Skip to contentSkip to conversation

serial: liteuart: add IRQ support

PATCH Superseded Gabriel Somlo 3 participants 39 messages Nov 18, 2022 – Nov 23, 2022

Subsystems (2)

v5 of 6 Superseded Latest is v6 →

lore
Share & tools
lore
#0 lore raw

v5 cover serial: liteuart: add IRQ support

Add IRQ support to the LiteX LiteUART serial interface

Changes from v4:
- picked up r/b Ilpo Järvinen on 12/14 and 13/14
- 12/14 ("add IRQ support for the RX path"): using dev_err() instead

    of combining pr_err() and pr_fmt(); also, remove superfluous comment
Quoted 35 lines

Changes from v3:
- picked up r/b Ilpo Järvinen on select subset of commits
- rebased entire series on top of tty-next tree
- 2/14 ("use bit number macros"): explicitly include <linux/bits.h>
- 3/14 ("remove unused uart_ops stubs"): don't add gratuitous comment removed later on in the series
- 12/14 ("add IRQ support for the RX path"): add shadow irq register to support polling mode and avoid reading hardware mmio irq register to learn which irq flags are enabled
- this also simplifies both liteuart_interrupt() and liteuart_startup()
- 13/14 ("add IRQ support for the TX path"):
- removed superfluous curly braces from liteuart_interrupt()
- simplified [start|stop]_tx() by using shadow irq register from 12/14
- simplified liteuart_tx_chars() by rebasing on top of tty-next tree

Changes from v2:
- further split out "separate RX loop from poll timer" into dedicated patches highlighting additional changes explicitly:
- factoring out tty_flip_buffer_push() (6/14)
- ack only RX events in RX loop (7/14)
- pass constant flag to uart_insert_char() directly (8/14)
- fix variable types in rx loop (9/14)
- separating RX loop from poll timer (10/14)
- added patch (11/14) to move function definitions to a more convenient location, making subsequent changes easier to read in diff patch form.
- fixes and clarifications for RX path IRQ support patch (now 12/14):
- only return IRQ_HANDLED for RX events
- use pr_fmt() to improve style of irq setup error message
- remove unnecessary locking from around single-register access
- modify TX path to support both IRQ-mode and polling (13/14)
- move polling-only liteuart_putchar() behind same conditional (CONFIG_SERIAL_LITEUART_CONSOLE) as the rest of the code that's still using it.

Quoted 4 lines

Changes from v1:
- split minor cosmetic changes out into individual patches

    (1/3 became 1..5/7)

- patches 6/7 and 7/7 unchanged (used to be 2/3 and 3/3)

Gabriel Somlo (14):
serial: liteuart: use KBUILD_MODNAME as driver name
serial: liteuart: use bit number macros
serial: liteuart: remove unused uart_ops stubs
serial: liteuart: don't set unused port fields
serial: liteuart: minor style fix in liteuart_init()
serial: liteuart: move tty_flip_buffer_push() out of rx loop
serial: liteuart: rx loop should only ack rx events
serial: liteuart: simplify passing of uart_insert_char() flag
serial: liteuart: fix rx loop variable types
serial: liteuart: separate rx loop from poll timer
serial: liteuart: move function definitions
serial: liteuart: add IRQ support for the RX path
serial: liteuart: add IRQ support for the TX path
serial: liteuart: move polling putchar() function

 drivers/tty/serial/liteuart.c | 210 ++++++++++++++++++++--------------
 1 file changed, 126 insertions(+), 84 deletions(-)
Signature · 1 line
2.38.1
#1 lore raw

PATCH v5 1/14 serial: liteuart: use KBUILD_MODNAME as driver name

Replace hard-coded instances of "liteuart" with KBUILD_MODNAME.

---
 drivers/tty/serial/liteuart.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/tty/serial/liteuart.c +3 -3
index 062812fe1b09..db898751ffe3 100644
@@ -57,7 +57,7 @@ static struct console liteuart_console;
 
 static struct uart_driver liteuart_driver = {
 	.owner = THIS_MODULE,
-	.driver_name = "liteuart",
+	.driver_name = KBUILD_MODNAME,
 	.dev_name = "ttyLXU",
 	.major = 0,
 	.minor = 0,
@@ -321,7 +321,7 @@ static struct platform_driver liteuart_platform_driver = {
 	.probe = liteuart_probe,
 	.remove = liteuart_remove,
 	.driver = {
-		.name = "liteuart",
+		.name = KBUILD_MODNAME,
 		.of_match_table = liteuart_of_match,
 	},
 };
@@ -367,7 +367,7 @@ static int liteuart_console_setup(struct console *co, char *options)
 }
 
 static struct console liteuart_console = {
-	.name = "liteuart",
+	.name = KBUILD_MODNAME,
 	.write = liteuart_console_write,
 	.device = uart_console_device,
 	.setup = liteuart_console_setup,
Signature · 1 line
2.38.1
Reviewed-by on v5 1/14
#2 lore raw

Re: [PATCH v5 01/14] serial: liteuart: use KBUILD_MODNAME as driver name

Quoted 3 lines · Gabriel Somlo

Replace hard-coded instances of "liteuart" with KBUILD_MODNAME.

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

#3 lore raw

PATCH v5 2/14 serial: liteuart: use bit number macros

In reply to Gabriel Somlo

Replace magic bit constants (e.g., 1, 2, 4) with BIT(x) expressions.

---
 drivers/tty/serial/liteuart.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
drivers/tty/serial/liteuart.c +3 -2
index db898751ffe3..18c1eb315ee9 100644
@@ -5,6 +5,7 @@
  * Copyright (C) 2019-2020 Antmicro <www.antmicro.com>
  */
 
+#include <linux/bits.h>
 #include <linux/console.h>
 #include <linux/litex.h>
 #include <linux/module.h>
@@ -38,8 +39,8 @@
 #define OFF_EV_ENABLE	0x14
 
 /* events */
-#define EV_TX		0x1
-#define EV_RX		0x2
+#define EV_TX		BIT(0)
+#define EV_RX		BIT(1)
 
 struct liteuart_port {
 	struct uart_port port;
Signature · 1 line
2.38.1
Reviewed-by on v5 2/14
#4 lore raw

Re: [PATCH v5 02/14] serial: liteuart: use bit number macros

Quoted 3 lines · Gabriel Somlo

Replace magic bit constants (e.g., 1, 2, 4) with BIT(x) expressions.

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

#5 lore raw

PATCH v5 3/14 serial: liteuart: remove unused uart_ops stubs

In reply to Gabriel Somlo

Remove stub uart_ops methods that are not called unconditionally from serial_core.

---
 drivers/tty/serial/liteuart.c | 17 -----------------
 1 file changed, 17 deletions(-)
drivers/tty/serial/liteuart.c +0 -17
index 18c1eb315ee9..989a4f8d5bd4 100644
@@ -154,11 +154,6 @@ static void liteuart_stop_rx(struct uart_port *port)
 	del_timer(&uart->timer);
 }
 
-static void liteuart_break_ctl(struct uart_port *port, int break_state)
-{
-	/* LiteUART doesn't support sending break signal */
-}
-
 static int liteuart_startup(struct uart_port *port)
 {
 	struct liteuart_port *uart = to_liteuart_port(port);
@@ -197,15 +192,6 @@ static const char *liteuart_type(struct uart_port *port)
 	return "liteuart";
 }
 
-static void liteuart_release_port(struct uart_port *port)
-{
-}
-
-static int liteuart_request_port(struct uart_port *port)
-{
-	return 0;
-}
-
 static void liteuart_config_port(struct uart_port *port, int flags)
 {
 	/*
@@ -232,13 +218,10 @@ static const struct uart_ops liteuart_ops = {
 	.stop_tx	= liteuart_stop_tx,
 	.start_tx	= liteuart_start_tx,
 	.stop_rx	= liteuart_stop_rx,
-	.break_ctl	= liteuart_break_ctl,
 	.startup	= liteuart_startup,
 	.shutdown	= liteuart_shutdown,
 	.set_termios	= liteuart_set_termios,
 	.type		= liteuart_type,
-	.release_port	= liteuart_release_port,
-	.request_port	= liteuart_request_port,
 	.config_port	= liteuart_config_port,
 	.verify_port	= liteuart_verify_port,
 };
Signature · 1 line
2.38.1
Reviewed-by on v5 3/14
#6 lore raw

Re: [PATCH v5 03/14] serial: liteuart: remove unused uart_ops stubs

Quoted 5 lines · Gabriel Somlo

Remove stub uart_ops methods that are not called unconditionally from serial_core.

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

#7 lore raw

PATCH v5 4/14 serial: liteuart: don't set unused port fields

In reply to Gabriel Somlo

Remove regshift and iobase port fields, since they are unused by the driver.

