[patch 10/12] lkdtm: fix for CONFIG_SCSI=n

[patch 10/12] lkdtm: fix for CONFIG_SCSI=n

[akpm]

From: Randy Dunlap <randy.dunlap@oracle.com>

From: Randy Dunlap <randy.dunlap@oracle.com>

Fix lkdtm test code for case of CONFIG_SCSI=n, to avoid build errors:

In file included from linux-next-20080618/tests/lkdtm.c:55:
linux-next-20080618/include/scsi/scsi_cmnd.h:28:25: warning: "BLK_MAX_CDB" is not defined
linux-next-20080618/include/scsi/scsi_cmnd.h:29:3: error: #error MAX_COMMAND_SIZE can not be bigger than BLK_MAX_CDB
In file included from linux-next-20080618/tests/lkdtm.c:55:
linux-next-20080618/include/scsi/scsi_cmnd.h: In function 'scsi_bidi_cmnd':
linux-next-20080618/include/scsi/scsi_cmnd.h:183: error: implicit declaration of function 'blk_bidi_rq'
linux-next-20080618/include/scsi/scsi_cmnd.h:184: error: dereferencing pointer to incomplete type
linux-next-20080618/include/scsi/scsi_cmnd.h: In function 'scsi_in':
linux-next-20080618/include/scsi/scsi_cmnd.h:190: error: dereferencing pointer to incomplete type
make[2]: *** [tests/lkdtm.o] Error 1

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Acked-by: Ankita Garg <ankita@in.ibm.com> 
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---

 tests/lkdtm.c |    8 ++++++++
 1 file changed, 8 insertions(+)

diff -puN tests/lkdtm.c~lkdtm-fix-for-config_scsi=n tests/lkdtm.c
--- a/tests/lkdtm.c~lkdtm-fix-for-config_scsi=n
+++ a/tests/lkdtm.c
@@ -53,7 +53,9 @@
 #include <linux/interrupt.h>
 #include <linux/hrtimer.h>
 #include <linux/ide.h>
+#ifdef CONFIG_SCSI
 #include <scsi/scsi_cmnd.h>
+#endif
 
 #define NUM_CPOINTS 8
 #define NUM_CPOINT_TYPES 5
@@ -172,12 +174,14 @@ static int jp_hrtimer_start(struct hrtim
 	return 0;
 }
 
+#ifdef CONFIG_SCSI
 static int jp_scsi_dispatch_cmd(struct scsi_cmnd *cmd)
 {
 	lkdtm_handler();
 	jprobe_return();
 	return 0;
 }
+#endif
 
 #ifdef CONFIG_IDE
 int jp_generic_ide_ioctl(ide_drive_t *drive, struct file *file,
@@ -304,8 +308,12 @@ static int __init lkdtm_module_init(void
 		lkdtm.entry = (kprobe_opcode_t*) jp_hrtimer_start;
 		break;
 	case SCSI_DISPATCH_CMD:
+#ifdef CONFIG_SCSI
 		lkdtm.kp.symbol_name = "scsi_dispatch_cmd";
 		lkdtm.entry = (kprobe_opcode_t*) jp_scsi_dispatch_cmd;
+#else
+		printk(KERN_INFO "lkdtm : Crash point not available\n");
+#endif
 		break;
 	case IDE_CORE_CP:
 #ifdef CONFIG_IDE
_

Re: [patch 10/12] lkdtm: fix for CONFIG_SCSI=n

[James Bottomley]

On Wed, 2008-07-30 at 12:44 -0700, akpm@linux-foundation.org wrote:
> From: Randy Dunlap <randy.dunlap@oracle.com>
> 
> From: Randy Dunlap <randy.dunlap@oracle.com>
> 
> Fix lkdtm test code for case of CONFIG_SCSI=n, to avoid build errors:
> 
> In file included from linux-next-20080618/tests/lkdtm.c:55:
> linux-next-20080618/include/scsi/scsi_cmnd.h:28:25: warning: "BLK_MAX_CDB" is not defined
> linux-next-20080618/include/scsi/scsi_cmnd.h:29:3: error: #error MAX_COMMAND_SIZE can not be bigger than BLK_MAX_CDB
> In file included from linux-next-20080618/tests/lkdtm.c:55:
> linux-next-20080618/include/scsi/scsi_cmnd.h: In function 'scsi_bidi_cmnd':
> linux-next-20080618/include/scsi/scsi_cmnd.h:183: error: implicit declaration of function 'blk_bidi_rq'
> linux-next-20080618/include/scsi/scsi_cmnd.h:184: error: dereferencing pointer to incomplete type
> linux-next-20080618/include/scsi/scsi_cmnd.h: In function 'scsi_in':
> linux-next-20080618/include/scsi/scsi_cmnd.h:190: error: dereferencing pointer to incomplete type
> make[2]: *** [tests/lkdtm.o] Error 1
> 
> Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
> Acked-by: Ankita Garg <ankita@in.ibm.com> 
> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
> ---
> 
>  tests/lkdtm.c |    8 ++++++++
>  1 file changed, 8 insertions(+)
> 
> diff -puN tests/lkdtm.c~lkdtm-fix-for-config_scsi=n tests/lkdtm.c
> --- a/tests/lkdtm.c~lkdtm-fix-for-config_scsi=n
> +++ a/tests/lkdtm.c

OK, you've sent me this one twice now ... what exactly is it?  There's
no tests directory in any kernel tree I have.

