Print this page
NEX-20218 Backport Illumos #9474 txg_kick() fails to see that we are quiescing, forcing transactions to their next stages without leaving them accumulate changes
MFV illumos-gate@fa41d87de9ec9000964c605eb01d6dc19e4a1abe
    9464 txg_kick() fails to see that we are quiescing, forcing transactions to their next stages without leaving them accumulate changes
    Reviewed by: Matt Ahrens <matt@delphix.com>
    Reviewed by: Brad Lewis <brad.lewis@delphix.com>
    Reviewed by: Andriy Gapon <avg@FreeBSD.org>
    Approved by: Dan McDonald <danmcd@joyent.com>


   8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
   9  * or http://www.opensolaris.org/os/licensing.
  10  * See the License for the specific language governing permissions
  11  * and limitations under the License.
  12  *
  13  * When distributing Covered Code, include this CDDL HEADER in each
  14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
  15  * If applicable, add the following below this CDDL HEADER, with the
  16  * fields enclosed by brackets "[]" replaced with your own identifying
  17  * information: Portions Copyright [yyyy] [name of copyright owner]
  18  *
  19  * CDDL HEADER END
  20  */
  21 
  22 /*
  23  * Copyright 2009 Sun Microsystems, Inc.  All rights reserved.
  24  * Use is subject to license terms.
  25  */
  26 
  27 /*
  28  * Copyright (c) 2013 by Delphix. All rights reserved.
  29  */
  30 
  31 #ifndef _SYS_TXG_IMPL_H
  32 #define _SYS_TXG_IMPL_H
  33 
  34 #include <sys/spa.h>
  35 #include <sys/txg.h>
  36 
  37 #ifdef  __cplusplus
  38 extern "C" {
  39 #endif
  40 
  41 /*
  42  * The tx_cpu structure is a per-cpu structure that is used to track
  43  * the number of active transaction holds (tc_count). As transactions
  44  * are assigned into a transaction group the appropriate tc_count is
  45  * incremented to indicate that there are pending changes that have yet
  46  * to quiesce. Consumers evenutally call txg_rele_to_sync() to decrement
  47  * the tc_count. A transaction group is not considered quiesced until all
  48  * tx_cpu structures have reached a tc_count of zero.


  75         list_t          tc_callbacks[TXG_SIZE]; /* commit cb list */
  76         char            tc_pad[8];              /* pad to fill 3 cache lines */
  77 };
  78 
  79 /*
  80  * The tx_state structure maintains the state information about the different
  81  * stages of the pool's transcation groups. A per pool tx_state structure
  82  * is used to track this information. The tx_state structure also points to
  83  * an array of tx_cpu structures (described above). Although the tx_sync_lock
  84  * is used to protect the members of this structure, it is not used to
  85  * protect the tx_open_txg. Instead a special lock in the tx_cpu structure
  86  * is used. Readers of tx_open_txg must grab the per-cpu tc_open_lock.
  87  * Any thread wishing to update tx_open_txg must grab the tc_open_lock on
  88  * every cpu (see txg_quiesce()).
  89  */
  90 typedef struct tx_state {
  91         tx_cpu_t        *tx_cpu;        /* protects access to tx_open_txg */
  92         kmutex_t        tx_sync_lock;   /* protects the rest of this struct */
  93 
  94         uint64_t        tx_open_txg;    /* currently open txg id */

  95         uint64_t        tx_quiesced_txg; /* quiesced txg waiting for sync */
  96         uint64_t        tx_syncing_txg; /* currently syncing txg id */
  97         uint64_t        tx_synced_txg;  /* last synced txg id */
  98 
  99         hrtime_t        tx_open_time;   /* start time of tx_open_txg */
 100 
 101         uint64_t        tx_sync_txg_waiting; /* txg we're waiting to sync */
 102         uint64_t        tx_quiesce_txg_waiting; /* txg we're waiting to open */
 103 
 104         kcondvar_t      tx_sync_more_cv;
 105         kcondvar_t      tx_sync_done_cv;
 106         kcondvar_t      tx_quiesce_more_cv;
 107         kcondvar_t      tx_quiesce_done_cv;
 108         kcondvar_t      tx_timeout_cv;
 109         kcondvar_t      tx_exit_cv;     /* wait for all threads to exit */
 110 
 111         uint8_t         tx_threads;     /* number of threads */
 112         uint8_t         tx_exiting;     /* set when we're exiting */
 113 
 114         kthread_t       *tx_sync_thread;


