本节简单介绍了PostgreSQL在执行插入过程中与缓存相关的函数RelationGetBufferForTuple,该函数返回满足空闲空间 >= 给定大小的page,并且该page对应的buffer状态为pinned和并持有独占锁。 ### 一、数据结构 **BufferDesc** 共享缓冲区的共享描述符(状态)数据 ``` /* * Flags for buffer descriptors * buffer描述器标记 * * Note: TAG_VALID essentially means that there is a buffer hashtable * entry associated with the buffer's tag. * 注意:TAG_VALID本质上意味着有一个与缓冲区的标记相关联的缓冲区散列表条目。 */ //buffer header锁定 #define BM_LOCKED (1U << 22) /* buffer header is locked */ //数据需要写入(标记为DIRTY) #define BM_DIRTY (1U << 23) /* data needs writing */ //数据是有效的 #define BM_VALID (1U << 24) /* data is valid */ //已分配buffer tag #define BM_TAG_VALID (1U << 25) /* tag is assigned */ //正在R/W #define BM_IO_IN_PROGRESS (1U << 26) /* read or write in progress */ //上一个I/O出现错误 #define BM_IO_ERROR (1U << 27) /* previous I/O failed */ //开始写则变DIRTY #define BM_JUST_DIRTIED (1U << 28) /* dirtied since write started */ //存在等待sole pin的其他进程 #define BM_PIN_COUNT_WAITER (1U << 29) /* have waiter for sole pin */ //checkpoint发生,必须刷到磁盘上 #define BM_CHECKPOINT_NEEDED (1U << 30) /* must write for checkpoint */ //持久化buffer(不是unlogged或者初始化fork) #define BM_PERMANENT (1U << 31) /* permanent buffer (not unlogged, * or init fork) */ /* * BufferDesc -- shared descriptor/state data for a single shared buffer. * BufferDesc -- 共享缓冲区的共享描述符(状态)数据 * * Note: Buffer header lock (BM_LOCKED flag) must be held to examine or change * the tag, state or wait_backend_pid fields. In general, buffer header lock * is a spinlock which is combined with flags, refcount and usagecount into * single atomic variable. This layout allow us to do some operations in a * single atomic operation, without actually acquiring and releasing spinlock; * for instance, increase or decrease refcount. buf_id field never changes * after initialization, so does not need locking. freeNext is protected by * the buffer_strategy_lock not buffer header lock. The LWLock can take care * of itself. The buffer header lock is *not* used to control access to the * data in the buffer! * 注意:必须持有Buffer header锁(BM_LOCKED标记)才能检查或修改tag/state/wait_backend_pid字段. * 通常来说,buffer header lock是spinlock,它与标记位/参考计数/使用计数组合到单个原子变量中. * 这个布局设计允许我们执行原子操作,而不需要实际获得或者释放spinlock(比如,增加或者减少参考计数). * buf_id字段在初始化后不会出现变化,因此不需要锁定. * freeNext通过buffer_strategy_lock锁而不是buffer header lock保护. * LWLock可以很好的处理自己的状态. * 务请注意的是:buffer header lock不用于控制buffer中的数据访问! * * It's assumed that nobody changes the state field while buffer header lock * is held. Thus buffer header lock holder can do complex updates of the * state variable in single write, simultaneously with lock release (cleaning * BM_LOCKED flag). On the other hand, updating of state without holding * buffer header lock is restricted to CAS, which insure that BM_LOCKED flag * is not set. Atomic increment/decrement, OR/AND etc. are not allowed. * 假定在持有buffer header lock的情况下,没有人改变状态字段. * 持有buffer header lock的进程可以执行在单个写操作中执行复杂的状态变量更新, * 同步的释放锁(清除BM_LOCKED标记). * 换句话说,如果没有持有buffer header lock的状态更新,会受限于CAS, * 这种情况下确保BM_LOCKED没有被设置. * 比如原子的增加/减少(AND/OR)等操作是不允许的. * * An exception is that if we have the buffer pinned, its tag can't change * underneath us, so we can examine the tag without locking the buffer header. * Also, in places we do one-time reads of the flags without bothering to * lock the buffer header; this is generally for situations where we don't * expect the flag bit being tested to be changing. * 一种例外情况是如果我们已有buffer pinned,该buffer的tag不能改变(在本进程之下), * 因此不需要锁定buffer header就可以检查tag了. * 同时,在执行一次性的flags读取时不需要锁定buffer header. * 这种情况通常用于我们不希望正在测试的flag bit将被改变. * * We can't physically remove items from a disk page if another backend has * the buffer pinned. Hence, a backend may need to wait for all other pins * to go away. This is signaled by storing its own PID into * wait_backend_pid and setting flag bit BM_PIN_COUNT_WAITER. At present, * there can be only one such waiter per buffer. * 如果其他进程有buffer pinned,那么进程不能物理的从磁盘页面中删除items. * 因此,后台进程需要等待其他pins清除.