[Devel] Re: [PATCH] Fix up headers so we can munge them for use by userspace.

[Oren Laadan]

Applied.

(but removed comment about sys/types.h as per Serge's comment).

Oren.

Matt Helsley wrote:
> On Mon, Oct 05, 2009 at 09:20:27PM -0500, Serge E. Hallyn wrote:
>> I'm sorry, I'm really pretty ignorant when it comes to userspace includes.
> 
> Me too. The shallow similarities in kernel/userspace header organization have,
> in the past, inspired plenty of bad assumptions on my part.
> 
>> Can you give a little more explanation as to:
>> 	1. why we don't need sys/user.h, and why it was included originally?
> 
> sys/user.h includes definitions for register structs/array lengths.
> 
> The comment above the include says:
> /* i387 structure seen from kernel/userspace */
> 
> but the checkpoint headers do not re-use existing register structs/array
> lengths from sys/user.h since those same definitions are not used
> in the kernel.
> 
> Similarly, s390 does not re-use existing register structs/array lengths
> from sys/user.h.
> 
> Honestly, I don't know why they were included originally when none of
> these structs/array lengths were used.
> 
>> 	2. why we need sys/types.h first?  I don't see that mentioned in
>> 		the linux/types.h.
> 
> Maybe I was wrong -- I can't seem to reproduce any build errors by
> re-ordering those two.
> 
>> 	3. how much of this will be distro-dependent?
> 
> Not much. I think since we're talking about headers it's more about glibc
> and kernel versions than distros -- I don't think SLES organizes its
> glibc headers differently than RHEL or debian. Are you aware of any
> distro differences that aren't just due to different upstream versions?
> 
> I've been building x86-32 on Fedora 10 and Ubuntu Jaunty. From time to
> time I've also been testing and/or looking at headers from RHEL for s390 and
> ppc. If anything I'm seeing "older" headers which I would expect to have
> fewer of the required definitions and hence more of the problems related to
> glibc headers.
> 
>> Don't read this as questioning their correctness, just as my curiosity.
> 
> Sure. They are good questions, and even questions about correctness
> are welcome.
> 
> Cheers,
> 	-Matt Helsley
> 
>> thanks,
>> -serge
>>
>> Quoting Matt Helsley (matthltc at us.ibm.com):
>>> Signed-off-by: Matt Helsley <matthltc at us.ibm.com>
>>> ---
>>>  arch/s390/include/asm/checkpoint_hdr.h |    7 +++++--
>>>  arch/x86/include/asm/checkpoint_hdr.h  |    2 --
>>>  include/linux/checkpoint_hdr.h         |    4 ++++
>>>  3 files changed, 9 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/arch/s390/include/asm/checkpoint_hdr.h b/arch/s390/include/asm/checkpoint_hdr.h
>>> index b6ea8ce..bc9f624 100644
>>> --- a/arch/s390/include/asm/checkpoint_hdr.h
>>> +++ b/arch/s390/include/asm/checkpoint_hdr.h
>>> @@ -19,8 +19,6 @@
>>>
>>>  #ifdef __KERNEL__
>>>  #include <asm/processor.h>
>>> -#else
>>> -#include <sys/user.h>
>>>  #endif
>>>
>>>  #ifdef CONFIG_64BIT
>>> @@ -34,7 +32,12 @@
>>>   * NUM_FPRS defined in <asm/ptrace.h> to be 16
>>>   * NUM_APRS defined in <asm/ptrace.h> to be 16
>>>   * NUM_CR_WORDS defined in <asm/ptrace.h> to be 3
>>> + *              but is not yet in glibc headers.
>>>   */
>>> +#ifndef NUM_CR_WORDS
>>> +#define NUM_CR_WORDS 3
>>> +#endif
>>> +
>>>  struct ckpt_hdr_cpu {
>>>  	struct ckpt_hdr h;
>>>  	__u64 args[1];
>>> diff --git a/arch/x86/include/asm/checkpoint_hdr.h b/arch/x86/include/asm/checkpoint_hdr.h
>>> index 7a24de5..282bd31 100644
>>> --- a/arch/x86/include/asm/checkpoint_hdr.h
>>> +++ b/arch/x86/include/asm/checkpoint_hdr.h
>>> @@ -34,8 +34,6 @@
>>>  /* i387 structure seen from kernel/userspace */
>>>  #ifdef __KERNEL__
>>>  #include <asm/processor.h>
>>> -#else
>>> -#include <sys/user.h>
>>>  #endif
>>>
>>>  #ifdef CONFIG_X86_32
>>> diff --git a/include/linux/checkpoint_hdr.h b/include/linux/checkpoint_hdr.h
>>> index f8bd4e1..b318cee 100644
>>> --- a/include/linux/checkpoint_hdr.h
>>> +++ b/include/linux/checkpoint_hdr.h
>>> @@ -10,6 +10,10 @@
>>>   *  distribution for more details.
>>>   */
>>>
>>> +#ifndef __KERNEL__
>>> +/* In userspace sys/types.h must be included before linux/types.h */
>>> +#include <sys/types.h>
>>> +#endif
>>>  #include <linux/types.h>
>>>
>>>  #ifdef __KERNEL__
>>> -- 
>>> 1.5.6.3
>>>
>>> _______________________________________________
>>> Containers mailing list
>>> Containers at lists.linux-foundation.org
>>> https://lists.linux-foundation.org/mailman/listinfo/containers
_______________________________________________
Containers mailing list
Containers at lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/containers