   8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
   9  * or http://www.opensolaris.org/os/licensing.
  10  * See the License for the specific language governing permissions
  11  * and limitations under the License.
  12  *
  13  * When distributing Covered Code, include this CDDL HEADER in each
  14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
  15  * If applicable, add the following below this CDDL HEADER, with the
  16  * fields enclosed by brackets "[]" replaced with your own identifying
  17  * information: Portions Copyright [yyyy] [name of copyright owner]
  18  *
  19  * CDDL HEADER END
  20  */
  21 
  22 /*
  23  * Copyright 2009 Sun Microsystems, Inc.  All rights reserved.
  24  * Use is subject to license terms.
  25  */
  26 
  27 /*
  28  * Copyright (c) 2013, 2017 by Delphix. All rights reserved.
  29  */
  30 
  31 #ifndef _SYS_TXG_IMPL_H
  32 #define _SYS_TXG_IMPL_H
  33 
  34 #include <sys/spa.h>
  35 #include <sys/txg.h>
  36 
  37 #ifdef  __cplusplus
  38 extern "C" {
  39 #endif
  40 
  41 /*
  42  * The tx_cpu structure is a per-cpu structure that is used to track
  43  * the number of active transaction holds (tc_count). As transactions
  44  * are assigned into a transaction group the appropriate tc_count is
  45  * incremented to indicate that there are pending changes that have yet
  46  * to quiesce. Consumers evenutally call txg_rele_to_sync() to decrement
  47  * the tc_count. A transaction group is not considered quiesced until all
  48  * tx_cpu structures have reached a tc_count of zero.


  75         list_t          tc_callbacks[TXG_SIZE]; /* commit cb list */
  76         char            tc_pad[8];              /* pad to fill 3 cache lines */
  77 };
  78 
  79 /*
  80  * The tx_state structure maintains the state information about the different
  81  * stages of the pool's transcation groups. A per pool tx_state structure
  82  * is used to track this information. The tx_state structure also points to
  83  * an array of tx_cpu structures (described above). Although the tx_sync_lock
  84  * is used to protect the members of this structure, it is not used to
  85  * protect the tx_open_txg. Instead a special lock in the tx_cpu structure
  86  * is used. Readers of tx_open_txg must grab the per-cpu tc_open_lock.
  87  * Any thread wishing to update tx_open_txg must grab the tc_open_lock on
  88  * every cpu (see txg_quiesce()).
  89  */
  90 typedef struct tx_state {
  91         tx_cpu_t        *tx_cpu;        /* protects access to tx_open_txg */
  92         kmutex_t        tx_sync_lock;   /* protects the rest of this struct */
  93 
  94         uint64_t        tx_open_txg;    /* currently open txg id */
  95         uint64_t        tx_quiescing_txg; /* currently quiescing txg id */
  96         uint64_t        tx_quiesced_txg; /* quiesced txg waiting for sync */
  97         uint64_t        tx_syncing_txg; /* currently syncing txg id */
  98         uint64_t        tx_synced_txg;  /* last synced txg id */
  99 
 100         hrtime_t        tx_open_time;   /* start time of tx_open_txg */
 101 
 102         uint64_t        tx_sync_txg_waiting; /* txg we're waiting to sync */
 103         uint64_t        tx_quiesce_txg_waiting; /* txg we're waiting to open */
 104 
 105         kcondvar_t      tx_sync_more_cv;
 106         kcondvar_t      tx_sync_done_cv;
 107         kcondvar_t      tx_quiesce_more_cv;
 108         kcondvar_t      tx_quiesce_done_cv;
 109         kcondvar_t      tx_timeout_cv;
 110         kcondvar_t      tx_exit_cv;     /* wait for all threads to exit */
 111 
 112         uint8_t         tx_threads;     /* number of threads */
 113         uint8_t         tx_exiting;     /* set when we're exiting */
 114 
 115         kthread_t       *tx_sync_thread;