From john.l.villalovos at intel.com  Mon Aug  2 14:08:22 2004
From: john.l.villalovos at intel.com (Villalovos, John L)
Date: Mon, 2 Aug 2004 14:08:22 -0700
Subject: [Ocfs-tools-devel] Error building ocfs-tools
Message-ID: <60C14C611F1DDD4198D53F2F43D8CA3B01864440@orsmsx410>

Is anyone else having problems building ocfs-tools?

I am building on Fedora Core 2 with all the patches installed.

I get the following output:

# make
<snip>
make -C ocfs2
make[1]: Entering directory `/root/ocfs2/tools/ocfs2'
make -C libocfs2
make[2]: Entering directory `/root/ocfs2/tools/ocfs2/libocfs2'
gcc  -fno-strict-aliasing -Wall -Wstrict-prototypes -Wno-format
-Wmissing-prototypes -Wmissing-declarations -O2   -DOCFS2_FLAT_INCLUDES
-Iinclude  -D__ILP32__  -o bitmap.o -c bitmap.c
bitmap.c:33:26: linux/bitops.h: No such file or directory
In file included from include/byteorder.h:36,
                 from include/ocfs2.h:45,
                 from bitmap.c:35:
/usr/include/asm/byteorder.h:6:2: warning: #warning using private kernel
header; include <endian.h> instead!
bitmap.c: In function `ocfs2_bitmap_set_generic':
bitmap.c:285: warning: implicit declaration of function
`__test_and_set_bit'
bitmap.c: In function `ocfs2_bitmap_clear_generic':
bitmap.c:311: warning: implicit declaration of function
`__test_and_clear_bit'
bitmap.c: In function `ocfs2_bitmap_test_generic':
bitmap.c:336: warning: implicit declaration of function `test_bit'
make[2]: *** [bitmap.o] Error 1
make[2]: Leaving directory `/root/ocfs2/tools/ocfs2/libocfs2'
make[1]: *** [libocfs2] Error 2
make[1]: Leaving directory `/root/ocfs2/tools/ocfs2'
make: *** [ocfs2] Error 2
#

From mark.fasheh at oracle.com  Mon Aug  2 14:53:13 2004
From: mark.fasheh at oracle.com (Mark Fasheh)
Date: Mon, 2 Aug 2004 14:53:13 -0700
Subject: [Ocfs-tools-devel] Error building ocfs-tools
In-Reply-To: <60C14C611F1DDD4198D53F2F43D8CA3B01864440@orsmsx410>
References: <60C14C611F1DDD4198D53F2F43D8CA3B01864440@orsmsx410>
Message-ID: <20040802215313.GJ8008@ca-server1.us.oracle.com>

On Mon, Aug 02, 2004 at 02:08:22PM -0700, Villalovos, John L wrote:
> Is anyone else having problems building ocfs-tools?
> 
> I am building on Fedora Core 2 with all the patches installed.
Hmm, it seems that linux/bitops.h doesn't exist on fc2 -- the stuff seems to
have moved to asm/bitops.h. The attached patch got things to compile on my
fc2 box (usually I compile on a RHEL3 machine):

I'm not sure what the proper fix ought to be...
	--Mark

--
Mark Fasheh
Software Developer, Oracle Corp
mark.fasheh at oracle.com

Index: ocfs2/libocfs2/bitmap.c
===================================================================
--- ocfs2/libocfs2/bitmap.c	(revision 189)
+++ ocfs2/libocfs2/bitmap.c	(working copy)
@@ -30,7 +30,7 @@
 #include <string.h>
 #include <limits.h>
 
-#include <linux/bitops.h>
+#include <asm/bitops.h>
 
 #include "ocfs2.h"
 
Index: ocfs2/mkfs.ocfs2/mkfs.c
===================================================================
--- ocfs2/mkfs.ocfs2/mkfs.c	(revision 189)
+++ ocfs2/mkfs.ocfs2/mkfs.c	(working copy)
@@ -38,7 +38,7 @@
 #include <malloc.h>
 #include <time.h>
 
-#include <linux/bitops.h>
+#include <asm/bitops.h>
 
 #include "ocfs2.h"
 

From john.l.villalovos at intel.com  Mon Aug  2 18:36:40 2004
From: john.l.villalovos at intel.com (Villalovos, John L)
Date: Mon, 2 Aug 2004 18:36:40 -0700
Subject: [Ocfs-tools-devel] Bugzilla component for ocfs-tools
Message-ID: <60C14C611F1DDD4198D53F2F43D8CA3B01864861@orsmsx410>

Could someone create a bugzilla component for ocfs-tools?

