RTC warning after "rtc: mc146818: Detect and handle broken RTCs"
Subsystems (2)
Hello,
On one of our home computers, the Linux RTC code since Linux 5.11 prints a warning message with a stacktrace in mc146818_get_time():
[ 0.351685] clocksource: jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 7645041785100000 ns
[ 0.351685] futex hash table entries: 2048 (order: 5, 131072 bytes, linear)
[ 0.351685] pinctrl core: initialized pinctrl subsystem
[ 0.351685] ------------[ cut here ]------------
[ 0.351685] WARNING: CPU: 0 PID: 1 at drivers/rtc/rtc-mc146818-lib.c:25 mc146818_get_time+0x1d5/0x230
[ 0.351685] Modules linked in:
[ 0.351685] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 5.11.0newdeb-mj4 #12
[ 0.351685] Hardware name: To Be Filled By O.E.M. To Be Filled By O.E.M./A780GXH/128M, BIOS P1.70 05/14/2010
[ 0.351685] RIP: 0010:mc146818_get_time+0x1d5/0x230
[ 0.351685] Code: 10 5b 41 5c 41 5d 41 5e 41 5f 5d c3 4c 89 e6 48 c7 c7 b0 87 56 8f e8 6a 9e 2a 00 bf 58 89 41 00 e8 30 f8 d4 ff e9 43 fe ff ff <0f> 0b 48 c7 c7 b0 87 56 8f 4c 89 e6 e8 4a 9e 2a 00 48 c7 03 ff ff [ 0.351685] RSP: 0018:ffffb618c001fda8 EFLAGS: 00010002
[ 0.351685] RAX: 00000000000000c0 RBX: ffffb618c001fde4 RCX: 0000000000000000
[ 0.351685] RDX: 0000000000000001 RSI: ffff9378c0456940 RDI: 000000000000000d
[ 0.351685] RBP: ffffb618c001fdd0 R08: ffff9378c0405b60 R09: 0000000000000000
[ 0.351685] R10: 0000000000000000 R11: 605640c07893ffff R12: 0000000000000246
[ 0.351685] R13: 0000000000000000 R14: ffffffff8f54f390 R15: 0000000000000000
[ 0.351685] FS: 0000000000000000(0000) GS:ffff9379d8000000(0000) knlGS:0000000000000000
[ 0.351685] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 0.351685] CR2: 0000000000000000 CR3: 0000000132810000 CR4: 00000000000006f0
[ 0.351685] Call Trace:
[ 0.351685] ? wakeup_sources_sysfs_init+0x35/0x35
[ 0.351685] early_resume_init+0x25/0x9c
[ 0.351685] ? cn_proc_init+0x3f/0x3f
[ 0.351685] do_one_initcall+0x48/0x1d0
[ 0.351685] kernel_init_freeable+0x1dc/0x23a
[ 0.351685] ? rest_init+0xba/0xba
[ 0.351685] kernel_init+0xe/0x116
[ 0.351685] ret_from_fork+0x22/0x30
[ 0.351685] ---[ end trace b30828ec3a66c652 ]---
[ 0.351685] PM: RTC time: 18446744073709551615:18446744073709551615:18446744073709551615, date: 1899-00-18446744073709551615
[ 0.351685] NET: Registered protocol family 16
This happens every boot on affected kernels. Surprisingly, it appears that the local clock gets synchronized from the RTC code nonetheless.
I have isolated the issue to the following commit:
commit 211e5db19d15a721b2953ea54b8f26c2963720eb
Author: Thomas Gleixner<[email protected]> Date: Tue Jan 26 18:02:11 2021 +0100
rtc: mc146818: Detect and handle broken RTCs
which was modified later by commit ebb22a05943666155e6da04407cc6e913974c78c
rtc: mc146818: Dont test for bit 0-5 in Register D
After reverting both commits on top of Linux 5.11, the warning disappears.
The sum of these commits causes Linux kernel to check whether bit 6 in CMOS register 0x0D is unset and abort otherwise.
This computer is a desktop with an ASRock A780GXH/128M motherboard and an AMD SB710 southbridge. The southbridge datasheet:
https://developer.amd.com/wordpress/media/2012/10/43009_sb7xx_rrg_pub_1.00.pdf
specifies (page 308) that bit 6 of register 0x0D is a scratchbit. Apparently, something had set it to 1.
Reading CMOS register 0x0d with a test patch consistently returns 0xc0, even when it is done multiple times consecutively.
