- 28 Jan, 2024 40 commits
-
-
Tudor Ambarus authored
Since the driver was introduced the port features flags never extended. As we don't expect more flags soon that would bypass the first cacheline of ``struct s3c24xx_uart_info``, change the type of ``has_divslot`` to bool. Bitfields operations incur performance penalty when set or read as compared to direct types. Signed-off-by:
Tudor Ambarus <tudor.ambarus@linaro.org> Link: https://lore.kernel.org/r/20240119104526.1221243-19-tudor.ambarus@linaro.orgSigned-off-by:
Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-
Tudor Ambarus authored
<linux/serial_s3c.h> provides a clock selection pool of maximum 4 clocks. Update the driver to consider a pool selection of maximum 8 clocks. u8 is large enough to allow more clocks than are supported by the driver now, and not too big to cause spanning of ``struct s3c24xx_uart_info`` through 2 cachelines when compiled for arm64. The goal is to reduce the memory footprint of ``struct s3c24xx_uart_info``. Reviewed-by:
Sam Protsenko <semen.protsenko@linaro.org> Signed-off-by:
-
Tudor Ambarus authored
Change the return type of the s3c24xx_serial_rx_fifocnt() method to ``unsigned int`` as the method only returns the fifo size and does not handle error codes. Reviewed-by:
-
Tudor Ambarus authored
s3c24xx_serial_console_txrdy() returned just 0 or 1 to indicate whether the TX is empty or not. Change its return type to bool. Reviewed-by:
-
Tudor Ambarus authored
s3c24xx_serial_txempty_nofifo() returned either 0 or BIT(2), which is counterintuitive. Make the method return bool, and return true when TX is empty and false otherwise. Reviewed-by:
-
Tudor Ambarus authored
The bitwise AND with the fifo mask is used to check if the fifo is empty or not, it doesn't care about the length, thus the comparison with zero is implicit. Rely on the implicit comparison instead. Reviewed-by:
-
Tudor Ambarus authored
``max_count`` negative values are not used. Since ``port->fifosize`` is an unsigned int, make ``max_count`` the same. Reviewed-by:
-
Tudor Ambarus authored
The comment brings no benefit as we can already see from the method's name, ``s3c24xx_serial_pm``, that it deals with power management. Drop the superfluous comment. Reviewed-by:
-
Tudor Ambarus authored
Move open brace '{' following function definition on the next line. Reviewed-by: -
Tudor Ambarus authored
Braces {} are not necessary for single statement blocks. Remove braces on single statement block. Reviewed-by: -
Tudor Ambarus authored
All registers of the IP have 32 bits. Use u32 variables when reading or writing from/to the registers. The purpose of those variables becomes clearer. Reviewed-by:
-
Tudor Ambarus authored
samsung_tty.c uses u32 and relies on <linux/console.h> to include <linux/types.h>. Explicitly include <linux/types.h>. We shall aim to have the driver self contained. Reviewed-by:
-
Tudor Ambarus authored
Sorting headers alphabetically helps locating duplicates, and makes it easier to figure out where to insert new headers. Reviewed-by:
-
Tudor Ambarus authored
The entire bus (PERIC) on which the GS101 serial resides only allows 32-bit register accesses. The reg-io-width dt property is disallowed for the "google,gs101-uart" compatible and instead the iotype is inferred from the compatible. Always set UPIO_MEM32 iotype for the gs101 earlycon. Reviewed-by:
Peter Griffin <peter.griffin@linaro.org> Reviewed-by:
Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> Reviewed-by:
-
Tudor Ambarus authored
