Patchwork USB: serial: console: fix reported terminal settings

login
register
mail settings
Submitter Johan Hovold
Date Dec. 4, 2018, 4 p.m.
Message ID <20181204160036.24625-1-johan@kernel.org>
Download mbox | patch
Permalink /patch/672141/
State New
Headers show

Comments

Johan Hovold - Dec. 4, 2018, 4 p.m.
The USB-serial console implementation has never reported the actual
terminal settings used. Despite storing the corresponding cflags in its
struct console, this was never honoured on later tty open() where the
tty termios would be left initialised to the driver defaults.

Unlike the serial console implementation, the USB-serial code calls
subdriver open() already at console setup. While calling set_termios()
before open() looks like it could work for some USB-serial drivers,
others definitely do not expect this, so modelling this after serial
core is going to be intrusive, if at all possible.

Instead, use a (renamed) tty helper to save the termios data used at
console setup, so that the tty termios reflects the actual terminal
settings after a subsequent tty open().

Note that the calls to tty_init_termios() (tty_driver_install()) and
tty_save_termios() are serialised using the disconnect mutex.

This specifically fixes a regression that was triggered by a recent
change adding software flow control to the pl2303 driver: a getty trying
to disable flow control while leaving the baud rate unchanged would now
also set the baud rate to the driver default (prior to the flow-control
change this had been a noop).

Fixes: 7041d9c3f01b ("USB: serial: pl2303: add support for tx xon/xoff flow control")
Cc: stable <stable@vger.kernel.org>	# 4.18
Reported-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Cc: Florian Zumbiehl <florz@florz.de>
Signed-off-by: Johan Hovold <johan@kernel.org>
---
 drivers/tty/tty_io.c         | 11 +++++++++--
 drivers/usb/serial/console.c |  2 +-
 include/linux/tty.h          |  1 +
 3 files changed, 11 insertions(+), 3 deletions(-)
Greg Kroah-Hartman - Dec. 4, 2018, 4:15 p.m.
On Tue, Dec 04, 2018 at 05:00:36PM +0100, Johan Hovold wrote:
> The USB-serial console implementation has never reported the actual
> terminal settings used. Despite storing the corresponding cflags in its
> struct console, this was never honoured on later tty open() where the
> tty termios would be left initialised to the driver defaults.
> 
> Unlike the serial console implementation, the USB-serial code calls
> subdriver open() already at console setup. While calling set_termios()
> before open() looks like it could work for some USB-serial drivers,
> others definitely do not expect this, so modelling this after serial
> core is going to be intrusive, if at all possible.
> 
> Instead, use a (renamed) tty helper to save the termios data used at
> console setup, so that the tty termios reflects the actual terminal
> settings after a subsequent tty open().
> 
> Note that the calls to tty_init_termios() (tty_driver_install()) and
> tty_save_termios() are serialised using the disconnect mutex.
> 
> This specifically fixes a regression that was triggered by a recent
> change adding software flow control to the pl2303 driver: a getty trying
> to disable flow control while leaving the baud rate unchanged would now
> also set the baud rate to the driver default (prior to the flow-control
> change this had been a noop).
> 
> Fixes: 7041d9c3f01b ("USB: serial: pl2303: add support for tx xon/xoff flow control")
> Cc: stable <stable@vger.kernel.org>	# 4.18
> Reported-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
> Cc: Florian Zumbiehl <florz@florz.de>
> Signed-off-by: Johan Hovold <johan@kernel.org>
> ---
>  drivers/tty/tty_io.c         | 11 +++++++++--
>  drivers/usb/serial/console.c |  2 +-
>  include/linux/tty.h          |  1 +
>  3 files changed, 11 insertions(+), 3 deletions(-)

Ah, messy :)

Want me to take this through my tty tree?