Everything was tested on Ubuntu 20.04.
Greetings,
Mateusz Jończyk
Last thought:
Instead of checking for bit 6 in the CMOS register 0x0d, it could be better to check if the hours register (CMOS register 0x04) contains a sane value (lower then 25). If it contained 0xff, something would be strange.
Also the stacktrace below should be sanely formatted:
W dniu 01.07.2021 o 19:34, Mateusz Jończyk pisze:
Quoted 56 lines
Hello,
On one of our home computers, the Linux RTC code since Linux 5.11 prints a warning message
with a stacktrace inmc146818_get_time():
[ 0.351685] clocksource: jiffies: mask:0xffffffffmax_cycles:0xffffffff,max_idle_ns: 7645041785100000 ns
[ 0.351685] futex hash table entries: 2048 (order: 5, 131072 bytes, linear)
[ 0.351685] pinctrl core: initialized pinctrl subsystem
[ 0.351685] ------------[ cut here ]------------
[ 0.351685] WARNING: CPU: 0 PID: 1 atdrivers/rtc/rtc-mc146818-lib.c:25mc146818_get_time+0x1d5/0x230
[ 0.351685] Modules linked in:
[ 0.351685] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 5.11.0newdeb-mj4 #12
[ 0.351685] Hardware name: To Be Filled By O.E.M. To Be Filled By O.E.M./A780GXH/128M, BIOS P1.70 05/14/2010
[ 0.351685] RIP: 0010:mc146818_get_time+0x1d5/0x230
[ 0.351685] Code: 10 5b 41 5c 41 5d 41 5e 41 5f 5d c3 4c 89 e6 48 c7 c7 b0 87 56 8f e8 6a 9e 2a 00 bf 58 89 41 00 e8 30 f8 d4 ff e9 43 fe ff ff <0f> 0b 48 c7 c7 b0 87 56 8f 4c 89 e6 e8 4a 9e 2a 00 48 c7 03 ff ff [ 0.351685] RSP: 0018:ffffb618c001fda8 EFLAGS: 00010002
[ 0.351685] RAX:00000000000000c0RBX:ffffb618c001fde4RCX: 0000000000000000
[ 0.351685] RDX: 0000000000000001 RSI:ffff9378c0456940RDI:000000000000000d
[ 0.351685] RBP:ffffb618c001fdd0R08:ffff9378c0405b60R09: 0000000000000000
[ 0.351685] R10: 0000000000000000 R11:605640c07893ffffR12: 0000000000000246
[ 0.351685] R13: 0000000000000000 R14:ffffffff8f54f390R15: 0000000000000000
[ 0.351685] FS: 0000000000000000(0000) GS:ffff9379d8000000(0000) knlGS:0000000000000000
[ 0.351685] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 0.351685] CR2: 0000000000000000 CR3: 0000000132810000 CR4:00000000000006f0
[ 0.351685] Call Trace:
[ 0.351685] ?wakeup_sources_sysfs_init+0x35/0x35
[ 0.351685]early_resume_init+0x25/0x9c
[ 0.351685] ?cn_proc_init+0x3f/0x3f
[ 0.351685]do_one_initcall+0x48/0x1d0
[ 0.351685]kernel_init_freeable+0x1dc/0x23a
[ 0.351685] ?rest_init+0xba/0xba
[ 0.351685]kernel_init+0xe/0x116
[ 0.351685]ret_from_fork+0x22/0x30
[ 0.351685] ---[ end traceb30828ec3a66c652]---
[ 0.351685] PM: RTC time: 18446744073709551615:18446744073709551615:18446744073709551615, date: 1899-00-18446744073709551615
[ 0.351685] NET: Registered protocol family 16
This happens every boot on affected kernels. Surprisingly, it appears that the local
clock gets synchronized from the RTC code nonetheless.
I have isolated the issue to the following commit:
commit211e5db19d15a721b2953ea54b8f26c2963720eb
Author: Thomas Gleixner<[email protected]>
Date: Tue Jan 26 18:02:11 2021 +0100
rtc: mc146818: Detect and handle broken RTCs
which was modified later by commitebb22a05943666155e6da04407cc6e913974c78c
rtc: mc146818: Dont test for bit 0-5 in Register D
After reverting both commits on top of Linux 5.11, the warning disappears.
The sum of these commits causes Linux kernel to check whether bit 6 in CMOS
register0x0Dis unset and abort otherwise.