---
 drivers/tty/serial/liteuart.c | 2 --
 1 file changed, 2 deletions(-)
drivers/tty/serial/liteuart.c +0 -2
index 989a4f8d5bd4..c6eb7eba5af8 100644
@@ -263,9 +263,7 @@ static int liteuart_probe(struct platform_device *pdev)
 	port->iotype = UPIO_MEM;
 	port->flags = UPF_BOOT_AUTOCONF;
 	port->ops = &liteuart_ops;
-	port->regshift = 2;
 	port->fifosize = 16;
-	port->iobase = 1;
 	port->type = PORT_UNKNOWN;
 	port->line = dev_id;
 	spin_lock_init(&port->lock);
Signature · 1 line
2.38.1
Reviewed-by on v5 4/14
#8 lore raw

Re: [PATCH v5 04/14] serial: liteuart: don't set unused port fields

Quoted 4 lines · Gabriel Somlo

Remove regshift and iobase port fields, since they are unused by the driver.

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

#9 lore raw

PATCH v5 5/14 serial: liteuart: minor style fix in liteuart_init()

In reply to Gabriel Somlo
---
 drivers/tty/serial/liteuart.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)
drivers/tty/serial/liteuart.c +2 -4
index c6eb7eba5af8..1e3429bcc2ad 100644
@@ -397,12 +397,10 @@ static int __init liteuart_init(void)
 		return res;
 
 	res = platform_driver_register(&liteuart_platform_driver);
-	if (res) {
+	if (res)
 		uart_unregister_driver(&liteuart_driver);
-		return res;
-	}
 
-	return 0;
+	return res;
 }
 
 static void __exit liteuart_exit(void)
Signature · 1 line
2.38.1
Reviewed-by on v5 5/14
#10 lore raw

Re: [PATCH v5 05/14] serial: liteuart: minor style fix in liteuart_init()

Quoted 2 lines · Gabriel Somlo

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

#11 lore raw

PATCH v5 6/14 serial: liteuart: move tty_flip_buffer_push() out of rx loop

In reply to Gabriel Somlo

Calling tty_flip_buffer_push() for each individual received character is overkill. Move it out of the rx loop, and only call it once per set of characters received together.

---
 drivers/tty/serial/liteuart.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/tty/serial/liteuart.c +2 -2
index 1e3429bcc2ad..81a86c5eb393 100644
@@ -87,10 +87,10 @@ static void liteuart_timer(struct timer_list *t)
 		/* no overflow bits in status */
 		if (!(uart_handle_sysrq_char(port, ch)))
 			uart_insert_char(port, status, 0, ch, flg);
-
-		tty_flip_buffer_push(&port->state->port);
 	}
 
+	tty_flip_buffer_push(&port->state->port);
+
 	mod_timer(&uart->timer, jiffies + uart_poll_timeout(port));
 }
 
Signature · 1 line
2.38.1
Reviewed-by on v5 6/14
#12 lore raw

Re: [PATCH v5 06/14] serial: liteuart: move tty_flip_buffer_push() out of rx loop

Quoted 6 lines · Gabriel Somlo

Calling tty_flip_buffer_push() for each individual received character is overkill. Move it out of the rx loop, and only call it once per set of characters received together.

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

#13 lore raw

PATCH v5 7/14 serial: liteuart: rx loop should only ack rx events

In reply to Gabriel Somlo

While receiving characters, it is necessary to acknowledge each one by writing to the EV_PENDING register's EV_RX bit. Ensure we do not also gratuitously set the EV_TX bit in the process.

---
 drivers/tty/serial/liteuart.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/tty/serial/liteuart.c +1 -1
index 81a86c5eb393..c90ab65fbdcf 100644
@@ -82,7 +82,7 @@ static void liteuart_timer(struct timer_list *t)
 		port->icount.rx++;
 
 		/* necessary for RXEMPTY to refresh its value */
-		litex_write8(membase + OFF_EV_PENDING, EV_TX | EV_RX);
+		litex_write8(membase + OFF_EV_PENDING, EV_RX);
 
 		/* no overflow bits in status */
 		if (!(uart_handle_sysrq_char(port, ch)))
Signature · 1 line
2.38.1
#14 lore raw

PATCH v5 8/14 serial: liteuart: simplify passing of uart_insert_char() flag

In reply to Gabriel Somlo

Simply provide the hard-coded TTY_NORMAL flag to uart_insert_char() directly -- no need to dedicate a variable for that exclusive purpose.

---
 drivers/tty/serial/liteuart.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/tty/serial/liteuart.c +1 -2
index c90ab65fbdcf..81aa7c1da73c 100644
@@ -73,7 +73,6 @@ static void liteuart_timer(struct timer_list *t)
 	struct liteuart_port *uart = from_timer(uart, t, timer);
 	struct uart_port *port = &uart->port;
 	unsigned char __iomem *membase = port->membase;
-	unsigned int flg = TTY_NORMAL;
 	int ch;
 	unsigned long status;
 
@@ -86,7 +85,7 @@ static void liteuart_timer(struct timer_list *t)
 
 		/* no overflow bits in status */
 		if (!(uart_handle_sysrq_char(port, ch)))
-			uart_insert_char(port, status, 0, ch, flg);
+			uart_insert_char(port, status, 0, ch, TTY_NORMAL);
 	}
 
 	tty_flip_buffer_push(&port->state->port);
Signature · 1 line
2.38.1
Reviewed-by on v5 8/14
#15 lore raw

Re: [PATCH v5 08/14] serial: liteuart: simplify passing of uart_insert_char() flag

Quoted 5 lines · Gabriel Somlo

Simply provide the hard-coded TTY_NORMAL flag to uart_insert_char() directly -- no need to dedicate a variable for that exclusive purpose.

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

#16 lore raw

PATCH v5 9/14 serial: liteuart: fix rx loop variable types

In reply to Gabriel Somlo

Update variable types to match the signature of uart_insert_char() which consumes them.

---
 drivers/tty/serial/liteuart.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/tty/serial/liteuart.c +1 -2
index 81aa7c1da73c..42ac9aee050a 100644
@@ -73,8 +73,7 @@ static void liteuart_timer(struct timer_list *t)
 	struct liteuart_port *uart = from_timer(uart, t, timer);
 	struct uart_port *port = &uart->port;
 	unsigned char __iomem *membase = port->membase;
-	int ch;
-	unsigned long status;
+	unsigned int status, ch;
 
 	while ((status = !litex_read8(membase + OFF_RXEMPTY)) == 1) {
 		ch = litex_read8(membase + OFF_RXTX);
Signature · 1 line
2.38.1
maintainer
#17 lore raw

Re: [PATCH v5 09/14] serial: liteuart: fix rx loop variable types

Quoted 14 lines · Gabriel Somlo

Update variable types to match the signature of uart_insert_char() which consumes them.

---
  drivers/tty/serial/liteuart.c | 3 +--
  1 file changed, 1 insertion(+), 2 deletions(-)
drivers/tty/serial/liteuart.c +0 -0
index 81aa7c1da73c..42ac9aee050a 100644

drivers/tty/serial/liteuart.c static void liteuart_timer(struct timer_list *t) · line 76

Quoted 6 lines
drivers/tty/serial/liteuart.c +1 -2
@@ -73,8 +73,7 @@ static void liteuart_timer(struct timer_list *t)
  	struct liteuart_port *uart = from_timer(uart, t, timer);
  	struct uart_port *port = &uart->port;
  	unsigned char __iomem *membase = port->membase;
-	int ch;
-	unsigned long status;
+	unsigned int status, ch;

These should be u8 after all, right?

Wait, status is a bool in the end:

drivers/tty/serial/liteuart.c static void liteuart_timer(struct timer_list *t) · line 78

Quoted 1 line
drivers/tty/serial/liteuart.c +0 -0
  	while ((status = !litex_read8(membase + OFF_RXEMPTY)) == 1) {

But why is it passed to uart_insert_char() as such? That's ugly. Maybe drop all of this "status" and pass LSR_RXC directly. The while's condition would simply look like (!litex_read8(membase + OFF_RXEMPTY)) then.

drivers/tty/serial/liteuart.c static void liteuart_timer(struct timer_list *t) · line 79

Quoted 1 line
drivers/tty/serial/liteuart.c +0 -0
  		ch = litex_read8(membase + OFF_RXTX);

thanks,

Signature · 2 lines
js
suse labs
maintainer
#18 lore raw

Re: [PATCH v5 09/14] serial: liteuart: fix rx loop variable types

Quoted 24 lines · Jiri Slaby
Quoted 14 lines · Gabriel Somlo

Update variable types to match the signature of uart_insert_char() which consumes them.

