Messages in this thread | | | Date | Sun, 27 Apr 2014 15:13:19 +0200 | From | Oleg Nesterov <> | Subject | Re: [RFC PATCH] uprobes: simplify rip-relative handling |
| |
On 04/27, Oleg Nesterov wrote: > > On 04/26, Denys Vlasenko wrote: > > > > @@ -46,7 +46,7 @@ struct arch_uprobe { > > > > union { > > #ifdef CONFIG_X86_64 > > - unsigned long rip_rela_target_address; > > + int insn_length; > > #endif > > in particular, we already have auprobe->def.ilen this patch can use.
And I forgot to mention...
I can easily misunderstand the change in handle_riprel_insn(), but it seems that since you removed memmove(cursor, ...) the changed insn has the same lenghth as the original insn?
If yes,
- the patch is wrong, it breaks the "correction" logic
- but, at the same time, we can simplify this logic and just kill the "long *correction" arg of riprel_post_xol.
Oleg.
| |