[Bug fortran/18475] New: Crash on a print

[at_gdb at mathalacarte dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=18475

Bug ID: 18475
Summary: Crash on a print
Product: gdb
Version: 7.9
Status: NEW
Severity: normal
Priority: P2
Component: fortran
Assignee: unassigned at sourceware dot org
Reporter: at_gdb at mathalacarte dot com
Target Milestone: ---

While running ddd 3.3.12 with gdb 7.9.1 on a Fortran program I encountered
this:

(gdb) p y(nv1:lrow)
cp-namespace.c:341: internal-error: cp_search_static_and_baseclasses: Assertion
`name[prefix_len + 1] == ':'' failed.
A problem internal to GDB has been detected,
further debugging may prove unreliable.

This is a bug, please report it. For instructions, see:
<http://bugs.gentoo.org/>.

At this point nv1 and lrow can both be printed and nv1=8 and lrow=9
p y(8:9) prints results just fine.

Just tried a similar thing with a different variable having a name ending in 1
and got the same kind of crash. Perhaps things are grouped like <nv><1:...
instead of <nv1><:...?

[vapier at gentoo dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=18475

Mike Frysinger <vapier at gentoo dot org> changed:

What |Removed |Added
----------------------------------------------------------------------------
CC| |vapier at gentoo dot org
Summary|Crash on a print |"p y(nv1:lrow)" hits
| |cp-namespace.c:341:
| |internal-error:
| |cp_search_static_and_basecl
| |asses: Assertion
| |`name[prefix_len + 1] ==
| |':'' failed

--- Comment #1 from Mike Frysinger <vapier at gentoo dot org> ---
can you post a reduced fortran test case ? that helps us track down and
include regression tests.

what target is this (i.e. your $CHOST) ?

[martin.bochud at geoazimut dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=18475

Bochud <martin.bochud at geoazimut dot com> changed:

What |Removed |Added
----------------------------------------------------------------------------
CC| |martin.bochud at geoazimut dot com

[keiths at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=18475

Keith Seitz <keiths at redhat dot com> changed:

What |Removed |Added
----------------------------------------------------------------------------
CC| |keiths at redhat dot com

--- Comment #2 from Keith Seitz <keiths at redhat dot com> ---
(In reply to Fred Krogh from comment #0)

Post by at_gdb at mathalacarte dot com
While running ddd 3.3.12 with gdb 7.9.1 on a Fortran program I encountered
(gdb) p y(nv1:lrow)
Assertion `name[prefix_len + 1] == ':'' failed.
A problem internal to GDB has been detected,
further debugging may prove unreliable.

Have you tried this on HEAD or 7.11.1? The patch for 18303 was committed in
March which removes this assert.

[thrasymache at psenemos dot nu]

https://sourceware.org/bugzilla/show_bug.cgi?id=18475

Sam Newbold <thrasymache at psenemos dot nu> changed:

What |Removed |Added
----------------------------------------------------------------------------
CC| |thrasymache at psenemos dot nu

--- Comment #3 from Sam Newbold <thrasymache at psenemos dot nu> ---
I could also trigger the assertion in gdb 7.9, and gdb 8.2 runs properly. My
source, however, is C++.

My steps to reproduce are:
1) build rwsh, accessible at https://github.com/thrasymache/rwsh
2) gdb ./rwsh
3) break Prototype::arg_to_param
4) run
5) p argm.str()

in 7.9 it produces the above assertion, in 8.2 in produces
$1 = ".init /path/to/rwsh/rwsh"