---
  drivers/tty/serial/liteuart.c | 3 +--

  1 file changed, 1 insertion(+), 2 deletions(-)
drivers/tty/serial/liteuart.c +0 -0
index 81aa7c1da73c..42ac9aee050a 100644

drivers/tty/serial/liteuart.c static void liteuart_timer(struct timer_list *t) · line 76

Quoted 6 lines
drivers/tty/serial/liteuart.c +1 -2
@@ -73,8 +73,7 @@ static void liteuart_timer(struct timer_list *t)
       struct liteuart_port *uart = from_timer(uart, t, timer);
       struct uart_port *port = &uart->port;
       unsigned char __iomem *membase = port->membase;
-    int ch;
-    unsigned long status;
+    unsigned int status, ch;

These should be u8 after all, right?

Quoted 5 lines

Wait, status is a bool in the end:

drivers/tty/serial/liteuart.c static void liteuart_timer(struct timer_list *t) · line 78

Quoted 1 line
drivers/tty/serial/liteuart.c +0 -0
       while ((status = !litex_read8(membase + OFF_RXEMPTY)) == 1) {

But why is it passed to uart_insert_char() as such? That's ugly. Maybe drop all of this "status" and pass LSR_RXC directly.

Actually, even 0 directly, provided overrun parameter is always 0.

Quoted 5 lines

The while's condition would simply look like (!litex_read8(membase + OFF_RXEMPTY)) then.

drivers/tty/serial/liteuart.c static void liteuart_timer(struct timer_list *t) · line 79

Quoted 1 line
drivers/tty/serial/liteuart.c +0 -0
           ch = litex_read8(membase + OFF_RXTX);

thanks,

Signature · 1 line
js
maintainer
#19 lore raw

Re: [PATCH v5 09/14] serial: liteuart: fix rx loop variable types

In reply to Jiri Slaby
Quoted 10 lines · Jiri Slaby
Quoted 9 lines · Gabriel Somlo

Update variable types to match the signature of uart_insert_char() which consumes them.

---
  drivers/tty/serial/liteuart.c | 3 +--

  1 file changed, 1 insertion(+), 2 deletions(-)
Quoted 13 lines

drivers/tty/serial/liteuart.c static void liteuart_timer(struct timer_list *t) · line 76

Quoted 11 lines
drivers/tty/serial/liteuart.c +1 -2
index 81aa7c1da73c..42ac9aee050a 100644
@@ -73,8 +73,7 @@ static void liteuart_timer(struct timer_list *t)
       struct liteuart_port *uart = from_timer(uart, t, timer);
       struct uart_port *port = &uart->port;
       unsigned char __iomem *membase = port->membase;
-    int ch;
-    unsigned long status;
+    unsigned int status, ch;

These should be u8 after all, right?

And can you change membase to u8 * too 8-)?

Signature · 2 lines
js
suse labs
#20 lore raw

Re: [PATCH v5 09/14] serial: liteuart: fix rx loop variable types

Hi Jiri,

Thanks for the feedback!

Quoted 11 lines · Jiri Slaby
Quoted 10 lines · Jiri Slaby
Quoted 9 lines · Gabriel Somlo

Update variable types to match the signature of uart_insert_char() which consumes them.

---
  drivers/tty/serial/liteuart.c | 3 +--

  1 file changed, 1 insertion(+), 2 deletions(-)
Quoted 14 lines
Quoted 6 lines
Quoted 6 lines
drivers/tty/serial/liteuart.c +0 -0
index 81aa7c1da73c..42ac9aee050a 100644
@@ -73,8 +73,7 @@ static void liteuart_timer(struct timer_list *t)
Quoted 8 lines

drivers/tty/serial/liteuart.c static void liteuart_timer(struct timer_list *t) · line 76

Quoted 6 lines
drivers/tty/serial/liteuart.c +1 -2
       struct liteuart_port *uart = from_timer(uart, t, timer);
       struct uart_port *port = &uart->port;
       unsigned char __iomem *membase = port->membase;
-    int ch;
-    unsigned long status;
+    unsigned int status, ch;

These should be u8 after all, right?

OK, so:

- I can hard-code status as 1, like so:

	while(!litex_read8(membase + OFF_RXEMPTY) {
		...
		if (!(uart_handle_sysrq_char(port, ch)))
			uart_insert_char(port, 1, 0, ch, TTY_NORMAL);

    ... since `status` would always be `1` inside the loop. So I'm
    basically going to get rid of it altogether.

- ch is indeed *produced* by litex_read8(), which would make it u8. It is subsequently *consumed* by uart_handle_sysrq_char() and uart_insert_char(), which both expect unsigned int.

If you think it's better to go with the type when the value is produced (as opposed to when it's consumed), I'm OK with that for the upcoming v6 of the series... :)

Quoted 1 line

And can you change membase to u8 * too 8-)?

Hmmm, in struct uart_port (in include/linux/serial_core.h), the membase field is declared as:

	unsigned char __iomem   *membase;

which is why I'm thinking we should leave it as-is? Unless there are plans (or a pending patch I'm unaware of) to switch the field in include/linux/serial_core.h to u8 as well? -- Please advise.

Thanks again,
--Gabriel

Quoted 4 lines
Signature · 2 lines
js
suse labs
maintainer
#21 lore raw

Re: [PATCH v5 09/14] serial: liteuart: fix rx loop variable types

Quoted 33 lines · Gabriel L. Somlo

Hi Jiri,

Thanks for the feedback!

Quoted 11 lines · Jiri Slaby
Quoted 10 lines · Jiri Slaby
Quoted 9 lines · Gabriel Somlo

Update variable types to match the signature of uart_insert_char() which consumes them.

---
   drivers/tty/serial/liteuart.c | 3 +--
   1 file changed, 1 insertion(+), 2 deletions(-)
Quoted 14 lines
Quoted 6 lines
Quoted 6 lines
drivers/tty/serial/liteuart.c +0 -0
index 81aa7c1da73c..42ac9aee050a 100644
@@ -73,8 +73,7 @@ static void liteuart_timer(struct timer_list *t)
Quoted 8 lines

drivers/tty/serial/liteuart.c static void liteuart_timer(struct timer_list *t) · line 76

Quoted 6 lines
drivers/tty/serial/liteuart.c +1 -2
       struct liteuart_port *uart = from_timer(uart, t, timer);
       struct uart_port *port = &uart->port;
       unsigned char __iomem *membase = port->membase;
-    int ch;
-    unsigned long status;
+    unsigned int status, ch;

These should be u8 after all, right?

OK, so:

Quoted 9 lines
   - I can hard-code `status` as `1`, like so:

	while(!litex_read8(membase + OFF_RXEMPTY) {
		...
		if (!(uart_handle_sysrq_char(port, ch)))
			uart_insert_char(port, 1, 0, ch, TTY_NORMAL);

     ... since `status` would always be `1` inside the loop. So I'm
     basically going to get rid of it altogether.

Yes, I had that in my mind. Except passing 1 to uart_insert_char() when overflow is hardwired to 0 makes no sense IMO :).

Quoted 3 lines

- ch is indeed *produced* by litex_read8(), which would make it u8. It is subsequently *consumed* by uart_handle_sysrq_char() and uart_insert_char(), which both expect unsigned int.

Ignore uart_handle_sysrq_char and uart_insert_char. They should be fixed one day. It should really be u8. All down the call chain (it magically turns into int in the sysrq handlers, then char is expected).

Quoted 3 lines

If you think it's better to go with the type when the value is produced (as opposed to when it's consumed), I'm OK with that for the upcoming v6 of the series... :)

Yes, please. We should slowly convert _all_ of them.

Quoted 10 lines
Quoted 1 line

And can you change membase to u8 * too 8-)?

Hmmm, in struct uart_port (in include/linux/serial_core.h), the membase field is declared as:

	unsigned char __iomem   *membase;

which is why I'm thinking we should leave it as-is? Unless there are plans (or a pending patch I'm unaware of) to switch the field in include/linux/serial_core.h to u8 as well? -- Please advise.

Ah, then keep it. I somehow thought it's void *. And yes, even this should be u8 __iomem *, eventually.

thanks,

Signature · 2 lines
js
suse labs
#22 lore raw

Re: [PATCH v5 09/14] serial: liteuart: fix rx loop variable types

Quoted 41 lines · Jiri Slaby
Quoted 40 lines · Gabriel L. Somlo

Hi Jiri,

Thanks for the feedback!

Quoted 11 lines · Jiri Slaby
Quoted 10 lines · Jiri Slaby
Quoted 9 lines · Gabriel Somlo

Update variable types to match the signature of uart_insert_char() which consumes them.

