Messages in this thread | | | Date | Thu, 10 Apr 2014 22:15:28 +0900 | From | Masami Hiramatsu <> | Subject | Re: [RFC PATCH v2 4/6] uprobes/x86: Emulate rip-relative call's |
| |
(2014/04/10 21:53), Denys Vlasenko wrote: > On 04/09/2014 09:44 PM, Oleg Nesterov wrote: >> +static void ttt_clear_displacement(struct arch_uprobe *auprobe, struct insn *insn) > > Branch instruction's offset isn't called "displacement" > on x86. > > How about ttt_clear_branch_offset?
I like his idea. "displacement" on x86 is so confused especially with using x86 insn decoder.
Thank you,
> >> +{ >> + /* >> + * Turn this insn into "call 1f; 1:", this is what we will execute >> + * out-of-line if ->emulate() fails. >> + * >> + * In the likely case this will lead to arch_uprobe_abort_xol(), but >> + * see the comment in ->emulate(). So we need to ensure that the new >> + * ->ip can't fall into non-canonical area and trigger #GP. >> + * >> + * We could turn it into (say) "pushf", but then we would need to >> + * divorce ->insn[] and ->ixol[]. We need to preserve the 1st byte >> + * of ->insn[] for set_orig_insn(). >> + */ >> + memset(auprobe->insn + insn_offset_immediate(insn), >> + 0, insn->immediate.nbytes); >> +} > >
-- Masami HIRAMATSU Software Platform Research Dept. Linux Technology Center Hitachi, Ltd., Yokohama Research Laboratory E-mail: masami.hiramatsu.pt@hitachi.com
| |