[linux-lvm] ext2online and 2.4.18

[linux-lvm] ext2online and 2.4.18

[Ian Prowell]

Has anyone gotten ext2online to work with 2.4.18?  I found an ext2online
patch for 2.4.17 at the Sourceforge site for ext2online, but it does not
seem to work for 2.4.18.  I have not been able to find any discussion of
it on the net or in the linux-lvm mailing list. 

Thanks,
Ian Prowell

Re: [linux-lvm] ext2online and 2.4.18

[Andres Salomon]

I had it work w/ 2.4.18-ac3.  I believe I had to manually fix a few
.rej's from the 2.4.17 patch, however.  I don't remember exactly at this
point..

On Wed, Jun 19, 2002 at 11:56:08AM -0700, Ian Prowell wrote:
> 
> Has anyone gotten ext2online to work with 2.4.18?  I found an ext2online
> patch for 2.4.17 at the Sourceforge site for ext2online, but it does not
> seem to work for 2.4.18.  I have not been able to find any discussion of
> it on the net or in the linux-lvm mailing list. 
> 
> Thanks,
> Ian Prowell
> 
> 
> 
> 
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

-- 
Broad surveillance is a mark of bad security.
	-- Bruce Schneier

Re: [linux-lvm] ext2online and 2.4.18

[Andreas Dilger]

On Jun 19, 2002  11:56 -0700, Ian Prowell wrote:
> Has anyone gotten ext2online to work with 2.4.18?  I found an ext2online
> patch for 2.4.17 at the Sourceforge site for ext2online, but it does not
> seem to work for 2.4.18.  I have not been able to find any discussion of
> it on the net or in the linux-lvm mailing list. 

I wasn't aware that there are patch conflicts with 2.4.18, as that is
what I'm running myself.  However, I had upgraded the kernel with
ext2online kernel patches in-place instead of applying the patch to
a new 2.4.18 tree.  I'll make a new 2.4.18 patch.

Cheers, Andreas

PS - I'm just finishing work on a 2.4.18 ext3 kernel patch, in time for
     my presentation on this subject at OLS.  Expect it soon.
--
Andreas Dilger
http://www-mddsp.enel.ucalgary.ca/People/adilger/
http://sourceforge.net/projects/ext2resize/

Re: [linux-lvm] ext2online and 2.4.18

[Ian Prowell]

Andreas,
Thanks for the prompt reply.  To be more specific I am using RedHat's
2.4.18-4 kernel.  When I apply the patch I get the following output.

[root@metaverse src]# cat /root/online-ext2-2.4.17.diff | patch -p0
patching file linux/CREDITS
Hunk #1 succeeded at 701 (offset 17 lines).
patching file linux/Documentation/Configure.help
Hunk #1 succeeded at 14316 (offset 441 lines).
patching file linux/fs/Config.in
Hunk #1 succeeded at 81 (offset 4 lines).
patching file linux/fs/ext2/balloc.c
patching file linux/fs/ext2/super.c
Hunk #9 succeeded at 730 (offset -1 lines).
Hunk #10 FAILED at 883.
Hunk #11 succeeded at 893 (offset 1 line).
Hunk #12 succeeded at 917 (offset -1 lines).
Hunk #13 succeeded at 979 (offset 1 line).
Hunk #14 succeeded at 987 (offset -1 lines).
Hunk #15 succeeded at 1024 (offset 1 line).
1 out of 15 hunks FAILED -- saving rejects to file
linux/fs/ext2/super.c.rej
patching file linux/include/linux/ext2_fs.h

It looks like most of the patch is working, but I am not a kernel
developer and really have no idea what failed on hunk 10.  If it would
be useful I can send the rejects file.

Once again thanks for the help and if it is a RedHat specific thing do
not worry about it too much.

-Ian

On Wed, 2002-06-19 at 12:55, Andreas Dilger wrote:
> On Jun 19, 2002  11:56 -0700, Ian Prowell wrote:
> > Has anyone gotten ext2online to work with 2.4.18?  I found an ext2online
> > patch for 2.4.17 at the Sourceforge site for ext2online, but it does not
> > seem to work for 2.4.18.  I have not been able to find any discussion of
> > it on the net or in the linux-lvm mailing list. 
> 
> I wasn't aware that there are patch conflicts with 2.4.18, as that is
> what I'm running myself.  However, I had upgraded the kernel with
> ext2online kernel patches in-place instead of applying the patch to
> a new 2.4.18 tree.  I'll make a new 2.4.18 patch.
> 
> Cheers, Andreas
> 
> PS - I'm just finishing work on a 2.4.18 ext3 kernel patch, in time for
>      my presentation on this subject at OLS.  Expect it soon.
> --
> Andreas Dilger
> http://www-mddsp.enel.ucalgary.ca/People/adilger/
> http://sourceforge.net/projects/ext2resize/

Re: [linux-lvm] ext2online and 2.4.18

[Andreas Dilger]

On Jun 19, 2002  14:10 -0700, Ian Prowell wrote:
> Andreas,
> Thanks for the prompt reply.  To be more specific I am using RedHat's
> 2.4.18-4 kernel.  When I apply the patch I get the following output.
> 
> [root@metaverse src]# cat /root/online-ext2-2.4.17.diff | patch -p0
> patching file linux/fs/ext2/super.c
> Hunk #9 succeeded at 730 (offset -1 lines).
> Hunk #10 FAILED at 883.
> Hunk #11 succeeded at 893 (offset 1 line).
> Hunk #12 succeeded at 917 (offset -1 lines).
> Hunk #13 succeeded at 979 (offset 1 line).
> Hunk #14 succeeded at 987 (offset -1 lines).
> Hunk #15 succeeded at 1024 (offset 1 line).
> 1 out of 15 hunks FAILED -- saving rejects to file
> linux/fs/ext2/super.c.rej
> 
> It looks like most of the patch is working, but I am not a kernel
> developer and really have no idea what failed on hunk 10.  If it would
> be useful I can send the rejects file.