---
   drivers/tty/serial/liteuart.c | 3 +--
   1 file changed, 1 insertion(+), 2 deletions(-)
Quoted 14 lines
Quoted 6 lines
Quoted 6 lines
drivers/tty/serial/liteuart.c +0 -0
index 81aa7c1da73c..42ac9aee050a 100644
@@ -73,8 +73,7 @@ static void liteuart_timer(struct timer_list *t)
Quoted 8 lines

drivers/tty/serial/liteuart.c static void liteuart_timer(struct timer_list *t) · line 76

Quoted 6 lines
drivers/tty/serial/liteuart.c +1 -2
       struct liteuart_port *uart = from_timer(uart, t, timer);
       struct uart_port *port = &uart->port;
       unsigned char __iomem *membase = port->membase;
-    int ch;
-    unsigned long status;
+    unsigned int status, ch;

These should be u8 after all, right?

OK, so:

   - I can hard-code `status` as `1`, like so:

	while(!litex_read8(membase + OFF_RXEMPTY) {
		...
		if (!(uart_handle_sysrq_char(port, ch)))
			uart_insert_char(port, 1, 0, ch, TTY_NORMAL);
Quoted 5 lines
Quoted 2 lines

... since status would always be 1 inside the loop. So I'm basically going to get rid of it altogether.

Yes, I had that in my mind. Except passing 1 to uart_insert_char() when overflow is hardwired to 0 makes no sense IMO :).

So, looking at what uart_insert_char() does, I could simply do this instead:

while(!litex_read8(membase + OFF_RXEMPTY) {
...

		/* LiteUART does not provide overrun bits */

if (!(uart_handle_sysrq_char(port, ch) ||
tty_insert_flip_char(&port->state->port, ch, TTY_NORMAL)))
++port->icount.buf_overrun;

That is, tty_insert_flip_char() is the portion of uart_insert_char()` that actually gets executed if status is 1 and overrun is 0...

I'm not quite confident about whether this is an improvement in legibility and/or code quality, but please let me know what *you* think... :)

Quoted 8 lines
Quoted 3 lines

- ch is indeed *produced* by litex_read8(), which would make it u8. It is subsequently *consumed* by uart_handle_sysrq_char() and uart_insert_char(), which both expect unsigned int.

Ignore uart_handle_sysrq_char and uart_insert_char. They should be fixed one day. It should really be u8. All down the call chain (it magically turns into int in the sysrq handlers, then char is expected).

Quoted 5 lines
Quoted 3 lines

If you think it's better to go with the type when the value is produced (as opposed to when it's consumed), I'm OK with that for the upcoming v6 of the series... :)

Yes, please. We should slowly convert _all_ of them.

OK, u8 it is, then.

Quoted 5 lines
Quoted 5 lines
Quoted 1 line

And can you change membase to u8 * too 8-)?

Hmmm, in struct uart_port (in include/linux/serial_core.h), the membase field is declared as:

Quoted 8 lines
Quoted 5 lines
	unsigned char __iomem   *membase;

which is why I'm thinking we should leave it as-is? Unless there are plans (or a pending patch I'm unaware of) to switch the field in include/linux/serial_core.h to u8 as well? -- Please advise.

Ah, then keep it. I somehow thought it's void *. And yes, even this should be u8 __iomem *, eventually.

OK, it should/will get updated in bulk once that change is made across the entire subsystem, leaving as-is for now.

Thanks again for all your help and advice!
--Gabriel

maintainer
#23 lore raw

Re: [PATCH v5 09/14] serial: liteuart: fix rx loop variable types

Quoted 10 lines · Gabriel L. Somlo

So, looking at what uart_insert_char() does, I could simply do this instead:

while(!litex_read8(membase + OFF_RXEMPTY) {
...

		/* LiteUART does not provide overrun bits */

if (!(uart_handle_sysrq_char(port, ch) ||
tty_insert_flip_char(&port->state->port, ch, TTY_NORMAL)))
++port->icount.buf_overrun;

Quoted 5 lines

That is, tty_insert_flip_char() is the portion of uart_insert_char()` that actually gets executed if status is 1 and overrun is 0...

I'm not quite confident about whether this is an improvement in legibility and/or code quality,

It's not :). Keep the uart_ helper.

Signature · 2 lines
js
suse labs
#24 lore raw

PATCH v5 10/14 serial: liteuart: separate rx loop from poll timer

In reply to Gabriel Somlo

Convert the rx loop into its own dedicated function, and (for now) call it from the poll timer. This is in preparation for adding irq support to the receive path.

---
 drivers/tty/serial/liteuart.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)
drivers/tty/serial/liteuart.c +9 -3
index 42ac9aee050a..76f8a09b82cd 100644
@@ -68,10 +68,8 @@ static struct uart_driver liteuart_driver = {
 #endif
 };
 
-static void liteuart_timer(struct timer_list *t)
+static void liteuart_rx_chars(struct uart_port *port)
 {
-	struct liteuart_port *uart = from_timer(uart, t, timer);
-	struct uart_port *port = &uart->port;
 	unsigned char __iomem *membase = port->membase;
 	unsigned int status, ch;
 
@@ -88,6 +86,14 @@ static void liteuart_timer(struct timer_list *t)
 	}
 
 	tty_flip_buffer_push(&port->state->port);
+}
+
+static void liteuart_timer(struct timer_list *t)
+{
+	struct liteuart_port *uart = from_timer(uart, t, timer);
+	struct uart_port *port = &uart->port;
+
+	liteuart_rx_chars(port);
 
 	mod_timer(&uart->timer, jiffies + uart_poll_timeout(port));
 }
Signature · 1 line
2.38.1
Reviewed-by on v5 10/14
#25 lore raw

Re: [PATCH v5 10/14] serial: liteuart: separate rx loop from poll timer

Hi Gabriel,

Quoted 6 lines · Gabriel Somlo

Convert the rx loop into its own dedicated function, and (for now) call it from the poll timer. This is in preparation for adding irq support to the receive path.

Thanks for your patch!

drivers/tty/serial/liteuart.c static struct uart_driver liteuart_driver = { · line 71

Quoted 8 lines
drivers/tty/serial/liteuart.c +1 -1
@@ -68,10 +68,8 @@ static struct uart_driver liteuart_driver = {
 #endif
 };
 
-static void liteuart_timer(struct timer_list *t)
+static void liteuart_rx_chars(struct uart_port *port)

So first you spin this off into a separate function, so it can be called from both the interrupt and polling paths.
Later, in "[PATCH v5 12/14] serial: liteuart: add IRQ support for the RX path", you remove the call from the polling path...

Quoted 11 lines
drivers/tty/serial/liteuart.c +1 -2
 {
-       struct liteuart_port *uart = from_timer(uart, t, timer);
-       struct uart_port *port = &uart->port;
        unsigned char __iomem *membase = port->membase;
        unsigned int status, ch;
 
@@ -88,6 +86,14 @@ static void liteuart_timer(struct timer_list *t)
        }
 
        tty_flip_buffer_push(&port->state->port);
+}

drivers/tty/serial/liteuart.c static void liteuart_timer(struct timer_list *t) · line 99

Quoted 10 lines
drivers/tty/serial/liteuart.c +7 -0
+
+static void liteuart_timer(struct timer_list *t)
+{
+       struct liteuart_port *uart = from_timer(uart, t, timer);
+       struct uart_port *port = &uart->port;
+
+       liteuart_rx_chars(port);
 
        mod_timer(&uart->timer, jiffies + uart_poll_timeout(port));
 }

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

#26 lore raw

Re: [PATCH v5 10/14] serial: liteuart: separate rx loop from poll timer

Quoted 20 lines · Geert Uytterhoeven

Hi Gabriel,

Quoted 6 lines · Gabriel Somlo

Convert the rx loop into its own dedicated function, and (for now) call it from the poll timer. This is in preparation for adding irq support to the receive path.

Thanks for your patch!

Quoted 6 lines
drivers/tty/serial/liteuart.c +0 -0
@@ -68,10 +68,8 @@ static struct uart_driver liteuart_driver = {
 #endif
 };
 
Quoted 7 lines

drivers/tty/serial/liteuart.c static struct uart_driver liteuart_driver = { · line 71

Quoted 2 lines
drivers/tty/serial/liteuart.c +1 -1
-static void liteuart_timer(struct timer_list *t)
+static void liteuart_rx_chars(struct uart_port *port)

So first you spin this off into a separate function, so it can be called from both the interrupt and polling paths.
Later, in "[PATCH v5 12/14] serial: liteuart: add IRQ support for the RX path", you remove the call from the polling path...

That's right -- the polling path now calls the IRQ handler, which, in turn, calls liteuart_rx_chars() (and later *_tx_chars() as well).

The polling timer becomes just an alternative way to invoke the irq handler, giving it a chance to take care of any pending transfers in either/both directions.

