[Bug gdb/27868] New: gdb

[Bug gdb/27868] New: gdb

[postix at posteo dot eu]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

            Bug ID: 27868
           Summary: gdb
           Product: gdb
           Version: 10.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: gdb
          Assignee: unassigned at sourceware dot org
          Reporter: postix at posteo dot eu
  Target Milestone: ---

Created attachment 13447
  --> https://sourceware.org/bugzilla/attachment.cgi?id=13447&action=edit
Output of coredumpctl info

When I tried to debug an app, gdb terminated with a SEGFAULT:

```
gdb /usr/bin/kdeinit5 kdeinit5.coredump
GNU gdb (GDB; openSUSE Tumbleweed) 10.1
(...)
Reading symbols from /usr/bin/kdeinit5...
Reading symbols from
/home/user/.cache/debuginfod_client/a412b812dc18d755bbb11a1880295f6b0b5b54dd/debuginfo...
[New LWP 2961]
[New LWP 2962]

warning: Section `.reg-xstate/2961' in core file too small.
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
Core was generated by `filenamesearch.so [kdeinit5]'.
Program terminated with signal SIGSEGV, Segmentation fault.

warning: Section `.reg-xstate/2961' in core file too small.
Aborted (core dumped)
```

Regarding "core file too small", "ulimit -c" says "unlimited".

Please find "coredumpctrl info" attached, I hope it's sufficient. If you need
any other info please let me know how I can help. 
I could also provide the whole "coredumpctrl dump" file, however with its ~43
MB filesize it's too big to attach it here.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb terminated with a SEGFAULT

[postix at posteo dot eu]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

postix at posteo dot eu changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|gdb                         |gdb terminated with a
                   |                            |SEGFAULT

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb terminated with a SEGFAULT

[postix at posteo dot eu]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

postix at posteo dot eu changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |postix at posteo dot eu

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb debug session: Aborted (core dumped)

[postix at posteo dot eu]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

Postix <postix at posteo dot eu> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|gdb terminated with a       |gdb debug session: Aborted
                   |SEGFAULT                    |(core dumped)

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb debug session: Aborted (core dumped)

[simark at simark dot ca]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

Simon Marchi <simark at simark dot ca> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |simark at simark dot ca

--- Comment #1 from Simon Marchi <simark at simark dot ca> ---
Hi,

What would be more useful would be to provide the binary and core you were
trying to load, that made GDB crash.

Simon

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb debug session: Aborted (core dumped)

[postix at posteo dot eu]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

--- Comment #2 from Postix <postix at posteo dot eu> ---
Created attachment 13448
  --> https://sourceware.org/bugzilla/attachment.cgi?id=13448&action=edit
Binary and coredump

> What would be more useful would be to provide the binary and core you were
> trying to load, that made GDB crash.

Excellent idea! Please find both attached.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb debug session: Aborted (core dumped)

[postix at posteo dot eu]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

--- Comment #3 from MH <postix at posteo dot eu> ---
Created attachment 13449
  --> https://sourceware.org/bugzilla/attachment.cgi?id=13449&action=edit
stracktrace (for gdb terminated by signal ABRT.)

Finally I could get a stracktrace via "coredumpctrl debug" from gdb. Please
find it attached.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb debug session: Aborted (core dumped)

[simark at simark dot ca]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

--- Comment #4 from Simon Marchi <simark at simark dot ca> ---
When I load your code on my machine, I do get the

  warning: Section `.reg-xstate/30883' in core file too small.

warning, but GDB doesn't crash.  It's perhaps related to the fact that I don't
have your whole sysroot, so GDB doesn't load any library.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb debug session: Aborted (core dumped)

[simark at simark dot ca]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

--- Comment #5 from Simon Marchi <simark at simark dot ca> ---
(In reply to MH from comment #3)
> Created attachment 13449 [details]
> stracktrace (for gdb terminated by signal ABRT.)
> 
> Finally I could get a stracktrace via "coredumpctrl debug" from gdb. Please
> find it attached.

Ah, thanks.  We see that it's while reading the DWARF debug info of some object
file, but we don't know which one.  The stack trace just cuts there?  Is this
because of a limit of systemd-coredump?  It would be useful to know which
object file is being read.  This way, you could attach it and we could see if
we can reproduce by reading just that file.

