Age | Commit message (Collapse) | Author |
|
Changelog
=========
Clean up stray whitespace in configure.ac (8d48b43)
Merge pull request #226 from nortiero/gcc6-darwin-workaround (ae59ac1)
Merge pull request #223 from juanfra684/patch-2 (c2e2e6a)
Fix linking error when gsc invoked with -:s flag (issue #227) (c86476d)
Add unit-tests for threads and mutexes (f72c967)
Improve unit tests for thread-start! and thread-join! (30748a6)
Better message (b0b11be)
Disable -fipa-ra optimization when the following conditions are met: (8ac3ca7)
Revert change to the OpenBSD test in configure (efa91eb)
|
|
Changelog
=========
Avoid heap overflows due to missing critical sections in multiple threaded VMs (2589a2b)
Avoid spurious heap overflows due to uninitialized variables (59d1f58)
|
|
Changelog
=========
Need to reset stack trip to properly detect stack overflows and interrupts after a GC (277666d)
|
|
Changelog
=========
Use correct parameter ordering for WIN32 InterlockedCompareExchangePointer function (3cd348c)
Use wrapping versions of ##fx* and ##fx+ when hashing to avoid overflows (861d4de)
Must use arithmetic shift right to properly serialize negative fixnums (1a130d1)
Replace calls to ##fxarithmetic-shift-right by more portable ##fxwraplogical-shift-right (67dc73d)
|
|
Changelog
=========
Fix issue #222 (from smp branch) (f14c393)
|
|
Changelog
=========
Better consistency with SMP branch (0c34159)
Merge branch 'master' of https://github.com/gambit/gambit (bd4c030)
Apply fix for issue #221 from SMP branch (3a2fe7e)
In FFT-based multiplication, speed up conversion to bignum. (db91c8b)
|
|
Changelog
=========
Avoid configuring WIN32 thread system in Microsoft Visual C++ build batch files (8f624c8)
Update Microsoft Visual C++ build batch files (7f760df)
|
|
|
|
Changelog
=========
Fix build process so that it can build multiple branches (0209073)
|
|
|
|
|
|
|
|
|
|
Changelog
=========
Avoid stale thread-resume-thunk when rescheduling thread (86ae4c4)
|
|
Changelog
=========
Merge branch 'master' of https://github.com/gambit/gambit (fe9acbc)
Fix issue #216 (crash when thread-interrupt! is called before body of thread has started executing) (cde319d)
|
|
Changelog
=========
Dump control-flow history for all processors of the VM. (433a723)
Rename trampoline function to ___trampoline for more consistent C backtraces (d858b0e)
|
|
Changelog
=========
Replace ill-named "run-queue" structure by "processor" and "vm" structures. (785cb74)
Refactor code to use a macro to define operations on double-ended-queues (3b7b23b)
|
|
Changelog
=========
Only use "volatile" qualifier when configured with --enable-multiple-threaded-vms (de3cd85)
|
|
Changelog
=========
Change numbering of interrupts (it is important for ___INTR_SYNC_OP to be first so that the barrier sync is entered reliably by all processors) (2f68abe)
The stack_trip pointer itself must be volatile because it is a modification of this pointer that informs a processor that it has been interrupted by another processor (a8604be)
Use ___WORD type to allow atomic increment/decrement of GC worker count (bfd04a2)
Smaller trace width in activity log so that large processor count fits on normal screens (457b104)
|
|
Changelog
=========
Proper cast for string conversion when --enable-cplusplus and --enable-openssl are used (318a156)
Fix incorrect uses of ___EXT macro (701101d)
|
|
Changelog
=========
Update README.md to make docs before installing (05575ba)
|
|
Changelog
=========
Better fix for ,(v procedure) issue (545f816)
Fix REPL command ,(v procedure) to visit the environment of a procedure (a6f5e57)
|
|
|
|
|
|
|
|
|
|
Changelog
=========
Put a time limit on the busy waiting in the barrier sync used by the garbage collector (bcaee11)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Changelog
=========
Correctly implement R5RS eval environments (see issue #204) (0928445)
Fix heap resizing calculation for very large heaps (> 4 GB) (1ca21f0)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Changelog
=========
When debugging, give location of call to fatal_heap_overflow (eb9f23f)
Fix miscomputation of heap size due to using int type (which may be 32 bits on a 64 bit machine) (3fed276)
|
|
|
|
|