Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
50 most recent check-ins tagged with "notworking"
|
2020-01-05
| ||
| 20:51 | Calculate effect of 'invoke' on variable analysis. Not yet 100% correct - breaks six additional test cases. check-in: 0fdceaac26 user: kbk tags: notworking, kbk-refactor-callframe | |
|
2020-01-03
| ||
| 19:48 | Patch out poly1305::compute for now - segfault! - and break out the 'vartest' tests. check-in: 13b2b5767e user: kbk tags: notworking, kbk-refactor-callframe | |
|
2020-01-02
| ||
| 20:48 | Detect that 'invoke' preserves some callframe variables. Add buffer flushing so that the identity of a test is reported before a crash. Remove some redundant test invocations. check-in: 65069fda8f user: kbk tags: notworking, kbk-refactor-callframe, current | |
|
2019-12-31
| ||
| 17:32 | Separate exists and doesn't-exist paths in 'translate' of read-modify-write operations. Fix overeager elimination of 'extractArray'. Move 'partialredundancy' off to its own specializer step, since 'tidy' passes may NOT introduce new variables. Only 'lsetest' is now failing. check-in: 09d85d696a user: kbk tags: notworking, kbk-refactor-callframe | |
| 00:10 | Finish up elision of 'moveFromCallFrame' in the simpler cases. check-in: bbb81cb21e user: kbk tags: notworking, kbk-refactor-callframe | |
|
2019-12-30
| ||
| 03:14 | Provide a hook on which to hang speculative phi generation. check-in: f635c5689f user: kbk tags: notworking, kbk-refactor-callframe | |
| 01:50 | Peephole optimization of 'moveFromCallFrame' when the value in the callframe is known. (Will be much more effective once more instructions are analyzed: notably 'invoke' and 'phi'.) check-in: 188df1dde4 user: kbk tags: notworking, kbk-refactor-callframe | |
|
2019-12-27
| ||
| 23:01 | Add the basic framework for forward analysis of available callframe variables check-in: 85ea960b98 user: kbk tags: notworking, kbk-refactor-callframe | |
|
2019-11-23
| ||
| 23:02 | merge trunk check-in: c3fc2d4137 user: kbk tags: notworking, kbk-refactor-callframe | |
| 22:35 | Merge trunk: a couple of ops that interact with the callframe and were missed while refactoring Leaf check-in: 19e26fa988 user: kbk tags: notworking, tcl_8_7_branch | |
|
2019-11-22
| ||
| 20:31 | Rejigger printing of results so that a crashing test can be identified more readily. check-in: c7402fd29e user: kennykb tags: notworking, tcl_8_7_branch | |
| 19:44 | Initial changes to support differences in the Tcl 8.7 ABI check-in: 668d4221ed user: kennykb tags: notworking, tcl_8_7_branch | |
|
2019-01-29
| ||
| 13:19 | Fix misorderered scalar check/copy to result check-in: 36e8177510 user: kbk tags: notworking, kbk-refactor-callframe | |
| 02:00 | Fix translation bugs that broke [dict] tests check-in: ab6da27502 user: kbk tags: notworking, kbk-refactor-callframe | |
|
2019-01-23
| ||
| 02:12 | Add conversion to IMPURE ZEROONE INT, needed in booltest. Keep code motion from messing with the callframe for now. Patch out the handful of remaining tests that abort. Next up: fix the dictest tests involving 'dict update', lsetest, and re-add support for direct ops. check-in: 27339e7e0c user: kbk tags: notworking, kbk-refactor-callframe | |
|
2019-01-22
| ||
| 02:52 | Restore the 'timings' list - if 'transform' doesn't produce it, there's a crash. check-in: 3e49a05889 user: kbk tags: notworking, kbk-refactor-callframe | |
| 02:36 | Run dead code elimination before copy propagation because copyprop explodes otherwise. check-in: 9e16d3b8d6 user: kbk tags: notworking, kbk-refactor-callframe | |
| 02:35 | Bug fixes to make all the 'expandtest' cases work check-in: 4f7b3a4d01 user: kbk tags: notworking, kbk-refactor-callframe | |
|
2019-01-21
| ||
| 22:33 | Enough partial implementation for expandtest::test1-test3 check-in: 89030dcc42 user: kbk tags: notworking, kbk-refactor-callframe | |