James

Re: [patch 10/12] lkdtm: fix for CONFIG_SCSI=n

[Andrew Morton]

On Wed, 30 Jul 2008 15:16:48 -0500
James Bottomley <James.Bottomley@HansenPartnership.com> wrote:

> On Wed, 2008-07-30 at 12:44 -0700, akpm@linux-foundation.org wrote:
> > From: Randy Dunlap <randy.dunlap@oracle.com>
> > 
> > From: Randy Dunlap <randy.dunlap@oracle.com>
> > 
> > Fix lkdtm test code for case of CONFIG_SCSI=n, to avoid build errors:
> > 
> > In file included from linux-next-20080618/tests/lkdtm.c:55:
> > linux-next-20080618/include/scsi/scsi_cmnd.h:28:25: warning: "BLK_MAX_CDB" is not defined
> > linux-next-20080618/include/scsi/scsi_cmnd.h:29:3: error: #error MAX_COMMAND_SIZE can not be bigger than BLK_MAX_CDB
> > In file included from linux-next-20080618/tests/lkdtm.c:55:
> > linux-next-20080618/include/scsi/scsi_cmnd.h: In function 'scsi_bidi_cmnd':
> > linux-next-20080618/include/scsi/scsi_cmnd.h:183: error: implicit declaration of function 'blk_bidi_rq'
> > linux-next-20080618/include/scsi/scsi_cmnd.h:184: error: dereferencing pointer to incomplete type
> > linux-next-20080618/include/scsi/scsi_cmnd.h: In function 'scsi_in':
> > linux-next-20080618/include/scsi/scsi_cmnd.h:190: error: dereferencing pointer to incomplete type
> > make[2]: *** [tests/lkdtm.o] Error 1
> > 
> > Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
> > Acked-by: Ankita Garg <ankita@in.ibm.com> 
> > Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
> > ---
> > 
> >  tests/lkdtm.c |    8 ++++++++
> >  1 file changed, 8 insertions(+)
> > 
> > diff -puN tests/lkdtm.c~lkdtm-fix-for-config_scsi=n tests/lkdtm.c
> > --- a/tests/lkdtm.c~lkdtm-fix-for-config_scsi=n
> > +++ a/tests/lkdtm.c
> 
> OK, you've sent me this one twice now ... what exactly is it?

Why didn't you tell me last time?

>  There's
> no tests directory in any kernel tree I have.

You don't have a copy of linux-next?  Wow.

I'll send it to Sam.

Re: [patch 10/12] lkdtm: fix for CONFIG_SCSI=n

[James Bottomley]

On Wed, 2008-07-30 at 13:42 -0700, Andrew Morton wrote:
> On Wed, 30 Jul 2008 15:16:48 -0500
> James Bottomley <James.Bottomley@HansenPartnership.com> wrote:
> 
> > On Wed, 2008-07-30 at 12:44 -0700, akpm@linux-foundation.org wrote:
> > > From: Randy Dunlap <randy.dunlap@oracle.com>
> > > 
> > > From: Randy Dunlap <randy.dunlap@oracle.com>
> > > 
> > > Fix lkdtm test code for case of CONFIG_SCSI=n, to avoid build errors:
> > > 
> > > In file included from linux-next-20080618/tests/lkdtm.c:55:
> > > linux-next-20080618/include/scsi/scsi_cmnd.h:28:25: warning: "BLK_MAX_CDB" is not defined
> > > linux-next-20080618/include/scsi/scsi_cmnd.h:29:3: error: #error MAX_COMMAND_SIZE can not be bigger than BLK_MAX_CDB
> > > In file included from linux-next-20080618/tests/lkdtm.c:55:
> > > linux-next-20080618/include/scsi/scsi_cmnd.h: In function 'scsi_bidi_cmnd':
> > > linux-next-20080618/include/scsi/scsi_cmnd.h:183: error: implicit declaration of function 'blk_bidi_rq'
> > > linux-next-20080618/include/scsi/scsi_cmnd.h:184: error: dereferencing pointer to incomplete type
> > > linux-next-20080618/include/scsi/scsi_cmnd.h: In function 'scsi_in':
> > > linux-next-20080618/include/scsi/scsi_cmnd.h:190: error: dereferencing pointer to incomplete type
> > > make[2]: *** [tests/lkdtm.o] Error 1
> > > 
> > > Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
> > > Acked-by: Ankita Garg <ankita@in.ibm.com> 
> > > Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
> > > ---
> > > 
> > >  tests/lkdtm.c |    8 ++++++++
> > >  1 file changed, 8 insertions(+)
> > > 
> > > diff -puN tests/lkdtm.c~lkdtm-fix-for-config_scsi=n tests/lkdtm.c
> > > --- a/tests/lkdtm.c~lkdtm-fix-for-config_scsi=n
> > > +++ a/tests/lkdtm.c
> > 
> > OK, you've sent me this one twice now ... what exactly is it?
> 
> Why didn't you tell me last time?

I assumed it was a piece of something I didn't have but which would
appear in mainline after the merge window opened.  Now the merge window
has opened and closed ...

> >  There's
> > no tests directory in any kernel tree I have.
> 
> You don't have a copy of linux-next?  Wow.

Erm .. what is it that was in linux-next before 2.6.26 and still isn't
in mainline?  Isn't that a linux-next protocol violation?

> I'll send it to Sam.

If he has the unmerged tree for this, sure.

James

Re: [patch 10/12] lkdtm: fix for CONFIG_SCSI=n

[Sam Ravnborg]

> 
> I assumed it was a piece of something I didn't have but which would
> appear in mainline after the merge window opened.  Now the merge window
> has opened and closed ...
> 
> > >  There's
> > > no tests directory in any kernel tree I have.
> > 
> > You don't have a copy of linux-next?  Wow.
> 
> Erm .. what is it that was in linux-next before 2.6.26 and still isn't
> in mainline?  Isn't that a linux-next protocol violation?
The pull request was sent late and ignored.

> 
> > I'll send it to Sam.
> 
> If he has the unmerged tree for this, sure.
I do.

	Sam