Messages in this thread | | | From | Roland McGrath <> | Subject | Re: [RFC Patch 2/9] x86 architecture implementation of Hardware Breakpoint interfaces | Date | Wed, 15 Oct 2008 19:57:28 -0700 (PDT) |
| |
I'm leaving aside the single-step stuff, which I just posted about.
I'm also concerned about getting RF right here. We need to make sure that every place that warps ->ip away somewhere also clears RF. In case of instruction breakpoints in user space, we might also want to mask RF from being seen/changed via ptrace/user_regset (a bit like is done for TF).
I'm a little inclined to do a first x86 version without instruction breakpoint support and hash out the rest of the code for a bit. Then add instruction breakpoints in a later patch. (Not that it has to be a lot later or anything. Just to table the detailed review of RF fiddling until after most of the meaty stuff is pretty baked.)
Thanks, Roland
| |