[Bug binutils/25803] cross compilation of glibc using mips64el-none-linux-gnu as the host

[nickc at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=25803

Nick Clifton <nickc at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|---                         |FIXED

--- Comment #28 from Nick Clifton <nickc at redhat dot com> ---
(In reply to gagan singh sidhu (gagz, broly, w/e u want) from comment #27)

> if you don’t mind, would you allow me to mail you in the future to learn
> about how exactly you pinpointed the issue?

You can - although there is no need to just email me - you can ask here, or on
the binutils mailing list: address@hidden.

Anyway to answer your question, I basically followed my standard bug
locating/fixing procedure which is:

  1. Reproduce the problem.  [This was actually the hardest step]
  2. Add print statements to the code to find out what is wrong.
     [Other people use debuggers, but I am old school].
  3. Generate possible fixes based on the knowledge of what is wrong.
  4. Test them out.

In this particular case I replaced the assertion that was being triggered with
a series of printf()s telling me about the symbols that were causing the
problems.  I then followed several blind alleys assuming that the symbols
should not have been there and trying to find out why.  Eventually however I
was able to prove to my satisfaction that the higher level code was correct and
that the symbols were indeed dynamic symbols.  So then I looked more closely at
them, saw their similar looking names, and hence created my post about "what is
special about these clock symbols".  Digging deeper I eventually noticed that
they were all IFUNC symbols and then is when the penny dropped.

Cheers
  Nick

-- 
You are receiving this mail because:
You are on the CC list for the bug.