It won't really be useful for you to send the failed hunk, because I
have no idea why it is failing.  I would suggest firing up a text editor
with fs/ext2/super.c and fs/ext2/super.c.rej and just manually cut-paste
the failed parts in.

Unified diffs are rather simple to understand.

Lines like "@@ -123,6 +234,8 @@" are telling you that the original file
had this hunk at line 123 and the modified file had the replacement
hunk at line 234.

Lines like "-                blah" are removing that line from the
original file and replacing it with "+           blarg" in the new file.

Hopefully hunk #10 isn't too large...  Sadly, the larger the hunk, the
more likely it is to not apply cleanly...

Cheers, Andreas
--
Andreas Dilger
http://www-mddsp.enel.ucalgary.ca/People/adilger/
http://sourceforge.net/projects/ext2resize/

Re: [linux-lvm] ext2online and 2.4.18

[Ian Prowell]

[-- Attachment #1: Type: text/plain, Size: 2060 bytes --]

Attached is the reject file.  It looks like hunk 10 removes 27 lines
from super.c around line 594, but these were not getting removed in
2.4.18.  Also attached is what I think the 2.4.18 patch should be.

-Ian


On Wed, 2002-06-19 at 14:47, Andreas Dilger wrote:
> On Jun 19, 2002  14:10 -0700, Ian Prowell wrote:
> > Andreas,
> > Thanks for the prompt reply.  To be more specific I am using RedHat's
> > 2.4.18-4 kernel.  When I apply the patch I get the following output.
> > 
> > [root@metaverse src]# cat /root/online-ext2-2.4.17.diff | patch -p0
> > patching file linux/fs/ext2/super.c
> > Hunk #9 succeeded at 730 (offset -1 lines).
> > Hunk #10 FAILED at 883.
> > Hunk #11 succeeded at 893 (offset 1 line).
> > Hunk #12 succeeded at 917 (offset -1 lines).
> > Hunk #13 succeeded at 979 (offset 1 line).
> > Hunk #14 succeeded at 987 (offset -1 lines).
> > Hunk #15 succeeded at 1024 (offset 1 line).
> > 1 out of 15 hunks FAILED -- saving rejects to file
> > linux/fs/ext2/super.c.rej
> > 
> > It looks like most of the patch is working, but I am not a kernel
> > developer and really have no idea what failed on hunk 10.  If it would
> > be useful I can send the rejects file.
> 
> It won't really be useful for you to send the failed hunk, because I
> have no idea why it is failing.  I would suggest firing up a text editor
> with fs/ext2/super.c and fs/ext2/super.c.rej and just manually cut-paste
> the failed parts in.
> 
> Unified diffs are rather simple to understand.
> 
> Lines like "@@ -123,6 +234,8 @@" are telling you that the original file
> had this hunk at line 123 and the modified file had the replacement
> hunk at line 234.
> 
> Lines like "-                blah" are removing that line from the
> original file and replacing it with "+           blarg" in the new file.
> 
> Hopefully hunk #10 isn't too large...  Sadly, the larger the hunk, the
> more likely it is to not apply cleanly...
> 
> Cheers, Andreas
> --
> Andreas Dilger
> http://www-mddsp.enel.ucalgary.ca/People/adilger/
> http://sourceforge.net/projects/ext2resize/


[-- Attachment #2: super.c.rej --]
[-- Type: application/x-reject, Size: 1460 bytes --]

[-- Attachment #3: online-ext2-2.4.18.diff --]
[-- Type: text/x-patch, Size: 20345 bytes --]

diff -ru linux-2.4.18.orig/CREDITS linux/CREDITS
--- linux-2.4.18.orig/CREDITS	Mon Feb 25 11:37:50 2002
+++ linux/CREDITS	Wed Jun 19 14:56:34 2002
@@ -694,6 +694,15 @@
 S: Warrendale, Pennsylvania 15086
 S: USA
 
+N: Andreas Dilger
+E: adilger@turbolinux.com
+W: http://www-mddsp.enel.ucalgary.ca/People/adilger/
+D: Ext2 filesystem online resize capability
+D: Ext3/LVM hacking
+S: 630 Schooner Cove N.W.
+S: Calgary, AB
+S: Canada   T3L 1Z1
+
 N: Alex deVries
 E: adevries@thepuffingroup.com
 D: Various SGI parts, bits of HAL2 and Newport, PA-RISC Linux.
diff -ru linux-2.4.18.orig/Documentation/Configure.help linux/Documentation/Configure.help
--- linux-2.4.18.orig/Documentation/Configure.help	Mon Feb 25 11:37:51 2002
+++ linux/Documentation/Configure.help	Wed Jun 19 14:56:34 2002
@@ -14016,6 +14016,20 @@
   be compiled as a module, and so this could be dangerous.  Most
   everyone wants to say Y here.
 
