From patchwork Thu Jun 10 19:36:37 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Adhemerval Zanella Netto X-Patchwork-Id: 43828 Return-Path: X-Original-To: patchwork@sourceware.org Delivered-To: patchwork@sourceware.org Received: from server2.sourceware.org (localhost [IPv6:::1]) by sourceware.org (Postfix) with ESMTP id 48B4939A2408 for ; Thu, 10 Jun 2021 19:40:58 +0000 (GMT) DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 48B4939A2408 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=sourceware.org; s=default; t=1623354058; bh=ee/cVCl5FcoUKi6VuRxc91WJ2q6uk2nVN1G5oy/vj/M=; h=To:Subject:Date:In-Reply-To:References:List-Id:List-Unsubscribe: List-Archive:List-Post:List-Help:List-Subscribe:From:Reply-To: From; b=x+4X8IU0DpLbIJcRTrHO7l16mOufR4EptcDYn/AVEJYApfdY0wpVrmvvskos5sdsp xoogDsErsRTNXIsSoVbNMeVB2ER5bm8ndNJSFCz/AYvQEzVaabjvvK8OVNgHPB4dpX SRUO/Bu3NaWIXsIz6ZYkpqzN31upxr5BkmQaN0FE= X-Original-To: libc-alpha@sourceware.org Delivered-To: libc-alpha@sourceware.org Received: from mail-qt1-x82a.google.com (mail-qt1-x82a.google.com [IPv6:2607:f8b0:4864:20::82a]) by sourceware.org (Postfix) with ESMTPS id 6BFBC399C8A4 for ; Thu, 10 Jun 2021 19:36:50 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.1 sourceware.org 6BFBC399C8A4 Received: by mail-qt1-x82a.google.com with SMTP id p21so776250qtw.6 for ; Thu, 10 Jun 2021 12:36:50 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:subject:date:message-id:in-reply-to :references:mime-version:content-transfer-encoding; bh=ee/cVCl5FcoUKi6VuRxc91WJ2q6uk2nVN1G5oy/vj/M=; b=sAshIBHTdlTKx55mjoGhKwB9IFphL7h1lJuiLdsF6TxHoa0rZvf9UInAOG19EJ2jiY +TnsN0+au2lQ2P19aeNsLTOok9fjHJ+Cc2Am375iBw6cyEQ88KZ3iTY2o8sqYHp9hBnL bKfKXObRAwRhgjBzMjm3bC3kxBATXkVXlahXxQtddRHYoiDNKsWVm+qYwkZ+jQcU5Mcw GDUtOlHn0HTovvfk+itDm1in4SluhHWOeXE4QQXExJAe2x/V6Im6tVZhd3hdqTdaX4ft KqYQkm6HeD9CZWBtrd9AFchTqylu2nPbR92kNZAermcgbLNwqbfotPnnsIi6SJgKfK6A JdkQ== X-Gm-Message-State: AOAM5311P/yHTbVVIkHLUcYi0wuv8gwwdNzmjKUhrXc2fIlACorbsdVP e9LVi5/qJ5+XVEx8g3NA+QkuwQc/lpjc6A== X-Google-Smtp-Source: ABdhPJxwG8J2NDTedIzBdRXezDkP4ZtJpF2QFmCuUCmW71up1/l/eNLrpGChVREKBcqGe4FhR50+AQ== X-Received: by 2002:a05:622a:392:: with SMTP id j18mr377436qtx.283.1623353809610; Thu, 10 Jun 2021 12:36:49 -0700 (PDT) Received: from birita.. ([177.194.59.218]) by smtp.googlemail.com with ESMTPSA id i21sm3012583qkl.20.2021.06.10.12.36.48 for (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Thu, 10 Jun 2021 12:36:49 -0700 (PDT) To: libc-alpha@sourceware.org Subject: [PATCH 4/6] nptl: Do not use pthread set_tid_address as state synchronization (BZ #19951) Date: Thu, 10 Jun 2021 16:36:37 -0300 Message-Id: <20210610193639.3650754-5-adhemerval.zanella@linaro.org> X-Mailer: git-send-email 2.30.2 In-Reply-To: <20210610193639.3650754-1-adhemerval.zanella@linaro.org> References: <20210610193639.3650754-1-adhemerval.zanella@linaro.org> MIME-Version: 1.0 X-Spam-Status: No, score=-12.4 required=5.0 tests=BAYES_00, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, GIT_PATCH_0, KAM_SHORT, RCVD_IN_DNSWL_NONE, SPF_HELO_NONE, SPF_PASS, TXREP autolearn=ham autolearn_force=no version=3.4.2 X-Spam-Checker-Version: SpamAssassin 3.4.2 (2018-09-13) on server2.sourceware.org X-BeenThere: libc-alpha@sourceware.