From patchwork Fri Aug 31 15:13:25 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Brian Foster X-Patchwork-Id: 10584065 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 88787112B for ; Fri, 31 Aug 2018 15:13:30 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 77E8B2BE51 for ; Fri, 31 Aug 2018 15:13:30 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id 6C0C62BE6C; Fri, 31 Aug 2018 15:13:30 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-7.9 required=2.0 tests=BAYES_00,MAILING_LIST_MULTI, RCVD_IN_DNSWL_HI autolearn=ham version=3.3.1 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id C2B712BE51 for ; Fri, 31 Aug 2018 15:13:29 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1727326AbeHaTVZ (ORCPT ); Fri, 31 Aug 2018 15:21:25 -0400 Received: from mx3-rdu2.redhat.com ([66.187.233.73]:35670 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1727618AbeHaTVZ (ORCPT ); Fri, 31 Aug 2018 15:21:25 -0400 Received: from smtp.corp.redhat.com (int-mx06.intmail.prod.int.rdu2.redhat.com [10.11.54.6]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id B642781663EA for ; Fri, 31 Aug 2018 15:13:27 +0000 (UTC) Received: from bfoster.bos.redhat.com (dhcp-41-2.bos.redhat.com [10.18.41.2]) by smtp.corp.redhat.com (Postfix) with ESMTP id 9863B2166B41 for ; Fri, 31 Aug 2018 15:13:27 +0000 (UTC) From: Brian Foster To: linux-xfs@vger.kernel.org Subject: [PATCH v4 1/3] xfs: don't unlock invalidated buf on aborted tx commit Date: Fri, 31 Aug 2018 11:13:25 -0400 Message-Id: <20180831151327.41225-2-bfoster@redhat.com> In-Reply-To: <20180831151327.41225-1-bfoster@redhat.com> References: <20180831151327.41225-1-bfoster@redhat.com> X-Scanned-By: MIMEDefang 2.78 on 10.11.54.6 X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.11.55.8]); Fri, 31 Aug 2018 15:13:27 +0000 (UTC) X-Greylist: inspected by milter-greylist-4.5.16 (mx1.redhat.com [10.11.55.8]); Fri, 31 Aug 2018 15:13:27 +0000 (UTC) for IP:'10.11.54.6' DOMAIN:'int-mx06.intmail.prod.int.rdu2.redhat.com' HELO:'smtp.corp.redhat.com' FROM:'bfoster@redhat.com' RCPT:'' Sender: linux-xfs-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-xfs@vger.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP xfstests generic/388,475 occasionally reproduce assertion failures in xfs_buf_item_unpin() when the final bli reference is dropped on an invalidated buffer and the buffer is not locked as it is expected to be. Invalidated buffers should remain locked on transaction commit until the final unpin, at which point the buffer is removed from the AIL and the bli is freed since stale buffers are not written back. The assert failures are associated with filesystem shutdown, typically due to log I/O errors injected by the test. The problematic situation can occur if the shutdown happens to cause a race between an active transaction that has invalidated a particular buffer and an I/O error on a log buffer that contains the bli associated with the same (now stale) buffer. Both transaction and log contexts acquire a bli reference. If the transaction has already invalidated the buffer by the time the I/O error occurs and ends up aborting due to shutdown, the transaction and log hold the last two references to a stale bli. If the transaction cancel occurs first, it treats the buffer as non-stale due to the aborted state: the bli reference is dropped and the buffer is released/unlocked. The log buffer I/O error handling eventually calls into xfs_buf_item_unpin(), drops the final reference to the bli and treats it as stale. The buffer wasn't left