Skip to contentSkip to conversation

rtc-ds1302: handle write protection

PATCH In review Sergey Yanovich 3 participants 6 messages May 20, 2013 – May 31, 2013 Real time clock (RTC)

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 rtc-ds1302: handle write protection

This chip has a control register and can prevent altering saved clock. Without this patch we could have:
----8<----

(arm)root@pac14:~# date
Tue May 21 03:08:27 MSK 2013
(arm)root@pac14:~# /etc/init.d/hwclock.sh show
Tue May 21 11:13:58 2013  -0.067322 seconds
(arm)root@pac14:~# /etc/init.d/hwclock.sh stop
[info] Saving the system clock.
[info] Hardware Clock updated to Tue May 21 03:09:01 MSK 2013.
(arm)root@pac14:~# /etc/init.d/hwclock.sh show
Tue May 21 11:14:15 2013  -0.624272 seconds
----8<----
---
 drivers/rtc/rtc-ds1302.c |    6 ++++++
 1 file changed, 6 insertions(+)
drivers/rtc/rtc-ds1302.c +6 -0
index 73eafb9..626aec2 100644
@@ -23,8 +23,12 @@
 #define	RTC_CMD_READ	0x81		/* Read command */
 #define	RTC_CMD_WRITE	0x80		/* Write command */
 
+#define	RTC_CMD_WRITE_ENABLE	0x00		/* Write enable */
+#define	RTC_CMD_WRITE_DISABLE	0x80		/* Write disable */
+
 #define RTC_ADDR_RAM0	0x20		/* Address of RAM0 */
 #define RTC_ADDR_TCR	0x08		/* Address of trickle charge register */
+#define	RTC_ADDR_CTRL	0x07		/* Address of control register */
 #define	RTC_ADDR_YEAR	0x06		/* Address of year register */
 #define	RTC_ADDR_DAY	0x05		/* Address of day of week register */
 #define	RTC_ADDR_MON	0x04		/* Address of month register */
@@ -313,6 +317,7 @@ static int __init ds1302_rtc_probe(struct platform_device *pdev)
 		return PTR_ERR(rtc);
 
 	platform_set_drvdata(pdev, rtc);
+	ds1302_writebyte(RTC_ADDR_CTRL, RTC_CMD_WRITE_ENABLE);
 
 	return 0;
 }
@@ -321,6 +326,7 @@ static int ds1302_rtc_remove(struct platform_device *pdev)
 {
 	struct rtc_device *rtc = platform_get_drvdata(pdev);
 
+	ds1302_writebyte(RTC_ADDR_CTRL, RTC_CMD_WRITE_DISABLE);
 	rtc_device_unregister(rtc);
 	platform_set_drvdata(pdev, NULL);
 
Signature · 1 line
1.7.10.4
#1 lore raw
Quoted 40 lines · On Tue, 21 May 2013 03:21:30 +0400 Sergey Yanovich <[email protected]> wrote:

This chip has a control register and can prevent altering saved clock. Without this patch we could have:
----8<----

(arm)root@pac14:~# date
Tue May 21 03:08:27 MSK 2013
(arm)root@pac14:~# /etc/init.d/hwclock.sh show
Tue May 21 11:13:58 2013  -0.067322 seconds
(arm)root@pac14:~# /etc/init.d/hwclock.sh stop
[info] Saving the system clock.
[info] Hardware Clock updated to Tue May 21 03:09:01 MSK 2013.
(arm)root@pac14:~# /etc/init.d/hwclock.sh show
Tue May 21 11:14:15 2013  -0.624272 seconds
----8<----

...

drivers/rtc/rtc-ds1302.c +5 -0
@@ -23,8 +23,12 @@
 #define	RTC_CMD_READ	0x81		/* Read command */
 #define	RTC_CMD_WRITE	0x80		/* Write command */
 
+#define	RTC_CMD_WRITE_ENABLE	0x00		/* Write enable */
+#define	RTC_CMD_WRITE_DISABLE	0x80		/* Write disable */
+
 #define RTC_ADDR_RAM0	0x20		/* Address of RAM0 */
 #define RTC_ADDR_TCR	0x08		/* Address of trickle charge register */
+#define	RTC_ADDR_CTRL	0x07		/* Address of control register */
 #define	RTC_ADDR_YEAR	0x06		/* Address of year register */
 #define	RTC_ADDR_DAY	0x05		/* Address of day of week register */
 #define	RTC_ADDR_MON	0x04		/* Address of month register */
@@ -313,6 +317,7 @@ static int __init ds1302_rtc_probe(struct platform_device *pdev)
 		return PTR_ERR(rtc);
 
 	platform_set_drvdata(pdev, rtc);
+	ds1302_writebyte(RTC_ADDR_CTRL, RTC_CMD_WRITE_ENABLE);
 
 	return 0;
 }

drivers/rtc/rtc-ds1302.c static int ds1302_rtc_remove(struct platform_device *pdev) · line 331

