Switch branches/tags
Commits on Sep 30, 2013
Commits on Sep 18, 2013
Commits on Sep 16, 2013
  1. Fixed a nasty bug in that I introduced last weel ( Print…

    lucasaiu committed Sep 16, 2013
    …f.fprintf stderr "QQQQQQQQ %s\n%!" s;)
  2. Conditionally moved some contextual C globals back out of the struct,…

    luca committed Sep 16, 2013
    … if they are accessed by the assembly runtime in some ports; this lets us keep existing assembly runtimes with no changes
  3. Minor fixes in the i386 GNU/Linux backend, for compatibility with the…

    luca committed Sep 16, 2013
    … multicontext system
Commits on Sep 12, 2013
  1. Stuff

    lucasaiu committed Sep 12, 2013
  2. Bugfixes and stuff

    Luca Saiu
    Luca Saiu committed Sep 12, 2013
Commits on Sep 10, 2013
  1. Fixed native code support for non-multiruntime architectures (develop…

    Luca Saiu
    Luca Saiu committed Sep 10, 2013
    …ed on PowerPC). Fixed PowerPC assembly runtime with a cute little hack, which unfortunately will have to go.
Commits on Sep 9, 2013
  1. Configuration improvements: now multi-context can be supported but st…

    lucasaiu committed Sep 9, 2013
    …ill disabled in an installation, and configuration macros reflect this duality
Commits on Sep 4, 2013
  1. Merge: trivial stuff

    lucasaiu committed Sep 4, 2013
  2. Trivial changes

    lucasaiu committed Sep 4, 2013
  3. Fixes for 32-bit and big-endian architectures, after testing on my ol…

    Luca Saiu
    Luca Saiu committed Sep 4, 2013
    …d PowerPC machine
Commits on Sep 3, 2013
Commits on Aug 6, 2013
  1. Two important changes: 1) changed the destruction of the ticker threa…

    lucasaiu committed Aug 6, 2013
    …ds at systhread context finalization to be synchronous: this is important for ensuring that the masterlock is not used after context destruction; 2) correctly destroyed the dynamically-allocated part of named_value_table, for each context.
  2. In the previous commit message I forgot to mention that I also implem…

    lucasaiu committed Aug 6, 2013
    …ented Context.at_exit, and updated the Callback machinery so that its state is correctly cloned at split time
  3. Implemented context finalization with reference counting. It works wi…

    lucasaiu committed Aug 6, 2013
    …th no threads and with systhread, but not yet (if ever) with vmthreads
Commits on Jul 25, 2013
  1. Added a thin abstraction layer over sem_wait and sem_post, because se…

    lucasaiu committed Jul 25, 2013
    …m_wait was being interrupted by signals in the bytecode runtime; now I correctly call it in a loop
Commits on Jul 23, 2013
  1. Fixed channel locking in the case of multi-context concurrent channel…

    lucasaiu committed Jul 23, 2013
    … access without threads, and vmthreads. The vmthreads part is ugly and inefficient in case of contention over busy channels
Commits on Jul 16, 2013
  1. Adapted VM threads to multi-context; threads work, now I only have to…

    lucasaiu committed Jul 16, 2013
    … move a few global variable to the context struct
Commits on Jul 15, 2013
Commits on Jul 12, 2013
  1. Channel refcount bug at least worked-around. No more known crashes at…

    lucasaiu committed Jul 12, 2013
    … this point. Win win. I deserve a beer.
Commits on Jul 5, 2013
  1. Important bugfix: the masterlock is now correctly initialized for eac…

    lucasaiu committed Jul 5, 2013
    …h context, rather than only for the first one
  2. Stuff

    lucasaiu committed Jul 5, 2013
Commits on Jul 3, 2013
  1. Important crash fix in st_stubs.c:caml_thread_start: the local clos m…

    lucasaiu committed Jul 3, 2013
    …ust NOT be GC-protected
Commits on Jun 26, 2013
  1. Fixed deadlock

    lucasaiu committed Jun 26, 2013
Commits on Jun 14, 2013
Commits on Jun 5, 2013
  1. Changed the initialization of ctx->caml_signal_handlers from 0 to Val…

    lucasaiu committed Jun 5, 2013
    …_int(0), to make the value serializable
Commits on Jun 4, 2013
  1. Big fix: I used to work on blobs from one context within st_stubs.c, …

    lucasaiu committed Jun 4, 2013
    …while at the same time that same context was running Caml code. This potentially used serialization structures in a non-reentrant way
Commits on May 27, 2013
Commits on May 17, 2013
  1. Stuff

    lucasaiu committed May 17, 2013
Commits on May 7, 2013
  1. Fixed the crazy bug only showing up without address randomization (th…

    lucasaiu committed May 7, 2013
    …anks to Fabrice). Other changes, comparatively minor
Commits on Apr 19, 2013
  1. Minor changes. Looking for a bizarre crash occurring when kernel.rand…

    lucasaiu committed Apr 19, 2013
    …omize_va_space is 0 or 1 (not 2)