| Message ID | 20220111132847.218193-1-asheplyakov@basealt.ru |
|---|---|
| State | New |
| Headers | show |
| Series | serial: 8250_dw: verify clock rate in dw8250_set_termios | expand |
Hello, Andy, On Tue, Jan 11, 2022 at 08:30:28PM +0200, Andy Shevchenko wrote: > On Tue, Jan 11, 2022 at 05:28:47PM +0400, asheplyakov@basealt.ru wrote: > > From: Alexey Sheplyakov <asheplyakov@basealt.ru> > > Thanks for your patch, my comments below. Thanks for the comments. > > Refuse to change the clock rate if clk_round_rate() returns > > a rate which is way too off (i.e. by more than 1/16 from the one > > necessary for a given baud rate). In particular this happens if > > the requested rate is below the minimum supported by the clock. > > > > Fixes the UART console on BE-M1000 SoC. Without this patch the > > console gets garbled immediately after loading the driver. > > dw8250_set_termios tries to configure the baud rate (115200), > > and calls clk_round_rate to figure out the supported rate closest > > to 1843200 Hz (which is 115200 * 16). However the (SoC-specific) > > clock driver returns 4705882 Hz. This frequency is way too off, > > hence after setting it the console gets garbled. > > So, the root cause is to understand _why_ the clock provider is uncapable > to fulfil the request. Any investigation has been conducted? Yes. On BE-M1000 SoC Linux has no direct control over (most) clocks. The registers of CMU (clock management unit) are accessible only from the secure world, therefore clocks are managed by the firmware (ARM-TF). Linux' driver is a shim which calls into the firmware. And that 4705882 Hz is exactly what is returned by firmware. > > Signed-off-by: Alexey Sheplyakov <asheplyakov@basealt.ru> > > Signed-off-by: Vadim V. Vlasov <vadim.vlasov@elpitech.ru> > > > > > Should be no blank lines in the tag block, but... > > > Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com> > > Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> > > Cc: Serge Semin <Sergey.Semin@baikalelectronics.ru> > > ...I recommend to use --cc parameter to git send-email instead of polluting > commit message. Got it. > ... > > > - if (rate > 0) { > > + if (rate > 0 && rate >= baud * 15 && rate <= baud * 17) { > > It doesn't fell like a correct fix. What is the correct way to check if the rate returned by clk_round_rate makes sense for the (new) baud rate? See, clk_round_rate is supposed to give a supported rate which is closest to the requested one. Usually it appears to be "close enough" to the requested rate (or an error), however (as far as I understand) there is no a formal requirement (and it's up to the driver to decide how close is "close enough"). Or should we just hope that the clock provider does the right thing? Best regards, Alexey
diff --git a/drivers/tty/serial/8250/8250_dw.c b/drivers/tty/serial/8250/8250_dw.c index 1769808031c5..ec7e8169c983 100644 --- a/drivers/tty/serial/8250/8250_dw.c +++ b/drivers/tty/serial/8250/8250_dw.c @@ -329,14 +329,15 @@ dw8250_do_pm(struct uart_port *port, unsigned int state, unsigned int old) static void dw8250_set_termios(struct uart_port *p, struct ktermios *termios, struct ktermios *old) { - unsigned long newrate = tty_termios_baud_rate(termios) * 16; + unsigned long baud = tty_termios_baud_rate(termios); + unsigned long newrate = baud * 16; struct dw8250_data *d = to_dw8250_data(p->private_data); long rate; int ret; clk_disable_unprepare(d->clk); rate = clk_round_rate(d->clk, newrate); - if (rate > 0) { + if (rate > 0 && rate >= baud * 15 && rate <= baud * 17) { /* * Note that any clock-notifer worker will block in * serial8250_update_uartclk() until we are done.