timing problem with building bounds.h?

timing problem with building bounds.h?

[Randy Dunlap]

2 times today I had problems with kernel builds using
O= and -j4 and include/linux/bounds.h being incomplete.
(using 2.6.29-rc5)


bounds.h was like this:

#ifndef __LINUX_BOUNDS_H__
#define __LINUX_BOUNDS_H__
/*
 * DO NOT MODIFY.
 *
 * This file was generated by Kbuild
 *
 */
	/// blank line here was the end of the file;



Is there something in top-level Kbuild or Makefile that
would explain this?

Thanks,
~Randy

Re: timing problem with building bounds.h?

[Sam Ravnborg]

On Mon, Feb 16, 2009 at 04:04:15PM -0800, Randy Dunlap wrote:
> 2 times today I had problems with kernel builds using
> O= and -j4 and include/linux/bounds.h being incomplete.
> (using 2.6.29-rc5)
> 
> 
> bounds.h was like this:
> 
> #ifndef __LINUX_BOUNDS_H__
> #define __LINUX_BOUNDS_H__
> /*
>  * DO NOT MODIFY.
>  *
>  * This file was generated by Kbuild
>  *
>  */
> 	/// blank line here was the end of the file;
> 
> 
> 
> Is there something in top-level Kbuild or Makefile that
> would explain this?

Not that I know of...
If you can reproduce it could you try:
1) What is the content of kernel/bounds.s
2) The output when building the kernel
[Even without V=1 this may give a clue where it goes wrong]

Was this with O=... build btw?

	Sam

Re: timing problem with building bounds.h?

[Randy Dunlap]

Sam Ravnborg wrote:
> On Mon, Feb 16, 2009 at 04:04:15PM -0800, Randy Dunlap wrote:
>> 2 times today I had problems with kernel builds using
>> O= and -j4 and include/linux/bounds.h being incomplete.
>> (using 2.6.29-rc5)
>>
>>
>> bounds.h was like this:
>>
>> #ifndef __LINUX_BOUNDS_H__
>> #define __LINUX_BOUNDS_H__
>> /*
>>  * DO NOT MODIFY.
>>  *
>>  * This file was generated by Kbuild
>>  *
>>  */
>> 	/// blank line here was the end of the file;
>>
>>
>>
>> Is there something in top-level Kbuild or Makefile that
>> would explain this?
> 
> Not that I know of...
> If you can reproduce it could you try:
> 1) What is the content of kernel/bounds.s
> 2) The output when building the kernel
> [Even without V=1 this may give a clue where it goes wrong]

I'll let you know if I can do that.

> Was this with O=... build btw?

Yes (see my 2nd line of text body :).

Thanks,
~Randy