This avoids two race conditions from the UART shutdown sequence both
leading to 'Machine check error in AXI2APB' and kernel oops.
One was that the clock was disabled before the DMA was terminated making
it possible for the DMA callbacks to be called after the clock was
disabled. These callbacks could write to the UART registers causing
timeout.
The second was that the clock was disabled before the UART was
completely flagged as closed. This is done after the shutdown is called
and a new write could be started after the clock was disabled.
tegra_uart_start_pio_tx could be called causing timeout.
Given that the baud rate is reset at the end of shutdown sequence, this
fix is to examine the baud rate to avoid register access from both race
conditions.
Besides, terminate the DMA before disabling the clock.
serial: sprd: correct the wrong sequence of arguments
The sequence of arguments which was passed to handle_lsr_errors() didn't
match the parameters defined in that function, &lsr was passed to flag
and &flag was passed to lsr, this patch fixed that.
serial: max310x: turn off transmitter before activating AutoCTS or auto transmitter flow control
As documented in the data-sheet, the transmitter must be disabled before
activating AutoCTS or auto transmitter flow control. Accordingly, the
transmitter must be enabled after AutoCTS or auto transmitter flow
control gets deactivated.
serial: max310x: Properly set flags in AutoCTS mode
Commit 391f93f2ec9f ("serial: core: Rework hw-assisted flow control
support") has changed the way the AutoCTS mode is handled.
According to that change, serial drivers which enable H/W AutoCTS mode must
set UPSTAT_AUTORTS, UPSTAT_AUTOCTS and UPSTAT_AUTOXOFF to prevent the
serial core from inadvertently disabling RX or TX. This patch adds proper
handling of UPSTAT_AUTORTS, UPSTAT_AUTOCTS and UPSTAT_AUTOXOFF flags.
Chunyan Zhang [Mon, 26 Aug 2019 07:29:29 +0000 (15:29 +0800)]
serial: sprd: keep console alive even if missing the 'enable' clock
The sprd serial console can work with only 26M fixed clock,
but the probe() is returning fail if the clock "enable" is not
configured in device tree.
This patch will fix the problem to let the uart device which is
used for console can be initialized even missing "enable" clock
configured in devicetree. We should make sure the debug function
as available as we can.
Chunyan Zhang [Mon, 26 Aug 2019 07:29:28 +0000 (15:29 +0800)]
serial: sprd: add console_initcall in sprd's uart driver
Use console_initcall to save the console index we selected on the
command line to sprd_console before probe finished. Thus we can
make different processes to the uart devices during initialization
according to whether it is used for console.
Richard Genoud [Mon, 26 Aug 2019 07:17:52 +0000 (09:17 +0200)]
tty/serial: atmel: remove unneeded atmel_get_lines_status function
Since commit 18dfef9c7f87 ("serial: atmel: convert to irq handling
provided mctrl-gpio"), the GPIOs interrupts are handled by
mctrl_gpio_irq_handle().
So, atmel_get_lines_status() can be completely killed and replaced by :
atmel_uart_readl(port, ATMEL_US_CSR);
Sergey Organov [Wed, 28 Aug 2019 18:37:54 +0000 (21:37 +0300)]
serial: imx: fix data breakage on termios change
imx_set_termios(): avoid writing baud rate divider registers when the
values to be written are the same as current. Any writing seems to
restart transmission/receiving logic in the hardware, that leads to
data breakage even when rate doesn't in fact change. E.g., user
switches RTS/CTS handshake and suddenly gets broken bytes.
Sergey Organov [Wed, 28 Aug 2019 18:37:52 +0000 (21:37 +0300)]
serial: imx: do not stop Rx/Tx on termios change
imx_set_termios(): stopping receiver and transmitter does harm when
something that doesn't touch transmission format/rate changes, such as
RTS/CTS handshake.
OTOH, it does no good on baud rate or format change, as
synchronization on upper-level protocols is still required to do it
right.
