Bug 216600

Summary: RTC wakealarm file is missing in v6.1-rc1
Product: Timers Reporter: Todd Brandt (todd.e.brandt)
Component: Realtime ClockAssignee: Rafael J. Wysocki (rjw)
Status: RESOLVED CODE_FIX    
Severity: normal CC: alexandre.belloni, kvalo, lenb, regressions, rjw, rui.zhang
Priority: P1    
Hardware: All   
OS: Linux   
Kernel Version: 6.1.0-rc1 Subsystem:
Regression: No Bisected commit-id:
Bug Depends on:    
Bug Blocks: 178231    
Attachments: [PATCH] Revert "rtc: cmos: Fix event handler registration ordering issue"

Description Todd Brandt 2022-10-18 01:54:29 UTC
In the v6.0-rc1 kernel release, the wakealarm file for the real time clock appears to be missing. Specifically:

/sys/class/rtc/rtc0/wakealarm

Without this file, the rtcwake and sleepgraph tools no longer function properly. This probably affects all tools that use the RTC through sysfs.

I've tested on 10 machines so far and this occurs on all of them regardless of the specific hardware.

The issue is in this merge for the rtc subsystem. I have not yet bisected further, but removing this commit fixes the issue.

commit b7270c69a36efc61ed6ebd31a8a458f354a6edc0
Merge: 4ce1b97949cb e5f12a398371
Author: Linus Torvalds <torvalds@linux-foundation.org>
Date:   Fri Oct 14 18:36:42 2022 -0700

    Merge tag 'rtc-6.1' of git://git.kernel.org/pub/scm/linux/kernel/git/abelloni/linux
    
    Pull RTC updates from Alexandre Belloni:
     "A great rework of the isl12022 driver makes up the bulk of the
      changes. There is also an important fix for CMOS and then the usual
      small fixes:
    
       - switch to devm_clk_get_enabled() where relevant
    
       - cmos: event handler registration fix
    
       - isl12022: code improvements"
    
    * tag 'rtc-6.1' of git://git.kernel.org/pub/scm/linux/kernel/git/abelloni/linux:
      rtc: rv3028: Fix codestyle errors
      rtc: cmos: Fix event handler registration ordering issue
      rtc: k3: Use devm_clk_get_enabled() helper
      rtc: jz4740: Use devm_clk_get_enabled() helper
      rtc: mpfs: Use devm_clk_get_enabled() helper
      rtc: ds1685: Fix spelling of function name in comment block
      rtc: isl12022: switch to using regmap API
      rtc: isl12022: drop redundant write to HR register
      rtc: isl12022: use dev_set_drvdata() instead of i2c_set_clientdata()
      rtc: isl12022: use %ptR
      rtc: isl12022: simplify some expressions
      rtc: isl12022: drop a dev_info()
      rtc: isl12022: specify range_min and range_max
      rtc: isl12022: stop using deprecated devm_rtc_device_register()
      rtc: stmp3xxx: Add failure handling for stmp3xxx_wdt_register()
      rtc: mxc: Use devm_clk_get_enabled() helper
      rtc: gamecube: Always reset HW_SRNPROT after read
      rtc: k3: detect SoC to determine erratum fix
      rtc: k3: wait until the unlock field is not zero
      rtc: mpfs: Remove printing of stray CR
Comment 1 Todd Brandt 2022-10-18 01:57:54 UTC
I meant the v6.1-rc1 release, sorry for the type in the first line.
Comment 2 Todd Brandt 2022-10-18 03:48:38 UTC
It looks like the offending commit is this, adding Raphael to the CC list:

commit 4919d3eb2ec0ee364f7e3cf2d99646c1b224fae8
Author: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Date:   Wed Oct 12 20:07:01 2022 +0200

    rtc: cmos: Fix event handler registration ordering issue
    
    Because acpi_install_fixed_event_handler() enables the event
    automatically on success, it is incorrect to call it before the
    handler routine passed to it is ready to handle events.
    
    Unfortunately, the rtc-cmos driver does exactly the incorrect thing
    by calling cmos_wake_setup(), which passes rtc_handler() to
    acpi_install_fixed_event_handler(), before cmos_do_probe(), because
    rtc_handler() uses dev_get_drvdata() to get to the cmos object
    pointer and the driver data pointer is only populated in
    cmos_do_probe().
    
