Re: linux-next: Tree for Jun 25

["Yann E. MORIN" <yann.morin.1998@free.fr>]

Michal, All,

[Sorry for duplicates, I inadvertently sent HTML]

On Tuesday 25 June 2013 11:21:53 Yann E. MORIN wrote:
> On Tuesday 25 June 2013 10:55:20 Sedat Dilek wrote:
> > Not sure what's going on, but I get this with today's Linux-Next.
> > ( Yesterday's next-20130624 was OK. )
> > 
> > $ LANG=C LC_ALL=C  yes "" | make oldconfig && LANG=C LC_ALL=C make
> > silentoldconfig </dev/null
> [--SNIP--]
> > RapidIO support (RAPIDIO) [Y/n/?] y
> >   IDT Tsi721 PCI Express SRIO Controller support (RAPIDIO_TSI721) [Y/n/?] y
> >   Discovery timeout duration (seconds) (RAPIDIO_DISC_TIMEOUT) [30] 30
> >   Enable RapidIO Input/Output Ports (RAPIDIO_ENABLE_RX_TX_PORTS) [N/y/?] n
> >   DMA Engine support for RapidIO (RAPIDIO_DMA_ENGINE) [Y/n/?] y
> >   RapidIO subsystem debug messages (RAPIDIO_DEBUG) [N/y/?] n
> >   Enumeration method [M/y/?] (NEW) aborted!
> > 
> > Console input/output is redirected. Run 'make oldconfig' to update
> > configuration.
> 
> Yes, the same error as reported by Fengguang earlier:
>     http://marc.info/?l=linux-kbuild&m=137214773928335&w=4
> 
> And caused by (yet once more):
>     8357b48: kconfig: fix randomising choice entries in presence of KCONFIG_ALLCONFIG
> 
> Basically, the issue is:
> 
>   - 'Enumeration method' is a tristate choice
>   - for tristate choices, it is possible to select zero, one or more of the
>     symbols in the choice
>   - if no symbol is selected (randomly or otherwise) in a tristate choice,
>     then the choice symbol itself has no assigned 'value'
>   - then silentoldconfig whines
> 
> It worked so far as a side effect of another bug (which 8357b48 tries to
> fix), and now this new one has been exposed.

OK, after battling on this during my lunch-break, I now better understand
the root cause. Consider this trivial Kconfig snippet:

    ---8<--- BEGIN Ktest ---8<---
    config MODULES
        bool "Modules"

    choice
        tristate "test"

    config A
        tristate "A"

    config B
        tristate "B"

    config C
        tristate "C"

    endchoice
    ---8<---  END Ktest  ---8<---

Now, this choice "test" is a tristate, so it can be either 'M' or 'Y'.

But since this is a choice _prompt_ *and* it does not have a symbol
associated, then its value is not stored in .config.

On a sub-sequent run of a kconfig utility, the Ktest file is parsed, and
the .config is read, associating read values to the symbols. But since
the choice prompt has no symbol, no value is assigned to it, so it is
believed to be new, and since it can be either 'M' or 'Y', it is asked
for.

Furthermore, even if the choice had a symbol, it would not be saved to
the .config either. And we'd be back to square-1.

> This one is starting to drive me crazy, really... :-(

Yes, it is! (head spining frantically, toons-like) ;-)

My lunch-break is over now, so more on this when I'mn back home, in ~6h.

Regards,
Yann E. MORIN.

-- 
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +0/33 662376056 | Software  Designer | \ / CAMPAIGN     |   ^                |
| --==< O_o >==-- '------------.-------:  X  AGAINST      |  /e\  There is no  |
| http://ymorin.is-a-geek.org/ | (*_*) | / \ HTML MAIL    |  """  conspiracy.  |
'------------------------------'-------'------------------'--------------------'