thanks,

greg k-h
Johan Hovold - Dec. 4, 2018, 4:31 p.m.
On Tue, Dec 04, 2018 at 05:15:18PM +0100, Greg Kroah-Hartman wrote:
> On Tue, Dec 04, 2018 at 05:00:36PM +0100, Johan Hovold wrote:
> > The USB-serial console implementation has never reported the actual
> > terminal settings used. Despite storing the corresponding cflags in its
> > struct console, this was never honoured on later tty open() where the
> > tty termios would be left initialised to the driver defaults.
> > 
> > Unlike the serial console implementation, the USB-serial code calls
> > subdriver open() already at console setup. While calling set_termios()
> > before open() looks like it could work for some USB-serial drivers,
> > others definitely do not expect this, so modelling this after serial
> > core is going to be intrusive, if at all possible.
> > 
> > Instead, use a (renamed) tty helper to save the termios data used at
> > console setup, so that the tty termios reflects the actual terminal
> > settings after a subsequent tty open().
> > 
> > Note that the calls to tty_init_termios() (tty_driver_install()) and
> > tty_save_termios() are serialised using the disconnect mutex.
> > 
> > This specifically fixes a regression that was triggered by a recent
> > change adding software flow control to the pl2303 driver: a getty trying
> > to disable flow control while leaving the baud rate unchanged would now
> > also set the baud rate to the driver default (prior to the flow-control
> > change this had been a noop).
> > 
> > Fixes: 7041d9c3f01b ("USB: serial: pl2303: add support for tx xon/xoff flow control")
> > Cc: stable <stable@vger.kernel.org>	# 4.18
> > Reported-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
> > Cc: Florian Zumbiehl <florz@florz.de>
> > Signed-off-by: Johan Hovold <johan@kernel.org>
> > ---
> >  drivers/tty/tty_io.c         | 11 +++++++++--
> >  drivers/usb/serial/console.c |  2 +-
> >  include/linux/tty.h          |  1 +
> >  3 files changed, 11 insertions(+), 3 deletions(-)
> 
> Ah, messy :)
> 
> Want me to take this through my tty tree?

If you prefer. I was planning on including this in a USB-serial pull
request for -rc6 since it fixes a user-reported regression, but perhaps
taking this through your tty-linus branch (which already holds a console
fix) is easier/faster.

We should wait for Jarkko to confirm that this fixes the problem he
reported first, though.

Thanks,
Johan
Jarkko Nikula - Dec. 5, 2018, 9:36 a.m.
On 12/4/18 6:31 PM, Johan Hovold wrote:
> On Tue, Dec 04, 2018 at 05:15:18PM +0100, Greg Kroah-Hartman wrote:
>> On Tue, Dec 04, 2018 at 05:00:36PM +0100, Johan Hovold wrote:
>>> The USB-serial console implementation has never reported the actual
>>> terminal settings used. Despite storing the corresponding cflags in its
>>> struct console, this was never honoured on later tty open() where the
>>> tty termios would be left initialised to the driver defaults.
>>>
>>> Unlike the serial console implementation, the USB-serial code calls
>>> subdriver open() already at console setup. While calling set_termios()
>>> before open() looks like it could work for some USB-serial drivers,
>>> others definitely do not expect this, so modelling this after serial
>>> core is going to be intrusive, if at all possible.
>>>
>>> Instead, use a (renamed) tty helper to save the termios data used at
>>> console setup, so that the tty termios reflects the actual terminal
>>> settings after a subsequent tty open().
>>>
>>> Note that the calls to tty_init_termios() (tty_driver_install()) and
>>> tty_save_termios() are serialised using the disconnect mutex.
>>>
>>> This specifically fixes a regression that was triggered by a recent
>>> change adding software flow control to the pl2303 driver: a getty trying
>>> to disable flow control while leaving the baud rate unchanged would now
>>> also set the baud rate to the driver default (prior to the flow-control
>>> change this had been a noop).
>>>
>>> Fixes: 7041d9c3f01b ("USB: serial: pl2303: add support for tx xon/xoff flow control")
>>> Cc: stable <stable@vger.kernel.org>	# 4.18
>>> Reported-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
>>> Cc: Florian Zumbiehl <florz@florz.de>
>>> Signed-off-by: Johan Hovold <johan@kernel.org>
>>> ---
>>>   drivers/tty/tty_io.c         | 11 +++++++++--
>>>   drivers/usb/serial/console.c |  2 +-
>>>   include/linux/tty.h          |  1 +
>>>   3 files changed, 11 insertions(+), 3 deletions(-)
>>
>> Ah, messy :)
>>
>> Want me to take this through my tty tree?
> 
> If you prefer. I was planning on including this in a USB-serial pull
> request for -rc6 since it fixes a user-reported regression, but perhaps
> taking this through your tty-linus branch (which already holds a console
> fix) is easier/faster.
> 
> We should wait for Jarkko to confirm that this fixes the problem he
> reported first, though.
> 
Great, this fixed the issue for both pl2303 based adapters I reported.

