Re: [patch 2/2] Displaced stepping across fork/vfork : test case

[Pedro Alves <pedro at codesourcery dot com>]

On Thursday 15 September 2011 16:27:04, Yao Qi wrote:
> This is improved by adding another matching in inserting breakpoint on 
> syscall insns like this,

...

> +    set syscall_insn_bp 0
> +    gdb_test_multiple "break \*$syscall_insn_addr"  "break on syscall insn" {
> +       -re "Breakpoint (\[0-9\]*) at .*$gdb_prompt $" {
> +           set syscall_insn_bp $expect_out(1,string)
> +           pass "break on syscall insns"
> +       }
> +       -re "Cannot insert breakpoint .*${gdb_prompt} $" {
> +           # This syscall insn may be in vsyscall page, and we can't set a breakpoint
> +           # there on some systems.
> +           fail "break on syscall insns"

Please don't make this is a fail.  It is not GDB's fault.  "unsupported"
or "xfail" would be better.

> +           set pf_prefix $old_pf_prefix
> +           return -1
> +       }
> +    }

But this won't work as is.  Breakpoints will only be inserted on the
next resume, not immediately.  Unless you enable always-inserted mode,
that is.

-- 
Pedro Alves