locked by xfs_buf_item_unlock(), however, so the assert fails and the buffer is double unlocked. The latter problem is mitigated by the fact that the fs is shutdown and no further damage is possible. ->iop_unlock() of an invalidated buffer should behave consistently with respect to the bli refcount, regardless of aborted state. If the refcount remains elevated on commit, we know the bli is awaiting an unpin (since it can't be in another transaction) and will be handled appropriately on log buffer completion. If the final bli reference of an invalidated buffer is dropped in ->iop_unlock(), we can assume the transaction has aborted because invalidation implies a dirty transaction. In the non-abort case, the log would have acquired a bli reference in ->iop_pin() and prevented bli release at ->iop_unlock() time. In the abort case the item must be freed and buffer unlocked because it wasn't pinned by the log. Rework xfs_buf_item_unlock() to simplify the currently circuitous and duplicate logic and leave invalidated buffers locked based on bli refcount, regardless of aborted state. This ensures that a pinned, stale buffer is always found locked when eventually unpinned. Signed-off-by: Brian Foster --- fs/xfs/xfs_buf_item.c | 91 ++++++++++++++++++++----------------------- fs/xfs/xfs_trace.h | 1 - 2 files changed, 42 insertions(+), 50 deletions(-) diff --git a/fs/xfs/xfs_buf_item.c b/fs/xfs/xfs_buf_item.c index 1c9d1398980b..42fce70b474d 100644 --- a/fs/xfs/xfs_buf_item.c +++ b/fs/xfs/xfs_buf_item.c @@ -556,73 +556,66 @@ xfs_buf_item_unlock( { struct xfs_buf_log_item *bip = BUF_ITEM(lip); struct xfs_buf *bp = bip->bli_buf; + bool freed; bool aborted; - bool hold = !!(bip->bli_flags & XFS_BLI_HOLD); - bool dirty = !!(bip->bli_flags & XFS_BLI_DIRTY); + bool hold = bip->bli_flags & XFS_BLI_HOLD; + bool dirty = bip->bli_flags & XFS_BLI_DIRTY; + bool stale = bip->bli_flags & XFS_BLI_STALE; #if defined(DEBUG) || defined(XFS_WARN) - bool ordered = !!(bip->bli_flags & XFS_BLI_ORDERED); + bool ordered = bip->bli_flags & XFS_BLI_ORDERED; #endif - aborted = test_bit(XFS_LI_ABORTED, &lip->li_flags); - - /* Clear the buffer's association with this transaction. */ - bp->b_transp = NULL; - - /* - * The per-transaction state has been copied above so clear it from the - * bli. - */ - bip->bli_flags &= ~(XFS_BLI_LOGGED | XFS_BLI_HOLD | XFS_BLI_ORDERED); - - /* - * If the buf item is marked stale, then don't do anything. We'll - * unlock the buffer and free the buf item when the buffer is unpinned - * for the last time. - */ - if (bip->bli_flags & XFS_BLI_STALE) { - trace_xfs_buf_item_unlock_stale(bip); - ASSERT(bip->__bli_format.blf_flags & XFS_BLF_CANCEL); - if (!aborted) { - atomic_dec(&bip->bli_refcount); - return; - } - } - trace_xfs_buf_item_unlock(bip); /* - * If the buf item isn't tracking any data, free it, otherwise drop the - * reference we hold to it. If we are aborting the transaction, this may - * be the only reference to the buf item, so we free it anyway - * regardless of whether it is dirty or not. A dirty abort implies a - * shutdown, anyway. - * * The bli dirty state should match whether the blf has logged segments * except for ordered buffers, where only the bli should be dirty. */ ASSERT((!ordered && dirty == xfs_buf_item_dirty_format(bip)) || (ordered && dirty && !xfs_buf_item_dirty_format(bip))); + ASSERT(!stale || (bip->__bli_format.blf_flags & XFS_BLF_CANCEL)); + + aborted = test_bit(XFS_LI_ABORTED, &lip->li_flags); /* - * Clean buffers, by definition, cannot be in the AIL. However, aborted - * buffers may be in the AIL regardless of dirty state. An aborted - * transaction that invalidates a buffer already in the AIL may have - * marked it stale and cleared the dirty state, for example. - * - * Therefore if we are aborting a buffer and we've just taken the last - * reference away, we have to check if it is in the AIL before freeing - * it. We need to free it in this case, because an aborted transaction - * has already shut the filesystem down and this is the last chance we - * will have to do so. + * Clear the buffer's association with this transaction and + * per-transaction state from the bli, which has been copied above. + */ + bp->b_transp = NULL; + bip->bli_flags &= ~(XFS_BLI_LOGGED | XFS_BLI_HOLD | XFS_BLI_ORDERED); + + /* + * Drop the transaction's bli reference and deal with the item if we had + * the last one. We must free the item if clean or aborted since it + * wasn't pinned by the log and this is the last chance to do so. If the + * bli is freed and dirty (but non-aborted), the buffer was not dirty in + * this transaction but modified by a previous one and still awaiting + * writeback. In that case, the bli is freed on buffer writeback + * completion. */ - if (atomic_dec_and_test(&bip->bli_refcount)) { - if (aborted) { - ASSERT(XFS_FORCED_SHUTDOWN(lip->li_mountp)); + freed = atomic_dec_and_test(&bip->bli_refcount); + if (freed) { + ASSERT(!aborted || XFS_FORCED_SHUTDOWN(lip->li_mountp)); + /* + * An aborted item may be in the AIL regardless of dirty state. + * For example, consider an aborted transaction that invalidated + * a dirty bli and cleared the dirty state. + */ + if (aborted) xfs_trans_ail_remove(lip, SHUTDOWN_LOG_IO_ERROR); + if (aborted || !dirty) xfs_buf_item_relse(bp); - } else if (!dirty) - xfs_buf_item_relse(bp); + } else if (stale) { + /* + * Stale buffers remain locked until final unpin unless the bli + * was freed in the branch above. A freed stale bli implies an + * abort because buffer invalidation dirties the bli and + * transaction. + */ + ASSERT(!freed); + return; } + ASSERT(!stale || (aborted && freed)); if (!hold) xfs_buf_relse(bp); diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h index ad315e83bc02..3043e5ed6495 100644 --- a/fs/xfs/xfs_trace.h +++ b/fs/xfs/xfs_trace.h @@ -473,7 +473,6 @@ DEFINE_BUF_ITEM_EVENT(xfs_buf_item_pin); DEFINE_BUF_ITEM_EVENT(xfs_buf_item_unpin); DEFINE_BUF_ITEM_EVENT(xfs_buf_item_unpin_stale); DEFINE_BUF_ITEM_EVENT(xfs_buf_item_unlock); -DEFINE_BUF_ITEM_EVENT(xfs_buf_item_unlock_stale); DEFINE_BUF_ITEM_EVENT(xfs_buf_item_committed); DEFINE_BUF_ITEM_EVENT(xfs_buf_item_push); DEFINE_BUF_ITEM_EVENT(xfs_trans_get_buf); From patchwork Fri Aug 31 15:13:26 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Brian Foster X-Patchwork-Id: 10584067 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 9F76B175A for ; Fri, 31 Aug 2018 15:13:30 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 91BC42BE5B for ; Fri, 31 Aug 2018 15:13:30 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id 839452BE55; Fri, 31 Aug 2018 15:13:30 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-7.9 required=2.0 tests=BAYES_00,MAILING_LIST_MULTI, RCVD_IN_DNSWL_HI autolearn=ham version=3.3.1 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id ED4DF2BE55 for ; Fri, 31 Aug 2018 15:13:29 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1727618AbeHaTVZ (ORCPT ); Fri, 31 Aug 2018 15:21:25 -0400 Received: from mx3-rdu2.redhat.com ([66.187.233.73]:35672 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1728148AbeHaTVZ (ORCPT ); Fri, 31 Aug 2018 15:21:25 -0400 Received: from smtp.corp.redhat.com (int-mx06.intmail.prod.int.rdu2.redhat.com [10.11.54.6]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id D94628185331 for ; Fri, 31 Aug 2018 15:13:27 +0000 (UTC) Received: from bfoster.bos.redhat.com (dhcp-41-2.bos.redhat.com [10.18.41.2]) by smtp.corp.redhat.com (Postfix) with ESMTP id BDBBE2166B41 for ; Fri, 31 Aug 2018 15:13:27 +0000 (UTC) From: Brian Foster To: linux-xfs@vger.kernel.org Subject: [PATCH v4 2/3] xfs: clean up xfs_trans_brelse() Date: Fri, 31 Aug 2018 11:13:26 -0400 Message-Id: <20180831151327.41225-3-bfoster@redhat.com> In-Reply-To: <20180831151327.41225-1-bfoster@redhat.com> References: <20180831151327.41225-1-bfoster@redhat.com> X-Scanned-By: MIMEDefang 2.78 on 10.11.54.6 X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.11.55.8]); Fri, 31 Aug 2018 15:13:27 +0000 (UTC) X-Greylist: inspected by milter-greylist-4.5.16 (mx1.redhat.com [10.11.55.8]); Fri, 31 Aug 2018 15:13:27 +0000 (UTC) for IP:'10.11.54.6' DOMAIN:'int-mx06.intmail.prod.int.rdu2.redhat.com' HELO:'smtp.corp.redhat.com' FROM:'bfoster@redhat.com' RCPT:'' Sender: linux-xfs-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-xfs@vger.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP xfs_trans_brelse() is a bit of a historical mess, similar to xfs_buf_item_unlock(). It is unnecessarily verbose, has snippets of commented out code, inconsistency with regard to stale items, etc. Clean up xfs_trans_brelse() to use similar logic and flow as xfs_buf_item_unlock() with regard to bli reference count handling. This patch makes no functional changes, but facilitates further refactoring of the common bli reference count handling code. Signed-off-by: Brian Foster --- fs/xfs/xfs_trans_buf.c | 110 +++++++++++++++-------------------------- 1 file changed, 39 insertions(+), 71 deletions(-) diff --git a/fs/xfs/xfs_trans_buf.c b/fs/xfs/xfs_trans_buf.c index 15919f67a88f..7498f87ceed3 100644 --- a/fs/xfs/xfs_trans_buf.c +++ b/fs/xfs/xfs_trans_buf.c @@ -322,49 +322,40 @@ xfs_trans_read_buf_map( } /* - * Release the buffer bp which was previously acquired with one of the - * xfs_trans_... buffer allocation routines if the buffer has not - * been modified within this transaction. If the buffer is modified - * within this transaction, do decrement the recursion count but do - * not release the buffer even if the count goes to 0. If the buffer is not - * modified within the transaction, decrement the recursion count and - * release the buffer if the recursion count goes to 0. + * Release a buffer previously joined to the transaction. If the buffer is + * modified within this transaction, decrement the recursion count but do not + * release the buffer even if the count goes to 0. If the buffer is not modified + * within the transaction, decrement the recursion count and release the buffer + * if the recursion count goes to 0. * - * If the buffer is to be released and it was not modified before - * this transaction began, then free the buf_log_item associated with it. + * If the buffer is to be released and it was not already dirty before this + * transaction began, then also free the buf_log_item associated with it. * - * If the transaction pointer is NULL, make this just a normal - * brelse() call. + * If the transaction pointer is NULL, this is a normal xfs_buf_relse() call. */ void xfs_trans_brelse( - xfs_trans_t *tp, - xfs_buf_t *bp) + struct xfs_trans *tp, + struct xfs_buf *bp) { - struct xfs_buf_log_item *bip; - int freed; + struct xfs_buf_log_item *bip = bp->b_log_item; + bool freed; + bool dirty; - /* - * Default to a normal brelse() call if the tp is NULL. - */ - if (tp == NULL) { - ASSERT(bp->b_transp == NULL); + ASSERT(bp->b_transp == tp); + + if (!tp) { xfs_buf_relse(bp); return; } - ASSERT(bp->b_transp == tp); - bip = bp->b_log_item; + trace_xfs_trans_brelse(bip); ASSERT(bip->bli_item.li_type == XFS_LI_BUF); - ASSERT(!(bip->bli_flags & XFS_BLI_STALE)); - ASSERT(!(bip->__bli_format.blf_flags & XFS_BLF_CANCEL)); ASSERT(atomic_read(&bip->bli_refcount) > 0); - trace_xfs_trans_brelse(bip); - /* - * If the release is just for a recursive lock, - * then decrement the count and return. + * If the release is for a recursive lookup, then decrement the count + * and return. */ if (bip->bli_recur > 0) { bip->bli_recur--; @@ -372,63 +363,40 @@ xfs_trans_brelse( } /* - * If the buffer is dirty within this transaction, we can't + * If the buffer is invalidated or dirty in this transaction, we can't * release it until we commit. */ if (test_bit(XFS_LI_DIRTY, &bip->bli_item.li_flags)) return; - - /* - * If the buffer has been invalidated, then we can't release - * it until the transaction commits to disk unless it is re-dirtied - * as part of this transaction. This prevents us from pulling - * the item from the AIL before we should. - */ if (bip->bli_flags & XFS_BLI_STALE) return; - ASSERT(!(bip->bli_flags & XFS_BLI_LOGGED)); - /* - * Free up the log item descriptor tracking the released item. + * Unlink the log item from the transaction and clear the hold flag, if + * set. We wouldn't want the next user of the buffer to get confused. */ + ASSERT(!(bip->bli_flags & XFS_BLI_LOGGED)); xfs_trans_del_item(&bip->bli_item); + bip->bli_flags &= ~XFS_BLI_HOLD; /* - * Clear the hold flag in the buf log item if it is set. - * We wouldn't want the next user of the buffer to - * get confused. - */ - if (bip->bli_flags & XFS_BLI_HOLD) { - bip->bli_flags &= ~XFS_BLI_HOLD; - } - - /* - * Drop our reference to the buf log item. + * Drop the reference to the bli. At this point, the bli must be either + * freed or dirty (or both). If freed, there are a couple cases where we + * are responsible to free the item. If the bli is clean, we're the last + * user of it. If the fs has shut down, the bli may be dirty and AIL + * resident, but won't ever be written back. We therefore may also need + * to remove it from the AIL before freeing it. */ freed = atomic_dec_and_test(&bip->bli_refcount); - - /* - * If the buf item is not tracking data in the log, then we must free it - * before releasing the buffer back to the free pool. - * - * If the fs has shutdown and we dropped the last reference, it may fall - * on us to release a (possibly dirty) bli if it never made it to the - * AIL (e.g., the aborted unpin already happened and didn't release it - * due to our reference). Since we're already shutdown and need - * ail_lock, just force remove from the AIL and release the bli here. - */ - if (XFS_FORCED_SHUTDOWN(tp->t_mountp) && freed) { - xfs_trans_ail_remove(&bip->bli_item, SHUTDOWN_LOG_IO_ERROR); - xfs_buf_item_relse(bp); - } else if (!(bip->bli_flags & XFS_BLI_DIRTY)) { -/*** - ASSERT(bp->b_pincount == 0); -***/ - ASSERT(atomic_read(&bip->bli_refcount) == 0); - ASSERT(!test_bit(XFS_LI_IN_AIL, &bip->bli_item.li_flags)); - ASSERT(!