Skip to contentSkip to conversation

param: add dev_add_param_tristate(_ro) helpers

PATCH In review Ahmad Fatoum 3 participants 7 messages Nov 21, 2019 – Dec 5, 2019

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/3 param: add dev_add_param_tristate(_ro) helpers

This is can be considered an extension to the dev_add_param_bool interfaces with a third value that's "unknown".
This can be used for cases, where barebox can flip a bit somewhere, but it has no way of knowing what the initial state of the bit was, e.g. turn on/off the watchdog, but no watchdog status. Turn on/off a co-processor, but no co-processor online status. And so on. Not providing a way to customize the "unknown" string is a deliberate choice, so future device parameters follow the same naming scheme.

---
 include/param.h | 24 ++++++++++++++++++++++++
 lib/parameter.c | 22 ++++++++++++++++++++++
 2 files changed, 46 insertions(+)
include/param.h +24 -0
index 4ac502e726c0..d75f50ea3e60 100644
@@ -63,6 +63,16 @@ struct param_d *dev_add_param_enum(struct device_d *dev, const char *name,
 		int (*get)(struct param_d *p, void *priv),
 		int *value, const char * const *names, int max, void *priv);
 
+enum param_tristate { PARAM_TRISTATE_UNKNOWN, PARAM_TRISTATE_TRUE, PARAM_TRISTATE_FALSE };
+
+struct param_d *dev_add_param_tristate(struct device_d *dev, const char *name,
+		int (*set)(struct param_d *p, void *priv),
+		int (*get)(struct param_d *p, void *priv),
+		int *value, void *priv);
+
+struct param_d *dev_add_param_tristate_ro(struct device_d *dev, const char *name,
+		int *value);
+
 struct param_d *dev_add_param_bitmask(struct device_d *dev, const char *name,
 		int (*set)(struct param_d *p, void *priv),
 		int (*get)(struct param_d *p, void *priv),
@@ -144,6 +154,20 @@ static inline struct param_d *dev_add_param_bitmask(struct device_d *dev, const
 	return ERR_PTR(-ENOSYS);
 }
 
+static inline struct param_d *dev_add_param_tristate(struct device_d *dev, const char *name,
+		int (*set)(struct param_d *p, void *priv),
+		int (*get)(struct param_d *p, void *priv),
+		int *value, void *priv)
+{
+	return ERR_PTR(-ENOSYS);
+}
+
+static inline struct param_d *dev_add_param_tristate_ro(struct device_d *dev, const char *name,
+		int *value)
+{
+	return ERR_PTR(-ENOSYS);
+}
+
 static inline struct param_d *dev_add_param_ip(struct device_d *dev, const char *name,
 		int (*set)(struct param_d *p, void *priv),
 		int (*get)(struct param_d *p, void *priv),
lib/parameter.c +22 -0
index fdbb2e71d15d..22695634e5f1 100644
@@ -588,6 +588,28 @@ struct param_d *dev_add_param_enum(struct device_d *dev, const char *name,
 	return &pe->param;
 }
 
+static const char *const tristate_names[] = {
+	[PARAM_TRISTATE_UNKNOWN] = "unknown",
+	[PARAM_TRISTATE_TRUE] = "1",
+	[PARAM_TRISTATE_FALSE] = "0",
+};
+
+struct param_d *dev_add_param_tristate(struct device_d *dev, const char *name,
+		int (*set)(struct param_d *p, void *priv),
+		int (*get)(struct param_d *p, void *priv),
+		int *value, void *priv)
+{
+	return dev_add_param_enum(dev, name, set, get, value, tristate_names,
+				  ARRAY_SIZE(tristate_names), priv);
+}
+
+struct param_d *dev_add_param_tristate_ro(struct device_d *dev, const char *name,
+		int *value)
+{
+	return dev_add_param_enum_ro(dev, name, value, tristate_names,
+				     ARRAY_SIZE(tristate_names));
+}
+
 struct param_bitmask {
 	struct param_d param;
 	unsigned long *value;
Signature · 7 lines
2.20.1


_______________________________________________
barebox mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/barebox
#1 lore raw

PATCH 2/3 watchdog: core: use new dev_add_param_tristate helper for .running param

Previous commit added a dev_add_param_tristate_ro that can be readily used instead of the enum parameter here. Use it.
This also fixes the issue that running_names had external linkage.

