Skip to contentSkip to conversation

eXcite nand flash driver

PATCH In review Thomas Koeller 5 participants 8 messages Feb 8, 2007 – Feb 11, 2007

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

PATCH 1/1 eXcite nand flash driver

This is a nand flash driver for the eXcite series of intelligent cameras manufactured by Basler Vision Technologies AG.

---
 drivers/mtd/nand/Kconfig            |   18 +++-
 drivers/mtd/nand/Makefile           |    1 +
 drivers/mtd/nand/excite_nandflash.c |  259 
+++++++++++++++++++++++++++++++++++
 3 files changed, 277 insertions(+), 1 deletions(-)
drivers/mtd/nand/Kconfig +17 -1
index 358f55a..5b50396 100644
@@ -216,10 +216,26 @@ config MTD_NAND_DISKONCHIP_BBTWRITE
 	  Even if you leave this disabled, you can enable BBT writes at module
 	  load time (assuming you build diskonchip as a module) with the module
 	  parameter "inftl_bbt_write=1".
-
+	  
 config MTD_NAND_SHARPSL
 	tristate "Support for NAND Flash on Sharp SL Series (C7xx + others)"
 	depends on MTD_NAND && ARCH_PXA
+ 
+config MTD_NAND_BASLER_EXCITE
+	tristate  "Support for NAND Flash on Basler eXcite"
+	depends on MTD_NAND && BASLER_EXCITE
+	help
+          This enables the driver for the NAND flash device found on the
+          Basler eXcite Smart Camera. If built as a module, the driver
+	  will be named "excite_nandflash.ko".
+
+config MTD_NAND_BASLER_EXCITE
+	tristate  "Support for NAND Flash on Basler eXcite"
+	depends on MTD_NAND && BASLER_EXCITE
+	help
+          This enables the driver for the NAND flash device found on the
+          Basler eXcite Smart Camera. If built as a module, the driver
+	  will be named "excite_nandflash.ko".
 
 config MTD_NAND_CAFE
        tristate "NAND support for OLPC CAF� chip"
drivers/mtd/nand/Makefile +1 -0
index f7a53f0..80f1dfc 100644
@@ -24,6 +24,7 @@ obj-$(CONFIG_MTD_NAND_NANDSIM)		+= nands
 obj-$(CONFIG_MTD_NAND_CS553X)		+= cs553x_nand.o
 obj-$(CONFIG_MTD_NAND_NDFC)		+= ndfc.o
 obj-$(CONFIG_MTD_NAND_AT91)		+= at91_nand.o
+obj-$(CONFIG_MTD_NAND_BASLER_EXCITE)	+= excite_nandflash.o
 
 nand-objs := nand_base.o nand_bbt.o
 cafe_nand-objs := cafe.o cafe_ecc.o
(quoted diff) +0 -0

b/drivers/mtd/nand/excite_nandflash.c
new file mode 100644
index 0000000..d683659

drivers/mtd/nand/excite_nandflash.c new file +259 -0

259 lines not on this page · raw message

Signature · 1 line
1.4.3
#1 lore raw

Just some cosmetic notes ...

drivers/mtd/nand/Kconfig config MTD_NAND_DISKONCHIP_BBTWRITE · line 219

Quoted 6 lines · Thomas Koeller
drivers/mtd/nand/Kconfig +1 -1
@@ -216,10 +216,26 @@ config MTD_NAND_DISKONCHIP_BBTWRITE
 	  Even if you leave this disabled, you can enable BBT writes at module
 	  load time (assuming you build diskonchip as a module) with the module
 	  parameter "inftl_bbt_write=1".
-
+

This just inserts whitespace.

