This is the mail archive of the gdb-patches@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[commit/Ada] Wrong value of packed array on big endian targets


This one is a fix from Paul were a the debugger was printing the wrong
value for a packed array. I will just quote his description of the
problem:

 The value of Fwl1.FWL_MESSAGE in these tests produces incorrect results
 on big-endian machines, because we incorrectly convert integer type moduli
 in the range 2**31 .. 2**32 to moduli in the range 2**64-2**31 .. 2**64 and
 we represent this packed array as a 32-bit unsigned integer.

The problem was exposed as part of a pretty complex set of sources,
so I don't have a small reproducer handy.  Let me know if the above
is not detailed enough, and I will dig further.

2008-09-30  Paul Hilfinger  <hilfinger@adacore.com>

        * ada-lang.c (ada_modulus): Correct to avoid sign problem with
        moduli >= 2**31.

Tested on x86-linux. No regression.
Checked in.

-- 
Joel

Attachment: moduli.diff
Description: Text document


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]