2007-12-20 04:53:13

by Yinghai Lu

[permalink] [raw]
Subject: not needed patch

Ingo.

commit fbdcf18df73758b2e187ab94678b30cd5f6ff9f9 is not needed.
another patch (by you !! commit 699d934d5f958d7944d195c03c334f28cc0b3669 x86: fixup cpu_info array conversion)
already removed clearing of c->cpu_index. in identify_cpu

also it is not consisent to smpboot_32.c. (it will assign id to cpu_index right after
*c = boot_cpu_data;
)

by revert commit fbdcf18df73758b2e187ab94678b30cd5f6ff9f9, we could use c->cpu_index in identify_cpu.

YH

commit fbdcf18df73758b2e187ab94678b30cd5f6ff9f9
Author: Mike Travis <[email protected]>
Date: Wed Dec 19 23:20:19 2007 +0100

x86: fix show cpuinfo cpu number always zero

when called by setup_arch) after smp_store_cpu_info() had set it to the
correct value.

The error shows up in 'cat /proc/cpuinfo' will all cpus = 0.

Signed-off-by: Mike Travis <[email protected]>
Cc: Andi Kleen <[email protected]>
Cc: Christoph Lameter <[email protected]>
Cc: Jack Steiner <[email protected]>
Cc: Suresh B Siddha <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>

diff --git a/arch/x86/kernel/smpboot_64.c b/arch/x86/kernel/smpboot_64.c
index 500670c..5948895 100644
--- a/arch/x86/kernel/smpboot_64.c
+++ b/arch/x86/kernel/smpboot_64.c
@@ -141,8 +141,8 @@ static void __cpuinit smp_store_cpu_info(int id)
struct cpuinfo_x86 *c = &cpu_data(id);

*c = boot_cpu_data;
- c->cpu_index = id;
identify_cpu(c);
+ c->cpu_index = id;
print_cpu_info(c);
}


2007-12-20 14:39:20

by Ingo Molnar

[permalink] [raw]
Subject: Re: not needed patch


* Yinghai Lu <[email protected]> wrote:

> Ingo.
>
> commit fbdcf18df73758b2e187ab94678b30cd5f6ff9f9 is not needed. another
> patch (by you !! commit 699d934d5f958d7944d195c03c334f28cc0b3669 x86:
> fixup cpu_info array conversion) already removed clearing of
> c->cpu_index. in identify_cpu
> also it is not consisent to smpboot_32.c. (it will assign id to
> cpu_index right after
>
> *c = boot_cpu_data;
> )

well, it might in the worst-case be a superfluous change, but not cause
any problems in 2.6.24, right?

> by revert commit fbdcf18df73758b2e187ab94678b30cd5f6ff9f9, we could
> use c->cpu_index in identify_cpu.

but that's 2.6.25 stuff, right? Travis?

Ingo

2007-12-20 14:56:41

by Mike Travis

[permalink] [raw]
Subject: Re: not needed patch

Ingo Molnar wrote:
> * Yinghai Lu <[email protected]> wrote:
>
>> Ingo.
>>
>> commit fbdcf18df73758b2e187ab94678b30cd5f6ff9f9 is not needed. another
>> patch (by you !! commit 699d934d5f958d7944d195c03c334f28cc0b3669 x86:
>> fixup cpu_info array conversion) already removed clearing of
>> c->cpu_index. in identify_cpu
>> also it is not consisent to smpboot_32.c. (it will assign id to
>> cpu_index right after
>>
>> *c = boot_cpu_data;
>> )
>
> well, it might in the worst-case be a superfluous change, but not cause
> any problems in 2.6.24, right?
>
>> by revert commit fbdcf18df73758b2e187ab94678b30cd5f6ff9f9, we could
>> use c->cpu_index in identify_cpu.
>
> but that's 2.6.25 stuff, right? Travis?

Since 6.5.24 needs to close soon, I'll queue up this change for 6.5.25.

Thanks everyone!