Review config MTD_NAND_DISKONCHIP_BBTWRITE · line 220

Quoted 1 line
(quoted diff) +0 -0
        tristate "NAND support for OLPC CAFÉ chip"

The accent on the E might cause problems, but I'm not sure.

drivers/mtd/nand/excite_nandflash.c line 51

Quoted 2 lines
drivers/mtd/nand/excite_nandflash.c new file +2 -0
+#define io_readb(__a__)		__raw_readb((__a__))
+#define io_writeb(__v__, __a__)	__raw_writeb((__v__), (__a__))

I would have used an inline function. Also, aren't there functions to use IOs already? What's the reason here?

drivers/mtd/nand/excite_nandflash.c line 56

Quoted 2 lines
drivers/mtd/nand/excite_nandflash.c new file +2 -0
+/* prefix for debug output */
+static const char module_id[] = "excite_nandflash";

I'd have used

  static const char* module_id = "excite_nandflash";

except if you need the sizeof it to capture the length.

drivers/mtd/nand/excite_nandflash.c line 71

Quoted 3 lines
drivers/mtd/nand/excite_nandflash.c new file +3 -0
+static inline const struct resource *excite_nand_get_resource
+    (struct platform_device *d, unsigned long flags, const char *basename){
+	const char fmt[] = "%s_%u";

Here, too, is no need for a local buffer, just use a pointer.

drivers/mtd/nand/excite_nandflash.c

Quoted 4 lines
drivers/mtd/nand/excite_nandflash.c new file +4 -0
+/* excite_nand_probe
+ *
+ * called by device layer when it finds a device matching
+ * one our driver can handled. [...]

"... when it finds a device our driver can handle."

Otherwise I'd say it looks very clean, better than quite some other stuff I have seen.

Uli

**************************************************************************************
Visit our website at <http://www.satorlaser.de/>
**************************************************************************************
Diese E-Mail einschließlich sämtlicher Anhänge ist nur für den Adressaten bestimmt und kann vertrauliche Informationen enthalten. Bitte benachrichtigen Sie den Absender umgehend, falls Sie nicht der beabsichtigte Empfänger sein sollten. Die E-Mail ist in diesem Fall zu löschen und darf weder gelesen, weitergeleitet, veröffentlicht oder anderweitig benutzt werden. E-Mails können durch Dritte gelesen werden und Viren sowie nichtautorisierte Änderungen enthalten. Sator Laser GmbH ist für diese Folgen nicht verantwortlich.

**************************************************************************************

#2 lore raw
Quoted 11 lines · Ulrich Eckhardt

Just some cosmetic notes ...

drivers/mtd/nand/Kconfig config MTD_NAND_DISKONCHIP_BBTWRITE · line 219

Quoted 6 lines · Thomas Koeller
drivers/mtd/nand/Kconfig +1 -1
@@ -216,10 +216,26 @@ config MTD_NAND_DISKONCHIP_BBTWRITE
 	  Even if you leave this disabled, you can enable BBT writes at module
 	  load time (assuming you build diskonchip as a module) with the module
 	  parameter "inftl_bbt_write=1".
-
+

This just inserts whitespace.

No, it actually _removes_ whitspace.

Quoted 4 lines

Review config MTD_NAND_DISKONCHIP_BBTWRITE · line 220

Quoted 2 lines
(quoted diff) +0 -0
        tristate "NAND support for OLPC CAFÉ chip"

The accent on the E might cause problems, but I'm not sure.

This has been in there before.

Quoted 7 lines

drivers/mtd/nand/excite_nandflash.c line 51

Quoted 3 lines
drivers/mtd/nand/excite_nandflash.c new file +2 -0
+#define io_readb(__a__)		__raw_readb((__a__))
+#define io_writeb(__v__, __a__)	__raw_writeb((__v__), (__a__))

I would have used an inline function. Also, aren't there functions to use IOs already? What's the reason here?

Quoted 8 lines

drivers/mtd/nand/excite_nandflash.c line 56

Quoted 2 lines
drivers/mtd/nand/excite_nandflash.c new file +2 -0
+/* prefix for debug output */
+static const char module_id[] = "excite_nandflash";

I'd have used

  static const char* module_id = "excite_nandflash";

except if you need the sizeof it to capture the length.

Then I'd have used

	static const char * const module_id = "excite_nandflash";

but what is the advantage of doing so?

Quoted 5 lines

Review config MTD_NAND_DISKONCHIP_BBTWRITE · line 227

Quoted 4 lines
(quoted diff) +2 -0
+static inline const struct resource *excite_nand_get_resource
+    (struct platform_device *d, unsigned long flags, const char
 *basename){ +	const char fmt[] = "%s_%u";
Quoted 8 lines

Here, too, is no need for a local buffer, just use a pointer.

drivers/mtd/nand/excite_nandflash.c

Quoted 4 lines
drivers/mtd/nand/excite_nandflash.c new file +4 -0
+/* excite_nand_probe
+ *
+ * called by device layer when it finds a device matching
+ * one our driver can handled. [...]