I can't seem to find one for the tools package.

Thanks,
John

From john.l.villalovos at intel.com  Mon Aug  2 18:57:34 2004
From: john.l.villalovos at intel.com (Villalovos, John L)
Date: Mon, 2 Aug 2004 18:57:34 -0700
Subject: [Ocfs-tools-devel] Error message patch
Message-ID: <60C14C611F1DDD4198D53F2F43D8CA3B0186486C@orsmsx410>

Simple little patch to tell people to try the "--no-raw" flag if the raw
bind fails.

Fedora Core 2 ships with /dev/rawctl disabled by default, so this will
probably be a common error.

John
-------------- next part --------------
A non-text attachment was scrubbed...
Name: rawerror.diff
Type: application/octet-stream
Size: 661 bytes
Desc: rawerror.diff
Url : http://oss.oracle.com/pipermail/ocfs-tools-devel/attachments/20040802/f2e409a0/attachment.obj 

From mark.fasheh at oracle.com  Mon Aug  2 19:17:26 2004
From: mark.fasheh at oracle.com (Mark Fasheh)
Date: Mon, 2 Aug 2004 19:17:26 -0700
Subject: [Ocfs-tools-devel] Error message patch
In-Reply-To: <60C14C611F1DDD4198D53F2F43D8CA3B0186486C@orsmsx410>
References: <60C14C611F1DDD4198D53F2F43D8CA3B0186486C@orsmsx410>
Message-ID: <20040803021726.GK8008@ca-server1.us.oracle.com>

On Mon, Aug 02, 2004 at 06:57:34PM -0700, Villalovos, John L wrote:
> Simple little patch to tell people to try the "--no-raw" flag if the raw
> bind fails.
> 
> Fedora Core 2 ships with /dev/rawctl disabled by default, so this will
> probably be a common error.
Yeah, I put that in there, along with the fix to compile -- it seems to work
fine in FC2 and RHEL3. If I broke some other compile, then I'll let someone
yell :)
	--Mark

--
Mark Fasheh
Software Developer, Oracle Corp
mark.fasheh at oracle.com

From manish.singh at oracle.com  Tue Aug  3 19:19:24 2004
From: manish.singh at oracle.com (Manish Singh)
Date: Tue, 3 Aug 2004 19:19:24 -0700
Subject: [Ocfs-tools-devel] Bugzilla component for ocfs-tools
In-Reply-To: <60C14C611F1DDD4198D53F2F43D8CA3B01864861@orsmsx410>
References: <60C14C611F1DDD4198D53F2F43D8CA3B01864861@orsmsx410>
Message-ID: <20040804021923.GA25650@ca-server1.us.oracle.com>

On Mon, Aug 02, 2004 at 06:36:40PM -0700, Villalovos, John L wrote:
> Could someone create a bugzilla component for ocfs-tools?
> 
> I can't seem to find one for the tools package.

There should be one:

http://oss.oracle.com/bugzilla/enter_bug.cgi?product=OCFS%20Tools

-Manish

From john.l.villalovos at intel.com  Wed Aug  4 10:58:19 2004
From: john.l.villalovos at intel.com (Villalovos, John L)
Date: Wed, 4 Aug 2004 10:58:19 -0700
Subject: [Ocfs-tools-devel] Bugzilla component for ocfs-tools
Message-ID: <60C14C611F1DDD4198D53F2F43D8CA3B018A90F6@orsmsx410>

manish.singh at oracle.com wrote:
> On Mon, Aug 02, 2004 at 06:36:40PM -0700, Villalovos, John L wrote:
>> Could someone create a bugzilla component for ocfs-tools?
>> 
>> I can't seem to find one for the tools package.
> 
> There should be one:
> 
> http://oss.oracle.com/bugzilla/enter_bug.cgi?product=OCFS%20Tools
> 

Error
	  	
Sorry; you do not have the permissions necessary to enter a bug against
this product.

