Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1722243 > unrolled thread

Re: tip -ENOBOOT - bisected to locking/refcounts, x86/asm: Implement fast refcount overflow protection

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: tip -ENOBOOT - bisected to locking/refcounts, x86/asm: Implement  fast refcount overflow protection Ingo Molnar <mingo@kernel.org> - 2017-08-29 11:00 +0200
    Re: tip -ENOBOOT - bisected to locking/refcounts, x86/asm:  Implement fast refcount overflow protection Mike Galbraith <efault@gmx.de> - 2017-08-29 11:30 +0200
      Re: tip -ENOBOOT - bisected to locking/refcounts, x86/asm: Implement  fast refcount overflow protection Ingo Molnar <mingo@kernel.org> - 2017-08-29 11:30 +0200
        Re: tip -ENOBOOT - bisected to locking/refcounts, x86/asm:  Implement fast refcount overflow protection Mike Galbraith <efault@gmx.de> - 2017-08-29 12:10 +0200
          Re: tip -ENOBOOT - bisected to locking/refcounts, x86/asm: Implement  fast refcount overflow protection Kees Cook <keescook@chromium.org> - 2017-08-29 17:40 +0200
            Re: tip -ENOBOOT - bisected to locking/refcounts, x86/asm:  Implement fast refcount overflow protection Mike Galbraith <efault@gmx.de> - 2017-08-29 18:40 +0200

#1722243 — Re: tip -ENOBOOT - bisected to locking/refcounts, x86/asm: Implement fast refcount overflow protection

* Mike Galbraith <efault@gmx.de> wrote:

> Greetings,
> 
> Take 2 of KVM bisect as you work fingered $subject.  Take 1 was stymied
> by build dependencies (aa5d1b81, df340524) which I foolishly tried to
> skip, leading git bisect to end up handing me a list of commits that
> might be busted.  During take 2, I added those two as required.
> 
> Symptom is a few splats as below, with box finally hanging.  Network
> comes up, but neither ssh nor console login is possible.

Just to double check, does the patch below also cure the regression?
If we don't find the bug it might be safer than a full revert.

Thanks,

	Ingo

 arch/x86/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index d04809921993..3b83ef8f1e9b 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -55,7 +55,7 @@ config X86
 	select ARCH_HAS_KCOV			if X86_64
 	select ARCH_HAS_MMIO_FLUSH
 	select ARCH_HAS_PMEM_API		if X86_64
-	select ARCH_HAS_REFCOUNT
+#	select ARCH_HAS_REFCOUNT
 	select ARCH_HAS_UACCESS_FLUSHCACHE	if X86_64
 	select ARCH_HAS_SET_MEMORY
 	select ARCH_HAS_SG_CHAIN

[toc] | [next] | [standalone]

#1722270 — Re: tip -ENOBOOT - bisected to locking/refcounts, x86/asm: Implement fast refcount overflow protection

On Tue, 2017-08-29 at 10:58 +0200, Ingo Molnar wrote:
> * Mike Galbraith <efault@gmx.de> wrote:
> 
> > Greetings,
> > 
> > Take 2 of KVM bisect as you work fingered $subject.  Take 1 was stymied
> > by build dependencies (aa5d1b81, df340524) which I foolishly tried to
> > skip, leading git bisect to end up handing me a list of commits that
> > might be busted.  During take 2, I added those two as required.
> > 
> > Symptom is a few splats as below, with box finally hanging.  Network
> > comes up, but neither ssh nor console login is possible.
> 
> Just to double check, does the patch below also cure the regression?
> If we don't find the bug it might be safer than a full revert.

I reverted it, and vbox now boots.  Should I still try the below?

> 
> Thanks,
> 
> 	Ingo
> 
>  arch/x86/Kconfig | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
> index d04809921993..3b83ef8f1e9b 100644
> --- a/arch/x86/Kconfig
> +++ b/arch/x86/Kconfig
> @@ -55,7 +55,7 @@ config X86
>  	select ARCH_HAS_KCOV			if X86_64
>  	select ARCH_HAS_MMIO_FLUSH
>  	select ARCH_HAS_PMEM_API		if X86_64
> -	select ARCH_HAS_REFCOUNT
> +#	select ARCH_HAS_REFCOUNT
>  	select ARCH_HAS_UACCESS_FLUSHCACHE	if X86_64
>  	select ARCH_HAS_SET_MEMORY
>  	select ARCH_HAS_SG_CHAIN

