From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jaegeuk Kim Subject: Re: [PATCH v2] f2fs: get victim segment again after new cp Date: Mon, 25 Jul 2016 11:40:05 -0700 Message-ID: <20160725184005.GB56718@jaegeuk> References: <1469185711-27932-1-git-send-email-heyunlei@huawei.com> <20160722185922.GA36408@jaegeuk> <6fbd66c3-ccfc-0389-dde5-81b901cb4586@kernel.org> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: Received: from sog-mx-2.v43.ch3.sourceforge.com ([172.29.43.192] helo=mx.sourceforge.net) by sfs-ml-1.v29.ch3.sourceforge.com with esmtp (Exim 4.76) (envelope-from ) id 1bRknT-0002BU-Pw for linux-f2fs-devel@lists.sourceforge.net; Mon, 25 Jul 2016 18:40:11 +0000 Received: from mail.kernel.org ([198.145.29.136]) by sog-mx-2.v43.ch3.sourceforge.com with esmtps (TLSv1:AES256-SHA:256) (Exim 4.76) id 1bRknS-0003WF-FH for linux-f2fs-devel@lists.sourceforge.net; Mon, 25 Jul 2016 18:40:11 +0000 Content-Disposition: inline In-Reply-To: <6fbd66c3-ccfc-0389-dde5-81b901cb4586@kernel.org> List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: linux-f2fs-devel-bounces@lists.sourceforge.net To: Chao Yu Cc: linux-f2fs-devel@lists.sourceforge.net On Sun, Jul 24, 2016 at 06:38:46AM +0800, Chao Yu wrote: ... > >> diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c > >> index de6c41c..ec17096 100644 > >> --- a/fs/f2fs/gc.c > >> +++ b/fs/f2fs/gc.c > >> @@ -908,10 +908,14 @@ gc_more: > >> * enough free sections, we should flush dent/node blocks and do > >> * garbage collections. > >> */ > >> - if (__get_victim(sbi, &segno, gc_type) || prefree_segments(sbi)) > >> + if (__get_victim(sbi, &segno, gc_type) || prefree_segments(sbi)) { > >> write_checkpoint(sbi, &cpc); > >> - else if (has_not_enough_free_secs(sbi, 0)) > > > > If segno is NULL_SEGNO, we get a panic when checking the below conditions. > > Anyway, I don't think we need this condition at all. > > I thinks the condition which trigger our f2fs_bug_on is really a corner case, > we'd better not waste the victim we found everytime, especially for these > victims from last background gc's victims. > > What about checking segno first? > > if(segno != NULL_SEGNO && (!get_valid_blocks(sbi, segno, sbi->segs_per_sec) || > IS_CURSEC(sbi, GET_SECNO(sbi, segno))) > segno = NULL_SEGNO; My viewpoint is write_checkpoint() flushes node and dentry pages, which provides another better victims. Another concern is get_valid_blocks and IS_CURSEG are not covered by any lock. Thanks, > > Thanks, > > > Let me remove them and just set NULL_SEGNO only after checkpoint. > > Please check the dev-test repo. > > > > Thanks, > > > >> + if(!get_valid_blocks(sbi, segno, sbi->segs_per_sec) > >> + || IS_CURSEC(sbi, segno / sbi->segs_per_sec)) > >> + segno = NULL_SEGNO; > >> + } else if (has_not_enough_free_secs(sbi, 0)) { > >> write_checkpoint(sbi, &cpc); > >> + } > >> } > >> > >> if (segno == NULL_SEGNO && !__get_victim(sbi, &segno, gc_type)) > >> -- > >> 1.9.1 > > > > ------------------------------------------------------------------------------ > > What NetFlow Analyzer can do for you? Monitors network bandwidth and traffic > > patterns at an interface-level. Reveals which users, apps, and protocols are > > consuming the most bandwidth. Provides multi-vendor support for NetFlow, > > J-Flow, sFlow and other flows. Make informed decisions using capacity planning > > reports.http://sdm.link/zohodev2dev > > _______________________________________________ > > Linux-f2fs-devel mailing list > > Linux-f2fs-devel@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel > > ------------------------------------------------------------------------------ What NetFlow Analyzer can do for you? Monitors network bandwidth and traffic patterns at an interface-level. Reveals which users, apps, and protocols are consuming the most bandwidth. Provides multi-vendor support for NetFlow, J-Flow, sFlow and other flows. Make informed decisions using capacity planning reports.http://sdm.link/zohodev2dev