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Libc-alpha mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-Patchwork-Original-From: Adhemerval Zanella via Libc-alpha From: Adhemerval Zanella Netto Reply-To: Adhemerval Zanella Errors-To: libc-alpha-bounces+patchwork=sourceware.org@sourceware.org Sender: "Libc-alpha" The use after free described in BZ#19951 is due the use of two different PD fields, 'joinid' and 'cancelhandling', to describe the thread state. And any state change potentially requires to check for both fields atomically to potentially handle partial state (such as pthread_join() with a cancellation handler and a joinstate rollback). This patch uses a different PD member with 4 possible states (JOINABLE, DETACHED, EXITING, and EXITED) to synchronize between pthread_create(), pthread_join(), pthread_detached(), and pthread_exit(): 1. On pthread_create() the inital state is set either to JOINABLE or DETACHED depending of the pthread attribute used. 2. On pthread_detach() call, a CAS is issued on the state. If the CAS fails it means that thread is already detached (DETACHED) or is being terminated (EXITING). For former an EINVAL is returned, while for latter pthread_detach() should be reponsible to join the thread (and deallocate the resources). 3. On pthread_create() exit phase (reached either if the thread function has returned, pthread_exit() has being called, or cancellation handled has been acted upon) we issue a CAS on state to set to EXITING Mode. If the thread is previous on DETACHED mode it will be the thread itself responsible to deallocate any resource, otherwise the threads needs to be joined. 4. The clear_tid_field on 'clone' call is changed to set the new 'state' field on thread exit (EXITED). This state ins only reached on thread termination. 4. The pthread_join() implementation is now simpler: the futex wait is done directly on thread state and there is no need to reset it in case of timeout (since the state is now set either by pthread_detach() or by the kernel on process termination). The race condition on pthread_detach is avoided with only one atomic operation on PD state: once the mode is set to THREAD_STATE_DETACHED it is up to thread itself to deallocate its memory (done on the exit phase at pthread_create()). This change trigger an invalid C11 thread tests: it crates a thread, which detaches itself, and after a timeout the creating thread checks if the join fails. The issue is once thrd_join() is called the thread lifetime has already expired. The test is changed so the slee is done by the thread itself, so the creating thread will try to join a valid thread. Checked on x86_64-linux-gnu, i686-linux-gnu, aarch64-linux-gnu, arm-linux-gnueabihf, and powerpc64-linux-gnu. --- nptl/descr.h | 26 +++--- nptl/nptl-stack.h | 2 +- nptl/pthreadP.h | 2 +- nptl/pthread_clockjoin.c | 2 +- nptl/pthread_create.c | 48 ++++++---- nptl/pthread_detach.c | 36 +++----- nptl/pthread_getattr_np.c | 2 +- nptl/pthread_join.c | 2 +- nptl/pthread_join_common.c | 131 ++++++++++------------------ nptl/pthread_timedjoin.c | 2 +- nptl/pthread_tryjoin.c | 18 ++-- sysdeps/nptl/dl-tls_init_tp.c | 4 +- sysdeps/nptl/libc_start_call_main.h | 7 ++ sysdeps/pthread/tst-thrd-detach.c | 16 ++-- 14 files changed, 138 insertions(+), 160 deletions(-) diff --git a/nptl/descr.h b/nptl/descr.h index c85778d449..4b2db6edab 100644 --- a/nptl/descr.h +++ b/nptl/descr.h @@ -124,6 +124,18 @@ struct priority_protection_data }; +/* Define a possible thread state on 'joinstate' field. The value will be + cleared by the kernel when the thread terminates (CLONE_CHILD_CLEARTID), + so THREAD_STATE_EXITED must be 0. */ +enum + { + THREAD_STATE_EXITED = 0, + THREAD_STATE_EXITING, + THREAD_STATE_JOINABLE, + THREAD_STATE_DETACHED, + }; + + /* Thread descriptor data structure. */ struct pthread { @@ -166,8 +178,7 @@ struct pthread GL (dl_stack_user) list. */ list_t list; - /* Thread ID - which is also a 'is this thread descriptor (and - therefore stack) used' flag. */ + /* Thread ID set by the kernel with CLONE_PARENT_SETTID. */ pid_t tid; /* Ununsed. */ @@ -335,15 +346,8 @@ struct pthread hp_timing_t cpuclock_offset_ununsed; #endif - /* If the thread waits to join another one the ID of the latter is - stored here. - - In case a thread is detached this field contains a pointer of the - TCB if the thread itself. This is something which cannot happen - in normal operation. */ - struct pthread *joinid; - /* Check whether a thread is detached. */ -#define IS_DETACHED(pd) ((pd)->joinid == (pd)) + /* The current thread state defined by the THREAD_STATE_* enumeration. */ + unsigned int joinstate; /* The result of the thread function. */ void *result; diff --git a/nptl/nptl-stack.h b/nptl/nptl-stack.h index a6bd8df77f..8f3cfbf71e 100644 --- a/nptl/nptl-stack.h +++ b/nptl/nptl-stack.h @@ -29,7 +29,7 @@ static inline bool __nptl_stack_in_use (struct pthread *pd) { - return pd->tid <= 0; + return atomic_load_relaxed (&pd->joinstate) == THREAD_STATE_EXITED; } /* Remove the stack ELEM from its list. */ diff --git a/nptl/pthreadP.h b/nptl/pthreadP.h index ef0b367120..bfb0e40b44 100644 --- a/nptl/pthreadP.h +++ b/nptl/pthreadP.h @@ -536,7 +536,7 @@ libc_hidden_proto (__pthread_setcanceltype) extern void __pthread_testcancel (void); libc_hidden_proto (__pthread_testcancel) extern int __pthread_clockjoin_ex (pthread_t, void **, clockid_t, - const struct __timespec64 *, bool) + const struct __timespec64 *) attribute_hidden; extern int __pthread_sigmask (int, const sigset_t *, sigset_t *); libc_hidden_proto (__pthread_sigmask); diff --git a/nptl/pthread_clockjoin.c b/nptl/pthread_clockjoin.c index f5007d7831..8798f26fcf 100644 --- a/nptl/pthread_clockjoin.c +++ b/nptl/pthread_clockjoin.c @@ -30,7 +30,7 @@ ___pthread_clockjoin_np64 (pthread_t threadid, void **thread_return, return EINVAL; return __pthread_clockjoin_ex (threadid, thread_return, - clockid, abstime, true); + clockid, abstime); } #if __TIMESIZE == 64 diff --git a/nptl/pthread_create.c b/nptl/pthread_create.c index 8f1a77b5a8..0c0f2c03e5 100644 --- a/nptl/pthread_create.c +++ b/nptl/pthread_create.c @@ -283,7 +283,8 @@ static int create_thread (struct pthread *pd, const struct pthread_attr *attr, # define ARCH_CLONE __clone #endif if (__glibc_unlikely (ARCH_CLONE (&start_thread, STACK_VARIABLES_ARGS, - clone_flags, pd, &pd->tid, tp, &pd->tid) + clone_flags, pd, &pd->tid, tp, + &pd->joinstate) == -1)) return errno; @@ -343,7 +344,7 @@ start_thread (void *arg) and free any resource prior return to the pthread_create caller. */ setup_failed = pd->setup_failed == 1; if (setup_failed) - pd->joinid = NULL; + pd->joinstate = THREAD_STATE_JOINABLE; /* And give it up right away. */ lll_unlock (pd->lock, LLL_PRIVATE); @@ -473,9 +474,20 @@ start_thread (void *arg) the breakpoint reports TD_THR_RUN state rather than TD_THR_ZOMBIE. */ atomic_bit_set (&pd->cancelhandling, EXITING_BIT); - if (__glibc_unlikely (atomic_decrement_and_test (&__nptl_nthreads))) - /* This was the last thread. */ - exit (0); + + /* CONCURRENCY NOTES: + + Concurrent pthread_detach() will either set state to + THREAD_STATE_DETACHED or wait the thread to terminate. The exiting state + set here is set so a pthread_join() wait until all the required cleanup + steps are done. + + The 'joinstate' field will be used to determine who is responsible to + call __nptl_free_tcb below. */ + + unsigned int joinstate = THREAD_STATE_JOINABLE; + atomic_compare_exchange_weak_acquire (&pd->joinstate, &joinstate, + THREAD_STATE_EXITING); #ifndef __ASSUME_SET_ROBUST_LIST /* We let the kernel do the notification if it is able to do so on the exit @@ -511,6 +523,10 @@ start_thread (void *arg) } #endif + if (__glibc_unlikely (atomic_decrement_and_test (&__nptl_nthreads))) + /* This was the last thread. */ + exit (0); + if (!pd->user_stack) advise_stack_range (pd->stackblock, pd->stackblock_size, (uintptr_t) pd, pd->guardsize); @@ -531,9 +547,7 @@ start_thread (void *arg) pd->setxid_futex = 0; } - /* If the thread is detached free the TCB. */ - if (IS_DETACHED (pd)) - /* Free the TCB. */ + if (joinstate == THREAD_STATE_DETACHED) __nptl_free_tcb (pd); out: @@ -541,7 +555,7 @@ out: The 'exit' implementation in the kernel will signal when the process is really dead since 'clone' got passed the CLONE_CHILD_CLEARTID - flag. The 'tid' field in the TCB will be set to zero. + flag. The 'joinstate' field in the TCB will be set to zero. The exit code is zero since in case all threads exit by calling 'pthread_exit' the exit status must be 0 (zero). */ @@ -635,10 +649,9 @@ __pthread_create_2_1 (pthread_t *newthread, const pthread_attr_t *attr, pd->flags = ((iattr->flags & ~(ATTR_FLAG_SCHED_SET | ATTR_FLAG_POLICY_SET)) | (self->flags & (ATTR_FLAG_SCHED_SET | ATTR_FLAG_POLICY_SET))); - /* Initialize the field for the ID of the thread which is waiting - for us. This is a self-reference in case the thread is created - detached. */ - pd->joinid = iattr->flags & ATTR_FLAG_DETACHSTATE ? pd : NULL; + pd->joinstate = iattr->flags & ATTR_FLAG_DETACHSTATE + ? THREAD_STATE_DETACHED + : THREAD_STATE_JOINABLE; /* The debug events are inherited from the parent. */ pd->eventbuf = self->eventbuf; @@ -797,10 +810,11 @@ __pthread_create_2_1 (pthread_t *newthread, const pthread_attr_t *attr, /* Similar to pthread_join, but since thread creation has failed at startup there is no need to handle all the steps. */ - pid_t tid; - while ((tid = atomic_load_acquire (&pd->tid)) != 0) - __futex_abstimed_wait_cancelable64 ((unsigned int *) &pd->tid, - tid, 0, NULL, LLL_SHARED); + unsigned int state; + while ((state = atomic_load_acquire (&pd->joinstate)) + != THREAD_STATE_EXITED) + __futex_abstimed_wait_cancelable64 (&pd->joinstate, state, 0, + NULL, LLL_SHARED); } /* State (c) or (d) and we have ownership of PD (see CONCURRENCY diff --git a/nptl/pthread_detach.c b/nptl/pthread_detach.c index ac50db9b0e..97d292cab8 100644 --- a/nptl/pthread_detach.c +++ b/nptl/pthread_detach.c @@ -26,32 +26,24 @@ ___pthread_detach (pthread_t th) { struct pthread *pd = (struct pthread *) th; - /* Make sure the descriptor is valid. */ - if (INVALID_NOT_TERMINATED_TD_P (pd)) - /* Not a valid thread handle. */ - return ESRCH; + /* CONCURRENCY NOTES: - int result = 0; + Concurrent pthread_detach() will return EINVAL for the case the thread + is already detached (THREAD_STATE_DETACHED). POSIX states it is + undefined to call pthread_detach if TH refers to a non joinable thread. - /* Mark the thread as detached. */ - if (atomic_compare_and_exchange_bool_acq (&pd->joinid, pd, NULL)) + For the case the thread is being terminated (THREAD_STATE_EXITING), + pthread_detach() will responsible to clean up the stack. */ + + int curstate = THREAD_STATE_JOINABLE; + if (!atomic_compare_exchange_weak_acquire (&pd->joinstate, &curstate, + THREAD_STATE_DETACHED)) { - /* There are two possibilities here. First, the thread might - already be detached. In this case we return EINVAL. - Otherwise there might already be a waiter. The standard does - not mention what happens in this case. */ - if (IS_DETACHED (pd)) - result = EINVAL; + if (curstate == THREAD_STATE_DETACHED) + return EINVAL; + return __pthread_join (th, 0); } - else - /* Check whether the thread terminated meanwhile. In this case we - will just free the TCB. */ - if ((pd->cancelhandling & EXITING_BITMASK) != 0) - /* Note that the code in __free_tcb