So I am working on trying to get mkfs doing all the initialization. So in the
process I am commenting the kernel module code so that I will know what is
happenning.
So here are some comments I have added.
John
Index: super.c
==================================================================--- super.c
(revision 860)
+++ super.c (working copy)
@@ -961,6 +961,8 @@
osb->publ_map |= (1 << osb->node_num);
osb->vol_state = VOLUME_INIT;
+ // Check if first time this partition has been mounted. If so create
+ // the root Oracle INode.
status = ocfs_create_root_oin (osb, NULL);
if (status >= 0)
osb->vol_state = VOLUME_ENABLED;
Index: oin.c
==================================================================--- oin.c
(revision 860)
+++ oin.c (working copy)
@@ -446,6 +446,11 @@
/* ocfs_create_root_oin()
*
+ * This function will create the root Oracle INode (OIN) on the file system if
+ * this is the first time that this partition has ever been mounted.
+ *
+ * If this is NOT the first time this partition has been mounted (meaning it
+ * has been mounted before) then this function does nothing.
*/
int ocfs_create_root_oin (ocfs_super * osb, struct inode *root)
{
@@ -459,7 +464,15 @@
LOG_ENTRY ();
+ // Is this the first time this partition has been mounted?
if (osb->vol_layout.root_start_off == 0) {
+ // This is our first time being mounted. We need to take care
+ // of setting up the root Oracle inode. In order to do this we
+ // are going to call ocfs_create_root_dir_node. So we need to
+ // get the disk offset to the root directory (root_dirnode).
+ // In order to get that we need to read in the volume header
+ // from the first sector of the disk.
+
status = ocfs_wait_for_disk_lock_release (osb, OCFS_VOLUME_LOCK_OFFSET,
10000, OCFS_DLM_NO_LOCK);
if (status < 0) {
@@ -471,12 +484,14 @@
OCFS_DLM_EXCLUSIVE_LOCK, FLAG_FILE_CREATE,
&lockres, &bh, NULL);
if (status >= 0) {
+ // Read the volume header from the first sector of the
+ // partition and put it into our buffer head
status = ocfs_read_bh(osb, 0, &hdr_bh, 0, NULL);
if (status < 0)
LOG_ERROR_STATUS (status);
tmpstat = ocfs_release_lock (osb, OCFS_VOLUME_LOCK_OFFSET,
OCFS_DLM_EXCLUSIVE_LOCK, 0, lockres, bh, NULL);
-
+
if (bh)
brelse(bh);
@@ -492,7 +507,13 @@
LOG_ERROR_STATUS (status);
goto finally;
}
-
+
+ // Get the volume header data from our buffer head. We need to
+ // know the following in order to call
+ // ocfs_create_root_dir_node():
+ // * root_off: disk offset to the ocfs root dirnode
+ // (root_directory)
+ // * internal_off: disk offset tot he data segment
volDiskHdr = (ocfs_vol_disk_hdr *) OCFS_BH_GET_DATA_READ(hdr_bh); /* read */
root_off = volDiskHdr->root_off;
int_off = volDiskHdr->internal_off;
@@ -502,10 +523,13 @@
ocfs_wait_for_disk_lock_release (osb,
OCFS_VOLUME_LOCK_OFFSET,
30000, OCFS_DLM_NO_LOCK);
+ // Setup the root_start_off with the correct value so
+ // that it will not run this function the next time the
+ // filesystem is mounted.
osb->vol_layout.root_start_off = root_off;
osb->vol_layout.root_int_off = int_off;
}
-
+ // Do all the work in creating our root directory
status = ocfs_create_root_dir_node (osb);
if (status < 0) {
LOG_ERROR_STATUS (status);
-------------- next part --------------
Index: super.c
==================================================================--- super.c
(revision 860)
+++ super.c (working copy)
@@ -961,6 +961,8 @@
osb->publ_map |= (1 << osb->node_num);
osb->vol_state = VOLUME_INIT;
+ // Check if first time this partition has been mounted. If so create
+ // the root Oracle INode.
status = ocfs_create_root_oin (osb, NULL);
if (status >= 0)
osb->vol_state = VOLUME_ENABLED;
Index: oin.c
==================================================================--- oin.c
(revision 860)
+++ oin.c (working copy)
@@ -446,6 +446,11 @@
/* ocfs_create_root_oin()
*
+ * This function will create the root Oracle INode (OIN) on the file system if
+ * this is the first time that this partition has ever been mounted.
+ *
+ * If this is NOT the first time this partition has been mounted (meaning it
+ * has been mounted before) then this function does nothing.
*/
int ocfs_create_root_oin (ocfs_super * osb, struct inode *root)
{
@@ -459,7 +464,15 @@
LOG_ENTRY ();
+ // Is this the first time this partition has been mounted?
if (osb->vol_layout.root_start_off == 0) {
+ // This is our first time being mounted. We need to take care
+ // of setting up the root Oracle inode. In order to do this we
+ // are going to call ocfs_create_root_dir_node. So we need to
+ // get the disk offset to the root directory (root_dirnode).
+ // In order to get that we need to read in the volume header
+ // from the first sector of the disk.
+
status = ocfs_wait_for_disk_lock_release (osb, OCFS_VOLUME_LOCK_OFFSET,
10000, OCFS_DLM_NO_LOCK);
if (status < 0) {
@@ -471,12 +484,14 @@
OCFS_DLM_EXCLUSIVE_LOCK, FLAG_FILE_CREATE,
&lockres, &bh, NULL);
if (status >= 0) {
+ // Read the volume header from the first sector of the
+ // partition and put it into our buffer head
status = ocfs_read_bh(osb, 0, &hdr_bh, 0, NULL);
if (status < 0)
LOG_ERROR_STATUS (status);
tmpstat = ocfs_release_lock (osb, OCFS_VOLUME_LOCK_OFFSET,
OCFS_DLM_EXCLUSIVE_LOCK, 0, lockres, bh, NULL);
-
+
if (bh)
brelse(bh);
@@ -492,7 +507,13 @@
LOG_ERROR_STATUS (status);
goto finally;
}
-
+
+ // Get the volume header data from our buffer head. We need to
+ // know the following in order to call
+ // ocfs_create_root_dir_node():
+ // * root_off: disk offset to the ocfs root dirnode
+ // (root_directory)
+ // * internal_off: disk offset tot he data segment
volDiskHdr = (ocfs_vol_disk_hdr *) OCFS_BH_GET_DATA_READ(hdr_bh); /* read */
root_off = volDiskHdr->root_off;
int_off = volDiskHdr->internal_off;
@@ -502,10 +523,13 @@
ocfs_wait_for_disk_lock_release (osb,
OCFS_VOLUME_LOCK_OFFSET,
30000, OCFS_DLM_NO_LOCK);
+ // Setup the root_start_off with the correct value so
+ // that it will not run this function the next time the
+ // filesystem is mounted.