Jan Kundrát [Wed, 28 Aug 2019 17:56:26 +0000 (19:56 +0200)]
tty: max310x: fix off-by-one buffer access when storing overrun
A recent change split the insertion loop into two parts. The first part
accessed bytes 0, 1, ... (rxlen - 2), and the second part by mistake
took offset `rxlen` instead of the correct `rxlen - 1`. So one byte was
not stored, and the final access wrote past the end of the rx_buf.
Ralf Ramsauer [Mon, 12 Aug 2019 11:21:52 +0000 (13:21 +0200)]
serial: 8250_pci: Implement MSI(-X) support
There may be setups, where legacy interrupts are not available. This is
the caese, e.g., when Linux runs as guest (aka. non-root cell) of the
partitioning hypervisor Jailhouse. There, only MSI(-X) interrupts are
available for guests.
But the 8250_pci driver currently only supports legacy ints. So let's
enable MSI(-X) interrupts.
Nevertheless, this needs to handled with care: while many 8250 devices
actually claim to support MSI(-X) interrupts it should not be enabled be
default. I had at least one device in my hands with broken MSI
implementation.
So better introduce a whitelist with devices that are known to support
MSI(-X) interrupts. I tested all devices mentioned in the patch.
Fintek F81504A/508A/512A is PCIE to 4/8/12 UARTs device. It's support
IO/MMIO/PCIE conf to access all functions. The old F81504/508/512 is
only support IO.
Andy Shevchenko [Wed, 14 Aug 2019 14:07:59 +0000 (17:07 +0300)]
serial: mctrl_gpio: Use gpiod flags directly
Description of the modem line control GPIOs contain a boolean type to set
direction of the line. Since GPIO library provides an enumerator type of flags,
we may utilize it and allow a bit more flexibility on the choice of the type of
the line parameters. It also removes an additional layer of value conversion.
Mao Wenan [Tue, 20 Aug 2019 12:40:15 +0000 (20:40 +0800)]
tty: serial: add dependence for CONFIG_SERIAL_FSL_LINFLEXUART
When CONFIG_SERIAL_FSL_LINFLEXUART=y and CONFIG_PRINTK is not set,
one compilation error is found as below:
drivers/tty/serial/fsl_linflexuart.c: In function linflex_earlycon_putchar:
drivers/tty/serial/fsl_linflexuart.c:608:31: error: CONFIG_LOG_BUF_SHIFT undeclared
(first use in this function); did you mean CONFIG_ISA_BUS_API?
if (earlycon_buf.len >= 1 << CONFIG_LOG_BUF_SHIFT)
^~~~~~~~~~~~~~~~~~~~
CONFIG_ISA_BUS_API
This because CONFIG_LOG_BUF_SHIFT is depended on CONFIG_PRINTK, fix this
by adding dependence for CONFIG_SERIAL_FSL_LINFLEXUART.
serial: mxs-auart: Don't check for mctrl_gpio_to_gpiod() returning error
Since commit 1d267ea6539f2663 ("serial: mctrl-gpio: simplify init
routine"), mctrl_gpio_init() returns failure if the assignment to any
member of the gpio array results in an error pointer.
Since commit c359522194593815 ("serial: mctrl_gpio: Avoid probe failures
in case of missing gpiolib"), mctrl_gpio_to_gpiod() returns NULL in the
!CONFIG_GPIOLIB case.
Hence there is no longer a need to check for mctrl_gpio_to_gpiod()
returning an error value. A simple NULL check is sufficient.
This follows the spirit of commit 445df7ff3fd1a0a9 ("serial: mctrl-gpio:
drop usages of IS_ERR_OR_NULL") in the mctrl-gpio core.
serial: sh-sci: Don't check for mctrl_gpio_to_gpiod() returning error
Since commit 1d267ea6539f2663 ("serial: mctrl-gpio: simplify init
routine"), mctrl_gpio_init() returns failure if the assignment to any
member of the gpio array results in an error pointer.
Since commit c359522194593815 ("serial: mctrl_gpio: Avoid probe failures
in case of missing gpiolib"), mctrl_gpio_to_gpiod() returns NULL in the
!CONFIG_GPIOLIB case.