"... when it finds a device our driver can handle."

Right.

Quoted 6 lines

Otherwise I'd say it looks very clean, better than quite some other stuff I have seen.

Uli

Thanks for your comments!

_______________________________

Thomas Köller, Software Developer

Basler Vision Technologies
An der Strusbek 60-62
22926 Ahrensburg
Germany

Tel. +49 (0) 4102 - 463 390
Fax +49 (0) 4102 - 463 390

mailto:[email protected]
http://www.baslerweb.com

Vorstand: Dr.-Ing. Dietmar Ley (Vorsitzender) · John P. Jennings · Peter Krumhoff · Aufsichtsratsvorsitzender: Norbert Basler Basler AG · Amtsgericht Ahrensburg HRB 4090 · Ust-IdNr.: DE 135 098 121 · Steuer-Nr.: 30 292 04497

_______________________________

#3 lore raw
In reply to Thomas Koeller
Quoted 20 lines · Thomas Koeller
drivers/mtd/nand/Kconfig +7 -1
index 358f55a..5b50396 100644
@@ -216,10 +216,26 @@ config MTD_NAND_DISKONCHIP_BBTWRITE
 	  Even if you leave this disabled, you can enable BBT writes at module
 	  load time (assuming you build diskonchip as a module) with the module
 	  parameter "inftl_bbt_write=1".
-
+	  
 config MTD_NAND_SHARPSL
 	tristate "Support for NAND Flash on Sharp SL Series (C7xx + others)"
 	depends on MTD_NAND && ARCH_PXA
+ 
+config MTD_NAND_BASLER_EXCITE
+	tristate  "Support for NAND Flash on Basler eXcite"
+	depends on MTD_NAND && BASLER_EXCITE
+	help
+          This enables the driver for the NAND flash device found on the

drivers/mtd/nand/Kconfig config MTD_NAND_DISKONCHIP_BBTWRITE · line 238

Quoted 10 lines
drivers/mtd/nand/Kconfig +10 -0
+          Basler eXcite Smart Camera. If built as a module, the driver
+	  will be named "excite_nandflash.ko".
+
+config MTD_NAND_BASLER_EXCITE
+	tristate  "Support for NAND Flash on Basler eXcite"
+	depends on MTD_NAND && BASLER_EXCITE
+	help
+          This enables the driver for the NAND flash device found on the
+          Basler eXcite Smart Camera. If built as a module, the driver
+	  will be named "excite_nandflash.ko".

You have the same config option twice... Cut and paste error?

Quoted 6 lines
drivers/mtd/nand/excite_nandflash.c new file +0 -0
new file mode 100644
index 0000000..d683659

<snip>

drivers/mtd/nand/excite_nandflash.c line 51

Quoted 3 lines
drivers/mtd/nand/excite_nandflash.c new file +3 -0
+
+#define io_readb(__a__)		__raw_readb((__a__))
+#define io_writeb(__v__, __a__)	__raw_writeb((__v__), (__a__))

Do you really need these defines? Why can't you call __raw_{readb,writeb} directly?

drivers/mtd/nand/excite_nandflash.c

Quoted 2 lines
drivers/mtd/nand/excite_nandflash.c new file +2 -0
+
+typedef void __iomem *io_reg_t;

Ugh... typdef?

<snip>

drivers/mtd/nand/excite_nandflash.c line 87

Quoted 8 lines
drivers/mtd/nand/excite_nandflash.c new file +8 -0
+static inline io_reg_t
+excite_nand_map_regs(struct platform_device *d, const char *basename)
+{
+	void *result = NULL;
+	const struct resource *const r =
+	    excite_nand_get_resource(d, IORESOURCE_MEM, basename);
+	if (likely(r))
+		result = ioremap_nocache(r->start, r->end + 1 - r->start);

Does this likely really buy you anything? I would think doing the converse (if any sort of *likely at all) would be better.

