[Buildroot] Randomization of kconfig choices?

[Buildroot] Randomization of kconfig choices?

[Thomas Petazzoni]

Hello Yann,

I'm coming to the news regarding the randomization of kconfig choices?
I know you've put a lot of efforts to try to fix this problem, but if I
remember correctly so far the proposed solutions had some remaining
issues.

I'm jumping back on this topic because yesterday I helped a friend set
up a Buildroot system with X.org for an industrial PC, and we built the
Modular X.org. And I discovered that the issue that our host-python
doesn't have libxml2 support is still there, and it causes a build
failure of Mesa. And since Mesa can only be compiled (currently) when
the Modular X.org is enabled, we never see this problem in the
autobuilders because we default to the Kdrive/TinyX variant.

Thanks,

Thomas
-- 
Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com

[Buildroot] Randomization of kconfig choices?

[Yann E. MORIN]

Thomas, All,

On 2013-06-18 09:31 +0200, Thomas Petazzoni spake thusly:
> I'm coming to the news regarding the randomization of kconfig choices?
> I know you've put a lot of efforts to try to fix this problem, but if I
> remember correctly so far the proposed solutions had some remaining
> issues.

No, I could not come up with a satisfiable solution so far. I've put it
a bit in a limbo, so I can start afresh from a new angle.

There have been some fixes in this area recently, but alas, none fixes
the randomisation of choices in presence of KCONFIG_ALLCONFIG.

But I was planning to work on this this week. Let's see where that will
get us...

Regards,
Yann E. MORIN.

-- 
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'

[Buildroot] Randomization of kconfig choices?

[Thomas Petazzoni]

Dear Yann E. MORIN,

On Tue, 18 Jun 2013 18:49:04 +0200, Yann E. MORIN wrote:

> No, I could not come up with a satisfiable solution so far. I've put it
> a bit in a limbo, so I can start afresh from a new angle.
> 
> There have been some fixes in this area recently, but alas, none fixes
> the randomisation of choices in presence of KCONFIG_ALLCONFIG.
> 
> But I was planning to work on this this week. Let's see where that will
> get us...

Aren't there some people that know better the history of this kconfig
that would be able to help? It seems like you're now more-or-less the
only one patching this stuff.

Best regards,

Thomas
-- 
Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com

[Buildroot] Randomization of kconfig choices?

[Yann E. MORIN]

Thomas, All,

On 2013-06-18 19:00 +0200, Thomas Petazzoni spake thusly:
> On Tue, 18 Jun 2013 18:49:04 +0200, Yann E. MORIN wrote:
> > No, I could not come up with a satisfiable solution so far. I've put it
> > a bit in a limbo, so I can start afresh from a new angle.
> > 
> > There have been some fixes in this area recently, but alas, none fixes
> > the randomisation of choices in presence of KCONFIG_ALLCONFIG.
> > 
> > But I was planning to work on this this week. Let's see where that will
> > get us...
> 
> Aren't there some people that know better the history of this kconfig
> that would be able to help?

It seems no one knowledgeable is willing to invest time in this.

There was someone (Sam?) who said he'd look, but that was over a month
ago, and he did not reply to my last mail. :-/

I have a small idea at the back of my mind that I have to put to
testing. I /may/ put you to contribution later tonight, mind you! ;-)

> It seems like you're now more-or-less the
> only one patching this stuff.

Oh, no. There have been a few new-comers lately who did provide some
further (good) fixes to things we were not even aware of.

I was even recently directed to a collection of kconfig patches from a
completely unrelated project (but somehow similar in spirit to what
Buildroot is, only to a much smaller scale).

There are good things there, and we're trying to shape them before
submitting (that won't be in 3.10, and probably not even in 3.11 since
it will probably not be ready before the next merge window, I'm afraid).

But, yes, no one is as interested as I am to get those being acutally
_merged_ upstream, it seems... :-/

But anyway, enough talk; let's do some coding! :-)

Regards,
Yann E. MORIN.

-- 
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'

[Buildroot] Randomization of kconfig choices?

[Thomas Petazzoni]

Dear Yann E. MORIN,

On Tue, 18 Jun 2013 19:15:08 +0200, Yann E. MORIN wrote:

> > Aren't there some people that know better the history of this kconfig
> > that would be able to help?
> 
> It seems no one knowledgeable is willing to invest time in this.

Yeah, that's unfortunate :-(

> There was someone (Sam?) who said he'd look, but that was over a month
> ago, and he did not reply to my last mail. :-/
> 
> I have a small idea at the back of my mind that I have to put to
> testing.

Great!

> I /may/ put you to contribution later tonight, mind you! ;-)

I will most likely not be available later tonight, but tomorrow I can
do some testing of whatever you came up with!

> > It seems like you're now more-or-less the
> > only one patching this stuff.
> 
> Oh, no. There have been a few new-comers lately who did provide some
> further (good) fixes to things we were not even aware of.

Being on the linux-kbuild list, I've seen those contributions, but many
of them are mostly on the surface, not hacking the low levels of the
kconfig magic.

> I was even recently directed to a collection of kconfig patches from a
> completely unrelated project (but somehow similar in spirit to what
> Buildroot is, only to a much smaller scale).

Interesting.

> There are good things there, and we're trying to shape them before
> submitting (that won't be in 3.10, and probably not even in 3.11 since
> it will probably not be ready before the next merge window, I'm afraid).

Yeah, 3.11 is approaching!

Best regards,

Thomas
-- 
Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com

[Buildroot] Randomization of kconfig choices?

[Yann E. MORIN]

Thomas, All,

On 2013-06-18 19:15 +0200, Yann E. MORIN spake thusly:
> I have a small idea at the back of my mind that I have to put to
> testing. I /may/ put you to contribution later tonight, mind you! ;-)

OK, ~2000 (two thousands!) tests, and counting. And still no breakage
observed. Yet, choices are now properly randomised!

For the records, the previous attempts would break as soon as ~100-200
tests.

So, I'm rather inclined and confident to say: Woohoo! :-)
But let's see...

Regards,
Yann E. MORIN.

-- 
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'

[Buildroot] Randomization of kconfig choices?

[Thomas Petazzoni]

Dear Yann E. MORIN,

On Tue, 18 Jun 2013 20:10:36 +0200, Yann E. MORIN wrote:

> On 2013-06-18 19:15 +0200, Yann E. MORIN spake thusly:
> > I have a small idea at the back of my mind that I have to put to
> > testing. I /may/ put you to contribution later tonight, mind you! ;-)
> 
> OK, ~2000 (two thousands!) tests, and counting. And still no breakage
> observed. Yet, choices are now properly randomised!
> 
> For the records, the previous attempts would break as soon as ~100-200
> tests.
> 
> So, I'm rather inclined and confident to say: Woohoo! :-)
> But let's see...

Nice! Let's integrate your patch in Buildroot then, and see what
happens, it should also be a good testing.

Best regards,

Thomas
-- 
Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com