---
 drivers/watchdog/wd_core.c | 9 +--------
 include/watchdog.h         | 5 ++++-
 2 files changed, 5 insertions(+), 9 deletions(-)
drivers/watchdog/wd_core.c +1 -8
index fcead1175558..b6e2a37b1f0c 100644
@@ -152,12 +152,6 @@ static unsigned int dev_get_watchdog_priority(struct device_d *dev)
 	return priority;
 }
 
-const char *running_names[] = {
-	[WDOG_HW_RUNNING_UNSUPPORTED] = "unknown",
-	[WDOG_HW_RUNNING] = "1",
-	[WDOG_HW_NOT_RUNNING] = "0",
-};
-
 int watchdog_register(struct watchdog *wd)
 {
 	struct param_d *p;
@@ -176,8 +170,7 @@ int watchdog_register(struct watchdog *wd)
 	if (ret)
 		return ret;
 
-	p = dev_add_param_enum_ro(&wd->dev, "running", &wd->running,
-				  running_names, ARRAY_SIZE(running_names));
+	p = dev_add_param_tristate_ro(&wd->dev, "running", &wd->running);
 	if (IS_ERR(p))
 		return PTR_ERR(p);
 
include/watchdog.h +4 -1
index 5790205a487b..9741570ce229 100644
@@ -15,9 +15,12 @@
 
 #include <poller.h>
 #include <driver.h>
+#include <param.h>
 
 enum wdog_hw_runnning {
-	 WDOG_HW_RUNNING_UNSUPPORTED, WDOG_HW_RUNNING, WDOG_HW_NOT_RUNNING
+	 WDOG_HW_RUNNING_UNSUPPORTED = PARAM_TRISTATE_UNKNOWN,
+	 WDOG_HW_RUNNING = PARAM_TRISTATE_TRUE,
+	 WDOG_HW_NOT_RUNNING = PARAM_TRISTATE_FALSE
 };
 
 struct watchdog {
Signature · 7 lines
2.20.1


_______________________________________________
barebox mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/barebox
#2 lore raw

PATCH 3/3 remoteproc: add .stop device parameter for stopping remote processor

In reply to Ahmad Fatoum

Both the STM32 and i.MX7 remote proc drivers populate the .stop member in the struct rproc, but it's not used anywhere. The firmware API is not really fitting to 'unload' firmware. Add instead a device parameter to stop a remote processor, e.g. remoteproc0.stop=1. This is similar to the probe command used with MMCs.

---
 drivers/remoteproc/remoteproc_core.c | 30 +++++++++++++++++++++++-----
 include/linux/remoteproc.h           |  2 ++
 2 files changed, 27 insertions(+), 5 deletions(-)
drivers/remoteproc/remoteproc_core.c +25 -5
index 8a28c1bafc1b..e031640bc7a0 100644
@@ -101,6 +101,8 @@ static int rproc_firmware_finish(struct firmware_handler *fh)
 	fw.size = rproc->fw_buf_ofs;
 
 	ret = rproc_start(rproc, &fw);
+	if (ret == 0)
+		rproc->stop = PARAM_TRISTATE_FALSE;
 
 	kfree(rproc->fw_buf);
 
@@ -120,6 +122,19 @@ static int rproc_register_dev(struct rproc *rproc, const char *alias)
 	return register_device(&rproc->dev);
 }
 
+static int rproc_set_stop(struct param_d *param, void *priv)
+{
+	struct rproc *rproc = priv;
+	int (*stop)(struct rproc *);
+
+	stop = rproc->ops->stop;
+
+	if (!stop)
+		return -ENOSYS;
+
+	return stop(rproc);
+}
+
 int rproc_add(struct rproc *rproc)
 {
 	struct device_d *dev = &rproc->dev;
@@ -142,12 +157,17 @@ int rproc_add(struct rproc *rproc)
 	fh->close = rproc_firmware_finish;
 
 	ret = firmwaremgr_register(fh);
-	if (ret)
-		dev_err(dev, "filed to register firmware handler %s\n", rproc->name);
-	else
-		dev_info(dev, "%s is available\n", rproc->name);
+	if (ret) {
+		dev_err(dev, "failed to register firmware handler %s\n", rproc->name);
+		return ret;
+	}
 
-	return ret;
+	rproc->stop = PARAM_TRISTATE_UNKNOWN;
+	dev_add_param_tristate(&rproc->dev, "stop", rproc_set_stop, NULL,
+			   &rproc->stop, rproc);
+
+	dev_info(dev, "%s is available\n", rproc->name);
+	return 0;
 }
 
 struct rproc *rproc_alloc(struct device_d *dev, const char *name,
include/linux/remoteproc.h +2 -0
index c6264d1c0a49..af35837fb39a 100644
@@ -41,6 +41,8 @@ struct rproc {
 
 	void *fw_buf;
 	size_t fw_buf_ofs;
+
+	int stop;
 };
 
 struct rproc *rproc_alloc(struct device_d *dev, const char *name,
Signature · 7 lines
2.20.1


_______________________________________________
barebox mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/barebox
#3 lore raw

Re: [PATCH 3/3] remoteproc: add .stop device parameter for stopping remote processor

Quoted 2 lines · Ahmad Fatoum

Both the STM32 and i.MX7 remote proc drivers populate the .stop member in the struct rproc, but it's not used anywhere.

