Re: less dwarf2 debug info stored in .o on ia64

[Rens Sankar <rens dot sankar at wipro dot com>]

Hi Jim,

We are working on the enhancement of the tool lcrash which can analyse a
linux kernel crash dump with the help of a .o file which has most of the 
information on the kernel data structures. This .o is created
at the time of compilation of the kernel.
There are commands within this utility to view the kernel data
structures. We are implementing dwarf2 support for this tool.
And with the .o file having dwarf2 info, we are unable to display the
info
on the unnamed padding fields. This leads to difference in the
declaration
of the structure in the header file and displayed by the tool.

We tried going through gcc source code but could not get much headway in 
dwarf2out.c. If you could kindly point to the general direction 
(e.g. function names etc.) where this decision is taken, we can try to 
study the code more.
 
Thanks and regards,
Rens.

Jim Wilson wrote:
> 
> On Fri, 2004-01-09 at 04:41, Rens Sankar wrote:
> > **here we can notice the information lose of "int :32", with dwarf-2 it
> > is the same case on i386 also... but stabs on i386 has "int :32" info.
> 
> Why do you need debug info for unnamed padding fields?  The user can't
> see them, and we don't need info on them in order to compute the
> location of following fields, so including them seems rather pointless.
> Though Devang does have a good point about the gdb ptype command, the
> output isn't quite right if we don't emit these unnamed fields.
> 
> As for the why, we have been ignoring nameless fields since the original
> version of the dwarf2out.c file was added in 1996.  That was based on
> the dwarf v1 dwarfout.c, which does the same thing, apparently since it
> was contributed in 1990.  So we have been doing it this way for over 13
> years now for dwarf debug info.
> --
> Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com