The chipset's pins are often multi-used because greedy usage compares to a finite pin number. This requests a carefully configuration in firmware or OS. If not, it may contribute to most of irq flood cases, which appears as a soft lockup issue on Linux. Strictly speaking, soft lockup and irq flood are different things to overcome. And it is helpful to make users aware of that situation for prime time.
This series shows the irq statistics when soft lockup. The statistics can be used to evaluate the possibility of irq flood and as a rough evaluated input to the kernel parameter "irqstorm_limit" in [1]. It is not easy to find a common way to work around irq flood, which may be raised by different root cause. To now, it is still a open question. Thomas and Guilherme suggested patches to suppress the odd irq in different situation. [1].[2] [1]: https://lore.kernel.org/lkml/87tuueftou....@nanos.tec.linutronix.de/ [2]: https://lore.kernel.org/linux-pci/20181018183721.27467-1-gpicc...@canonical.com/ Cc: Thomas Gleixner <t...@linutronix.de> Cc: Jisheng Zhang <jisheng.zh...@synaptics.com> Cc: "Peter Zijlstra (Intel)" <pet...@infradead.org> Cc: Vlastimil Babka <vba...@suse.cz> Cc: Andrew Morton <a...@linux-foundation.org> Cc: "Guilherme G. Piccoli" <gpicc...@canonical.com> Cc: Petr Mladek <pmla...@suse.com> Cc: ke...@lists.infradead.org To: linux-kernel@vger.kernel.org Pingfan Liu (3): x86/irq: account the unused irq kernel/watchdog: make watchdog_touch_ts more accurate by using nanosecond kernel/watchdog: use soft lockup to detect irq flood arch/x86/kernel/irq.c | 1 + include/linux/kernel_stat.h | 1 + kernel/watchdog.c | 37 ++++++++++++++++++++++++++----------- 3 files changed, 28 insertions(+), 11 deletions(-) -- 2.7.5