That is what I get.

John

From john.l.villalovos at intel.com  Thu Aug  5 14:14:52 2004
From: john.l.villalovos at intel.com (Villalovos, John L)
Date: Thu, 5 Aug 2004 14:14:52 -0700
Subject: [Ocfs-tools-devel] Patch for byteorder stuff
Message-ID: <60C14C611F1DDD4198D53F2F43D8CA3B018FA601@orsmsx410>

I saw a comment in the byteorder.h file that says that we should do
something different.

So I did.

John
-------------- next part --------------
A non-text attachment was scrubbed...
Name: byteorder.diff
Type: application/octet-stream
Size: 16936 bytes
Desc: byteorder.diff
Url : http://oss.oracle.com/pipermail/ocfs-tools-devel/attachments/20040805/3104b086/attachment.obj 

From john.l.villalovos at intel.com  Thu Aug  5 15:02:59 2004
From: john.l.villalovos at intel.com (Villalovos, John L)
Date: Thu, 5 Aug 2004 15:02:59 -0700
Subject: [Ocfs-tools-devel] Patch for byteorder stuff
Message-ID: <60C14C611F1DDD4198D53F2F43D8CA3B018FA69F@orsmsx410>

ocfs-tools-devel-bounces at oss.oracle.com wrote:
> I saw a comment in the byteorder.h file that says that we should do
> something different. 
> 
> So I did.

Not very informative after reading that :(

Anyway I took the CPU_TO_DISK & DISK_TO_CPU macros from the GPL EVMS
code and used them instead of the kernel ones.

Plus it uses the byteswap.h file from glibc, which takes care of the
actual byte swapping.

John

From Joel.Becker at oracle.com  Thu Aug  5 15:23:46 2004
From: Joel.Becker at oracle.com (Joel Becker)
Date: Thu, 5 Aug 2004 15:23:46 -0700
Subject: [Ocfs-tools-devel] Patch for byteorder stuff
In-Reply-To: <60C14C611F1DDD4198D53F2F43D8CA3B018FA69F@orsmsx410>
References: <60C14C611F1DDD4198D53F2F43D8CA3B018FA69F@orsmsx410>
Message-ID: <20040805222346.GB2595@ca-server1.us.oracle.com>

On Thu, Aug 05, 2004 at 03:02:59PM -0700, Villalovos, John L wrote:
> Anyway I took the CPU_TO_DISK & DISK_TO_CPU macros from the GPL EVMS
> code and used them instead of the kernel ones.

	DISK being little endian and size proper?

Joel

-- 

"The opposite of a correct statement is a false statement. The
 opposite of a profound truth may well be another profound truth."
         - Niels Bohr 

Joel Becker
Senior Member of Technical Staff
Oracle Corporation
E-mail: joel.becker at oracle.com
Phone: (650) 506-8127

From john.l.villalovos at intel.com  Thu Aug  5 15:34:00 2004
From: john.l.villalovos at intel.com (Villalovos, John L)
Date: Thu, 5 Aug 2004 15:34:00 -0700
Subject: [Ocfs-tools-devel] Patch for byteorder stuff
Message-ID: <60C14C611F1DDD4198D53F2F43D8CA3B018FA722@orsmsx410>

joel.becker at oracle.com wrote:
> On Thu, Aug 05, 2004 at 03:02:59PM -0700, Villalovos, John L wrote:
>> Anyway I took the CPU_TO_DISK & DISK_TO_CPU macros from the GPL EVMS
>> code and used them instead of the kernel ones.
> 
> 	DISK being little endian and size proper?

Yes.  DISK is in little endian format.

They have the following macros:

CPU_TO_DISK16
CPU_TO_DISK32
CPU_TO_DISK64

DISK_TO_CPU16
DISK_TO_CPU32
DISK_TO_CPU64

John

From john.l.villalovos at intel.com  Fri Aug  6 11:34:26 2004
From: john.l.villalovos at intel.com (Villalovos, John L)
Date: Fri, 6 Aug 2004 11:34:26 -0700
Subject: [Ocfs-tools-devel] fsck for OCFS2 yet?
Message-ID: <60C14C611F1DDD4198D53F2F43D8CA3B018FAE06@orsmsx410>

Is there an fsck.ocfs2 yet?

