Reversible debugging
Reversible debugging is the ability to use a debugger to step an application
backward as well as forward. Familiar debugger operations to step
the debuggee line-by-line, into, out-of or over functions calls are agumented
with reverse counterparts: e.g. step back line-by-line or out of a function to
its call-site.
Reversible debugging is a game-changing technology for software development. It
is an unpalatable truth that debugging dominates software development. Not only
do developers spend the vast majority of their time debugging, but the
unpredictability of bugs makes managing software projects a thankless task.
Debugging is like solving a murder mystery: the developer must think backwards
to work out what went wrong. A reversible debugger is to a developer as CCTV
footage of the murder scene is to a detective. Bugs that would have taken
weeks to solve can now be solved in minutes.
Read more on reversible debugging
|
|
UndoDB
UndoDB was the first commercially-available reversible debugger for programs
running "native" code. We defined the technology and UndoDB continues to
represent the state-of-the-art.
UndoDB's "snapshot-and-replay" technique stores periodic copy-on-write snapshots
of the application and only non-deterministic inputs (system calls,
thread-switches, shared memory reads, etc). This means a typical CPU-intensive
application can run with just 1.7x slow-down compared to native execution, and
IO bound applications at close to native speed. Memory requirements are also
comparatively very low.
Almost any application can be debugged, including applications that use threads,
signals, and/or shared memory. Applications using custom system libraries (e.g.
malloc or threading) are supported. We even support applications that directly
access hardware (e.g. openGL or Infiniband). UndoDB brings reversible
debugging to complex, real-world applications. In short, it's reversible
debugging that really works.
Read what's new in UndoDB 3.0
Read more on UndoDB's technical details
|
