Release 2.2
Dmitriy Vyukov
http://groups.google.com/group/relacy/files
This is mostly bug fix release. Here is a list of changes:
Features:
+ Support for pthread_mutex_timedlock()
+ Support for ETIMEDOUT, EINTR in pthread_cond_timedwait()/
pthread_cond_wait()
+ rl::hash_ptr(p, sz) function which provides deterministic hashing of
pointers
Fixes:
+ Win32 mutex is now recursive
+ Compilation issue on MSVC x64 when RL_DEBUGBREAK_ON_ASSERT/
RL_DEBUGBREAK_ON_FAILURE defined
+ Fixed OOM crash when execution history is very large
+ Fixed rare crash during iteration count estimation in context bound
scheduler
+ Fixed bug in pthread_rwlock/SRWLOCK that at most 2 readers may
acquire it simultaneously
+ Fixed bug regarding false race detection when simulation runs for
very long time (int overflow happened)
Open issues are:
- Problem with modeling global variables
http://groups.google.com/group/relacy/browse_frm/thread/0f35245abae1db94
- Minor problem with Win32 TLS simulation on Win64 platforms:
http://groups.google.com/group/relacy/browse_frm/thread/349c0a96c6032873
- Problem with timed waits under full/context bound schedulers:
http://groups.google.com/group/relacy/browse_frm/thread/6ad2c1c22d325c5a
(there is a fundamental problem between verification completeness and
fairness when timed waits are used)
All the issues are pointed by Edward. Thank you Edward. I am working
on the issues to the extent of my abilities. I just hope that Relacy
in it's current state is already useful and of benefit to users.
Stay tuned!
--
Dmitriy V'jukov
Chris M. Thomasson
> Just uploaded Release 2.2 to the Files section:http://groups.google.com/group/relacy/files
>
> This is mostly bug fix release. Here is a list of changes:
[...]
> + Fixed bug regarding false race detection when simulation runs for
> very long time (int overflow happened)
[...]
Thank you so very much for taking the time to address these issues
Dmitriy!
I just wanted to let everybody know that version 2.2 fixed basically
all of my long running intensive testing models of various algorithms
I am developing and experimenting with. For instance, it solved the
false race detection in the model of my lock-based vZOOM algorithm
posted here:
http://groups.google.com/group/relacy/msg/c2fb28c1d32ab7a6
I have to admit, this bug in Relacy was literally going to drive me
insane Dmitriy! I was getting race detections when I just "knew" that
it was "impossible", then I would start to second guess myself; ouch!
God Bless Relacy 2.2!
;^D
edward
On Jan 15, 6:42 am, Dmitriy Vyukov <dvyu...@gmail.com> wrote:
> Just uploaded Release 2.2 to the Files section:http://groups.google.com/group/relacy/files
Much thanks for the fixes and new features!
-Edward
Dmitriy Vyukov
Chris and Edward, thank you for your kind words too!
--
Dmitriy V'jukov