Thanks,
--Gabriel

Quoted 32 lines
Quoted 11 lines
drivers/tty/serial/liteuart.c +1 -2
 {
-       struct liteuart_port *uart = from_timer(uart, t, timer);
-       struct uart_port *port = &uart->port;
        unsigned char __iomem *membase = port->membase;
        unsigned int status, ch;
 
@@ -88,6 +86,14 @@ static void liteuart_timer(struct timer_list *t)
        }
 
        tty_flip_buffer_push(&port->state->port);
+}

drivers/tty/serial/liteuart.c static void liteuart_timer(struct timer_list *t) · line 99

Quoted 10 lines
drivers/tty/serial/liteuart.c +7 -0
+
+static void liteuart_timer(struct timer_list *t)
+{
+       struct liteuart_port *uart = from_timer(uart, t, timer);
+       struct uart_port *port = &uart->port;
+
+       liteuart_rx_chars(port);
 
        mod_timer(&uart->timer, jiffies + uart_poll_timeout(port));
 }

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

#27 lore raw

PATCH v5 11/14 serial: liteuart: move function definitions

In reply to Gabriel Somlo

Move definitions for liteuart_[stop|start]_tx(), liteuart_stop_rx(), and liteuart_putchar() to a more convenient location in preparation for adding IRQ support. This patch contains no functional changes.

---
 drivers/tty/serial/liteuart.c | 82 +++++++++++++++++------------------
 1 file changed, 41 insertions(+), 41 deletions(-)
drivers/tty/serial/liteuart.c +41 -41
index 76f8a09b82cd..8a6e176be08e 100644
@@ -68,6 +68,47 @@ static struct uart_driver liteuart_driver = {
 #endif
 };
 
+static void liteuart_putchar(struct uart_port *port, unsigned char ch)
+{
+	while (litex_read8(port->membase + OFF_TXFULL))
+		cpu_relax();
+
+	litex_write8(port->membase + OFF_RXTX, ch);
+}
+
+static void liteuart_stop_tx(struct uart_port *port)
+{
+}
+
+static void liteuart_start_tx(struct uart_port *port)
+{
+	struct circ_buf *xmit = &port->state->xmit;
+	unsigned char ch;
+
+	if (unlikely(port->x_char)) {
+		litex_write8(port->membase + OFF_RXTX, port->x_char);
+		port->icount.tx++;
+		port->x_char = 0;
+	} else if (!uart_circ_empty(xmit)) {
+		while (xmit->head != xmit->tail) {
+			ch = xmit->buf[xmit->tail];
+			uart_xmit_advance(port, 1);
+			liteuart_putchar(port, ch);
+		}
+	}
+
+	if (uart_circ_chars_pending(xmit) < WAKEUP_CHARS)
+		uart_write_wakeup(port);
+}
+
+static void liteuart_stop_rx(struct uart_port *port)
+{
+	struct liteuart_port *uart = to_liteuart_port(port);
+
+	/* just delete timer */
+	del_timer(&uart->timer);
+}
+
 static void liteuart_rx_chars(struct uart_port *port)
 {
 	unsigned char __iomem *membase = port->membase;
@@ -98,14 +139,6 @@ static void liteuart_timer(struct timer_list *t)
 	mod_timer(&uart->timer, jiffies + uart_poll_timeout(port));
 }
 
-static void liteuart_putchar(struct uart_port *port, unsigned char ch)
-{
-	while (litex_read8(port->membase + OFF_TXFULL))
-		cpu_relax();
-
-	litex_write8(port->membase + OFF_RXTX, ch);
-}
-
 static unsigned int liteuart_tx_empty(struct uart_port *port)
 {
 	/* not really tx empty, just checking if tx is not full */
@@ -125,39 +158,6 @@ static unsigned int liteuart_get_mctrl(struct uart_port *port)
 	return TIOCM_CTS | TIOCM_DSR | TIOCM_CAR;
 }
 
-static void liteuart_stop_tx(struct uart_port *port)
-{
-}
-
-static void liteuart_start_tx(struct uart_port *port)
-{
-	struct circ_buf *xmit = &port->state->xmit;
-	unsigned char ch;
-
-	if (unlikely(port->x_char)) {
-		litex_write8(port->membase + OFF_RXTX, port->x_char);
-		port->icount.tx++;
-		port->x_char = 0;
-	} else if (!uart_circ_empty(xmit)) {
-		while (xmit->head != xmit->tail) {
-			ch = xmit->buf[xmit->tail];
-			uart_xmit_advance(port, 1);
-			liteuart_putchar(port, ch);
-		}
-	}
-
-	if (uart_circ_chars_pending(xmit) < WAKEUP_CHARS)
-		uart_write_wakeup(port);
-}
-
-static void liteuart_stop_rx(struct uart_port *port)
-{
-	struct liteuart_port *uart = to_liteuart_port(port);
-
-	/* just delete timer */
-	del_timer(&uart->timer);
-}
-
 static int liteuart_startup(struct uart_port *port)
 {
 	struct liteuart_port *uart = to_liteuart_port(port);
Signature · 1 line
2.38.1
Reviewed-by on v5 11/14
#28 lore raw

Re: [PATCH v5 11/14] serial: liteuart: move function definitions

Quoted 6 lines · Gabriel Somlo

Move definitions for liteuart_[stop|start]_tx(), liteuart_stop_rx(), and liteuart_putchar() to a more convenient location in preparation for adding IRQ support. This patch contains no functional changes.

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

#29 lore raw

PATCH v5 12/14 serial: liteuart: add IRQ support for the RX path

In reply to Gabriel Somlo

Add support for IRQ-driven RX. Support for the TX path will be added in a separate commit.

---

Changes from v4:
- using dev_err() instead of a combo of pr_err() and pr_fmt()
- dropped "get irq" comment in probe()

Quoted 4 lines

Changes from v3:
- add shadow irq register to support polling mode and avoid reading hardware mmio irq register to learn which irq flags are enabled
- this also simplifies both liteuart_interrupt() and liteuart_startup()

 drivers/tty/serial/liteuart.c | 76 +++++++++++++++++++++++++++++++----
 1 file changed, 69 insertions(+), 7 deletions(-)
drivers/tty/serial/liteuart.c +69 -7
index 8a6e176be08e..678c37c952cf 100644
@@ -7,6 +7,7 @@
 
 #include <linux/bits.h>
 #include <linux/console.h>
+#include <linux/interrupt.h>
 #include <linux/litex.h>
 #include <linux/module.h>
 #include <linux/of.h>
@@ -46,6 +47,7 @@ struct liteuart_port {
 	struct uart_port port;
 	struct timer_list timer;
 	u32 id;
+	u8 irq_reg;
 };
 
 #define to_liteuart_port(port)	container_of(port, struct liteuart_port, port)
@@ -76,6 +78,19 @@ static void liteuart_putchar(struct uart_port *port, unsigned char ch)
 	litex_write8(port->membase + OFF_RXTX, ch);
 }
 
+static void liteuart_update_irq_reg(struct uart_port *port, bool set, u8 mask)
+{
+	struct liteuart_port *uart = to_liteuart_port(port);
+
+	if (set)
+		uart->irq_reg |= mask;
+	else
+		uart->irq_reg &= ~mask;
+
+	if (port->irq)
+		litex_write8(port->membase + OFF_EV_ENABLE, uart->irq_reg);
+}
+
 static void liteuart_stop_tx(struct uart_port *port)
 {
 }
@@ -129,13 +144,27 @@ static void liteuart_rx_chars(struct uart_port *port)
 	tty_flip_buffer_push(&port->state->port);
 }
 
+static irqreturn_t liteuart_interrupt(int irq, void *data)
+{
+	struct liteuart_port *uart = data;
+	struct uart_port *port = &uart->port;
+	u8 isr;
+
+	spin_lock(&port->lock);
+	isr = litex_read8(port->membase + OFF_EV_PENDING) & uart->irq_reg;
+	if (isr & EV_RX)
+		liteuart_rx_chars(port);
+	spin_unlock(&port->lock);
+
+	return IRQ_RETVAL(isr);
+}
+
 static void liteuart_timer(struct timer_list *t)
 {
 	struct liteuart_port *uart = from_timer(uart, t, timer);
 	struct uart_port *port = &uart->port;
 
-	liteuart_rx_chars(port);
-
+	liteuart_interrupt(0, port);
 	mod_timer(&uart->timer, jiffies + uart_poll_timeout(port));
 }
 
@@ -161,19 +190,46 @@ static unsigned int liteuart_get_mctrl(struct uart_port *port)
 static int liteuart_startup(struct uart_port *port)
 {
 	struct liteuart_port *uart = to_liteuart_port(port);
+	unsigned long flags;
+	int ret;
 