这可以通过存储它自己的PID到wait_backend_pid中, * 并设置标记位BM_PIN_COUNT_WAITER. * 目前,每个缓冲区只能由一个等待进程. * * We use this same struct for local buffer headers, but the locks are not * used and not all of the flag bits are useful either. To avoid unnecessary * overhead, manipulations of the state field should be done without actual * atomic operations (i.e. only pg_atomic_read_u32() and * pg_atomic_unlocked_write_u32()). * 本地缓冲头部使用同样的结构,但并不需要使用locks,而且并不是所有的标记位都使用. * 为了避免不必要的负载,状态域的维护不需要实际的原子操作 * (比如只有pg_atomic_read_u32() and pg_atomic_unlocked_write_u32()) * * Be careful to avoid increasing the size of the struct when adding or * reordering members. Keeping it below 64 bytes (the most common CPU * cache line size) is fairly important for performance. * 在增加或者记录成员变量时,小心避免增加结构体的大小. * 保持结构体大小在64字节内(通常的CPU缓存线大小)对于性能是非常重要的. */ typedef struct BufferDesc { //buffer tag BufferTag tag; /* ID of page contained in buffer */ //buffer索引编号(0开始),指向相应的buffer pool slot int buf_id; /* buffer's index number (from 0) */ /* state of the tag, containing flags, refcount and usagecount */ //tag状态,包括flags/refcount和usagecount pg_atomic_uint32 state; //pin-count等待进程ID int wait_backend_pid; /* backend PID of pin-count waiter */ //空闲链表链中下一个空闲的buffer int freeNext; /* link in freelist chain */ //缓冲区内容锁 LWLock content_lock; /* to lock access to buffer contents */ } BufferDesc; ``` **BufferTag** Buffer tag标记了buffer存储的是磁盘中哪个block ``` /* * Buffer tag identifies which disk block the buffer contains. * Buffer tag标记了buffer存储的是磁盘中哪个block * * Note: the BufferTag data must be sufficient to determine where to write the * block, without reference to pg_class or pg_tablespace entries. It's * possible that the backend flushing the buffer doesn't even believe the * relation is visible yet (its xact may have started before the xact that * created the rel). The storage manager must be able to cope anyway. * 注意:BufferTag必须足以确定如何写block而不需要参照pg_class或者pg_tablespace数据字典信息. * 有可能后台进程在刷新缓冲区的时候深圳不相信关系是可见的(事务可能在创建rel的事务之前). * 存储管理器必须可以处理这些事情. * * Note: if there's any pad bytes in the struct, INIT_BUFFERTAG will have * to be fixed to zero them, since this struct is used as a hash key. * 注意:如果在结构体中有填充的字节,INIT_BUFFERTAG必须将它们固定为零，因为这个结构体用作散列键. */ typedef struct buftag { //物理relation标识符 RelFileNode rnode; /* physical relation identifier */ ForkNumber forkNum; //相对于relation起始的块号 BlockNumber blockNum; /* blknum relative to begin of reln */ } BufferTag; ``` ### 二、源码解读 RelationGetBufferForTuple函数返回满足空闲空间>=给定大小的page,并且该page对应的buffer状态为pinned和并持有独占锁 输入： relation-数据表 len-需要的空间大小 otherBuffer-用于update场景，上一次pinned的buffer options-处理选项 bistate-BulkInsert标记 vmbuffer-第1个vm(visibilitymap) vmbuffer_other-用于update场景，上一次pinned的buffer对应的vm(visibilitymap) 注意: otherBuffer这个参数让人觉得困惑，原因是PG的机制使然 Update时，不是原地更新，而是原数据保留（更新xmax），新数据插入 原数据&新数据如果在不同Block中，锁定Block的时候可能会出现Deadlock 举个例子：Session A更新表T的第一行，第一行在Block 0中，新数据存储在Block 2中 Session B更新表T的第二行，第二行在Block 0中，新数据存储在Block 2中 Block 0/2均要锁定才能完整实现Update操作： 如果Session A先锁定了Block 2，Session B先锁定了Block 0， 然后Session A尝试锁定Block 0，Session B尝试锁定Block 2，这时候就会出现死锁 为了避免这种情况，PG规定锁定时，同一个Relation，按Block的编号顺序锁定， 如需要锁定0和2，那必须先锁定Block 0，再锁定2 输出： 为Tuple分配的Buffer 其主要实现逻辑如下: 1.初始化相关变量 2.获取预留空间 3.如为Update操作,则获取上次pinned buffer对应的Block 4.获取目标page:targetBlock 5.如targetBlock非法,并且使用FSM,则使用FSM寻找 6.如targetBlock仍非法,则循环遍历page检索合适的Block 6.1.读取并独占锁定目标block,以及给定的otherBuffer(如给出) 6.2.获取vm 6.3.