[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#344205: linux-source-2.6.14: drivers/net/wireless/airo.c fails to compile



dann frazier, 2006-01-06 16:50:08 -0700 :

> tags 344205 + unreproducible

[...]

> I followed your instructions, and I can't reproduce.  I used the 686
> config from linux-image-2.6.12-1,

Maybe I'm misunderstanding you, but that's not what I did.  I'm not
trying to rebuild kernels with the config provided by Debian packages
(which I don't doubt would work), I'm trying to build my own kernel
packages from my own customised config.  Which I linked to in my
original bug report.

> The longrun.ko module exists in 2.6.15-1, is anything else needed?

  That was merely an example of why I wanted to build my own kernel
packages.  That one turned out to be a bad reason, but there might be
others (like, all the reasons justifying the existence of the
linux-source-* packages).

[End of "He doesn't even try to understand me, whine whine"]

  Anyway.  When reading the build log in more detail, then airo.c, I
understood the reason for the bug.  The function 'flashpchar' is
explicitly defined at the beginning of the file, but only in an
#ifdef/#endif block (testing for MICSUPPORT).  Unfortunately,
MICSUPPORT is #undefed earlier if CONFIG_CRYPTO is not defined.  Also,
the code using the flashpchar function seems not to be in a
#ifdef/#endif block.  So when the Crypto API is not enabled, you get a
call for a function whose prototype is missing.  If I just enable
that, then I can build my kernels again.

  I don't know what should be done, whether it's adding #ifdefs to the
function call or turning "#warning MIC support requires Crypto API"
into an #error, or tweaking the rules so that you can't enable the
airo module without the crypto API.  Your call, but I suggest this
should be at least signaled to upstream.

Roland.
-- 
Roland Mas

Just a little bit of you every day will surely keep the doctors away.
  -- Just a little bit of you (The Jackson Five)



Reply to: