This also fixes ADCE to not remove possibly needed OpTypeForwardPointer.
The bug, its fix and the corresponding test have a circular dependency
with the extension, so they are packaged together.
If a member of a struct has a relaxed precision, sroa will not split the
struct. This means we do not get all cases. This commit handles these
cases. The other part is that the decoration needs to be passed on to
the new variables.
Fixes#2786
Fixes#2768
* In scalar replacement, interpret access chain indexes as signed counts
* Use Constant::GetSignExtendedValue and Constant::GetZeroExtendedValue
where appropriate
* new tests
spirv-opt: Add --graphics-robust-access
Clamps access chain indices so they are always
in bounds.
Assumes:
- Logical addressing mode
- No runtime-array-descriptor-indexing
- No variable pointers
Adds stub code for clamping coordinate and samples
for OpImageTexelPointer.
Adds SinglePassRunAndFail optimizer test fixture.
Android.mk: add source/opt/graphics_robust_access_pass.cpp
Adds Constant::GetSignExtendedValue, Constant::GetZeroExtendedValue
This fixes#2608.
The original test case had an out-of-bounds reference that ended up
folding into OpCompositeExtract that was indexing right outside the
constant composite.
The returned constant would then cause a segfault during constant
propagation.
Fixes#2764
* Don't replace all uses when simplifying instructions, instead only
update non-debug, non-decoration uses
* added a test
* Add a new version of RAUW that takes a predicate to decide whether to
replace the use or not
* used in simplification pass
* Fix#2609 - Handle out-of-bounds scalar replacements.
When SROA tries to do a replacement for an OpAccessChain that is exactly
one element out of bounds, the code was trying to access its internal
array of replacements and segfaulting.
This protects the code from doing this, and it additionally fixes the
way SROA works by not returning failure when it refuses to do a
replacement. Instead of failing the optimization pass, SROA will now
simply refuse to do the replacement and keep going.
Additionally, this patch fixes the SROA logic to now return a proper status so we can
correctly state that the pass made no changes to the IR if it only found
invalid references.
Merge return expects unreachable merge block to look a certain way, and
unreachable continue blocks to look a certain way. What if an
unreachable block is both a merge and a continue? The continue is
suppose to take precedent, but merge-return implements it with the merge
taking precedent. This change flips that around.
Fixes#2746
* Process OpDecorateId in ADCE
When there is an OpDecorateId instruction that is live,
the ids that is references must be kept live. This change
adds them to the worklist.
I've also updated a validator check to allow OpDecorateId
to be able to apply to decoration groups.
Fixes#1759.
* Remove dead code.
In merge return, we need to know the original dominator for a block in order to
traverse code from the original dominator to the new dominator and add
appropriate Phi nodes. The current code gets this wrong because the dominator
tree is build as needed. The first time we get the immediate dominator for a
function we just built the dominator tree and it takes into account that a
block has been split. The second time it does not.
This inconsistency needs to be fixed. We do that by recording the original
dominator for all blocks at the start of the pass.
If we were to record just the basic block, that could change if the block is
split. We want to traverse the code in the body of the original dominator,
whatever block it ends up in. To make this easy to track, we not save the
terminator instruction to represent the original dominator.
Fixes#2745
When a phi candidate is marked as trivial, we are suppose to update all
of its uses to the reference the value that it is being folded to.
However, the code updates the uses misses `defs_at_block_`. So at a
later time, the id for the trivial phi can reemerge.
Fixes#2744
* Bindless Instrument: Make init check depend solely on input_init_enabled
Previously was dependent on presense of descriptor_indexing extension
in SPIR-V, but this missed some cases. Tests updated to refect this new
policy.
* Fix format.
* Fix bug in merge return
The merge return pass seems to assume that the only new edges in the cfg
are from return block to merge blocks. However, it is possible that a
merge block branches to a merge block when it did not before.
This change add a new variable to track all of the new edges. It also
renames some other variables and cleans us the code to make it a bit
easier to read.
Fixes#2702.
Dead branch elimination needs to know about the constructs that a block is contained it when determining what to do with its merge instruction. We currently fold branches in block as we see them, which is parent constructs before their children. This causes the struct cfg analysis to crash because it tries to get the parent construct for a block after the parent has been folded.
This can be fixed by folding the branch of the children before the parents.
Fixes#2667.
There are a couple spots where we are not looking at decorations when we should.
1. Value numbering is suppose to assign a different value number to ids if they have different decorations. However that is not being done for OpCopyObject and OpPhi.
1. Instruction simplification is propagating OpCopyObject instruction without checking for decorations. It should only do that if no decorations are being lost.
Add a new function to the decoration manager to check if the decorations of one id are a subset of the decorations of another.
Fixes#2715.
