bug#40271: 26.3; gdb-mode fails to initialize if the target executable is built without symbols

[Vladimir Nikishkin]

Sorry, I misunderstood you.

lockywolf@delllaptop:~$ gdb -i=mi a.out
=thread-group-added,id="i1"
~"GNU gdb (GDB) 9.1\n"
~"Copyright (C) 2020 Free Software Foundation, Inc.\n"
~"License GPLv3+: GNU GPL version 3 or later
<http://gnu.org/licenses/gpl.html>\nThis is free software: you are
free to change and redistribute it.\nThere is NO WARRANTY, to the
extent permitted by law."
~"\nType \"show copying\" and \"show warranty\" for details.\n"
~"This GDB was configured as \"x86_64-slackware-linux\".\n"
~"Type \"show configuration\" for configuration details.\n"
~"For bug reporting instructions, please see:\n"
~"<http://www.gnu.org/software/gdb/bugs/>.\n"
~"Find the GDB manual and other documentation resources online at:\n
 <http://www.gnu.org/software/gdb/documentation/>."
~"\n\n"
~"For help, type \"help\".\n"
~"Type \"apropos word\" to search for commands related to \"word\"...\n"
=cmd-param-changed,param="logging file",value="/home/lockywolf/gdb-log.log"
=cmd-param-changed,param="print pretty",value="on"
=cmd-param-changed,param="print array-indexes",value="on"
=cmd-param-changed,param="print array",value="on"
=cmd-param-changed,param="print symbol-filename",value="on"
=cmd-param-changed,param="print object",value="on"
=cmd-param-changed,param="print vtbl",value="on"
=cmd-param-changed,param="history save",value="on"
~"Reading symbols from a.out...\n"
(gdb)
-file-list-exec-source-file
^done,line="1",file="../sysdeps/x86_64/start.S",fullname="/root/glibc-2a0c65c5167af50893952729ba38cc68/glibc-2.30/csu/../sysdeps/x86_64/start.S",macro-info="0"
(gdb)

I am confused, because libc is also expected to be stripped. Is this a gdb bug?

Frankly speaking, I just found this by mistake. I forgot to issue
-ggdb when compiling my code. I'm not a reverse engineer.

What to do... well, I am not an expert, but the opposite case, that is
when non-stripped code calls a stripped code function and encounters
and exception, certainly is handled some way. Maybe gdb-mode can say
something like "the most recent known file is claimed (by the code) to
be /root/blah.S, but this file is also missing (just as your debugging
symbols)". GDB itself does something like this:
(gdb) list
1    ../sysdeps/x86_64/start.S: No such file or directory.

In a perfect world it could ask for an external source address, but
perhaps this is an overkill



сб, 28 мар. 2020 г. в 21:12, Eli Zaretskii <eliz@gnu.org>:
>
> [Please keep the bug address on the CC line, so that this discussion
> gets recorded by the Emacs issue tracker.]
>
> > From: Vladimir Nikishkin <lockywolf@gmail.com>
> > Date: Sat, 28 Mar 2020 20:38:50 +0800
> >
> > I am not super sure how to debug this.
> >
> > Firsly, regarding your question:
> >
> > lockywolf@delllaptop:~/DevLinux/chibi-sicp$ gdb -i=mi a.out
> > =thread-group-added,id="i1"
> > ~"GNU gdb (GDB) 9.1\n"
> > ~"Copyright (C) 2020 Free Software Foundation, Inc.\n"
> > ~"License GPLv3+: GNU GPL version 3 or later
> > <http://gnu.org/licenses/gpl.html>\nThis is free software: you are
> > free to change and redistribute it.\nThere is NO WARRANTY, to the
> > extent permitted by law."
> > ~"\nType \"show copying\" and \"show warranty\" for details.\n"
> > ~"This GDB was configured as \"x86_64-slackware-linux\".\n"
> > ~"Type \"show configuration\" for configuration details.\n"
> > ~"For bug reporting instructions, please see:\n"
> > ~"<http://www.gnu.org/software/gdb/bugs/>.\n"
> > ~"Find the GDB manual and other documentation resources online at:\n
> >  <http://www.gnu.org/software/gdb/documentation/>."
> > ~"\n\n"
> > ~"For help, type \"help\".\n"
> > ~"Type \"apropos word\" to search for commands related to \"word\"...\n"
> > =cmd-param-changed,param="logging file",value="/home/lockywolf/gdb-log.log"
> > =cmd-param-changed,param="print pretty",value="on"
> > =cmd-param-changed,param="print array-indexes",value="on"
> > =cmd-param-changed,param="print array",value="on"
> > =cmd-param-changed,param="print symbol-filename",value="on"
> > =cmd-param-changed,param="print object",value="on"
> > =cmd-param-changed,param="print vtbl",value="on"
> > =cmd-param-changed,param="history save",value="on"
> > ~"Reading symbols from a.out...\n"
> > (gdb)
> >
> > Secondly:
> > The place where gdb-mode fail, IMO, is line 4688 in the gdb-mi.el,
> > function gdb-get-source-file
> > I inserted (write-file "/home/lockywolf/debugging-buffer.txt") right
> > after (goto-char (min-point)), and the contents of the file are:
> > lockywolf@delllaptop:~$ cat debugging-buffer.1.txt
> > line="1",file="../sysdeps/x86_64/start.S",fullname="/root/glibc-2a0c65c5167af50893952729ba38cc68/glibc-2.30/csu/../sysdeps/x86_64/start.S",macro-info="0"
> >
> > /root/glibc-* is the place where Slackware developers compile glibc.
> > Doesn't seem like the correct file. The correct file is called
> > "second.f90", and is in the same directory.
>
> That seems to be the problem: in my case I get an error message:
>
>   (gdb)
>   -file-list-exec-source-file
>   ^error,msg="No symbol table is loaded.  Use the \"file\" command."
>
> which is quite expected.
>
> So I guess the problem is that your program is stripped, but the
> standard C library isn't.  What do you suggest Emacs should do in this
> case?  How can it know that the source file it's been fed isn't the
> right one?
>
> More generally, what is the use case where you want to debug a
> stripped program in gdb-mi? why did you strip the program in the first
> place?



-- 
Yours sincerely, Vladimir Nikishkin