Skip to contentSkip to conversation

2.4.23-pre8-pac1 and -rc1-pac1 NFSv3 problem

DISCUSSION Greg Louis 6 participants 10 messages Nov 11, 2003 – Dec 4, 2003 NFS client

lore
Share & tools
lore

Greg Louis

author
#0 lore raw

Kernels 2.4.23-pre7-pac1 and 2.4.23-rc1 are ok but -pre8-pac1 and -rc1-pac1 behave as follows: mounting a remote directory via NFS with v3 enabled (client and server) seems to work ok, and running mount with no parameters shows the NFS mount, but any attempt at access fails with a message like
/bin/ls: reading directory /whatever/it/was: Input/output error

Signature · 3 lines
| G r e g  L o u i s         | gpg public key: 0x400B1AA86D9E3E64 |
|  http://www.bgl.nu/~glouis |   (on my website or any keyserver) |
|  http://wecanstopspam.org in signatures helps fight junk email. |
author
#1 lore raw
Quoted 6 lines · Greg Louis

Kernels 2.4.23-pre7-pac1 and 2.4.23-rc1 are ok but -pre8-pac1 and -rc1-pac1 behave as follows: mounting a remote directory via NFS with v3 enabled (client and server) seems to work ok, and running mount with no parameters shows the NFS mount, but any attempt at access fails with a message like
/bin/ls: reading directory /whatever/it/was: Input/output error

Reverting all changes to fs/nfs/* since 2.4.23-pre7-pac1, and only those, corrects the problem.

Signature · 3 lines
| G r e g  L o u i s         | gpg public key: 0x400B1AA86D9E3E64 |
|  http://www.bgl.nu/~glouis |   (on my website or any keyserver) |
|  http://wecanstopspam.org in signatures helps fight junk email. |
#2 lore raw
Quoted 11 lines · Greg Louis
Quoted 6 lines · Greg Louis

Kernels 2.4.23-pre7-pac1 and 2.4.23-rc1 are ok but -pre8-pac1 and -rc1-pac1 behave as follows: mounting a remote directory via NFS with v3 enabled (client and server) seems to work ok, and running mount with no parameters shows the NFS mount, but any attempt at access fails with a message like
/bin/ls: reading directory /whatever/it/was: Input/output error

Reverting all changes to fs/nfs/* since 2.4.23-pre7-pac1, and only those, corrects the problem.

/metoo

annwn:~> bpsh 0 mount
none on /proc type proc (rw)
none on /dev/pts type devpts (rw,mode=0620)
none on /dev/shm type tmpfs (rw)
none on /tmp type tmpfs (rw)
192.168.0.1:/lib on /lib type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/bin on /bin type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/sbin on /sbin type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/usr on /usr type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/opt on /opt type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/home on /home type nfs (rw,nfsvers=3,noac,addr=192.168.0.1) 192.168.0.1:/work on /work/shared type nfs (rw,nfsvers=3,noac,addr=192.168.0.1)

annwn:~> bpsh 0 pwd
/home/magallon

annwn:~> bpsh 0 ls
ls: reading directory .: Invalid argument

It works for some time, and then it breaks.
Could you send me the patch you used to revert those changes ? I will try to make a diff from rc1 to pre7 and reverse.

TIA

Signature · 4 lines
J.A. Magallon <jamagallon()able!es>     \                 Software is like sex:
werewolf!able!es                         \           It's better when it's free
Mandrake Linux release 10.0 (Cooker) for i586
Linux 2.4.23-rc1-jam1 (gcc 3.3.1 (Mandrake Linux 9.2 3.3.1-4mdk))
#3 lore raw
Quoted 32 lines · J.A. Magallon
Quoted 11 lines · Greg Louis
Quoted 6 lines · Greg Louis

Kernels 2.4.23-pre7-pac1 and 2.4.23-rc1 are ok but -pre8-pac1 and -rc1-pac1 behave as follows: mounting a remote directory via NFS with v3 enabled (client and server) seems to work ok, and running mount with no parameters shows the NFS mount, but any attempt at access fails with a message like
/bin/ls: reading directory /whatever/it/was: Input/output error

Reverting all changes to fs/nfs/* since 2.4.23-pre7-pac1, and only those, corrects the problem.

