This is the mail archive of the
ecos-discuss@sources.redhat.com
mailing list for the eCos project.
Re: Is this the compiler or debugger issue?
- To: lingsu at palmmicro dot com
- Subject: Re: [ECOS] Is this the compiler or debugger issue?
- From: Bart Veer <bartv at redhat dot com>
- Date: Thu, 10 Aug 2000 13:05:11 +0100
- CC: ecos-discuss at sources dot redhat dot com
- References: <024701c00253$5d0fc6b0$0201a8c0@raccoon>
- Reply-to: bartv at redhat dot com
>>>>> "Ling" == Ling Su <lingsu@palmmicro.com> writes:
Ling> Hello, everyone,
Ling> I experienced some problems to run the mips64vr4300-elf
Ling> toolchain on eCos, hope anyone can kindly help me. I used
Ling> binutils2.0, egcs-2000-5-30 and insight 5.0. I successfully
Ling> download the code, but even I just try to set "break" at
Ling> main, it complains, "Cannot access memory at address 0x0". I
Ling> doubt if I have some setting problems.
Ling> The sample session to run a hello.exe in ecos example
Ling> directry is shown in below. The strange thing is, even I try
Ling> to run the hello.exe in simulator target, some error message
Ling> shows "mips-core: 1 byte write to unmapped address
Ling> 0xc1000007 at 0x0". Actually I use mips64vr4300-elf-run to
Ling> run the test example hello.exe, the log is also posted
Ling> below. Thanks for anyone who can show me some light to solve
Ling> this. Thanks a lot!
This may be the same problem as was reported in
http://sources.redhat.com/ml/ecos-discuss/2000-08/msg00007.html and
the subsequent thread. The debug information in the executable is
incorrect, so gdb thinks that main() is at location 0 and tries to set
the breakpoint there - which is not going to work very well.
The problem may be in binutils, gcc, or possibly gdb. I suggest
looking in the mailing lists for those projects in that order and see
if anything relevant is being discussed. According to Chris Morrow
removing -gc-sections from the link line made the problem go away, so
that might be a temporary workaround.
Bart Veer // eCos net maintainer