GS101's Connectivity Peripheral blocks (peric0/1 blocks) which include the I3C and USI (I2C, SPI, UART) only allow 32-bit register accesses. Instead of specifying the reg-io-width = 4 everywhere, for each node, the requirement should be deduced from the compatible. Infer UPIO_MEM32 iotype from the "google,gs101-uart" compatible. Update the uart info name to be GS101 specific in order to differentiate from the other exynos platforms. All the other settings are not changed. exynos_fifoszdt_serial_drv_data was replaced by gs101_serial_drv_data because the iotype restriction is gs101 specific and there was no other user of exynos_fifoszdt_serial_drv_data. Reviewed-by:
-
Tudor Ambarus authored
GS101's Connectivity Peripheral blocks (peric0/1 blocks) which include the I3C and USI (I2C, SPI, UART) only allow 32-bit register accesses. If using 8-bit register accesses, a SError Interrupt is raised causing the system unusable. Instead of specifying the reg-io-width = 4 everywhere, for each node, the requirement should be deduced from the compatible. Prepare the samsung tty driver to allow IO types different than UPIO_MEM. ``struct uart_port::iotype`` is an unsigned char where all its 8 bits are exposed to uapi. We can't make NULL checks on it to verify if it's set, thus always set it from the driver's data. Use u8 for the ``iotype`` member of ``struct s3c24xx_uart_info`` to emphasize that the iotype is an 8 bit mask. Reviewed-by:
-
Tudor Ambarus authored
All gs101 serial ports are restricted to 32-bit register accesses. This requirement will be inferred from the compatible. Do not allow the reg-io-width property for the google,gs101-uart compatible. Suggested-by:
-
Tudor Ambarus authored
The core expects for tx_empty() either TIOCSER_TEMT when the tx is empty or 0 otherwise. s3c24xx_serial_txempty_nofifo() might return 0x4, and at least uart_get_lsr_info() tries to clear exactly TIOCSER_TEMT (BIT(1)). Fix tx_empty() to return TIOCSER_TEMT. Fixes: 1da177e4 ("Linux-2.6.12-rc2") Signed-off-by:
-
Hugo Villeneuve authored
Fix indentation and add line after do/while() block. Reviewed-by:
Andy Shevchenko <andy.shevchenko@gmail.com> Signed-off-by:
Hugo Villeneuve <hvilleneuve@dimonoff.com> Link: https://lore.kernel.org/r/20240118152213.2644269-18-hugo@hugovil.comSigned-off-by:
-
Hugo Villeneuve authored
Add comments about I2C slave address structure, and reformat to improve readability. Also reformat some comments according to kernel coding style. Reviewed-by:
-
Hugo Villeneuve authored
Fixes the following checkpatch warnings: WARNING: Prefer 'unsigned int' to bare use of 'unsigned' With this change, the affected functions now match the prototypes in struct gpio_chip. Reviewed-by: -
Hugo Villeneuve authored
Simplify driver by defining a common function to handle the power control of all variants. Reviewed-by:
-
Hugo Villeneuve authored
Simplify driver by defining a common function to handle the detection of all variants. Reviewed-by:
-
Hugo Villeneuve authored
GENMASK() is preferred when defining bitmasks. Of all the masks changed, only MAX310x_REV_MASK is actually used. No functional change. Reviewed-by:
-
Hugo Villeneuve authored
Replace dev_err() with dev_err_probe(). This helps in simplifing code and standardizing the error output. Reviewed-by:
-
Hugo Villeneuve authored
Allows to simplify code by removing the break statement in the default switch/case in some functions. Suggested-by:
-
Hugo Villeneuve authored
Simplify error handling and only call uart_remove_one_port() if line bit is set, instead of having to manually set s->p[i].port.dev to NULL. Reviewed-by:
-
Hugo Villeneuve authored
Use a separate regmap name for each port so they can each have their own debugfs entry, allowing to access each port registers independently. For example, a four channels/ports device like the MAX14830 will have four entries in its regmap debugfs: $ find /sys/kernel/debug/regmap -type d | grep spi0.0 /sys/kernel/debug/regmap/spi0.0-port0 /sys/kernel/debug/regmap/spi0.0-port1 /sys/kernel/debug/regmap/spi0.0-port2 /sys/kernel/debug/regmap/spi0.0-port3 Cc: Jan Kundrát <jan.kundrat@cesnet.cz> Link: https://lore.kernel.org/all/77f101f1-897d-4e6d-a8fd-27b818caf768@cesnet.cz/Reviewed-by:
