tree: https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master head: 0b58e108042b0ed28a71cd7edf5175999955b233 commit: 3be88389f46263f166973e80e528dcc9268e24cb [9885/10451] rcu: Summarize RCU CPU stall warnings during CSD-lock stalls config: x86_64-randconfig-a014-20211016 (https://download.01.org/0day-ci/archive/20240703/202407031722.nBIh2u7x-lkp@xxxxxxxxx/config) compiler: gcc-13 (Ubuntu 13.2.0-4ubuntu3) 13.2.0 reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20240703/202407031722.nBIh2u7x-lkp@xxxxxxxxx/reproduce) If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags | Reported-by: kernel test robot <lkp@xxxxxxxxx> | Closes: https://lore.kernel.org/oe-kbuild-all/202407031722.nBIh2u7x-lkp@xxxxxxxxx/ All errors (new ones prefixed by >>): ld: vmlinux.o: in function `check_cpu_stall': >> kernel/rcu/tree_stall.h:797:(.text+0x273b89): undefined reference to `csd_lock_is_stuck' vim +797 kernel/rcu/tree_stall.h 724 725 static void check_cpu_stall(struct rcu_data *rdp) 726 { 727 bool self_detected; 728 unsigned long gs1; 729 unsigned long gs2; 730 unsigned long gps; 731 unsigned long j; 732 unsigned long jn; 733 unsigned long js; 734 struct rcu_node *rnp; 735 736 lockdep_assert_irqs_disabled(); 737 if ((rcu_stall_is_suppressed() && !READ_ONCE(rcu_kick_kthreads)) || 738 !rcu_gp_in_progress()) 739 return; 740 rcu_stall_kick_kthreads(); 741 742 /* 743 * Check if it was requested (via rcu_cpu_stall_reset()) that the FQS 744 * loop has to set jiffies to ensure a non-stale jiffies value. This 745 * is required to have good jiffies value after coming out of long 746 * breaks of jiffies updates. Not doing so can cause false positives. 747 */ 748 if (READ_ONCE(rcu_state.nr_fqs_jiffies_stall) > 0) 749 return; 750 751 j = jiffies; 752 753 /* 754 * Lots of memory barriers to reject false positives. 755 * 756 * The idea is to pick up rcu_state.gp_seq, then 757 * rcu_state.jiffies_stall, then rcu_state.gp_start, and finally 758 * another copy of rcu_state.gp_seq. These values are updated in 759 * the opposite order with memory barriers (or equivalent) during 760 * grace-period initialization and cleanup. Now, a false positive 761 * can occur if we get an new value of rcu_state.gp_start and a old 762 * value of rcu_state.jiffies_stall. But given the memory barriers, 763 * the only way that this can happen is if one grace period ends 764 * and another starts between these two fetches. This is detected 765 * by comparing the second fetch of rcu_state.gp_seq with the 766 * previous fetch from rcu_state.gp_seq. 767 * 768 * Given this check, comparisons of jiffies, rcu_state.jiffies_stall, 769 * and rcu_state.gp_start suffice to forestall false positives. 770 */ 771 gs1 = READ_ONCE(rcu_state.gp_seq); 772 smp_rmb(); /* Pick up ->gp_seq first... */ 773 js = READ_ONCE(rcu_state.jiffies_stall); 774 smp_rmb(); /* ...then ->jiffies_stall before the rest... */ 775 gps = READ_ONCE(rcu_state.gp_start); 776 smp_rmb(); /* ...and finally ->gp_start before ->gp_seq again. */ 777 gs2 = READ_ONCE(rcu_state.gp_seq); 778 if (gs1 != gs2 || 779 ULONG_CMP_LT(j, js) || 780 ULONG_CMP_GE(gps, js)) 781 return; /* No stall or GP completed since entering function. */ 782 rnp = rdp->mynode; 783 jn = jiffies + ULONG_MAX / 2; 784 self_detected = READ_ONCE(rnp->qsmask) & rdp->grpmask; 785 if (rcu_gp_in_progress() && 786 (self_detected || ULONG_CMP_GE(j, js + RCU_STALL_RAT_DELAY)) && 787 cmpxchg(&rcu_state.jiffies_stall, js, jn) == js) { 788 /* 789 * If a virtual machine is stopped by the host it can look to 790 * the watchdog like an RCU stall. Check to see if the host 791 * stopped the vm. 792 */ 793 if (kvm_check_and_clear_guest_paused()) 794 return; 795 796 rcu_stall_notifier_call_chain(RCU_STALL_NOTIFY_NORM, (void *)j - gps); > 797 if (READ_ONCE(csd_lock_suppress_rcu_stall) && csd_lock_is_stuck()) { 798 pr_err("INFO: %s detected stall, but suppressed full report due to a stuck CSD-lock.\n", rcu_state.name); 799 } else if (self_detected) { 800 /* We haven't checked in, so go dump stack. */ 801 print_cpu_stall(gps); 802 } else { 803 /* They had a few time units to dump stack, so complain. */ 804 print_other_cpu_stall(gs2, gps); 805 } 806 807 if (READ_ONCE(rcu_cpu_stall_ftrace_dump)) 808 rcu_ftrace_dump(DUMP_ALL); 809 810 if (READ_ONCE(rcu_state.jiffies_stall) == jn) { 811 jn = jiffies + 3 * rcu_jiffies_till_stall_check() + 3; 812 WRITE_ONCE(rcu_state.jiffies_stall, jn); 813 } 814 } 815 } 816 -- 0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki
