aboutsummaryrefslogtreecommitdiffstats
path: root/src/qml/compiler/qv4ssa.cpp
Commit message (Collapse)AuthorAgeFilesLines
...
* | V4 IR: make statement numbering fixed and clean up statement worklists.Erik Verbruggen2014-05-281-165/+178
| | | | | | | | | | | | | | | | | | | | | | | | | | | | Every statement in the IR now gets a fixed unique number. This number can be used to store statements or information for a statement into an array where the number is used as an index. This removes the need for many hashes. In the process of changing the code the two statement worklists in the optimizer are merged into one. A single instance can be (and is) re-used by the various algorithms. Change-Id: I8f49ec7b1df79cf6914c5747f5d2c994dad110b2 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* | V4 IR: replace QHash in DefUses with a vector.Erik Verbruggen2014-05-271-41/+65
| | | | | | | | | | Change-Id: Ibf21f5fe0f8ab035add5354f45f7869f4cdfead8 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* | V4 IR: removed a couple of hashes and sets.Erik Verbruggen2014-05-261-57/+87
| | | | | | | | | | Change-Id: I09f9aa1921745b9aa323349d90c334b156f690cb Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* | V4: Split arguments/locals from temps.Erik Verbruggen2014-05-231-92/+63
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | There are a couple of reasons to split the temporaries off from the arguments and locals: Temporaries are invisible, and changes to them cannot be observed. On the other hand, arguments and locals are visible, and writes to them can be seen from other places (nested functions), or by using the arguments array. So, in practice these correspond to memory locations. (One could argue that if neither nested functions, nor eval(), nor arguments[] is used, the loads/stores are invisible too. But that's an optimization, and changing locals/arguments to temporaries can be done in a separate pass.) Because of the "volatile" nature of arguments and locals, their usage cannot be optimized. All optimizations (SSA construction, register allocation, copy elimination, etc.) work on temporaries. Being able to easily ignore all non-temporaries has the benefit that optimizations can be faster. Previously, Temps were not uniquely numbered: argument 1, local 1, and temporary 1 all had the same number and were distinguishable by their type. So, for any mapping from Temp to something else, a QHash was used. Now that Temps only hold proper temporaries, the indexes do uniquely identify them. Add to that the fact that after transforming to SSA form all temporaries are renumbered starting from 0 and without any holes in the numbering, many of those datastructures can be changed to simple vectors. That change gives a noticeable performance improvement. One implication of this change is that a number of functions that took a Temp as their argument, now need to take Temp-or-ArgLocal, so Expr. However, it turns out that there are very few places where that applies, as many of those places also need to take constants or names. However, explicitly separating memory loads/stores for arguments/locals from temporaries adds the benefit that it's now easier to do a peep-hole optimizer for those load/store operations in the future: when a load is directly preceded by a store, it can be eliminated if the value is still available in a temporary. Change-Id: I4114006b076795d9ea9fe3649cdb3b9d7b7508f0 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* | Merge "Merge remote-tracking branch 'origin/stable' into dev" into ↵Simon Hausmann2014-05-221-20/+12
|\ \ | | | | | | | | | refs/staging/dev
| * | Merge remote-tracking branch 'origin/stable' into devSimon Hausmann2014-05-221-20/+12
| |\| | | | | | | | | | Change-Id: I0dd91626837276f5811e4830f4a4e9f89bf1e1bd
| | * V4 IR: change BasicBlockSet::const_iterator to use std::find.Erik Verbruggen2014-05-151-20/+12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When the storage in the BasicBlockSet is a std::vector<bool>, the begin() iterator and the ++operator now use std::find. Good STL implementations have an optimized version specifically for std::vector<bool> that is a lot faster than the iterating that was used before. The change also makes the code more readable. Change-Id: Ic37bac694c133c597b3d61a91b04a0b758516dc3 Reviewed-by: Lars Knoll <lars.knoll@digia.com>
