Loading ...
Sorry, an error occurred while loading the content.

13820Re: stupid n00b question on kernel compiles

Expand Messages
  • Alan Clucas
    Jul 4, 2002
    • 0 Attachment
      > On Tue, 2 Jul 2002 21:16:28 +0100, Chris Howells <chrish@...> said:
      > > On Tuesday 02 July 2002 7:30 pm, david carlton wrote:
      >
      > >> Is there any more reason to do this when compiling the kernel than
      > >> there is when compiling other pieces of software? (Other than the
      > >> fact that the kernel's more important than most pieces of software, of
      > >> course.)
      >
      > > I think it's just standard practice to ensure that there aren't any
      > > stale object files around that could cause weird breakages.
      >
      > Sure, but is there any more reason for that to happen when compiling
      > the kernel than when compiling anything else? Because sometimes I go
      > through several compiles in a row when tinkering with my .config; so
      > far I haven't been doing a "make clean", and I haven't noticed
      > anything bad happen, but I'd like to know if I should be more careful.

      There is one really good reason - if the kernel is badly compiled then it is
      much worse than if mpg123 is badly compiled. It is a much larger and more
      complex beast than most programs, and unlike many normal programs, you are
      actually only going to compile in a very small subset of the whole thing. If
      the dependancies in the hundreds of makefiles in the kernel are all correct
      for all the millions of possible configurations (make menuconfig results)
      then you're OK. But if not then you'll just get weird bugs. Back when I
      hacked at the kernel I was just playing with the ide driver, and then just
      with a single file, but it still caused funny results until I recompiled the
      whole thing.

      Alan
    • Show all 11 messages in this topic