/metoo

annwn:~> bpsh 0 mount
none on /proc type proc (rw)
none on /dev/pts type devpts (rw,mode=0620)
none on /dev/shm type tmpfs (rw)
none on /tmp type tmpfs (rw)
192.168.0.1:/lib on /lib type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/bin on /bin type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/sbin on /sbin type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/usr on /usr type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/opt on /opt type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/home on /home type nfs (rw,nfsvers=3,noac,addr=192.168.0.1) 192.168.0.1:/work on /work/shared type nfs (rw,nfsvers=3,noac,addr=192.168.0.1)

annwn:~> bpsh 0 pwd
/home/magallon

Quoted 6 lines

annwn:~> bpsh 0 ls
ls: reading directory .: Invalid argument

It works for some time, and then it breaks.
Could you send me the patch you used to revert those changes ? I will try to make a diff from rc1 to pre7 and reverse.

Just this (pre7 -> pre8|rc1):

fs/nfs/nfs3proc.c +1 -7
diff -ruN linux-2.4.23-pre7/fs/nfs/nfs3proc.c linux-2.4.23-rc1/fs/nfs/nfs3proc.c
@@ -433,8 +433,6 @@
  * The decode function itself doesn't perform any decoding, it just makes
  * sure the reply is syntactically correct.
  *
- * Also note that this implementation handles both plain readdir and
- * readdirplus.
  */
 static int
 nfs3_proc_readdir(struct inode *dir, struct rpc_cred *cred,
@@ -448,11 +446,7 @@
 	struct rpc_message	msg = { NFS3PROC_READDIR, &arg, &res, cred };
 	int			status;
 
-	if (plus)
-		msg.rpc_proc = NFS3PROC_READDIRPLUS;
-
-	dprintk("NFS call  readdir%s %d\n",
-			plus? "plus" : "", (unsigned int) cookie);
+	dprintk("NFS call  readdir %d\n", (unsigned int) cookie);
 
 	dir_attr.valid = 0;
 	status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0);
fs/nfs/nfs3xdr.c +4 -14
diff -ruN linux-2.4.23-pre7/fs/nfs/nfs3xdr.c linux-2.4.23-rc1/fs/nfs/nfs3xdr.c
@@ -599,8 +599,6 @@
 u32 *
 nfs3_decode_dirent(u32 *p, struct nfs_entry *entry, int plus)
 {
-	struct nfs_entry old = *entry;
-
 	if (!*p++) {
 		if (!*p)
 			return ERR_PTR(-EAGAIN);
@@ -616,20 +614,12 @@
 	p = xdr_decode_hyper(p, &entry->cookie);
 
 	if (plus) {
-		p = xdr_decode_post_op_attr(p, &entry->fattr);
+		struct nfs_fattr fattr;
+		p = xdr_decode_post_op_attr(p, &fattr);
 		/* In fact, a post_op_fh3: */
 		if (*p++) {
-			p = xdr_decode_fhandle(p, &entry->fh);
-			/* Ugh -- server reply was truncated */
-			if (p == NULL) {
-				dprintk("NFS: FH truncated\n");
-				*entry = old;
-				return ERR_PTR(-EAGAIN);
-			}
-		} else {
-			/* If we don't get a file handle, the attrs
-			 * aren't worth a lot. */
-			entry->fattr.valid = 0;
+			struct nfs_fh fh;
+			p = xdr_decode_fhandle(p, &fh);
 		}
 	}
 
fs/nfs/write.c +16 -2
diff -ruN linux-2.4.23-pre7/fs/nfs/write.c linux-2.4.23-rc1/fs/nfs/write.c
@@ -225,8 +225,19 @@
 	struct inode *inode = page->mapping->host;
 	unsigned long end_index;
 	unsigned offset = PAGE_CACHE_SIZE;
+	int inode_referenced = 0;
 	int err;
 
+	/*
+	 * Note: We need to ensure that we have a reference to the inode
+	 *       if we are to do asynchronous writes. If not, waiting
+	 *       in nfs_wait_on_request() may deadlock with clear_inode().
+	 *
+	 *       If igrab() fails here, then it is in any case safe to
+	 *       call nfs_wb_page(), since there will be no pending writes.
+	 */
+	if (igrab(inode) != 0)
+		inode_referenced = 1;
 	end_index = inode->i_size >> PAGE_CACHE_SHIFT;
 
