The patch titled block: prevent possible io_context->refcount overflow has been added to the -mm tree. Its filename is block-prevent-possible-io_context-refcount-overflow.patch Before you just go and hit "reply", please: a) Consider who else should be cc'ed b) Prefer to cc a suitable mailing list as well c) Ideally: find the original patch on the mailing list and do a reply-to-all to that, adding suitable additional cc's *** Remember to use Documentation/SubmitChecklist when testing your code *** See http://userweb.kernel.org/~akpm/stuff/added-to-mm.txt to find out what to do about this The current -mm tree may be found at http://userweb.kernel.org/~akpm/mmotm/ ------------------------------------------------------ Subject: block: prevent possible io_context->refcount overflow From: Nikanth Karthikesan <knikanth@xxxxxxxxxx> Currently io_context has an atomic_t(32-bit) as refcount. In the case of cfq, for each device against whcih a task does I/O, a reference to the io_context would be taken. And when there are multiple process sharing io_contexts(CLONE_IO) would also have a reference to the same io_context. Theoretically the possible maximum number of processes sharing the same io_context + the number of disks/cfq_data referring to the same io_context can overflow the 32-bit counter on a very high-end machine. Even though it is an improbable case, let us make it atomic_long_t. Signed-off-by: Nikanth Karthikesan <knikanth@xxxxxxx> Cc: Jens Axboe <jens.axboe@xxxxxxxxxx> Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> --- block/blk-ioc.c | 12 ++++++------ block/cfq-iosched.c | 2 +- include/linux/iocontext.h | 6 +++--- 3 files changed, 10 insertions(+), 10 deletions(-) diff -puN block/blk-ioc.c~block-prevent-possible-io_context-refcount-overflow block/blk-ioc.c --- a/block/blk-ioc.c~block-prevent-possible-io_context-refcount-overflow +++ a/block/blk-ioc.c @@ -35,9 +35,9 @@ int put_io_context(struct io_context *io if (ioc == NULL) return 1; - BUG_ON(atomic_read(&ioc->refcount) == 0); + BUG_ON(atomic_long_read(&ioc->refcount) == 0); - if (atomic_dec_and_test(&ioc->refcount)) { + if (atomic_long_dec_and_test(&ioc->refcount)) { rcu_read_lock(); if (ioc->aic && ioc->aic->dtor) ioc->aic->dtor(ioc->aic); @@ -90,7 +90,7 @@ struct io_context *alloc_io_context(gfp_ ret = kmem_cache_alloc_node(iocontext_cachep, gfp_flags, node); if (ret) { - atomic_set(&ret->refcount, 1); + atomic_long_set(&ret->refcount, 1); atomic_set(&ret->nr_tasks, 1); spin_lock_init(&ret->lock); ret->ioprio_changed = 0; @@ -151,7 +151,7 @@ struct io_context *get_io_context(gfp_t ret = current_io_context(gfp_flags, node); if (unlikely(!ret)) break; - } while (!atomic_inc_not_zero(&ret->refcount)); + } while (!atomic_long_inc_not_zero(&ret->refcount)); return ret; } @@ -163,8 +163,8 @@ void copy_io_context(struct io_context * struct io_context *dst = *pdst; if (src) { - BUG_ON(atomic_read(&src->refcount) == 0); - atomic_inc(&src->refcount); + BUG_ON(atomic_long_read(&src->refcount) == 0); + atomic_long_inc(&src->refcount); put_io_context(dst); *pdst = src; } diff -puN block/cfq-iosched.c~block-prevent-possible-io_context-refcount-overflow block/cfq-iosched.c --- a/block/cfq-iosched.c~block-prevent-possible-io_context-refcount-overflow +++ a/block/cfq-iosched.c @@ -1282,7 +1282,7 @@ static void cfq_dispatch_request(struct if (!cfqd->active_cic) { struct cfq_io_context *cic = RQ_CIC(rq); - atomic_inc(&cic->ioc->refcount); + atomic_long_inc(&cic->ioc->refcount); cfqd->active_cic = cic; } } diff -puN include/linux/iocontext.h~block-prevent-possible-io_context-refcount-overflow include/linux/iocontext.h --- a/include/linux/iocontext.h~block-prevent-possible-io_context-refcount-overflow +++ a/include/linux/iocontext.h @@ -64,7 +64,7 @@ struct cfq_io_context { * and kmalloc'ed. These could be shared between processes. */ struct io_context { - atomic_t refcount; + atomic_long_t refcount; atomic_t nr_tasks; /* all the fields below are protected by this lock */ @@ -91,8 +91,8 @@ static inline struct io_context *ioc_tas * if ref count is zero, don't allow sharing (ioc is going away, it's * a race). */ - if (ioc && atomic_inc_not_zero(&ioc->refcount)) { - atomic_inc(&ioc->nr_tasks); + if (ioc && atomic_long_inc_not_zero(&ioc->refcount)) { + atomic_long_inc(&ioc->refcount); return ioc; } _ Patches currently in -mm which might be from knikanth@xxxxxxxxxx are block-prevent-possible-io_context-refcount-overflow.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