Bug in MI command -data-dissasemble ??

Bug in MI command -data-dissasemble ??

[Carlos Eduardo Rodrigues de Almeida]

Hi all,
I was trying the -data-disassemble command and I found something strange.
I think with a session I can explain better:


[eduardo@~/lab3/Debug]$gdb -i=mi2
~"GNU gdb 6.4\n"
~"Copyright 2005 Free Software Foundation, Inc.\n"
~"GDB is free software, covered by the GNU General Public License, and
you are\n"
~"welcome to change it and/or distribute copies of it under certain
conditions.\n"
~"Type \"show copying\" to see the conditions.\n"
~"There is absolutely no warranty for GDB.  Type \"show warranty\" for
details.\n"
~"This GDB was configured as \"x86_64-pc-linux-gnu\"."
~"\n"
(gdb)
-file-exec-and-symbols lab3
~"Using host libthread_db library \"/lib/libthread_db.so.1\".\n"
^done
(gdb)
-data-disassemble -f /home/eduardo/libgip/src/gip_app.cpp -l 68 -n 1 -- 1
&"mi_cmd_disassemble: Invalid line number\n"
^error,msg="mi_cmd_disassemble: Invalid line number"
(gdb)
-data-disassemble -f gip_app.cpp -l 68 -n 1 -- 1
^done,asm_insns=[{address="0x000000000040d7e0",func-name="_Z12gip_app_initPc",offset="0",inst="mov
   %rbx,0xfffffffffffffff0(%rsp)"}]
(gdb)
-data-disassemble -f /home/eduardo/libgip/src/gip_app1.cpp -l 68 -n 1 -- 1
&"mi_cmd_disassemble: Invalid filename.\n"
^error,msg="mi_cmd_disassemble: Invalid filename."
(gdb)


The gip_app.cpp is a file of a library I used to compile lab3.
When I use the complete path (/home/eduardo/libgip/src/gip_app.cpp)
GDB gives me error messge invalid LINE number. When I use only the
file name(gip_app.cpp) the disassemble works. The path is right
because when I specify a wrong file GDB gives me another
message(Invalid filename)

I think this is a bug.. at least the message is wrong.. the line
number is valid.

Thank you,
Eduardo

Bug in MI command -data-dissasemble ??

[Nick Roberts]

 ...
 > -data-disassemble -f /home/eduardo/libgip/src/gip_app.cpp -l 68 -n 1 -- 1
 > &"mi_cmd_disassemble: Invalid line number\n"
 > ^error,msg="mi_cmd_disassemble: Invalid line number"
 > (gdb)
 > -data-disassemble -f gip_app.cpp -l 68 -n 1 -- 1
 > ^done,asm_insns=[{address="0x000000000040d7e0",func-name="_Z12gip_app_initPc",offset="0",inst="mov
 >    %rbx,0xfffffffffffffff0(%rsp)"}]
 > (gdb)
 > -data-disassemble -f /home/eduardo/libgip/src/gip_app1.cpp -l 68 -n 1 -- 1
 > &"mi_cmd_disassemble: Invalid filename.\n"
 > ^error,msg="mi_cmd_disassemble: Invalid filename."
 > (gdb)
 > 
 > 
 > The gip_app.cpp is a file of a library I used to compile lab3.
 > When I use the complete path (/home/eduardo/libgip/src/gip_app.cpp)
 > GDB gives me error messge invalid LINE number. When I use only the
 > file name(gip_app.cpp) the disassemble works. The path is right
 > because when I specify a wrong file GDB gives me another
 > message(Invalid filename)
 > 
 > I think this is a bug.. at least the message is wrong.. the line
 > number is valid.

It is also odd that you specified mixed source and disassembly (--1) but got
only disassembly in the output (--0).  Perhaps it has something to with 64 bit
opeartion.  It seems to work fine for me with i686-pc-linux-gnu.

Incidently, for Emacs I currently just use the CLI command "disassemble".


-- 
Nick                                           http://www.inet.net.nz/~nickrob

Re: Bug in MI command -data-dissasemble ??

[Carlos Eduardo Rodrigues de Almeida]

Hi all,
I'll try to figure out what is going on.. Is there some document that
describes the organization of the GDB code?? Is there some place where
I can grab more information?

Thank you.