<snip>

Quoted 13 lines

+ */

drivers/mtd/nand/excite_nandflash.c new file +12 -0
+static int __exit excite_nand_remove(struct device *dev)
+{
+	struct excite_nand_drvdata * const this = dev_get_drvdata(dev);
+
+	dev_set_drvdata(dev, NULL);
+
+	if (unlikely(!this)) {
+		printk(KERN_ERR "%s: called %s without private data!!",
+		       module_id, __func__);
+		return -EINVAL;
+	}
+

drivers/mtd/nand/excite_nandflash.c line 160

Quoted 10 lines
drivers/mtd/nand/excite_nandflash.c new file +10 -0
+	/* first thing we need to do is release our mtd
+	 * then go through freeing the resource used
+	 */
+	nand_release(&this->board_mtd);
+
+	/* free the common resources */
+	if (likely(this->regs)) {
+		iounmap(this->regs);
+		this->regs = NULL;
+	}

Same likely usage comment as above.

josh

#4 lore raw
Quoted 7 lines · Josh Boyer

drivers/mtd/nand/excite_nandflash.c line 160

Quoted 5 lines
drivers/mtd/nand/excite_nandflash.c new file +5 -0
+	/* free the common resources */
+	if (likely(this->regs)) {
+		iounmap(this->regs);
+		this->regs = NULL;
+	}

Same likely usage comment as above.

I agree, this function will be called one or very few times, so this hint is not reasonable in this case.

Signature · 2 lines
Best regards,
Artem Bityutskiy (Битюцкий Артём)
#5 lore raw
In reply to Thomas Koeller

drivers/mtd/nand/excite_nandflash.c line 70

Quoted 2 lines · Thomas Koeller
drivers/mtd/nand/excite_nandflash.c new file +2 -0
+static inline const struct resource *excite_nand_get_resource
+    (struct platform_device *d, unsigned long flags, const char *basename) {

Move curly bracket to new line.

drivers/mtd/nand/excite_nandflash.c line 71

Quoted 1 line
drivers/mtd/nand/excite_nandflash.c new file +1 -0
+	const char fmt[] = "%s_%u";

Please move the format into the snprintf

drivers/mtd/nand/excite_nandflash.c line 76

Quoted 5 lines
drivers/mtd/nand/excite_nandflash.c new file +5 -0
+	char buf[80];
+
+	if (unlikely
+	    (snprintf(buf, sizeof buf, fmt, basename, d->id) >= sizeof buf))
+		return NULL;

Useless unlikely

drivers/mtd/nand/excite_nandflash.c line 85

Quoted 9 lines
drivers/mtd/nand/excite_nandflash.c new file +9 -0
+	return platform_get_resource_byname(d, flags, buf);
+}
+
+static inline io_reg_t
+excite_nand_map_regs(struct platform_device *d, const char *basename)
+{
+	void *result = NULL;
+	const struct resource *const r =
+	    excite_nand_get_resource(d, IORESOURCE_MEM, basename);

Blank line between variable declaration and code.

drivers/mtd/nand/excite_nandflash.c line 87

Quoted 2 lines
drivers/mtd/nand/excite_nandflash.c new file +2 -0
+	if (likely(r))
+		result = ioremap_nocache(r->start, r->end + 1 - r->start);

Useless likely

Quoted 13 lines
drivers/mtd/nand/excite_nandflash.c new file +11 -0
+	return result;
+}
 
 
+/* command and control functions */
+static void excite_nand_control(struct mtd_info *mtd, int cmd,
+				       unsigned int ctrl)
+{
+	io_reg_t regs =
+	    container_of(mtd, struct excite_nand_drvdata, board_mtd)->regs;
+	static void __iomem *tgt = NULL;
+
+	switch (ctrl) {

drivers/mtd/nand/excite_nandflash.c line 116

Quoted 10 lines
drivers/mtd/nand/excite_nandflash.c new file +10 -0
+	case NAND_CTRL_CHANGE | NAND_CTRL_CLE:
+		tgt = regs + EXCITE_NANDFLASH_CMD_BYTE;
+		break;
+	case NAND_CTRL_CHANGE | NAND_CTRL_ALE:
+		tgt = regs + EXCITE_NANDFLASH_ADDR_BYTE;
+		break;
+	case NAND_CTRL_CHANGE | NAND_NCE:
+		tgt = regs + EXCITE_NANDFLASH_DATA_BYTE;
+		break;
+	}