In svn the directory is empty on the trunk.  The directory being:
ocfs2/fsck.ocfs2/

John

From john.l.villalovos at intel.com  Fri Aug  6 11:27:26 2004
From: john.l.villalovos at intel.com (Villalovos, John L)
Date: Fri, 6 Aug 2004 11:27:26 -0700
Subject: [Ocfs-tools-devel] Patch for byteorder stuff
Message-ID: <60C14C611F1DDD4198D53F2F43D8CA3B018FADE3@orsmsx410>

joel.becker at oracle.com wrote:
> On Thu, Aug 05, 2004 at 03:02:59PM -0700, Villalovos, John L wrote:
>> Anyway I took the CPU_TO_DISK & DISK_TO_CPU macros from the GPL EVMS
>> code and used them instead of the kernel ones.
> 
> 	DISK being little endian and size proper?

I also have a patch to have it use the C99 standard for integers of
fixed sizes.

For example:

32 bit unsigned integer = unint32_t
32 bit signed integer   = int32_t

These come from #include <stdint.h> which is part of the C99 standard.

It appears to be present on AS 3 and above.

Let me know if any interest.

John

From mark.fasheh at oracle.com  Fri Aug  6 14:28:07 2004
From: mark.fasheh at oracle.com (Mark Fasheh)
Date: Fri, 6 Aug 2004 14:28:07 -0700
Subject: [Ocfs-tools-devel] fsck for OCFS2 yet?
In-Reply-To: <60C14C611F1DDD4198D53F2F43D8CA3B018FAE06@orsmsx410>
References: <60C14C611F1DDD4198D53F2F43D8CA3B018FAE06@orsmsx410>
Message-ID: <20040806212807.GV8008@ca-server1.us.oracle.com>

AFAIR, Joel's working on libocfs2 1st as fsck basically becomes a wrapper
around that.
	--Mark

On Fri, Aug 06, 2004 at 11:34:26AM -0700, Villalovos, John L wrote:
> Is there an fsck.ocfs2 yet?
> 
> In svn the directory is empty on the trunk.  The directory being:
> ocfs2/fsck.ocfs2/
> 
> John
> _______________________________________________
> Ocfs-tools-devel mailing list
> Ocfs-tools-devel at oss.oracle.com
> http://oss.oracle.com/mailman/listinfo/ocfs-tools-devel
--
Mark Fasheh
Software Developer, Oracle Corp
mark.fasheh at oracle.com

From Joel.Becker at oracle.com  Fri Aug  6 17:00:37 2004
From: Joel.Becker at oracle.com (Joel Becker)
Date: Fri, 6 Aug 2004 17:00:37 -0700
Subject: [Ocfs-tools-devel] fsck for OCFS2 yet?
In-Reply-To: <20040806212807.GV8008@ca-server1.us.oracle.com>
References: <60C14C611F1DDD4198D53F2F43D8CA3B018FAE06@orsmsx410>
	<20040806212807.GV8008@ca-server1.us.oracle.com>
Message-ID: <20040807000036.GI2595@ca-server1.us.oracle.com>

On Fri, Aug 06, 2004 at 02:28:07PM -0700, Mark Fasheh wrote:
> AFAIR, Joel's working on libocfs2 1st as fsck basically becomes a wrapper
> around that.

	Better answer: I think just about all of libocfs2 required for
fsck is done, but I've just barely started on the port of fsck.ext2, and
I've been working on other fires the past week.

Joel

-- 

"Three o'clock is always too late or too early for anything you
 want to do."
        - Jean-Paul Sartre

Joel Becker
Senior Member of Technical Staff
Oracle Corporation
E-mail: joel.becker at oracle.com
Phone: (650) 506-8127


From Joel.Becker at oracle.com  Fri Aug  6 16:59:46 2004
From: Joel.Becker at oracle.com (Joel Becker)
Date: Fri, 6 Aug 2004 16:59:46 -0700
Subject: [Ocfs-tools-devel] Patch for byteorder stuff
In-Reply-To: <60C14C611F1DDD4198D53F2F43D8CA3B018FADE3@orsmsx410>
References: <60C14C611F1DDD4198D53F2F43D8CA3B018FADE3@orsmsx410>
Message-ID: <20040806235946.GH2595@ca-server1.us.oracle.com>

On Fri, Aug 06, 2004 at 11:27:26AM -0700, Villalovos, John L wrote:
> I also have a patch to have it use the C99 standard for integers of
> fixed sizes.

