tue, Jun 5, 2012 at 6:26 PM, Thorsten Glaser <tg@mirbsd.de> wrote:
>>On Tue, Jun 5, 2012 at 5:51 PM, Thorsten Glaser <tg@mirbsd.org> wrote:
>>> ② $EDITOR 00*
>>> ③ Add a To: line, and possibly a In-Reply-To: line referencing
>>>  the Message-ID from the introductory mail (to get them threaded),
>>>  to each file; save and exit https://www.mirbsd.org/jupp.htm ^Wthe editor
>>> ④ for a in 00*; do sendmail -oem -oi -t -f sender@mydomain.com <"$a"; done
>>
>>"git send-email --to user@domain 00*"?
>
> You conveniently cut off Item #1 above, which involves copying
> the patches to a machine that can mail out.

Rsync over ssh is the easy part ;-)

Alternatively, you probably can configure git to use ssh to call into sendmail
on a remote machine for the actual sending.

>>Does git send-email handle that correctly?
>
> There is no way to handle 8bit correctly in SMTP, that does not
> involve converting them to QP or Base64.
>
>>It will not use base64, as that's rejected by vger.kernel.org.
>
> Interesting. How are QP encoded patches handled (other than,
> what I assume, spit from Linux developers)?

Decent mail clients allow to save the mail body to a file, after
QP decoding?

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] Atari: support for NetUSBee register access to isp116x.h

[Thorsten Glaser]

Andreas Schwab dixit:

>How about using formail and mailx instead?

mksh: formail: not found

Also, mailx would mangle headers; directly invoking the MSA works,
at least for me. (And none of these can fix the 8bit issue.)

bye,
//mirabilos
-- 
13:37⎜«Natureshadow» Deep inside, I hate mirabilos. I mean, he's a good
guy. But he's always right! In every fsckin' situation, he's right. Even
with his deeply perverted taste in software and borked ambition towards
broken OSes - in the end, he's damn right about it :(! […] works in mksh

Re: [PATCH] [m68k] Atari: support for NetUSBee register access to isp116x.h

[Andreas Schwab]

Thorsten Glaser <tg@mirbsd.de> writes:

> There is no way to handle 8bit correctly in SMTP, that does not
> involve converting them to QP or Base64.

That is not true.  8BITMIME will handle that correctly.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: [PATCH] [m68k] Atari: support for NetUSBee register access to isp116x.h

[Thorsten Glaser]

Andreas Schwab dixit:

>That is not true.  8BITMIME will handle that correctly.

8BITMIME is not ubiquitous. Worse, most current MTAs do *not* handle
conversion of an 8bit mail to QP/Base64 correctly and violate the RFC.

bye,
//mirabilos
-- 
13:37⎜«Natureshadow» Deep inside, I hate mirabilos. I mean, he's a good
guy. But he's always right! In every fsckin' situation, he's right. Even
with his deeply perverted taste in software and borked ambition towards
broken OSes - in the end, he's damn right about it :(! […] works in mksh

Re: [PATCH] [m68k] Atari: support for NetUSBee register access to isp116x.h

[Andreas Schwab]

You don't *have* to use non-ASCII characters if you can't stand the
heat.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: [PATCH] [m68k] Atari: support for NetUSBee register access to isp116x.h

[Geert Uytterhoeven]

On Tue, Jun 5, 2012 at 6:51 PM, Thorsten Glaser <tg@mirbsd.de> wrote:
>>How about using formail and mailx instead?
>
> mksh: formail: not found

Wrong tool ;-)

$ dpkg -S formail
procmail: /usr/share/man/man1/formail.1.gz
procmail: /usr/bin/formail
$

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] Atari: support for NetUSBee register access to isp116x.h

[Geert Uytterhoeven]

On Tue, Jun 5, 2012 at 7:57 PM, Thorsten Glaser <tg@mirbsd.de> wrote:
> Andreas Schwab dixit:
>>That is not true.  8BITMIME will handle that correctly.
>
> 8BITMIME is not ubiquitous. Worse, most current MTAs do *not* handle
> conversion of an 8bit mail to QP/Base64 correctly and violate the RFC.

I added a Russian word in Cyrilic (UTF8) to a file in a Linux kernel
git repository.
I commited it with a commit message containing Cyrillic, too.
I used "git format-patch" to create a patch, which contains a.o. the following
headers:

| Content-Type: text/plain; charset=UTF-8
| Content-Transfer-Encoding: 8bit

I used "git send-email" to send it to me (it goes out via my provider, passes
by linux-m68k.org/nerdnet.nl, and arrives at gmail).
The patch looks fine in the gmail web interface.
I used "Show original" to open the original email incl. headers. It still has

| Content-Type: text/plain; charset=UTF-8
| Content-Transfer-Encoding: 8bit

No Base64 or QP involved.
I saved it to a file and removed the first line.
I could apply the patch fine using "git am".

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] Atari: support for NetUSBee register access to isp116x.h

[Thorsten Glaser]

Geert Uytterhoeven dixit:

>I used "git format-patch" to create a patch, which contains a.o. the following
>headers:
>
>| Content-Type: text/plain; charset=UTF-8
>| Content-Transfer-Encoding: 8bit

Yes, but that will not pass through all SMTP servers, since 8bit is
an optional extension which, for example, exim4 does not support¹ by
default, and other servers² can be configured to be strict, too (or
simply not show³ 8BITMIME support).

[…]
>No Base64 or QP involved.

Yes, but you cannot _depend_ on that.

① exim4 can be configured to either display 8BITMIME in the EHLO
  response or not, but that doesn’t change its handling – it’s
  simply oblivios of whether the message contains 8bit or not,
  just passes them on in either case (violating the SMTP RFC)

② for curiosity reasons, I run such servers

③ e.g. those that don’t do EHLO at all; I *bet* there are still
  commercial SMTP implementations like that around

bye,
//mirabilos
-- 
“It is inappropriate to require that a time represented as
 seconds since the Epoch precisely represent the number of
 seconds between the referenced time and the Epoch.”
	-- IEEE Std 1003.1b-1993 (POSIX) Section B.2.2.2