| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <202206050354.7iGBqref-lkp@intel.com>
Date: Sun, 5 Jun 2022 03:43:09 +0800
From: kernel test robot <lkp@...el.com>
To: David Howells <dhowells@...hat.com>
Cc: kbuild-all@...ts.01.org,
GNU/Weeb Mailing List <gwml@...r.gnuweeb.org>,
linux-kernel@...r.kernel.org
Subject: [ammarfaizi2-block:dhowells/linux-fs/netfs-linked-list 58/59]
fs/netfs/buffered_flush.c:408:30: sparse: sparse: incompatible types in
comparison expression (different type sizes):
tree: https://github.com/ammarfaizi2/linux-block dhowells/linux-fs/netfs-linked-list
head: 22ecc2fcdab4616e624408911ec1d54644e82030
commit: ff509dc49891f934a7a774e277b87729018db753 [58/59] netfs: Use flush from afs_writeback
config: i386-randconfig-s002 (https://download.01.org/0day-ci/archive/20220605/202206050354.7iGBqref-lkp@intel.com/config)
compiler: gcc-11 (Debian 11.3.0-1) 11.3.0
reproduce:
# apt-get install sparse
# sparse version: v0.6.4-18-g56afb504-dirty
# https://github.com/ammarfaizi2/linux-block/commit/ff509dc49891f934a7a774e277b87729018db753
git remote add ammarfaizi2-block https://github.com/ammarfaizi2/linux-block
git fetch --no-tags ammarfaizi2-block dhowells/linux-fs/netfs-linked-list
git checkout ff509dc49891f934a7a774e277b87729018db753
# save the config file
mkdir build_dir && cp config build_dir/.config
make W=1 C=1 CF='-fdiagnostic-prefix -D__CHECK_ENDIAN__' O=build_dir ARCH=i386 SHELL=/bin/bash fs/netfs/
If you fix the issue, kindly add following tag where applicable
Reported-by: kernel test robot <lkp@...el.com>
sparse warnings: (new ones prefixed by >>)
>> fs/netfs/buffered_flush.c:408:30: sparse: sparse: incompatible types in comparison expression (different type sizes):
>> fs/netfs/buffered_flush.c:408:30: sparse: unsigned int *
>> fs/netfs/buffered_flush.c:408:30: sparse: unsigned long *
fs/netfs/buffered_flush.c:411:30: sparse: sparse: incompatible types in comparison expression (different type sizes):
fs/netfs/buffered_flush.c:411:30: sparse: unsigned int *
fs/netfs/buffered_flush.c:411:30: sparse: unsigned long *
fs/netfs/buffered_flush.c:413:17: sparse: sparse: incompatible types in comparison expression (different type sizes):
fs/netfs/buffered_flush.c:413:17: sparse: unsigned int *
fs/netfs/buffered_flush.c:413:17: sparse: unsigned long *
vim +408 fs/netfs/buffered_flush.c
372
373 /*
374 * Extend the region to be written back to include subsequent contiguously
375 * dirty pages if possible, but don't sleep while doing so.
376 *
377 * If this page holds new content, then we can include filler zeros in the
378 * writeback.
379 */
380 static void netfs_extend_writeback(struct netfs_io_request *wreq,
381 struct writeback_control *wbc,
382 struct netfs_i_context *ctx,
383 struct netfs_dirty_region *region)
384 {
385 struct folio_batch fbatch;
386 struct folio *folio;
387 unsigned int i;
388 pgoff_t index = wreq->last + 1, stop_mask, stop_at, max_pages;
389 ssize_t max_size;
390 size_t align, hard_align;
391 bool stop = true, dirty;
392 int ret;
393
394 XA_STATE(xas, &wreq->mapping->i_pages, index);
395
396 _enter("%lx", index);
397
398 /* We have a number of criteria by which we can decide where to stop
399 * extension of this writeback:
400 *
401 * 1) The maximum I/O size (but wbacks can be subdivided)
402 * 2) Boundaries within the filesystem (eg. ceph object size)
403 * 3) Local boundaries (cache granules, VM radix node sizes)
404 * 4) Content crypto/compression boundaries
405 */
406 hard_align = PAGE_SIZE;
407 if (ctx->min_bshift || ctx->crypto_bshift)
> 408 hard_align = max(hard_align,
409 1UL << max(ctx->min_bshift, ctx->crypto_bshift));
410 if (ctx->cache_order)