 	/* Ensure we've flushed out any previous writes */
@@ -244,7 +255,8 @@
 		goto out;
 do_it:
 	lock_kernel();
-	if (NFS_SERVER(inode)->wsize >= PAGE_CACHE_SIZE && !IS_SYNC(inode)) {
+	if (NFS_SERVER(inode)->wsize >= PAGE_CACHE_SIZE && !IS_SYNC(inode) &&
+			inode_referenced) {
 		err = nfs_writepage_async(NULL, inode, page, 0, offset);
 		if (err >= 0)
 			err = 0;
@@ -256,7 +268,9 @@
 	unlock_kernel();
 out:
 	UnlockPage(page);
-	return err; 
+	if (inode_referenced)
+		iput(inode);
+	return err;
 }
 
 /*
Signature · 4 lines
J.A. Magallon <jamagallon()able!es>     \                 Software is like sex:
werewolf!able!es                         \           It's better when it's free
Mandrake Linux release 10.0 (Cooker) for i586
Linux 2.4.23-rc1-jam1 (gcc 3.3.1 (Mandrake Linux 9.2 3.3.1-4mdk))
#4 lore raw

PowerBook shutdown

Hi,

I've got a new 12" PowerBook G4 that is happily running Ben's 2.4.23-pre5 kernel. One odd thing happens, though. When the lid is closed, after a couple of minutes the machine turns off. It doesn't /sbin/shutdown, it just powers off. Has anyone else had this happen?

Cheers,
dstn.

author
#5 lore raw
In reply to J.A. Magallon
Quoted 10 lines · J.A. Magallon
Quoted 8 lines · Greg Louis
Quoted 6 lines · Greg Louis

Kernels 2.4.23-pre7-pac1 and 2.4.23-rc1 are ok but -pre8-pac1 and -rc1-pac1 behave as follows: mounting a remote directory via NFS with v3 enabled (client and server) seems to work ok, and running mount with no parameters shows the NFS mount, but any attempt at access fails with a message like
/bin/ls: reading directory /whatever/it/was: Input/output error

Quoted 8 lines
Quoted 2 lines

Reverting all changes to fs/nfs/* since 2.4.23-pre7-pac1, and only those, corrects the problem.

/metoo

It works for some time, and then it breaks.
Could you send me the patch you used to revert those changes ? I will try to make a diff from rc1 to pre7 and reverse.

Actually all I did was
rm fs/nfs/*.c
cp ../linux-2.4.23pre7/fs/nfs/*.c fs/nfs

Could well be overkill but it seems to work fine on a farm of mixed 2.4.22 and 2.4.23rc1-with-this-change machines.

Signature · 3 lines
| G r e g  L o u i s         | gpg public key: 0x400B1AA86D9E3E64 |
|  http://www.bgl.nu/~glouis |   (on my website or any keyserver) |
|  http://wecanstopspam.org in signatures helps fight junk email. |
#6 lore raw

2.4.23-pac1

$SUBJECT is at ftp.kernel.org/pub/linux/kernel/people/bero/2.4/2.4.23/

Changes since 2.4.23-rc5-pac1:
- Increase version number

Signature · 5 lines
Ark Linux - Linux for the masses
http://www.arklinux.org/

Redistribution and processing of this message is subject to
http://www.arklinux.org/terms.php
author
#7 lore raw

2.4.23-pre8-pac1 and since: NFSv3 problem

This is still present in kernel 2.4.23-pac1.

Quoted 155 lines · J.A. Magallon
Quoted 32 lines · J.A. Magallon
Quoted 11 lines · Greg Louis
Quoted 6 lines · Greg Louis

Kernels 2.4.23-pre7-pac1 and 2.4.23-rc1 are ok but -pre8-pac1 and -rc1-pac1 behave as follows: mounting a remote directory via NFS with v3 enabled (client and server) seems to work ok, and running mount with no parameters shows the NFS mount, but any attempt at access fails with a message like
/bin/ls: reading directory /whatever/it/was: Input/output error

Reverting all changes to fs/nfs/* since 2.4.23-pre7-pac1, and only those, corrects the problem.