Tested on top of 0072a0c14d5b ("Merge tag 'media/v4.20-4' of 
git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media")

Tested-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Johan Hovold - Dec. 5, 2018, 9:50 a.m.
On Wed, Dec 05, 2018 at 11:36:52AM +0200, Jarkko Nikula wrote:
> On 12/4/18 6:31 PM, Johan Hovold wrote:
> > On Tue, Dec 04, 2018 at 05:15:18PM +0100, Greg Kroah-Hartman wrote:
> >> On Tue, Dec 04, 2018 at 05:00:36PM +0100, Johan Hovold wrote:
> >>> The USB-serial console implementation has never reported the actual
> >>> terminal settings used. Despite storing the corresponding cflags in its
> >>> struct console, this was never honoured on later tty open() where the
> >>> tty termios would be left initialised to the driver defaults.
> >>>
> >>> Unlike the serial console implementation, the USB-serial code calls
> >>> subdriver open() already at console setup. While calling set_termios()
> >>> before open() looks like it could work for some USB-serial drivers,
> >>> others definitely do not expect this, so modelling this after serial
> >>> core is going to be intrusive, if at all possible.

> >>> This specifically fixes a regression that was triggered by a recent
> >>> change adding software flow control to the pl2303 driver: a getty trying
> >>> to disable flow control while leaving the baud rate unchanged would now
> >>> also set the baud rate to the driver default (prior to the flow-control
> >>> change this had been a noop).
> >>>
> >>> Fixes: 7041d9c3f01b ("USB: serial: pl2303: add support for tx xon/xoff flow control")
> >>> Cc: stable <stable@vger.kernel.org>	# 4.18
> >>> Reported-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
> >>> Cc: Florian Zumbiehl <florz@florz.de>
> >>> Signed-off-by: Johan Hovold <johan@kernel.org>
> >>> ---
> >>>   drivers/tty/tty_io.c         | 11 +++++++++--
> >>>   drivers/usb/serial/console.c |  2 +-
> >>>   include/linux/tty.h          |  1 +
> >>>   3 files changed, 11 insertions(+), 3 deletions(-)
> >>
> >> Ah, messy :)
> >>
> >> Want me to take this through my tty tree?
> > 
> > If you prefer. I was planning on including this in a USB-serial pull
> > request for -rc6 since it fixes a user-reported regression, but perhaps
> > taking this through your tty-linus branch (which already holds a console
> > fix) is easier/faster.
> > 
> > We should wait for Jarkko to confirm that this fixes the problem he
> > reported first, though.
> > 
> Great, this fixed the issue for both pl2303 based adapters I reported.
> 
> Tested on top of 0072a0c14d5b ("Merge tag 'media/v4.20-4' of 
> git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media")
> 
> Tested-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>

Great, thanks for testing.

Greg, I noticed I left out the part about subdrivers not expecting
*write()* to be called before open() so I'll amend the commit message
when applying and include this one in a pull-request tomorrow, if that's
ok with you?

Note that the changes to tty are trivial; I'm just renaming and
exporting an existing helper.