读取buffer,判断是否有足够的空闲空间,如足够,则返回 6.4.如仍不足够,则调用RecordAndGetPageWithFreeSpace获取targetBlock,再次循环 7.遍历完毕,仍找不到block,则扩展表 8.扩展表后,以P_NEW模式读取buffer并锁定 9.获取该buffer对应的page,执行相关校验 10.校验不通过报错,校验通过则返回buffer ``` /* * RelationGetBufferForTuple * * Returns pinned and exclusive-locked buffer of a page in given relation * with free space >= given len. * 返回满足空闲空间>=给定大小的page,并且该page对应的buffer状态为pinned和并持有独占锁 * * If otherBuffer is not InvalidBuffer, then it references a previously * pinned buffer of another page in the same relation; on return, this * buffer will also be exclusive-locked. (This case is used by heap_update; * the otherBuffer contains the tuple being updated.) * 如果otherBuffer不是InvalidBuffer, * 那么otherBuffer依赖的是先前同一个relation但是其他page的pinned buffer. * 返回时,该buffer同时被独占锁定. * (heap_update会出现这种情况,otherBuffer存储正update的tuple) * * The reason for passing otherBuffer is that if two backends are doing * concurrent heap_update operations, a deadlock could occur if they try * to lock the same two buffers in opposite orders. To ensure that this * can't happen, we impose the rule that buffers of a relation must be * locked in increasing page number order. This is most conveniently done * by having RelationGetBufferForTuple lock them both, with suitable care * for ordering. * 传递otherBuffer的原因是如果两个进程在并发heap_update操作, * 如果它们尝试以相反的顺序锁定相同的两个buffer,那会出现死锁. * 为了确保这种情况不会出现,我们规定，关系缓冲区必须按page的编号顺序锁定. * 要做到这一点,最方便的方法是让RelationGetBufferForTuple注意顺序锁定它们. * * NOTE: it is unlikely, but not quite impossible, for otherBuffer to be the * same buffer we select for insertion of the new tuple (this could only * happen if space is freed in that page after heap_update finds there's not * enough there). In that case, the page will be pinned and locked only once. * 注意:这不太可能,但又不是不可能,为了让otherBuffer与我们选择插入新元组的buffer一致. * (这只会发生在在执行heap_update检索page发现没有足够的空闲空间,但随后空间被释放的情况) * 在这种情况下,page会被pinned并且只会lock一次. * * For the vmbuffer and vmbuffer_other arguments, we avoid deadlock by * locking them only after locking the corresponding heap page, and taking * no further lwlocks while they are locked. * 对于vmbuffer和vmbuffer_other参数,通过在锁定相应的heap page后再锁定它们来避免死锁, * 同时,在被锁定后,不再持有lwlocks. * * We normally use FSM to help us find free space. However, * if HEAP_INSERT_SKIP_FSM is specified, we just append a new empty page to * the end of the relation if the tuple won't fit on the current target page. * This can save some cycles when we know the relation is new and doesn't * contain useful amounts of free space. * 通常来说,使用FSM检索空闲空间.但是,如果指定了HEAP_INSERT_SKIP_FSM, * 那么如果当前的目标page不适合,则直接在relation的最后追加空page. * 这样可以在知道relation是新的情况下,节省一些处理时间,而且不需要持有有用的空闲空间计数信息. * * HEAP_INSERT_SKIP_FSM is also useful for non-WAL-logged additions to a * relation, if the caller holds exclusive lock and is careful to invalidate * relation's smgr_targblock before the first insertion --- that ensures that * all insertions will occur into newly added pages and not be intermixed * with tuples from other transactions. That way, a crash can't risk losing * any committed data of other transactions. (See heap_insert's comments * for additional constraints needed for safe usage of this behavior.) * HEAP_INSERT_SKIP_FSM同时对于非WAL logged关系也是有用的, * 如果调用者持有独占锁并且在首次插入前使得关系的smgr_targblock无效 --- * 这可以确保所有的插入会出现在新增加的pages中,而不会与其他事务的tuple混起来. * 按这种方式,如果出现宕机,那么就不会有丢失其他事务提交的数据的风险. * (详细参考heap_insert的注释,里面提到了使用该动作的其他约束) * * The caller can also provide a BulkInsertState object to optimize many * insertions into the same relation. This keeps a pin on the current * insertion target page (to save pin/unpin cycles) and also passes a * BULKWRITE buffer selection strategy object to the buffer manager. * Passing NULL