Quoted 6 lines
drivers/rtc/rtc-ds1302.c +1 -0
@@ -321,6 +326,7 @@ static int ds1302_rtc_remove(struct platform_device *pdev)
 {
 	struct rtc_device *rtc = platform_get_drvdata(pdev);
 
+	ds1302_writebyte(RTC_ADDR_CTRL, RTC_CMD_WRITE_DISABLE);
 	rtc_device_unregister(rtc);
 	platform_set_drvdata(pdev, NULL);

ds1302_rtc_remove() no longer exists in my tree - it got whittled away to nothing by
http://ozlabs.org/~akpm/mmots/broken-out/rtc-rtc-ds1302-remove-unnecessary-platform_set_drvdata.patch
and
http://ozlabs.org/~akpm/mmots/broken-out/drivers-rtc-rtc-ds1302c-remove-empty-function.patch

Perhaps it should be re-added for this?

#2 lore raw
Quoted 15 lines · Andrew Morton

On Tue, 21 May 2013 03:21:30 +0400 Sergey Yanovich <[email protected]> wrote:

(quoted diff) +0 -0
@@ -321,6 +326,7 @@ static int ds1302_rtc_remove(struct platform_device *pdev)
 >  {
 >  	struct rtc_device *rtc = platform_get_drvdata(pdev);

drivers/rtc/rtc-ds1302.c static int ds1302_rtc_remove(struct platform_device *pdev) · line 331

Quoted 4 lines
drivers/rtc/rtc-ds1302.c +1 -0
+	ds1302_writebyte(RTC_ADDR_CTRL, RTC_CMD_WRITE_DISABLE);
 	rtc_device_unregister(rtc);
 	platform_set_drvdata(pdev, NULL);

ds1302_rtc_remove() no longer exists in my tree - it got whittled away to nothing by
http://ozlabs.org/~akpm/mmots/broken-out/rtc-rtc-ds1302-remove-unnecessary-platform_set_drvdata.patch
and
http://ozlabs.org/~akpm/mmots/broken-out/drivers-rtc-rtc-ds1302c-remove-empty-function.patch

Quoted 1 line

Perhaps it should be re-added for this?

There are 2 options. I would be happy with either.

1. I've chosen 'probe/remove' to enable/disable write access.

2. Another option is to wrap enable/disable around ds1302_rtc_set_time().

IIUC, the former saves a few bytes of memory. However, now, when ds1302_rtc_remove() is gone, the latter looks better. So I could rewrite the patch either way.

#3 lore raw
Quoted 8 lines · Sergey Yanovich
Quoted 7 lines · Andrew Morton

On Tue, 21 May 2013 03:21:30 +0400 Sergey Yanovich <[email protected]> wrote:

(quoted diff) +0 -0
@@ -321,6 +326,7 @@ static int ds1302_rtc_remove(struct platform_device
 *pdev)
 >  {
 >  	struct rtc_device *rtc = platform_get_drvdata(pdev);
Quoted 1 line
Quoted 7 lines
Quoted 7 lines

drivers/rtc/rtc-ds1302.c static int ds1302_rtc_remove(struct platform_device *pdev) · line 331

Quoted 3 lines
drivers/rtc/rtc-ds1302.c +1 -0
+	ds1302_writebyte(RTC_ADDR_CTRL, RTC_CMD_WRITE_DISABLE);
 	rtc_device_unregister(rtc);
 	platform_set_drvdata(pdev, NULL);

ds1302_rtc_remove() no longer exists in my tree - it got whittled away to nothing by

http://ozlabs.org/~akpm/mmots/broken-out/rtc-rtc-ds1302-remove-unnecessary-platform_set_drvdata.patch

Quoted 2 lines
Quoted 2 lines

and

http://ozlabs.org/~akpm/mmots/broken-out/drivers-rtc-rtc-ds1302c-remove-empty-function.patch

Quoted 5 lines
Quoted 2 lines

Perhaps it should be re-added for this?

There are 2 options. I would be happy with either.

Quoted 8 lines

1. I've chosen 'probe/remove' to enable/disable write access.

2. Another option is to wrap enable/disable around ds1302_rtc_set_time().

IIUC, the former saves a few bytes of memory. However, now, when ds1302_rtc_remove() is gone, the latter looks better. So I could rewrite the patch either way.

Option two looks actually safer to me, as it ensures that an unexpected reboot outside of the set_time section doesn't leave write access enabled. You never know what firmware could do while you're not looking...