/metoo

annwn:~> bpsh 0 mount
none on /proc type proc (rw)
none on /dev/pts type devpts (rw,mode=0620)
none on /dev/shm type tmpfs (rw)
none on /tmp type tmpfs (rw)
192.168.0.1:/lib on /lib type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/bin on /bin type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/sbin on /sbin type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/usr on /usr type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/opt on /opt type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/home on /home type nfs (rw,nfsvers=3,noac,addr=192.168.0.1) 192.168.0.1:/work on /work/shared type nfs (rw,nfsvers=3,noac,addr=192.168.0.1)

annwn:~> bpsh 0 pwd
/home/magallon

Quoted 6 lines

annwn:~> bpsh 0 ls
ls: reading directory .: Invalid argument

It works for some time, and then it breaks.
Could you send me the patch you used to revert those changes ? I will try to make a diff from rc1 to pre7 and reverse.

Just this (pre7 -> pre8|rc1):

diff -ruN linux-2.4.23-pre7/fs/nfs/nfs3proc.c linux-2.4.23-rc1/fs/nfs/nfs3proc.c

fs/nfs/nfs3proc.c +1 -7
@@ -433,8 +433,6 @@
  * The decode function itself doesn't perform any decoding, it just makes
  * sure the reply is syntactically correct.
  *
- * Also note that this implementation handles both plain readdir and
- * readdirplus.
  */
 static int
 nfs3_proc_readdir(struct inode *dir, struct rpc_cred *cred,
@@ -448,11 +446,7 @@
 	struct rpc_message	msg = { NFS3PROC_READDIR, &arg, &res, cred };
 	int			status;
 
-	if (plus)
-		msg.rpc_proc = NFS3PROC_READDIRPLUS;
-
-	dprintk("NFS call  readdir%s %d\n",
-			plus? "plus" : "", (unsigned int) cookie);
+	dprintk("NFS call  readdir %d\n", (unsigned int) cookie);
 
 	dir_attr.valid = 0;
 	status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0);

diff -ruN linux-2.4.23-pre7/fs/nfs/nfs3xdr.c linux-2.4.23-rc1/fs/nfs/nfs3xdr.c

fs/nfs/nfs3xdr.c +4 -14
@@ -599,8 +599,6 @@
 u32 *
 nfs3_decode_dirent(u32 *p, struct nfs_entry *entry, int plus)
 {
-	struct nfs_entry old = *entry;
-
 	if (!*p++) {
 		if (!*p)
 			return ERR_PTR(-EAGAIN);
@@ -616,20 +614,12 @@
 	p = xdr_decode_hyper(p, &entry->cookie);
 
 	if (plus) {
-		p = xdr_decode_post_op_attr(p, &entry->fattr);
+		struct nfs_fattr fattr;
+		p = xdr_decode_post_op_attr(p, &fattr);
 		/* In fact, a post_op_fh3: */
 		if (*p++) {
-			p = xdr_decode_fhandle(p, &entry->fh);
-			/* Ugh -- server reply was truncated */
-			if (p == NULL) {
-				dprintk("NFS: FH truncated\n");
-				*entry = old;
-				return ERR_PTR(-EAGAIN);
-			}
-		} else {
-			/* If we don't get a file handle, the attrs
-			 * aren't worth a lot. */
-			entry->fattr.valid = 0;
+			struct nfs_fh fh;
+			p = xdr_decode_fhandle(p, &fh);
 		}
 	}
 

diff -ruN linux-2.4.23-pre7/fs/nfs/write.c linux-2.4.23-rc1/fs/nfs/write.c

fs/nfs/write.c +16 -2
@@ -225,8 +225,19 @@
 	struct inode *inode = page->mapping->host;
 	unsigned long end_index;
 	unsigned offset = PAGE_CACHE_SIZE;
+	int inode_referenced = 0;
 	int err;
 