-	/* disable events */
-	litex_write8(port->membase + OFF_EV_ENABLE, 0);
+	if (port->irq) {
+		ret = request_irq(port->irq, liteuart_interrupt, 0,
+				  KBUILD_MODNAME, uart);
+		if (ret) {
+			dev_err(port->dev,
+				"line %d irq %d failed: switch to polling\n",
+				port->line, port->irq);
+			port->irq = 0;
+		}
+	}
 
-	/* prepare timer for polling */
-	timer_setup(&uart->timer, liteuart_timer, 0);
-	mod_timer(&uart->timer, jiffies + uart_poll_timeout(port));
+	spin_lock_irqsave(&port->lock, flags);
+	/* only enabling rx irqs during startup */
+	liteuart_update_irq_reg(port, true, EV_RX);
+	spin_unlock_irqrestore(&port->lock, flags);
+
+	if (!port->irq) {
+		timer_setup(&uart->timer, liteuart_timer, 0);
+		mod_timer(&uart->timer, jiffies + uart_poll_timeout(port));
+	}
 
 	return 0;
 }
 
 static void liteuart_shutdown(struct uart_port *port)
 {
+	struct liteuart_port *uart = to_liteuart_port(port);
+	unsigned long flags;
+
+	spin_lock_irqsave(&port->lock, flags);
+	liteuart_update_irq_reg(port, false, EV_RX | EV_TX);
+	spin_unlock_irqrestore(&port->lock, flags);
+
+	if (port->irq)
+		free_irq(port->irq, port);
+	else
+		del_timer_sync(&uart->timer);
 }
 
 static void liteuart_set_termios(struct uart_port *port, struct ktermios *new,
@@ -262,6 +318,12 @@ static int liteuart_probe(struct platform_device *pdev)
 		goto err_erase_id;
 	}
 
+	ret = platform_get_irq_optional(pdev, 0);
+	if (ret < 0 && ret != -ENXIO)
+		return ret;
+	if (ret > 0)
+		port->irq = ret;
+
 	/* values not from device tree */
 	port->dev = &pdev->dev;
 	port->iotype = UPIO_MEM;
Signature · 1 line
2.38.1
maintainer
#30 lore raw

Re: [PATCH v5 12/14] serial: liteuart: add IRQ support for the RX path

Quoted 76 lines · Gabriel Somlo

Add support for IRQ-driven RX. Support for the TX path will be added in a separate commit.

---

Changes from v4:
- using dev_err() instead of a combo of pr_err() and pr_fmt()
- dropped "get irq" comment in probe()

Quoted 4 lines

Changes from v3:
- add shadow irq register to support polling mode and avoid reading hardware mmio irq register to learn which irq flags are enabled
- this also simplifies both liteuart_interrupt() and liteuart_startup()

  drivers/tty/serial/liteuart.c | 76 +++++++++++++++++++++++++++++++----
  1 file changed, 69 insertions(+), 7 deletions(-)
drivers/tty/serial/liteuart.c +28 -0
index 8a6e176be08e..678c37c952cf 100644
@@ -7,6 +7,7 @@
 
  #include <linux/bits.h>
  #include <linux/console.h>
+#include <linux/interrupt.h>
  #include <linux/litex.h>
  #include <linux/module.h>
  #include <linux/of.h>
@@ -46,6 +47,7 @@ struct liteuart_port {
  	struct uart_port port;
  	struct timer_list timer;
  	u32 id;
+	u8 irq_reg;
  };
 
  #define to_liteuart_port(port)	container_of(port, struct liteuart_port, port)
@@ -76,6 +78,19 @@ static void liteuart_putchar(struct uart_port *port, unsigned char ch)
  	litex_write8(port->membase + OFF_RXTX, ch);
  }
 
+static void liteuart_update_irq_reg(struct uart_port *port, bool set, u8 mask)
+{
+	struct liteuart_port *uart = to_liteuart_port(port);
+
+	if (set)
+		uart->irq_reg |= mask;
+	else
+		uart->irq_reg &= ~mask;
+
+	if (port->irq)
+		litex_write8(port->membase + OFF_EV_ENABLE, uart->irq_reg);
+}
+
  static void liteuart_stop_tx(struct uart_port *port)
  {
  }
@@ -129,13 +144,27 @@ static void liteuart_rx_chars(struct uart_port *port)
  	tty_flip_buffer_push(&port->state->port);
  }
 
+static irqreturn_t liteuart_interrupt(int irq, void *data)
+{
+	struct liteuart_port *uart = data;
+	struct uart_port *port = &uart->port;
+	u8 isr;
+
+	spin_lock(&port->lock);
+	isr = litex_read8(port->membase + OFF_EV_PENDING) & uart->irq_reg;
+	if (isr & EV_RX)
+		liteuart_rx_chars(port);
+	spin_unlock(&port->lock);
+
+	return IRQ_RETVAL(isr);

drivers/tty/serial/liteuart.c static void liteuart_rx_chars(struct uart_port *port) · line 167

Quoted 10 lines
drivers/tty/serial/liteuart.c +3 -2
+}
+
  static void liteuart_timer(struct timer_list *t)
  {
  	struct liteuart_port *uart = from_timer(uart, t, timer);
  	struct uart_port *port = &uart->port;
 
-	liteuart_rx_chars(port);
-
+	liteuart_interrupt(0, port);

Are you sure spin_lock() is safe from this path? I assume so, but have you thought about it?

Quoted 5 lines
drivers/tty/serial/liteuart.c +0 -0
  	mod_timer(&uart->timer, jiffies + uart_poll_timeout(port));
  }
 
@@ -161,19 +190,46 @@ static unsigned int liteuart_get_mctrl(struct uart_port *port)
  static int liteuart_startup(struct uart_port *port)

drivers/tty/serial/liteuart.c static unsigned int liteuart_get_mctrl(struct uart_port *port) · line 198

Quoted 10 lines
drivers/tty/serial/liteuart.c +5 -2
  {
  	struct liteuart_port *uart = to_liteuart_port(port);
+	unsigned long flags;
+	int ret;
 
-	/* disable events */
-	litex_write8(port->membase + OFF_EV_ENABLE, 0);
+	if (port->irq) {
+		ret = request_irq(port->irq, liteuart_interrupt, 0,
+				  KBUILD_MODNAME, uart);

Just asking: cannot the irq be shared?

drivers/tty/serial/liteuart.c static unsigned int liteuart_get_mctrl(struct uart_port *port) · line 202

Quoted 4 lines
drivers/tty/serial/liteuart.c +4 -0
+		if (ret) {
+			dev_err(port->dev,
+				"line %d irq %d failed: switch to polling\n",
+				port->line, port->irq);

That is, it should be only dev_warn(), or?

drivers/tty/serial/liteuart.c static unsigned int liteuart_get_mctrl(struct uart_port *port) · line 205

Quoted 3 lines
drivers/tty/serial/liteuart.c +3 -0
+			port->irq = 0;
+		}
+	}

thanks,

Signature · 2 lines
js
suse labs
#31 lore raw

Re: [PATCH v5 12/14] serial: liteuart: add IRQ support for the RX path

Hi Jiri,

Quoted 77 lines · Jiri Slaby
Quoted 76 lines · Gabriel Somlo

Add support for IRQ-driven RX. Support for the TX path will be added in a separate commit.

---

Changes from v4:
- using dev_err() instead of a combo of pr_err() and pr_fmt()
- dropped "get irq" comment in probe()

Quoted 4 lines

Changes from v3:
- add shadow irq register to support polling mode and avoid reading hardware mmio irq register to learn which irq flags are enabled
- this also simplifies both liteuart_interrupt() and liteuart_startup()

  drivers/tty/serial/liteuart.c | 76 +++++++++++++++++++++++++++++++----
  1 file changed, 69 insertions(+), 7 deletions(-)
drivers/tty/serial/liteuart.c +30 -0
index 8a6e176be08e..678c37c952cf 100644
@@ -7,6 +7,7 @@
  #include <linux/bits.h>
  #include <linux/console.h>
+#include <linux/interrupt.h>
  #include <linux/litex.h>
  #include <linux/module.h>
  #include <linux/of.h>
@@ -46,6 +47,7 @@ struct liteuart_port {
  	struct uart_port port;
  	struct timer_list timer;
  	u32 id;
+	u8 irq_reg;
  };
  #define to_liteuart_port(port)	container_of(port, struct liteuart_port, port)
@@ -76,6 +78,19 @@ static void liteuart_putchar(struct uart_port *port, unsigned char ch)
  	litex_write8(port->membase + OFF_RXTX, ch);
  }
+static void liteuart_update_irq_reg(struct uart_port *port, bool set, u8 mask)
+{
+	struct liteuart_port *uart = to_liteuart_port(port);
+
+	if (set)
+		uart->irq_reg |= mask;
+	else
+		uart->irq_reg &= ~mask;
+
+	if (port->irq)
+		litex_write8(port->membase + OFF_EV_ENABLE, uart->irq_reg);
+}
+
  static void liteuart_stop_tx(struct uart_port *port)
  {
  }
@@ -129,13 +144,27 @@ static void liteuart_rx_chars(struct uart_port *port)
  	tty_flip_buffer_push(&port->state->port);
  }
+static irqreturn_t liteuart_interrupt(int irq, void *data)
+{
+	struct liteuart_port *uart = data;
+	struct uart_port *port = &uart->port;
+	u8 isr;
+
+	spin_lock(&port->lock);
+	isr = litex_read8(port->membase + OFF_EV_PENDING) & uart->irq_reg;
+	if (isr & EV_RX)
+		liteuart_rx_chars(port);
+	spin_unlock(&port->lock);
+
+	return IRQ_RETVAL(isr);
+}
+
  static void liteuart_timer(struct timer_list *t)
  {
Quoted 8 lines

drivers/tty/serial/liteuart.c static void liteuart_rx_chars(struct uart_port *port) · line 167