+Online resize for ext2 filesystems
+CONFIG_EXT2_RESIZE
+  This option gives you the ability to increase the size of an ext2
+  filesystem while it is mounted (in use).  In order to do this, you
+  must also be able to resize the underlying disk partition, probably
+  via a Logical Volume Manager (LVM), metadevice (MD), or hardware
+  RAID device - none of that capability is included in this feature.
+  If you don't know what any of these things are, or you haven't
+  configured your kernel for them, you should probably say N here.  If
+  you choose Y, then your kernel will be about 3k larger, and you need
+  to get some more software (http://ext2resize.sourceforge.net/) in
+  order to actually resize your filesystem, otherwise this feature
+  will just sit unused inside the kernel.
+
 Ext3 journalling file system support (EXPERIMENTAL)
 CONFIG_EXT3_FS
   This is the journalling version of the Second extended file system
diff -ru linux-2.4.18.orig/fs/Config.in linux/fs/Config.in
--- linux-2.4.18.orig/fs/Config.in	Mon Feb 25 11:38:07 2002
+++ linux/fs/Config.in	Wed Jun 19 14:56:34 2002
@@ -77,6 +77,7 @@
 tristate 'ROM file system support' CONFIG_ROMFS_FS
 
 tristate 'Second extended fs support' CONFIG_EXT2_FS
+dep_mbool '  Online ext2 resize support (DANGEROUS)' CONFIG_EXT2_RESIZE $CONFIG_EXT2_FS $CONFIG_EXPERIMENTAL
 
 tristate 'System V/Xenix/V7/Coherent file system support' CONFIG_SYSV_FS
 
diff -ru linux-2.4.18.orig/fs/ext2/balloc.c linux/fs/ext2/balloc.c
--- linux-2.4.18.orig/fs/ext2/balloc.c	Mon Feb 25 11:38:08 2002
+++ linux/fs/ext2/balloc.c	Wed Jun 19 14:56:34 2002
@@ -616,13 +616,13 @@
 
 unsigned long ext2_count_free_blocks (struct super_block * sb)
 {
-#ifdef EXT2FS_DEBUG
 	struct ext2_super_block * es;
 	unsigned long desc_count, bitmap_count, x;
 	int bitmap_nr;
 	struct ext2_group_desc * gdp;
 	int i;
-	
+
+	if (test_opt(sb, DEBUG) && test_opt(sb, CHECK)) {
 	lock_super (sb);
 	es = sb->u.ext2_sb.s_es;
 	desc_count = 0;
@@ -643,13 +643,12 @@
 			i, le16_to_cpu(gdp->bg_free_blocks_count), x);
 		bitmap_count += x;
 	}
-	printk("ext2_count_free_blocks: stored = %lu, computed = %lu, %lu\n",
+	printk(__FUNCTION__": stored = %u, computed gdt = %lu, bitmap = %lu\n",
 	       le32_to_cpu(es->s_free_blocks_count), desc_count, bitmap_count);
 	unlock_super (sb);
 	return bitmap_count;
-#else
+	} else
 	return le32_to_cpu(sb->u.ext2_sb.s_es->s_free_blocks_count);
-#endif
 }
 
 static inline int block_in_use (unsigned long block,
diff -ru linux-2.4.18.orig/fs/ext2/super.c linux/fs/ext2/super.c
--- linux-2.4.18.orig/fs/ext2/super.c	Mon Feb 25 11:38:08 2002
+++ linux/fs/ext2/super.c	Wed Jun 19 15:16:33 2002
@@ -14,6 +14,7 @@
  *
  *  Big-endian to little-endian byte-swapping/bitmaps by
  *        David S. Miller (davem@caip.rutgers.edu), 1995
+ *  Online resize by Andreas Dilger (adilger@turbolinux.com), July 1999
  */
 
 #include <linux/config.h>
@@ -163,7 +164,9 @@
  */
 static int parse_options (char * options, unsigned long * sb_block,
 			  unsigned short *resuid, unsigned short * resgid,
-			  unsigned long * mount_options)
+			  unsigned long *mount_options,
+			  unsigned long *n_blocks_count,
+			  unsigned long *resgdt)
 {
 	char * this_char;
 	char * value;
@@ -229,6 +232,31 @@
 		else if (!strcmp (this_char, "nogrpid") ||
 			 !strcmp (this_char, "sysvgroups"))
 			clear_opt (*mount_options, GRPID);
+#ifdef CONFIG_EXT2_RESIZE
+		else if (!strcmp(this_char, "resize")) {
+			printk("EXT2-fs: parse_options: resize=%s\n", value);
+			if (!n_blocks_count) {
+				printk("EXT2-fs: resize option only available "
+				       "for remount\n");
+				return 0;
+			}
+			if (!value || !*value) {
+				printk("EXT2-fs: resize requires number of "
+				       "blocks\n");
+				return 0;
+			}
+			*n_blocks_count = simple_strtoul(value, &value, 0);
+			if (*value == ':') {
+				value++;
+				*resgdt = simple_strtoul(value, &value, 0);
+			}
+			if (*value) {
+				printk("EXT2-fs: invalid resize option: %s\n",
+				       value);
+				return 0;
+			}
+		}
+#endif /* CONFIG_EXT2_RESIZE */
 		else if (!strcmp (this_char, "resgid")) {
 			if (!value || !*value) {
 				printk ("EXT2-fs: the resgid option requires "
@@ -314,10 +342,10 @@
 	es->s_mnt_count=cpu_to_le16(le16_to_cpu(es->s_mnt_count) + 1);
 	ext2_write_super(sb);
 	if (test_opt (sb, DEBUG))
-		printk ("[EXT II FS %s, %s, bs=%lu, fs=%lu, gc=%lu, "
+		printk ("[EXT II FS %s, %s, bs=%lu, bc=%u, gc=%lu, "
 			"bpg=%lu, ipg=%lu, mo=%04lx]\n",
 			EXT2FS_VERSION, EXT2FS_DATE, sb->s_blocksize,
-			sb->u.ext2_sb.s_frag_size,
+			le32_to_cpu(es->s_blocks_count),
 			sb->u.ext2_sb.s_groups_count,
 			EXT2_BLOCKS_PER_GROUP(sb),
 			EXT2_INODES_PER_GROUP(sb),
@@ -331,7 +359,9 @@
 	return res;
 }
 