(bip->bli_flags & XFS_BLI_INODE_ALLOC_BUF)); - xfs_buf_item_relse(bp); + dirty = bip->bli_flags & XFS_BLI_DIRTY; + ASSERT(freed || dirty); + if (freed) { + bool abort = XFS_FORCED_SHUTDOWN(tp->t_mountp); + ASSERT(abort || !test_bit(XFS_LI_IN_AIL, &bip->bli_item.li_flags)); + if (abort) + xfs_trans_ail_remove(&bip->bli_item, SHUTDOWN_LOG_IO_ERROR); + if (!dirty || abort) + xfs_buf_item_relse(bp); } bp->b_transp = NULL; From patchwork Fri Aug 31 15:13:27 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Brian Foster X-Patchwork-Id: 10584069 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 5D545112B for ; Fri, 31 Aug 2018 15:13:31 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 4D9C42BE51 for ; Fri, 31 Aug 2018 15:13:31 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id 41C2D2BE5B; Fri, 31 Aug 2018 15:13:31 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-7.9 required=2.0 tests=BAYES_00,MAILING_LIST_MULTI, RCVD_IN_DNSWL_HI autolearn=ham version=3.3.1 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 8F6742BE51 for ; Fri, 31 Aug 2018 15:13:30 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1728148AbeHaTV0 (ORCPT ); Fri, 31 Aug 2018 15:21:26 -0400 Received: from mx3-rdu2.redhat.com ([66.187.233.73]:54104 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1728529AbeHaTVZ (ORCPT ); Fri, 31 Aug 2018 15:21:25 -0400 Received: from smtp.corp.redhat.com (int-mx06.intmail.prod.int.rdu2.redhat.com [10.11.54.6]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id 08E077DAC9 for ; Fri, 31 Aug 2018 15:13:28 +0000 (UTC) Received: from bfoster.bos.redhat.com (dhcp-41-2.bos.redhat.com [10.18.41.2]) by smtp.corp.redhat.com (Postfix) with ESMTP id DC0342166BA1 for ; Fri, 31 Aug 2018 15:13:27 +0000 (UTC) From: Brian Foster To: linux-xfs@vger.kernel.org Subject: [PATCH v4 3/3] xfs: refactor xfs_buf_log_item reference count handling Date: Fri, 31 Aug 2018 11:13:27 -0400 Message-Id: <20180831151327.41225-4-bfoster@redhat.com> In-Reply-To: <20180831151327.41225-1-bfoster@redhat.com> References: <20180831151327.41225-1-bfoster@redhat.com> X-Scanned-By: MIMEDefang 2.78 on 10.11.54.6 X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.11.55.2]); Fri, 31 Aug 2018 15:13:28 +0000 (UTC) X-Greylist: inspected by milter-greylist-4.5.16 (mx1.redhat.com [10.11.55.2]); Fri, 31 Aug 2018 15:13:28 +0000 (UTC) for IP:'10.11.54.6' DOMAIN:'int-mx06.intmail.prod.int.rdu2.redhat.com' HELO:'smtp.corp.redhat.com' FROM:'bfoster@redhat.com' RCPT:'' Sender: linux-xfs-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-xfs@vger.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP The xfs_buf_log_item structure has a reference counter with slightly tricky semantics. In the common case, a buffer is logged and committed in a transaction, committed to the on-disk log (added to the AIL) and then finally written back and removed from the AIL. The bli refcount covers two potentially overlapping timeframes: 1. the bli is held in an active transaction 2. the bli is pinned by the log The caveat to this approach is that the reference counter does not purely dictate the lifetime of the bli. IOW, when a dirty buffer is physically logged and unpinned, the bli refcount may go to zero as the log item is inserted into the AIL. Only once the buffer is written back can the bli finally be freed. The above semantics means that it is not enough for the various refcount decrementing contexts to release the bli on decrement to zero. xfs_trans_brelse(), transaction commit (->iop_unlock()) and unpin (->iop_unpin()) must all drop the associated reference and make additional checks to determine if the current context is responsible for freeing the item. For example, if a transaction holds but does not dirty a particular bli, the commit