Dynamic analysis using rr

Octav Chipara ochipara at gmail.com
Mon Feb 19 19:39:25 UTC 2018

Dear Rob,

Thank you for the quick reply. I am reading through the technical report. I do realize that the values of variables are not saved. I see two options to obtain them:

(1) I noticed that you are having a gdb session as part of rr replay. Would it be possible to retrieve variables by interacting with the gdb session?
(2) Alternatively, I could use an emulator (e.g., unicorn) to playback the traces after a syscall since they should be deterministic have no I/O.

Do you have any suggestions which approach may be easier?

— Octav

> On Feb 19, 2018, at 1:01 PM, Robert O'Callahan <robert at ocallahan.org> wrote:
> On Tue, Feb 20, 2018 at 5:58 AM, Octav Chipara <ochipara at gmail.com <mailto:ochipara at gmail.com>> wrote:
> I want to build a dynamic analysis tool that leverages rr's record capabilities. Specifically, I would like to playback the traces to determine the value of variables in the program at different points in the execution. To do this, I would like understand what is the format that rr uses to save the data that it collects. Is there documentation describing the storage format?
> rr_trace.capnp contains some useful information. https://arxiv.org/abs/1705.05937 <https://arxiv.org/abs/1705.05937> is very helpful for understanding how rr works and what it records.
> Determining the values of variables during replay would require actually replaying the program rather than just reading the trace file, since the values of variables aren't in the trace file. You may be able to use the ReplaySession API to do this.
> Rob
> -- 
> Su ot deraeppa sah dna Rehtaf eht htiw saw hcihw, efil lanrete eht uoy ot mialcorp ew dna, ti ot yfitset dna ti nees evah ew; deraeppa efil eht. Efil fo Drow eht gninrecnoc mialcorp ew siht - dehcuot evah sdnah ruo dna ta dekool evah ew hcihw, seye ruo htiw nees evah ew hcihw, draeh evah ew hcihw, gninnigeb eht morf saw hcihw taht.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.mozilla.org/pipermail/rr-dev/attachments/20180219/ca3e42d6/attachment.html>

More information about the rr-dev mailing list