|Debug optimized code linuxkaffee_@_gmx.net (Stephan Ceram) (2008-09-16)|
|Re: Debug optimized code firstname.lastname@example.org (RR) (2008-09-17)|
|Re: Debug optimized code email@example.com (cr88192) (2008-09-17)|
|Re: Debug optimized code firstname.lastname@example.org (Walter Banks) (2008-09-17)|
|Re: Debug optimized code email@example.com (rcmetzger) (2008-09-17)|
|Re: Debug optimized code firstname.lastname@example.org (Stephen Horne) (2008-09-18)|
|Re: Debug optimized code email@example.com (Barry Kelly) (2008-09-19)|
|Re: Debug optimized code firstname.lastname@example.org (George Neuner) (2008-09-20)|
|[6 later articles]|
|Date:||Wed, 17 Sep 2008 18:24:33 +0800|
|Posted-Date:||17 Sep 2008 07:52:57 EDT|
You can have a look at
"This white paper provides an outline on the concept of debugging
optimized code. This paper discusses the way compiler transformations
affect debugging, the practical problems in debugging optimizing code
and techniques that can be used to overcome the problems in debugging
There is a lot of activity currently on this subject. But after all
this, no usable product has evolved till date to debug optimized code.
This is due to the practical problems involved in debugging optimized
code. This white paper analyzes the practical problems in debugging
optimized code and presents a clear interface between a compiler and
debugger in achieving debugging of optimized code."
On Wed, Sep 17, 2008 at 5:37 AM, Stephan Ceram <linuxkaffee_@_gmx.net> wrote:
I've a question about compilers' symbolic debugging information (like
DWARF2) and compiler optimizations. Should the code be always compiled
without any optimizations, i.e. -O0, to get accurate debug information
where all source code line number are correct or would I also get accurate
results for higher optimization levels? Usually, at least -O1 does
not modify the code too much, so I would assume that this optimization
level could be applied for a successful debugging afterwards.
What are you experiences?
Return to the
Search the comp.compilers archives again.