-static int ext2_check_descriptors (struct super_block * sb)
+static int ext2_check_descriptors(struct super_block *sb,
+				  struct buffer_head **group_desc,
+				  unsigned long groups_count)
 {
 	int i;
 	int desc_block = 0;
@@ -340,10 +370,10 @@
 
 	ext2_debug ("Checking group descriptors");
 
-	for (i = 0; i < sb->u.ext2_sb.s_groups_count; i++)
+	for (i = 0; i < groups_count; i++)
 	{
 		if ((i % EXT2_DESC_PER_BLOCK(sb)) == 0)
-			gdp = (struct ext2_group_desc *) sb->u.ext2_sb.s_group_desc[desc_block++]->b_data;
+			gdp = (struct ext2_group_desc *)group_desc[desc_block++]->b_data;
 		if (le32_to_cpu(gdp->bg_block_bitmap) < block ||
 		    le32_to_cpu(gdp->bg_block_bitmap) >= block + EXT2_BLOCKS_PER_GROUP(sb))
 		{
@@ -378,6 +408,267 @@
 	return 1;
 }
 
+static int ext2_read_descriptors(struct super_block *sb,
+				 unsigned long blocks_count)
+{
+	struct buffer_head **group_desc;
+	struct buffer_head **o_group_desc;
+	unsigned long groups_count;
+	unsigned long o_groups_count;
+	unsigned long gdb_count;
+	unsigned long o_gdb_count;
+
+	o_group_desc = EXT2_SB(sb)->s_group_desc;
+	o_groups_count = EXT2_SB(sb)->s_groups_count;
+	o_gdb_count = EXT2_SB(sb)->s_gdb_count;
+
+	groups_count = (blocks_count -
+			le32_to_cpu(EXT2_SB(sb)->s_es->s_first_data_block) +
+			EXT2_BLOCKS_PER_GROUP(sb) - 1) /
+			EXT2_BLOCKS_PER_GROUP(sb);
+	gdb_count = (groups_count + EXT2_DESC_PER_BLOCK(sb) - 1) /
+		   EXT2_DESC_PER_BLOCK(sb);
+	if (test_opt(sb, DEBUG)) {
+		printk("EXT2-fs: ext2_read_descriptors: o_groups_count=%lu, "
+		       "groups_count=%lu, blocks_count=%lu\n", o_groups_count,
+		       groups_count, blocks_count);
+		printk("EXT2-fs: ext2_read_descriptors: o_gdb_count=%lu, "
+		       "gdb_count=%lu\n", o_gdb_count, gdb_count);
+	}
+
+	if (o_gdb_count != gdb_count) {
+		unsigned long logic_gd_block = EXT2_SB(sb)->s_sbh->b_blocknr +1;
+		unsigned long i;
+
+		group_desc = (struct buffer_head **)kmalloc(gdb_count *
+				sizeof(struct buffer_head *), GFP_KERNEL);
+		if (group_desc == NULL) {
+			ext2_warning(sb, __FUNCTION__,
+				     "not enough memory for %ld group blocks",
+				     gdb_count);
+			return -ENOMEM;
+		}
+		for (i = o_gdb_count; i < gdb_count; i++) {	/* Add blocks */
+			group_desc[i] = bread(sb->s_dev, logic_gd_block + i,
+					      sb->s_blocksize);
+			if (!group_desc[i]) {
+				unsigned long j;
+
+				for (j = o_gdb_count; j < i; j++)
+					brelse(group_desc[j]);
+				kfree(group_desc);
+				ext2_warning(sb, __FUNCTION__,
+					     "can't read group block %ld",
+					     i);
+				return -EIO;
+			}
+		}
+
+		/*
+		 * Copy over pointers to old descriptor blocks already loaded.
+		 */
+		memcpy(group_desc, o_group_desc,
+		       (o_gdb_count < gdb_count ? o_gdb_count : gdb_count) *
+		       sizeof(struct buffer_head *));
+	} else
+		group_desc = o_group_desc; /* this will never be NULL */
+
+	if (!ext2_check_descriptors(sb, group_desc, groups_count)) {
+		unsigned long j;
+
+		for (j = o_gdb_count; j < gdb_count; j++)
+			brelse(group_desc[j]);
+		if (group_desc != o_group_desc)
+			kfree(group_desc);
+		return -EINVAL;
+	}
+
+	EXT2_SB(sb)->s_group_desc = group_desc;
+	EXT2_SB(sb)->s_groups_count = groups_count;
+	EXT2_SB(sb)->s_gdb_count = gdb_count;
+
+	if (o_group_desc && o_group_desc != EXT2_SB(sb)->s_group_desc)
+		kfree(o_group_desc);
+
+	return 0;
+} /* ext2_read_descriptors */
+
+#ifdef CONFIG_EXT2_RESIZE
+/* Make the disk blocks in a new group available to the filesystem */
+static int ext2_update_group(struct super_block *sb, unsigned int block_group,
+			     unsigned int reserved, unsigned int resgdt)
+{
+	struct ext2_group_desc *gdp;
+	struct ext2_super_block *es;
+	int blocks;
+	int m_blocks;
+	int inodes;
+	unsigned long gdb;
+	int shrink = reserved > 100 ? 1 : 0;
+
+	es = EXT2_SB(sb)->s_es;
+	gdp = ext2_get_group_desc(sb, block_group, NULL);
+	gdb = ext2_bg_num_gdb(sb, block_group);
+
+	inodes = le32_to_cpu(gdp->bg_free_inodes_count);
+	blocks = le32_to_cpu(gdp->bg_free_blocks_count);
+	m_blocks = EXT2_SB(sb)->s_itb_per_group + 2 +
+		ext2_bg_has_super(sb, block_group) +
+		(gdb ? gdb + resgdt : 0) + blocks;
+
+	if (block_group < EXT2_SB(sb)->s_groups_count - 1 &&
+	    m_blocks != EXT2_BLOCKS_PER_GROUP(sb)) {
+		ext2_warning(sb, __FUNCTION__,
+			     "bad group size for %s group %d (%d blocks) "
+			     "(iblk = %ld, GDT blk %ld, resgdt %d, blocks %d\n",
+			     ext2_bg_has_super(sb, block_group) ? "normal" :
+			     "no-backup", block_group, m_blocks,
+			     EXT2_SB(sb)->s_itb_per_group, gdb, resgdt, blocks);
+		return -EINVAL;
+	}
+
+	es->s_free_inodes_count =
+		cpu_to_le32(le32_to_cpu(es->s_free_inodes_count) + inodes);
+	es->s_inodes_count = cpu_to_le32(le32_to_cpu(es->s_inodes_count) +
+		inodes);
+