makes sure each thread - control block is freed only once. */ - __nptl_free_tcb (pd); - - return result; + return 0; } versioned_symbol (libc, ___pthread_detach, pthread_detach, GLIBC_2_34); libc_hidden_ver (___pthread_detach, __pthread_detach) diff --git a/nptl/pthread_getattr_np.c b/nptl/pthread_getattr_np.c index 25807cb529..560ba9ab98 100644 --- a/nptl/pthread_getattr_np.c +++ b/nptl/pthread_getattr_np.c @@ -53,7 +53,7 @@ __pthread_getattr_np (pthread_t thread_id, pthread_attr_t *attr) iattr->flags = thread->flags; /* The thread might be detached by now. */ - if (IS_DETACHED (thread)) + if (atomic_load_acquire (&thread->joinstate) == THREAD_STATE_DETACHED) iattr->flags |= ATTR_FLAG_DETACHSTATE; /* This is the guardsize after adjusting it. */ diff --git a/nptl/pthread_join.c b/nptl/pthread_join.c index d2b33de73d..195a537029 100644 --- a/nptl/pthread_join.c +++ b/nptl/pthread_join.c @@ -23,7 +23,7 @@ int ___pthread_join (pthread_t threadid, void **thread_return) { return __pthread_clockjoin_ex (threadid, thread_return, 0 /* Ignored */, - NULL, true); + NULL); } versioned_symbol (libc, ___pthread_join, pthread_join, GLIBC_2_34); libc_hidden_ver (___pthread_join, __pthread_join) diff --git a/nptl/pthread_join_common.c b/nptl/pthread_join_common.c index 7303069316..0870f36ecb 100644 --- a/nptl/pthread_join_common.c +++ b/nptl/pthread_join_common.c @@ -22,113 +22,74 @@ #include #include -static void -cleanup (void *arg) +/* Check for a possible deadlock situation where the threads are waiting for + each other to finish. Note that this is a "may" error. To be 100% sure we + catch this error we would have to lock the data structures but it is not + necessary. In the unlikely case that two threads are really caught in this + situation they will deadlock. It is the programmer's problem to figure + this out. */ +static inline bool +check_for_deadlock (int state, struct pthread *pd) { - /* If we already changed the waiter ID, reset it. The call cannot - fail for any reason but the thread not having done that yet so - there is no reason for a loop. */ struct pthread *self = THREAD_SELF; - atomic_compare_exchange_weak_acquire (&arg, &self, NULL); + return ((pd == self + || (atomic_load_acquire (&self->joinstate) == THREAD_STATE_DETACHED + && (pd->cancelhandling + & (CANCELED_BITMASK | EXITING_BITMASK + | TERMINATED_BITMASK)) == 0)) + && !(self->cancelstate == PTHREAD_CANCEL_ENABLE + && (pd->cancelhandling & (CANCELED_BITMASK | EXITING_BITMASK + | TERMINATED_BITMASK)) + == CANCELED_BITMASK)); } int __pthread_clockjoin_ex (pthread_t threadid, void **thread_return, clockid_t clockid, - const struct __timespec64 *abstime, bool block) + const struct __timespec64 *abstime) { struct pthread *pd = (struct pthread *) threadid; - /* Make sure the descriptor is valid. */ - if (INVALID_NOT_TERMINATED_TD_P (pd)) - /* Not a valid thread handle. */ - return ESRCH; - - /* Is the thread joinable?. */ - if (IS_DETACHED (pd)) - /* We cannot wait for the thread. */ - return EINVAL; - - struct pthread *self = THREAD_SELF; - int result = 0; - LIBC_PROBE (pthread_join, 1, threadid); - if ((pd == self - || (self->joinid == pd - && (pd->cancelhandling - & (CANCELED_BITMASK | EXITING_BITMASK - | TERMINATED_BITMASK)) == 0)) - && !