| 19:45 | Fix mislinking of phi operations on the error branch of 'invoke' check-in: 407e1ef055 user: kbk tags: notworking, kbk-refactor-callframe | |
| 18:14 | Enough changes to get through first two 'expandtest' tests check-in: 5557b1e592 user: kbk tags: notworking, kbk-refactor-callframe | |
|
2019-01-18
| ||
| 04:22 | Add the final processing in 'varargs' - next, emit the error path. check-in: 305328fa6b user: kbk tags: notworking, kbk-refactor-callframe | |
|
2019-01-16
| ||
| 02:30 | More argument preparation code in 'varargs' check-in: 76b943ad4a user: kbk tags: notworking, kbk-refactor-callframe | |
|
2019-01-14
| ||
| 03:46 | Further development of varargs. Note that the invocation sequence is much, much simpler than it used to be, so 'invoke.tcl' is no more. check-in: 90e908dae3 user: kbk tags: notworking, kbk-refactor-callframe | |
|
2019-01-13
| ||
| 15:43 | Clean out dead 'exists.tcl' source check-in: f283d28ebd user: kbk tags: notworking, kbk-refactor-callframe | |
| 15:38 | Merge the (not-working) vararg reform branch. It appears that both these tasks need to be attacked at the same time because the changes are tightly interwoven. check-in: 05c93c9cc5 user: kbk tags: notworking, kbk-refactor-callframe | |
|
2019-01-09
| ||
| 01:52 | Add support for compiled procs that alter the callframe, then fail. check-in: 30d6a54fdb user: kbk tags: notworking, kbk-refactor-callframe | |
|
2019-01-08
| ||
| 04:17 | Fix more trivial errors in 'translate' and missing cases in code gen check-in: fba0ec7241 user: kbk tags: notworking, kbk-refactor-callframe | |
|
2019-01-05
| ||
| 22:48 | Generate only one moveFromCallFrame per bytecode instruction check-in: 9dbaca7922 user: kbk tags: notworking, kbk-refactor-callframe | |
|
2019-01-02
| ||
| 02:20 | Oops, one extra callframe ref in 'callframe-error-quads' check-in: cdc4d10e86 user: kbk tags: notworking, kbk-refactor-callframe | |
|
2019-01-01
| ||
| 23:06 | Update quadcode generation and documentation so that 'direct' operations refer to the callframe (which is necessary in the event that they are emitted to deal with double-dereference). CODE ISSUER IS NOT YET UPDATED TO COPE WITH THE CHANGE. check-in: 104debc734 user: kbk tags: notworking, kbk-refactor-callframe | |
| 21:57 | Move callframe moves to the actual points where callframe values are moved. They can then be optimized by avaiability and liveness analysis. check-in: dbd9571178 user: kbk tags: notworking, kbk-refactor-callframe | |
|
2018-12-27
| ||
| 19:47 | Add 'buiilder.tcl' inadvertently omitted from previous commit Closed-Leaf check-in: 2e3b3ea76f user: kbk tags: notworking, kbk-refactor-varargs | |
| 19:44 | Very rough beginning of a refactored 'varargs' pass check-in: 7c9d400e5e user: kbk tags: notworking, kbk-refactor-varargs | |
|
2018-12-17
| ||
| 22:08 | Finish jump threading - actually do the block duplication and redirection of jumps. Add the logic for SSA deconstruction (required by jump threading) and make SSA construction work with the deconstructed result. check-in: a41b93130e user: kbk tags: notworking, kbk-jumpthread | |
|
2018-12-16
| ||
| 05:18 | Add constant folding for unary +. It probably ought to be folded out altogether, since it serves only to trigger type checking. Add an outline of the actual process of jump threading once the threads have been determined. check-in: c388c8737b user: kbk tags: notworking, kbk-jumpthread | |
| 04:55 | Add a 'cos2' test case to illustrate the cost of non-numeric ordering comparisons check-in: 86167d6917 user: kbk tags: notworking, kbk-jumpthread | |
| 04:43 | Finish forward analysis in the jump threading pass. check-in: dc90afa15d user: kbk tags: notworking, kbk-jumpthread | |
|
2018-12-10
| ||