Thanks,
Johan
Greg Kroah-Hartman - Dec. 5, 2018, 10:05 a.m.
On Wed, Dec 05, 2018 at 10:50:49AM +0100, Johan Hovold wrote:
> On Wed, Dec 05, 2018 at 11:36:52AM +0200, Jarkko Nikula wrote:
> > On 12/4/18 6:31 PM, Johan Hovold wrote:
> > > On Tue, Dec 04, 2018 at 05:15:18PM +0100, Greg Kroah-Hartman wrote:
> > >> On Tue, Dec 04, 2018 at 05:00:36PM +0100, Johan Hovold wrote:
> > >>> The USB-serial console implementation has never reported the actual
> > >>> terminal settings used. Despite storing the corresponding cflags in its
> > >>> struct console, this was never honoured on later tty open() where the
> > >>> tty termios would be left initialised to the driver defaults.
> > >>>
> > >>> Unlike the serial console implementation, the USB-serial code calls
> > >>> subdriver open() already at console setup. While calling set_termios()
> > >>> before open() looks like it could work for some USB-serial drivers,
> > >>> others definitely do not expect this, so modelling this after serial
> > >>> core is going to be intrusive, if at all possible.
> 
> > >>> This specifically fixes a regression that was triggered by a recent
> > >>> change adding software flow control to the pl2303 driver: a getty trying
> > >>> to disable flow control while leaving the baud rate unchanged would now
> > >>> also set the baud rate to the driver default (prior to the flow-control
> > >>> change this had been a noop).
> > >>>
> > >>> Fixes: 7041d9c3f01b ("USB: serial: pl2303: add support for tx xon/xoff flow control")
> > >>> Cc: stable <stable@vger.kernel.org>	# 4.18
> > >>> Reported-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
> > >>> Cc: Florian Zumbiehl <florz@florz.de>
> > >>> Signed-off-by: Johan Hovold <johan@kernel.org>
> > >>> ---
> > >>>   drivers/tty/tty_io.c         | 11 +++++++++--
> > >>>   drivers/usb/serial/console.c |  2 +-
> > >>>   include/linux/tty.h          |  1 +
> > >>>   3 files changed, 11 insertions(+), 3 deletions(-)
> > >>
> > >> Ah, messy :)
> > >>
> > >> Want me to take this through my tty tree?
> > > 
> > > If you prefer. I was planning on including this in a USB-serial pull
> > > request for -rc6 since it fixes a user-reported regression, but perhaps
> > > taking this through your tty-linus branch (which already holds a console
> > > fix) is easier/faster.
> > > 
> > > We should wait for Jarkko to confirm that this fixes the problem he
> > > reported first, though.
> > > 
> > Great, this fixed the issue for both pl2303 based adapters I reported.
> > 
> > Tested on top of 0072a0c14d5b ("Merge tag 'media/v4.20-4' of 
> > git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media")
> > 
> > Tested-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
> 
> Great, thanks for testing.
> 
> Greg, I noticed I left out the part about subdrivers not expecting
> *write()* to be called before open() so I'll amend the commit message
> when applying and include this one in a pull-request tomorrow, if that's
> ok with you?

Sure, that's fine.

> Note that the changes to tty are trivial; I'm just renaming and
> exporting an existing helper.

Yeah, I have no objections to them.  We can take them in my usb tree.

thanks,

greg k-h
Johan Hovold - Dec. 5, 2018, 10:10 a.m.
On Wed, Dec 05, 2018 at 11:05:45AM +0100, Greg Kroah-Hartman wrote:
> On Wed, Dec 05, 2018 at 10:50:49AM +0100, Johan Hovold wrote:

> > Note that the changes to tty are trivial; I'm just renaming and
> > exporting an existing helper.
> 
> Yeah, I have no objections to them.  We can take them in my usb tree.

Thanks. Can I add your ack to the patch as well?

Johan
Greg Kroah-Hartman - Dec. 5, 2018, 10:18 a.m.
On Wed, Dec 05, 2018 at 11:10:20AM +0100, Johan Hovold wrote:
> On Wed, Dec 05, 2018 at 11:05:45AM +0100, Greg Kroah-Hartman wrote:
> > On Wed, Dec 05, 2018 at 10:50:49AM +0100, Johan Hovold wrote:
> 
> > > Note that the changes to tty are trivial; I'm just renaming and
> > > exporting an existing helper.
> > 
> > Yeah, I have no objections to them.  We can take them in my usb tree.
> 
> Thanks. Can I add your ack to the patch as well?

Please do:

Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Johan Hovold - Dec. 5, 2018, 10:33 a.m.
On Wed, Dec 05, 2018 at 11:18:50AM +0100, Greg Kroah-Hartman wrote:
> On Wed, Dec 05, 2018 at 11:10:20AM +0100, Johan Hovold wrote:
> > On Wed, Dec 05, 2018 at 11:05:45AM +0100, Greg Kroah-Hartman wrote:
> > > On Wed, Dec 05, 2018 at 10:50:49AM +0100, Johan Hovold wrote:
> > 
> > > > Note that the changes to tty are trivial; I'm just renaming and
> > > > exporting an existing helper.
> > > 
> > > Yeah, I have no objections to them.  We can take them in my usb tree.
> > 
> > Thanks. Can I add your ack to the patch as well?
> 
> Please do:
> 
> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Thanks, now applied.

Johan

Patch

diff --git a/drivers/tty/tty_io.c b/drivers/tty/tty_io.c
index ee80dfbd5442..687250ec8032 100644
--- a/drivers/tty/tty_io.c
+++ b/drivers/tty/tty_io.c
@@ -1373,7 +1373,13 @@  struct tty_struct *tty_init_dev(struct tty_driver *driver, int idx)
 	return ERR_PTR(retval);
 }
 
-static void tty_free_termios(struct tty_struct *tty)
+/**
+ * tty_save_termios() - save tty termios data in driver table
+ * @tty: tty whose termios data to save
+ *
+ * Locking: Caller guarantees serialisation with tty_init_termios().
+ */
+void tty_save_termios(struct tty_struct *tty)
 {
 	struct ktermios *tp;
 	int idx = tty->index;
@@ -1392,6 +1398,7 @@  static void tty_free_termios(struct tty_struct *tty)
 	}
 	*tp = tty->termios;
 }
+EXPORT_SYMBOL_GPL(tty_save_termios);
 
 /**
  *	tty_flush_works		-	flush all works of a tty/pty pair
@@ -1491,7 +1498,7 @@  static void release_tty(struct tty_struct *tty, int idx)
 	WARN_ON(!mutex_is_locked(&tty_mutex));
 	if (tty->ops->shutdown)
 		tty->ops->shutdown(tty);
-	tty_free_termios(tty);
+	tty_save_termios(tty);
 	tty_driver_remove_tty(tty->driver, tty);
 	tty->port->itty = NULL;
 	if (tty->link)
diff --git a/drivers/usb/serial/console.c b/drivers/usb/serial/console.c
index 17940589c647..7d289302ff6c 100644
--- a/drivers/usb/serial/console.c
+++ b/drivers/usb/serial/console.c
@@ -101,7 +101,6 @@  static int usb_console_setup(struct console *co, char *options)
 		cflag |= PARENB;
 		break;
 	}
-	co->cflag = cflag;
 
 	/*
 	 * no need to check the index here: if the index is wrong, console
@@ -164,6 +163,7 @@  static int usb_console_setup(struct console *co, char *options)
 			serial->type->set_termios(tty, port, &dummy);
 
 			tty_port_tty_set(&port->port, NULL);
+			tty_save_termios(tty);
 			tty_kref_put(tty);
 		}
 		tty_port_set_initialized(&port->port, 1);
diff --git a/include/linux/tty.h b/include/linux/tty.h
index 414db2bce715..392138fe59b6 100644
--- a/include/linux/tty.h
+++ b/include/linux/tty.h
@@ -556,6 +556,7 @@  extern struct tty_struct *tty_init_dev(struct tty_driver *driver, int idx);
 extern void tty_release_struct(struct tty_struct *tty, int idx);
 extern int tty_release(struct inode *inode, struct file *filp);
 extern void tty_init_termios(struct tty_struct *tty);
+extern void tty_save_termios(struct tty_struct *tty);
 extern int tty_standard_install(struct tty_driver *driver,
 		struct tty_struct *tty);