rtc-m48t59 on an Ultra 5

rtc-m48t59 on an Ultra 5

[Josip Rodin]

Hi,

Current sparc-next-2.6 on my Ultra 5 works fine, but the new RTC driver
prints the following during boot:

rtc-m48t59 rtc-m48t59.0: rtc core: registered m48t59 as rtc0
rtc-m48t59 rtc-m48t59.0: hctosys: invalid date/time

The system time is okay after boot, though.

% sudo ln -s rtc0 /dev/rtc
% sudo hwclock --debug; echo; date
hwclock from util-linux-2.12r
Using /dev/rtc interface to clock.
Last drift adjustment done at 1220663753 seconds after 1969
Last calibration done at 1220663753 seconds after 1969
Hardware clock is on UTC time
Assuming hardware clock is kept in UTC time.
Waiting for clock tick...
/dev/rtc does not have interrupt functions. Waiting in loop for time from /dev/rtc to change
...got clock tick
Time read from Hardware Clock: 1940/09/06 01:42:03
Hw clock time : 1940/09/06 01:42:03 = -925251477 seconds since 1969
Fri Sep  6 02:42:03 1940  -0.646109 seconds

Sat Sep  6 03:42:03 CEST 2008

Should I be worried? :)

-- 
     2. That which causes joy or happiness.

Re: rtc-m48t59 on an Ultra 5

[David Miller]

From: Josip Rodin <joy@entuzijast.net>
Date: Sat, 6 Sep 2008 03:44:32 +0200

> Current sparc-next-2.6 on my Ultra 5 works fine, but the new RTC driver
> prints the following during boot:
> 
> rtc-m48t59 rtc-m48t59.0: rtc core: registered m48t59 as rtc0
> rtc-m48t59 rtc-m48t59.0: hctosys: invalid date/time
> 
> The system time is okay after boot, though.

It might be related to this bug fix I just added to the
sparc-next-2.6 tree:

sparc64: fix wrong m48t59 RTC year

Correctly calculate offset to the year register for
Mostek RTC.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 arch/sparc64/kernel/time.c |    6 ++++--
 1 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/arch/sparc64/kernel/time.c b/arch/sparc64/kernel/time.c
index 209e7d2..226a004 100644
--- a/arch/sparc64/kernel/time.c
+++ b/arch/sparc64/kernel/time.c
@@ -500,6 +500,7 @@ static struct of_platform_driver bq4802_driver = {
 static unsigned char mostek_read_byte(struct device *dev, u32 ofs)
 {
 	struct platform_device *pdev = to_platform_device(dev);
+	struct m48t59_plat_data *pdata = pdev->dev.platform_data;
 	void __iomem *regs;
 	unsigned char val;
 
@@ -507,7 +508,7 @@ static unsigned char mostek_read_byte(struct device *dev, u32 ofs)
 	val = readb(regs + ofs);
 
 	/* the year 0 is 1968 */
-	if (ofs = M48T59_YEAR) {
+	if (ofs = pdata->offset + M48T59_YEAR) {
 		val += 0x68;
 		if ((val & 0xf) > 9)
 			val += 6;
@@ -518,10 +519,11 @@ static unsigned char mostek_read_byte(struct device *dev, u32 ofs)
 static void mostek_write_byte(struct device *dev, u32 ofs, u8 val)
 {
 	struct platform_device *pdev = to_platform_device(dev);
+	struct m48t59_plat_data *pdata = pdev->dev.platform_data;
 	void __iomem *regs;
 
 	regs = (void __iomem *) pdev->resource[0].start;
-	if (ofs = M48T59_YEAR) {
+	if (ofs = pdata->offset + M48T59_YEAR) {
 		if (val < 0x68)
 			val += 0x32;
 		else
-- 
1.5.6.5.GIT

Re: rtc-m48t59 on an Ultra 5

[Josip Rodin]

On Sun, Sep 07, 2008 at 06:14:09PM -0700, David Miller wrote:
> > Current sparc-next-2.6 on my Ultra 5 works fine, but the new RTC driver
> > prints the following during boot:
> > 
> > rtc-m48t59 rtc-m48t59.0: rtc core: registered m48t59 as rtc0
> > rtc-m48t59 rtc-m48t59.0: hctosys: invalid date/time
> > 
> > The system time is okay after boot, though.
> 
> It might be related to this bug fix I just added to the
> sparc-next-2.6 tree:
> 
> sparc64: fix wrong m48t59 RTC year
> 
> Correctly calculate offset to the year register for
> Mostek RTC.

Oh yeah, I forgot to report my test of that patch, sorry. Now the kernel said:

rtc-m48t59 rtc-m48t59.0: setting system clock to 1970-09-06 17:38:55 UTC (21490735)

Userland says:

% sudo hwclock --debug; echo; date
hwclock from util-linux-2.12r
Using /dev/rtc interface to clock.
Last drift adjustment done at 1220722547 seconds after 1969
Last calibration done at 1220722547 seconds after 1969
Hardware clock is on UTC time
Assuming hardware clock is kept in UTC time.
Waiting for clock tick...
/dev/rtc does not have interrupt functions. Waiting in loop for time from /dev/rtc to change
...got clock tick
Time read from Hardware Clock: 1970/09/08 01:49:11
Hw clock time : 1970/09/08 01:49:11 = 21606551 seconds since 1969
Tue Sep  8 02:49:11 1970  -0.803853 seconds

Mon Sep  8 03:49:10 CEST 2008

Is this now ready for a hwclock --systohc or?

-- 
     2. That which causes joy or happiness.

Re: rtc-m48t59 on an Ultra 5

[Robert Reif]

Josip Rodin wrote:
> % sudo hwclock --debug; echo; date
> hwclock from util-linux-2.12r
> Using /dev/rtc interface to clock.
>   
I have similar problems with debian etch on sparc32.  hwclock
only supports /dev/rtc.  Upgrading hwclock is problematic because there
are 3 different versions (original hwclock, linux-util and linux-util-ng)
and they require init.d script changes.  Superblock times are also
corrupted with bad dates and fails fsck which immediately causes a
reboot.  Fixing it requires mounting the drive on another system and
running fsck to fix the times.  This is a real pain because I haven't
been able to find a combination that boots both the stock debian etch
kernel and sparc-next.

Is there any way to have the the rtc device show up as /dev/rtc?

Re: rtc-m48t59 on an Ultra 5

[Krzysztof Helt]

On Sun, 07 Sep 2008 23:25:13 -0400
Robert Reif <reif@earthlink.net> wrote:

> Josip Rodin wrote:
> > % sudo hwclock --debug; echo; date
> > hwclock from util-linux-2.12r
> > Using /dev/rtc interface to clock.
> >   
> I have similar problems with debian etch on sparc32.  hwclock
> only supports /dev/rtc.  Upgrading hwclock is problematic because there
> are 3 different versions (original hwclock, linux-util and linux-util-ng)
> and they require init.d script changes.  Superblock times are also
> corrupted with bad dates and fails fsck which immediately causes a
> reboot.  Fixing it requires mounting the drive on another system and
> running fsck to fix the times.  This is a real pain because I haven't
> been able to find a combination that boots both the stock debian etch
> kernel and sparc-next.
> 
> Is there any way to have the the rtc device show up as /dev/rtc?
> 

There are few things which can improve your situation:

1. Rebuild the kernel with  CONFIG_RTC_HCTOSYS option set.
It is named "Set system time from RTC on startup and resume"
This make system use the rtc0 (you can set it up in configuration
as well) to get initial time and date.

2. Add "ln -s /dev/rtc0 /dev/rtc" to your booting scripts before
the hwclock (or ntp) is used for the first time.

3. Define udev rule (if you use udev) to create /dev/rtc link to 
the /dev/rtc0.

Regards,
Krzysztof

Re: rtc-m48t59 on an Ultra 5

[David Miller]

From: Krzysztof Helt <krzysztof.h1@wp.pl>
Date: Mon, 8 Sep 2008 17:13:13 +0200

> There are few things which can improve your situation:
> 
> 1. Rebuild the kernel with  CONFIG_RTC_HCTOSYS option set.
> It is named "Set system time from RTC on startup and resume"
> This make system use the rtc0 (you can set it up in configuration
> as well) to get initial time and date.
> 
> 2. Add "ln -s /dev/rtc0 /dev/rtc" to your booting scripts before
> the hwclock (or ntp) is used for the first time.
> 
> 3. Define udev rule (if you use udev) to create /dev/rtc link to 
> the /dev/rtc0.

These all work, but I really think there should be at least a kernel
config controlled way for an automatic alias to show up.

Even something as simple as just creating a /sys/class/misc/rtc/dev
file that outputs the major and minor device numbers of /dev/rtc0,
udev would just "do the right thing" and create that alias /dev/rtc
node for the user.

This whole /dev/rtc0 business is just unnecessary pain, it makes
transition of a platform to the RTC subsystem harder rather than
easier.