for bistate selects the default behavior. * 调用者同时提供了BulkInsertState对象用于优化大量插入到同一个relation的情况. * 这会在当前插入的目标page保持pin(节省pin/unpin处理过程) * 同时会传递BULKWRITE缓冲区选择器策略对象到buffer manager中. * 如使用默认模式,则设置bitstate为NULL. * * We always try to avoid filling existing pages further than the fillfactor. * This is OK since this routine is not consulted when updating a tuple and * keeping it on the same page, which is the scenario fillfactor is meant * to reserve space for. * 我们通常尝试避免填充现有页面超过填充因子设定的范围. * 这是没有问题的,因为在更新元组并将其保存在同一个page中时,不会参考此例程, * 该场景下填充因子会用到. * * ereport(ERROR) is allowed here, so this routine *must* be called * before any (unlogged) changes are made in buffer pool. * ereport(ERROR)可在这允许使用,因此该例程必须在buffer pool出现任何变化前调用. */ /* 输入： relation-数据表 len-需要的空间大小 otherBuffer-用于update场景，上一次pinned的buffer options-处理选项 bistate-BulkInsert标记 vmbuffer-第1个vm(visibilitymap) vmbuffer_other-用于update场景，上一次pinned的buffer对应的vm(visibilitymap) 注意: otherBuffer这个参数让人觉得困惑，原因是PG的机制使然 Update时，不是原地更新，而是原数据保留（更新xmax），新数据插入 原数据&新数据如果在不同Block中，锁定Block的时候可能会出现Deadlock 举个例子：Session A更新表T的第一行，第一行在Block 0中，新数据存储在Block 2中 Session B更新表T的第二行，第二行在Block 0中，新数据存储在Block 2中 Block 0/2均要锁定才能完整实现Update操作： 如果Session A先锁定了Block 2，Session B先锁定了Block 0， 然后Session A尝试锁定Block 0，Session B尝试锁定Block 2，这时候就会出现死锁 为了避免这种情况，PG规定锁定时，同一个Relation，按Block的编号顺序锁定， 如需要锁定0和2，那必须先锁定Block 0，再锁定2 输出： 为Tuple分配的Buffer 附： Pinned buffers：means buffers are currently being used,it should not be flushed out. */ Buffer RelationGetBufferForTuple(Relation relation, Size len, Buffer otherBuffer, int options, BulkInsertState bistate, Buffer *vmbuffer, Buffer *vmbuffer_other) { bool use_fsm = !(options & HEAP_INSERT_SKIP_FSM);//是否使用FSM寻找空闲空间 Buffer buffer = InvalidBuffer;// Page page;// Size pageFreeSpace = 0,//page空闲空间 saveFreeSpace = 0;//page需要预留的空间 BlockNumber targetBlock,//目标Block otherBlock;//上一次pinned的buffer对应的Block bool needLock;//是否需要上锁 //大小对齐 len = MAXALIGN(len); /* be conservative */ /* Bulk insert is not supported for updates, only inserts. */ //otherBuffer有效，说明是update操作，不支持bi(BulkInsert) //bulk操作仅支持插入 Assert(otherBuffer == InvalidBuffer || !bistate); /* * If we're gonna fail for oversize tuple, do it right away * 对于超限的元组,直接报错 */ //#define MaxHeapTupleSize (BLCKSZ - MAXALIGN(SizeOfPageHeaderData + sizeof(ItemIdData))) //#define MinHeapTupleSize MAXALIGN(SizeofHeapTupleHeader) if (len > MaxHeapTupleSize) ereport(ERROR, (errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED), errmsg("row is too big: size %zu, maximum size %zu", len, MaxHeapTupleSize))); /* Compute desired extra freespace due to fillfactor option */ //获取预留空间 // #define RelationGetTargetPageFreeSpace(relation, defaultff) \ (BLCKSZ * (100 - RelationGetFillFactor(relation, defaultff)) / 100) saveFreeSpace = RelationGetTargetPageFreeSpace(relation, HEAP_DEFAULT_FILLFACTOR); //update操作,获取上次pinned buffer对应的Block if (otherBuffer != InvalidBuffer) otherBlock = BufferGetBlockNumber(otherBuffer); else otherBlock = InvalidBlockNumber; /* just to keep compiler quiet */ /* * We first try to put the tuple on the same page we last inserted a tuple * on, as cached in the BulkInsertState or relcache entry. If that * doesn't work, we ask the Free Space Map to locate a suitable page. * Since the FSM's info might be out of date, we have to be prepared to * loop around and retry multiple times. (To insure this isn't an infinite * loop, we must update the FSM with the correct amount of free space on * each page that proves not to be suitable.) If the FSM has no record of * a page with enough free space, we give up and extend the relation. * 首先会尝试把元组放在最后插入元组的page上,比如BulkInsertState或者relcache条目. * 如果找不到,那么我们通过FSM来定位合适的page. * 