The patch titled fix BUG at include/linux/mm.h:220! has been removed from the -mm tree. Its filename was mm-use-pagevec-to-rotate-reclaimable-page-fix-bug-at-include-linux-mmh220.patch This patch was dropped because it was folded into mm-use-pagevec-to-rotate-reclaimable-page.patch ------------------------------------------------------ Subject: fix BUG at include/linux/mm.h:220! From: Hugh Dickins <hugh@xxxxxxxxxxx> a.k.a. mm-use-pagevec-to-rotate-reclaimable-page-fix-3.patch put_page_testzero's VM_BUG_ON(atomic_read(&page->_count) == 0) fired a couple of times while heavily swapping (on -rc7-mm1 and -rc8-mm1). release_pages() has been using spin_lock_irq/spin_unlock_irq: but now that it's being called by pagevec_move_tail() with interrupts disabled, it must use spin_lock_irqsave/spin_unlock_irqrestore instead. Aside from it being generally wrong to enable IRQs behind the caller's back, here a pagevec might overflow, or (as I think must have happened to cause those BUGs) release_pages() may be preempted to another cpu while still acting on the pagevec of the cpu it started out with, and that owning cpu also come in to drain it. Hmm, no, one of the BUGs was with preemption disabled: so I don't fully understand it. Tested for 39 hours with rc8-mm1 on three 4cpu machines: no more BUGs. Signed-off-by: Hugh Dickins <hugh@xxxxxxxxxxx> Cc: Hisashi Hifumi <hifumi.hisashi@xxxxxxxxxxxxx> Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> --- diff -puN mm/swap.c~mm-use-pagevec-to-rotate-reclaimable-page-fix-bug-at-include-linux-mmh220 mm/swap.c --- a/mm/swap.c~mm-use-pagevec-to-rotate-reclaimable-page-fix-bug-at-include-linux-mmh220 +++ a/mm/swap.c @@ -296,6 +296,7 @@ void release_pages(struct page **pages, int i; struct pagevec pages_to_free; struct zone *zone = NULL; + unsigned long uninitialized_var(flags); pagevec_init(&pages_to_free, cold); for (i = 0; i < nr; i++) { @@ -303,7 +304,7 @@ void release_pages(struct page **pages, if (unlikely(PageCompound(page))) { if (zone) { - spin_unlock_irq(&zone->lru_lock); + spin_unlock_irqrestore(&zone->lru_lock, flags); zone = NULL; } put_compound_page(page); @@ -317,9 +318,10 @@ void release_pages(struct page **pages, struct zone *pagezone = page_zone(page); if (pagezone != zone) { if (zone) - spin_unlock_irq(&zone->lru_lock); + spin_unlock_irqrestore(&zone->lru_lock, + flags); zone = pagezone; - spin_lock_irq(&zone->lru_lock); + spin_lock_irqsave(&zone->lru_lock, flags); } VM_BUG_ON(!PageLRU(page)); __ClearPageLRU(page); @@ -328,7 +330,7 @@ void release_pages(struct page **pages, if (!pagevec_add(&pages_to_free, page)) { if (zone) { - spin_unlock_irq(&zone->lru_lock); + spin_unlock_irqrestore(&zone->lru_lock, flags); zone = NULL; } __pagevec_free(&pages_to_free); @@ -336,7 +338,7 @@ void release_pages(struct page **pages, } } if (zone) - spin_unlock_irq(&zone->lru_lock); + spin_unlock_irqrestore(&zone->lru_lock, flags); pagevec_free(&pages_to_free); } _ Patches currently in -mm which might be from hugh@xxxxxxxxxxx are origin.patch mm-clarify-__add_to_swap_cache-locking.patch radix-tree-use-indirect-bit.patch mm-use-pagevec-to-rotate-reclaimable-page.patch mm-use-pagevec-to-rotate-reclaimable-page-fix-bug-at-include-linux-mmh220.patch mm-use-pagevec-to-rotate-reclaimable-page-kill-redundancy-in-rotate_reclaimable_page.patch mm-use-pagevec-to-rotate-reclaimable-page-move_tail_pages-into-lru_add_drain.patch introduce-write_begin-write_end-aops-important-fix.patch reiserfs-convert-to-new-aops-fix.patch hostfs-convert-to-new-aops-fix.patch affs-convert-to-new-aops-fix.patch flush-cache-before-installing-new-page-at-migraton.patch flush-icache-before-set_pte-on-ia64-flush-icache-at-set_pte.patch flush-icache-before-set_pte-on-ia64-flush-icache-at-set_pte-fix.patch flush-icache-before-set_pte-on-ia64-flush-icache-at-set_pte-fix-update.patch maps-pssproportional-set-size-accounting-in-smaps.patch mm-shmemc-make-3-functions-static.patch lib-percpu_counter_sub.patch mm-per-device-dirty-threshold-fix.patch ext2-reservations.patch fix-for-ext2-reservation.patch ext2-balloc-use-io_error-label.patch memory-controller-memory-accounting-v7-fix-swapoff-breakage-however.patch exportfs-add-fid-type.patch exportfs-add-new-methods.patch shmem-new-export-ops.patch exportfs-remove-old-methods.patch exportfs-make-struct-export_operations-const.patch exportfs-update-documentation.patch prio_tree-debugging-patch.patch - To unsubscribe from this list: send the line "unsubscribe mm-commits" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html