From: Johan Hovold <johan@kernel.org>
To: Michael Hanselmann <public@hansmi.ch>
Cc: Johan Hovold <johan@kernel.org>,
linux-usb@vger.kernel.org, Michael Dreher <michael@5dot1.de>,
Jonathan Olds <jontio@i4free.co.nz>
Subject: Re: [PATCH v2 6/6] USB: serial: ch341: Simulate break condition if not supported
Date: Tue, 30 Jun 2020 13:39:06 +0200 [thread overview]
Message-ID: <20200630113906.GA3334@localhost> (raw)
In-Reply-To: <6e29707b-6774-9f25-25ac-4b4cd202a017@msgid.hansmi.ch>
On Thu, May 28, 2020 at 12:21:11AM +0200, Michael Hanselmann wrote:
> On 14.05.20 16:47, Johan Hovold wrote:
> > On Tue, Mar 31, 2020 at 11:37:22PM +0000, Michael Hanselmann wrote:
> >> static int ch341_set_baudrate_lcr(struct usb_device *dev,
> >> - struct ch341_private *priv, u8 lcr)
> >> + struct ch341_private *priv,
> >> + unsigned baud_rate, u8 lcr)
> >
> > Use speed_t.
>
> Done. ch341_private.baud_rate and ch341_set_termios also use unsigned
> though. Should those also be changed to speed_t?
I'd just leave it for now.
> >> + r = ch341_set_baudrate_lcr(port->serial->dev, priv,
> >> + CH341_MIN_BPS,
> >> + CH341_LCR_ENABLE_RX | CH341_LCR_ENABLE_TX |
> >
> > Hmm. This would corrupt incoming data as well during the break.
>
> Yes, there's no way to avoid that. In my opinion being able to send a
> break signal for a serial console interrupt (SysRq) outweighs the
> corruption. Updated the comment on ch341_simulate_break to mention
> the caveat.
>
> >> + dev_dbg(&port->dev, "%s - Leave break state requested\n", __func__);
> >> +
> >> + if (time_before(jiffies, priv->break_end)) {
> >> + /* Wait until NUL byte is written */
> >> + delay = min_t(unsigned long, HZ, priv->break_end - jiffies);
> >
> > Looks like this can underflow if you're preempted after the check.
>
> Moved the subtraction before the min_t macro to only evaluate it once.
That's not sufficient; the problem is that jiffies may be updated after
time_before().
> >> + if (priv->quirks & CH341_QUIRK_SIMULATE_BREAK) {
> >> + dev_warn_once(&port->dev,
> >> + "%s - hardware doesn't support real break"
> >> + " condition, simulating instead\n",
> >> + __func__);
> >
> > Don't break the string, and drop the __func__.
>
> Done, also for the other error messages you pointed out.
>
> Michael
>
> ---
> From 94fec46e814276491c9a027c5d3912b68deb9c55 Mon Sep 17 00:00:00 2001
> From: Michael Hanselmann <public@hansmi.ch>
> Date: Thu, 5 Mar 2020 01:50:35 +0100
> Subject: [PATCH 2/2] USB: serial: ch341: Simulate break condition if not
> supported
>
> A subset of all CH341 devices don't support a real break condition. This
> fact is already used in the "ch341_detect_quirks" function. With this
> change a quirk is implemented to simulate a break condition by
> temporarily lowering the baud rate and sending a NUL byte.
>
> The primary drawbacks of this approach are that the duration of the
> break can't be controlled by userland and that data incoming during
> a simulated break is corrupted.
>
> The "TTY_DRIVER_HARDWARE_BREAK" serial driver flag was investigated as
> an alternative. It's a driver-wide flag and would've required
> significant changes to the serial and USB-serial driver frameworks to
> expose it for individual USB-serial adapters.
>
> Tested by sending a break condition and watching the TX pin using an
> oscilloscope.