411 hard_align = max(hard_align, PAGE_SIZE << ctx->cache_order);
412
413 align = min(hard_align, XA_CHUNK_SIZE * PAGE_SIZE);
414 if (wreq->alignment > align)
415 align = wreq->alignment;
416
417 stop_mask = (align - 1) / PAGE_SIZE;
418
419 max_size = XA_CHUNK_SIZE * PAGE_SIZE;
420 if (wreq->wsize > max_size)
421 max_size = roundup_pow_of_two(wreq->wsize);
422
423 _debug("LIMITS al=%zx ha=%zx mx=%zx", align, hard_align, max_size);
424
425 max_pages = max_size / PAGE_SIZE;
426 max_pages &= ~stop_mask;
427 _debug("MAX_PAGES %lx %lx", max_pages, stop_mask);
428 if (wreq->last - wreq->first + 1 >= max_pages) {
429 _leave(" [already hit wsize %lx %lx]",
430 wreq->last - wreq->first + 1, max_pages);
431 return;
432 }
433
434 stop_at = wreq->first + max_pages;
435 if (stop_at < wreq->first)
436 stop_at = ULONG_MAX;
437 else
438 stop_at = round_down(stop_at, align / PAGE_SIZE);
439 _debug("STOP_AT %lx (%lx %lx %lx)", stop_at, wreq->first, max_pages, align / PAGE_SIZE);
440
441 if (index >= stop_at || wbc->nr_to_write <= 0 || wreq->len > max_size) {
442 _leave(" [prestop]");
443 return;
444 }
445
446 do {
447 /* Firstly, we gather up a batch of contiguous dirty folios
448 * under the RCU read lock - but we can't clear the dirty flags
449 * there if any of those folios are mapped.
450 */
451 folio_batch_init(&fbatch);
452 _debug("extend %lx %lx", index, xas.xa_index);
453 rcu_read_lock();
454
455 xas_for_each(&xas, folio, ULONG_MAX) {
456 stop = true;
457 if (xas_retry(&xas, folio))
458 continue;
459 if (xa_is_value(folio))
460 break;
461 if (folio_index(folio) != index)
462 break;
463
464 if (!folio_try_get_rcu(folio)) {
465 xas_reset(&xas);
466 continue;
467 }
468
469 /* Has the folio moved or been split? */
470 if (unlikely(folio != xas_reload(&xas))) {
471 folio_put(folio);
472 break;
473 }
474
475 if (!folio_trylock(folio)) {
476 folio_put(folio);
477 break;
478 }
479 if (!folio_test_dirty(folio) ||
480 folio_test_writeback(folio)) {
481 folio_unlock(folio);
482 folio_put(folio);
483 break;
484 }
485
486 wreq->len += folio_size(folio);
487 index += folio_nr_pages(folio);
488 stop = index >= stop_at || wbc->nr_to_write <= 0;
489
490 if (!folio_batch_add(&fbatch, folio))
491 break;
492 if (stop)
493 break;
494 }
495
496 if (!stop)
497 xas_pause(&xas);
498 rcu_read_unlock();
499
500 /* Now, if we obtained any pages, we can shift them to being
501 * writable and mark them for caching.
502 */
503 if (!folio_batch_count(&fbatch))
504 break;
505
506 for (i = 0; i < folio_batch_count(&fbatch); i++) {
507 folio = fbatch.folios[i];
508 //_debug("found [%x] %lx", i, folio->index);
509 trace_netfs_folio_dirty(wreq->mapping, folio,
510 netfs_folio_trace_store_ex);
511
512 dirty = folio_clear_dirty_for_io(folio);
513 if (folio_start_writeback(folio)) {
514 _debug("! no wb");
515 goto nomem_redirty;
516 }
517
518 ret = netfs_xa_store_and_mark(&wreq->buffer, folio->index, folio,
519 true, true, dirty, GFP_NOFS);
520 if (ret < 0) {
521 _debug("! no buffer");
522 goto nomem_cancel_wb;
523 }
524 wbc->nr_to_write -= folio_nr_pages(folio);
525 index = folio_next_index(folio);
526 wreq->last = index - 1;
527 folio_unlock(folio);
528 }
529
530 cond_resched();
531 } while (!stop);
532
533 _leave(" ok [%zx]", wreq->last);
534 return;
535
536 nomem_cancel_wb:
537 folio_end_writeback(folio);
538 nomem_redirty:
539 if (dirty)
540 folio_redirty_for_writepage(wbc, folio);
541 for (; i < folio_batch_count(&fbatch); i++) {
542 folio_unlock(folio);
543 folio_put(folio);
544 }
545 _leave(" cancel [%zx]", wreq->last);
546 }
547
--
0-DAY CI Kernel Test Service
https://01.org/lkp
Powered by blists - more mailing lists