| Message ID | 20230526184139.2.Ic6ebbf307ca0efe91f08ce2c1eb4a037ba6b0700@changeid |
|---|---|
| State | New |
| Headers | show |
| Series | watchdog: Cleanup / fixes after buddy series v5 reviews | expand |
On Fri 2023-05-26 18:41:32, Douglas Anderson wrote: > Right now there is one arch (sparc64) that selects HAVE_NMI_WATCHDOG > without selecting HAVE_HARDLOCKUP_DETECTOR_ARCH. Because of that one > architecture, we have some special case code in the watchdog core to > handle the fact that watchdog_hardlockup_probe() isn't implemented. > > Let's implement watchdog_hardlockup_probe() for sparc64 and get rid of > the special case. > > As a side effect of doing this, code inspection tells us that we could > fix a minor bug where the system won't properly realize that NMI > watchdogs are disabled. Specifically, on powerpc if > CONFIG_PPC_WATCHDOG is turned off the arch might still select > CONFIG_HAVE_HARDLOCKUP_DETECTOR_ARCH which selects > CONFIG_HAVE_NMI_WATCHDOG. Since CONFIG_PPC_WATCHDOG was off then > nothing will override the "weak" watchdog_hardlockup_probe() and we'll > fallback to looking at CONFIG_HAVE_NMI_WATCHDOG. > > Suggested-by: Petr Mladek <pmladek@suse.com> > Signed-off-by: Douglas Anderson <dianders@chromium.org> Looks good: Reviewed-by: Petr Mladek <pmladek@suse.com> > --- > Though this does fix a minor bug, I didn't mark this as "Fixes" > because it's super minor. One could also argue that this wasn't a bug > at all but simply was never an implemented feature. The code that > added some amount of dynamicness here was commit a994a3147e4c > ("watchdog/hardlockup/perf: Implement init time detection of perf") > which, as per the title, was only intending to make "perf" > dynamic. The old NMI watchdog presumably has never been handled > dynamically. I agree that it is a minor bug and Fixes tag is not needed. Another motivation is that all the dependencies are quite complicated. IMHO, it is not worth spending time on reviewing potential backports. That said, I am afraid that the artificial intelligence will nominate this patch for stable backports even without the Fixes tag. You know, there are the words "fix" and "bug" in the commit message ;-) Best Regards, Petr
diff --git a/arch/Kconfig b/arch/Kconfig index 64d771855ecd..b4f6387b12fe 100644 --- a/arch/Kconfig +++ b/arch/Kconfig @@ -428,7 +428,8 @@ config HAVE_NMI_WATCHDOG bool help The arch provides a low level NMI watchdog. It provides - asm/nmi.h, and defines its own arch_touch_nmi_watchdog(). + asm/nmi.h, and defines its own watchdog_hardlockup_probe() and + arch_touch_nmi_watchdog(). config HAVE_HARDLOCKUP_DETECTOR_ARCH bool diff --git a/arch/sparc/kernel/nmi.c b/arch/sparc/kernel/nmi.c index 9d9e29b75c43..17cdfdbf1f3b 100644 --- a/arch/sparc/kernel/nmi.c +++ b/arch/sparc/kernel/nmi.c @@ -65,6 +65,11 @@ void arch_touch_nmi_watchdog(void) } EXPORT_SYMBOL(arch_touch_nmi_watchdog); +int __init watchdog_hardlockup_probe(void) +{ + return 0; +} + static void die_nmi(const char *str, struct pt_regs *regs, int do_panic) { int this_cpu = smp_processor_id(); diff --git a/kernel/watchdog.c b/kernel/watchdog.c index 4b9e31edb47f..62230f5b8878 100644 --- a/kernel/watchdog.c +++ b/kernel/watchdog.c @@ -217,19 +217,6 @@ void __weak watchdog_hardlockup_disable(unsigned int cpu) { } */ int __weak __init watchdog_hardlockup_probe(void) { - /* - * If CONFIG_HAVE_NMI_WATCHDOG is defined then an architecture - * is assumed to have the hard watchdog available and we return 0. - */ - if (IS_ENABLED(CONFIG_HAVE_NMI_WATCHDOG)) - return 0; - - /* - * Hardlockup detectors other than those using CONFIG_HAVE_NMI_WATCHDOG - * are required to implement a non-weak version of this probe function - * to tell whether they are available. If they don't override then - * we'll return -ENODEV. - */ return -ENODEV; }
Right now there is one arch (sparc64) that selects HAVE_NMI_WATCHDOG without selecting HAVE_HARDLOCKUP_DETECTOR_ARCH. Because of that one architecture, we have some special case code in the watchdog core to handle the fact that watchdog_hardlockup_probe() isn't implemented. Let's implement watchdog_hardlockup_probe() for sparc64 and get rid of the special case. As a side effect of doing this, code inspection tells us that we could fix a minor bug where the system won't properly realize that NMI watchdogs are disabled. Specifically, on powerpc if CONFIG_PPC_WATCHDOG is turned off the arch might still select CONFIG_HAVE_HARDLOCKUP_DETECTOR_ARCH which selects CONFIG_HAVE_NMI_WATCHDOG. Since CONFIG_PPC_WATCHDOG was off then nothing will override the "weak" watchdog_hardlockup_probe() and we'll fallback to looking at CONFIG_HAVE_NMI_WATCHDOG. Suggested-by: Petr Mladek <pmladek@suse.com> Signed-off-by: Douglas Anderson <dianders@chromium.org> --- Though this does fix a minor bug, I didn't mark this as "Fixes" because it's super minor. One could also argue that this wasn't a bug at all but simply was never an implemented feature. The code that added some amount of dynamicness here was commit a994a3147e4c ("watchdog/hardlockup/perf: Implement init time detection of perf") which, as per the title, was only intending to make "perf" dynamic. The old NMI watchdog presumably has never been handled dynamically. arch/Kconfig | 3 ++- arch/sparc/kernel/nmi.c | 5 +++++ kernel/watchdog.c | 13 ------------- 3 files changed, 7 insertions(+), 14 deletions(-)