| 04:51 | Remove vestiges of the old node splitter from 'inline.tcl'. Make console dribble in 'jumpthread.tcl' contingent on debug-jumpthread. check-in: fd2ea3e6f1 user: kbk tags: notworking, kbk-jumpthread | |
| 04:09 | OOPS! Didn't add the new 'jumpthread.tcl' module! Remove the old 'nodesplit.tcl', and remove the 'renameTemps' pass since 'nodesplit' was the only thing that ever depended on it. check-in: 67b989853b user: kbk tags: notworking, kbk-jumpthread | |
| 01:34 | Calculation of anticipable tests for jump threading check-in: f138663832 user: kbk tags: notworking, kbk-jumpthread | |
|
2018-12-06
| ||
| 01:42 | Remove speculative phis if they turned out not to be useful check-in: bd6294fade user: kbk tags: notworking, kbk-pre | |
|
2018-12-05
| ||
| 05:37 | Rewrite pre_insert and pre_phi_translate to NOT use a persistent cache. The missing piece in VanD04 is that pre_insert must call pre_phi_translate for all the anticipated expressions first, and then use the dictionary that results for the translated values. There are still further bugs, but we're over this hump at least. check-in: 0bbe5a5dd3 user: kbk tags: notworking, kbk-pre | |
|
2018-12-03
| ||
| 05:08 | Stop constant folding from leaving dead code behind, Add a test for simple nested iterations, using [lmap]. Temporarily patch 'foreach' operations from being hoistable - I don't think this will be necessary, but it's tickling other bugs. Make translation of values across a phi work if one of the inputs to the phi is a literal. Put 'bbidom' and 'bblevel' directly after dead code elimination, because virtually everything depends on having dominators, which deadcode destroys. check-in: 4f50ed77b2 user: kbk tags: notworking, kbk-pre | |
| 01:06 | Fix crashing bug in creating variable name for a phi operation. Fix bug where partial redundancy elimination got stale dominators, causing much weirdness. Add additional tracing to available expression calculation. check-in: ce1088b7a7 user: kbk tags: notworking, kbk-pre | |
|
2018-12-02
| ||
| 23:27 | Turn on partial redundancy elimination. Correct bug in availability analysis that led to infinite loop in 'msrange2' check-in: 68da56cfee user: kbk tags: notworking, kbk-pre | |
|
2018-11-29
| ||
| 00:56 | merge trunk check-in: 815387c202 user: kbk tags: notworking, kbk-pre | |
|
2018-11-26
| ||
| 04:57 | Added a couple of tests to trigger loop-invariant code motion. Corrected du-chain mismanagement in callframe.tcl (deleting moveToCallFrame), copyprop.tcl (deleting any quads), uselessphis (deleting anything), and narrow.tcl (cleanupNarrow, deleting anything). NOT WORKING: During node splitting, at least 'flightawarebench', 'msrange', 'bug7c59', and 'cse-caller' are still coming out with mismanaged chains. Improved auditing in the 'tidy' pass of the transformer, and made it less aggressive about rechecking everything. NOT WORKING: in the 'wordcounter2' test case, it's missing at least one invocation of 'initArrayIfNotExists' because we're falling into the code gen in 'compile.tcl' for initArrayIfNotExists(ARRAY). Advanced partial redundancy elimination to 'code complete' status. The stuff surrounding it is in bad enough shape at the moment that it is far from tested, and known not to be working for everything, but is patched out at the moment. check-in: 02ab4301aa user: kbk tags: notworking, kbk-pre | |
|
2018-11-13
| ||
| 02:57 | Code complete for PRE - except that I don't know how to introduce a phi if PRE has made an expression available on two afferent flow graph edges rather than in a dominator. check-in: fa9f7e0eb3 user: kbk tags: notworking, kbk-pre | |
|
2018-11-10
| ||
| 22:17 | Use the lexical-name algorithm for 'altered' for now. May change to value-driven later. check-in: 592020ed7a user: kbk tags: notworking, kbk-pre | |