This computer is a desktop with an ASRock A780GXH/128M motherboard and
an AMD SB710 southbridge. The southbridge datasheet:
https://developer.amd.com/wordpress/media/2012/10/43009_sb7xx_rrg_pub_1.00.pdf
specifies (page 308) that bit 6 of register0x0Dis a scratchbit.
Apparently, something had set it to 1.
Reading CMOS register0x0dwith a test patch consistently returns0xc0, even
when it is done multiple times consecutively.
Everything was tested on Ubuntu 20.04.
Greetings,
Mateusz Jończyk
PATCH 1/1 rtc: mc146818: Use hours for checking RTC availability
To prevent an infinite loop, it is necessary to ascertain the RTC is present. Previous code was checking if bit 6 in register 0x0D is cleared. This caused a false negative on a motherboard with an AMD SB710 southbridge; according to the specification [1], bit 6 of register 0x0D on this chipset is a scratchbit.
Use the RTC_HOURS register instead, which is expected to contain a value not larger then 24, in BCD format.
Caveat: when I was playing with
while true; do cat /sys/class/rtc/rtc0/time; done
I sometimes triggered this mechanism on my HP laptop. It appears that CMOS_READ(RTC_VALID) was sometimes reading the number of seconds from previous loop iteration. This happens very rarely, though, and this patch does not make it any more likely.
[1] AMD SB700/710/750 Register Reference Guide, page 308,
https://developer.amd.com/wordpress/media/2012/10/43009_sb7xx_rrg_pub_1.00.pdf
--- drivers/rtc/rtc-cmos.c | 6 +++--- drivers/rtc/rtc-mc146818-lib.c | 10 ++++++++-- 2 files changed, 11 insertions(+), 5 deletions(-)
drivers/rtc/rtc-cmos.c +3 -3
@@ -798,10 +798,10 @@ cmos_do_probe(struct device *dev, struct resource *ports, int rtc_irq)spin_lock_irq (&rtc_lock); -/* Ensure that the RTC is accessible. Bit 6 must be 0! */ -if ((CMOS_READ (RTC_VALID ) &0x40 ) !=0 ) { +/* Ensure that the RTC is accessible (RTC_HOURS is in BCD format) */ +if (CMOS_READ (RTC_HOURS ) >0x24 ) {spin_unlock_irq (&rtc_lock); -dev_warn (dev,"not accessible\n" ); +dev_warn (dev,"not accessible or not working correctly\n" ); retval = -ENXIO ; goto cleanup1; }
drivers/rtc/rtc-mc146818-lib.c +8 -2
@@ -21,9 +21,15 @@ unsigned int mc146818_get_time(struct rtc_time *time) again:spin_lock_irqsave (&rtc_lock, flags); -/* Ensure that the RTC is accessible. Bit 6 must be 0! */ -if (WARN_ON_ONCE ((CMOS_READ (RTC_VALID ) &0x40 ) !=0 )) { + +/* +* Ensure that the RTC is accessible, to avoid an infinite loop. +* RTC_HOURS is in BCD format. +*/ +if (CMOS_READ (RTC_HOURS ) >0x24 ) {spin_unlock_irqrestore (&rtc_lock, flags); +pr_warn_once ("Real-time clock is not accessible or not " +"working correctly\n" );memset (time,0xff ,sizeof (*time));return 0 ; }
Signature · 1 line
2.25.1
Re: [PATCH] rtc: mc146818: Use hours for checking RTC availability
Hello,
Please do not apply this patch, it causes trouble on my Intel Kaby Lake laptop. (see below).
W dniu 07.07.2021 o 23:38, Mateusz Jończyk pisze:
Quoted 9 lines
To prevent an infinite loop, it is necessary to ascertain the RTC is present. Previous code was checking if bit 6 in register
0x0Dis cleared. This caused a false negative on a motherboard with an AMD SB710 southbridge; according to the specification [1], bit 6 of register0x0Don this chipset is a scratchbit.Use the
RTC_HOURSregister instead, which is expected to contain a value not larger then 24, in BCD format.
Quoted 6 lines
Caveat: when I was playing with
while true; do cat/sys/class/rtc/rtc0/time; done
I sometimes triggered this mechanism on my HP laptop. It appears thatCMOS_READ(RTC_VALID)was sometimes reading the number of seconds from previous loop iteration. This happens very rarely, though, and this patch does not make it any more likely.