-Mike

2007-12-20 16:29:27

by Mike Travis

[permalink] [raw]
Subject: Re: not needed patch

Ingo Molnar wrote:
> * Yinghai Lu <[email protected]> wrote:
>
>> Ingo.
>>
>> commit fbdcf18df73758b2e187ab94678b30cd5f6ff9f9 is not needed. another
>> patch (by you !! commit 699d934d5f958d7944d195c03c334f28cc0b3669 x86:
>> fixup cpu_info array conversion) already removed clearing of
>> c->cpu_index. in identify_cpu
>> also it is not consisent to smpboot_32.c. (it will assign id to
>> cpu_index right after
>>
>> *c = boot_cpu_data;
>> )
>
> well, it might in the worst-case be a superfluous change, but not cause
> any problems in 2.6.24, right?
>
>> by revert commit fbdcf18df73758b2e187ab94678b30cd5f6ff9f9, we could
>> use c->cpu_index in identify_cpu.
>
> but that's 2.6.25 stuff, right? Travis?
>

Looking at this more closely, yes my change is not needed and should be
removed. I'm not sure what caused my cpu # to be all zeros when I was
testing, but it now works ok without my change.

Thanks,
Mike

2007-12-20 16:34:25

by Ingo Molnar

[permalink] [raw]
Subject: Re: not needed patch


* Mike Travis <[email protected]> wrote:

> >> by revert commit fbdcf18df73758b2e187ab94678b30cd5f6ff9f9, we could
> >> use c->cpu_index in identify_cpu.
> >
> > but that's 2.6.25 stuff, right? Travis?
> >
>
> Looking at this more closely, yes my change is not needed and should
> be removed. I'm not sure what caused my cpu # to be all zeros when I
> was testing, but it now works ok without my change.

how about this question:

> > well, it might in the worst-case be a superfluous change, but not
> > cause any problems in 2.6.24, right?

Ingo

2007-12-20 16:54:26

by Mike Travis

[permalink] [raw]
Subject: Re: not needed patch

Ingo Molnar wrote:
> * Mike Travis <[email protected]> wrote:
>
>>>> by revert commit fbdcf18df73758b2e187ab94678b30cd5f6ff9f9, we could
>>>> use c->cpu_index in identify_cpu.
>>> but that's 2.6.25 stuff, right? Travis?
>>>
>> Looking at this more closely, yes my change is not needed and should
>> be removed. I'm not sure what caused my cpu # to be all zeros when I
>> was testing, but it now works ok without my change.
>
> how about this question:
>
>>> well, it might in the worst-case be a superfluous change, but not
>>> cause any problems in 2.6.24, right?
>
> Ingo

I don't think it hurts anything except as noted, it wouldn't be available
to the subfunctions (like identify_cpu) that might need the cpu_index
(which also is noted that none currently do.)

-Mike

2007-12-20 19:45:38

by Yinghai Lu

[permalink] [raw]
Subject: Re: not needed patch

On Thursday 20 December 2007 06:29:06 am Ingo Molnar wrote:
>
> * Yinghai Lu <[email protected]> wrote:
>
> > Ingo.
> >
> > commit fbdcf18df73758b2e187ab94678b30cd5f6ff9f9 is not needed. another
> > patch (by you !! commit 699d934d5f958d7944d195c03c334f28cc0b3669 x86:
> > fixup cpu_info array conversion) already removed clearing of
> > c->cpu_index. in identify_cpu
> > also it is not consisent to smpboot_32.c. (it will assign id to
> > cpu_index right after
> >
> > *c = boot_cpu_data;
> > )
>
> well, it might in the worst-case be a superfluous change, but not cause
> any problems in 2.6.24, right?

now it is ok with 2.6.24.

>
> > by revert commit fbdcf18df73758b2e187ab94678b30cd5f6ff9f9, we could
> > use c->cpu_index in identify_cpu.
>
> but that's 2.6.25 stuff, right? Travis?
or at least before bewfore merging smpboot_32.c and smpboot_64.c

YH