The patch titled rwsem: fully separate code paths to wake writers vs readers has been added to the -mm tree. Its filename is rwsem-fully-separate-code-pathes-to-wake-writers-vs-readers.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: rwsem: fully separate code paths to wake writers vs readers From: Michel Lespinasse <walken@xxxxxxxxxx> This is in preparation for later changes in the series. In __rwsem_do_wake(), the first queued waiter is checked first in order to determine whether it's a writer or a reader. The code paths diverge at this point. The code that checks and increments the rwsem active count is duplicated on both sides - the point is that later changes in the series will be able to independently modify both sides. Signed-off-by: Michel Lespinasse <walken@xxxxxxxxxx> Cc: David Howells <dhowells@xxxxxxxxxx> Cc: Mike Waychison <mikew@xxxxxxxxxx> Cc: Suleiman Souhlal <suleiman@xxxxxxxxxx> Cc: Ying Han <yinghan@xxxxxxxxxx> Cc: Ingo Molnar <mingo@xxxxxxx> Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx> Cc: "H. Peter Anvin" <hpa@xxxxxxxxx> Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> --- lib/rwsem.c | 61 +++++++++++++++++++++++++++----------------------- 1 file changed, 34 insertions(+), 27 deletions(-) diff -puN lib/rwsem.c~rwsem-fully-separate-code-pathes-to-wake-writers-vs-readers lib/rwsem.c --- a/lib/rwsem.c~rwsem-fully-separate-code-pathes-to-wake-writers-vs-readers +++ a/lib/rwsem.c @@ -41,7 +41,7 @@ struct rwsem_waiter { * - if we come here from up_xxxx(), then: * - the 'active part' of count (&0x0000ffff) reached 0 (but may have changed) * - the 'waiting part' of count (&0xffff0000) is -ve (and will still be so) - * - there must be someone on the queue + * - there must be someone on the queue * - the spinlock must be held by the caller * - woken process blocks are discarded from the list after having task zeroed * - writers are only woken if downgrading is false @@ -54,26 +54,23 @@ __rwsem_do_wake(struct rw_semaphore *sem struct list_head *next; signed long oldcount, woken, loop; + waiter = list_entry(sem->wait_list.next, struct rwsem_waiter, list); + if (!(waiter->flags & RWSEM_WAITING_FOR_WRITE)) + goto readers_only; + if (downgrading) - goto dont_wake_writers; + goto out; - /* if we came through an up_xxxx() call, we only only wake someone up - * if we can transition the active part of the count from 0 -> 1 + /* There's a writer at the front of the queue - try to grant it the + * write lock. However, we only wake this writer if we can transition + * the active part of the count from 0 -> 1 */ - try_again: + try_again_write: oldcount = rwsem_atomic_update(RWSEM_ACTIVE_BIAS, sem) - RWSEM_ACTIVE_BIAS; if (oldcount & RWSEM_ACTIVE_MASK) - goto undo; - - waiter = list_entry(sem->wait_list.next, struct rwsem_waiter, list); - - /* try to grant a single write lock if there's a writer at the front - * of the queue - note we leave the 'active part' of the count - * incremented by 1 and the waiting part incremented by 0x00010000 - */ - if (!(waiter->flags & RWSEM_WAITING_FOR_WRITE)) - goto readers_only; + /* Someone grabbed the sem already */ + goto undo_write; /* We must be careful not to touch 'waiter' after we set ->task = NULL. * It is an allocated on the waiter's stack and may become invalid at @@ -87,18 +84,24 @@ __rwsem_do_wake(struct rw_semaphore *sem put_task_struct(tsk); goto out; - /* don't want to wake any writers */ - dont_wake_writers: - waiter = list_entry(sem->wait_list.next, struct rwsem_waiter, list); - if (waiter->flags & RWSEM_WAITING_FOR_WRITE) - goto out; + readers_only: + if (downgrading) + goto wake_readers; + + /* if we came through an up_xxxx() call, we only only wake someone up + * if we can transition the active part of the count from 0 -> 1 */ + try_again_read: + oldcount = rwsem_atomic_update(RWSEM_ACTIVE_BIAS, sem) + - RWSEM_ACTIVE_BIAS; + if (oldcount & RWSEM_ACTIVE_MASK) + /* Someone grabbed the sem already */ + goto undo_read; - /* grant an infinite number of read locks to the readers at the front - * of the queue - * - note we increment the 'active part' of the count by the number of - * readers before waking any processes up + wake_readers: + /* Grant an infinite number of read locks to the readers at the front + * of the queue. Note we increment the 'active part' of the count by + * the number of readers before waking any processes up. */ - readers_only: woken = 0; do { woken++; @@ -138,10 +141,14 @@ __rwsem_do_wake(struct rw_semaphore *sem /* undo the change to the active count, but check for a transition * 1->0 */ - undo: + undo_write: + if (rwsem_atomic_update(-RWSEM_ACTIVE_BIAS, sem) & RWSEM_ACTIVE_MASK) + goto out; + goto try_again_write; + undo_read: if (rwsem_atomic_update(-RWSEM_ACTIVE_BIAS, sem) & RWSEM_ACTIVE_MASK) goto out; - goto try_again; + goto try_again_read; } /* _ Patches currently in -mm which might be from walken@xxxxxxxxxx are x86-rwsem-stay-on-fast-path-when-count0-in-__up_write.patch x86-rwsem-minor-cleanups.patch rwsem-fully-separate-code-pathes-to-wake-writers-vs-readers.patch rwsem-lighter-active-count-checks-when-waking-up-readers.patch rwsem-let-rwsem_waiting_bias-represent-any-number-of-waiting-threads.patch rwsem-wake-queued-readers-when-writer-blocks-on-active-read-lock.patch rwsem-smaller-wrappers-around-rwsem_down_failed_common.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