+	es->s_free_blocks_count =
+		cpu_to_le32(le32_to_cpu(es->s_free_blocks_count) + blocks);
+	es->s_blocks_count = cpu_to_le32(le32_to_cpu(es->s_blocks_count) +
+		m_blocks);
+	es->s_r_blocks_count = cpu_to_le32(le32_to_cpu(es->s_r_blocks_count) +
+		m_blocks * reserved / 100);
+
+	mark_buffer_dirty(EXT2_SB(sb)->s_sbh);
+	sb->s_dirt = 1;
+
+	if (test_opt(sb, DEBUG)) {
+		printk("EXT2-fs: ext2_update_group: %s %s group %u: "
+		       "%u inodes, %u blocks (%u free), %d (%d%%) reserved\n",
+		       shrink ? "removing" : "adding",
+		       ext2_bg_has_super(sb, block_group) ? "normal" :
+		       "no-super", block_group, inodes, m_blocks, blocks,
+		       m_blocks * reserved / 100, reserved);
+		printk("EXT2-fs: ext2_update_group: %u inodes (%u free), "
+		       "%u blocks (%u free)\n",
+		       le32_to_cpu(es->s_inodes_count),
+		       le32_to_cpu(es->s_free_inodes_count),
+		       le32_to_cpu(es->s_blocks_count),
+		       le32_to_cpu(es->s_free_blocks_count));
+	}
+
+	return 0;
+} /* ext2_update_group */
+
+/* Resize the filesystem to the new number of blocks specified.  If required,
+ * the new group descriptors should have already been configured for us (we
+ * only check that they are valid).
+ */
+static int ext2_resize_fs(struct super_block *sb,
+			  struct ext2_super_block *es,
+			  unsigned long n_blocks_count, unsigned long resgdt)
+{
+	unsigned long o_blocks_count;
+	unsigned long o_groups_count;
+	unsigned long last, add;
+	unsigned long reserved;
+	struct buffer_head *bh;
+	unsigned long i;
+	int err = 0;
+
+	o_blocks_count = le32_to_cpu(es->s_blocks_count);
+	o_groups_count = EXT2_SB(sb)->s_groups_count;
+
+	if (test_opt(sb, DEBUG))
+		printk("EXT2-fs: ext2_resize_fs: from %lu to %lu blocks\n",
+		       o_blocks_count, n_blocks_count);
+
+	if (n_blocks_count == 0 || n_blocks_count == o_blocks_count)
+		return 0;
+
+	if (n_blocks_count < o_blocks_count) {
+		ext2_warning(sb, __FUNCTION__, "error: can't shrink FS!");
+		return -EBUSY;
+	}
+
+	/* See if the device is actually as big as what was requested */
+	bh = bread(sb->s_dev, n_blocks_count - 1, EXT2_BLOCK_SIZE(sb));
+	if (!bh) {
+		ext2_warning(sb, __FUNCTION__,
+			     "unable to read last block, resize aborted");
+		return -ENOSPC;
+	}
+	brelse(bh);
+
+	/* For reserved percentage calculation, we avoid 32-bit overflow. */
+	reserved = o_blocks_count > 10000000 ?
+		(le32_to_cpu(es->s_r_blocks_count) + o_blocks_count / 200) /
+			(o_blocks_count / 100) :
+		(le32_to_cpu(es->s_r_blocks_count) * 100 + o_blocks_count / 2) /
+			o_blocks_count;
+
+	if ((err = ext2_read_descriptors(sb, n_blocks_count))) {
+		ext2_warning(sb, __FUNCTION__,
+			     "group descriptor error %d, resize aborted", err);
+		return err;
+	}
+
+	/* Handle the remaining blocks in the last partial group. */
+	last = (o_blocks_count - le32_to_cpu(es->s_first_data_block)) %
+		EXT2_BLOCKS_PER_GROUP(sb);
+	if (last != 0) {	/* The last group isn't full yet */
+		struct inode inode;
+		add = EXT2_BLOCKS_PER_GROUP(sb) - last;
+		if (add + o_blocks_count > n_blocks_count)
+			add = n_blocks_count - o_blocks_count;
+		es->s_blocks_count =
+			cpu_to_le32(le32_to_cpu(es->s_blocks_count) + add);
+		es->s_r_blocks_count =
+			cpu_to_le32(le32_to_cpu(es->s_r_blocks_count) +
+				    add * reserved / 100);
+		mark_buffer_dirty(EXT2_SB(sb)->s_sbh);
+		sb->s_dirt = 1;
+		unlock_super(sb);
+		/*
+		 * Fake out an inode enough to "free" the new blocks in this
+		 * group.  Turn off quotas for this inode so it doesn't get
+		 * confused by freeing blocks that don't really exist yet.
+		 */
+		inode.i_sb = sb;
+		inode.i_state = I_DIRTY;
+		INIT_LIST_HEAD(&inode.i_hash);
+		for (i = 0; i < MAXQUOTAS; i++)
+			inode.i_dquot[i] = NODQUOT;
+		ext2_free_blocks(&inode, o_blocks_count, add);
+		if (test_opt(sb, DEBUG))
+			printk("EXT2-fs: ext2_resize_fs: added %lu new blocks "
+			       "to %lu blocks in last group\n",
+			       add, o_blocks_count);
+		lock_super(sb);
+	}
+
+	/*
+	 * Update superblock with remaining new group block/inode counts
+	 */
+	for (i = o_groups_count; i < EXT2_SB(sb)->s_groups_count &&
+	     !(err = ext2_update_group(sb, i, reserved, resgdt)); i++)
+		/* empty loop */;
+
+	if (err || le32_to_cpu(es->s_blocks_count) != n_blocks_count) {
+		ext2_warning(sb, __FUNCTION__,
+			     "specified size does not match new block count "
+			     "(%lu != %u) (err %d)", n_blocks_count,
+			     le32_to_cpu(es->s_blocks_count), err);
+		EXT2_SB(sb)->s_mount_state &= ~EXT2_VALID_FS;
+	} else
+		EXT2_SB(sb)->s_mount_state |= EXT2_VALID_FS;
+
+	return err;
+} /* ext2_resize_fs */
+#else  /* !CONFIG_EXT2_RESIZE */
+#define ext2_resize_fs(sb, es, n_blocks_count, resgdt) 0
+#endif /* CONFIG_EXT2_RESIZE */
+
 #define log2(n) ffz(~(n))
  