由于FSM的信息可能过期,这时候不得不循环并尝试多次. * (为了确保这不是一个无限循环,必须使用正确的页面空闲空间信息更新不靠谱的FSM) * 如果FSM中信息提示没有page有空闲空间,放弃并扩展relation. * * When use_fsm is false, we either put the tuple onto the existing target * page or extend the relation. * 如use_fsm为F,我们要不把元组放在现存的目标page上,要不扩展relation. */ if (len + saveFreeSpace > MaxHeapTupleSize) { //如果需要的大小+预留空间大于可容纳的最大Tuple大小，不使用FSM，扩展后再尝试 /* can't fit, don't bother asking FSM */ targetBlock = InvalidBlockNumber; use_fsm = false; } else if (bistate && bistate->current_buf != InvalidBuffer)//BulkInsert模式 targetBlock = BufferGetBlockNumber(bistate->current_buf); else targetBlock = RelationGetTargetBlock(relation);//普通Insert模式 if (targetBlock == InvalidBlockNumber && use_fsm) { //还没有找到合适的BlockNumber，并且需要使用FSM /* * We have no cached target page, so ask the FSM for an initial * target. * 没有缓存目标page,使用FSM获取初始目标page */ //使用FSM申请空闲空间=len + saveFreeSpace的块 targetBlock = GetPageWithFreeSpace(relation, len + saveFreeSpace); /* * If the FSM knows nothing of the rel, try the last page before we * give up and extend. This avoids one-tuple-per-page syndrome during * bootstrapping or in a recently-started system. * 如果FSM对rel一无所知,在放弃并扩展前尝试下最后那个page. * 这可以避免在bootstrapping或者最近已启动系统时一个元组一个page的情况. */ //申请不到，使用最后一个块，否则扩展或者放弃 if (targetBlock == InvalidBlockNumber) { BlockNumber nblocks = RelationGetNumberOfBlocks(relation); if (nblocks > 0) targetBlock = nblocks - 1; } } loop: while (targetBlock != InvalidBlockNumber) { //---------- 循环直至成功获取插入数据的块号 /* * Read and exclusive-lock the target block, as well as the other * block if one was given, taking suitable care with lock ordering and * the possibility they are the same block. * 读取并独占锁定目标block,以及给定的另外一个快(如给出),需要适当的关注锁的顺序 * 并关注它们是否同一个块. * * If the page-level all-visible flag is set, caller will need to * clear both that and the corresponding visibility map bit. However, * by the time we return, we'll have x-locked the buffer, and we don't * want to do any I/O while in that state. So we check the bit here * before taking the lock, and pin the page if it appears necessary. * Checking without the lock creates a risk of getting the wrong * answer, so we'll have to recheck after acquiring the lock. * 如果设置了块级别的all-visible flag,调用者需要清空该块的标记和相应的vm标记. * 但是,在返回时,我们将持有buffer的独占锁,并且我们不希望在这种情况下执行I/O操作. * 因此,我们在获取锁前检查标记位,如看起来需要的话,pin page. * 没有持有锁执行检查会出现错误,因此我们将不得不在获取锁后重新执行检查. */ if (otherBuffer == InvalidBuffer) { //----------- 非Update操作 /* easy case */ //这种情况比较简单 //获取Buffer buffer = ReadBufferBI(relation, targetBlock, bistate); if (PageIsAllVisible(BufferGetPage(buffer))) //如果Page可见，那么把Page Pin在内存中（Pin的意思是固定/保留） visibilitymap_pin(relation, targetBlock, vmbuffer); LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE);//锁定buffer } else if (otherBlock == targetBlock) { //----------- Update操作，新记录跟原记录在同一个Block中 //这种情况也比较简单 /* also easy case */ buffer = otherBuffer; if (PageIsAllVisible(BufferGetPage(buffer))) visibilitymap_pin(relation, targetBlock, vmbuffer); LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE); } else if (otherBlock < targetBlock) { //----------- Update操作，原记录所在的Block < 新记录的Block /* lock other buffer first */ //首先锁定otherBlock buffer = ReadBuffer(relation, targetBlock); if (PageIsAllVisible(BufferGetPage(buffer))) visibilitymap_pin(relation, targetBlock, vmbuffer); //优先锁定BlockNumber小的那个 LockBuffer(otherBuffer, BUFFER_LOCK_EXCLUSIVE); LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE); } else { //------------ Update操作，原记录所在的Block > 新记录的Block /* lock target buffer first */ buffer = ReadBuffer(relation, targetBlock); if (PageIsAllVisible(BufferGetPage(buffer))) visibilitymap_pin(relation, targetBlock, vmbuffer); //优先锁定BlockNumber小的那个 LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE); LockBuffer(otherBuffer, BUFFER_LOCK_EXCLUSIVE); } /* * We now have the target page (and the other buffer, if any) pinned * and locked. However, since our initial PageIsAllVisible checks * were performed before acquiring the lock, the results might now be * out of date, either for the selected victim buffer, or for the * other buffer passed by the caller. In that case, we'll need to * give up our locks, go get the pin(s) we failed to get earlier, and * re-lock. That's pretty painful, but hopefully shouldn't happen * often. * 现在已有了target page,并且该page(包括other buffer,如存在)已缓存到内存中(pinned)且已锁定. * 但是,由于初始的PageIsAllVisible在获取锁前执行,结果可能已经过期, * 这时候可能选择了需要被淘汰的buffer或者otherBuffer出现了变化. * 在这种情况下,需要放弃锁,回到先前曾经失败的pin的地方,重新锁定. * 这蛮吐血的,希望不要经常出现. * * Note that there's a small possibility that we didn't pin the page * above but still have the correct page pinned anyway, either because * we've already made a previous pass through this loop, or because * caller passed us the right page anyway. * 注意存在较小的可能是我们在上面不需要pin page,但仍然需要持有正确的pinned page, * 这一方面是因为我们已经通过该循环执行了一遍,另外一方面是调用者通过其他方式传入了正确的page. * * Note also that it's possible that by the time we get the pin and * retake the buffer locks, the visibility map bit will have been * cleared by some other backend anyway. In that case, we'll have * done a bit of extra work for no gain, but there's no real harm * done. * 同时要注意在我们获取pin并且重新获取buffer lock时,vm位已被其他后台进程清除了. * 在这种情况下,我们需要执行一些额外的工作以避免重复工作,但这实质上并没有什么危害. */ if (otherBuffer == InvalidBuffer || buffer <= otherBuffer) GetVisibilityMapPins(relation, buffer, otherBuffer, targetBlock, otherBlock, vmbuffer, vmbuffer_other);//Pin VM在内存中 else GetVisibilityMapPins(relation, otherBuffer, buffer, otherBlock, targetBlock, vmbuffer_other, vmbuffer);//Pin VM在内存中 /* * Now we can check to see if there's enough free space here. If so, * we're done. * 现在我们可以检查是否有足够的空闲空间. * 如有,则我们已完成所有工作了. */ page = BufferGetPage(buffer); pageFreeSpace = PageGetHeapFreeSpace(page); if (len + saveFreeSpace <= pageFreeSpace) { //有足够的空间存储数据，返回此Buffer /* use this page as future insert target, too */ //用这个page作为未来插入的目标page /* #define RelationSetTargetBlock(relation, targblock) \ do { \ RelationOpenSmgr(relation); \ (relation)->rd_smgr->smgr_targblock = (targblock); \ } while (0) */ RelationSetTargetBlock(relation, targetBlock); return buffer; } /* * Not enough space, so we must give up our page locks and pin (if * any) and prepare to look elsewhere. We don't care which order we * unlock the two buffers in, so this can be slightly simpler than the * code above. * 空间不够,必须放弃持有的page locks和pin,准备检索其他地方. * 在解锁时不需要关注两个buffer的顺序,这个逻辑比先前的逻辑要简单. */ LockBuffer(buffer, BUFFER_LOCK_UNLOCK); if (otherBuffer == InvalidBuffer) ReleaseBuffer(buffer); else if (otherBlock != targetBlock) { LockBuffer(otherBuffer, BUFFER_LOCK_UNLOCK); ReleaseBuffer(buffer); } /* Without FSM, always fall out of the loop and extend */ //不使用FSM定位空闲空间，跳出循环，执行扩展 if (!use_fsm) break; /* * Update FSM as to condition of this page, and ask for another page * to try. */ //使用FSM获取下一个备选的Block //注意：如果全部扫描后发现没有满足条件的Block，targetBlock = InvalidBlockNumber，跳出循环 targetBlock = RecordAndGetPageWithFreeSpace(relation, targetBlock, pageFreeSpace, len + saveFreeSpace); } //--------- 没有获取满足条件的Block，扩展表 /* * Have to extend the relation. * * We have to use a lock to ensure no one else is extending the rel at the * same time, else we will both try to initialize the same new page. We * can skip locking for new or temp relations, however, since no one else * could be accessing them. * 必须锁定以确保其他进程不能扩展rel,否则我们会同时尝试初始化新的page. * 