6.7 build breaks on i686

2007-10-16 Thread Ilya N. Golubev
Version: 6.7

Compiler: gcc (GCC) 3.3.2 (Mandrake Linux 10.0 3.3.2-6mdk)

System: i686-pc-linux-gnu

While compiling the following file, `-Werror' is passed to compiler.
Still

gdb-6.7/bfd/elf32-i386.c: In function `elf_i386_tls_transition':
gdb-6.7/bfd/elf32-i386.c:1174: warning: duplicate `const'
gdb-6.7/bfd/elf32-i386.c:1174: warning: duplicate `const'
make[4]: *** [elf32-i386.lo] Error 1

Preprocessed code contains the following.

`bfd.h':

typedef const struct reloc_howto_struct reloc_howto_type;

`elf32-i386.c':

  const reloc_howto_type *from, *to;

`reloc_howto_type' appears to be quite generic, the same in all
builds.  So this error warning should occur whenever compiling any
code like that.


___
bug-gdb mailing list
bug-gdb@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-gdb


(void(**)()) still in 6.6

2006-12-22 Thread Ilya N. Golubev
Version: 6.3, 6.4, 6.4.50.20060307-cvs, 6.6

In 6.6 it is just the same as in previous versions, and as described
on Fri, 31 Mar 2006 22:12:41 +0400 ([EMAIL PROTECTED]).

(gdb) p*(void(**)())$addr
A syntax error in expression, near `*)())$addr'.


___
bug-gdb mailing list
bug-gdb@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-gdb


crash in expr eval

2006-11-12 Thread Ilya N. Golubev
Version: 6.5, no changes

Configuration: all `configure' args are srcdir, installation prefix.

System: i686-pc-linux-gnu

#0  check_typedef (type=0x0) at ../../share/src/gdb-6.5/gdb/gdbtypes.c:1384
#1  0x080ddc99 in coerce_ref (arg=0x84a13b0)
at ../../share/src/gdb-6.5/gdb/value.c:1615
#2  0x080e3456 in value_assign (toval=0x84a13b0, fromval=0x84ca610)
at ../../share/src/gdb-6.5/gdb/valops.c:535
#3  0x080e1603 in evaluate_subexp_standard (expect_type=0x0, exp=0x8dd0b18, 
pos=0xb484, noside=EVAL_NORMAL)
at ../../share/src/gdb-6.5/gdb/eval.c:1486
#4  0x080ddf1d in evaluate_subexp (expect_type=0x0, exp=0x8dd0b18, 
pos=0xb484, noside=EVAL_NORMAL)
at ../../share/src/gdb-6.5/gdb/eval.c:72
#5  0x080de0dd in evaluate_expression (exp=0x8dd0b18)
at ../../share/src/gdb-6.5/gdb/eval.c:162
#6  0x080ec5ff in set_command (
exp=0x8dc14c7 
$reva=Feval(Fcar(Fread_from_string(build_string(\(condition-case nil (princ 
(buffer-substring nil nil (get-buffer \\\*Disassemble*\\\))) (error 
nil))\),Qnil,Qnil))), from_tty=0)
at ../../share/src/gdb-6.5/gdb/printcmd.c:1003

In `coerce_ref' `arg-type' is 0x0.  Have saved core file, will post
further details on request.


___
bug-gdb mailing list
bug-gdb@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-gdb


can not step

2006-11-12 Thread Ilya N. Golubev
Versions: 6.3, 6.5.

System: i686-pc-linux-gnu.  Packages:

kernel-2.6.3.27mdk-1-1mdk
glibc-2.3.3-12.8.100mdk
gcc-3.3.2-6mdk

Built emacs 21.4a with that gcc, `CFLAGS=-g -O2'.  Stopped in
`decode_coding_iso2022', which does no system calls, consists entirely
of cpu use.  All attempts to step further would fail.  `step', `next',
`stepi' would behave exactly like `continue'.

Observed that only on system with set of packages as above, and even
there only with that debugged program.


___
bug-gdb mailing list
bug-gdb@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-gdb


gags on -pie executables

2006-09-17 Thread Ilya N. Golubev
Version: 6.5

System: x86_64-unknown-linux-gnu fedora core 4

Compiler: gcc-4.0.0-8

gdb can not handle one sort of binary executables that are otherwise
ran normally by the system.  They are those linked with `-pie' passed
to linking `cc'.  They may also be detected by `file' reporting
executable to be

ELF 64-bit LSB shared object, AMD x86-64, version 1 (SYSV), not stripped

(Otherwise, it reports them as)

ELF 64-bit LSB executable, AMD x86-64, version 1 (SYSV), for GNU/Linux 2.4.0, 
dynamically linked (uses shared libs), not stripped

Otherwise they are regular executables, having `main' and `_start'
symbols, and having entry point at `_start'.  System runs program
built this way normally.  Program may be built with debugging
information (`-g'), and later assuming that.

`at' gdb command completes normally, but `bt' does not show line
numbers.  Even `i sh' shows no shared libraries, though `ldd' shows
that there should be some.  `b' with line numbers completes normally,
and then `c' outputs:

Continuing.
Warning:
Cannot insert breakpoint 2.
Error accessing memory address 0x6f66e: Input/output error.

All of these commands break the same way even if using `r' instead of
`at'.

Once the program is relinked without `-pie', even from the same object
files, gdb works normally.