Eduardo
On 3/1/06, Carlos Eduardo Rodrigues de Almeida
<eduardo.almeida@gmail.com> wrote:
> Hi all,
> I was trying the -data-disassemble command and I found something strange.
> I think with a session I can explain better:
>
>
> [eduardo@~/lab3/Debug]$gdb -i=mi2
> ~"GNU gdb 6.4\n"
> ~"Copyright 2005 Free Software Foundation, Inc.\n"
> ~"GDB is free software, covered by the GNU General Public License, and
> you are\n"
> ~"welcome to change it and/or distribute copies of it under certain
> conditions.\n"
> ~"Type \"show copying\" to see the conditions.\n"
> ~"There is absolutely no warranty for GDB.  Type \"show warranty\" for
> details.\n"
> ~"This GDB was configured as \"x86_64-pc-linux-gnu\"."
> ~"\n"
> (gdb)
> -file-exec-and-symbols lab3
> ~"Using host libthread_db library \"/lib/libthread_db.so.1\".\n"
> ^done
> (gdb)
> -data-disassemble -f /home/eduardo/libgip/src/gip_app.cpp -l 68 -n 1 -- 1
> &"mi_cmd_disassemble: Invalid line number\n"
> ^error,msg="mi_cmd_disassemble: Invalid line number"
> (gdb)
> -data-disassemble -f gip_app.cpp -l 68 -n 1 -- 1
> ^done,asm_insns=[{address="0x000000000040d7e0",func-name="_Z12gip_app_initPc",offset="0",inst="mov
>   %rbx,0xfffffffffffffff0(%rsp)"}]
> (gdb)
> -data-disassemble -f /home/eduardo/libgip/src/gip_app1.cpp -l 68 -n 1 -- 1
> &"mi_cmd_disassemble: Invalid filename.\n"
> ^error,msg="mi_cmd_disassemble: Invalid filename."
> (gdb)
>
>
> The gip_app.cpp is a file of a library I used to compile lab3.
> When I use the complete path (/home/eduardo/libgip/src/gip_app.cpp)
> GDB gives me error messge invalid LINE number. When I use only the
> file name(gip_app.cpp) the disassemble works. The path is right
> because when I specify a wrong file GDB gives me another
> message(Invalid filename)
>
> I think this is a bug.. at least the message is wrong.. the line
> number is valid.
>
> Thank you,
> Eduardo
>

Re: Bug in MI command -data-dissasemble ??

[Daniel Jacobowitz]

On Fri, Mar 03, 2006 at 02:21:09PM -0300, Carlos Eduardo Rodrigues de Almeida wrote:
> Hi all,
> I'll try to figure out what is going on.. Is there some document that
> describes the organization of the GDB code?? Is there some place where
> I can grab more information?

There's the GDB internals manual (gdbint.texinfo, format it as info or
pdf or whatever's convenient for you); but, I don't think it has much
to say about MI issues.

-- 
Daniel Jacobowitz
CodeSourcery

Re: Bug in MI command -data-dissasemble ??

[Carlos Eduardo Rodrigues de Almeida]

Well... I got the current (6.4) gdb sources and compiled them.
The problem does not happen in the current.
Should I contact someone at gentoo.org?
Eduardo

On 3/3/06, Daniel Jacobowitz <drow@false.org> wrote:
> On Fri, Mar 03, 2006 at 02:21:09PM -0300, Carlos Eduardo Rodrigues de Almeida wrote:
> > Hi all,
> > I'll try to figure out what is going on.. Is there some document that
> > describes the organization of the GDB code?? Is there some place where
> > I can grab more information?
>
> There's the GDB internals manual (gdbint.texinfo, format it as info or
> pdf or whatever's convenient for you); but, I don't think it has much
> to say about MI issues.
>
> --
> Daniel Jacobowitz
> CodeSourcery
>

Re: Bug in MI command -data-dissasemble ??

[Daniel Jacobowitz]

On Fri, Mar 03, 2006 at 07:39:01PM +0000, Carlos Eduardo Rodrigues de Almeida wrote:
> Well... I got the current (6.4) gdb sources and compiled them.
> The problem does not happen in the current.
> Should I contact someone at gentoo.org?

Yes, probably.

-- 
Daniel Jacobowitz
CodeSourcery

Re: Bug in MI command -data-dissasemble ??

[Nick Roberts]

Daniel Jacobowitz writes:
 > On Fri, Mar 03, 2006 at 02:21:09PM -0300, Carlos Eduardo Rodrigues de Almeida wrote:
 > > Hi all,
 > > I'll try to figure out what is going on.. Is there some document that
 > > describes the organization of the GDB code?? Is there some place where
 > > I can grab more information?
 > 
 > There's the GDB internals manual (gdbint.texinfo, format it as info or
 > pdf or whatever's convenient for you); but, I don't think it has much
 > to say about MI issues.

But I suspect (possibly ignorantly) that the underlying problem is to do with
symbol table information, which the internals manual does describe.  An easy
way to test this might be to look at the output of of the CLI command
disassemble.  Anyway it looks like its not present in current CVS.  Perhaps 64
bit operation is a recent feature.


-- 
Nick                                           http://www.inet.net.nz/~nickrob