An alternative would be to bundle all the libraries mentioned by "info shared"
and attach that.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb debug session: Aborted (core dumped)

[postix at posteo dot eu]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

--- Comment #6 from MH <postix at posteo dot eu> ---
Created attachment 13452
  --> https://sourceware.org/bugzilla/attachment.cgi?id=13452&action=edit
"info shared" libraries for coredump of /usr/bin/gdb (Comment 3)

First of all, the stacktrace mentioned and attached in Comment 3 does not
belong to the core dump of kdeinit5, but to the coredump of gdb itself. I'm
sorry that I have confused it yesterday. However, let's go on with the case of
gdb as we have more information here.

(In reply to Simon Marchi from comment #5)
> The stack trace just cuts there? 
Yes.

> Is this because of a limit of systemd-coredump? 
I'm not sure. ulimit -c and ulimit -f both say "unlimited". There's also plenty
of disk space. I don't know where a limit would come from.


> It would be useful to know which object file is being read.
> This way, you could attach it and we could
> see if we can reproduce by reading just that file.
How can I figure that out, which object is read?

> An alternative would be to bundle all the libraries mentioned by "info
> shared" and attach that.
Please see the archive attached.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb debug session: Aborted (core dumped)

[postix at posteo dot eu]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

--- Comment #7 from MH <postix at posteo dot eu> ---
Created attachment 13453
  --> https://sourceware.org/bugzilla/attachment.cgi?id=13453&action=edit
/usr/bin/gdb (10.1 OpenSUSE TW)

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb debug session: Aborted (core dumped)

[postix at posteo dot eu]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

--- Comment #8 from MH <postix at posteo dot eu> ---
As there core dump of 

Sun 2021-05-16 12:24:48 CEST   8465  1000   100   6 present   /usr/bin/gdb

is too big (even compressed), I have uploaded it to a cloud. [1] 
You access it with "gdb27868" (without the ")

I can trigger the crash by
> gdb /usr/bin/gdb gdb.dump
> bt

[1] https://rwth-aachen.sciebo.de/s/GBAQS0kK51LRwyd

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb debug session: Aborted (core dumped)

[simark at simark dot ca]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

--- Comment #9 from Simon Marchi <simark at simark dot ca> ---
(In reply to MH from comment #6)
> Created attachment 13452 [details]
> "info shared" libraries for coredump of /usr/bin/gdb (Comment 3)
> 
> First of all, the stacktrace mentioned and attached in Comment 3 does not
> belong to the core dump of kdeinit5, but to the coredump of gdb itself.

Yes, that was clear.

> > It would be useful to know which object file is being read.
> > This way, you could attach it and we could
> > see if we can reproduce by reading just that file.
> How can I figure that out, which object is read?

One way would be to do "set debug dwarf-die 1" before reproducing.  It will
spit out a bunch of info about what is being read, we can figure out which
library is read at the moment of the crash.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb debug session: Aborted (core dumped)

[simark at simark dot ca]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

--- Comment #10 from Simon Marchi <simark at simark dot ca> ---
(In reply to MH from comment #8)
> As there core dump of 
> 
> Sun 2021-05-16 12:24:48 CEST   8465  1000   100   6 present   /usr/bin/gdb
> 
> is too big (even compressed), I have uploaded it to a cloud. [1] 
> You access it with "gdb27868" (without the ")
> 
> I can trigger the crash by
> > gdb /usr/bin/gdb gdb.dump
> > bt
> 
> [1] https://rwth-aachen.sciebo.de/s/GBAQS0kK51LRwyd

Thanks!

However, I think that your binaries don't have embedded DWARF debug info.  The
distributions typically distribute debug info as separate files, usually in
/usr/lib/debug.  You can probably reproduce the crash because GDB on your
system finds them, but GDB on my system doesn't (so doesn't read any DWARF).

