Skip to contentSkip to conversation

pcmcia vs. MECR on pxa25x/sa1111

PATCH Pavel Machek 3 participants 4 messages Feb 1, 2011 – Feb 16, 2011

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

After 2.6.34 changes, __pxa2xx_drv_pcmcia_probe() was replaced by sa1111_pcmcia_add(). That unfortunately means that configure_sockets() is not called, leading to MECR not being set properly, leading to strange crashes.

Tested on pxa255+sa1111, I do not have lubbock board nearby. Perhaps cleaner solution exists?

drivers/pcmcia/pxa2xx_base.c +1 -1
@@ -204,10 +204,10 @@ pxa2xx_pcmcia_frequency_change(struct soc_pcmcia_socket *skt,
 }
 #endif
 
-static void pxa2xx_configure_sockets(struct device *dev)
+void pxa2xx_configure_sockets(struct device *dev)
 {
 	struct pcmcia_low_level *ops = dev->platform_data;
 
 	/*
 	 * We have at least one socket, so set MECR:CIT
 	 * (Card Is There)
drivers/pcmcia/pxa2xx_lubbock.c +3 -0
@@ -209,6 +209,8 @@ static struct pcmcia_low_level lubbock_pcmcia_ops = {
 
 #include "pxa2xx_base.h"
 
+extern void pxa2xx_configure_sockets(struct device *dev);
+
 int pcmcia_lubbock_init(struct sa1111_dev *sadev)
 {
 	int ret = -ENODEV;
@@ -228,6 +230,7 @@ int pcmcia_lubbock_init(struct sa1111_dev *sadev)
 		pxa2xx_drv_pcmcia_ops(&lubbock_pcmcia_ops);
 		ret = sa1111_pcmcia_add(sadev, &lubbock_pcmcia_ops,
 				pxa2xx_drv_pcmcia_add_one);
+		pxa2xx_configure_sockets(&(sadev->dev));
 	}
 
 	return ret;
#1 lore raw
Quoted 30 lines · Pavel Machek

After 2.6.34 changes, __pxa2xx_drv_pcmcia_probe() was replaced by sa1111_pcmcia_add(). That unfortunately means that configure_sockets() is not called, leading to MECR not being set properly, leading to strange crashes.

Tested on pxa255+sa1111, I do not have lubbock board nearby. Perhaps cleaner solution exists?

drivers/pcmcia/pxa2xx_base.c +1 -1
@@ -204,10 +204,10 @@ pxa2xx_pcmcia_frequency_change(struct soc_pcmcia_socket *skt,
 }
 #endif
 
-static void pxa2xx_configure_sockets(struct device *dev)
+void pxa2xx_configure_sockets(struct device *dev)
 {
 	struct pcmcia_low_level *ops = dev->platform_data;
 
 	/*
 	 * We have at least one socket, so set MECR:CIT
 	 * (Card Is There)

drivers/pcmcia/pxa2xx_lubbock.c static struct pcmcia_low_level lubbock_pcmcia_ops = { · line 213

Quoted 5 lines
drivers/pcmcia/pxa2xx_lubbock.c +2 -0
@@ -209,6 +209,8 @@ static struct pcmcia_low_level lubbock_pcmcia_ops = {
 
 #include "pxa2xx_base.h"
 
+extern void pxa2xx_configure_sockets(struct device *dev);
+

Please put function prototypes in header files.

drivers/pcmcia/pxa2xx_lubbock.c int pcmcia_lubbock_init(struct sa1111_dev *sadev) · line 233

Quoted 8 lines
drivers/pcmcia/pxa2xx_lubbock.c +1 -0
 int pcmcia_lubbock_init(struct sa1111_dev *sadev)
 {
 	int ret = -ENODEV;
@@ -228,6 +230,7 @@ int pcmcia_lubbock_init(struct sa1111_dev *sadev)
 		pxa2xx_drv_pcmcia_ops(&lubbock_pcmcia_ops);
 		ret = sa1111_pcmcia_add(sadev, &lubbock_pcmcia_ops,
 				pxa2xx_drv_pcmcia_add_one);
+		pxa2xx_configure_sockets(&(sadev->dev));

Additional parens not required.

While making this a visible function, it may also make sense to change it from taking a struct device pointer to taking the struct pcmcia_low_level pointer which is really what its after.

Lastly the ordering of pxa2xx_configure_sockets() is wrong - doing setup after adding sockets is racy. That's probably something which should fixed in the pxa2xx code too.

Signature · 3 lines
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:
#2 lore raw

After 2.6.34 changes, __pxa2xx_drv_pcmcia_probe() was replaced by sa1111_pcmcia_add(). That unfortunately means that configure_sockets() is not called, leading to MECR not being set properly, leading to strange crashes.

Tested on pxa255+sa1111, I do not have lubbock board nearby. Perhaps cleaner solution exists?

---

Here's updated version. I'd prefer not to do too extensive changes as I do not have lubbock nearby.

drivers/pcmcia/pxa2xx_base.c +1 -1
@@ -204,7 +204,7 @@ pxa2xx_pcmcia_frequency_change(struct soc_pcmcia_socket *skt,
 }
 #endif
 
-static void pxa2xx_configure_sockets(struct device *dev)
+void pxa2xx_configure_sockets(struct device *dev)
 {
 	struct pcmcia_low_level *ops = dev->platform_data;
 
drivers/pcmcia/pxa2xx_base.h +1 -0
@@ -1,3 +1,4 @@
 int pxa2xx_drv_pcmcia_add_one(struct soc_pcmcia_socket *skt);
 void pxa2xx_drv_pcmcia_ops(struct pcmcia_low_level *ops);
+void pxa2xx_configure_sockets(struct device *dev);
 
drivers/pcmcia/pxa2xx_lubbock.c +1 -0
@@ -226,6 +226,7 @@ int pcmcia_lubbock_init(struct sa1111_dev *sadev)
 		lubbock_set_misc_wr((1 << 15) | (1 << 14), 0);
 
 		pxa2xx_drv_pcmcia_ops(&lubbock_pcmcia_ops);
+		pxa2xx_configure_sockets(&sadev->dev);
 		ret = sa1111_pcmcia_add(sadev, &lubbock_pcmcia_ops,
 				pxa2xx_drv_pcmcia_add_one);
 	}
#3 lore raw

Russell,

Quoted 9 lines · Pavel Machek

After 2.6.34 changes, __pxa2xx_drv_pcmcia_probe() was replaced by sa1111_pcmcia_add(). That unfortunately means that configure_sockets() is not called, leading to MECR not being set properly, leading to strange crashes.

Tested on pxa255+sa1111, I do not have lubbock board nearby. Perhaps cleaner solution exists?

Quoted 6 lines

---

Here's updated version. I'd prefer not to do too extensive changes as I do not have lubbock nearby.

will you pick this up, or shall I?

Best,
Dominik