osb->vol_layout.root_start_off = root_off;
osb->vol_layout.root_int_off = int_off;
}
-
+ // Do all the work in creating our root directory
status = ocfs_create_root_dir_node (osb);
if (status < 0) {
LOG_ERROR_STATUS (status);
John, Eeeeek! Sorry I should have said something! I pretty much finished mkfs over the last few days. I couldn't take it anymore so I just went ahead and did it over the weekend ;-) How far along are you? We should sync up. -kurt On Wed, Apr 21, 2004 at 04:48:36PM -0700, John L. Villalovos wrote:> So I am working on trying to get mkfs doing all the initialization. So in > the process I am commenting the kernel module code so that I will know what > is happenning. > > So here are some comments I have added. > > John > > Index: super.c > ==================================================================> --- super.c (revision 860) > +++ super.c (working copy) > @@ -961,6 +961,8 @@ > > osb->publ_map |= (1 << osb->node_num); > osb->vol_state = VOLUME_INIT; > + // Check if first time this partition has been mounted. If so create > + // the root Oracle INode. > status = ocfs_create_root_oin (osb, NULL); > if (status >= 0) > osb->vol_state = VOLUME_ENABLED; > Index: oin.c > ==================================================================> --- oin.c (revision 860) > +++ oin.c (working copy) > @@ -446,6 +446,11 @@ > > /* ocfs_create_root_oin() > * > + * This function will create the root Oracle INode (OIN) on the file > system if > + * this is the first time that this partition has ever been mounted. > + * > + * If this is NOT the first time this partition has been mounted (meaning > it > + * has been mounted before) then this function does nothing. > */ > int ocfs_create_root_oin (ocfs_super * osb, struct inode *root) > { > @@ -459,7 +464,15 @@ > > LOG_ENTRY (); > > + // Is this the first time this partition has been mounted? > if (osb->vol_layout.root_start_off == 0) { > + // This is our first time being mounted. We need to take > care > + // of setting up the root Oracle inode. In order to do this > we > + // are going to call ocfs_create_root_dir_node. So we need > to > + // get the disk offset to the root directory (root_dirnode). > + // In order to get that we need to read in the volume header > + // from the first sector of the disk. > + > status = ocfs_wait_for_disk_lock_release (osb, > OCFS_VOLUME_LOCK_OFFSET, > 10000, OCFS_DLM_NO_LOCK); > if (status < 0) { > @@ -471,12 +484,14 @@ > OCFS_DLM_EXCLUSIVE_LOCK, > FLAG_FILE_CREATE, > &lockres, &bh, NULL); > if (status >= 0) { > + // Read the volume header from the first sector of > the > + // partition and put it into our buffer head > status = ocfs_read_bh(osb, 0, &hdr_bh, 0, NULL); > if (status < 0) > LOG_ERROR_STATUS (status); > > tmpstat = ocfs_release_lock (osb, > OCFS_VOLUME_LOCK_OFFSET, OCFS_DLM_EXCLUSIVE_LOCK, 0, > lockres, bh, NULL); > - > + > if (bh) > brelse(bh); > > @@ -492,7 +507,13 @@ > LOG_ERROR_STATUS (status); > goto finally; > } > - > + > + // Get the volume header data from our buffer head. We need > to > + // know the following in order to call > + // ocfs_create_root_dir_node(): > + // * root_off: disk offset to the ocfs root dirnode > + // (root_directory) > + // * internal_off: disk offset tot he data segment > volDiskHdr = (ocfs_vol_disk_hdr *) > OCFS_BH_GET_DATA_READ(hdr_bh); /* read */ > root_off = volDiskHdr->root_off; > int_off = volDiskHdr->internal_off; > @@ -502,10 +523,13 @@ > ocfs_wait_for_disk_lock_release (osb, > OCFS_VOLUME_LOCK_OFFSET, > 30000, > OCFS_DLM_NO_LOCK); > + // Setup the root_start_off with the correct value so > + // that it will not run this function the next time > the > + // filesystem is mounted. > osb->vol_layout.root_start_off = root_off; > osb->vol_layout.root_int_off = int_off; > } > - > + // Do all the work in creating our root directory > status = ocfs_create_root_dir_node (osb); > if (status < 0) { > LOG_ERROR_STATUS (status); >> Index: super.c > ==================================================================> --- super.c (revision 860) > +++ super.c (working copy) > @@ -961,6 +961,8 @@ > > osb->publ_map |= (1 << osb->node_num); > osb->vol_state = VOLUME_INIT; > + // Check if first time this partition has been mounted. If so create > + // the root Oracle INode. > status = ocfs_create_root_oin (osb, NULL); > if (status >= 0) > osb->vol_state = VOLUME_ENABLED; > Index: oin.c > ==================================================================> --- oin.c (revision 860) > +++ oin.c (working copy) > @@ -446,6 +446,11 @@ > > /* ocfs_create_root_oin() > * > + * This function will create the root Oracle INode (OIN) on the file system if > + * this is the first time that this partition has ever been mounted. > + * > + * If this is NOT the first time this partition has been mounted (meaning it > + * has been mounted before) then this function does nothing. > */ > int ocfs_create_root_oin (ocfs_super * osb, struct inode *root) > { > @@ -459,7 +464,15 @@ > > LOG_ENTRY (); > > + // Is this the first time this partition has been mounted? > if (osb->vol_layout.root_start_off == 0) { > + // This is our first time being mounted. We need to take care > + // of setting up the root Oracle inode. In order to do this we > + // are going to call ocfs_create_root_dir_node. So we need to > + // get the disk offset to the root directory (root_dirnode). > + // In order to get that we need to read in the volume header > + // from the first sector of the disk. > + > status = ocfs_wait_for_disk_lock_release (osb, OCFS_VOLUME_LOCK_OFFSET, > 10000, OCFS_DLM_NO_LOCK); > if (status < 0) { > @@ -471,12 +484,14 @@ > OCFS_DLM_EXCLUSIVE_LOCK, FLAG_FILE_CREATE, > &lockres, &bh, NULL); > if (status >= 0) { > + // Read the volume header from the first sector of the > + // partition and put it into our buffer head > status = ocfs_read_bh(osb, 0, &hdr_bh, 0, NULL); > if (status < 0) > LOG_ERROR_STATUS (status); > > tmpstat = ocfs_release_lock (osb, OCFS_VOLUME_LOCK_OFFSET, OCFS_DLM_EXCLUSIVE_LOCK, 0, lockres, bh, NULL); > - > + > if (bh) > brelse(bh); > > @@ -492,7 +507,13 @@ > LOG_ERROR_STATUS (status); > goto finally; > } > - > + > + // Get the volume header data from our buffer head. We need to > + // know the following in order to call > + // ocfs_create_root_dir_node(): > + // * root_off: disk offset to the ocfs root dirnode > + // (root_directory) > + // * internal_off: disk offset tot he data segment > volDiskHdr = (ocfs_vol_disk_hdr *) OCFS_BH_GET_DATA_READ(hdr_bh); /* read */ > root_off = volDiskHdr->root_off; > int_off = volDiskHdr->internal_off; > @@ -502,10 +523,13 @@ > ocfs_wait_for_disk_lock_release (osb, > OCFS_VOLUME_LOCK_OFFSET, > 30000, OCFS_DLM_NO_LOCK); > + // Setup the root_start_off with the correct value so > + // that it will not run this function the next time the > + // filesystem is mounted. > osb->vol_layout.root_start_off = root_off; > osb->vol_layout.root_int_off = int_off; > } > - > + // Do all the work in creating our root directory > status = ocfs_create_root_dir_node (osb); > if (status < 0) { > LOG_ERROR_STATUS (status);> _______________________________________________ > Ocfs2-devel mailing list > Ocfs2-devel@oss.oracle.com > http://oss.oracle.com/mailman/listinfo/ocfs2-devel
I thought Kurt was already done with mkfs. But comments can't hurt... unless it is C++ style. :-) Please could you use C style comments... /* */. Thanks Sunil John L. Villalovos wrote:> So I am working on trying to get mkfs doing all the initialization. > So in the process I am commenting the kernel module code so that I > will know what is happenning. > > So here are some comments I have added. > > John > > Index: super.c > ==================================================================> --- super.c (revision 860) > +++ super.c (working copy) > @@ -961,6 +961,8 @@ > > osb->publ_map |= (1 << osb->node_num); > osb->vol_state = VOLUME_INIT; > + // Check if first time this partition has been mounted. If so > create > + // the root Oracle INode. > status = ocfs_create_root_oin (osb, NULL); > if (status >= 0) > osb->vol_state = VOLUME_ENABLED; > Index: oin.c > ==================================================================> --- oin.c (revision 860) > +++ oin.c (working copy) > @@ -446,6 +446,11 @@ > > /* ocfs_create_root_oin() > * > + * This function will create the root Oracle INode (OIN) on the file > system if > + * this is the first time that this partition has ever been mounted. > + * > + * If this is NOT the first time this partition has been mounted > (meaning it > + * has been mounted before) then this function does nothing. > */ > int ocfs_create_root_oin (ocfs_super * osb, struct inode *root) > { > @@ -459,7 +464,15 @@ > > LOG_ENTRY (); > > + // Is this the first time this partition has been mounted? > if (osb->vol_layout.root_start_off == 0) { > + // This is our first time being mounted. We need to take care > + // of setting up the root Oracle inode. In order to do this we > + // are going to call ocfs_create_root_dir_node. So we need to > + // get the disk offset to the root directory (root_dirnode). > + // In order to get that we need to read in the volume header > + // from the first sector of the disk. > + > status = ocfs_wait_for_disk_lock_release (osb, > OCFS_VOLUME_LOCK_OFFSET, > 10000, OCFS_DLM_NO_LOCK); > if (status < 0) { > @@ -471,12 +484,14 @@ > OCFS_DLM_EXCLUSIVE_LOCK, FLAG_FILE_CREATE, > &lockres, &bh, NULL); > if (status >= 0) { > + // Read the volume header from the first sector of the > + // partition and put it into our buffer head > status = ocfs_read_bh(osb, 0, &hdr_bh, 0, NULL); > if (status < 0) > LOG_ERROR_STATUS (status); > > tmpstat = ocfs_release_lock (osb, > OCFS_VOLUME_LOCK_OFFSET, OCFS_DLM_EXCLUSIVE_LOCK, 0, lockres, bh, NULL); > - > + > if (bh) > brelse(bh); > > @@ -492,7 +507,13 @@ > LOG_ERROR_STATUS (status); > goto finally; > } > - > + > + // Get the volume header data from our buffer head. We need to > + // know the following in order to call > + // ocfs_create_root_dir_node(): > + // * root_off: disk offset to the ocfs root dirnode > + // (root_directory) > + // * internal_off: disk offset tot he data segment > volDiskHdr = (ocfs_vol_disk_hdr *) > OCFS_BH_GET_DATA_READ(hdr_bh); /* read */ > root_off = volDiskHdr->root_off; > int_off = volDiskHdr->internal_off; > @@ -502,10 +523,13 @@ > ocfs_wait_for_disk_lock_release (osb, > OCFS_VOLUME_LOCK_OFFSET, > 30000, OCFS_DLM_NO_LOCK); > + // Setup the root_start_off with the correct value so > + // that it will not run this function the next time the > + // filesystem is mounted. > osb->vol_layout.root_start_off = root_off; > osb->vol_layout.root_int_off = int_off; > } > - > + // Do all the work in creating our root directory > status = ocfs_create_root_dir_node (osb); > if (status < 0) { > LOG_ERROR_STATUS (status); > >------------------------------------------------------------------------ > >Index: super.c >==================================================================>--- super.c (revision 860) >+++ super.c (working copy) >@@ -961,6 +961,8 @@ > > osb->publ_map |= (1 << osb->node_num); > osb->vol_state = VOLUME_INIT; >+ // Check if first time this partition has been mounted. If so create >+ // the root Oracle INode. > status = ocfs_create_root_oin (osb, NULL); > if (status >= 0) > osb->vol_state = VOLUME_ENABLED; >Index: oin.c >==================================================================>--- oin.c (revision 860) >+++ oin.c (working copy) >@@ -446,6 +446,11 @@ > > /* ocfs_create_root_oin() > * >+ * This function will create the root Oracle INode (OIN) on the file system if >+ * this is the first time that this partition has ever been mounted. >+ * >+ * If this is NOT the first time this partition has been mounted (meaning it >+ * has been mounted before) then this function does nothing. > */ > int ocfs_create_root_oin (ocfs_super * osb, struct inode *root) > { >@@ -459,7 +464,15 @@ > > LOG_ENTRY (); > >+ // Is this the