linux-stable.git/drivers/tty/serial/sh-sci.c, branch linux-3.14.y

serial: sh-sci: Remove cpufreq notifier to fix crash/deadlock

2016-05-04T21:51:25+00:00

commit ff1cab374ad98f4b9f408525ca9c08992b4ed784 upstream.

The BSP team noticed that there is spin/mutex lock issue on sh-sci when
CPUFREQ is used.  The issue is that the notifier function may call
mutex_lock() while the spinlock is held, which can lead to a BUG().
This may happen if CPUFREQ is changed while another CPU calls
clk_get_rate().

Taking the spinlock was added to the notifier function in commit
e552de2413edad1a ("sh-sci: add platform device private data"), to
protect the list of serial ports against modification during traversal.
At that time the Common Clock Framework didn't exist yet, and
clk_get_rate() just returned clk->rate without taking a mutex.
Note that since commit d535a2305facf9b4 ("serial: sh-sci: Require a
device per port mapping."), there's no longer a list of serial ports to
traverse, and taking the spinlock became superfluous.

To fix the issue, just remove the cpufreq notifier:
  1. The notifier doesn't work correctly: all it does is update stored
     clock rates; it does not update the divider in the hardware.
     The divider will only be updated when calling sci_set_termios().
     I believe this was broken back in 2004, when the old
     drivers/char/sh-sci.c driver (where the notifier did update the
     divider) was replaced by drivers/serial/sh-sci.c (where the
     notifier just updated port->uartclk).
     Cfr. full-history-linux commits 6f8deaef2e9675d9 ("[PATCH] sh: port
     sh-sci driver to the new API") and 3f73fe878dc9210a ("[PATCH]
     Remove old sh-sci driver").
  2. On modern SoCs, the sh-sci parent clock rate is no longer related
     to the CPU clock rate anyway, so using a cpufreq notifier is
     futile.

Signed-off-by: Geert Uytterhoeven 
Signed-off-by: Greg Kroah-Hartman

serial: sh-sci: Add OF support

2013-12-24T12:50:54+00:00

Extend the driver to with support for SCIx device tree bindings. A
minimal set of features is supported, additional properties can be added
later should the need to describe more device features arise.

Signed-off-by: Bastian Hecht 
Signed-off-by: Laurent Pinchart 
Acked-by: Simon Horman 
Acked-by: Greg Kroah-Hartman 
Signed-off-by: Simon Horman

serial: sh-sci: Remove platform data mapbase and irqs fields

2013-12-24T12:50:54+00:00

The fields are not used anymore by board files, remove them.

Signed-off-by: Laurent Pinchart 
Acked-by: Greg Kroah-Hartman 
Signed-off-by: Simon Horman

serial: sh-sci: Remove platform data scbrr_algo_id field

2013-12-24T12:50:54+00:00

The field isn't set by any board, remote it.

Signed-off-by: Laurent Pinchart 
Acked-by: Greg Kroah-Hartman 
Signed-off-by: Simon Horman

serial: sh-sci: Rework baud rate calculation

2013-12-24T02:17:53+00:00

Computing the baud rate register value requires knowledge of the
hardware sampling rate. This information is currently encoded in a baud
rate calculation algorithm ID passed through platform data. However, it
can be derived from the port type directly in most cases.

Compute the sampling rate internally in the driver if the baud rate
calculation algorithm ID isn't specified, and allow platforms to
override the sampling rate through platform data in special cases (this
is only required for SCIFA ports on sh7723 and sh7724, the reason needs
to be investigated).

Signed-off-by: Laurent Pinchart 
Acked-by: Greg Kroah-Hartman 
Signed-off-by: Simon Horman

serial: sh-sci: Compute overrun_bit without using baud rate algo

2013-12-24T02:17:52+00:00

The overrun bit index is a property of the hardware. It's currently
computed based on a different and unrelated hardware property, the baud
rate calculation algorithm. Compute it using hardware identification
information only.

Signed-off-by: Laurent Pinchart 
Acked-by: Greg Kroah-Hartman 
Signed-off-by: Simon Horman

serial: sh-sci: Remove unused GPIO request code

2013-12-24T02:17:50+00:00

The driver requests at initialization time GPIOs passed through platform
data. No platform makes use of this feature, remove it.

Signed-off-by: Laurent Pinchart 
Acked-by: Greg Kroah-Hartman 
Signed-off-by: Simon Horman

serial: sh-sci: Move overrun_bit and error_mask fields out of pdata

2013-12-24T02:17:48+00:00

None of the fields is ever set by board code, and both of them are set
in the driver at probe time. Move them out of struct plat_sci_port to
struct sci_port.

Signed-off-by: Laurent Pinchart 
Acked-by: Greg Kroah-Hartman 
Signed-off-by: Simon Horman

serial: sh-sci: Support resources passed through platform resources

2013-12-24T02:17:44+00:00

Memory and IRQ resources are currently passed to the driver through
platform data. Support passing them through the standard platform
resources mechanism instead. This deprecates platform data resources.

Signed-off-by: Laurent Pinchart 
Acked-by: Greg Kroah-Hartman 
Signed-off-by: Simon Horman

serial: sh-sci: Don't check IRQ in verify port operation

2013-12-24T02:17:41+00:00

The IRQ number can't be modified by the user as the port is fixed.
There's no need to check the new IRQ number as it will be ignored by the
core.

Signed-off-by: Laurent Pinchart 
Acked-by: Greg Kroah-Hartman 
Signed-off-by: Simon Horman