但是,我们可以为新的或者临时关系跳过锁定,这时候没有其他进程可以访问它们. */ //新创建的数据表或者临时表，无需Lock needLock = !RELATION_IS_LOCAL(relation); /* * If we need the lock but are not able to acquire it immediately, we'll * consider extending the relation by multiple blocks at a time to manage * contention on the relation extension lock. However, this only makes * sense if we're using the FSM; otherwise, there's no point. * 如果需要锁定但不能够马上获取,考虑通过一次性多个blocks的方式扩展关系, * 这样可以在关系扩展锁上管理竞争. * 但是,这在使用FSM的时候才会奇效,否则没有其他太好的办法. */ if (needLock)//需要锁定 { if (!use_fsm) //不使用FSM LockRelationForExtension(relation, ExclusiveLock); else if (!ConditionalLockRelationForExtension(relation, ExclusiveLock)) { /* Couldn't get the lock immediately; wait for it. */ //不能马上获取锁,等待 LockRelationForExtension(relation, ExclusiveLock); /* * Check if some other backend has extended a block for us while * we were waiting on the lock. */ //如有其它进程扩展了数据表，那么可以成功获取满足条件的targetBlock targetBlock = GetPageWithFreeSpace(relation, len + saveFreeSpace); /* * If some other waiter has already extended the relation, we * don't need to do so; just use the existing freespace. * 如果其他等待进程已经扩展了关系,那么我们不需要再扩展了,使用现成的空闲空间即可. */ if (targetBlock != InvalidBlockNumber) { UnlockRelationForExtension(relation, ExclusiveLock); goto loop; } /* Time to bulk-extend. */ //其它进程没有扩展 //Just extend it! RelationAddExtraBlocks(relation, bistate); } } /* * In addition to whatever extension we performed above, we always add at * least one block to satisfy our own request. * 处理上面执行的扩展,我们总是添加了至少一个block用以满足自身需要. * * XXX This does an lseek - rather expensive - but at the moment it is the * only way to accurately determine how many blocks are in a relation. Is * it worth keeping an accurate file length in shared memory someplace, * rather than relying on the kernel to do it for us? * XXX 这相当于做了一次lseek - 相当昂贵的操作! - 在这时候这也是唯一可以准确确定关系有多少blocks的方法. * 相对于不是使用内核来完成这个事情,在内存的某个地方保存准确的文件尺寸是否更好? */ //扩展表后，New Page！ buffer = ReadBufferBI(relation, P_NEW, bistate); /* * We can be certain that locking the otherBuffer first is OK, since it * must have a lower page number. * 这时候可以确定首先锁定的otherBuffer没有问题,因为它有一个较小的page编号 */ if (otherBuffer != InvalidBuffer) ////otherBuffer的顺序一定在扩展的Block之前，Lock it！ LockBuffer(otherBuffer, BUFFER_LOCK_EXCLUSIVE); /* * Now acquire lock on the new page. * 现在可以尝试为新page上锁 */ //锁定New Page LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE); /* * Release the file-extension lock; it's now OK for someone else to extend * the relation some more. Note that we cannot release this lock before * we have buffer lock on the new page, or we risk a race condition * against vacuumlazy.c --- see comments therein. * 是否文件扩展锁.现在对于其他进程来说可以扩展relation了. * 注意不能在持有新page的buffer lock前释放该锁,否则将会在vacuumlazy.c中存在条件竞争. * 详细可参见注释. */ if (needLock) //释放扩展锁 UnlockRelationForExtension(relation, ExclusiveLock); /* * We need to initialize the empty new page. Double-check that it really * is empty (this should never happen, but if it does we don't want to * risk wiping out valid data). * 我们需要初始化空的新page. * 需再次检查该page是空的(这应该不会出现,但执行这个操作是因为我们不希望冒删除有效数据的风险) */ //获取相应的Page page = BufferGetPage(buffer); if (!PageIsNew(page)) //不是New Page，那一定某个地方搞错了！ elog(ERROR, "page %u of relation \"%s\" should be empty but is not", BufferGetBlockNumber(buffer), RelationGetRelationName(relation)); //初始化New Page PageInit(page, BufferGetPageSize(buffer), 0); //New Page也满足不了要求的大小，报错 if (len > PageGetHeapFreeSpace(page)) { /* We should not get here given the test at the top */ elog(PANIC, "tuple is too big: size %zu", len); } /* * Remember the new page as our target for future insertions. * 记录新page为未来插入的目标page. * * XXX should we enter the new page into the free space map immediately, * or just keep it for this backend's exclusive use in the short run * (until VACUUM sees it)? Seems to depend on whether you expect the * current backend to make more insertions or not, which is probably a * good bet most of the time. So for now, don't add it to FSM yet. * XXX 我们应该马上把新的page放到FSM中吗, * 或者只是把该page放在后台进程的私有空间中在很短时间内独占使用(直至vacuum可以看到它位置)? * 看起来这依赖于你希望当前的后台进程是否执行更多的插入操作,这在大多数时间下会更好. * 因此,现在还没有把它添加到FSM中. */ //终于找到了可用于存储数据的Block RelationSetTargetBlock(relation, BufferGetBlockNumber(buffer)); //返回 return buffer; } ``` ### 三、跟踪分析 测试脚本 ``` 15:54:13 (xdb@[local]:5432)testdb=# insert into t1 values (1,'1','1'); ``` 调用栈 ``` (gdb) b RelationGetBufferForTuple Breakpoint 1 at 0x4ef179: file hio.c, line 318. (gdb) c Continuing. Breakpoint 1, RelationGetBufferForTuple (relation=0x7f4f51fe39b8, len=32, otherBuffer=0, options=0, bistate=0x0, vmbuffer=0x7ffea95dbf6c, vmbuffer_other=0x0) at hio.c:318 318 bool use_fsm = !(options & HEAP_INSERT_SKIP_FSM); (gdb) bt #0 RelationGetBufferForTuple (relation=0x7f4f51fe39b8, len=32, otherBuffer=0, options=0, bistate=0x0, vmbuffer=0x7ffea95dbf6c, vmbuffer_other=0x0) at hio.c:318 #1 0x00000000004df1f8 in heap_insert (relation=0x7f4f51fe39b8, tup=0x178a478, cid=0, options=0, bistate=0x0) at heapam.c:2468 #2 0x0000000000709dda in ExecInsert (mtstate=0x178a220, slot=0x178a680, planSlot=0x178a680, estate=0x1789eb8, canSetTag=true) at nodeModifyTable.c:529 #3 0x000000000070c475 in ExecModifyTable (pstate=0x178a220) at nodeModifyTable.c:2159 #4 0x00000000006e05cb in ExecProcNodeFirst (node=0x178a220) at execProcnode.c:445 #5 0x00000000006d552e in ExecProcNode (node=0x178a220) at ../../../src/include/executor/executor.h:247 #6 0x00000000006d7d66 in ExecutePlan (estate=0x1789eb8, planstate=0x178a220, use_parallel_mode=false, operation=CMD_INSERT, sendTuples=false, numberTuples=0, direction=ForwardScanDirection, dest=0x17a7688, execute_once=true) at execMain.c:1723 #7 0x00000000006d5af8 in standard_ExecutorRun (queryDesc=0x178e458, direction=ForwardScanDirection, count=0, execute_once=true) at execMain.c:364 #8 0x00000000006d5920 in ExecutorRun (queryDesc=0x178e458, direction=ForwardScanDirection, count=0, execute_once=true) at execMain.c:307 #9 0x00000000008c1092 in ProcessQuery (plan=0x16b3ac0, sourceText=0x16b1ec8 "insert into t1 values (1,'1','1');", params=0x0, queryEnv=0x0, dest=0x17a7688, completionTag=0x7ffea95dc500 "") at pquery.c:161 #10 0x00000000008c29a1 in PortalRunMulti (portal=0x1717488, isTopLevel=true, setHoldSnapshot=false, dest=0x17a7688, altdest=0x17a7688, completionTag=0x7ffea95dc500 "") at pquery.c:1286 #11 0x00000000008c1f7a in PortalRun (portal=0x1717488, count=9223372036854775807, isTopLevel=true, run_once=true, dest=0x17a7688, altdest=0x17a7688, completionTag=0x7ffea95dc500 "") at pquery.c:799 #12 0x00000000008bbf16 in exec_simple_query (query_string=0x16b1ec8 "insert into t1 values (1,'1','1');") at postgres.c:1145 #13 0x00000000008c01a1 in PostgresMain (argc=1, argv=0x16dbaf8, dbname=0x16db960 "testdb", username=0x16aeba8 "xdb") at postgres.c:4182 #14 0x000000000081e07c in BackendRun (port=0x16d3940) at postmaster.c:4361 #15 0x000000000081d7ef in BackendStartup (port=0x16d3940) at postmaster.c:4033 ---Type <return> to continue, or q <return> to quit--- #16 0x0000000000819be9 in ServerLoop () at postmaster.c:1706 #17 0x000000000081949f in PostmasterMain (argc=1, argv=0x16acb60) at postmaster.c:1379 #18 0x0000000000742941 in main (argc=1, argv=0x16acb60) at main.c:228 (gdb) ``` 详细跟踪分析待续... ### 四、参考资料 [PG Source Code](https://doxygen.postgresql.org) [Buffer Manager](http://www.interdb.jp/pg/pgsql08.html)