Hence there is no longer a need to check for mctrl_gpio_to_gpiod()
returning an error value. A simple NULL check is sufficient.
This follows the spirit of commit 445df7ff3fd1a0a9 ("serial: mctrl-gpio:
drop usages of IS_ERR_OR_NULL") in the mctrl-gpio core.
tty/serial: atmel: reschedule TX after RX was started
When half-duplex RS485 communication is used, after RX is started, TX
tasklet still needs to be scheduled tasklet. This avoids console freezing
when more data is to be transmitted, if the serial communication is not
closed.
Fixes: 69646d7a3689 ("tty/serial: atmel: RS485 HD w/DMA: enable RX after TX is stopped") Signed-off-by: Razvan Stefanescu <razvan.stefanescu@microchip.com> Cc: stable <stable@vger.kernel.org> Link: https://lore.kernel.org/r/20190813074025.16218-1-razvan.stefanescu@microchip.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Introduce support for LINFlex driver, based on:
- the version of Freescale LPUART driver after commit b3e3bf2ef2c7 ("Merge
4.0-rc7 into tty-next");
- commit abf1e0a98083 ("tty: serial: fsl_lpuart: lock port on console
write").
In this basic version, the driver can be tested using initramfs and relies
on the clocks and pin muxing set up by U-Boot.
Remarks concerning the earlycon support:
- LinFlexD does not allow character transmissions in the INIT mode (see
section 47.4.2.1 in the reference manual[1]). Therefore, a mutual
exclusion between the first linflex_setup_watermark/linflex_set_termios
executions and linflex_earlycon_putchar was employed and the characters
normally sent to earlycon during initialization are kept in a buffer and
sent afterwards.
- Empirically, character transmission is also forbidden within the last 1-2
ms before entering the INIT mode, so we use an explicit timeout
(PREINIT_DELAY) between linflex_earlycon_putchar and the first call to
linflex_setup_watermark.
- U-Boot currently uses the UART FIFO mode, while this driver makes the
transition to the buffer mode. Therefore, the earlycon putchar function
matches the U-Boot behavior before initializations and the Linux behavior
after.
Rahul Tanwar [Thu, 8 Aug 2019 10:02:07 +0000 (18:02 +0800)]
serial: lantiq: Make IRQ & ISR assignment dynamic
This driver/IP is reused across multiple SoCs. Older SoCs supported three
separate IRQs for tx, rx & err interrupts. Newer Lightning Mountain SoC
supports single IRQ for all of tx/rx/err interrupts. This patch modifies
the driver design to support dynamic assignment of IRQ resources & ISRs
based on devicetree node compatible entries.
Rahul Tanwar [Thu, 8 Aug 2019 10:02:06 +0000 (18:02 +0800)]
serial: lantiq: Use proper DT compatible string
Use explicit string instead of a macro for devicetree compatible string.
This series of patches is to add support for multiple SoCs which reuse the same
serial controller IP. The following patches will add another compatible string
to support new Lightning Mountain(LGM) SoC. So it makes sense to have the
compatible strings explicitly mentioned instead of a fixed macro.
Andy Shevchenko [Tue, 6 Aug 2019 09:43:22 +0000 (12:43 +0300)]
serial: 8250_lpss: Enable HS UART on Elkhart Lake
Intel Elkhart Lake may use High Speed UART from OSE IP block.
This is different to what we have in main LPSS, though compatible
with older version of it, which is handled by this driver.
Enable OSE HS UART on Intel Elkhart Lake by adding PCI IDs.
Andy Shevchenko [Tue, 6 Aug 2019 09:43:19 +0000 (12:43 +0300)]
serial: 8250_lpss: add fractional divisor support
For Synopsys DesignWare 8250 uart which version >= 4.00a, there's a
valid divisor latch fraction register.
Now the preparation is done, it's easy to add the feature support.
This patch firstly tries to get the fractional divisor width during
probe, then setups specific get_divisor() and set_divisor() hook.
Among other changes the FIFO size is now retrieved from the hardware.