Quoted 5 lines
drivers/tty/serial/liteuart.c +1 -2
  	struct liteuart_port *uart = from_timer(uart, t, timer);
  	struct uart_port *port = &uart->port;
-	liteuart_rx_chars(port);
-
+	liteuart_interrupt(0, port);

Are you sure spin_lock() is safe from this path? I assume so, but have you thought about it?

I checked and at that point in_serving_softirq() is true.

*However*, after studying spin_lock() and friends for a while, I'm not quite clear on whether that unequivocally translates to "yes, we're safe" :)

As such, I'm inclined to switch to spin_lock_irqsave() and spin_unlock_irqrestore() even in the interrupt handler, which is explicitly stated to be "safe from any context":
https://www.kernel.org/doc/html/v4.15/kernel-hacking/locking.html#cheat-sheet-for-locking

The alternative could be to set TIMER_IRQSAFE in timer_setup(), but no other tty driver seems to be doing that, so I'd be a bit off the beaten path there... :)

Please do let me know what you think about this, particularly if you consider going the spin_lock_irqsave-everywhere-just-to-be-safe route overkill... :)

Quoted 7 lines
Quoted 7 lines
drivers/tty/serial/liteuart.c +1 -0
  	mod_timer(&uart->timer, jiffies + uart_poll_timeout(port));
  }
@@ -161,19 +190,46 @@ static unsigned int liteuart_get_mctrl(struct uart_port *port)
  static int liteuart_startup(struct uart_port *port)
  {
  	struct liteuart_port *uart = to_liteuart_port(port);
+	unsigned long flags;
Quoted 8 lines

drivers/tty/serial/liteuart.c static unsigned int liteuart_get_mctrl(struct uart_port *port) · line 198

Quoted 6 lines
drivers/tty/serial/liteuart.c +4 -2
+	int ret;
-	/* disable events */
-	litex_write8(port->membase + OFF_EV_ENABLE, 0);
+	if (port->irq) {
+		ret = request_irq(port->irq, liteuart_interrupt, 0,
+				  KBUILD_MODNAME, uart);

Just asking: cannot the irq be shared?

Given the way LiteX gateware is currently generated, each irq-triggering device is given its own separate line. I don't think setting the IRQF_SHARED flag actually *hurts* anything (no difference in behavior while testing), but I don't think it's needed ATM.

Quoted 6 lines

drivers/tty/serial/liteuart.c static unsigned int liteuart_get_mctrl(struct uart_port *port) · line 202

Quoted 4 lines
drivers/tty/serial/liteuart.c +4 -0
+		if (ret) {
+			dev_err(port->dev,
+				"line %d irq %d failed: switch to polling\n",
+				port->line, port->irq);

That is, it should be only dev_warn(), or?

Makes sense, will use dev_warn() in v6.

Please LMK what you think about spin_lock[_irqsave] (and IRQF_SHARED), and I'll send out v6 with all the necessary chances right after that.

Thanks much,
--Gabriel

Quoted 9 lines

drivers/tty/serial/liteuart.c static unsigned int liteuart_get_mctrl(struct uart_port *port) · line 205

Quoted 3 lines
drivers/tty/serial/liteuart.c +3 -0
+			port->irq = 0;
+		}
+	}

thanks,

Signature · 2 lines
js
suse labs
maintainer
#32 lore raw

Re: [PATCH v5 12/14] serial: liteuart: add IRQ support for the RX path

Quoted 11 lines · Gabriel L. Somlo
Quoted 10 lines

drivers/tty/serial/liteuart.c static void liteuart_rx_chars(struct uart_port *port) · line 167

Quoted 7 lines
   static void liteuart_timer(struct timer_list *t)
   {
   	struct liteuart_port *uart = from_timer(uart, t, timer);
   	struct uart_port *port = &uart->port;
drivers/tty/serial/liteuart.c +1 -2
-	liteuart_rx_chars(port);
-
+	liteuart_interrupt(0, port);

Are you sure spin_lock() is safe from this path? I assume so, but have you thought about it?

Quoted 5 lines

I checked and at that point in_serving_softirq() is true.

*However*, after studying spin_lock() and friends for a while, I'm not quite clear on whether that unequivocally translates to "yes, we're safe" :)

Depends whether some hard irq context is grabbing the port lock too. If it does, it will spin forever waiting for this soft irq to finish (never happens as it was interrupted).

Quoted 10 lines

As such, I'm inclined to switch to spin_lock_irqsave() and spin_unlock_irqrestore() even in the interrupt handler, which is explicitly stated to be "safe from any context":
https://www.kernel.org/doc/html/v4.15/kernel-hacking/locking.html#cheat-sheet-for-locking

The alternative could be to set TIMER_IRQSAFE in timer_setup(), but no other tty driver seems to be doing that, so I'd be a bit off the beaten path there... :)

Ah, no.

Quoted 3 lines

Please do let me know what you think about this, particularly if you consider going the spin_lock_irqsave-everywhere-just-to-be-safe route overkill... :)

If you are unsure about the other contexts, irqsave/restore is the way to go. It can be lifted later, if someone investigates harder.

thanks,

Signature · 1 line
js
#33 lore raw

Re: [PATCH v5 12/14] serial: liteuart: add IRQ support for the RX path

Quoted 33 lines · Jiri Slaby
Quoted 8 lines · Gabriel L. Somlo
Quoted 8 lines
Quoted 7 lines
   static void liteuart_timer(struct timer_list *t)
   {
    struct liteuart_port *uart = from_timer(uart, t, timer);
    struct uart_port *port = &uart->port;
drivers/tty/serial/liteuart.c +1 -2
-   liteuart_rx_chars(port);
-
+   liteuart_interrupt(0, port);
Quoted 8 lines
Quoted 2 lines

Are you sure spin_lock() is safe from this path? I assume so, but have you thought about it?

I checked and at that point in_serving_softirq() is true.

*However*, after studying spin_lock() and friends for a while, I'm not quite clear on whether that unequivocally translates to "yes, we're safe" :)

Depends whether some hard irq context is grabbing the port lock too. If it does, it will spin forever waiting for this soft irq to finish (never happens as it was interrupted).

Quoted 10 lines

As such, I'm inclined to switch to spin_lock_irqsave() and spin_unlock_irqrestore() even in the interrupt handler, which is explicitly stated to be "safe from any context":
https://www.kernel.org/doc/html/v4.15/kernel-hacking/locking.html#cheat-sheet-for-locking

The alternative could be to set TIMER_IRQSAFE in timer_setup(), but no other tty driver seems to be doing that, so I'd be a bit off the beaten path there... :)

Quoted 8 lines

Ah, no.

Quoted 3 lines

Please do let me know what you think about this, particularly if you consider going the spin_lock_irqsave-everywhere-just-to-be-safe route overkill... :)

If you are unsure about the other contexts, irqsave/restore is the way to go. It can be lifted later, if someone investigates harder.

Inside the interrupt handler, plain spin_lock() should be OK. Inside the timer handler, I think you need to use spin_lock_irqsave(), as e.g. liteuart_set_termios() (and lots of code in serial_core.c) already uses spin_lock_irqsave().
Besides, on non-SMP, spin_lock() doesn't do anything, so you need to rely on disabling the local interrupt.

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

#34 lore raw

Re: [PATCH v5 12/14] serial: liteuart: add IRQ support for the RX path

Quoted 43 lines · Geert Uytterhoeven
Quoted 33 lines · Jiri Slaby
Quoted 8 lines · Gabriel L. Somlo
Quoted 8 lines
Quoted 7 lines
   static void liteuart_timer(struct timer_list *t)
   {
    struct liteuart_port *uart = from_timer(uart, t, timer);
    struct uart_port *port = &uart->port;
drivers/tty/serial/liteuart.c +1 -2
-   liteuart_rx_chars(port);
-
+   liteuart_interrupt(0, port);
Quoted 8 lines
Quoted 2 lines

Are you sure spin_lock() is safe from this path? I assume so, but have you thought about it?