According to Intel's documentation [2]:
The update cycle [of CMOS RTC registers, which happens every second] will start at least 488 μs after the UIP bit of register A is asserted, and the entire cycle does not take more than 1984 μs to complete. The time and date RAM locations (0–9) are disconnected from the external bus during this time.
Therefore, trying to access the RTC_HOURS register during the update time is undefined behavior. In my tests, it frequently returned values from the wrong register, such as the seconds or minutes register instead of the hours register and the values failed the "< 24" comparison.
Indeed, the wrong readouts happen when the UIP bit is set.
I was not aware of this and blamed my firmware for accessing the RTC clock independently.
BTW, cmos_set_alarm() in drivers/rtc/rtc-cmos.c does not check for UIP bit. As it writes to registers in the 0-9 range, it may fail to set the alarm time.
Greetings,
Mateusz
[2] 7th Generation Intel® Processor Family I/O for U/Y Platforms [...] Datasheet – Volume 1 of 2, page 209
Intel's Document Number: 334658-006,
https://www.intel.com/content/dam/www/public/us/en/documents/datasheets/7th-and-8th-gen-core-family-mobile-u-y-processor-lines-i-o-datasheet-vol-1.pdf
Quoted 55 lines
[1] AMD SB700/710/750 Register Reference Guide, page 308,
https://developer.amd.com/wordpress/media/2012/10/43009_sb7xx_rrg_pub_1.00.pdfFixes 211e5db19d15 ("rtc: mc146818: Detect and handle broken RTCs")Fixes ebb22a059436 ("rtc: mc146818: Dont test for bit 0-5 in Register D")Signed-off-by Mateusz Jończyk <[email protected]>Cc Thomas Gleixner <[email protected]>Cc Alessandro Zummo <[email protected]>Cc Alexandre Belloni <[email protected]>--- drivers/rtc/rtc-cmos.c | 6 +++--- drivers/rtc/rtc-mc146818-lib.c | 10 ++++++++-- 2 files changed, 11 insertions(+), 5 deletions(-)drivers/rtc/rtc-cmos.c +3
-3@@ -798,10 +798,10 @@ cmos_do_probe(struct device *dev, struct resource *ports, int rtc_irq)spin_lock_irq (&rtc_lock); -/* Ensure that the RTC is accessible. Bit 6 must be 0! */ -if ((CMOS_READ (RTC_VALID ) &0x40 ) !=0 ) { +/* Ensure that the RTC is accessible (RTC_HOURS is in BCD format) */ +if (CMOS_READ (RTC_HOURS ) >0x24 ) {spin_unlock_irq (&rtc_lock); -dev_warn (dev,"not accessible\n" ); +dev_warn (dev,"not accessible or not working correctly\n" ); retval = -ENXIO ; goto cleanup1; }drivers/rtc/rtc-mc146818-lib.c +8
-2@@ -21,9 +21,15 @@ unsigned int mc146818_get_time(struct rtc_time *time) again:spin_lock_irqsave (&rtc_lock, flags); -/* Ensure that the RTC is accessible. Bit 6 must be 0! */ -if (WARN_ON_ONCE ((CMOS_READ (RTC_VALID ) &0x40 ) !=0 )) { + +/* +* Ensure that the RTC is accessible, to avoid an infinite loop. +* RTC_HOURS is in BCD format. +*/ +if (CMOS_READ (RTC_HOURS ) >0x24 ) {spin_unlock_irqrestore (&rtc_lock, flags); +pr_warn_once ("Real-time clock is not accessible or not " +"working correctly\n" );memset (time,0xff ,sizeof (*time));return 0 ; }
PATCH 1/1 rtc-mc146818: fix RTC presence check
To prevent an infinite loop in mc146818_get_time(),
commit 211e5db19d15 ("rtc: mc146818: Detect and handle broken RTCs") added a check for RTC availability. Together with a later fix, it checked if bit 6 in register 0x0d is cleared. This, however, caused a false negative on a motherboard with an AMD SB710 southbridge; according to the specification [1], bit 6 of register 0x0d of this chipset is a scratchbit.
This caused a regression in Linux 5.11 - the RTC was not used by the kernel.
As a better alternative, check whether the UIP ("Update-in-progress") bit is set for longer then 10ms. If that is the case, then apparently the RTC is either absent (and all register reads return 0xff) or broken. Also limit the number of loop iterations in mc146818_get_time() to 10.