Andy Shevchenko [Tue, 6 Aug 2019 09:43:16 +0000 (12:43 +0300)]
serial: 8250_dw: split Synopsys DesignWare 8250 common functions
We would like to use same functions in the couple of drivers for
Synopsys DesignWare 8250 UART. Split them from 8250_dw into new brand
library module which users will select explicitly.
Andy Shevchenko [Tue, 6 Aug 2019 09:43:14 +0000 (12:43 +0300)]
serial: 8250_dw: Use a unified new dev variable in remove
The commit 2cb78eab2376 ("serial: 8250_dw: Use a unified new dev variable in
probe") introduced a local dev variable in ->probe(). Do the same in ->remove()
in order to prepare for sequential patches.
Rahul Tanwar [Wed, 7 Aug 2019 09:21:31 +0000 (17:21 +0800)]
serial: lantiq: Add SMP support
The existing driver can only support single core SoC. But new multicore
platforms which reuse the same driver/IP need SMP support. This patch adds
multicore support in the driver.
Andrey Smirnov [Mon, 5 Aug 2019 18:57:01 +0000 (11:57 -0700)]
tty: serial: fsl_lpuart: Ignore TX/RX interrupts if DMA is enabled
In a mixed DMA/IRQ use-case (e.g.: DMA for TX, IRQ for RX), interrupt
handler might try to handle Rx/Tx condition it shouldn't. Change the
code to only handle TX/RX event if corresponding path isn't being
handled by DMA.
Andrey Smirnov [Mon, 5 Aug 2019 18:57:00 +0000 (11:57 -0700)]
tty: serial: fsl_lpuart: Don't enable TIE in .startup() or .resume()
Enabling TIE in .startup() callback causes the driver to start (or at
least try) to transmit data before .start_tx() is called. Which, while
harmless (since TIE handler will immediately disable it), is a no-op
and shouldn't really happen. Drop UARTCR2_TIE from list of bits set in
lpuart_startup().
This change will also not enable TIE in .resume(), but it seems that,
similart to .startup(), transmit interrupt shouldn't be enabled there
either.
Most users of lpuart*_setup_watermark() enable identical set of flags
right after the call, so combine those two action into a subroutine
and make use of it.
Code configure DMA TX path in lpuart_startup(), lpuart32_startup() and
lpuart_resume() is doing exactly the same thing, so move it into a
standalone subroutine.
serial: mctrl_gpio: Avoid probe failures in case of missing gpiolib
If CONFIG_GPIOLIB is not enabled, mctrl_gpio_init() and
mctrl_gpio_init_noauto() will currently return an error pointer with
-ENOSYS. As the mctrl GPIOs are usually optional, drivers need to
check for this condition to allow continue probing.
To avoid the need for this check in each driver, we return NULL
instead, as all the mctrl_gpio_*() functions are skipped anyway.
We also adapt mctrl_gpio_to_gpiod() to be in line with this change.
Aaron Sierra [Thu, 1 Aug 2019 18:59:56 +0000 (13:59 -0500)]
serial: 8250_exar: Absorb remaining 8250_port INT0 support
Move INT0 clearing out of common, per-port serial8250_do_startup()
into PCI device probe/resume.
As described in commit 2c0ac5b48a35 ("serial: exar: Fix stuck MSIs"),
the purpose of clearing INT0 is to prevent the PCI interrupt line from
becoming stuck asserted, "which is fatal with edge-triggered MSIs".
Like the clearing via interrupt handler that moved from common code in
commit c7e1b4059075 ("tty: serial: exar: Relocate sleep wake-up
handling"), this clearing at startup can be better handled at the PCI
device level.
Cc: Jan Kiszka <jan.kiszka@siemens.com> Cc: Sudip Mukherjee <sudip.mukherjee@codethink.co.uk> Signed-off-by: Aaron Sierra <asierra@xes-inc.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Link: https://lore.kernel.org/r/20190801185956.3222-1-asierra@xes-inc.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
serial: sh-sci: Use DEVICE_ATTR_RW() for rx_fifo_trigger
While commit b6b996b6cdeecf7e ("treewide: Use DEVICE_ATTR_RW") converted
the rx_fifo_timeout attribute, it forgot to convert rx_fifo_trigger due
to a slightly different function naming.