+	/*
+	 * Note: We need to ensure that we have a reference to the inode
+	 *       if we are to do asynchronous writes. If not, waiting
+	 *       in nfs_wait_on_request() may deadlock with clear_inode().
+	 *
+	 *       If igrab() fails here, then it is in any case safe to
+	 *       call nfs_wb_page(), since there will be no pending writes.
+	 */
+	if (igrab(inode) != 0)
+		inode_referenced = 1;
 	end_index = inode->i_size >> PAGE_CACHE_SHIFT;
 
 	/* Ensure we've flushed out any previous writes */
@@ -244,7 +255,8 @@
 		goto out;
 do_it:
 	lock_kernel();
-	if (NFS_SERVER(inode)->wsize >= PAGE_CACHE_SIZE && !IS_SYNC(inode)) {
+	if (NFS_SERVER(inode)->wsize >= PAGE_CACHE_SIZE && !IS_SYNC(inode) &&
+			inode_referenced) {
 		err = nfs_writepage_async(NULL, inode, page, 0, offset);
 		if (err >= 0)
 			err = 0;
@@ -256,7 +268,9 @@
 	unlock_kernel();
 out:
 	UnlockPage(page);
-	return err; 
+	if (inode_referenced)
+		iput(inode);
+	return err;
 }
 
 /*
Signature · 5 lines
-- 
J.A. Magallon <jamagallon()able!es>     \                 Software is like sex:
werewolf!able!es                         \           It's better when it's free
Mandrake Linux release 10.0 (Cooker) for i586
Linux 2.4.23-rc1-jam1 (gcc 3.3.1 (Mandrake Linux 9.2 3.3.1-4mdk))
Signature · 3 lines
| G r e g  L o u i s         | gpg public key: 0x400B1AA86D9E3E64 |
|  http://www.bgl.nu/~glouis |   (on my website or any keyserver) |
|  http://wecanstopspam.org in signatures helps fight junk email. |
#8 lore raw

Re: 2.4.23-pre8-pac1 and since: NFSv3 problem

Could someone please send a patch to fix this against 2.4.23-pac1?

Thanks in advance.

Quoted 169 lines · Greg Louis

This is still present in kernel 2.4.23-pac1.

Quoted 155 lines · J.A. Magallon
Quoted 32 lines · J.A. Magallon
Quoted 11 lines · Greg Louis
Quoted 6 lines · Greg Louis

Kernels 2.4.23-pre7-pac1 and 2.4.23-rc1 are ok but -pre8-pac1 and -rc1-pac1 behave as follows: mounting a remote directory via NFS with v3 enabled (client and server) seems to work ok, and running mount with no parameters shows the NFS mount, but any attempt at access fails with a message like
/bin/ls: reading directory /whatever/it/was: Input/output error

Reverting all changes to fs/nfs/* since 2.4.23-pre7-pac1, and only those, corrects the problem.

/metoo

annwn:~> bpsh 0 mount
none on /proc type proc (rw)
none on /dev/pts type devpts (rw,mode=0620)
none on /dev/shm type tmpfs (rw)
none on /tmp type tmpfs (rw)
192.168.0.1:/lib on /lib type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/bin on /bin type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/sbin on /sbin type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/usr on /usr type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/opt on /opt type nfs (ro,noatime,nfsvers=3,nolock,addr=192.168.0.1) 192.168.0.1:/home on /home type nfs (rw,nfsvers=3,noac,addr=192.168.0.1) 192.168.0.1:/work on /work/shared type nfs (rw,nfsvers=3,noac,addr=192.168.0.1)

annwn:~> bpsh 0 pwd
/home/magallon

Quoted 6 lines

annwn:~> bpsh 0 ls
ls: reading directory .: Invalid argument

It works for some time, and then it breaks.
Could you send me the patch you used to revert those changes ? I will try to make a diff from rc1 to pre7 and reverse.