However, I noticed you mentioned earlier you were on OpenSuse Tumbleweed.  So I
started a container with that, installed kinit and its debug info, and was able
to reproduce:

# gdb -q /usr/bin/kdeinit5 -readnow
Reading symbols from /usr/bin/kdeinit5...
Reading symbols from /usr/lib/debug/usr/bin/kdeinit5-5.82.0-1.1.x86_64.debug...
Expanding full symbols from
/usr/lib/debug/usr/bin/kdeinit5-5.82.0-1.1.x86_64.debug...
Aborted (core dumped)

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb debug session: Aborted (core dumped)

[simark at simark dot ca]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

--- Comment #11 from Simon Marchi <simark at simark dot ca> ---
(In reply to Simon Marchi from comment #10)
> # gdb -q /usr/bin/kdeinit5 -readnow
> Reading symbols from /usr/bin/kdeinit5...
> Reading symbols from
> /usr/lib/debug/usr/bin/kdeinit5-5.82.0-1.1.x86_64.debug...
> Expanding full symbols from
> /usr/lib/debug/usr/bin/kdeinit5-5.82.0-1.1.x86_64.debug...
> Aborted (core dumped)

Damn, I can't reproduce it anymore, I don't know what changed :(

# gdb -q /usr/bin/kdeinit5 -readnow
Reading symbols from /usr/bin/kdeinit5...
Reading symbols from /usr/lib/debug/usr/bin/kdeinit5-5.82.0-1.1.x86_64.debug...
Expanding full symbols from
/usr/lib/debug/usr/bin/kdeinit5-5.82.0-1.1.x86_64.debug...
Expanding full symbols from /usr/bin/kdeinit5...
(gdb)

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb debug session: Aborted (core dumped)

[postix at posteo dot eu]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

--- Comment #12 from MH <postix at posteo dot eu> ---
(In reply to Simon Marchi from comment #9)
> One way would be to do "set debug dwarf-die 1" before reproducing.  It will
> spit out a bunch of info about what is being read, we can figure out which
> library is read at the moment of the crash.

I have done that and put both, the output (gdb_dwarf.txt) and the corresponding
dwarf file (debuginfo) to the cloud space.

> Damn, I can't reproduce it anymore, I don't know what changed :(
Weird, I still can and my system is up to date (20210513). Hopefully the dwarf
mentioned before will help us further.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb debug session: Aborted (core dumped)

[postix at posteo dot eu]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

--- Comment #13 from MH <postix at posteo dot eu> ---
> Damn, I can't reproduce it anymore, I don't know what changed :(

With the latest update today I can no longer reproduce it myself. Not sure if
we shall be happy or sad about it. ;)

Does the debuginfo file and gdb_dwarf.txt output save the day to find the bug
nevertheless?

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb debug session: Aborted (core dumped)

[simark at simark dot ca]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

--- Comment #14 from Simon Marchi <simark at simark dot ca> ---
(In reply to MH from comment #13)
> > Damn, I can't reproduce it anymore, I don't know what changed :(
> 
> With the latest update today I can no longer reproduce it myself. Not sure
> if we shall be happy or sad about it. ;)
> 
> Does the debuginfo file and gdb_dwarf.txt output save the day to find the
> bug nevertheless?

I'm confused once again.  Did you observe gdb crashing while debugging gdb
itself, or while debugging kdeinit?

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb debug session: Aborted (core dumped)

[postix at posteo dot eu]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

--- Comment #15 from MH <postix at posteo dot eu> ---
(In reply to Simon Marchi from comment #14)
> (In reply to MH from comment #13)
> > > Damn, I can't reproduce it anymore, I don't know what changed :(
> > 
> > With the latest update today I can no longer reproduce it myself. Not sure
> > if we shall be happy or sad about it. ;)
> > 
> > Does the debuginfo file and gdb_dwarf.txt output save the day to find the
> > bug nevertheless?
> 
> I'm confused once again.  Did you observe gdb crashing while debugging gdb
> itself, or while debugging kdeinit?

Both, but beginning with comment 3, I only had focused on gdb crashing while
debugging gdb itself.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb debug session: Aborted (core dumped)

[postix at posteo dot eu]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

postix <postix at posteo dot eu> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |WORKSFORME
             Status|UNCONFIRMED                 |RESOLVED

--- Comment #16 from postix <postix at posteo dot eu> ---
Let's close it.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27868] gdb debug session: Aborted (core dumped)

[simark at simark dot ca]

https://sourceware.org/bugzilla/show_bug.cgi?id=27868

Simon Marchi <simark at simark dot ca> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|WORKSFORME                  |INVALID

-- 
You are receiving this mail because:
You are on the CC list for the bug.