The patch titled Subject: percpu_rw_semaphore-reimplement-to-not-block-the-readers-unnecessarilyfix has been removed from the -mm tree. Its filename was percpu_rw_semaphore-reimplement-to-not-block-the-readers-unnecessari-lyfix.patch This patch was dropped because it was folded into percpu_rw_semaphore-reimplement-to-not-block-the-readers-unnecessarily.patch ------------------------------------------------------ From: Oleg Nesterov <oleg@xxxxxxxxxx> Subject: percpu_rw_semaphore-reimplement-to-not-block-the-readers-unnecessarilyfix More include's and more comments, no changes in code. To remind, once/if I am sure you agree with this patch I'll send 2 additional and simple patches: 1. lockdep annotations 2. CONFIG_PERCPU_RWSEM It seems that we can do much more improvements to a) speedup the writers and b) make percpu_rw_semaphore more useful, but not right now. Signed-off-by: Oleg Nesterov <oleg@xxxxxxxxxx> Cc: Paul E. McKenney <paulmck@xxxxxxxxxxxxxxxxxx> Cc: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx> Cc: Mikulas Patocka <mpatocka@xxxxxxxxxx> Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx> Cc: Ingo Molnar <mingo@xxxxxxx> Cc: Srikar Dronamraju <srikar@xxxxxxxxxxxxxxxxxx> Cc: Ananth N Mavinakayanahalli <ananth@xxxxxxxxxx> Cc: Anton Arapov <anton@xxxxxxxxxx> Cc: Jens Axboe <axboe@xxxxxxxxx> Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> --- lib/percpu-rwsem.c | 35 +++++++++++++++++++++++++++++++++-- 1 file changed, 33 insertions(+), 2 deletions(-) diff -puN lib/percpu-rwsem.c~percpu_rw_semaphore-reimplement-to-not-block-the-readers-unnecessari-lyfix lib/percpu-rwsem.c --- a/lib/percpu-rwsem.c~percpu_rw_semaphore-reimplement-to-not-block-the-readers-unnecessari-lyfix +++ a/lib/percpu-rwsem.c @@ -1,6 +1,11 @@ +#include <linux/mutex.h> +#include <linux/rwsem.h> +#include <linux/percpu.h> +#include <linux/wait.h> #include <linux/percpu-rwsem.h> #include <linux/rcupdate.h> #include <linux/sched.h> +#include <linux/errno.h> int percpu_init_rwsem(struct percpu_rw_semaphore *brw) { @@ -21,6 +26,29 @@ void percpu_free_rwsem(struct percpu_rw_ brw->fast_read_ctr = NULL; /* catch use after free bugs */ } +/* + * This is the fast-path for down_read/up_read, it only needs to ensure + * there is no pending writer (!mutex_is_locked() check) and inc/dec the + * fast per-cpu counter. The writer uses synchronize_sched_expedited() to + * serialize with the preempt-disabled section below. + * + * The nontrivial part is that we should guarantee acquire/release semantics + * in case when + * + * R_W: down_write() comes after up_read(), the writer should see all + * changes done by the reader + * or + * W_R: down_read() comes after up_write(), the reader should see all + * changes done by the writer + * + * If this helper fails the callers rely on the normal rw_semaphore and + * atomic_dec_and_test(), so in this case we have the necessary barriers. + * + * But if it succeeds we do not have any barriers, mutex_is_locked() or + * __this_cpu_add() below can be reordered with any LOAD/STORE done by the + * reader inside the critical section. See the comments in down_write and + * up_write below. + */ static bool update_fast_ctr(struct percpu_rw_semaphore *brw, unsigned int val) { bool success = false; @@ -98,6 +126,7 @@ void percpu_down_write(struct percpu_rw_ * * 3. Ensures that if any reader has exited its critical section via * fast-path, it executes a full memory barrier before we return. + * See R_W case in the comment above update_fast_ctr(). */ synchronize_sched_expedited(); @@ -116,8 +145,10 @@ void percpu_up_write(struct percpu_rw_se /* allow the new readers, but only the slow-path */ up_write(&brw->rw_sem); - /* insert the barrier before the next fast-path in down_read */ + /* + * Insert the barrier before the next fast-path in down_read, + * see W_R case in the comment above update_fast_ctr(). + */ synchronize_sched_expedited(); - mutex_unlock(&brw->writer_mutex); } _ Patches currently in -mm which might be from oleg@xxxxxxxxxx are origin.patch percpu_rw_semaphore-reimplement-to-not-block-the-readers-unnecessarily.patch percpu_rw_semaphore-kill-writer_mutex-add-write_ctr.patch percpu_rw_semaphore-add-the-lockdep-annotations.patch percpu_rw_semaphore-introduce-config_percpu_rwsem.patch ptrace-introduce-ptrace_o_exitkill.patch procfs-add-ability-to-plug-in-auxiliary-fdinfo-providers.patch fs-eventfd-add-procfs-fdinfo-helper.patch fs-epoll-add-procfs-fdinfo-helper-v2.patch fs-epoll-add-procfs-fdinfo-helper-v2-fs-epoll-drop-enabled-field-from-fdinfo-output.patch fdinfo-show-sigmask-for-signalfd-fd-v3.patch fs-exportfs-escape-nil-dereference-if-no-s_export_op-present.patch fs-exportfs-add-exportfs_encode_inode_fh-helper.patch fs-notify-add-procfs-fdinfo-helper-v7.patch fs-notify-add-procfs-fdinfo-helper-v7-add-missing-space-after-prefix.patch fs-notify-add-procfs-fdinfo-helper-v7-dont-forget-to-provide-fhandle-for-inode-fanotify.patch fs-notify-add-procfs-fdinfo-helper-v7-fs-fanotify-ddd-missing-pieces-in-fdinfo-for-ability-to-call-fanotify_init.patch docs-add-documentation-about-proc-pid-fdinfo-fd-output.patch docs-add-documentation-about-proc-pid-fdinfo-fd-output-fix.patch fs-fanotify-add-mflags-field-to-fanotify-output.patch docs-update-documentation-about-proc-pid-fdinfo-fd-fanotify-output.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