[pypy-commit] pypy stmgc-c4: update TODO

Raemi noreply at buildbot.pypy.org
Tue Oct 15 11:32:46 CEST 2013


Author: Remi Meier <remi.meier at gmail.com>
Branch: stmgc-c4
Changeset: r67390:bafcb0cdff48
Date: 2013-10-15 10:35 +0200
http://bitbucket.org/pypy/pypy/changeset/bafcb0cdff48/

Log:	update TODO

diff --git a/TODO b/TODO
--- a/TODO
+++ b/TODO
@@ -1,7 +1,3 @@
-------------------------------------------------------------
-
-kill INEVITABLE in et.c, replace with "global_cur_time & 1" again
-
 ------------------------------------------------------------
 
 try to let non-atomic inevitable transactions run for longer, until
@@ -21,11 +17,7 @@
 
 ------------------------------------------------------------
 
-implement thread-locals in RPython (for the executioncontext)
-
-------------------------------------------------------------
-
-optimize the static placement of the STM_XxxBARRIERs
+optimize the static placement of the STM_XxxBARRIERs and use them in JIT
 
 ------------------------------------------------------------
 
@@ -39,8 +31,7 @@
 to the PyFrame object (make sure it's always written and don't put more
 barriers)
 
-in parallel, tweak the API of stmgc: think about adding
-stm_repeat_read_barrier, and support "tentative" write_barrier calls
+in parallel, tweak the API of stmgc: support "tentative" write_barrier calls
 that are not actually followed by a write (checked by comparing the
 object contents)
 
@@ -58,4 +49,16 @@
 fast-path, as well as splitting it based e.g. on the RPython type of
 object.  See also vtune.
 
-reimplement the fast-path of the nursery allocations in the GC
+JIT
+~~~
+
+* reimplement the fast-path of the nursery allocations in the GC
+** use this for frame allocation in stmrewrite for call_assembler
+* use specialized barriers in JIT
+* optimize produced assembler code
+* avoid calling aroundstate.after() for call_release_gil and instead
+  start a normal transaction after the call
+* maybe GUARD_NOT_INEVITABLE after call_may_force, call_assembler
+  which is a small check if we are inevitable and does a transaction_break
+  if we are.
+* look at XXXs for STM everywhere


More information about the pypy-commit mailing list