Re: The gdb x86 function prologue parser

[Daniel Jacobowitz <drow@false.org>]

On Tue, Jun 14, 2005 at 12:34:55AM +0200, Mark Kettenis wrote:
>    From: Jason Molenda <jmolenda@apple.com>
>    Date: Mon, 13 Jun 2005 15:04:51 -0700
> 
>    Ah!  Now it starts to make sense.  I couldn't understand how this had  
>    been so untested. :)
> 
>    The one part I'm curious about -- does gdb get the CFI information  
>    out of gcc's eh_frame section or something?  How do developers debug  
>    KDE/GNOME applications, where many functions on their stack are from  
>    optimized libraries that don't have any debug info (except maybe  
>    eh_frame)?  It seems like these users should be tripping on these  
>    problems all the time.
> 
> Yup.  We prefer .debug_frame but if that's not available we suck in
> .eh_frame.  So anything that's compiled with -fexceptions (wich
> implies all C++ code) basically has usable CFI.

> I sometimes wonder whether people are using gdb at all...

(me too)

> ...then I find it incredibly stupid that vendors of an Open Source
> operating system ship libraries without debugging information.

More and more vendors are shipping libraries with optional debugging
information.  This is what objcopy --only-keep-debug was invented for. 
For Debian, I also have a couple of hacks to ship unwind information
for some libraries we don't want to provide debug information for by
default (like glibc).

If you install libc6-dbg, backtraces will suddenly Work Better.  No
other intervention required.

For the next release of Debian I hope we'll be using this feature even
more heavily.

-- 
Daniel Jacobowitz
CodeSourcery, LLC