Vivek Gautam [Thu, 1 Aug 2019 12:11:53 +0000 (17:41 +0530)]
tty: serial: qcom_geni_serial: Update the oversampling rate
For QUP IP versions 2.5 and above the oversampling rate is halved
from 32 to 16. Update this rate after reading hardware version
register, so that the clock divider value is correctly set to
achieve required baud rate.
When closing and shutting down the exar serial port, if the chip
has not finished sending all of the data in its buffer, the
remaining bytes will be lost. Hold off on the shutdown until the
bytes have all been sent.
Andy Shevchenko [Wed, 31 Jul 2019 17:05:58 +0000 (20:05 +0300)]
serial: 8250_exar: Move custom divisor support out from 8250_port
There are Exar custom divisor support in 8250_port which belongs to
8250_exar module. Move it out to the correct module and do not contaminate
generic code with it.
Cc: Aaron Sierra <asierra@xes-inc.com> Cc: Jan Kiszka <jan.kiszka@siemens.com> Cc: Sudip Mukherjee <sudip.mukherjee@codethink.co.uk> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Link: https://lore.kernel.org/r/20190731170558.52897-3-andriy.shevchenko@linux.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Andy Shevchenko [Wed, 31 Jul 2019 17:05:57 +0000 (20:05 +0300)]
serial: 8250_exar: Extract PM routine from 8250_port
There are Exar quirks in 8250_port which belong to 8250_exar module.
Extract PM routine to the correct module and do not contaminate generic code
with it.
Cc: Aaron Sierra <asierra@xes-inc.com> Cc: Jan Kiszka <jan.kiszka@siemens.com> Cc: Sudip Mukherjee <sudip.mukherjee@codethink.co.uk> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Link: https://lore.kernel.org/r/20190731170558.52897-2-andriy.shevchenko@linux.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
serial: sh-sci: use driver core functions, not sysfs ones.
This is a driver, do not call "raw" sysfs functions, instead call driver
core ones. Specifically convert the use of sysfs_create_file() and
sysfs_remove_file() to use device_create_file() and device_remove_file()
Stephen Rothwell [Wed, 31 Jul 2019 06:05:57 +0000 (16:05 +1000)]
xilinx_uartps.c: suppress "may be used uninitialised" warning
A powerpc allyesconfig build produces this warning:
In file included from include/linux/radix-tree.h:16,
from include/linux/idr.h:15,
from include/linux/kernfs.h:13,
from include/linux/sysfs.h:16,
from include/linux/kobject.h:20,
from include/linux/device.h:16,
from include/linux/platform_device.h:13,
from drivers/tty/serial/xilinx_uartps.c:16:
drivers/tty/serial/xilinx_uartps.c: In function 'cdns_uart_console_write':
include/linux/spinlock.h:288:3: warning: 'flags' may be used uninitialized in this function [-Wmaybe-uninitialized]
_raw_spin_unlock_irqrestore(lock, flags); \
^~~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/tty/serial/xilinx_uartps.c:1197:16: note: 'flags' was declared here
unsigned long flags;
^~~~~
It looks like gcc just can't track the relationship between "locked"
and "flags", and it is obvious that "flags" won't be used when "locked"
is zero, so the simplest thing is to initialise flags.
Cc: Jiri Slaby <jslaby@suse.com> Cc: Michal Simek <michal.simek@xilinx.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Link: https://lore.kernel.org/r/20190731160557.6a09c3e1@canb.auug.org.au Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Stephen Boyd [Tue, 30 Jul 2019 18:15:44 +0000 (11:15 -0700)]
tty: Remove dev_err() usage after platform_get_irq()
We don't need dev_err() messages when platform_get_irq() fails now that
platform_get_irq() prints an error message itself when something goes
wrong. Let's remove these prints with a simple semantic patch.
tty: serial: fsl_lpuart: Drop unnecessary lpuart*_stop_tx()
By the time lpuart_shutdown() calls lpuart_stop_tx() UARTCR2_TE and
UARTCR2_TIE (which the latter will clear) are already cleared, so that
function call should effectively be a no-op. Moreso, lpuart_stop_tx()
is expected to be executed with port spinlock held, which the caller
doesn't. Given all that, drop the call to lpuart_stop_tx() in
lpuart_shutdown().
