This is the mail archive of the gdb@sources.redhat.com mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

bogus bkpt, using #line directives


I'm using gdb (6.1.1) to debug applications that are compiled from C++ code which is, in turn, generated from special higher-level code (whose source files have ".mmm" extensions).

I've inserted "#line" directives at each .cpp line that claim that that line is really from the .mmm file it was generated from.

Here's a snippet of code from P.cpp to talk about:

#line 39 "/tmp/allenh/metro/examples/mgdb/P.mmm"
              try {
#line 41 "/tmp/allenh/metro/examples/mgdb/P.mmm"
                port1->writeInt(this, w);
#line 42 "/tmp/allenh/metro/examples/mgdb/P.mmm"
                w = w + 1;
#line 42 "/tmp/allenh/metro/examples/mgdb/P.mmm"
              } catch (STUCK) {
> ...

A breakpoint set at line 42 ends up in a bogus place:

(gdb) b P.mmm:41
Breakpoint 6 at 0x805edde: file P.mmm, line 41.
(gdb) b P.mmm:42
Breakpoint 7 at 0x804d54a: file P.mmm, line 42.
(gdb) info b
Num Type Disp Enb Address What
6 breakpoint keep y 0x0805edde in P::thread() at P.mmm:41
7 breakpoint keep y 0x0804d54a in ~node at P.mmm:42


The breakpoint at "~node" is just plain wrong. The breakpoint is set correctly if the "#line" directives are removed and the breakpoints are set on the actual .cpp line numbers.

The program is very complex & I haven't yet been able to make a simple test case to demonstrate this. I'm sort of desperately hoping this looks familiar to somebody who can offer me some hope, or suggestions. I'm left wondering if I have to debug gdb, but I'd be in really unfamiliar territory.

-Allen


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]