 /*
@@ -409,8 +700,7 @@
 	unsigned long offset = 0;
 	kdev_t dev = sb->s_dev;
 	int blocksize = BLOCK_SIZE;
-	int db_count;
-	int i, j;
+	int i;
 
 	/*
 	 * See what the current blocksize for the device is, and
@@ -425,7 +715,7 @@
 
 	sb->u.ext2_sb.s_mount_opt = 0;
 	if (!parse_options ((char *) data, &sb_block, &resuid, &resgid,
-	    &sb->u.ext2_sb.s_mount_opt)) {
+	    &sb->u.ext2_sb.s_mount_opt, NULL, NULL)) {
 		return NULL;
 	}
 
@@ -596,32 +886,6 @@
 		goto failed_mount;
 	}
 
-	sb->u.ext2_sb.s_groups_count = (le32_to_cpu(es->s_blocks_count) -
-				        le32_to_cpu(es->s_first_data_block) +
-				       EXT2_BLOCKS_PER_GROUP(sb) - 1) /
-				       EXT2_BLOCKS_PER_GROUP(sb);
-	db_count = (sb->u.ext2_sb.s_groups_count + EXT2_DESC_PER_BLOCK(sb) - 1) /
-		   EXT2_DESC_PER_BLOCK(sb);
-	sb->u.ext2_sb.s_group_desc = kmalloc (db_count * sizeof (struct buffer_head *), GFP_KERNEL);
-	if (sb->u.ext2_sb.s_group_desc == NULL) {
-		printk ("EXT2-fs: not enough memory\n");
-		goto failed_mount;
-	}
-	for (i = 0; i < db_count; i++) {
-		sb->u.ext2_sb.s_group_desc[i] = sb_bread(sb, logic_sb_block + i + 1);
-		if (!sb->u.ext2_sb.s_group_desc[i]) {
-			for (j = 0; j < i; j++)
-				brelse (sb->u.ext2_sb.s_group_desc[j]);
-			kfree(sb->u.ext2_sb.s_group_desc);
-			printk ("EXT2-fs: unable to read group descriptors\n");
-			goto failed_mount;
-		}
-	}
-	if (!ext2_check_descriptors (sb)) {
-		printk ("EXT2-fs: group descriptors corrupted!\n");
-		db_count = i;
-		goto failed_mount2;
-	}
 	for (i = 0; i < EXT2_MAX_GROUP_LOADED; i++) {
 		sb->u.ext2_sb.s_inode_bitmap_number[i] = 0;
 		sb->u.ext2_sb.s_inode_bitmap[i] = NULL;
@@ -630,11 +894,18 @@
 	}
 	sb->u.ext2_sb.s_loaded_inode_bitmaps = 0;
 	sb->u.ext2_sb.s_loaded_block_bitmaps = 0;
-	sb->u.ext2_sb.s_gdb_count = db_count;
 	/*
 	 * set up enough so that it can read an inode
 	 */
 	sb->s_op = &ext2_sops;
+	sb->u.ext2_sb.s_group_desc = NULL;
+	sb->u.ext2_sb.s_groups_count = 0;
+	sb->u.ext2_sb.s_gdb_count = 0;
+	if ((i = ext2_read_descriptors(sb, le32_to_cpu(es->s_blocks_count)))) {
+		ext2_error(sb, __FUNCTION__,
+			   "group descriptor error %d, unable to mount", i);
+		goto failed_mount;
+	}
 	sb->s_root = d_alloc_root(iget(sb, EXT2_ROOT_INO));
 	if (!sb->s_root || !S_ISDIR(sb->s_root->d_inode->i_mode) ||
 	    !sb->s_root->d_inode->i_blocks || !sb->s_root->d_inode->i_size) {
@@ -649,7 +920,7 @@
 	ext2_setup_super (sb, es, sb->s_flags & MS_RDONLY);
 	return sb;
 failed_mount2:
-	for (i = 0; i < db_count; i++)
+	for (i = 0; i < sb->u.ext2_sb.s_gdb_count; i++)
 		brelse(sb->u.ext2_sb.s_group_desc[i]);
 	kfree(sb->u.ext2_sb.s_group_desc);
 failed_mount:
@@ -709,6 +980,8 @@
 	struct ext2_super_block * es;
 	unsigned short resuid = sb->u.ext2_sb.s_resuid;
 	unsigned short resgid = sb->u.ext2_sb.s_resgid;
+	unsigned long n_blocks_count = 0;
+	unsigned long resgdt = 0;
 	unsigned long new_mount_opt;
 	unsigned long tmp;
 
