Hi, Thomas, On Thu, Aug 24, 2023 at 5:37 PM Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote: > > On Tue, Aug 22 2023 at 12:02, Huacai Chen wrote: > > + * This function may be called in NMI context, so we cannot use ktime_get_ns() > > + * and ktime_get_coarse_ns(). Instead, we use their inaccurate but safe friends > > + * ktime_get_mono_fast_ns() and ktime_get_seconds() which will cause rcu_state. > > + * jiffies_stall to be a little large than expected (harmless and safer). > > What's inaccurate about ktime_get_mono_fast_ns()? Bogus comments are > even worse than no comments. ktime_get_mono_fast_ns() is not as accurate as ktime_get_ns(), I get this conclusion from: * So if the update lowers the slope, readers who are forced to the * not yet updated second array are still using the old steeper slope. > > > */ > > void rcu_cpu_stall_reset(void) > > { > > + u64 curr, last, delta; > > + > > + curr = ktime_get_mono_fast_ns(); > > + last = ktime_get_seconds() * NSEC_PER_SEC; > > So this will trigger a warning when someone debugs suspend with KGDB. Yes, ktime_get_seconds() may cause a warning, I haven't noticed this before. > > It seems the approach taken here seems to be to throw stuff at the wall > and see what sticks. I don't understand what's meaning, but I believe your advice in another thread is the best solution, so let me try. Huacai > > Thanks, > > tglx