breakpoint setting bug

breakpoint setting bug

[Bruce Korb]

I've got too much stuff on my plate at the moment.
If this is insufficient, I'll try to do a better
test case later.  Sorry.


In the following code fragment, I set a breakpoint at "emit_file_text"
only to have insight/gdb to set it on line 1227 instead of 1219 or 1222.
Oops.

 	1211	static void
 	1212	emit_file_text (input, remaining_size, restore_name,
 	1213	                quoted_restore_name, split_flag)
 	1214	     FILE * input;
 	1215	     off_t remaining_size;
 	1216	     char const * restore_name;
 	1217	     char const * quoted_restore_name;
 	1218	     int * split_flag;
-	1219	{
 	1220	  char buffer[BUFSIZ];
 	1221	
-	1222	  while (fgets (buffer, BUFSIZ, input))
 	1223	    {
 	1224	
 	1225	      /* Output a line and test the length.  */
 	1226	
	1227	      if (!mandatory_prefix_mode

Re: breakpoint setting bug

[Keith Seitz]

On 02/21/2010 09:48 AM, Bruce Korb wrote:
> I've got too much stuff on my plate at the moment.
> If this is insufficient, I'll try to do a better
> test case later.  Sorry.

Yeah, this is definitely not enough information to appropriately assess 
the bug. I used your example code to try to build a test case, but I 
could not reproduce the bug.

> In the following code fragment, I set a breakpoint at "emit_file_text"
> only to have insight/gdb to set it on line 1227 instead of 1219 or 1222.
> Oops.
>
>   	1211	static void
>   	1212	emit_file_text (input, remaining_size, restore_name,
>   	1213	                quoted_restore_name, split_flag)
>   	1214	     FILE * input;
>   	1215	     off_t remaining_size;
>   	1216	     char const * restore_name;
>   	1217	     char const * quoted_restore_name;
>   	1218	     int * split_flag;
> -	1219	{
>   	1220	  char buffer[BUFSIZ];
>   	1221	
> -	1222	  while (fgets (buffer, BUFSIZ, input))
>   	1223	    {
>   	1224	
>   	1225	      /* Output a line and test the length.  */
>   	1226	
> 	1227	      if (!mandatory_prefix_mode

Was this application compiled with optimization? How did you set the 
breakpoint? Clicked on the line in the source window or used the 
console's "break" command? If the former, what happens when you type, 
"break emit_file_text" in the console window? What version of 
gdb/insight? What host and target?

If you could pare this down to a small test case, that would be of 
considerable help.

Keith

Re: breakpoint setting bug

[Bruce Korb]

Hi Keith,

On Sun, Feb 21, 2010 at 10:28 AM, Keith Seitz <keiths@redhat.com> wrote:
> Yeah, this is definitely not enough information to appropriately assess the

Figured.

> bug. I used your example code to try to build a test case, but I could not
> reproduce the bug.

Figured.  :(

>> In the following code fragment, I set a breakpoint at "emit_file_text"
>> only to have insight/gdb to set it on line 1227 instead of 1219 or 1222.
>> Oops.

> Was this application compiled with optimization?

sharutils compiles thus:
> gcc -DLOCALEDIR=\"/old-home/gnu/proj/sharutils-inst/share/locale\" \
> -DHAVE_CONFIG_H -I. -I..  -I. -I.. -I../lib -I../intl   -g -Wall -MT shar.o \
> -MD -MP -MF .deps/shar.Tpo -c -o shar.o shar.c
though I had the same symptom when adding "-O0" to the command.

> How did you set the breakpoint?
In the command window:
> gdb> b emit_file_text

> What version of gdb/insight?
$ insight --version
GNU gdb 6.8
Copyright (C) 2008 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.  Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-unknown-linux-gnu"
> What host and target?
host == target

> If you could pare this down to a small test case, that would be of
> considerable help.

I'm sure.  That's the part that will take me a little time to get to....
Sorry to just tease you.  ;)
If nothing pops up as obvious, I'll try to get to it in a week or so.

Thank you!  Regards, Bruce

Re: breakpoint setting bug

[Keith Seitz]

On 02/21/2010 10:40 AM, Bruce Korb wrote:
>> bug. I used your example code to try to build a test case, but I could not
>> reproduce the bug.
>
> Figured.  :(

Isn't that how it always goes? O:-)

>> Was this application compiled with optimization?
>
> sharutils compiles thus:
>> gcc -DLOCALEDIR=\"/old-home/gnu/proj/sharutils-inst/share/locale\" \
>> -DHAVE_CONFIG_H -I. -I..  -I. -I.. -I../lib -I../intl   -g -Wall -MT shar.o \
>> -MD -MP -MF .deps/shar.Tpo -c -o shar.o shar.c
> though I had the same symptom when adding "-O0" to the command.

I don't see anything suspicious in that, so it's definitely a bug.

>> How did you set the breakpoint?
> In the command window:
>> gdb>  b emit_file_text

Okay, that makes life a lot easier: It's a gdb bug. :-)

> GNU gdb 6.8

Ah. 6.8. I don't suppose I could cajole you into trying something a 
little more recent? Gdb is currently on it's 7.1 release, so you're 
using something that is almost three releases behind.

FWIW, I use insight pretty much daily, and I only ever use a snapshot. I 
seldom run into problems.

> I'm sure.  That's the part that will take me a little time to get to....
> Sorry to just tease you.  ;)
> If nothing pops up as obvious, I'll try to get to it in a week or so.

No problem. I'll be here.

Keith