Bug#648996: linux-2.6: FTBFS [m68k] in modpost: "strlen" [iscsi_target_mod.ko] undefined! (and others)

[Thorsten Glaser <tg@mirbsd.de>]

Ben Hutchings dixit:

>But there are no warnings in the log for use of these functions
>without declarations.  So if <linux/string.h> is not included
>already (indirectly), where are they declared?  I don't believe

I fear they might be correct and gcc replaces certain function
calls with others, or – worse – replaces certain “expanded” code
with function calls, AGAIN. I’ve seen this behaviour in my other
kernel code sometimes and build with -ffreestanding (but does that
guarantee this won’t happen?) to maybe fix that.

Since the Linux Kernel build system is terminally broken¹, I don’t
even know whether it uses -ffreestanding, all I see is “CC”…

bye,
//mirabilos
① unless you pass V=1 or something (can never seem to recall which
  of these applies to which of the broken build system set), but I
  can’t really do that without changing the source package, which,
  in turn, I don’t do if I want to produce standard, reproducible,
  Debian packages ready for uploading to the archive, as close to
  the rules for main as possible.
-- 
Solange man keine schmutzigen Tricks macht, und ich meine *wirklich*
schmutzige Tricks, wie bei einer doppelt verketteten Liste beide
Pointer XORen und in nur einem Word speichern, funktioniert Boehm ganz
hervorragend.		-- Andreas Bogk über boehm-gc in d.a.s.r