[GIT PULL] please pull one bug fix for v4.4

[GIT PULL] please pull one bug fix for v4.4

[Jeff Merkey]

Hi Linus,

The following changes since commit 527e9316f8ec44bd53d90fb9f611fa7ffff52bb9:

  Linux 4.4-rc4 (2015-12-06 15:43:12 -0800)

are available in the git repository at:

  https://github.com/jeffmerkey/linux/tree/fixes-for-linus

hw_breakpoint.c:  fix INT1 recursion and system hard hang.

Fixes a 13 year old bug in the int1 handler path that results in a
hard system lockup is someone triggers an int1 breakpoint in the
hardware and no perf event has been registered.   Prints a log message
and sets the resume flag in x86 and x86_64 to prevent the system from
locking up and gracefully prints a rate limited message.

Re: [GIT PULL] please pull one bug fix for v4.4

[Jeff Merkey]

Raw commit address for the diff is:

https://github.com/jeffmerkey/linux/commit/e1d3c76814c839a835286843f0bca33b0c5d0dd8.diff


On 12/13/15, Jeff Merkey <linux.mdb@gmail.com> wrote:
> Hi Linus,
>
> The following changes since commit
> 527e9316f8ec44bd53d90fb9f611fa7ffff52bb9:
>
>   Linux 4.4-rc4 (2015-12-06 15:43:12 -0800)
>
> are available in the git repository at:
>
>   https://github.com/jeffmerkey/linux/tree/fixes-for-linus
>
> hw_breakpoint.c:  fix INT1 recursion and system hard hang.
>
> Fixes a 13 year old bug in the int1 handler path that results in a
> hard system lockup is someone triggers an int1 breakpoint in the
> hardware and no perf event has been registered.   Prints a log message
> and sets the resume flag in x86 and x86_64 to prevent the system from
> locking up and gracefully prints a rate limited message.
>

Re: [GIT PULL] please pull one bug fix for v4.4

[Jeff Merkey]

Patch address with email included:

https://github.com/jeffmerkey/linux/commit/e1d3c76814c839a835286843f0bca33b0c5d0dd8.patch

Re: [GIT PULL] please pull one bug fix for v4.4

[Linus Torvalds]

The merge window is open, so I started looking at this that I had
ignored as being too late last time..

On Sun, Dec 13, 2015 at 7:02 PM, Jeff Merkey <linux.mdb@gmail.com> wrote:
> Patch address with email included:
>
> https://github.com/jeffmerkey/linux/commit/e1d3c76814c839a835286843f0bca33b0c5d0dd8.patch

So at a minimum, I'd like a sign-off for the patch. But also, can you
describe how to actually trigger the problem? Is this actually
triggerable without an ICE?

                Linus

Re: [GIT PULL] please pull one bug fix for v4.4

[Thomas Gleixner]

On Mon, 11 Jan 2016, Linus Torvalds wrote:

> The merge window is open, so I started looking at this that I had
> ignored as being too late last time..
> 
> On Sun, Dec 13, 2015 at 7:02 PM, Jeff Merkey <linux.mdb@gmail.com> wrote:
> > Patch address with email included:
> >
> > https://github.com/jeffmerkey/linux/commit/e1d3c76814c839a835286843f0bca33b0c5d0dd8.patch
> 
> So at a minimum, I'd like a sign-off for the patch. But also, can you
> describe how to actually trigger the problem? Is this actually
> triggerable without an ICE?

This patch has been discussed and the version above has issues. It's still not
decided how we are going to deal with that.

Thanks,

	tglx

Re: [GIT PULL] please pull one bug fix for v4.4

[Jeff Merkey]

On 1/11/16, Linus Torvalds <torvalds@linux-foundation.org> wrote:
> The merge window is open, so I started looking at this that I had
> ignored as being too late last time..
>
> On Sun, Dec 13, 2015 at 7:02 PM, Jeff Merkey <linux.mdb@gmail.com> wrote:
>> Patch address with email included:
>>
>> https://github.com/jeffmerkey/linux/commit/e1d3c76814c839a835286843f0bca33b0c5d0dd8.patch
>
> So at a minimum, I'd like a sign-off for the patch. But also, can you
> describe how to actually trigger the problem? Is this actually
> triggerable without an ICE?
>
>                 Linus
>

Hi Linus,

It is triggerable without an ice, any module can call set_debugreg dr7
and trigger it, and some do, and I suspect a lot of cases where folks
see hard hangs are related to this issue if some program happens to
stuff bogus data into dr6 -- most folks just don't know this is what
causes it.

That being said,  it seems to require an ICE of some sort fiddling
with the debug registers to cause it, though there are places in the
code I based on what I reviewed where someone running GDB from
userspace could trigger it because thread.debugreg6 gets whacked.
Not all of the code paths clear the trap flag in debugreg6 after a
trap occurs and this bit seems to get recycled.

I will send you an updated pull location properly signed off.

Jeff

Re: [GIT PULL] please pull one bug fix for v4.4

[Jeff Merkey]

On 1/11/16, Thomas Gleixner <tglx@linutronix.de> wrote:
> On Mon, 11 Jan 2016, Linus Torvalds wrote:
>
>> The merge window is open, so I started looking at this that I had
>> ignored as being too late last time..
>>
>> On Sun, Dec 13, 2015 at 7:02 PM, Jeff Merkey <linux.mdb@gmail.com> wrote:
>> > Patch address with email included:
>> >
>> > https://github.com/jeffmerkey/linux/commit/e1d3c76814c839a835286843f0bca33b0c5d0dd8.patch
>>
>> So at a minimum, I'd like a sign-off for the patch. But also, can you
>> describe how to actually trigger the problem? Is this actually
>> triggerable without an ICE?
>
> This patch has been discussed and the version above has issues. It's still
> not
> decided how we are going to deal with that.
>
> Thanks,
>
> 	tglx
>

Thank you for the prompt reply.  It's a complex issue in a heavily
traversed and stable code path.  That being said, the fix is trivial
and non-intrusive.

Jeff