If a system experiences a lot of memory failures, then any associated printk() output really needs to be rate-limited. I noticed this while running selftests/mm/uffd-unit-tests, which logs 12,305 lines of output, adding (on my system) an extra 97 seconds of runtime due to printk time. The test normally only takes about 10 seconds, and the enclosing set of mm selftests normally takes 305 seconds, so the additional 97 seconds really hurts. Generating lots of memory poisoning events seems like a valid use case, by which I mean that this is not just a testing artifact. And that's why the fix applies to the code that directly generates the output, rather than the selftest that triggers it. With this patch, all but 10 lines are suppressed, thus speeding up that particular selftest by 90% (runtime drops from 107 seconds, to 10.6 seconds). Signed-off-by: John Hubbard <jhubbard@xxxxxxxxxx> --- arch/x86/mm/fault.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c index bba4e020dd64..e4f3c7721f45 100644 --- a/arch/x86/mm/fault.c +++ b/arch/x86/mm/fault.c @@ -928,7 +928,7 @@ do_sigbus(struct pt_regs *regs, unsigned long error_code, unsigned long address, struct task_struct *tsk = current; unsigned lsb = 0; - pr_err( + pr_err_ratelimited( "MCE: Killing %s:%d due to hardware memory corruption fault at %lx\n", tsk->comm, tsk->pid, address); if (fault & VM_FAULT_HWPOISON_LARGE) base-commit: dccb07f2914cdab2ac3a5b6c98406f765acab803 prerequisite-patch-id: b901ece2a5b78503e2fb5480f20e304d36a0ea27 -- 2.45.0