         M.
Signature · 1 line
Who you jivin' with that Cosmik Debris?
#4 lore raw

PATCH 1/1 rtc-ds1302: handle write protection

Quoted 1 line

From f1cd048a066b249082752a96abce7d33a0cd4ea3 Mon Sep 17 00:00:00 2001

From: Sergey Yanovich <[email protected]>
Date: Tue, 21 May 2013 03:06:31 +0400
Subject: [PATCH] rtc-ds1302: handle write protection

This chip has a control register and can prevent altering saved clock. Without this patch we could have:
----8<----

(arm)root@pac14:~# date
Tue May 21 03:08:27 MSK 2013
(arm)root@pac14:~# /etc/init.d/hwclock.sh show
Tue May 21 11:13:58 2013  -0.067322 seconds
(arm)root@pac14:~# /etc/init.d/hwclock.sh stop
[info] Saving the system clock.
[info] Hardware Clock updated to Tue May 21 03:09:01 MSK 2013.
(arm)root@pac14:~# /etc/init.d/hwclock.sh show
Tue May 21 11:14:15 2013  -0.624272 seconds
----8<----

The patch enables write access to rtc before the driver tries to write time and re-disables when time data is written.

# Changes to be committed:
---
changes for v2:
- do enable/disable around set_time() instead of in probe()/remove()

 drivers/rtc/rtc-ds1302.c |    7 +++++++
 1 file changed, 7 insertions(+)
drivers/rtc/rtc-ds1302.c +7 -0
index 7533b72..07e8d79 100644
@@ -23,8 +23,12 @@
 #define	RTC_CMD_READ	0x81		/* Read command */
 #define	RTC_CMD_WRITE	0x80		/* Write command */
 
+#define	RTC_CMD_WRITE_ENABLE	0x00		/* Write enable */
+#define	RTC_CMD_WRITE_DISABLE	0x80		/* Write disable */
+
 #define RTC_ADDR_RAM0	0x20		/* Address of RAM0 */
 #define RTC_ADDR_TCR	0x08		/* Address of trickle charge register */
+#define	RTC_ADDR_CTRL	0x07		/* Address of control register */
 #define	RTC_ADDR_YEAR	0x06		/* Address of year register */
 #define	RTC_ADDR_DAY	0x05		/* Address of day of week register */
 #define	RTC_ADDR_MON	0x04		/* Address of month register */
@@ -161,6 +165,7 @@ static int ds1302_rtc_read_time(struct device *dev, struct rtc_time *tm)
 
 static int ds1302_rtc_set_time(struct device *dev, struct rtc_time *tm)
 {
+	ds1302_writebyte(RTC_ADDR_CTRL, RTC_CMD_WRITE_ENABLE);
 	/* Stop RTC */
 	ds1302_writebyte(RTC_ADDR_SEC, ds1302_readbyte(RTC_ADDR_SEC) | 0x80);
 
@@ -175,6 +180,8 @@ static int ds1302_rtc_set_time(struct device *dev, struct rtc_time *tm)
 	/* Start RTC */
 	ds1302_writebyte(RTC_ADDR_SEC, ds1302_readbyte(RTC_ADDR_SEC) & ~0x80);
 
+	ds1302_writebyte(RTC_ADDR_CTRL, RTC_CMD_WRITE_DISABLE);
+
 	return 0;
 }
 
Signature · 1 line
1.7.10.4
Acked-by on v1 1/1
#5 lore raw
Quoted 19 lines · Sergey Yanovich

From f1cd048a066b249082752a96abce7d33a0cd4ea3 Mon Sep 17 00:00:00 2001
From: Sergey Yanovich <[email protected]>
Date: Tue, 21 May 2013 03:06:31 +0400
Subject: [PATCH] rtc-ds1302: handle write protection

This chip has a control register and can prevent altering saved clock. Without this patch we could have:
----8<----

(arm)root@pac14:~# date
Tue May 21 03:08:27 MSK 2013
(arm)root@pac14:~# /etc/init.d/hwclock.sh show
Tue May 21 11:13:58 2013  -0.067322 seconds
(arm)root@pac14:~# /etc/init.d/hwclock.sh stop
[info] Saving the system clock.
[info] Hardware Clock updated to Tue May 21 03:09:01 MSK 2013.
(arm)root@pac14:~# /etc/init.d/hwclock.sh show
Tue May 21 11:14:15 2013  -0.624272 seconds
----8<----
Quoted 1 line

The patch enables write access to rtc before the driver tries to write

time

Quoted 3 lines

and re-disables when time data is written.

FWIW,

        M.
Signature · 1 line
Who you jivin' with that Cosmik Debris?