I checked and at that point in_serving_softirq() is true.

*However*, after studying spin_lock() and friends for a while, I'm not quite clear on whether that unequivocally translates to "yes, we're safe" :)

Depends whether some hard irq context is grabbing the port lock too. If it does, it will spin forever waiting for this soft irq to finish (never happens as it was interrupted).

Quoted 10 lines

As such, I'm inclined to switch to spin_lock_irqsave() and spin_unlock_irqrestore() even in the interrupt handler, which is explicitly stated to be "safe from any context":
https://www.kernel.org/doc/html/v4.15/kernel-hacking/locking.html#cheat-sheet-for-locking

The alternative could be to set TIMER_IRQSAFE in timer_setup(), but no other tty driver seems to be doing that, so I'd be a bit off the beaten path there... :)

Quoted 8 lines

Ah, no.

Quoted 3 lines

Please do let me know what you think about this, particularly if you consider going the spin_lock_irqsave-everywhere-just-to-be-safe route overkill... :)

If you are unsure about the other contexts, irqsave/restore is the way to go. It can be lifted later, if someone investigates harder.

Quoted 6 lines

Inside the interrupt handler, plain spin_lock() should be OK. Inside the timer handler, I think you need to use spin_lock_irqsave(), as e.g. liteuart_set_termios() (and lots of code in serial_core.c) already uses spin_lock_irqsave().
Besides, on non-SMP, spin_lock() doesn't do anything, so you need to rely on disabling the local interrupt.

Thanks Geert for the clarification! I could write two wrappers around the actual code doing the interrupt handler work, one with spin_lock() for the actual irq handler and another with spin_lock_irqsave() for the timer codepath. But to keep things simple I'm inclined to simply use spin_lock_irqsave() and add a comment saying it's because we need it when polling and it's not actually harmful when using IRQ.

Thanks,
--Gabriel

Quoted 10 lines

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

#35 lore raw

PATCH v5 13/14 serial: liteuart: add IRQ support for the TX path

In reply to Gabriel Somlo

Switch the TX path to IRQ-driven operation, while maintaining support for polling mode via the poll timer.

---
 drivers/tty/serial/liteuart.c | 30 +++++++++++++-----------------
 1 file changed, 13 insertions(+), 17 deletions(-)
drivers/tty/serial/liteuart.c +13 -17
index 678c37c952cf..850125870efb 100644
@@ -93,27 +93,12 @@ static void liteuart_update_irq_reg(struct uart_port *port, bool set, u8 mask)
 
 static void liteuart_stop_tx(struct uart_port *port)
 {
+	liteuart_update_irq_reg(port, false, EV_TX);
 }
 
 static void liteuart_start_tx(struct uart_port *port)
 {
-	struct circ_buf *xmit = &port->state->xmit;
-	unsigned char ch;
-
-	if (unlikely(port->x_char)) {
-		litex_write8(port->membase + OFF_RXTX, port->x_char);
-		port->icount.tx++;
-		port->x_char = 0;
-	} else if (!uart_circ_empty(xmit)) {
-		while (xmit->head != xmit->tail) {
-			ch = xmit->buf[xmit->tail];
-			uart_xmit_advance(port, 1);
-			liteuart_putchar(port, ch);
-		}
-	}
-
-	if (uart_circ_chars_pending(xmit) < WAKEUP_CHARS)
-		uart_write_wakeup(port);
+	liteuart_update_irq_reg(port, true, EV_TX);
 }
 
 static void liteuart_stop_rx(struct uart_port *port)
@@ -144,6 +129,15 @@ static void liteuart_rx_chars(struct uart_port *port)
 	tty_flip_buffer_push(&port->state->port);
 }
 
+static void liteuart_tx_chars(struct uart_port *port)
+{
+	u8 ch;
+
+	uart_port_tx(port, ch,
+		!litex_read8(port->membase + OFF_TXFULL),
+		litex_write8(port->membase + OFF_RXTX, ch));
+}
+
 static irqreturn_t liteuart_interrupt(int irq, void *data)
 {
 	struct liteuart_port *uart = data;
@@ -154,6 +148,8 @@ static irqreturn_t liteuart_interrupt(int irq, void *data)
 	isr = litex_read8(port->membase + OFF_EV_PENDING) & uart->irq_reg;
 	if (isr & EV_RX)
 		liteuart_rx_chars(port);
+	if (isr & EV_TX)
+		liteuart_tx_chars(port);
 	spin_unlock(&port->lock);
 
 	return IRQ_RETVAL(isr);
Signature · 1 line
2.38.1
maintainer
#36 lore raw

Re: [PATCH v5 13/14] serial: liteuart: add IRQ support for the TX path

Quoted 5 lines · Gabriel Somlo

Switch the TX path to IRQ-driven operation, while maintaining support for polling mode via the poll timer.

...

drivers/tty/serial/liteuart.c static irqreturn_t liteuart_interrupt(int irq, void *data) · line 152

Quoted 6 lines
drivers/tty/serial/liteuart.c +2 -0
@@ -154,6 +148,8 @@ static irqreturn_t liteuart_interrupt(int irq, void *data)
  	isr = litex_read8(port->membase + OFF_EV_PENDING) & uart->irq_reg;
  	if (isr & EV_RX)
  		liteuart_rx_chars(port);
+	if (isr & EV_TX)
+		liteuart_tx_chars(port);

Wait, how do you ensure the OFF_EV_PENDING reg contains EV_RX and/or EV_TX in the polling mode?

thanks,

Signature · 2 lines
js
suse labs
#37 lore raw

Re: [PATCH v5 13/14] serial: liteuart: add IRQ support for the TX path

Quoted 7 lines · Jiri Slaby
Quoted 5 lines · Gabriel Somlo

Switch the TX path to IRQ-driven operation, while maintaining support for polling mode via the poll timer.

...

Quoted 9 lines

drivers/tty/serial/liteuart.c static irqreturn_t liteuart_interrupt(int irq, void *data) · line 152

Quoted 6 lines
drivers/tty/serial/liteuart.c +2 -0
@@ -154,6 +148,8 @@ static irqreturn_t liteuart_interrupt(int irq, void *data)
  	isr = litex_read8(port->membase + OFF_EV_PENDING) & uart->irq_reg;
  	if (isr & EV_RX)
  		liteuart_rx_chars(port);
+	if (isr & EV_TX)
+		liteuart_tx_chars(port);

Wait, how do you ensure the OFF_EV_PENDING reg contains EV_RX and/or EV_TX in the polling mode?

The hardware (well, *gateware*) is coded to populate the EV_PENDING register regardless of whether IRQs are enabled via the EV_ENABLE register (or indeed, whether IRQs are even "wired into" the design at all). See

https://github.com/enjoy-digital/litex/blob/master/litex/soc/cores/uart.py
and
https://github.com/enjoy-digital/litex/blob/master/litex/soc/interconnect/csr_eventmanager.py

for a starting point on why it's OK to just assume that...

Thanks,
--Gabriel

#38 lore raw

PATCH v5 14/14 serial: liteuart: move polling putchar() function

In reply to Gabriel Somlo

The polling liteuart_putchar() function is only called from methods conditionally enabled by CONFIG_SERIAL_LITEUART_CONSOLE. Move its definition closer to the console code where it is dependent on the same config option.

---
 drivers/tty/serial/liteuart.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)
drivers/tty/serial/liteuart.c +8 -8
index 850125870efb..47d12bb1b9e0 100644
@@ -70,14 +70,6 @@ static struct uart_driver liteuart_driver = {
 #endif
 };
 
-static void liteuart_putchar(struct uart_port *port, unsigned char ch)
-{
-	while (litex_read8(port->membase + OFF_TXFULL))
-		cpu_relax();
-
-	litex_write8(port->membase + OFF_RXTX, ch);
-}
-
 static void liteuart_update_irq_reg(struct uart_port *port, bool set, u8 mask)
 {
 	struct liteuart_port *uart = to_liteuart_port(port);
@@ -372,6 +364,14 @@ static struct platform_driver liteuart_platform_driver = {
 
 #ifdef CONFIG_SERIAL_LITEUART_CONSOLE
 
+static void liteuart_putchar(struct uart_port *port, unsigned char ch)
+{
+	while (litex_read8(port->membase + OFF_TXFULL))
+		cpu_relax();
+
+	litex_write8(port->membase + OFF_RXTX, ch);
+}
+
 static void liteuart_console_write(struct console *co, const char *s,
 	unsigned int count)
 {
Signature · 1 line
2.38.1