* | | V4 IR: change datastructure of the worklist used by type inference.Erik Verbruggen2014-05-221-48/+119
|/ / | | | | | | | | | | | | | | | | | | No sets, no cry, and no mallocs of QHashNodes. QBitArray is not used on purpose: good STL implementations specialize std::find for std::vector<bool>, with a significant speed gain. Change-Id: Ic986bbd746e96eb494496f0508acb17318f4f0fb Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* / V4 IR: change IR printing to use a visitor.Erik Verbruggen2014-05-211-97/+10
|/ | | | | | | | This allows for overriding methods to customize the printing of nodes. It also removes some duplicate code. Change-Id: Ieb9eec2fa7d4e211932d7772586a1d62b119a90a Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* Merge remote-tracking branch 'origin/release' into stableFrederik Gladhorn2014-05-011-5/+2
|\ | | | | | | Change-Id: I996a85744753598bb48c7e0d7954049202f4f037
| * Extend the QML bootstrap library by the IR buildersSimon Hausmann2014-04-231-5/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | This is among other things needed to fix the qml import scanner to detect dependencies from .js files correctly. The patch also fixes the use of Q_QML_EXPORT towards Q_QML_PRIVATE_EXPORT where appropriate and corrects the wrong include path for the double conversion code to actually be relative to the file it is included from. This worked by accident because of other include paths present in the build. Change-Id: I338583dad2f76300819af8ab0dae8e5724c84430 Reviewed-by: Lars Knoll <lars.knoll@digia.com>
* | V4 IR: more copy removal.Erik Verbruggen2014-04-301-29/+47
| | | | | | | | | | Change-Id: Ie86bf5902328ad105fb07eb3e2809db21ff024d9 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* | V4 IR: replace QSets with QBitArray and QVector.Erik Verbruggen2014-04-291-20/+19
| | | | | | | | | | Change-Id: I565e0a22d4e94495eb427b85a59a62733a815527 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* | V4 IR: reduce runtime cost.Erik Verbruggen2014-04-151-32/+55
| | | | | | | | | | | | | | | | | | | | | | - Replace 2 QHash<BasicBlock *, ...> with QVector<...>, where the basic-block index is the index of the vector. - Nearly all QHashes and QSets will have a minimal fill rate. So, initialize/reserve all of them with a reasonable minimal size to prevent re-allocations and re-hashing. Change-Id: Iade857991d73fddd0b92cecb8d458064b253a08d Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* | V4 IR: prevent accidental detaches of QVectors.Erik Verbruggen2014-04-151-2/+3
| | | | | | | | | | Change-Id: I20ebf44ff0609f6833f7e59a4f2fb312be11b8c1 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* | V4 IR: clean up basic-block management and statement access.Erik Verbruggen2014-04-151-254/+259
|/ | | | | | | | | | | | | | | | BasicBlocks have an index property which points to the index of that basic block in the container array in Function. This property can be used to store calculated information about basic blocks in a vector, where the vector index corresponds to the basic block index. This is a lot cheaper than storing any information in a QHash<BasicBlock *, ....>. However, this numbering requires that no re-ordering or deletion of blocks happens. This change cleans up all that handling which was scattered over a number of places. Change-Id: I337abd39c030b9d30c82b7bbcf2ba89e50a08e63 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* Enable constant propagation for all typesLars Knoll2014-03-191-13/+20
| | | | | | | | | So far constant propagation was only enabled for numbers and booleans. Enable it for all types now and make sure the propagation does the right thing. Change-Id: I202b0073f463d8a42e34931a736544207284b6dc Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* Cleanup our runtime methodsLars Knoll2014-03-111-8/+8
| | | | | | | | | Move all our runtime methods into the QV4::Runtime struct and give them nicer names without underscores. Sort them logically and remove a few unused methods. Change-Id: Ib69b71764ff194d0ba211aac581f9a99734d8180 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* V4 SSA: remove unused code.Erik Verbruggen2014-03-081-49/+1
| | | | | Change-Id: I56b3e5400e7b9880b9534117ac17a80436ff1733 Reviewed-by: Lars Knoll <lars.knoll@digia.com>