Such an executables are actually part of some systems.  They are
abundant in one installed from
http://download.fedoraproject.org/pub/fedora/linux/core/4/x86_64/,
(see
http://fedoraproject.org/wiki/Distribution/Download#head-e82afb53c564fe4d3c977246c5760fad2edfccd9
for other ways to download).  This includes, but not limited to,
`/usr/bin/at', `/usr/bin/dig', `/usr/bin/host', `/usr/bin/omshell',
`/usr/bin/svn'.  (Hope that this is enough.)


___
bug-gdb mailing list
bug-gdb@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-gdb


(void(**)())

2006-04-02 Thread Ilya N. Golubev
Versions: 6.3, 6.4, 6.4.50.20060307-cvs

When current expression language is c, gdb considers the following
type cast an error

(gdb) p*(void(**)())$addr
A syntax error in expression, near `*)())$addr'.

Version 5.1 would process it the same way standard c implementations,
that is, as `(void(*(*))())' cast.


___
Bug-gdb mailing list
Bug-gdb@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-gdb


reading data members: `this' reference

2005-09-30 Thread Ilya N. Golubev
Version: 6.3

On request will compose reproducing instructions with a smaller code
example.  For now, observed that on code of `kdecore' from kdelibs
3.2.

(gdb) p(NETRootInfoPrivate*)(*(void**)($esp)-0x28)
$69 = (NETRootInfoPrivate *) 0x810f5c0
(gdb) p $69-workarea
Variable this is not available.

`workarea' is a (data) field of `NETRootInfoPrivate' structure.
Reading it should require nothing except address of that structure
variable.  But, instead, for some reason gdb tries to use `this'
(pointer to class instance passed to class method that currently
executes) - which is indeed not available at the moment due to some
compiler optimizations.  The worst is that the whole gdb expression
evaluation terminates, not just error / warning is issued.


___
Bug-gdb mailing list
Bug-gdb@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-gdb


`distclean' incomplete

2005-09-30 Thread Ilya N. Golubev
Version: 6.3

`make distclean' leaves the following files in build directory.

intl/config.h
intl/config.status
intl/stamp-h
gdb/doc/gdb.info
gdb/doc/gdb.info-1
gdb/doc/gdb.info-2
gdb/doc/gdb.info-3
gdb/doc/gdb.info-4
gdb/observer.h
gdb/ada-lex.c
gdb/observer.inc
utils/sparclite/Makefile
utils/sparclite/config.status
utils/wince/config.status
utils/wince/Makefile

If a new gdb version is configured to build in the same directory, it
may even break the build, in particular, because files left in build
directory are obsolete wrt new version and gdb makefiles in that new
version fail to notice that and rebuild them.


___
Bug-gdb mailing list
Bug-gdb@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-gdb


Re: never builds gdb.info

2005-06-01 Thread Ilya N. Golubev
 (although personally I think it's a wrong design):

So do I.  In particular, if `gdb.texinfo' or other `gdb.info'
prerequisite was changed, will `make install' install updated
`gdb.info' version?  Putting checking this in my `todo' list.

The worst is that distribution does not even contain `INSTALL' file to
document this, um, irregularity.

 make info to get the documentation built.

Or remove `gdb.info' from src dir, and `make gdb-cfg.texi gdb.info' in
build one.


___
Bug-gdb mailing list
Bug-gdb@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-gdb


never builds gdb.info

2005-05-28 Thread Ilya N. Golubev
Neither
http://sources.redhat.com/cgi-bin/gnatsweb.pl?cmd=createdatabase=gdbreturn_url=http%3A%2F%2Fsources.redhat.com%2Fcgi-bin%2Fgnatsweb.pl%3Fdatabase%3Dgdb%26user%3Dguest%26password%3Dguest%26cmd%3Dquery

referenced by http://www.gnu.org/software/gdb/bugs/ sends
confirmation message about submitted report, nor the submitted report
can be found in the database.  Sending it this way.

Version: 6.3

`make' does not even try to build `gdb.info'.

../share/gdb-6.3/configuremake
rm ../share/gdb-6.3/gdb/doc/gdb.info*
make
ls ../share/gdb-6.3/gdb/doc/gdb.info*

ls: ../share/gdb-6.3/gdb/doc/gdb.info*: No such file or directory


___
Bug-gdb mailing list
Bug-gdb@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-gdb


gdb.info: no gdb-cfg.texi

2005-05-28 Thread Ilya N. Golubev
Version: 6.3

Can not rebuild `gdb.info' even by explicit command.  At least one
required file is neither in distribution nor gets built.

bash ../share/gdb-6.3/configuremake
bash rm ../share/gdb-6.3/gdb/doc/gdb.info*
bash pushd gdb/doc;make ../../../share/gdb-6.3/gdb/doc/gdb.info;popd

makeinfo  ../../../share/gdb-6.3/gdb/doc/gdb.texinfo -o 
../../../share/gdb-6.3/gdb/doc/gdb.info
../../../share/gdb-6.3/gdb/doc/gdb.texinfo:11: @include `gdb-cfg.texi': No such 
file or directory.

Will post the ensuing cascade of warnings and errors on request.

bash find . ../share/gdb-6.3 -type f -name gdb-cfg.texi

outputs nothing.


___
Bug-gdb mailing list
Bug-gdb@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-gdb