Err, did this ever work ? I doubt it. From nand_base.c:

                chip->cmd_ctrl(mtd, page_addr, ctrl);                                                                                                        
                ctrl &= ~NAND_CTRL_CHANGE;                                                                                                                   
                chip->cmd_ctrl(mtd, page_addr >> 8, ctrl);  

So I expect an OOPS happens on a regular base.

drivers/mtd/nand/excite_nandflash.c line 129

Quoted 4 lines
drivers/mtd/nand/excite_nandflash.c new file +4 -0
+static int excite_nand_devready(struct mtd_info *mtd)
+{
+	struct excite_nand_drvdata * const drvdata =
+	    container_of(mtd, struct excite_nand_drvdata, board_mtd);

Blank line missing

Quoted 18 lines
drivers/mtd/nand/excite_nandflash.c new file +18 -0
+	return io_readb(drvdata->regs + EXCITE_NANDFLASH_STATUS_BYTE);
+}
+
+/* excite_nand_remove
+ *
+ * called by device layer to remove the driver
+ * the binding to the mtd and all allocated
+ * resources are released
+ */
+static int __exit excite_nand_remove(struct device *dev)
+{
+	struct excite_nand_drvdata * const this = dev_get_drvdata(dev);
+
+	dev_set_drvdata(dev, NULL);
+
+	if (unlikely(!this)) {
+		printk(KERN_ERR "%s: called %s without private data!!",
+		       module_id, __func__);

drivers/mtd/nand/excite_nandflash.c line 157

Quoted 10 lines
drivers/mtd/nand/excite_nandflash.c new file +10 -0
+		return -EINVAL;
+	}
+
+	/* first thing we need to do is release our mtd
+	 * then go through freeing the resource used
+	 */
+	nand_release(&this->board_mtd);
+
+	/* free the common resources */
+	if (likely(this->regs)) {

Why should you ever come here with no mapping ?

tglx

#6 lore raw
Quoted 22 lines · Thomas Gleixner
Quoted 9 lines
drivers/mtd/nand/excite_nandflash.c new file +9 -0
+/* command and control functions */
+static void excite_nand_control(struct mtd_info *mtd, int cmd,
+				       unsigned int ctrl)
+{
+	io_reg_t regs =
+	    container_of(mtd, struct excite_nand_drvdata, board_mtd)->regs;
+	static void __iomem *tgt = NULL;
+
+	switch (ctrl) {

drivers/mtd/nand/excite_nandflash.c line 116

Quoted 10 lines
drivers/mtd/nand/excite_nandflash.c new file +10 -0
+	case NAND_CTRL_CHANGE | NAND_CTRL_CLE:
+		tgt = regs + EXCITE_NANDFLASH_CMD_BYTE;
+		break;
+	case NAND_CTRL_CHANGE | NAND_CTRL_ALE:
+		tgt = regs + EXCITE_NANDFLASH_ADDR_BYTE;
+		break;
+	case NAND_CTRL_CHANGE | NAND_NCE:
+		tgt = regs + EXCITE_NANDFLASH_DATA_BYTE;
+		break;
+	}

Err, did this ever work ? I doubt it. From nand_base.c:

Quoted 6 lines
                chip->cmd_ctrl(mtd, page_addr, ctrl);
                ctrl &= ~NAND_CTRL_CHANGE;
                chip->cmd_ctrl(mtd, page_addr >> 8, ctrl);

So I expect an OOPS happens on a regular base.

I guess it is the 'static void __iomem *tgt = NULL' part that worries you? Think about it, that value is never used.

However, I admit it is somewhat unclean, and therefore I am changing it. Updated patch follows.

tk

#7 lore raw
Quoted 3 lines · Thomas Koeller
Quoted 1 line

So I expect an OOPS happens on a regular base.

I guess it is the 'static void __iomem *tgt = NULL' part that worries you? Think about it, that value is never used.

True.

Quoted 2 lines

However, I admit it is somewhat unclean, and therefore I am changing it. Updated patch follows.

Yes, it is confusing and looks horrible.

tglx