Commit d936cf9b39b06c8d2e0d7fb5e7b4f176e18dec69

Authored by Hugh Dickins
Committed by Linus Torvalds
1 parent edc315fd22

badpage: ratelimit print_bad_pte and bad_page

print_bad_pte() and bad_page() might each need ratelimiting - especially
for their dump_stacks, almost never of interest, yet not quite
dispensible.  Correlating corruption across neighbouring entries can be
very helpful, so allow a burst of 60 reports before keeping quiet for the
remainder of that minute (or allow a steady drip of one report per
second).

Signed-off-by: Hugh Dickins <hugh@veritas.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Showing 2 changed files with 48 additions and 1 deletions Side-by-side Diff

... ... @@ -383,6 +383,29 @@
383 383 pmd_t *pmd = pmd_offset(pud, addr);
384 384 struct address_space *mapping;
385 385 pgoff_t index;
  386 + static unsigned long resume;
  387 + static unsigned long nr_shown;
  388 + static unsigned long nr_unshown;
  389 +
  390 + /*
  391 + * Allow a burst of 60 reports, then keep quiet for that minute;
  392 + * or allow a steady drip of one report per second.
  393 + */
  394 + if (nr_shown == 60) {
  395 + if (time_before(jiffies, resume)) {
  396 + nr_unshown++;
  397 + return;
  398 + }
  399 + if (nr_unshown) {
  400 + printk(KERN_EMERG
  401 + "Bad page map: %lu messages suppressed\n",
  402 + nr_unshown);
  403 + nr_unshown = 0;
  404 + }
  405 + nr_shown = 0;
  406 + }
  407 + if (nr_shown++ == 0)
  408 + resume = jiffies + 60 * HZ;
386 409  
387 410 mapping = vma->vm_file ? vma->vm_file->f_mapping : NULL;
388 411 index = linear_page_index(vma, addr);
... ... @@ -223,6 +223,30 @@
223 223  
224 224 static void bad_page(struct page *page)
225 225 {
  226 + static unsigned long resume;
  227 + static unsigned long nr_shown;
  228 + static unsigned long nr_unshown;
  229 +
  230 + /*
  231 + * Allow a burst of 60 reports, then keep quiet for that minute;
  232 + * or allow a steady drip of one report per second.
  233 + */
  234 + if (nr_shown == 60) {
  235 + if (time_before(jiffies, resume)) {
  236 + nr_unshown++;
  237 + goto out;
  238 + }
  239 + if (nr_unshown) {
  240 + printk(KERN_EMERG
  241 + "Bad page state: %lu messages suppressed\n",
  242 + nr_unshown);
  243 + nr_unshown = 0;
  244 + }
  245 + nr_shown = 0;
  246 + }
  247 + if (nr_shown++ == 0)
  248 + resume = jiffies + 60 * HZ;
  249 +
226 250 printk(KERN_EMERG "Bad page state in process %s pfn:%05lx\n",
227 251 current->comm, page_to_pfn(page));
228 252 printk(KERN_EMERG
... ... @@ -232,7 +256,7 @@
232 256 printk(KERN_EMERG "Trying to fix it up, but a reboot is needed\n");
233 257  
234 258 dump_stack();
235   -
  259 +out:
236 260 /* Leave bad fields for debug, except PageBuddy could make trouble */
237 261 __ClearPageBuddy(page);
238 262 add_taint(TAINT_BAD_PAGE);