In a previous approach to this problem, I implemented a check whether the RTC_HOURS register contains a value <= 24. This, however, sometimes did not work correctly on my Intel Kaby Lake laptop. According to Intel's documentation [2], "the time and date RAM locations (0-9) are disconnected from the external bus" during the update cycle so reading this register without checking the UIP bit is incorrect.
[1] AMD SB700/710/750 Register Reference Guide, page 308,
https://developer.amd.com/wordpress/media/2012/10/43009_sb7xx_rrg_pub_1.00.pdf
[2] 7th Generation Intel ® Processor Family I/O for U/Y Platforms [...] Datasheet Volume 1 of 2, page 209
Intel's Document Number: 334658-006,
https://www.intel.com/content/dam/www/public/us/en/documents/datasheets/7th-and-8th-gen-core-family-mobile-u-y-processor-lines-i-o-datasheet-vol-1.pdf
---
Hello,
Please check this commit carefully as the RTC is a finicky beast and I'm quite new to kernel development.
I have tested this on three computers, though, and found no problems. Also, using the UIP (Update-in-progress) bit should be safe.
Greetings,
Mateusz
drivers/rtc/rtc-cmos.c | 10 ++++------ drivers/rtc/rtc-mc146818-lib.c | 34 ++++++++++++++++++++++++++++++---- include/linux/mc146818rtc.h | 1 + 3 files changed, 35 insertions(+), 10 deletions(-)
drivers/rtc/rtc-cmos.c +4 -6
@@ -796,16 +796,14 @@ cmos_do_probe(struct device *dev, struct resource *ports, int rtc_irq)rename_region (ports,dev_name (&cmos_rtc.rtc->dev)); -spin_lock_irq (&rtc_lock); - -/* Ensure that the RTC is accessible. Bit 6 must be 0! */ -if ((CMOS_READ (RTC_VALID ) &0x40 ) !=0 ) { -spin_unlock_irq (&rtc_lock); -dev_warn (dev,"not accessible\n" ); +if (!mc146818_does_rtc_work ()) { +dev_warn (dev,"broken or not accessible\n" ); retval = -ENXIO ; goto cleanup1; } +spin_lock_irq (&rtc_lock); +if (!(flags &CMOS_RTC_FLAGS_NOFREQ )) {/* force periodic irq to CMOS reset default of 1024Hz; *
drivers/rtc/rtc-mc146818-lib.c +30 -4
@@ -8,10 +8,35 @@#include <linux/acpi.h> #endif +/* +* If the UIP (Update-in-progress) bit of the RTC is set for more then +* 10ms, the RTC apparently is broken or not present. +*/ +unsigned int mc146818_does_rtc_work (void ) +{ +int i; +unsigned char val; +unsigned long flags; + +for (i =0 ; i <20 ; i++) { +spin_lock_irqsave (&rtc_lock, flags); + val =CMOS_READ (RTC_FREQ_SELECT ); +spin_unlock_irqrestore (&rtc_lock, flags); + +if ((val &RTC_UIP ) ==0 ) +return 1 ; + +udelay (500 ); + } + +return 0 ; +} +unsigned int mc146818_get_time (struct rtc_time *time) {unsigned char ctrl;unsigned long flags; +unsigned int iter_count =0 ;unsigned char century =0 ;bool retry;
@@ -20,13 +45,14 @@ unsigned int mc146818_get_time(struct rtc_time *time)#endif again: -spin_lock_irqsave (&rtc_lock, flags); -/* Ensure that the RTC is accessible. Bit 6 must be 0! */ -if (WARN_ON_ONCE ((CMOS_READ (RTC_VALID ) &0x40 ) !=0 )) { -spin_unlock_irqrestore (&rtc_lock, flags); +if (iter_count >10 ) { +pr_err_ratelimited ("Unable to read current time from RTC\n" );memset (time,0xff ,sizeof (*time));return 0 ; } + iter_count++; + +spin_lock_irqsave (&rtc_lock, flags);/* * Check whether there is an update in progress during which the
include/linux/mc146818rtc.h +1 -0
@@ -123,6 +123,7 @@ struct cmos_rtc_board_info {#define RTC_IO_EXTENT_USED RTC_IO_EXTENT#endif /* ARCH_RTC_LOCATION */ +unsigned int mc146818_does_rtc_work (void );unsigned int mc146818_get_time (struct rtc_time *time);int mc146818_set_time (struct rtc_time *time);
Signature · 1 line
2.25.1