@@ -717,14 +990,16 @@
 	 */
 	new_mount_opt = sb->u.ext2_sb.s_mount_opt;
 	if (!parse_options (data, &tmp, &resuid, &resgid,
-			    &new_mount_opt))
+			    &new_mount_opt, &n_blocks_count, &resgdt))
 		return -EINVAL;
 
 	sb->u.ext2_sb.s_mount_opt = new_mount_opt;
 	sb->u.ext2_sb.s_resuid = resuid;
 	sb->u.ext2_sb.s_resgid = resgid;
 	es = sb->u.ext2_sb.s_es;
-	if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
+	if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY) &&
+	    (n_blocks_count == 0 ||
+	     n_blocks_count == le32_to_cpu(es->s_blocks_count)))
 		return 0;
 	if (*flags & MS_RDONLY) {
 		if (le16_to_cpu(es->s_state) & EXT2_VALID_FS ||
@@ -751,6 +1026,8 @@
 		 * by e2fsck since we originally mounted the partition.)
 		 */
 		sb->u.ext2_sb.s_mount_state = le16_to_cpu(es->s_state);
+		if ((ret = ext2_resize_fs(sb, es, n_blocks_count, resgdt)))
+			return ret;
 		if (!ext2_setup_super (sb, es, 0))
 			sb->s_flags &= ~MS_RDONLY;
 	}
diff -ru linux-2.4.18.orig/include/linux/ext2_fs.h linux/include/linux/ext2_fs.h
--- linux-2.4.18.orig/include/linux/ext2_fs.h	Thu Nov 22 11:46:52 2001
+++ linux/include/linux/ext2_fs.h	Wed Jun 19 14:56:34 2002
@@ -26,6 +26,7 @@
  * Define EXT2FS_DEBUG to produce debug messages
  */
 #undef EXT2FS_DEBUG
+#define CONFIG_EXT2_CHECK
 
 /*
  * Define EXT2_PREALLOCATE to preallocate data blocks for expanding files
@@ -61,6 +62,7 @@
 #define EXT2_ACL_DATA_INO	 4	/* ACL inode */
 #define EXT2_BOOT_LOADER_INO	 5	/* Boot loader inode */
 #define EXT2_UNDEL_DIR_INO	 6	/* Undelete directory inode */
+#define EXT2_RESIZE_INO		 7	/* Reserved group descriptors inode */
 
 /* First non-reserved inode for old ext2 filesystems */
 #define EXT2_GOOD_OLD_FIRST_INO	11

Re: [linux-lvm] ext2online and 2.4.18

[sander]

Hi Ian,

I applied your patch to 2.4.18 kernel. It applied cleanly. Compiled the new
kernel and no problems doing that but... ext2online still fails while in the
kernel.

Hope this helps,
Sander de Boer

Citeren Ian Prowell <ip_lvm@prowell.org>:

> Attached is the reject file.  It looks like hunk 10 removes 27 lines
> from super.c around line 594, but these were not getting removed in
> 2.4.18.  Also attached is what I think the 2.4.18 patch should be.
> 
> -Ian
> 
> 






__________________________________________________________

Snel en makkelijk mailen

Wanadoo webmail || http://webmail.wanadoo.nl

Re: [linux-lvm] ext2online and 2.4.18

[Andreas Dilger]

On Jun 20, 2002  13:28 +0200, sander@sanderscorner.com wrote:
> I applied your patch to 2.4.18 kernel. It applied cleanly. Compiled the new
> kernel and no problems doing that but... ext2online still fails while in the
> kernel.

Could you be more specific?  Do you get any messages in syslog or from
dmesg?  Does the command just lock up and never return?  If you mount
the filesystem with "-o debug,check" you get a few extra messages that
may help narrow down the problem.

Cheers, Andreas
--
Andreas Dilger
http://www-mddsp.enel.ucalgary.ca/People/adilger/
http://sourceforge.net/projects/ext2resize/

Re[2]: [linux-lvm] ext2online and 2.4.18

[Sander]

(Response to message from Andreas Dilger, adilger@clusterfs.com on Thursday, June 20, 2002 6:34:20 PM)

Hallo Andreas,

One message in dmesg which I don't see in a kernel without the patch:
"EXT2-fs warning: maximal mount count reached, running e2fsck is
recommended", strange since I have fs checking automatically.
I have seen other messages on a few previous boots with a patched
kernel, that I unfortunately can not reproduce, where there are complaints about an
option 'resize' that is not understood.

