The main goal is to make the call graph of a huge program like cc1 (20k functions).
For timing measurements, I have tried:
nm cc1 | awk '$1 == "T" {print "break *0x" $2 }' > t.gdb
gdb -x t.gdb
run
Sourcing t.gdb took around 2 minutes, but seemed to set all breakpoints.
But running run seemed took a very long time.
After it regains control, continue is responsive, but doing:
while 1
continue
bt
end
runs very slowly.