[Devel] [PATCH RHEL7 COMMIT] ext4: replace ext4_kvmalloc() with kvmalloc()

Konstantin Khorenko khorenko at virtuozzo.com
Thu Jul 26 18:14:51 MSK 2018


The commit is pushed to "branch-rh7-3.10.0-862.9.1.vz7.63.x-ovz" and will appear at https://src.openvz.org/scm/ovz/vzkernel.git
after rh7-3.10.0-862.9.1.vz7.63.3
------>
commit d695abe411ce7e1e89736278aea76dc576e73612
Author: Oleg Babin <obabin at virtuozzo.com>
Date:   Thu Jul 26 18:14:51 2018 +0300

    ext4: replace ext4_kvmalloc() with kvmalloc()
    
    ext4_kvmalloc() is used to allocate the table of group descritors
    blocks. It is called in GFP_NOFS context which is not vmalloc()
    compatible so it cannot be directly replaced with kvmalloc().
    
    In order to use kvmalloc() with GFP_KERNEL flag the memory allocation
    is moved from add_new_gdb() / add_new_gdb_meta_bg() functions to the
    beginning of the resize process before any journaling is started and
    any FS locks are taken.
    
    After this we do not need ext4_kvmalloc/ext4_kvfree functions any more.
    
    https://jira.sw.ru/browse/PSBM-83044
    
    Signed-off-by: Oleg Babin <obabin at virtuozzo.com>
    Signed-off-by: Jan Dakinevich <jan.dakinevich at virtuozzo.com>
---
 fs/ext4/ext4.h   |  4 ++--
 fs/ext4/resize.c | 44 ++++++++++----------------------------------
 fs/ext4/super.c  | 29 +++++++++++++++++++++++++++++
 3 files changed, 41 insertions(+), 36 deletions(-)

diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index ef07fa0fe6a4..22e463c0b7e9 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -2298,10 +2298,10 @@ extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
 /* super.c */
 extern int ext4_calculate_overhead(struct super_block *sb);
 extern void ext4_superblock_csum_set(struct super_block *sb);
-extern void *ext4_kvmalloc(size_t size, gfp_t flags);
-extern void *ext4_kvzalloc(size_t size, gfp_t flags);
 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
 				    ext4_group_t ngroup);
+extern int ext4_alloc_group_desc_bh_array(struct super_block *sb,
+					  ext4_group_t ngroup);
 extern const char *ext4_decode_error(struct super_block *sb, int errno,
 				     char nbuf[16]);
 
diff --git a/fs/ext4/resize.c b/fs/ext4/resize.c
index 76d0e0bd1911..99682f90d7b3 100644
--- a/fs/ext4/resize.c
+++ b/fs/ext4/resize.c
@@ -757,7 +757,6 @@ static int add_new_gdb(handle_t *handle, struct inode *inode,
 	struct ext4_super_block *es = EXT4_SB(sb)->s_es;
 	unsigned long gdb_num = group / EXT4_DESC_PER_BLOCK(sb);
 	ext4_fsblk_t gdblock = EXT4_SB(sb)->s_sbh->b_blocknr + 1 + gdb_num;
-	struct buffer_head **o_group_desc, **n_group_desc;
 	struct buffer_head *dind;
 	struct buffer_head *gdb_bh;
 	int gdbackups;
@@ -815,16 +814,6 @@ static int add_new_gdb(handle_t *handle, struct inode *inode,
 	if (unlikely(err))
 		goto exit_dind;
 
-	n_group_desc = ext4_kvmalloc((gdb_num + 1) *
-				     sizeof(struct buffer_head *),
-				     GFP_NOFS);
-	if (!n_group_desc) {
-		err = -ENOMEM;
-		ext4_warning(sb, "not enough memory for %lu groups",
-			     gdb_num + 1);
-		goto exit_inode;
-	}
-
 	/*
 	 * Finally, we have all of the possible failures behind us...
 	 *
@@ -850,13 +839,8 @@ static int add_new_gdb(handle_t *handle, struct inode *inode,
 	}
 	brelse(dind);
 