>
> Signed-off-by: Michael Hanselmann <public@hansmi.ch>
> ---
> drivers/usb/serial/ch341.c | 105 ++++++++++++++++++++++++++++++++++---
> 1 file changed, 97 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/usb/serial/ch341.c b/drivers/usb/serial/ch341.c
> index 202cfa4ef6c7..1e63310cfd9c 100644
> --- a/drivers/usb/serial/ch341.c
> +++ b/drivers/usb/serial/ch341.c
> @@ -78,6 +78,7 @@
> #define CH341_LCR_CS5 0x00
>
> #define CH341_QUIRK_LIMITED_PRESCALER BIT(0)
> +#define CH341_QUIRK_SIMULATE_BREAK BIT(1)
>
> static const struct usb_device_id id_table[] = {
> { USB_DEVICE(0x4348, 0x5523) },
> @@ -94,6 +95,7 @@ struct ch341_private {
> u8 msr;
> u8 lcr;
> unsigned long quirks;
> + unsigned long break_end;
> };
>
> static void ch341_set_termios(struct tty_struct *tty,
> @@ -168,10 +170,9 @@ static const speed_t ch341_min_rates[] = {
> * 2 <= div <= 256 if fact = 0, or
> * 9 <= div <= 256 if fact = 1
> */
> -static int ch341_get_divisor(struct ch341_private *priv)
> +static int ch341_get_divisor(struct ch341_private *priv, speed_t speed)
> {
> unsigned int fact, div, clk_div;
> - speed_t speed = priv->baud_rate;
> bool force_fact0 = false;
> int ps;
>
> @@ -234,15 +235,16 @@ static int ch341_get_divisor(struct ch341_private *priv)
> }
>
> static int ch341_set_baudrate_lcr(struct usb_device *dev,
> - struct ch341_private *priv, u8 lcr)
> + struct ch341_private *priv,
> + speed_t baud_rate, u8 lcr)
> {
> int val;
> int r;
>
> - if (!priv->baud_rate)
> + if (!baud_rate)
> return -EINVAL;
>
> - val = ch341_get_divisor(priv);
> + val = ch341_get_divisor(priv, baud_rate);
> if (val < 0)
> return -EINVAL;
>
> @@ -322,7 +324,7 @@ static int ch341_configure(struct usb_device *dev, struct ch341_private *priv)
> if (r < 0)
> goto out;
>
> - r = ch341_set_baudrate_lcr(dev, priv, priv->lcr);
> + r = ch341_set_baudrate_lcr(dev, priv, priv->baud_rate, priv->lcr);
> if (r < 0)
> goto out;
>
> @@ -356,7 +358,7 @@ static int ch341_detect_quirks(struct usb_serial_port *port)
> CH341_REG_BREAK, 0, buffer, size, DEFAULT_TIMEOUT);
> if (r == -EPIPE) {
> dev_dbg(&port->dev, "break control not supported\n");
> - quirks = CH341_QUIRK_LIMITED_PRESCALER;
> + quirks = CH341_QUIRK_LIMITED_PRESCALER | CH341_QUIRK_SIMULATE_BREAK;
> r = 0;
> goto out;
> }
> @@ -537,7 +539,8 @@ static void ch341_set_termios(struct tty_struct *tty,
> if (baud_rate) {
> priv->baud_rate = baud_rate;
>
> - r = ch341_set_baudrate_lcr(port->serial->dev, priv, lcr);
> + r = ch341_set_baudrate_lcr(port->serial->dev, priv,
> + priv->baud_rate, lcr);
> if (r < 0 && old_termios) {
> priv->baud_rate = tty_termios_baud_rate(old_termios);
> tty_termios_copy_hw(&tty->termios, old_termios);
> @@ -556,15 +559,101 @@ static void ch341_set_termios(struct tty_struct *tty,
> ch341_set_handshake(port->serial->dev, priv->mcr);
> }
>
> +/*
> + * A subset of all CH34x devices don't support a real break condition and
> + * reading CH341_REG_BREAK fails (see also ch341_detect_quirks). This function
> + * simulates a break condition by lowering the baud rate to the minimum
> + * supported by the hardware upon enabling the break condition and sending
> + * a NUL byte.
> + *
> + * Incoming data is corrupted while the break condition is being simulated.
> + *
> + * Normally the duration of the break condition can be controlled individually
> + * by userspace using TIOCSBRK and TIOCCBRK or by passing an argument to
> + * TCSBRKP. Due to how the simulation is implemented the duration can't be
> + * controlled. The duration is always about (1s / 46bd * 9bit) = 196ms.
> + */
> +static void ch341_simulate_break(struct tty_struct *tty, int break_state)
> +{
> + struct usb_serial_port *port = tty->driver_data;
> + struct ch341_private *priv = usb_get_serial_port_data(port);
> + unsigned long delay;
> + int r;
> +
> + if (break_state != 0) {
> + dev_dbg(&port->dev, "enter break state requested\n");
> +
> + r = ch341_set_baudrate_lcr(port->serial->dev, priv,
> + CH341_MIN_BPS,
> + CH341_LCR_ENABLE_RX | CH341_LCR_ENABLE_TX |
> + CH341_LCR_CS8);
Continuation lines should be indented at least two tabs further.
And let's just merge the last two lines.
> + if (r < 0) {
> + dev_err(&port->dev,
> + "failed to change baud rate to %u: %d\n",
> + CH341_MIN_BPS, r);
> + goto restore;
> + }
> +
> + r = tty_put_char(tty, '\0');
> + if (r < 0) {
> + dev_err(&port->dev,
> + "failed to write NUL byte for simulated break condition: %d\n",
> + r);
> + goto restore;
> + }
> +
> + /*
> + * Compute how long transmission will take and add a bit of
> + * safety margin.