In case of lpuart32_shutdown()/lpuart32_stop_tx(), TIE won't even be
set if lpuart_dma_tx_use is true. Drop it there as well.
tty: serial: fsl_lpuart: Use cpu_relax() instead of barrier()
Use cpu_relax() instead of barrier() in a tight polling loops to make
them a bit more idiomatic. Should also improve things on ARM64 a bit
since cpu_relax() will expand into "yield" instruction there.
tty: serial: fsl_lpuart: Use appropriate lpuart32_* I/O funcs
When dealing with 32-bit variant of LPUART IP block appropriate I/O
helpers have to be used to properly deal with endianness
differences. Change all of the offending code to do that.
Fixes: a5fa2660d787 ("tty/serial/fsl_lpuart: Add CONSOLE_POLL support
for lpuart32.") Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com> Cc: Stefan Agner <stefan@agner.ch> Cc: Bhuvanchandra DV <bhuvanchandra.dv@toradex.com> Cc: Chris Healy <cphealy@gmail.com> Cc: Cory Tusar <cory.tusar@zii.aero> Cc: Lucas Stach <l.stach@pengutronix.de> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Jiri Slaby <jslaby@suse.com> Cc: linux-imx@nxp.com Cc: linux-serial@vger.kernel.org Cc: linux-kernel@vger.kernel.org Link: https://lore.kernel.org/r/20190729195226.8862-14-andrew.smirnov@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
tty: serial: fsl_lpuart: Drop no-op bit opearation
The check for termios->c_cflag & CRTSCTS ensure that if we reach else
branch, CRTSCTS in termios->c_cflag is already going to be
cleard. Doing so explicitly there is not necessary. Drop it.
tty: serial: fls_lpuart: Split shared TX IRQ handler into two
While sharing code for Tx interrupt handler between 8 and 32 bit
variant of the peripheral saves a bit of code duplication it also adds
quite a number of lpuart_is_32() checks which makes it harder to
understand. Move shared bits back into corresponding
lpuart*_transmit_buffer functions, split lpuart_txint into
lpuart_txint and lpuart32_txint so we can drop all extra
lpuart_is_32() check and make the code flow more linear.
tty: serial: fsl_lpuart: Fix issue in software flow control
Although I haven't observed this bug in practice, it seems that the
code for handling x_char of LPUART is pretty much identical to that of
i.MX. So the fix found in commit 7e2fb5aa8d81 ("serial: imx: Fix issue
in software flow control"):
serial: imx: Fix issue in software flow control
After send out x_char in UART driver, x_char needs to be cleared
by UART driver itself, otherwise data in TXFIFO can no longer be
sent out.
Also tx counter needs to be increased to keep track of correct
number of transmitted data.
Signed-off-by: Jiada Wang <jiada_wang@mentor.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
should apply here as well.
tty: serial: fsl_lpuart: Drop unnecessary uart_write_wakeup()
Uart_write_wakeup() will already be called as a part of
lpuart*_transmit_buffer() call, so there doesn't seem to be a reason
to call it again right after.
It also appears that second uart_write_wakeup() might potentially
cause unwanted write wakeup when transmitting an x_char. See commit 5e42e9a30cda ("serial: imx: Fix x_char handling and tx flow control")
where this problem was fixed in a very similarly structured i.MX UART
driver.
It appears that lpuart_rxint, lpuart_txint and lpuart32_rxint were
modelled after identical function found in UART driver for
i.MX. However, while said functions are used as individual IRQ
handlers in i.MX driver (in case of i.MX1), it is not the case for
LPUART. Given that, there's no need for us to restrict the prototype
of the handler to irqreturn_t foo(int, void *) and we can drop all of
uneened boilerplate code by changing it void foo(struct lpuart_port *).
