ROOT NFS fixes ...

ROOT NFS fixes ...

[Herbert Pötzl]

Hi Trond!

please have a look at the attached patch, which
addresses the root=/dev/wossname issue, when ROOT_NFS
is configured. this is a NFS only issue, and IMHO
fixes an erroneous behaviour.

if it seems okay to you, please let Marcelo know,
so he could add this patch to the next -pre kernel

MTIA,
Herbert


------------- NFS ROOT FIX

diff -NurbP --minimal linux-2.4.22-pre7/fs/nfs/nfsroot.c linux-2.4.22-pre7-fix/fs/nfs/nfsroot.c
--- linux-2.4.22-pre7/fs/nfs/nfsroot.c	2003-06-13 16:51:37.000000000 +0200
+++ linux-2.4.22-pre7-fix/fs/nfs/nfsroot.c	2003-07-21 22:15:12.000000000 +0200
@@ -305,7 +305,7 @@
  */
 int __init nfs_root_setup(char *line)
 {
-	ROOT_DEV = MKDEV(UNNAMED_MAJOR, 255);
+	ROOT_DEV = MKDEV(NFS_MAJOR, NFS_MINOR);
 	if (line[0] == '/' || line[0] == ',' || (line[0] >= '0' && line[0] <= '9')) {
 		strncpy(nfs_root_name, line, sizeof(nfs_root_name));
 		nfs_root_name[sizeof(nfs_root_name)-1] = '\0';
diff -NurbP --minimal linux-2.4.22-pre7/include/linux/nfs.h linux-2.4.22-pre7-fix/include/linux/nfs.h
--- linux-2.4.22-pre7/include/linux/nfs.h	2000-04-01 18:04:27.000000000 +0200
+++ linux-2.4.22-pre7-fix/include/linux/nfs.h	2003-07-21 22:13:12.000000000 +0200
@@ -30,6 +30,9 @@
 #define NFS_MNT_PROGRAM	100005
 #define NFS_MNT_PORT	627
 
+#define NFS_MAJOR   	UNNAMED_MAJOR
+#define NFS_MINOR   	0xff
+
 /*
  * NFS stats. The good thing with these values is that NFSv3 errors are
  * a superset of NFSv2 errors (with the exception of NFSERR_WFLUSH which
diff -NurbP --minimal linux-2.4.22-pre7/init/do_mounts.c linux-2.4.22-pre7-fix/init/do_mounts.c
--- linux-2.4.22-pre7/init/do_mounts.c	2003-07-19 14:14:31.000000000 +0200
+++ linux-2.4.22-pre7-fix/init/do_mounts.c	2003-07-21 22:13:12.000000000 +0200
@@ -88,7 +88,7 @@
 	const char *name;
 	const int num;
 } root_dev_names[] __initdata = {
-	{ "nfs",     0x00ff },
+	{ "nfs",     MKDEV(NFS_MAJOR, NFS_MINOR) },
 	{ "hda",     0x0300 },
 	{ "hdb",     0x0340 },
 	{ "loop",    0x0700 },
@@ -759,7 +759,8 @@
 static void __init mount_root(void)
 {
 #ifdef CONFIG_ROOT_NFS
-	if (MAJOR(ROOT_DEV) == UNNAMED_MAJOR) {
+       if (MAJOR(ROOT_DEV) == NFS_MAJOR
+           && MINOR(ROOT_DEV) == NFS_MINOR) {
 		if (mount_nfs_root()) {
 			sys_chdir("/root");
 			ROOT_DEV = current->fs->pwdmnt->mnt_sb->s_dev;
diff -NurbP --minimal linux-2.4.22-pre7/net/ipv4/ipconfig.c linux-2.4.22-pre7-fix/net/ipv4/ipconfig.c
--- linux-2.4.22-pre7/net/ipv4/ipconfig.c	2003-07-19 14:14:31.000000000 +0200
+++ linux-2.4.22-pre7-fix/net/ipv4/ipconfig.c	2003-07-21 22:15:50.000000000 +0200
@@ -1234,7 +1234,7 @@
 			 * 				-- Chip
 			 */
 #ifdef CONFIG_ROOT_NFS
-			if (ROOT_DEV == MKDEV(UNNAMED_MAJOR, 255)) {
+			if (ROOT_DEV == MKDEV(NFS_MAJOR, NFS_MINOR)) {
 				printk(KERN_ERR 
 					"IP-Config: Retrying forever (NFS root)...\n");
 				goto try_try_again;

Re: ROOT NFS fixes ...