	For what code?  libocfs2 uses those types.

Joel

-- 

"One of the symptoms of an approaching nervous breakdown is the
 belief that one's work is terribly important."
         - Bertrand Russell 

Joel Becker
Senior Member of Technical Staff
Oracle Corporation
E-mail: joel.becker at oracle.com
Phone: (650) 506-8127

From john.l.villalovos at intel.com  Fri Aug  6 17:26:59 2004
From: john.l.villalovos at intel.com (Villalovos, John L)
Date: Fri, 6 Aug 2004 17:26:59 -0700
Subject: [Ocfs-tools-devel] Patch for byteorder stuff
Message-ID: <60C14C611F1DDD4198D53F2F43D8CA3B0194A46A@orsmsx410>

joel.becker at oracle.com wrote:
> On Fri, Aug 06, 2004 at 11:27:26AM -0700, Villalovos, John L wrote:
>> I also have a patch to have it use the C99 standard for integers of
>> fixed sizes.
> 
> 	For what code?  libocfs2 uses those types.

A LOT of things use the kernel based ones.

Try a:
# grep -r '__\(u\|s\)\(8\|16\|32\|64\)' ocfs-tools/ocfs2/

This is a list of files that I have modified at this time.

[root at linuxjohn tools]# svn st
M      ocfs2/libocfs2/include/kernel-jbd.h
M      ocfs2/libocfs2/include/ocfs2_disk_dlm.h
M      ocfs2/libocfs2/include/ocfs2_fs.h
M      ocfs2/libocfs2/include/jfs_compat.h
M      ocfs2/libocfs2/include/ocfs1_fs_compat.h
M      ocfs2/debugfs.ocfs2/utils.c
M      ocfs2/debugfs.ocfs2/commands.c
M      ocfs2/debugfs.ocfs2/include/journal.h
M      ocfs2/debugfs.ocfs2/include/jbd.h
M      ocfs2/debugfs.ocfs2/include/utils.h
M      ocfs2/debugfs.ocfs2/include/readfs.h
M      ocfs2/debugfs.ocfs2/include/main.h
M      ocfs2/debugfs.ocfs2/include/dump.h
M      ocfs2/debugfs.ocfs2/readfs.c
M      ocfs2/debugfs.ocfs2/dump.c
M      ocfs2/debugfs.ocfs2/bindraw.c
M      ocfs2/debugfs.ocfs2/journal.c
M      ocfs2/mkfs.ocfs2/mkfs.c
M      ocfs2/mkfs.ocfs2/mkfs2.c
M      ocfs2/mkfs.ocfs2/readdir.c

John

From Joel.Becker at oracle.com  Fri Aug  6 17:56:06 2004
From: Joel.Becker at oracle.com (Joel Becker)
Date: Fri, 6 Aug 2004 17:56:06 -0700
Subject: [Ocfs-tools-devel] Patch for byteorder stuff
In-Reply-To: <60C14C611F1DDD4198D53F2F43D8CA3B0194A46A@orsmsx410>
References: <60C14C611F1DDD4198D53F2F43D8CA3B0194A46A@orsmsx410>
Message-ID: <20040807005605.GL2595@ca-server1.us.oracle.com>

On Fri, Aug 06, 2004 at 05:26:59PM -0700, Villalovos, John L wrote:
> M      ocfs2/libocfs2/include/kernel-jbd.h
> M      ocfs2/libocfs2/include/ocfs2_disk_dlm.h
> M      ocfs2/libocfs2/include/ocfs2_fs.h
> M      ocfs2/libocfs2/include/jfs_compat.h
> M      ocfs2/libocfs2/include/ocfs1_fs_compat.h

	These are kernel compatibility files.  They are copied directly
from the kernel, and should use the kernel types.  That's why the kernel
exports __u32, etc.
	IOW, they should be left as-is.

