Re: [PATCH] Revert 2fbd07a5f so machines with BSPs phsyical apicid != 0 can boot

[Suresh Siddha]

On Mon, 2010-01-11 at 16:46 -0800, Linus Torvalds wrote:
> 
> On Mon, 11 Jan 2010, Suresh Siddha wrote:
> > 
> > Linus, We are in -rc3 and thought we have few days atleast to sort it
> > out and post the correct fix to the problem, rather than do a quick
> > revert (as we know that the current code is not fundamentally broken).
> 
> You seem to think that -rc3 is "early". It's not. 
> 
> Also, you seem to dismiss the fact that the commit has been reported to 
> break real machines, and then you try to blame the MACHINE instead of 
> blaming the commit. 

Linus, I spent two hours in the morning reviewing code/testing this on
different platforms (removing/re-arranging sockets on different
platforms so that I am closer to Ananth's failing config) and I haven't
seen the failure. Even Yinghai tried it separately and couldn't see this
on his platform. 

And From Ananth's report, we do know there is a problem some where. I am
not blaming his MACHINE. I was trying to understand what is specific to
his platform/configuration, so that I can better understand where the
issue is. Sorry if my words sounded like blaming. Didn't really mean to.

> That makes me irritated. I don't understand why it's so hard for people to 
> see that if there is a problem IT NEEDS TO BE FIXED.
> 
> The default action should not be "let's keep the problem and then try to 
> figure it out". No, the default action is "let's FIX the problem first!"
> 
> Once the problem is fixed, you have as much time as you want to try to 
> figure out why it happened in the first time. But we do _not_ just keep a 
> broken kernel around because you don't know what is broken. 

Ok. I read the problem report today morning and after my testing, I had
some confidence that this is a not a widespread problem. So thought I
will take a day more to get more analysis/information from Ananth before
I ask you/x86 folks for revert. Didn't really mean to hold on to the
broken fix. And hence the ack when you wanted to revert.

> Quite frankly, I hope the "re-submit" is not actually that. There's no 
> point in submitting something like this again. I still think that the 
> whole "let's have different code-paths for Intel and AMD" thing is just 
> plain crazy. There's no reason to do this.
> 
> For example, quite apart from the actual problem report, your patch causes 
> the x86-64 code to simply become UGLIER AND LESS MAINTAINABLE. That whole 
> intel-vs-amd issue is total black magic, with no comments and no reason.

I will work with Yinghai who first observed the failure on AMD platform
and introduced this fix.

commit e0da33646826b66ef933d47ea2fb7a693fd849bf
Author: Yinghai Lu <yhlu.kernel@xxxxxxxxx>
Date:   Sun Jun 8 18:29:22 2008 -0700

    x86: introduce max_physical_apicid for bigsmp switching
    
    a multi-socket test-system with 3 or 4 ioapics, when 4 dualcore cpus or
    2 quadcore cpus installed, needs to switch to bigsmp or physflat.
    
    CPU apic id is [4,11] instead of [0,7], and we need to check max apic
    id instead of cpu numbers.
    
    also add check for 32 bit when acpi is not compiled in or acpi=off.
    

> So no. I'm not going to take a resubmission. 

I will work with Yinghai and Ananth to come up with a clean solution.

thanks,
suresh

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/