No messages of this kind ended up in the messages file, or the debug
log. They only show when booting and in dmesg.

Guess this won't help much, but that's all I have for the moment.

-- 

Regards,
Sander

Thursday, June 20, 2002, 6:34:20 PM, you wrote:

> Could you be more specific?  Do you get any messages in syslog or from
> dmesg?  Does the command just lock up and never return?  If you mount
> the filesystem with "-o debug,check" you get a few extra messages that
> may help narrow down the problem.

> Cheers, Andreas
> --
> Andreas Dilger
> http://www-mddsp.enel.ucalgary.ca/People/adilger/
> http://sourceforge.net/projects/ext2resize/


> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

Re: [linux-lvm] ext2online and 2.4.18

[Jan-Benedict Glaw]

[-- Attachment #1: Type: text/plain, Size: 755 bytes --]

On Fri, 2002-06-21 23:49:18 +0200, Sander <boeraa@wanadoo.nl>
wrote in message <482222938.20020621234918@wanadoo.nl>:
> (Response to message from Andreas Dilger, adilger@clusterfs.com on Thursday, June 20, 2002 6:34:20 PM)
> One message in dmesg which I don't see in a kernel without the patch:
> "EXT2-fs warning: maximal mount count reached, running e2fsck is
> recommended", strange since I have fs checking automatically.

Are you sure that this/those filesystem(s) have a non-zero value in
/etc/fstab's "fs_passno" field (it's the 6th and last value in each
row)?

MfG, JBG

-- 
Jan-Benedict Glaw   .   jbglaw@lug-owl.de   .   +49-172-7608481
	 -- New APT-Proxy written in shell script --
	   http://lug-owl.de/~jbglaw/software/ap2/

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [linux-lvm] ext2online and 2.4.18

[Sander]

(Response to message from Andreas Dilger, adilger@clusterfs.com on Thursday, June 20, 2002 6:34:20 PM)

Hallo Andreas,

One message in dmesg which I don't see in a kernel without the patch:
"EXT2-fs warning: maximal mount count reached, running e2fsck is
recommended", strange since I have fs checking automatically.
I have seen other messages on a few previous boots with a patched
kernel, that I unfortunately can not reproduce, where there are complaints about an
option 'resize' that is not understood.

No messages of this kind ended up in the messages file, or the debug
log. They only show when booting and in dmesg.

Guess this won't help much, but that's all I have for the moment.

-- 

Regards,
Sander

Thursday, June 20, 2002, 6:34:20 PM, you wrote:

> Could you be more specific?  Do you get any messages in syslog or from
> dmesg?  Does the command just lock up and never return?  If you mount
> the filesystem with "-o debug,check" you get a few extra messages that
> may help narrow down the problem.

> Cheers, Andreas
> --
> Andreas Dilger
> http://www-mddsp.enel.ucalgary.ca/People/adilger/
> http://sourceforge.net/projects/ext2resize/


> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

Re: [linux-lvm] ext2online and 2.4.18

[Andreas Dilger]

On Jun 21, 2002  23:49 +0200, Sander wrote:
> One message in dmesg which I don't see in a kernel without the patch:
> "EXT2-fs warning: maximal mount count reached, running e2fsck is
> recommended", strange since I have fs checking automatically.

This may happen if you are doing lots of testing.  The interface to the
online resize is done via "mount -o remount,resize=<new_size> /device"
so each time you do a resize it counts as an extra mount.  At one time
I thought I should fix this, so that you don't get filesystem checks
more often, but online resizing is a bit dangerous in the end, so if
you do lots of resizing it probably isn't a bad idea to do extra
checks on the filesystem.

> I have seen other messages on a few previous boots with a patched
> kernel, that I unfortunately can not reproduce, where there are
> complaints about an option 'resize' that is not understood.

This would indicate that the filesystem you are trying to resize does
not have the kernel patch applied.

Cheers, Andreas
--
Andreas Dilger
http://www-mddsp.enel.ucalgary.ca/People/adilger/
http://sourceforge.net/projects/ext2resize/

Re[2]: [linux-lvm] ext2online and 2.4.18

[Sander]

(Response to message from Jan-Benedict Glaw, jbglaw@lug-owl.de on Friday, June 21, 2002 11:59:03 PM)

Hallo Jan-Benedict,

Yes, definately non-zero. This message never shows in an un-patched kernel and
fsck'ing happens automatically. On a previous occasion I checked the mount counts too.
They were perfectly ok.

-- 

Regards,
Sander

Friday, June 21, 2002, 11:59:03 PM, you wrote:

> Are you sure that this/those filesystem(s) have a non-zero value in
> /etc/fstab's "fs_passno" field (it's the 6th and last value in each
> row)?

> MfG, JBG

Re[2]: [linux-lvm] ext2online and 2.4.18

[sander]

(Response to message from Andreas Dilger, adilger@clusterfs.com on Saturday, June 22, 2002 10:11:38 AM)

Hello Andreas,

Got it working! I had previously just put CONFIG_EXT2_RESIZE=y in the
.config manually and now let the configure menu do the work.

I guess that's what happens if you take a shortcut.

-- 

Regards,
sander

Saturday, June 22, 2002, 10:11:38 AM, you wrote:

> This would indicate that the filesystem you are trying to resize does
> not have the kernel patch applied.

> Cheers, Andreas
> --
> Andreas Dilger
> http://www-mddsp.enel.ucalgary.ca/People/adilger/
> http://sourceforge.net/projects/ext2resize/


> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html