-	o_group_desc = EXT4_SB(sb)->s_group_desc;
-	memcpy(n_group_desc, o_group_desc,
-	       EXT4_SB(sb)->s_gdb_count * sizeof(struct buffer_head *));
-	n_group_desc[gdb_num] = gdb_bh;
-	EXT4_SB(sb)->s_group_desc = n_group_desc;
+	EXT4_SB(sb)->s_group_desc[gdb_num] = gdb_bh;
 	EXT4_SB(sb)->s_gdb_count++;
-	kvfree(o_group_desc);
 
 	le16_add_cpu(&es->s_reserved_gdt_blocks, -1);
 	err = ext4_handle_dirty_super(handle, sb);
@@ -866,7 +850,6 @@ static int add_new_gdb(handle_t *handle, struct inode *inode,
 	return err;
 
 exit_inode:
-	kvfree(n_group_desc);
 	brelse(iloc.bh);
 exit_dind:
 	brelse(dind);
@@ -884,7 +867,6 @@ static int add_new_gdb_meta_bg(struct super_block *sb,
 			       handle_t *handle, ext4_group_t group) {
 	ext4_fsblk_t gdblock;
 	struct buffer_head *gdb_bh;
-	struct buffer_head **o_group_desc, **n_group_desc;
 	unsigned long gdb_num = group / EXT4_DESC_PER_BLOCK(sb);
 	int err;
 
@@ -893,23 +875,9 @@ static int add_new_gdb_meta_bg(struct super_block *sb,
 	gdb_bh = sb_bread(sb, gdblock);
 	if (!gdb_bh)
 		return -EIO;
-	n_group_desc = ext4_kvmalloc((gdb_num + 1) *
-				     sizeof(struct buffer_head *),
-				     GFP_NOFS);
-	if (!n_group_desc) {
-		err = -ENOMEM;
-		ext4_warning(sb, "not enough memory for %lu groups",
-			     gdb_num + 1);
-		return err;
-	}
 
-	o_group_desc = EXT4_SB(sb)->s_group_desc;
-	memcpy(n_group_desc, o_group_desc,
-	       EXT4_SB(sb)->s_gdb_count * sizeof(struct buffer_head *));
-	n_group_desc[gdb_num] = gdb_bh;
-	EXT4_SB(sb)->s_group_desc = n_group_desc;
+	EXT4_SB(sb)->s_group_desc[gdb_num] = gdb_bh;
 	EXT4_SB(sb)->s_gdb_count++;
-	kvfree(o_group_desc);
 	BUFFER_TRACE(gdb_bh, "get_write_access");
 	err = ext4_journal_get_write_access(handle, gdb_bh);
 	if (unlikely(err))
@@ -1627,6 +1595,10 @@ int ext4_group_add(struct super_block *sb, struct ext4_new_group_data *input)
 	if (err)
 		goto out;
 
+	err = ext4_alloc_group_desc_bh_array(sb, input->group + 1);
+	if (err)
+		goto out;
+
 	err = ext4_mb_alloc_groupinfo(sb, input->group + 1);
 	if (err)
 		goto out;
@@ -1977,6 +1949,10 @@ int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count)
 	if (err)
 		return err;
 
+	err = ext4_alloc_group_desc_bh_array(sb, n_group + 1);
+	if (err)
+		goto out;
+
 	err = ext4_mb_alloc_groupinfo(sb, n_group + 1);
 	if (err)
 		goto out;
diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index 5d4e01d161df..8c90717d4a26 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -2190,6 +2190,35 @@ int ext4_alloc_flex_bg_array(struct super_block *sb, ext4_group_t ngroup)
 	return 0;
 }
 
+/*
+ * Allocate the top-level s_group_desc array for the specified number
+ * of groups. As the memory is allocated before the journaling is started
+ * we can safely use kvmalloc() with GFP_KERNEL flag here.
+ */
+int ext4_alloc_group_desc_bh_array(struct super_block *sb, ext4_group_t ngroup)
+{
+	struct ext4_sb_info *sbi = EXT4_SB(sb);
+	unsigned long num_desc = DIV_ROUND_UP(ngroup,  EXT4_DESC_PER_BLOCK(sb));
+	struct buffer_head **n_group_desc;
+
+	if (num_desc <= sbi->s_gdb_count)
+		return 0;
+
+	n_group_desc = kvmalloc(num_desc * sizeof(struct buffer_head *),
+				GFP_KERNEL);
+	if (!n_group_desc) {
+		ext4_warning(sb, "not enough memory for %lu groups", num_desc);
+		return -ENOMEM;
+	}
+
+	memcpy(n_group_desc, sbi->s_group_desc,
+	       num_desc * sizeof(struct buffer_head *));
+	kvfree(sbi->s_group_desc);
+
+	sbi->s_group_desc = n_group_desc;
+	return 0;
+}
+
 static int ext4_fill_flex_info(struct super_block *sb)
 {
 	struct ext4_sb_info *sbi = EXT4_SB(sb);


More information about the Devel mailing list