[Marcelo Tosatti]

On Wed, 30 Jul 2003, Herbert Pötzl wrote:

> On Wed, Jul 30, 2003 at 11:28:46AM -0300, Marcelo Tosatti wrote:
> >
> > On Wed, 30 Jul 2003, Marcelo Tosatti wrote:
> > > On Wed, 30 Jul 2003, Herbert Pötzl wrote:
> > > > On Wed, Jul 30, 2003 at 10:57:52AM -0300, Marcelo Tosatti wrote:
> > > > > On Tue, 29 Jul 2003, Herbert Pötzl wrote:
> > > > >
> > > > > >
> > > > > > Hi Marcelo!
> > > > > >
> > > > > > just verified that the NFS root bug-fix was not
> > > > > > included in 2.4.22-pre9, unfortunately I have
> > > > > > to ask you again, why you do not want to fix
> > > > > > this issue in 2.4.22 ...
> > > > > >
> > > > > > I do not understand why Trond obviously is
> > > > > > ignoring my mails, regarding this particular
> > > > > > issue, maybe he is just too busy to look at
> > > > > > four twoline changes, and more, I do not
> > > > > > understand why this isn't accepted into the
> > > > > > marcelo kernel tree, as it obviously fixes a
> > > > > > misbehaviour?
> > > > > >
> > > > > > please explain!
> > > > > >
> > > > > > It is okay for me, if your argumentation goes
> > > > > > like "I don't like you, that's reason enough
> > > > > > for me to not include your patches ...", but I
> > > > > > would like to know ...
> > > > >
> > > > > I do not consider the patch critical enough.
> > > > >
> > > > > Get it in 2.5 first, then come back :)
> > > >
> > > > I hope this is a joke, and you are still reading
> > > > your mail ...
> > >
> > > No, this is not a joke, at all.
> > >
> > > Let me repeat: I (and Trond) do not consider this patch critical.
> >
> > Ok, I'm wrong. I just read Trond's mail saying the patch is OK.
> >
> > I'll apply it to -pre10.
>
> hmm, maybe it _is_ too early to do this ;)
> but on the other hand, I'm a fearless creature ...

What do you mean too early ?

Trond is OK with it, it seems.

Re: ROOT NFS fixes ...

[Herbert Pötzl]

On Wed, Jul 30, 2003 at 11:28:46AM -0300, Marcelo Tosatti wrote:
> 
> On Wed, 30 Jul 2003, Marcelo Tosatti wrote:
> > On Wed, 30 Jul 2003, Herbert Pötzl wrote:
> > > On Wed, Jul 30, 2003 at 10:57:52AM -0300, Marcelo Tosatti wrote:
> > > > On Tue, 29 Jul 2003, Herbert Pötzl wrote:
> > > >
> > > > >
> > > > > Hi Marcelo!
> > > > >
> > > > > just verified that the NFS root bug-fix was not
> > > > > included in 2.4.22-pre9, unfortunately I have
> > > > > to ask you again, why you do not want to fix
> > > > > this issue in 2.4.22 ...
> > > > >
> > > > > I do not understand why Trond obviously is
> > > > > ignoring my mails, regarding this particular
> > > > > issue, maybe he is just too busy to look at
> > > > > four twoline changes, and more, I do not
> > > > > understand why this isn't accepted into the
> > > > > marcelo kernel tree, as it obviously fixes a
> > > > > misbehaviour?
> > > > >
> > > > > please explain!
> > > > >
> > > > > It is okay for me, if your argumentation goes
> > > > > like "I don't like you, that's reason enough
> > > > > for me to not include your patches ...", but I
> > > > > would like to know ...
> > > >
> > > > I do not consider the patch critical enough.
> > > >
> > > > Get it in 2.5 first, then come back :)
> > >
> > > I hope this is a joke, and you are still reading
> > > your mail ...
> >
> > No, this is not a joke, at all.
> >
> > Let me repeat: I (and Trond) do not consider this patch critical.
> 
> Ok, I'm wrong. I just read Trond's mail saying the patch is OK.
> 
> I'll apply it to -pre10.

hmm, maybe it _is_ too early to do this ;)
but on the other hand, I'm a fearless creature ...

many thanks,
Herbert

Re: ROOT NFS fixes ...

[Marcelo Tosatti]

On Wed, 30 Jul 2003, Marcelo Tosatti wrote:

>
> On Wed, 30 Jul 2003, Herbert Pötzl wrote:
>
> > On Wed, Jul 30, 2003 at 10:57:52AM -0300, Marcelo Tosatti wrote:
> > >
> > >
> > > On Tue, 29 Jul 2003, Herbert Pötzl wrote:
> > >
> > > >
> > > > Hi Marcelo!
> > > >
> > > > just verified that the NFS root bug-fix was not
> > > > included in 2.4.22-pre9, unfortunately I have
> > > > to ask you again, why you do not want to fix
> > > > this issue in 2.4.22 ...
> > > >
> > > > I do not understand why Trond obviously is
> > > > ignoring my mails, regarding this particular
> > > > issue, maybe he is just too busy to look at
> > > > four twoline changes, and more, I do not
> > > > understand why this isn't accepted into the
> > > > marcelo kernel tree, as it obviously fixes a
> > > > misbehaviour?
> > > >
> > > > please explain!
> > > >
> > > > It is okay for me, if your argumentation goes
> > > > like "I don't like you, that's reason enough
> > > > for me to not include your patches ...", but I
> > > > would like to know ...
> > >
> > > I do not consider the patch critical enough.
> > >
> > > Get it in 2.5 first, then come back :)
> >
> > I hope this is a joke, and you are still reading
> > your mail ...
>
> No, this is not a joke, at all.
>
> Let me repeat: I (and Trond) do not consider this patch critical.

