From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1759029Ab3LBVGR (ORCPT ); Mon, 2 Dec 2013 16:06:17 -0500 Received: from mail.linuxfoundation.org ([140.211.169.12]:38402 "EHLO mail.linuxfoundation.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754021Ab3LBTNS (ORCPT ); Mon, 2 Dec 2013 14:13:18 -0500 From: Greg Kroah-Hartman To: linux-kernel@vger.kernel.org Cc: Greg Kroah-Hartman , stable@vger.kernel.org, Manfred Spraul , Mike Galbraith , Davidlohr Bueso , Andrew Morton , Linus Torvalds Subject: [PATCH 3.10 044/173] ipc/sem.c: synchronize semop and semctl with IPC_RMID Date: Mon, 2 Dec 2013 11:10:27 -0800 Message-Id: <20131202191147.728085484@linuxfoundation.org> X-Mailer: git-send-email 1.8.4.3.gca3854a In-Reply-To: <20131202191142.873808297@linuxfoundation.org> References: <20131202191142.873808297@linuxfoundation.org> User-Agent: quilt/0.60-8.1.3 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org 3.10-stable review patch. If anyone has any objections, please let me know. ------------------ From: Manfred Spraul commit 6e224f94597842c5eb17f1fc2208d20b6f7f7d49 upstream. After acquiring the semlock spinlock, operations must test that the array is still valid. - semctl() and exit_sem() would walk stale linked lists (ugly, but should be ok: all lists are empty) - semtimedop() would sleep forever - and if woken up due to a signal - access memory after free. The patch also: - standardizes the tests for .deleted, so that all tests in one function leave the function with the same approach. - unconditionally tests for .deleted immediately after every call to sem_lock - even it it means that for semctl(GETALL), .deleted will be tested twice. Both changes make the review simpler: After every sem_lock, there must be a test of .deleted, followed by a goto to the cleanup code (if the function uses "goto cleanup"). The only exception is semctl_down(): If sem_ids().rwsem is locked, then the presence in ids->ipcs_idr is equivalent to !.deleted, thus no additional test is required. Signed-off-by: Manfred Spraul Cc: Mike Galbraith Acked-by: Davidlohr Bueso Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds Signed-off-by: Greg Kroah-Hartman --- ipc/sem.c | 42 +++++++++++++++++++++++++++++------------- 1 file changed, 29 insertions(+), 13 deletions(-) --- a/ipc/sem.c +++ b/ipc/sem.c @@ -1282,6 +1282,12 @@ static int semctl_setval(struct ipc_name sem_lock(sma, NULL, -1); + if (sma->sem_perm.deleted) { + sem_unlock(sma, -1); + rcu_read_unlock(); + return -EIDRM; + } + curr = &sma->sem_base[semnum]; ipc_assert_locked_object(&sma->sem_perm); @@ -1336,12 +1342,14 @@ static int semctl_main(struct ipc_namesp int i; sem_lock(sma, NULL, -1); + if (sma->sem_perm.deleted) { + err = -EIDRM; + goto out_unlock; + } if(nsems > SEMMSL_FAST) { if (!ipc_rcu_getref(sma)) { - sem_unlock(sma, -1); - rcu_read_unlock(); err = -EIDRM; - goto out_free; + goto out_unlock; } sem_unlock(sma, -1); rcu_read_unlock(); @@ -1354,10 +1362,8 @@ static int semctl_main(struct ipc_namesp rcu_read_lock(); sem_lock_and_putref(sma); if (sma->sem_perm.deleted) { - sem_unlock(sma, -1); - rcu_read_unlock(); err = -EIDRM; - goto out_free; + goto out_unlock; } } for (i = 0; i < sma->sem_nsems; i++) @@ -1375,8 +1381,8 @@ static int semctl_main(struct ipc_namesp struct sem_undo *un; if (!ipc_rcu_getref(sma)) { - rcu_read_unlock(); - return -EIDRM; + err = -EIDRM; + goto out_rcu_wakeup; } rcu_read_unlock(); @@ -1404,10 +1410,8 @@ static int semctl_main(struct ipc_namesp rcu_read_lock(); sem_lock_and_putref(sma); if (sma->sem_perm.deleted) { - sem_unlock(sma, -1); - rcu_read_unlock(); err = -EIDRM; - goto out_free; + goto out_unlock; } for (i = 0; i < nsems; i++) @@ -1431,6 +1435,10 @@ static int semctl_main(struct ipc_namesp goto out_rcu_wakeup; sem_lock(sma, NULL, -1); + if (sma->sem_perm.deleted) { + err = -EIDRM; + goto out_unlock; + } curr = &sma->sem_base[semnum]; switch (cmd) { @@ -1836,6 +1844,10 @@ SYSCALL_DEFINE4(semtimedop, int, semid, if (error) goto out_rcu_wakeup; + error = -EIDRM; + locknum = sem_lock(sma, sops, nsops); + if (sma->sem_perm.deleted) + goto out_unlock_free; /* * semid identifiers are not unique - find_alloc_undo may have * allocated an undo structure, it was invalidated by an RMID @@ -1843,8 +1855,6 @@ SYSCALL_DEFINE4(semtimedop, int, semid, * This case can be detected checking un->semid. The existence of * "un" itself is guaranteed by rcu. */ - error = -EIDRM; - locknum = sem_lock(sma, sops, nsops); if (un && un->semid == -1) goto out_unlock_free; @@ -2057,6 +2067,12 @@ void exit_sem(struct task_struct *tsk) } sem_lock(sma, NULL, -1); + /* exit_sem raced with IPC_RMID, nothing to do */ + if (sma->sem_perm.deleted) { + sem_unlock(sma, -1); + rcu_read_unlock(); + continue; + } un = __lookup_undo(ulp, semid); if (un == NULL) { /* exit_sem raced with IPC_RMID+semget() that created