(self->cancelstate == PTHREAD_CANCEL_ENABLE - && (pd->cancelhandling & (CANCELED_BITMASK | EXITING_BITMASK - | TERMINATED_BITMASK)) - == CANCELED_BITMASK)) - /* This is a deadlock situation. The threads are waiting for each - other to finish. Note that this is a "may" error. To be 100% - sure we catch this error we would have to lock the data - structures but it is not necessary. In the unlikely case that - two threads are really caught in this situation they will - deadlock. It is the programmer's problem to figure this - out. */ - return EDEADLK; - - /* Wait for the thread to finish. If it is already locked something - is wrong. There can only be one waiter. */ - else if (__glibc_unlikely (atomic_compare_exchange_weak_acquire (&pd->joinid, - &self, - NULL))) - /* There is already somebody waiting for the thread. */ - return EINVAL; - - /* BLOCK waits either indefinitely or based on an absolute time. POSIX also - states a cancellation point shall occur for pthread_join, and we use the - same rationale for posix_timedjoin_np. Both clockwait_tid and the futex - call use the cancellable variant. */ - if (block) + int result = 0; + unsigned int state; + while ((state = atomic_load_acquire (&pd->joinstate)) + != THREAD_STATE_EXITED) { - /* During the wait we change to asynchronous cancellation. If we - are cancelled the thread we are waiting for must be marked as - un-wait-ed for again. */ - pthread_cleanup_push (cleanup, &pd->joinid); - - /* We need acquire MO here so that we synchronize with the - kernel's store to 0 when the clone terminates. (see above) */ - pid_t tid; - while ((tid = atomic_load_acquire (&pd->tid)) != 0) - { - /* The kernel notifies a process which uses CLONE_CHILD_CLEARTID via - futex wake-up when the clone terminates. The memory location - contains the thread ID while the clone is running and is reset to - zero by the kernel afterwards. The kernel up to version 3.16.3 - does not use the private futex operations for futex wake-up when - the clone terminates. */ - int ret = __futex_abstimed_wait_cancelable64 ( - (unsigned int *) &pd->tid, tid, clockid, abstime, LLL_SHARED); - if (ret == ETIMEDOUT || ret == EOVERFLOW) - { - result = ret; - break; - } + if (check_for_deadlock (state, pd)) + return EDEADLK; + + /* POSIX states calling pthread_join() on a non joinable thread is + undefined. However, if PD is still in the cache we can still warn + the caller. */ + if (state == THREAD_STATE_DETACHED) + return EINVAL; + + /* pthread_join() is a cancellation entrypoint and we use the same + rationale for pthread_timedjoin_np(). + + The kernel notifies a process which uses CLONE_CHILD_CLEARTID via + a memory zerouing and futex wake up when the process terminates. + The futex operation is not private. */ + int ret = __futex_abstimed_wait_cancelable64 (&pd->joinstate, state, + clockid, abstime, + LLL_SHARED); + if (ret == ETIMEDOUT || ret == EOVERFLOW) + { + result = ret; + break; } - - pthread_cleanup_pop (0); } void *pd_result = pd->result; - if (__glibc_likely (result == 0)) + if (result == 0) { - /* We mark the thread as terminated and as joined. */ - pd->tid = -1; - - /* Store the return value if the caller is interested. */ + pd->tid = 0; if (thread_return != NULL) *thread_return = pd_result; - - /* Free the TCB. */ __nptl_free_tcb (pd); } - else - pd->joinid = NULL; LIBC_PROBE (pthread_join_ret, 3, threadid, result, pd_result); diff --git a/nptl/pthread_timedjoin.c b/nptl/pthread_timedjoin.c index ebc31f935a..b35c8f4279 100644 --- a/nptl/pthread_timedjoin.c +++ b/nptl/pthread_timedjoin.c @@ -25,7 +25,7 @@ ___pthread_timedjoin_np64 (pthread_t threadid, void **thread_return, const struct __timespec64 *abstime) { return __pthread_clockjoin_ex (threadid, thread_return, - CLOCK_REALTIME, abstime, true); + CLOCK_REALTIME, abstime); } #if __TIMESIZE == 64 diff --git a/nptl/pthread_tryjoin.c b/nptl/pthread_tryjoin.c index fd938e8780..726f2abc68 100644 --- a/nptl/pthread_tryjoin.c +++ b/nptl/pthread_tryjoin.c @@ -22,15 +22,17 @@ int __pthread_tryjoin_np (pthread_t threadid, void **thread_return) { - /* Return right away if the thread hasn't terminated yet. */ - struct pthread *pd = (struct pthread *) threadid; - if (pd->tid != 0) - return EBUSY; + /* The joinable state (THREAD_STATE_JOINABLE) is straigthforward since the + thread hasn't finished yet and trying to join might block. + Both detached (THREAD_STATE_DETACHED) and exiting (THREAD_STATE_EXITING) + might also result in a possible blocking call: a detached thread might + change its state to exiting and a exiting thread my take some time to + exit (and thus let the kernel set the state to THREAD_STATE_EXITED). */ - /* If pd->tid == 0 then lll_wait_tid will not block on futex - operation. */ - return __pthread_clockjoin_ex (threadid, thread_return, 0 /* Ignored */, - NULL, false); + struct pthread *pd = (struct pthread *) threadid; + return atomic_load_acquire (&pd->joinstate) != THREAD_STATE_EXITED + ? EBUSY + : __pthread_clockjoin_ex (threadid, thread_return, 0, NULL); } versioned_symbol (libc, __pthread_tryjoin_np, pthread_tryjoin_np, GLIBC_2_34); diff --git a/sysdeps/nptl/dl-tls_init_tp.c b/sysdeps/nptl/dl-tls_init_tp.c index b7b3bb1cdb..32a9acd8f7 100644 --- a/sysdeps/nptl/dl-tls_init_tp.c +++ b/sysdeps/nptl/dl-tls_init_tp.c @@ -62,7 +62,7 @@ __tls_init_tp (void) /* Early initialization of the TCB. */ struct pthread *pd = THREAD_SELF; - pd->tid = INTERNAL_SYSCALL_CALL (set_tid_address, &pd->tid); + pd->tid = INTERNAL_SYSCALL_CALL (set_tid_address, &pd->joinstate); THREAD_SETMEM (pd, specific[0], &pd->specific_1stblock[0]); THREAD_SETMEM (pd, user_stack, true); @@ -97,4 +97,6 @@ __tls_init_tp (void) THREAD_SETMEM (pd, cancelstate, PTHREAD_CANCEL_ENABLE); THREAD_SETMEM (pd, canceltype, PTHREAD_CANCEL_DEFERRED); + + THREAD_SETMEM (pd, joinstate, THREAD_STATE_JOINABLE); } diff --git a/sysdeps/nptl/libc_start_call_main.h b/sysdeps/nptl/libc_start_call_main.h index b56bf34325..592326e977 100644 --- a/sysdeps/nptl/libc_start_call_main.h +++ b/sysdeps/nptl/libc_start_call_main.h @@ -17,6 +17,7 @@ . */ #include +#include #include _Noreturn static void @@ -65,6 +66,12 @@ __libc_start_call_main (int (*main) (int, char **, char ** MAIN_AUXVEC_DECL), /* One less thread. Decrement the counter. If it is zero we terminate the entire process. */ result = 0; + + /* For the case a thread is waiting for the main thread to finish. */ + struct pthread *self = THREAD_SELF; + atomic_store_release (&self->joinstate, THREAD_STATE_EXITED); + futex_wake (&self->joinstate, 1, FUTEX_SHARED); + if (! atomic_decrement_and_test (&__nptl_nthreads)) /* Not much left to do but to exit the thread, not the process. */ while (1) diff --git a/sysdeps/pthread/tst-thrd-detach.c b/sysdeps/pthread/tst-thrd-detach.c index c844767748..e1906a0e10 100644 --- a/sysdeps/pthread/tst-thrd-detach.c +++ b/sysdeps/pthread/tst-thrd-detach.c @@ -20,14 +20,14 @@ #include #include #include - +#include #include static int detach_thrd (void *arg) { - if (thrd_detach (thrd_current ()) != thrd_success) - FAIL_EXIT1 ("thrd_detach failed"); + thrd_sleep (&(struct timespec) { .tv_sec = INT_MAX }, NULL); + thrd_exit (thrd_success); } @@ -36,15 +36,11 @@ do_test (void) { thrd_t id; - /* Create new thread. */ - if (thrd_create (&id, detach_thrd, NULL) != thrd_success) - FAIL_EXIT1 ("thrd_create failed"); + TEST_COMPARE (thrd_create (&id, detach_thrd, NULL), thrd_success); - /* Give some time so the thread can finish. */ - thrd_sleep (&(struct timespec) {.tv_sec = 2}, NULL); + TEST_COMPARE (thrd_detach (id), thrd_success); - if (thrd_join (id, NULL) == thrd_success) - FAIL_EXIT1 ("thrd_join succeed where it should fail"); + TEST_COMPARE (thrd_join (id, NULL), thrd_error); return 0; }