* Fix more MSVC2012 compiler warnings.Erik Verbruggen2014-03-071-4/+7
| | | | | | | All are conversions from size_t to int or to unsigned. Change-Id: Ic94c938dcad6d50a32dd6ec62da2341869cf994d Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* Clean up our internal namespacesLars Knoll2014-02-231-58/+58
| | | | | | | | QQmlJS::MASM -> QV4::JIT QQmlJS::V4IR -> QV4::IR Change-Id: I707e8990459114a699c200fe3c22cec3c8df1afc Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* Really eliminate a|0 and b&(-1)Lars Knoll2014-02-221-8/+8
| | | | | | | | The old code wasn't doing what it promised to do and failed to remove these expressions. Change-Id: I6718539fd528f293db537e47ff1c9ac4b27ada55 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* Mask rhs of shift operations in the IR for constantsLars Knoll2014-02-221-0/+12
| | | | | | | Saves an instruction for shifts with contants Change-Id: Ia12355d2fe2b9f80631056cda5edd79b45189e99 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* Remove unused text streamLars Knoll2014-02-221-3/+0
| | | | | Change-Id: I23a1f5f2c6f782fab315db4725412a9473178b70 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* Remove some code duplicationLars Knoll2014-02-221-85/+27
| | | | | Change-Id: If336731a49bcb5e1812d50a39cb97e263f23b183 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* Cleanup: Pass the function into the various SSA stagesLars Knoll2014-02-221-37/+33
| | | | | | | | Pass in the function object instead of the variablesCanEscape flag. Cleans up the code a bit. Change-Id: If2d24b30fb223fa03e27f75f40ad42b881b2b3a0 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* Merge remote-tracking branch 'origin/stable' into devSimon Hausmann2014-02-181-1/+2
|\ | | | | | | | | | | | | | | Conflicts: src/qml/jsruntime/qv4functionobject.cpp src/quick/scenegraph/coreapi/qsgbatchrenderer_p.h Change-Id: Id164f6c3b45501aa466908659ec4e3b957323753
| * Merge remote-tracking branch 'origin/release' into stableFrederik Gladhorn2014-02-111-1/+2
| |\ | | | | | | | | | Change-Id: I5ac68cc3ad3926190817f0d3d5b4526e70badff6
| | * Remove static initialization of QObjectsv5.2.1Eskil Abrahamsen Blomfeldt2014-01-291-1/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Statically initializing QObjects will cause e.g. the main thread pointer in QCoreApplication to be set when the library is loading. On Android, this is never the real main thread. The effect was a warning about QApplication not being initialized on main thread, and a race condition which sometimes caused apps to hang on startup. [ChangeLog][Android] Fixed possible hang on startup for Qt Quick applications. Task-number: QTBUG-36426 Change-Id: I7bd8a8f35ef1a2548949978563e3157f8dc854c7 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* | | V4: Do not reverse-propagate int32 conversion through unary minus.Erik Verbruggen2014-02-141-4/+29
| | | | | | | | | | | | | | | | | | | | | | | | | | | Conversion to int32 might truncate, and because int32 is two's complement, INT_MIN might get converted incorrectly. Change-Id: Iaf893d3bd619f4c5791654e609f96cffca5c6917 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com> Reviewed-by: Lars Knoll <lars.knoll@digia.com>
* | | V4 IR: remove unused phi targets.Erik Verbruggen2014-02-141-0/+12
| | | | | | | | | | | | | | | | | | | | | | | | | | | Remove all phi nodes whose target temps have no uses. This obvious optimization was missing. Change-Id: I24354e225ba7b01a3c2a6f4b2e40dd78d6ee3d7d Reviewed-by: Simon Hausmann <simon.hausmann@digia.com> Reviewed-by: Lars Knoll <lars.knoll@digia.com>
* | | V4 IR: changed worklist used while doing optimizations.Erik Verbruggen2014-02-141-36/+138
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Change the worklist in the optimizeSSA function to allow only one entry for a statement. This prevents re-processing a statement multiple times, and could lead to crashes when the statement was removed in an earlier pass. Change-Id: I2f09cf74525cfe19708ec7a8bc6d555218625e87 Reviewed-by: Lars Knoll <lars.knoll@digia.com>
* | | Merge remote-tracking branch 'origin/stable' into devSimon Hausmann2014-02-111-3/+6
|\| | | | | | | | | | | | | | | | | | | | | | | | | | Conflicts: src/qml/compiler/qv4ssa.cpp src/qml/jsruntime/qv4arrayobject.cpp src/qml/jsruntime/qv4context.cpp Change-Id: Ied5b23bec4dc14abe51127c507aed668f855c1e1
