From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from lists.sourceforge.net (lists.sourceforge.net [216.105.38.7]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id 40F4EC76188 for ; Wed, 5 Apr 2023 15:56:04 +0000 (UTC) Received: from [127.0.0.1] (helo=sfs-ml-4.v29.lw.sourceforge.com) by sfs-ml-4.v29.lw.sourceforge.com with esmtp (Exim 4.95) (envelope-from ) id 1pk5UQ-0008MM-QH; Wed, 05 Apr 2023 15:56:03 +0000 Received: from [172.30.20.202] (helo=mx.sourceforge.net) by sfs-ml-4.v29.lw.sourceforge.com with esmtps (TLS1.2) tls TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384 (Exim 4.95) (envelope-from ) id 1pk5UO-0008M7-4c for linux-f2fs-devel@lists.sourceforge.net; Wed, 05 Apr 2023 15:56:01 +0000 DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=sourceforge.net; s=x; h=In-Reply-To:Content-Type:MIME-Version:References: Message-ID:Subject:Cc:To:From:Date:Sender:Reply-To:Content-Transfer-Encoding: Content-ID:Content-Description:Resent-Date:Resent-From:Resent-Sender: Resent-To:Resent-Cc:Resent-Message-ID:List-Id:List-Help:List-Unsubscribe: List-Subscribe:List-Post:List-Owner:List-Archive; bh=FzNMEaD/bfHX29ABhmTkmCmd8JTyK2ZVLIATWx+KG0A=; b=gDeNfl5JTcQS23o19x5k61lPEq 4BxlqfCJOb2JAFaZ6R5AGAUOZV1tNB8nHJM6ylM51FD4kQLcVhmzU79mMdeOYNit24/CvrCZryl/r 1HzR5kdaUfV5FVT6UTDH+lkkspafftdXfzWEQ6OSld1r366fLmds13Pbx94ljCdZ4v4A=; DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=sf.net; s=x ; h=In-Reply-To:Content-Type:MIME-Version:References:Message-ID:Subject:Cc:To :From:Date:Sender:Reply-To:Content-Transfer-Encoding:Content-ID: Content-Description:Resent-Date:Resent-From:Resent-Sender:Resent-To:Resent-Cc :Resent-Message-ID:List-Id:List-Help:List-Unsubscribe:List-Subscribe: List-Post:List-Owner:List-Archive; bh=FzNMEaD/bfHX29ABhmTkmCmd8JTyK2ZVLIATWx+KG0A=; b=nH589IiLvwxSR4hDdQlkUiVpRE cQZ9GL/yGN3ZqkpshyxsTTGpnH+9WeM/wQbqMn2h+KF0oFmCGW3z2DIqaNddDtfnGAwDKM6eI17fm zuK6vJd8bs/PpbFi20Ba1hhF1u3OuucM7opkToXLiSsk8lxbdo2RSKzhoydQXMRHoPl8=; Received: from dfw.source.kernel.org ([139.178.84.217]) by sfi-mx-2.v28.lw.sourceforge.com with esmtps (TLS1.2:ECDHE-RSA-AES256-GCM-SHA384:256) (Exim 4.95) id 1pk5UJ-0001wq-D3 for linux-f2fs-devel@lists.sourceforge.net; Wed, 05 Apr 2023 15:56:00 +0000 Received: from smtp.kernel.org (relay.kernel.org [52.25.139.140]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by dfw.source.kernel.org (Postfix) with ESMTPS id F29C061D4A for ; Wed, 5 Apr 2023 15:55:49 +0000 (UTC) Received: by smtp.kernel.org (Postfix) with ESMTPSA id 4EAD5C433D2; Wed, 5 Apr 2023 15:55:49 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=k20201202; t=1680710149; bh=YYDw0RuzTK3iw0hXHKgSXbk0yvOIgNiO+TT8XJvU7GY=; h=Date:From:To:Cc:Subject:References:In-Reply-To:From; b=fdPH2fhPJz3vAlXSaoPyW4UTbkLVVi0H+IHUaXTmwThxRofluxBVBnt6znjjFPdyw BAeI/YwpID3g6RgSFB8OU3ex7XFGFhMb3rg2Moohn1ycWtnnMpXBmiNnNEtjrYPfW/ qZSFrcxzHMndvVpivUYnhcJS/j8T2togNsRgifYybAuAHt8xYHnBRWzL5JpeU3PMlM BEixxAdul0D9ZDJekcLRatv7T1mJAsA2vjAyybts0YyvK4m6rW/Aql8VlxPtJWUcy7 5As/QtByH5ioBC7dfARZ0RSwE46tftdu92jN7Sj3vFV1n9UHQq/N+q1TOxTcI40rvf CgWKsB5ICqjLA== Date: Wed, 5 Apr 2023 08:55:47 -0700 From: Jaegeuk Kim To: Chao Yu Message-ID: References: <20230324071028.336982-1-chao@kernel.org> MIME-Version: 1.0 Content-Disposition: inline In-Reply-To: X-Headers-End: 1pk5UJ-0001wq-D3 Subject: Re: [f2fs-dev] [PATCH] f2fs: fix to trigger a checkpoint in the end of foreground garbage collection X-BeenThere: linux-f2fs-devel@lists.sourceforge.net X-Mailman-Version: 2.1.21 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: linux-kernel@vger.kernel.org, linux-f2fs-devel@lists.sourceforge.net Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Errors-To: linux-f2fs-devel-bounces@lists.sourceforge.net On 04/05, Chao Yu wrote: > On 2023/4/5 5:39, Jaegeuk Kim wrote: > > Can we do like this? > > > > From 9a58f0e59364241aa31b555cfe793d278e39b0dc Mon Sep 17 00:00:00 2001 > > From: Jaegeuk Kim > > Date: Tue, 4 Apr 2023 14:36:00 -0700 > > Subject: [PATCH] f2fs: do checkpoint when there's not enough free sections > > > > We didn't do checkpoint in FG_GC case, which may cause losing to reclaim prefree > > sctions in time. > > > > Fixes: 6f8d4455060d ("f2fs: avoid fi->i_gc_rwsem[WRITE] lock in f2fs_gc") > > Signed-off-by: Chao Yu > > Signed-off-by: Jaegeuk Kim > > --- > > fs/f2fs/gc.c | 24 +++++++++++------------- > > 1 file changed, 11 insertions(+), 13 deletions(-) > > > > diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c > > index 56c53dbe05c9..f1d0dd9c5a6c 100644 > > --- a/fs/f2fs/gc.c > > +++ b/fs/f2fs/gc.c > > @@ -1806,6 +1806,7 @@ int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control) > > }; > > unsigned int skipped_round = 0, round = 0; > > unsigned int upper_secs; > > + bool stop_gc = false; > > trace_f2fs_gc_begin(sbi->sb, gc_type, gc_control->no_bg_gc, > > gc_control->nr_free_secs, > > @@ -1876,19 +1877,15 @@ int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control) > > (gc_type == FG_GC) ? sec_freed : 0, 0)) { > > if (gc_type == FG_GC && sec_freed < gc_control->nr_free_secs) > > goto go_gc_more; > > - goto stop; > > - } > > - > > - /* FG_GC stops GC by skip_count */ > > - if (gc_type == FG_GC) { > > + stop_gc = true; > > I guess below condition is for emergency recycle of prefree segments during > foreground GC, in order to avoid exhausting free sections due to to many > metadata allocation during CP. > > if (free_sections(sbi) <= upper_secs + NR_GC_CHECKPOINT_SECS && > prefree_segments(sbi)) { > > But for common case, free_sections() is close to reserved_segments(), and > upper_secs + NR_GC_CHECKPOINT_SECS value may be far smaller than free_sections(), > so checkpoint may not be trggered as expected, IIUC. > > So it's fine to just trigger CP in the end of foreground garbage collection? My major concern is to avoid unnecessary checkpointing given multiple FG_GC requests were pending in parallel. And, I don't want to add so many combination which gives so many corner cases, and feel f2fs_gc() needs to call checkpoint automatically in the worst case scenario only. By the way, do we just need to call checkpoint here including FG_GC as well? 1832 1833 if (gc_type == BG_GC && has_not_enough_free_secs(sbi, 0, 0)) { 1834 /* 1835 * For example, if there are many prefree_segments below given 1836 * threshold, we can make them free by checkpoint. Then, we 1837 * secure free segments which doesn't need fggc any more. 1838 */ 1839 if (prefree_segments(sbi)) { 1840 ret = f2fs_write_checkpoint(sbi, &cpc); 1841 if (ret) 1842 goto stop; 1843 } 1844 if (has_not_enough_free_secs(sbi, 0, 0)) 1845 gc_type = FG_GC; 1846 } > > One other concern is for those path as below: > - disable_checkpoint > - ioc_gc > - ioc_gc_range > - ioc_resize > ... I think the upper caller should decide to call checkpoint, if they want to reclaim the prefree likewise f2fs_disable_checkpoint. > > We've passed gc_type as FG_GC, but the demand here is to migrate block in time, > rather than dirtying blocks, and callers don't expect checkpoint in f2fs_gc(), > instead the callers will do the checkpoit as it needs. > > That means it's better to decouple FG_GC and write_checkpoint behavior, so I > added another parameter .reclaim_space to just let f2fs_balance_fs() to trigger > checkpoit in the end of f2fs_gc(). > > Thanks, > > > + } else if (gc_type == FG_GC) { > > + /* FG_GC stops GC by skip_count */ > > if (sbi->skipped_gc_rwsem) > > skipped_round++; > > round++; > > if (skipped_round > MAX_SKIP_GC_COUNT && > > - skipped_round * 2 >= round) { > > - ret = f2fs_write_checkpoint(sbi, &cpc); > > - goto stop; > > - } > > + skipped_round * 2 >= round) > > + stop_gc = true; > > } > > __get_secs_required(sbi, NULL, &upper_secs, NULL); > > @@ -1901,12 +1898,13 @@ int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control) > > prefree_segments(sbi)) { > > ret = f2fs_write_checkpoint(sbi, &cpc); > > if (ret) > > - goto stop; > > + stop_gc = true; > > } > > go_gc_more: > > - segno = NULL_SEGNO; > > - goto gc_more; > > - > > + if (!stop_gc) { > > + segno = NULL_SEGNO; > > + goto gc_more; > > + } > > stop: > > SIT_I(sbi)->last_victim[ALLOC_NEXT] = 0; > > SIT_I(sbi)->last_victim[FLUSH_DEVICE] = gc_control->victim_segno; _______________________________________________ Linux-f2fs-devel mailing list Linux-f2fs-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel