Re: breakpoints/1848: breakpoint (Error accessing memory address)

[Mark Kettenis <kettenis at gnu dot org>]

The following reply was made to PR breakpoints/1848; it has been noted by GNATS.

From: Mark Kettenis <kettenis@gnu.org>
To: jacopo.desimoi@sns.it
Cc: gdb-gnats@sources.redhat.com
Subject: Re: breakpoints/1848: breakpoint  (Error accessing memory address)
Date: Fri, 28 Jan 2005 21:38:43 +0100 (CET)

    From: Jacopo De Simoi <jacopo.desimoi@sns.it>
    Date: Fri, 28 Jan 2005 08:51:53 +0000
 
 [Jacopo, please CC'ing gdb-gnats@sources.redhat.com, that way this
  discussion will be automatically attached to the bug report]
 
    I have some more infos on the programs I am trying to debug
    The executables are probably relocated. Using another debugger
    (totalview demo) there is a fatal error saying
 
    Base executable '' was relocated:
      Linked at 0x00000000, loaded at 0x552aaaa000
 
    Actually, when I tried to debug gdb i found that the difference
    between any address output-ted IN gdb with printf("....."); and the
    address that gdb tries to use is exactly 0x552aaaa000. But as I
    think there should be no direct correspondence between these two
    addresses and the routines bfd_set/get_blalblah should manage this
    difference, is it right?
 
    Maybe this can help in locating the problem in gdb.
 
 I'm suspecting that this is a so position-independent executable
 (PIE).  I thought gdb supported such executables, but apparently there
 are still some problems.  I have no systems that use PIEs.  Can you
 attach a small executable that exhibits the problem to the bug-report?
 I think you can just reply to this message with the executable as a
 MIME attachment.  Hopefully the binary will run on my amd64 Linux
 system.  In any case, please report the problem to the Gentoo people.
 They did a poor job by not verifying that their toolchain completely
 supported PIE.
 
 Mark