first time this partition has been mounted? > if (osb->vol_layout.root_start_off == 0) { >+ // This is our first time being mounted. We need to take care >+ // of setting up the root Oracle inode. In order to do this we >+ // are going to call ocfs_create_root_dir_node. So we need to >+ // get the disk offset to the root directory (root_dirnode). >+ // In order to get that we need to read in the volume header >+ // from the first sector of the disk. >+ > status = ocfs_wait_for_disk_lock_release (osb, OCFS_VOLUME_LOCK_OFFSET, > 10000, OCFS_DLM_NO_LOCK); > if (status < 0) { >@@ -471,12 +484,14 @@ > OCFS_DLM_EXCLUSIVE_LOCK, FLAG_FILE_CREATE, > &lockres, &bh, NULL); > if (status >= 0) { >+ // Read the volume header from the first sector of the >+ // partition and put it into our buffer head > status = ocfs_read_bh(osb, 0, &hdr_bh, 0, NULL); > if (status < 0) > LOG_ERROR_STATUS (status); > > tmpstat = ocfs_release_lock (osb, OCFS_VOLUME_LOCK_OFFSET, OCFS_DLM_EXCLUSIVE_LOCK, 0, lockres, bh, NULL); >- >+ > if (bh) > brelse(bh); > >@@ -492,7 +507,13 @@ > LOG_ERROR_STATUS (status); > goto finally; > } >- >+ >+ // Get the volume header data from our buffer head. We need to >+ // know the following in order to call >+ // ocfs_create_root_dir_node(): >+ // * root_off: disk offset to the ocfs root dirnode >+ // (root_directory) >+ // * internal_off: disk offset tot he data segment > volDiskHdr = (ocfs_vol_disk_hdr *) OCFS_BH_GET_DATA_READ(hdr_bh); /* read */ > root_off = volDiskHdr->root_off; > int_off = volDiskHdr->internal_off; >@@ -502,10 +523,13 @@ > ocfs_wait_for_disk_lock_release (osb, > OCFS_VOLUME_LOCK_OFFSET, > 30000, OCFS_DLM_NO_LOCK); >+ // Setup the root_start_off with the correct value so >+ // that it will not run this function the next time the >+ // filesystem is mounted. > osb->vol_layout.root_start_off = root_off; > osb->vol_layout.root_int_off = int_off; > } >- >+ // Do all the work in creating our root directory > status = ocfs_create_root_dir_node (osb); > if (status < 0) { > LOG_ERROR_STATUS (status); > > >------------------------------------------------------------------------ > >_______________________________________________ >Ocfs2-devel mailing list >Ocfs2-devel@oss.oracle.com >http://oss.oracle.com/mailman/listinfo/ocfs2-devel > >
Sunil Mushran wrote:> I thought Kurt was already done with mkfs. > > But comments can't hurt... unless it is C++ style. :-) > Please could you use C style comments... /* */.Just curious why you don't want C++ style comments. They are part of the C standard since 1999 and all the compilers that can compile the kernel support them. I noticed that other code in the kernel uses them too. I'm just curious since I am a fan of those style comments. But I'll be happy to redo it in the /* */ style. John -- A: Because it messes up the order in which people normally read text. Q: Why is top-posting such a bad thing? A: Top-posting Q: What is the most annoying thing on usenet and in e-mail?
khackel@ca2.us.oracle.com wrote:> John, > > Eeeeek! Sorry I should have said something! I pretty much finished > mkfs over the last few days. I couldn't take it anymore so I > just went > ahead and did it over the weekend ;-) > > How far along are you? We should sync up.Not that far along :( Or maybe that should be :) I was starting to try to figure out the code. I went through reading the mkfs code and then was going through the OCFS2 code. I hadn't done any coding yet so I don't have anything to contribute. We have moved from one campus to another campus two weeks ago. So I haven't done a lot on OCFS the last two weeks. I have been making sure that our infrastructure came up successfully. On a positive note it did and on another positive note I am now 1/2 mile from work so my commute is much shorter :) John -- A: Because it messes up the order in which people normally read text. Q: Why is top-posting such a bad thing? A: Top-posting Q: What is the most annoying thing on usenet and in e-mail?