Ok, I'm wrong. I just read Trond's mail saying the patch is OK.

I'll apply it to -pre10.

Re: ROOT NFS fixes ...

[Marc-Christian Petersen]

On Wednesday 30 July 2003 15:57, Marcelo Tosatti wrote:

Hi Marcelo, Herbert,

> > just verified that the NFS root bug-fix was not
> > included in 2.4.22-pre9, unfortunately I have
> > to ask you again, why you do not want to fix
> > this issue in 2.4.22 ...
> > I do not understand why Trond obviously is
> > ignoring my mails, regarding this particular
> > issue, maybe he is just too busy to look at
> > four twoline changes, and more, I do not
> > understand why this isn't accepted into the
> > marcelo kernel tree, as it obviously fixes a
> > misbehaviour?

> I do not consider the patch critical enough.
> Get it in 2.5 first, then come back :)
err, correct me if I am totally wrong here but does "ROOT NFS fixes" means 
somewhat like

append="nfsroot=10.0.0.7:/home/nfsroot \
    ip=10.0.0.254:10.0.0.7:10.0.0.7:255.255.255.0:temp:eth0:off root=/dev/nfs"

does not work w/o the fix?

ciao, Marc

Re: ROOT NFS fixes ...

[Marcelo Tosatti]

On Wed, 30 Jul 2003, Herbert Pötzl wrote:

> On Wed, Jul 30, 2003 at 10:57:52AM -0300, Marcelo Tosatti wrote:
> >
> >
> > On Tue, 29 Jul 2003, Herbert Pötzl wrote:
> >
> > >
> > > Hi Marcelo!
> > >
> > > just verified that the NFS root bug-fix was not
> > > included in 2.4.22-pre9, unfortunately I have
> > > to ask you again, why you do not want to fix
> > > this issue in 2.4.22 ...
> > >
> > > I do not understand why Trond obviously is
> > > ignoring my mails, regarding this particular
> > > issue, maybe he is just too busy to look at
> > > four twoline changes, and more, I do not
> > > understand why this isn't accepted into the
> > > marcelo kernel tree, as it obviously fixes a
> > > misbehaviour?
> > >
> > > please explain!
> > >
> > > It is okay for me, if your argumentation goes
> > > like "I don't like you, that's reason enough
> > > for me to not include your patches ...", but I
> > > would like to know ...
> >
> > I do not consider the patch critical enough.
> >
> > Get it in 2.5 first, then come back :)
>
> I hope this is a joke, and you are still reading
> your mail ...

No, this is not a joke, at all.

Let me repeat: I (and Trond) do not consider this patch critical.

Re: ROOT NFS fixes ...

[Marcelo Tosatti]

On Tue, 29 Jul 2003, Herbert Pötzl wrote:

>
> Hi Marcelo!
>
> just verified that the NFS root bug-fix was not
> included in 2.4.22-pre9, unfortunately I have
> to ask you again, why you do not want to fix
> this issue in 2.4.22 ...
>
> I do not understand why Trond obviously is
> ignoring my mails, regarding this particular
> issue, maybe he is just too busy to look at
> four twoline changes, and more, I do not
> understand why this isn't accepted into the
> marcelo kernel tree, as it obviously fixes a
> misbehaviour?
>
> please explain!
>
> It is okay for me, if your argumentation goes
> like "I don't like you, that's reason enough
> for me to not include your patches ...", but I
> would like to know ...

I do not consider the patch critical enough.

Get it in 2.5 first, then come back :)

ROOT NFS fixes ...

[Herbert Pötzl]

Hi Marcelo!

just verified that the NFS root bug-fix was not
included in 2.4.22-pre9, unfortunately I have 
to ask you again, why you do not want to fix 
this issue in 2.4.22 ...

I do not understand why Trond obviously is
ignoring my mails, regarding this particular
issue, maybe he is just too busy to look at 
four twoline changes, and more, I do not 
understand why this isn't accepted into the 
marcelo kernel tree, as it obviously fixes a
misbehaviour?

please explain!

It is okay for me, if your argumentation goes 
like "I don't like you, that's reason enough
for me to not include your patches ...", but I
would like to know ...

best,
Herbert