    This leads to a NULL pointer dereference in rtc_handler() on boot
    if the RTC fixed event happens to be active at the init time.
    
    To address this issue, change the initialization ordering of the
    driver so that cmos_wake_setup() is always called after a successful
    cmos_do_probe() call.
    
    While at it, change cmos_pnp_probe() to call cmos_do_probe() after
    the initial if () statement used for computing the IRQ argument to
    be passed to cmos_do_probe() which is cleaner than calling it in
    each branch of that if () (local variable "irq" can be of type int,
    because it is passed to that function as an argument of type int).
    
    Note that commit 6492fed7d8c9 ("rtc: rtc-cmos: Do not check
    ACPI_FADT_LOW_POWER_S0") caused this issue to affect a larger number
    of systems, because previously it only affected systems with
    ACPI_FADT_LOW_POWER_S0 set, but it is present regardless of that
    commit.
    
    Fixes: 6492fed7d8c9 ("rtc: rtc-cmos: Do not check ACPI_FADT_LOW_POWER_S0")
    Fixes: a474aaedac99 ("rtc-cmos: move wake setup from ACPI glue into RTC driver")
    Link: https://lore.kernel.org/linux-acpi/20221010141630.zfzi7mk7zvnmclzy@techsingularity.net/
    Reported-by: Mel Gorman <mgorman@techsingularity.net>
    Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
    Reviewed-by: Bjorn Helgaas <bhelgaas@google.com>
    Tested-by: Mel Gorman <mgorman@techsingularity.net>
    Link: https://lore.kernel.org/r/5629262.DvuYhMxLoT@kreacher
    Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Comment 3 Zhang Rui 2022-10-18 03:50:28 UTC
This seems to be introduced by commit
commit 4919d3eb2ec0ee364f7e3cf2d99646c1b224fae8
Author:     Rafael J. Wysocki <rafael.j.wysocki@intel.com>
AuthorDate: Wed Oct 12 20:07:01 2022 +0200
Commit:     Alexandre Belloni <alexandre.belloni@bootlin.com>
CommitDate: Thu Oct 13 23:27:52 2022 +0200

    rtc: cmos: Fix event handler registration ordering issue


with the commit, the rtc device is registered (cmos_do_probe()) before wakeup-capability enabled in cmos_wake_setup(), thus "wakealarm" attribute is not created because rtc_does_wakealarm() returns failure.

I have confirmed the workaround below can bring the wakealarm attribute back.

diff --git a/drivers/rtc/rtc-cmos.c b/drivers/rtc/rtc-cmos.c
index 343f31ab7a01..76a5c7d968a4 100644
--- a/drivers/rtc/rtc-cmos.c
+++ b/drivers/rtc/rtc-cmos.c
@@ -1368,6 +1368,7 @@ static int cmos_pnp_probe(struct pnp_dev *pnp, const struct pnp_device_id *id)
                irq = pnp_irq(pnp, 0);
        }

+       device_init_wakeup(&pnp->dev, 1);
        ret = cmos_do_probe(&pnp->dev, pnp_get_resource(pnp, IORESOURCE_IO, 0), irq);
        if (ret)
Comment 4 The Linux kernel's regression tracker (Thorsten Leemhuis) 2022-10-18 06:48:23 UTC
TWIMC and for proper Link: and reported-by tags, there afaics is another report about this issue here: https://lore.kernel.org/all/87edv6mlow.fsf@kernel.org/
Comment 5 Len Brown 2022-10-18 13:44:03 UTC
Created attachment 303029 [details]
[PATCH] Revert "rtc: cmos: Fix event handler registration ordering issue"

The change in comment #3 did not work for me,
but the revert (attached) does work for me.
Comment 6 Rafael J. Wysocki 2022-10-18 16:21:27 UTC
Should be fixed by:

https://lore.kernel.org/linux-acpi/5887691.lOV4Wx5bFT@kreacher/
Comment 7 Todd Brandt 2022-10-18 18:39:05 UTC
Just tested Raphael's patch, it appears to work on the 6 machines I've tried.
Comment 8 Todd Brandt 2022-10-24 20:11:39 UTC
This patch still does not appear in 6.1.0-rc2, so I'm keeping it open until it actually makes it into the upstream kernel.
Comment 9 Todd Brandt 2022-10-31 22:53:52 UTC
I just verified the fix has made it into 6.1-rc3. This bug is now closed.