þri., 26. feb. 2019 kl. 13:49 skrifaði Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx>: > > On 2019-02-22 14:10:09 [+0000], Andri Yngvason wrote: > > Has someone run into the same issue? > > Does this make a difference: > > diff --git a/arch/x86/entry/entry_32.S b/arch/x86/entry/entry_32.S > index 0169c257cfffe..e6f61c813baf6 100644 > --- a/arch/x86/entry/entry_32.S > +++ b/arch/x86/entry/entry_32.S > @@ -773,15 +773,15 @@ ENTRY(resume_kernel) > > # atleast preempt count == 0 ? > cmpl $_PREEMPT_ENABLED,PER_CPU_VAR(__preempt_count) > - jne restore_all > + jne restore_all_kernel > > movl PER_CPU_VAR(current_task), %ebp > cmpl $0,TASK_TI_preempt_lazy_count(%ebp) # non-zero preempt_lazy_count ? > - jnz restore_all > + jnz restore_all_kernel > > testl $_TIF_NEED_RESCHED_LAZY, TASK_TI_flags(%ebp) > - jz restore_all > - test_int_off: > + jz restore_all_kernel > +test_int_off: > #endif > testl $X86_EFLAGS_IF, PT_EFLAGS(%esp) # interrupts off (exception path) ? > jz restore_all_kernel > Yes, indeed it makes a lot of difference. Thanks! Cheers, Andri
