Hi,
These test cases are to expose the problems I mentioned in patch #1. 
Without patch #1 applied, there are three new failures,

+FAIL: gdb.base/disp-step-fork.exp: single step over fork (timeout)
+FAIL: gdb.base/disp-step-vfork.exp: continue to marker (the program is 
no longer running)
+FAIL: gdb.base/disp-step-vfork.exp: print global

After patch #1 applied, these three failures are fixed,

+PASS: gdb.base/disp-step-fork.exp: single step over fork
+PASS: gdb.base/disp-step-vfork.exp: continue to marker
+PASS: gdb.base/disp-step-vfork.exp: print global

In short, these two test cases do something similar.  First, they will 
look for syscall instruction in vfork () or fork (), and set breakpoint 
on it.  Then, continue to syscall insn, turn displaced stepping on, and 
step over that breakpoint.

Regression tested on x86_64-unknow-linux-gnu.

-- 
Yao (齐尧)