Stefan Agner [Mon, 29 Jul 2019 19:52:04 +0000 (12:52 -0700)]
tty: serial: fsl_lpuart: flush receive FIFO after overruns
After overruns the FIFO pointers become misaligned. This
typically shows by characters still being stuck in the FIFO
despite the empty flag being asserted. After the first
assertion of the overrun flag the empty flag still seems to
indicate FIFO state correctly and all data can be read.
However, after another overrun assertion the FIFO seems to
be off by one such that the last received character is still
in the FIFO (despite the empty flag being asserted).
Flushing the receive FIFO reinitializes pointers. Hence it
is recommended to flush the FIFO after overruns, see also:
https://community.nxp.com/thread/321175
Hence, on assertion of the overrun flag read the remaining
data from the FIFO and flush buffers.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com> Acked-by: Max Krummenacher <max.krummenacher@toradex.com> Cc: Stefan Agner <stefan@agner.ch> Cc: Bhuvanchandra DV <bhuvanchandra.dv@toradex.com> Cc: Chris Healy <cphealy@gmail.com> Cc: Cory Tusar <cory.tusar@zii.aero> Cc: Lucas Stach <l.stach@pengutronix.de> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Jiri Slaby <jslaby@suse.com> Cc: linux-imx@nxp.com Cc: linux-serial@vger.kernel.org Cc: linux-kernel@vger.kernel.org Link: https://lore.kernel.org/r/20190729195226.8862-3-andrew.smirnov@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Stefan Agner [Mon, 29 Jul 2019 19:52:03 +0000 (12:52 -0700)]
tty: serial: fsl_lpuart: fix framing error handling when using DMA
When using DMA framing error get cleared properly. However, due
to the additional read from the data register, an underflow in
the receive FIFO buffer occurs (the FIFO pointer gets out of
sync).
Clear the FIFO in case an underflow has occurred. Also disable the
receiver during this operation and when reading the data register to
minimize potential interference.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com> Acked-by: Max Krummenacher <max.krummenacher@toradex.com> Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com> Cc: Stefan Agner <stefan@agner.ch> Cc: Bhuvanchandra DV <bhuvanchandra.dv@toradex.com> Cc: Chris Healy <cphealy@gmail.com> Cc: Cory Tusar <cory.tusar@zii.aero> Cc: Lucas Stach <l.stach@pengutronix.de> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Jiri Slaby <jslaby@suse.com> Cc: linux-imx@nxp.com Cc: linux-serial@vger.kernel.org Cc: linux-kernel@vger.kernel.org Link: https://lore.kernel.org/r/20190729195226.8862-2-andrew.smirnov@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Called in only one place, for RS232, it only obscures things, as it
doesn't go well with 2 similar named functions,
imx_uart_rts_inactive() and imx_uart_rts_active(), that both are
RS485-specific.
serial: imx: set_mctrl(): correctly restore autoRTS state
imx_uart_set_mctrl() happened to set UCR2_CTSC bit whenever TIOCM_RTS
was set, no matter if RTS/CTS handshake is enabled or not. Now fixed by
turning handshake on only when CRTSCTS bit for the port is set.
serial: imx: set_termios(): do not enable autoRTS if RTS is unset
Don't let receiver hardware automatically control RTS output if it
was requested to be inactive.
To ensure this, set_termios() shouldn't set UCR2_CTSC bit if UCR2_CTS
(=TIOCM_RTS) is cleared. Added corresponding check in imx_uart_rts_auto()
to fix this.
8250_lpss: check null return when calling pci_ioremap_bar
pci_ioremap_bar may return null. This is eventually de-referenced at
drivers/dma/dw/core.c:1154 and drivers/dma/dw/core.c:1168. A null check
is needed to prevent null de-reference. I am adding the check and in
case of failure. Thanks to Andy Shevchenko for the hint on the necessity
of pci_iounmap when exiting.
projects / linux.git / log