Just this (pre7 -> pre8|rc1):

diff -ruN linux-2.4.23-pre7/fs/nfs/nfs3proc.c linux-2.4.23-rc1/fs/nfs/nfs3proc.c

fs/nfs/nfs3proc.c +1 -7
@@ -433,8 +433,6 @@
  * The decode function itself doesn't perform any decoding, it just makes
  * sure the reply is syntactically correct.
  *
- * Also note that this implementation handles both plain readdir and
- * readdirplus.
  */
 static int
 nfs3_proc_readdir(struct inode *dir, struct rpc_cred *cred,
@@ -448,11 +446,7 @@
 	struct rpc_message	msg = { NFS3PROC_READDIR, &arg, &res, cred };
 	int			status;
 
-	if (plus)
-		msg.rpc_proc = NFS3PROC_READDIRPLUS;
-
-	dprintk("NFS call  readdir%s %d\n",
-			plus? "plus" : "", (unsigned int) cookie);
+	dprintk("NFS call  readdir %d\n", (unsigned int) cookie);
 
 	dir_attr.valid = 0;
 	status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0);

diff -ruN linux-2.4.23-pre7/fs/nfs/nfs3xdr.c linux-2.4.23-rc1/fs/nfs/nfs3xdr.c

fs/nfs/nfs3xdr.c +4 -14
@@ -599,8 +599,6 @@
 u32 *
 nfs3_decode_dirent(u32 *p, struct nfs_entry *entry, int plus)
 {
-	struct nfs_entry old = *entry;
-
 	if (!*p++) {
 		if (!*p)
 			return ERR_PTR(-EAGAIN);
@@ -616,20 +614,12 @@
 	p = xdr_decode_hyper(p, &entry->cookie);
 
 	if (plus) {
-		p = xdr_decode_post_op_attr(p, &entry->fattr);
+		struct nfs_fattr fattr;
+		p = xdr_decode_post_op_attr(p, &fattr);
 		/* In fact, a post_op_fh3: */
 		if (*p++) {
-			p = xdr_decode_fhandle(p, &entry->fh);
-			/* Ugh -- server reply was truncated */
-			if (p == NULL) {
-				dprintk("NFS: FH truncated\n");
-				*entry = old;
-				return ERR_PTR(-EAGAIN);
-			}
-		} else {
-			/* If we don't get a file handle, the attrs
-			 * aren't worth a lot. */
-			entry->fattr.valid = 0;
+			struct nfs_fh fh;
+			p = xdr_decode_fhandle(p, &fh);
 		}
 	}
 

diff -ruN linux-2.4.23-pre7/fs/nfs/write.c linux-2.4.23-rc1/fs/nfs/write.c

fs/nfs/write.c +16 -2
@@ -225,8 +225,19 @@
 	struct inode *inode = page->mapping->host;
 	unsigned long end_index;
 	unsigned offset = PAGE_CACHE_SIZE;
+	int inode_referenced = 0;
 	int err;
 
+	/*
+	 * Note: We need to ensure that we have a reference to the inode
+	 *       if we are to do asynchronous writes. If not, waiting
+	 *       in nfs_wait_on_request() may deadlock with clear_inode().
+	 *
+	 *       If igrab() fails here, then it is in any case safe to
+	 *       call nfs_wb_page(), since there will be no pending writes.
+	 */
+	if (igrab(inode) != 0)
+		inode_referenced = 1;
 	end_index = inode->i_size >> PAGE_CACHE_SHIFT;
 
 	/* Ensure we've flushed out any previous writes */
@@ -244,7 +255,8 @@
 		goto out;
 do_it:
 	lock_kernel();
-	if (NFS_SERVER(inode)->wsize >= PAGE_CACHE_SIZE && !IS_SYNC(inode)) {
+	if (NFS_SERVER(inode)->wsize >= PAGE_CACHE_SIZE && !IS_SYNC(inode) &&
+			inode_referenced) {
 		err = nfs_writepage_async(NULL, inode, page, 0, offset);
 		if (err >= 0)
 			err = 0;
@@ -256,7 +268,9 @@
 	unlock_kernel();
 out:
 	UnlockPage(page);
-	return err; 
+	if (inode_referenced)
+		iput(inode);
+	return err;
 }
 
 /*
Signature · 5 lines
-- 
J.A. Magallon <jamagallon()able!es>     \                 Software is like sex:
werewolf!able!es                         \           It's better when it's free
Mandrake Linux release 10.0 (Cooker) for i586
Linux 2.4.23-rc1-jam1 (gcc 3.3.1 (Mandrake Linux 9.2 3.3.1-4mdk))
Signature · 8 lines
| G r e g  L o u i s         | gpg public key: 0x400B1AA86D9E3E64 |
|  http://www.bgl.nu/~glouis |   (on my website or any keyserver) |
|  http://wecanstopspam.org in signatures helps fight junk email. |
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/
Signature · 2 lines
-- 
pozsy
#9 lore raw