| * | V4: fix range sortingErik Verbruggen2014-01-311-3/+6
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When a life-time interval is split from another interval, it has to come before an interval that starts at the same position but is not split. This also means that a means that the all ranges in a split interval are uses, which is important for allocation: all incoming parameters for an operation need to be allocated before allocating a register for the result as the result will only start its life "at the end" of the operation. This patch fixes a problem register allocation is done in a function where register pressure is high (e.g. on platforms that have few registers to start with). Specifically, crypto.js on x86 triggered it. Change-Id: Iee3e5d82a887b8de573dfc23513844143d0c8073 Reviewed-by: Albert Astals Cid <albert.astals@canonical.com> Reviewed-by: Simon Hausmann <simon.hausmann@digia.com> Reviewed-by: Fawzi Mohamed <fawzi.mohamed@digia.com> Reviewed-by: Lars Knoll <lars.knoll@digia.com>
| * | V4: remove unused field.Erik Verbruggen2014-01-291-5/+3
| | | | | | | | | | | | | | | | | | Change-Id: Id1edc073db349cfc7e4b5a9dca045760016ebacf Reviewed-by: Simon Hausmann <simon.hausmann@digia.com> Reviewed-by: Fawzi Mohamed <fawzi.mohamed@digia.com>
* | | V4: stack slot allocator for the interpreter.Erik Verbruggen2014-02-071-2/+8
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Use the life-time intervals to track which temps go out of scope, in order to re-use the stack-slots they occupied. This reduces the memory consumption on the JavaScript stack and allows for deeper call stacks. For the ECMA tests, this reduces the number of slots needed for the main/entry function from more than 650 to about 10 (depending on the test). Change-Id: Iff4044f5ff7f25e1f88ff1a04f4e80dd99a67590 Reviewed-by: Lars Knoll <lars.knoll@digia.com>
* | | V4: fix life range for phi node target that is never used.Erik Verbruggen2014-02-071-1/+7
| | | | | | | | | | | | | | | | | | | | | | | | When a temp is defined by a phi-node, but never used, still insert the (very short) life range. Change-Id: Ia976f496736a1606108fab7597c5d90048d9d55a Reviewed-by: Lars Knoll <lars.knoll@digia.com>
* | | V4: remove unused field.Erik Verbruggen2014-01-301-5/+3
|/ / | | | | | | | | | | Change-Id: Ic62ac6be99b79aa2f8c37fc386fef6b04b480247 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com> Reviewed-by: Lars Knoll <lars.knoll@digia.com>
* / V4: fix range splitting when split is between intervals.Erik Verbruggen2014-01-171-8/+39
|/ | | | | | | | | | | Also added some "white-box" unit tests and sprinkled in a bit of documentation. The case that went wrong is covered by the test rangeSplitting_1: before the fix, the new interval would have two ranges: [66-64],[70-71]. The first range is invalid and should not be there at all. Change-Id: If0742f4e6a96d98ea5d696f95126886ba66f92bb Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* V4 IR: do edge splitting after SSA transformationErik Verbruggen2014-01-161-8/+58
| | | | | | | | | | | | | | | This reduces the work for the dominator tree/frontier calculations, because there are less blocks to consider. All blocks inserted by splitting the critical edges, have (by definition) no effect on the dominator calculations. However, the immediate dominators for all new blocks needs to be added, because this information is used by the block scheduling. This change reduces memory/time usage during optimization passes, especially when processing excessively big switch statements. Change-Id: Ia69882e9dabdddffa1c98b1079012d8d988e1e8f Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* V4: lower memory allocator pressure.Erik Verbruggen2014-01-161-12/+18
| | | | | | | | | | | | | Changes to datastructures and more re-using of locally used temporary vectors. For the test regress-74474-002.js this lowers the total allocated memory from 1.98GB to 158MB. Thse peak memory usage stays at 75MB. There is no functional change. This should give a modest performance improvement which mainly depends on the speed of malloc()/free(). Change-Id: I1877c1903e59a33ee79ff2b801ef6f2c1cee30a6 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* V4 IR: update immediate dominators when purging unreachable basic-blocksErik Verbruggen2014-01-161-13/+19