-
Hugo Villeneuve authored
Add macro to hold the maximum number of UART ports per IC/device. Reviewed-by:
Jan Kundrát <jan.kundrat@cesnet.cz> Tested-by:
-
Hugo Villeneuve authored
Running pahole shows that there are some holes within the max310x_devtype structure. Remove holes and optimize alignment by reorganizing structure members. This can also lead to data structure size reduction for some CPUs. On 64-bit CPU (arm64): Before: /* size: 40, cachelines: 1, members: 6 */ /* sum members: 34, holes: 2, sum holes: 6 */ /* last cacheline: 40 bytes */ After: /* size: 40, cachelines: 1, members: 6 */ /* padding: 6 */ /* last cacheline: 40 bytes */ On 32-bit CPU (i386): Before: /* size: 32, cachelines: 1, members: 6 */ /* sum members: 26, holes: 2, sum holes: 6 */ /* last cacheline: 32 bytes */ After: /* size: 24, cachelines: 1, members: 8 */ /* padding: 2 */ /* last cacheline: 24 bytes */ Suggested-by: -
Hugo Villeneuve authored
Replace g with q. Helpful when grepping thru source code or logs for "request" keyword. Fixes: f6544418 ("serial: New serial driver MAX310X") Reviewed-by:
-
Hugo Villeneuve authored
Use preferred spi_get_device_match_data() instead of device_get_match_data() and spi_get_device_id() to get the driver match data. Suggested-by:
-
Hugo Villeneuve authored
Use preferred i2c_get_match_data() instead of device_get_match_data() to get the driver match data. Suggested-by:
-
Hugo Villeneuve authored
This allows to instantiate a max14830 I2C device from userspace. Helpful when testing driver with i2c-stub. Reviewed-by:
-
Hugo Villeneuve authored
When trying to instantiate a max14830 device from userspace: echo max14830 0x60 > /sys/bus/i2c/devices/i2c-2/new_device we get the following error: Unable to handle kernel NULL pointer dereference at virtual address... ... Call trace: max310x_i2c_probe+0x48/0x170 [max310x] i2c_device_probe+0x150/0x2a0 ... Add check for validity of devtype to prevent the error, and abort probe with a meaningful error message. Fixes: 2e1f2d9a ("serial: max310x: implement I2C support") Cc: stable@vger.kernel.org Reviewed-by: -
Valentin Caron authored
Since STM32MP25, FIFO size could vary regarding the STM32MPxx version. So we get this size from "hwcfgr1" register and compute threshold values corresponding to the ratio given by reference manual. As STM32MP1x, STM32MP25 and STM32H7 share the same compatible and STM32H7 doesn't have a register to get FIFO size, we force FIFO size to 16 in case of zero read from hwcfgr1 register. Signed-off-by:
Valentin Caron <valentin.caron@foss.st.com> Link: https://lore.kernel.org/r/20240112095300.2004878-5-valentin.caron@foss.st.comSigned-off-by:
-
Valentin Caron authored
USART has registers above 0xff offset, so extend variable type to u16. And change UNDEF_REG to 0xffff. Signed-off-by:
-
Valentin Caron authored
STM32MP25x got 9 instances of U(S)ART. So extend STM32_MAX_PORTS to 9, in order to handle all instances. Signed-off-by:
-
Valentin Caron authored
In the case of high USART input clock and low baud rate, BRR value is not enough to get correct baud rate. So here we use USART prescaler to divide USART input clock to get the correct baud rate. PRESC register is only available since stm32h7. Signed-off-by:
-
Lukas Bulwahn authored
The config HW_CONSOLE is always identical to the config VT and is not visible in the kernel's build menuconfig. So, CONFIG_HW_CONSOLE is redundant. Replace all references to CONFIG_HW_CONSOLE with CONFIG_VT and remove CONFIG_HW_CONSOLE. Signed-off-by:
Lukas Bulwahn <lukas.bulwahn@gmail.com> Reviewed-by:
Javier Martinez Canillas <javierm@redhat.com> Acked-by:
Dmitry Torokhov <dmitry.torokhov@gmail.com> Reviewed-by:
Geert Uytterhoeven <geert@linux-m68k.org> Acked-by:
-