2.4.23-pac1 VIA/S3 DRM compilation error

In reply to Bernhard Rosenkraenzer
Quoted 6 lines · Bernhard Rosenkraenzer

$SUBJECT is at ftp.kernel.org/pub/linux/kernel/people/bero/2.4/2.4.23/

Changes since 2.4.23-rc5-pac1:
- Increase version number

similar error messages for the via driver. 2.4.22-ac4 worked fine.

make[4]: Entering directory `/usr/src/linux-2.4.22/drivers/char/drm'
gcc -D__KERNEL__ -I/usr/src/linux-2.4.22/include -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -fomit-frame-pointer -pipe -mpreferred-stack-boundary=2 -march=c3 -falign-functions=0 -falign-jumps=0 -falign-loops=0 -DDO_MUNMAP_4_ARGS -nostdinc -iwithprefix include -DKBUILD_BASENAME=savage_drv -c -o savage_drv.o savage_drv.c savage_drv.c: In function `savage_alloc_continuous_mem':
savage_drv.c:55: error: `drm_savage_alloc_cont_mem_t' undeclared (first use in this function)
savage_drv.c:55: error: (Each undeclared identifier is reported only once
savage_drv.c:55: error: for each function it appears in.)
savage_drv.c:55: error: syntax error before "cont_mem"
savage_drv.c:70: error: `cont_mem' undeclared (first use in this function)
savage_drv.c:70: error: syntax error before ')' token
savage_drv.c:70: error: syntax error before ')' token
savage_drv.c:94: error: `DRM_SAVAGE_MEM_LOCATION_PCI' undeclared (first use in this function)
savage_drv.c:136: error: syntax error before ')' token
savage_drv.c:136: error: syntax error before ')' token
savage_drv.c:139:2: warning: #warning "Race at the very least"
savage_drv.c: In function `savage_get_physics_address':
savage_drv.c:150: error: `drm_savage_get_physcis_address_t' undeclared (first use in this function)
savage_drv.c:150: error: syntax error before "req"
savage_drv.c:157: error: `req' undeclared (first use in this function)
savage_drv.c:157: error: syntax error before ')' token
savage_drv.c:157: error: syntax error before ')' token
savage_drv.c:161:2: warning: #warning "FIXME: need to redo logic for this"
savage_drv.c:182: error: syntax error before ')' token
savage_drv.c:182: error: syntax error before ')' token
savage_drv.c: In function `savage_free_cont_mem':
savage_drv.c:191: error: `drm_savage_alloc_cont_mem_t' undeclared (first use in this function)
savage_drv.c:191: error: syntax error before "cont_mem"
savage_drv.c:201: error: `cont_mem' undeclared (first use in this function)
savage_drv.c:201: error: syntax error before ')' token
savage_drv.c:201: error: syntax error before ')' token
savage_drv.c:203:2: warning: #warning "fix size overflow check"
savage_drv.c:225: error: too few arguments to function `do_munmap'
In file included from savage_drv.c:240:
drm_drv.h: At top level:
drm_drv.h:251: error: `DRM_IOCTL_SAVAGE_ALLOC_CONTINUOUS_MEM' undeclared here (not in a function)
drm_drv.h:251: error: nonconstant array index in initializer
drm_drv.h:251: error: (near initialization for `savage_ioctls')
drm_drv.h:251: error: `DRM_IOCTL_SAVAGE_GET_PHYSICS_ADDRESS' undeclared here (not in a function)
drm_drv.h:251: error: nonconstant array index in initializer
drm_drv.h:251: error: (near initialization for `savage_ioctls')
drm_drv.h:251: error: `DRM_IOCTL_SAVAGE_FREE_CONTINUOUS_MEM' undeclared here (not in a function)
drm_drv.h:251: error: nonconstant array index in initializer
drm_drv.h:251: error: (near initialization for `savage_ioctls')
make[4]: *** [savage_drv.o] Error 1