[toc] | [prev] | [next] | [standalone]

#1722271

* Mike Galbraith <efault@gmx.de> wrote:

> On Tue, 2017-08-29 at 10:58 +0200, Ingo Molnar wrote:
> > * Mike Galbraith <efault@gmx.de> wrote:
> > 
> > > Greetings,
> > > 
> > > Take 2 of KVM bisect as you work fingered $subject.  Take 1 was stymied
> > > by build dependencies (aa5d1b81, df340524) which I foolishly tried to
> > > skip, leading git bisect to end up handing me a list of commits that
> > > might be busted.  During take 2, I added those two as required.
> > > 
> > > Symptom is a few splats as below, with box finally hanging.  Network
> > > comes up, but neither ssh nor console login is possible.
> > 
> > Just to double check, does the patch below also cure the regression?
> > If we don't find the bug it might be safer than a full revert.
> 
> I reverted it, and vbox now boots.  Should I still try the below?

Would be nice, so I can stick a Tested-by on it and such!

Thanks,

	Ingo

[toc] | [prev] | [next] | [standalone]

#1722296 — Re: tip -ENOBOOT - bisected to locking/refcounts, x86/asm: Implement fast refcount overflow protection

On Tue, 2017-08-29 at 11:27 +0200, Ingo Molnar wrote:
> * Mike Galbraith <efault@gmx.de> wrote:
> 
> > On Tue, 2017-08-29 at 10:58 +0200, Ingo Molnar wrote:
> > > * Mike Galbraith <efault@gmx.de> wrote:
> > > 
> > > > Greetings,
> > > > 
> > > > Take 2 of KVM bisect as you work fingered $subject.  Take 1 was stymied
> > > > by build dependencies (aa5d1b81, df340524) which I foolishly tried to
> > > > skip, leading git bisect to end up handing me a list of commits that
> > > > might be busted.  During take 2, I added those two as required.
> > > > 
> > > > Symptom is a few splats as below, with box finally hanging.  Network
> > > > comes up, but neither ssh nor console login is possible.
> > > 
> > > Just to double check, does the patch below also cure the regression?
> > > If we don't find the bug it might be safer than a full revert.
> > 
> > I reverted it, and vbox now boots.  Should I still try the below?
> 
> Would be nice, so I can stick a Tested-by on it and such!

Done, works.

	-Mike

[toc] | [prev] | [next] | [standalone]

#1722567

On Tue, Aug 29, 2017 at 3:03 AM, Mike Galbraith <efault@gmx.de> wrote:
> On Tue, 2017-08-29 at 11:27 +0200, Ingo Molnar wrote:
>> * Mike Galbraith <efault@gmx.de> wrote:
>>
>> > On Tue, 2017-08-29 at 10:58 +0200, Ingo Molnar wrote:
>> > > * Mike Galbraith <efault@gmx.de> wrote:
>> > >
>> > > > Greetings,
>> > > >
>> > > > Take 2 of KVM bisect as you work fingered $subject.  Take 1 was stymied
>> > > > by build dependencies (aa5d1b81, df340524) which I foolishly tried to
>> > > > skip, leading git bisect to end up handing me a list of commits that
>> > > > might be busted.  During take 2, I added those two as required.
>> > > >
>> > > > Symptom is a few splats as below, with box finally hanging.  Network
>> > > > comes up, but neither ssh nor console login is possible.
>> > >
>> > > Just to double check, does the patch below also cure the regression?
>> > > If we don't find the bug it might be safer than a full revert.
>> >
>> > I reverted it, and vbox now boots.  Should I still try the below?
>>
>> Would be nice, so I can stick a Tested-by on it and such!
>
> Done, works.

Was this the x86 refcount implementation, or a atomic -> refcount conversion?

-Kees

-- 
Kees Cook
Pixel Security

[toc] | [prev] | [next] | [standalone]

#1722602 — Re: tip -ENOBOOT - bisected to locking/refcounts, x86/asm: Implement fast refcount overflow protection

On Tue, 2017-08-29 at 08:39 -0700, Kees Cook wrote:
> 
> Was this the x86 refcount implementation, or a atomic -> refcount conversion?

mmm.... <oija board> yes.

# CONFIG_REFCOUNT_FULL is not set

	-Mike

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web