The .stop member in struct rproc is introduced in this patch.

Quoted 16 lines

The firmware API is not
really fitting to 'unload' firmware. Add instead a device parameter to stop a remote processor, e.g. remoteproc0.stop=1. This is similar to the probe command used with MMCs.

---
 drivers/remoteproc/remoteproc_core.c | 30 +++++++++++++++++++++++-----
 include/linux/remoteproc.h           |  2 ++
 2 files changed, 27 insertions(+), 5 deletions(-)
drivers/remoteproc/remoteproc_core.c +0 -0
index 8a28c1bafc1b..e031640bc7a0 100644

drivers/remoteproc/remoteproc_core.c static int rproc_firmware_finish(struct firmware_handler *fh) · line 105

Quoted 5 lines
drivers/remoteproc/remoteproc_core.c +2 -0
@@ -101,6 +101,8 @@ static int rproc_firmware_finish(struct firmware_handler *fh)
 	fw.size = rproc->fw_buf_ofs;
 
 	ret = rproc_start(rproc, &fw);
+	if (ret == 0)
+		rproc->stop = PARAM_TRISTATE_FALSE;

Can we use positive logic here? "Status Stopped is false" is harder to read than just "running" or "started".

Quoted 8 lines
drivers/remoteproc/remoteproc_core.c +2 -0
 	kfree(rproc->fw_buf);
 
@@ -120,6 +122,19 @@ static int rproc_register_dev(struct rproc *rproc, const char *alias)
 	return register_device(&rproc->dev);
 }
 
+static int rproc_set_stop(struct param_d *param, void *priv)
+{

drivers/remoteproc/remoteproc_core.c static int rproc_register_dev(struct rproc *rproc, const char *alias) · line 136

Quoted 10 lines
drivers/remoteproc/remoteproc_core.c +10 -0
+	struct rproc *rproc = priv;
+	int (*stop)(struct rproc *);
+
+	stop = rproc->ops->stop;
+
+	if (!stop)
+		return -ENOSYS;
+
+	return stop(rproc);
+}

I would assume that when I can stop the remote processor with this parameter I should be able to start it here as well, no?

Sascha

Signature · 9 lines
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/barebox
#4 lore raw

Re: [PATCH 3/3] remoteproc: add .stop device parameter for stopping remote processor

Quoted 5 lines · Sascha Hauer
Quoted 2 lines · Ahmad Fatoum

Both the STM32 and i.MX7 remote proc drivers populate the .stop member in the struct rproc, but it's not used anywhere.

The .stop member in struct rproc is introduced in this patch.

Indeed. I was referring to the stop member in the ops struct, which is a so-far unused function pointer.

Quoted 6 lines

drivers/remoteproc/remoteproc_core.c static int rproc_firmware_finish(struct firmware_handler *fh) · line 105

Quoted 3 lines
drivers/remoteproc/remoteproc_core.c +2 -0
 	ret = rproc_start(rproc, &fw);
+	if (ret == 0)
+		rproc->stop = PARAM_TRISTATE_FALSE;

Can we use positive logic here? "Status Stopped is false" is harder to read than just "running" or "started".

Naming it .stop emphasizes the fact that it's only meant to stop execution, not start it. See below.

Quoted 5 lines

drivers/remoteproc/remoteproc_core.c static int rproc_register_dev(struct rproc *rproc, const char *alias) · line 136

Quoted 2 lines
drivers/remoteproc/remoteproc_core.c +2 -0
+	return stop(rproc);
+}

I would assume that when I can stop the remote processor with this parameter I should be able to start it here as well, no?

Which firmware would the processor execute when started via parameter? I see no benefit in powering up the co-processor without specifying a firmware image.

Thoughts?
Ahmad

Signature · 9 lines
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | https://www.pengutronix.de/ |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/barebox
#5 lore raw

Re: [PATCH 3/3] remoteproc: add .stop device parameter for stopping remote processor

In reply to Sascha Hauer

Hello Sascha,

Quoted 2 lines · Sascha Hauer

I would assume that when I can stop the remote processor with this parameter I should be able to start it here as well, no?

I've yet to think some more about this, but could you merge the first two commits now?

Cheers
Ahmad

Signature · 9 lines
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | https://www.pengutronix.de/ |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/barebox
#6 lore raw

Re: [PATCH 3/3] remoteproc: add .stop device parameter for stopping remote processor

Quoted 8 lines · Ahmad Fatoum

Hello Sascha,

Quoted 2 lines · Sascha Hauer

I would assume that when I can stop the remote processor with this parameter I should be able to start it here as well, no?

I've yet to think some more about this, but could you merge the first two commits now?

Did that.

Sascha

Signature · 9 lines
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/barebox