Received-SPF: pass (google.com: domain of linux-ext4-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   "Ritesh Harjani (IBM)" <ritesh.list@gmail.com>
To:     Theodore Ts'o <tytso@mit.edu>
Cc:     linux-ext4@vger.kernel.org,
        Harshad Shirwadkar <harshadshirwadkar@gmail.com>,
        Wang Shilong <wshilong@ddn.com>,
        Andreas Dilger <adilger.kernel@dilger.ca>,
        Li Xi <lixi@ddn.com>, Ritesh Harjani <ritesh.list@gmail.com>
Subject: [RFCv1 50/72] e2fsck: fix readahead for pfsck of pass1
Date:   Mon,  7 Nov 2022 17:51:38 +0530
Message-Id: <7fce79fc82c97b0ff5c9b7e81f3c2f343316c706.1667822611.git.ritesh.list@gmail.com>
In-Reply-To: <cover.1667822611.git.ritesh.list@gmail.com>
References: <cover.1667822611.git.ritesh.list@gmail.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

From: Wang Shilong <wshilong@ddn.com>

Several improvments for this patch:

1) move readahead_kb detection to preparing phase.
2) inode readahead should be aware of thread block group
boundary.
3) make readahead_kb aware of multiple threads.

Signed-off-by: Wang Shilong <wshilong@ddn.com>
Signed-off-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
---
 e2fsck/pass1.c | 29 ++++++++++++++++++++---------
 1 file changed, 20 insertions(+), 9 deletions(-)

diff --git a/e2fsck/pass1.c b/e2fsck/pass1.c
index 1d4f576c..5d07daec 100644
--- a/e2fsck/pass1.c
+++ b/e2fsck/pass1.c
@@ -1098,16 +1098,20 @@ out:
 static void pass1_readahead(e2fsck_t ctx, dgrp_t *group, ext2_ino_t *next_ino)
 {
 	ext2_ino_t inodes_in_group = 0, inodes_per_block, inodes_per_buffer;
-	dgrp_t start = *group, grp;
+	dgrp_t start = *group, grp, grp_end = ctx->fs->group_desc_count;
 	blk64_t blocks_to_read = 0;
 	errcode_t err = EXT2_ET_INVALID_ARGUMENT;
 
+#ifdef HAVE_PTHREAD
+	if (ctx->fs->fs_num_threads > 1)
+		grp_end = ctx->thread_info.et_group_end;
+#endif
 	if (ctx->readahead_kb == 0)
 		goto out;
 
 	/* Keep iterating groups until we have enough to readahead */
 	inodes_per_block = EXT2_INODES_PER_BLOCK(ctx->fs->super);
-	for (grp = start; grp < ctx->fs->group_desc_count; grp++) {
+	for (grp = start; grp < grp_end; grp++) {
 		if (ext2fs_bg_flags_test(ctx->fs, grp, EXT2_BG_INODE_UNINIT))
 			continue;
 		inodes_in_group = ctx->fs->super->s_inodes_per_group -
@@ -1300,12 +1304,25 @@ static errcode_t e2fsck_pass1_prepare(e2fsck_t ctx)
 {
 	struct problem_context pctx;
 	ext2_filsys fs = ctx->fs;
+	unsigned long long readahead_kb;
 
 	init_ext2_max_sizes();
-#ifdef	HAVE_PTHREAD
+#ifdef HAVE_PTHREAD
 	e2fsck_pass1_set_thread_num(ctx);
 #endif
+	/* If we can do readahead, figure out how many groups to pull in. */
+	if (!e2fsck_can_readahead(ctx->fs))
+		ctx->readahead_kb = 0;
+	else if (ctx->readahead_kb == ~0ULL)
+		ctx->readahead_kb = e2fsck_guess_readahead(ctx->fs);
 
+#ifdef HAVE_PTHREAD
+	/* don't use more than 1/10 of memory for threads checking */
+	readahead_kb = get_memory_size() / (10 * ctx->fs_num_threads);
+	/* maybe better disable RA if this is too small? */
+	if (ctx->readahead_kb > readahead_kb)
+		ctx->readahead_kb = readahead_kb;
+#endif
 	clear_problem_context(&pctx);
 	if (!(ctx->options & E2F_OPT_PREEN))
 		fix_problem(ctx, PR_1_PASS_HEADER, &pctx);
@@ -1482,13 +1499,7 @@ void e2fsck_pass1_run(e2fsck_t ctx)
 	init_resource_track(&rtrack, ctx->fs->io);
 	clear_problem_context(&pctx);
 
-	/* If we can do readahead, figure out how many groups to pull in. */
-	if (!e2fsck_can_readahead(ctx->fs))
-		ctx->readahead_kb = 0;
-	else if (ctx->readahead_kb == ~0ULL)
-		ctx->readahead_kb = e2fsck_guess_readahead(ctx->fs);
 	pass1_readahead(ctx, &ra_group, &ino_threshold);
-
 	if (ext2fs_has_feature_dir_index(fs->super) &&
 	    !(ctx->options & E2F_OPT_NO)) {
 		if (ext2fs_u32_list_create(&ctx->dirs_to_hash, 50))
-- 
2.37.3