> + */
Where's that margin?
> + priv->break_end = jiffies + (10 * HZ / CH341_MIN_BPS);
> +
> + return;
> + }
> +
> + dev_dbg(&port->dev, "leave break state requested\n");
> +
> + if (time_before(jiffies, priv->break_end)) {
> + /*
> + * Wait until NUL byte is written and limit delay to one second
> + * at most.
> + */
So you could still be preempted here so that delay would wrap. Just
store jiffies in a "now" temporary before the conditional?
> + delay = priv->break_end - jiffies;
> + delay = min_t(unsigned long, HZ, delay);
And then you can skip the min_t() here.
> +
> + dev_dbg(&port->dev,
> + "wait %d ms while transmitting NUL byte at %u baud\n",
> + jiffies_to_msecs(delay), CH341_MIN_BPS);
> + schedule_timeout_interruptible(delay);
> + }
> +
> +restore:
> + /* Restore original baud rate */
> + r = ch341_set_baudrate_lcr(port->serial->dev, priv, priv->baud_rate,
> + priv->lcr);
> + if (r < 0)
> + dev_err(&port->dev,
> + "restoring original baud rate of %u failed: %d\n",
> + priv->baud_rate, r);
> +}
> +
> static void ch341_break_ctl(struct tty_struct *tty, int break_state)
> {
> const uint16_t ch341_break_reg =
> ((uint16_t) CH341_REG_LCR << 8) | CH341_REG_BREAK;
> struct usb_serial_port *port = tty->driver_data;
> + struct ch341_private *priv = usb_get_serial_port_data(port);
> int r;
> uint16_t reg_contents;
> uint8_t *break_reg;
>
> + if (priv->quirks & CH341_QUIRK_SIMULATE_BREAK) {
> + dev_warn_once(&port->dev,
> + "hardware doesn't support real break condition, simulating instead\n");
This should probably be moved to probe and quirk_detect() and be a
dev_info (e.g. consider having two of these devices in a system).
> + ch341_simulate_break(tty, break_state);
> + return;
> + }
> +
> break_reg = kmalloc(2, GFP_KERNEL);
> if (!break_reg)
> return;
Care to respin as a v4?
Johan
next prev parent reply other threads:[~2020-06-30 11:39 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-03-31 23:37 [PATCH v2 0/6] USB: serial: ch341: Add support for limited chips (was: Add support for HL340 devices) Michael Hanselmann
2020-03-31 23:37 ` [PATCH v2 1/6] USB: serial: ch341: Reduce special cases in clock calculation Michael Hanselmann
2020-03-31 23:37 ` [PATCH v2 2/6] USB: serial: ch341: Add basis for quirk detection Michael Hanselmann
2020-05-14 14:09 ` Johan Hovold
2020-03-31 23:37 ` [PATCH v2 3/6] USB: serial: ch341: Limit prescaler on quirky chips Michael Hanselmann
2020-05-14 14:17 ` Johan Hovold
2020-05-27 13:16 ` Johan Hovold
2020-05-27 15:41 ` Michael Hanselmann
2020-05-29 7:15 ` Johan Hovold
2020-03-31 23:37 ` [PATCH v2 4/6] USB: serial: ch341: Name prescaler, divisor registers Michael Hanselmann
2020-05-14 14:24 ` Johan Hovold
2020-05-27 20:59 ` Michael Hanselmann
2020-06-29 9:51 ` Johan Hovold
2020-03-31 23:37 ` [PATCH v2 5/6] USB: serial: ch341: Compute minimum baud rate Michael Hanselmann
2020-05-27 22:19 ` Michael Hanselmann
2020-06-30 9:57 ` Johan Hovold
2020-03-31 23:37 ` [PATCH v2 6/6] USB: serial: ch341: Simulate break condition if not supported Michael Hanselmann
2020-05-14 14:47 ` Johan Hovold
2020-05-27 22:21 ` Michael Hanselmann
2020-06-30 11:39 ` Johan Hovold [this message]
2020-07-04 18:25 ` Michael Hanselmann
2020-07-06 9:31 ` Johan Hovold
2020-05-14 14:02 ` [PATCH v2 0/6] USB: serial: ch341: Add support for limited chips (was: Add support for HL340 devices) Johan Hovold
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200630113906.GA3334@localhost \
--to=johan@kernel.org \
--cc=jontio@i4free.co.nz \
--cc=linux-usb@vger.kernel.org \
--cc=michael@5dot1.de \
--cc=public@hansmi.ch \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).