GDB threading/FP bug

[bennet]

There's a floating point bug in GDB that occurs when debugging a
program with thread libraries linked in.  I've seen it occur with gcc
3.0.4/gdb5.0 3.1.1/gdb 5.2.1 and I believe also gcc 3.0.2/gdb
19991004.  As shown below, the first FP operation that GDB does is
turned into a NAN.  Subsequent operations succeed.  The non-gdb
execution is listed for comparison.  Anyone know if there's a few in
the works for this?  (I tried to search the newsgroup for relevant
reports but its hard to see through all the spam.)

-----------example test.cpp--------------------
#include <stdio.h>

void foo() {
    double bar = (double)0;
    printf("<< %f >>\n", bar);
}

int main() {
    foo();
    foo();
    return 0;
}
-----------example test.cpp--------------------

$ gdb -v
GNU gdb 5.2.1

$ gcc -v
Reading specs from /usr/local/lib/gcc-lib/i686-pc-linux-gnu/3.1.1/specs
Configured with: ./configure --enable-threads=posix
Thread model: posix
gcc version 3.1.1

$ echo run > gdbcmd && c++ -lpthread -o test test.cpp && gdb -x gdbcmd
-batch test && ./test
[New Thread 1024 (runnable)]
<< nan >>
<< 0.000000 >>

Program exited normally.
<< 0.000000 >>
<< 0.000000 >>