> M      ocfs2/debugfs.ocfs2/utils.c
> M      ocfs2/debugfs.ocfs2/commands.c
> M      ocfs2/debugfs.ocfs2/include/journal.h
> M      ocfs2/debugfs.ocfs2/include/jbd.h
> M      ocfs2/debugfs.ocfs2/include/utils.h
> M      ocfs2/debugfs.ocfs2/include/readfs.h
> M      ocfs2/debugfs.ocfs2/include/main.h
> M      ocfs2/debugfs.ocfs2/include/dump.h
> M      ocfs2/debugfs.ocfs2/readfs.c
> M      ocfs2/debugfs.ocfs2/dump.c
> M      ocfs2/debugfs.ocfs2/bindraw.c
> M      ocfs2/debugfs.ocfs2/journal.c
> M      ocfs2/mkfs.ocfs2/mkfs.c
> M      ocfs2/mkfs.ocfs2/mkfs2.c
> M      ocfs2/mkfs.ocfs2/readdir.c

	These are not my fault :-)

Joel

-- 

"Hell is oneself, hell is alone, the other figures in it, merely projections."
        - T. S. Eliot

Joel Becker
Senior Member of Technical Staff
Oracle Corporation
E-mail: joel.becker at oracle.com
Phone: (650) 506-8127

From Sunil.Mushran at oracle.com  Sat Aug  7 15:58:46 2004
From: Sunil.Mushran at oracle.com (Sunil Mushran)
Date: 07 Aug 2004 15:58:46 -0700
Subject: [Ocfs-tools-devel] Patch for byteorder stuff
In-Reply-To: <60C14C611F1DDD4198D53F2F43D8CA3B018FADE3@orsmsx410>
References: <60C14C611F1DDD4198D53F2F43D8CA3B018FADE3@orsmsx410>
Message-ID: <1091919527.1274.18.camel@localhost.localdomain>

I am definitely interested in the endian fix. That's real value...
something I was planning to do at some point.

Not too sure about the other considering we had standardized on __u32
kernel format from get go. But I really don't care this way or that.

BTW, does this endian fix cover debugfs too or only libocfs2?

On Fri, 2004-08-06 at 11:27, Villalovos, John L wrote:
> joel.becker at oracle.com wrote:
> > On Thu, Aug 05, 2004 at 03:02:59PM -0700, Villalovos, John L wrote:
> >> Anyway I took the CPU_TO_DISK & DISK_TO_CPU macros from the GPL EVMS
> >> code and used them instead of the kernel ones.
> > 
> > 	DISK being little endian and size proper?
> 
> I also have a patch to have it use the C99 standard for integers of
> fixed sizes.
> 
> For example:
> 
> 32 bit unsigned integer = unint32_t
> 32 bit signed integer   = int32_t
> 
> These come from #include <stdint.h> which is part of the C99 standard.
> 
> It appears to be present on AS 3 and above.
> 
> Let me know if any interest.
> 
> John
> _______________________________________________
> Ocfs-tools-devel mailing list
> Ocfs-tools-devel at oss.oracle.com
> http://oss.oracle.com/mailman/listinfo/ocfs-tools-devel




From Joel.Becker at oracle.com  Thu Aug 12 12:45:39 2004
From: Joel.Becker at oracle.com (Joel Becker)
Date: Thu, 12 Aug 2004 12:45:39 -0700
Subject: [Ocfs-tools-devel] Patch for byteorder stuff
In-Reply-To: <60C14C611F1DDD4198D53F2F43D8CA3B018FA601@orsmsx410>
References: <60C14C611F1DDD4198D53F2F43D8CA3B018FA601@orsmsx410>
Message-ID: <20040812194539.GC3462@ca-server1.us.oracle.com>

On Thu, Aug 05, 2004 at 02:14:52PM -0700, Villalovos, John L wrote:
> I saw a comment in the byteorder.h file that says that we should do
> something different.

John,
	Ok, I've done a proper byteorder setup based on endian.h and
byteswap.h.  I use cpu_to_leN, because it is an understood idiom, but I
do a userspace glibc implementation therof.

Joel

-- 

Life's Little Instruction Book #306

	"Take a nap on Sunday afternoons."

Joel Becker
Senior Member of Technical Staff
Oracle Corporation
E-mail: joel.becker at oracle.com
Phone: (650) 506-8127