may drop the refcount to zero. If the bli itself is clean, it is also not AIL resident and must be freed at this time. The same is true for xfs_trans_brelse(). If the transaction dirties a bli and then aborts or an unpin results in an abort due to a log I/O error, the last reference count holder is expected to explicitly remove the item from the AIL and release it (since an abort means filesystem shutdown and metadata writeback will never occur). This leads to fairly complex checks being replicated in a few different places. Since ->iop_unlock() and xfs_trans_brelse() are nearly identical, refactor the logic into a common helper that implements and documents the semantics in one place. This patch does not change behavior. Signed-off-by: Brian Foster --- fs/xfs/xfs_buf_item.c | 90 +++++++++++++++++++++++++----------------- fs/xfs/xfs_buf_item.h | 1 + fs/xfs/xfs_trans_buf.c | 23 +---------- 3 files changed, 56 insertions(+), 58 deletions(-) diff --git a/fs/xfs/xfs_buf_item.c b/fs/xfs/xfs_buf_item.c index 42fce70b474d..12d8455bfbb2 100644 --- a/fs/xfs/xfs_buf_item.c +++ b/fs/xfs/xfs_buf_item.c @@ -531,6 +531,49 @@ xfs_buf_item_push( return rval; } +/* + * Drop the buffer log item refcount and take appropriate action. This helper + * determines whether the bli must be freed or not, since a decrement to zero + * does not necessarily mean the bli is unused. + * + * Return true if the bli is freed, false otherwise. + */ +bool +xfs_buf_item_put( + struct xfs_buf_log_item *bip) +{ + struct xfs_log_item *lip = &bip->bli_item; + bool aborted; + bool dirty; + + /* drop the bli ref and return if it wasn't the last one */ + if (!atomic_dec_and_test(&bip->bli_refcount)) + return false; + + /* + * We dropped the last ref and must free the item if clean or aborted. + * If the bli is dirty and non-aborted, the buffer was clean in the + * transaction but still awaiting writeback from previous changes. In + * that case, the bli is freed on buffer writeback completion. + */ + aborted = test_bit(XFS_LI_ABORTED, &lip->li_flags) || + XFS_FORCED_SHUTDOWN(lip->li_mountp); + dirty = bip->bli_flags & XFS_BLI_DIRTY; + if (dirty && !aborted) + return false; + + /* + * The bli is aborted or clean. An aborted item may be in the AIL + * regardless of dirty state. For example, consider an aborted + * transaction that invalidated a dirty bli and cleared the dirty + * state. + */ + if (aborted) + xfs_trans_ail_remove(lip, SHUTDOWN_LOG_IO_ERROR); + xfs_buf_item_relse(bip->bli_buf); + return true; +} + /* * Release the buffer associated with the buf log item. If there is no dirty * logged data associated with the buffer recorded in the buf log item, then @@ -556,13 +599,12 @@ xfs_buf_item_unlock( { struct xfs_buf_log_item *bip = BUF_ITEM(lip); struct xfs_buf *bp = bip->bli_buf; - bool freed; - bool aborted; + bool released; bool hold = bip->bli_flags & XFS_BLI_HOLD; - bool dirty = bip->bli_flags & XFS_BLI_DIRTY; bool stale = bip->bli_flags & XFS_BLI_STALE; #if defined(DEBUG) || defined(XFS_WARN) bool ordered = bip->bli_flags & XFS_BLI_ORDERED; + bool dirty = bip->bli_flags & XFS_BLI_DIRTY; #endif trace_xfs_buf_item_unlock(bip); @@ -575,8 +617,6 @@ xfs_buf_item_unlock( (ordered && dirty && !xfs_buf_item_dirty_format(bip))); ASSERT(!stale || (bip->__bli_format.blf_flags & XFS_BLF_CANCEL)); - aborted = test_bit(XFS_LI_ABORTED, &lip->li_flags); - /* * Clear the buffer's association with this transaction and * per-transaction state from the bli, which has been copied above. @@ -585,40 +625,16 @@ xfs_buf_item_unlock( bip->bli_flags &= ~(XFS_BLI_LOGGED | XFS_BLI_HOLD | XFS_BLI_ORDERED); /* - * Drop the transaction's bli reference and deal with the item if we had - * the last one. We must free the item if clean or aborted since it - * wasn't pinned by the log and this is the last chance to do so. If the - * bli is freed and dirty (but non-aborted), the buffer was not dirty in - * this transaction but modified by a previous one and still awaiting - * writeback. In that case, the bli is freed on buffer writeback - * completion. + * Unref the item and unlock the buffer unless held or stale. Stale + * buffers remain locked until final unpin unless the bli is freed by + * the unref call. The latter implies shutdown because buffer + * invalidation dirties the bli and transaction. */ - freed = atomic_dec_and_test(&bip->bli_refcount); - if (freed) { - ASSERT(!aborted || XFS_FORCED_SHUTDOWN(lip->li_mountp)); - /* - * An aborted item may be in the AIL regardless of dirty state. - * For example, consider an aborted transaction that invalidated - * a dirty bli and cleared the dirty state. - */ - if (aborted) - xfs_trans_ail_remove(lip, SHUTDOWN_LOG_IO_ERROR); - if (aborted || !dirty) - xfs_buf_item_relse(bp); - } else if (stale) { - /* - * Stale buffers remain locked until final unpin unless the bli - * was freed in the branch above. A freed stale bli implies an - * abort because buffer invalidation dirties the bli and - * transaction. - */ - ASSERT(!freed); + released = xfs_buf_item_put(bip); + if (hold || (stale && !released)) return; - } - ASSERT(!stale || (aborted && freed)); - - if (!hold) - xfs_buf_relse(bp); + ASSERT(!stale || test_bit(XFS_LI_ABORTED, &lip->li_flags)); + xfs_buf_relse(bp); } /* diff --git a/fs/xfs/xfs_buf_item.h b/fs/xfs/xfs_buf_item.h index 3f7d7b72e7e6..90f65f891fab 100644 --- a/fs/xfs/xfs_buf_item.h +++ b/fs/xfs/xfs_buf_item.h @@ -51,6 +51,7 @@ struct xfs_buf_log_item { int xfs_buf_item_init(struct xfs_buf *, struct xfs_mount *); void xfs_buf_item_relse(struct xfs_buf *); +bool xfs_buf_item_put(struct xfs_buf_log_item *); void xfs_buf_item_log(struct xfs_buf_log_item *, uint, uint); bool xfs_buf_item_dirty_format(struct xfs_buf_log_item *); void xfs_buf_attach_iodone(struct xfs_buf *, diff --git a/fs/xfs/xfs_trans_buf.c b/fs/xfs/xfs_trans_buf.c index 7498f87ceed3..286a287ac57a 100644 --- a/fs/xfs/xfs_trans_buf.c +++ b/fs/xfs/xfs_trans_buf.c @@ -339,8 +339,6 @@ xfs_trans_brelse( struct xfs_buf *bp) { struct xfs_buf_log_item *bip = bp->b_log_item; - bool freed; - bool dirty; ASSERT(bp->b_transp == tp); @@ -379,25 +377,8 @@ xfs_trans_brelse( xfs_trans_del_item(&bip->bli_item); bip->bli_flags &= ~XFS_BLI_HOLD; - /* - * Drop the reference to the bli. At this point, the bli must be either - * freed or dirty (or both). If freed, there are a couple cases where we - * are responsible to free the item. If the bli is clean, we're the last - * user of it. If the fs has shut down, the bli may be dirty and AIL - * resident, but won't ever be written back. We therefore may also need - * to remove it from the AIL before freeing it. - */ - freed = atomic_dec_and_test(&bip->bli_refcount); - dirty = bip->bli_flags & XFS_BLI_DIRTY; - ASSERT(freed || dirty); - if (freed) { - bool abort = XFS_FORCED_SHUTDOWN(tp->t_mountp); - ASSERT(abort || !test_bit(XFS_LI_IN_AIL, &bip->bli_item.li_flags)); - if (abort) - xfs_trans_ail_remove(&bip->bli_item, SHUTDOWN_LOG_IO_ERROR); - if (!dirty || abort) - xfs_buf_item_relse(bp); - } + /* drop the reference to the bli */ + xfs_buf_item_put(bip); bp->b_transp = NULL; xfs_buf_relse(bp);