| | | | | | | | | The basic block scheduling uses this information to place loops. When the immediate dominator information is invalid, the scheduling can be sub-optimal, or will sometimes forget to schedule some blocks. Change-Id: Iaeb45f2b757b676310be25a658ceadc07d5722ec Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* V4: remove unnecessary spills and order them correctly.Erik Verbruggen2014-01-161-9/+13
| | | | | | | | | | | | | | | | | | | When doing edge resolving, too many spills were generated, and the dependency tracking of moves was not complete. Now we only insert spills that are caused by phi-nodes (because any other spill would be generated at the point a variable was defined). However, there can still be multiple dependencies between the moves generated by the edge resolving. Instead of only checking the first dependency, all of them are tracked. The bug report was a case where an unneccesary spill was generated, that got tracked, but "suppressed" the other (valid!) dependent move. The randomness was caused by the hash seeding of QHash. Task-number: QTBUG-35840 Change-Id: Ifbc3c8fc13de53c46a8b5859721b2497189921a3 Reviewed-by: Fawzi Mohamed <fawzi.mohamed@digia.com> Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* V4: optimize dominator frontier storage.Erik Verbruggen2014-01-081-7/+164
| | | | | | | | | | | | | | | | | | Changes the dominator frontier storage from a set of basic-blocks for every basic-block to a BasicBlockSet for every basic-block. This new class stores a maximum of 8 nodes in a vector, and switches to a bit vector when going beyond 8 nodes. This is important in two cases: most basic-blocks have 2-3 nodes in the frontier, and an array is faster than a set in these cases. The few cases where the frontier goes beyond 8 nodes, is when a switch statement is used with lots of cases that all fall-through. On regress-74474-003.js this reduces peak memory usage from 1.68G to 60M. The switch statement in this test results in 27000 basic-blocks. Change-Id: I42646522ba9f8642d42a5d70fc6b760bb47ae69f Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* V4: remove class field in DominatorTree that was used only once.Erik Verbruggen2014-01-021-21/+18
| | | | | | | | | Calculation of all the children of nodes in the dominator tree is now calculated as a local variable right before computing the dominator frontier. The effect is that they are not retained after their only use. Change-Id: I83c962c691b78cb767708eb04cf30d3b7a760deb Reviewed-by: Lars Knoll <lars.knoll@digia.com>
* V4 IR: do not add unconditional jumps to work listsErik Verbruggen2014-01-021-1/+4
| | | | | | | | | | Both type inference and the optimization pass do not do anything with unconditional jumps. So, instead of adding them to the worklist and later on removing them again, it’s faster to never add them in the first place. Change-Id: Ib81d43e9ea6df2b1a70e9dd1e9b9c29cb6d345d2 Reviewed-by: Lars Knoll <lars.knoll@digia.com>
* V4 IR: change datastructures for dominator calculations.Erik Verbruggen2014-01-021-112/+172
| | | | | | | | Replace hashes from basic-block to basic-block with vectors that associate basic-block index to basic-block index. Change-Id: I834ea3d825e4d2b02c075b1b0f080f5a65f41317 Reviewed-by: Fawzi Mohamed <fawzi.mohamed@digia.com>
* V4 SSA: add some more literature references.Erik Verbruggen2013-12-191-5/+13
| | | | | | | Also fixed some comments. Change-Id: I4aedff84bdbf8de248025bc48805a859704bdd8a Reviewed-by: Lars Knoll <lars.knoll@digia.com>
* V4 IR: change block scheduling algorithm from recursive to iterative.Erik Verbruggen2013-12-121-102/+189
| | | | | | | | This makes time- and memory-complexity a lot better when compiling big JavaScript functions. Change-Id: I2a7cb9b5979844254747fa5cf7355cca0b113904 Reviewed-by: Fawzi Mohamed <fawzi.mohamed@digia.com>
* V4: change variable renumbering algorithm from recursive to iterative.Erik Verbruggen2013-12-101-89/+237
| | | | | | | | | Replace the recursive calls and subsequent clean-ups by pushing actions on a to-do stack, and processing that stack in a loop. Change-Id: I83536e88d400592b6e9f5fda3d795e41711a131a Reviewed-by: Simon Hausmann <simon.hausmann@digia.com> Reviewed-by: Fawzi Mohamed <fawzi.mohamed@digia.com>
generated by cgit v1.2.3 (git 2.25.1) at 2025-12-24 03:29:09 +0000