Inlining does not inline functions that have a single return that is in a loop. This is because the return cannot be replaced by a branch outside of the loop easily. Merge return knows how to rewrite the function so the return is replaced by a branch.
Fixes#2038.
It is illegal to inline an OpKill instruction into a continue construct because the continue header will no longer dominate the backedge.
This commit adds a check for this, and does not inline.
If we still want to be able to inline a function that contains an OpKill, we can add a new pass that will wrap OpKill instructions into its own function with just the single instruction.
I do not believe that this is a common case right now, so I will not do that yet.
Fixes#2433.
When working on descriptor indexing validation for compute shaders, the
gl_GlobalInvocationID builtin was being loaded as uint which would cause
compute shaders instrumented by the bindless check pass to have:
%83 = OpLoad %uint %gl_GlobalInvocationID
%84 = OpCompositeExtract %uint %83 0
%85 = OpCompositeExtract %uint %83 1
%86 = OpCompositeExtract %uint %83 2
which results in validation failures:
error: line 127: Reached non-composite type while indexes still remain
to be traversed.
%84 = OpCompositeExtract %uint %83 0
for trying to extract a uint from a uint.
When it's an OpConstant or OpSpecConstant, then the literal
values are compared. If the OpSpecConstant also has a SpecId
decoration, then that's also compared.
Otherwise, it's an OpSpecConstantOp and we only compare the
ID of the OpSpecConstantOp instruction itself.
Fixes#2649
* Types: Avoid comparing IDs for in Type::IsSameImpl
When linking, we end up with duplicate types for imported and exported
types, that needs to be removed. The current code would reject valid
import/export pairs of symbols due to IDs mismatch, even if the types or
constants behind those ID were the same.
Enabled remaining type_match_test
Fixes#2442
New version has additional word in stage-specific section. Also
some changes in content for tesselation and compute shaders. Either
version can be invoked at pass creation. This is done to ease integration
and updating of validation layers. Version 1 is deprecated and eventually
will go away.
Also sneaking in fix to version 1 compute shaders.
* Handle nested breaks from switches.
There was a recent decision made to allow branches to the merge node of
a switch even if the switch is not the first enclosing construct. They
can be generated by glslang from break statements in switches.
Dead branch elimination seems to be the only optimization that will
break because of this change, so I will update that optimizations.
The change made are:
- Track switches in structured cfg analysis.
- In Dead branch elimination:
- Look for nested breaks that will require a switch instruction.
- Rewrite, but don't delete, switchs that are required even if it
could be replaced by an unconditional branch.
- When looking for the first break, consider the merge of a switch
as well.
See #2612.
* Fix variable names and comments.
* Add tests for the struct cfg analysis and switches.
* Fix typos in comments.
In order to try to reduce code duplication and to be able
to fold more cases, we want to use the instruction folder
when folding an OpSpecConstantOp with constant operands.
A couple other changes are need to make this work. First
GetDefiningInstruction| in the constant manager is able
to handle |type_id| being logically equivalent to another
type, so we updated the interface, and removed the assert.
Some tests were also updated because we not generate
better code because constants are not duplicated as much
as before.
No need for new tests. The functionality of the instruction folder is
already tested. There are tests check that the instruction folder is
being used correctly for OpCompositeExtract and OpVectorShuffle in the
existing test cases.
Fixes#2585.
* Update memory model support for SPIR-V 1.4
Fixes#2552
* Upgrade memory model now supports two memory access operands for
OpCopyMemory*
* in all cases the pass will first generate two operands by either
adding them or copying
* updates accounts for multiple operands
* tests
There is a case where sroa is not handling id overflow gracefully. It
is handled and an error message is output when the ids overflow.
Fixes https://crbug.com/961030.
Fixes#2555
* Fix a bug in validation where interfaces were considered non-unique
between different entry points targeting the same function
* added a test
* Update private to local pass to remove localized private variables
from entry point interfaces
* added tests
Fixes#2551
* Add support for 1.4 entry point interface lists
* only input and output variables are automatically live
* can clean up interfaces after DCE
* added tests
* allow opt tests to specify a target environment
* SPIR-V 1.4 headers, add SPV_ENV_UNIVERSAL_1_4
* Support --target-env spv1.4 in help for command line tools
* Support asm/dis of UniformId decoration
* Validate UniformId decoration
* Fix version check on instructions and operands
Also register decorations used with OpDecorateId
* Extension lists can differ between enums that match
Example: SubgroupMaskEq vs SubgroupMaskEqKHR
* Validate scope value for Uniform decoration, for SPIR-V 1.4
* More unioning of exts
* Preserve grammar order within an enum value
* 1.4: Validate OpSelect over composites
* Tools default to 1.4
* Add asm/dis test for OpCopyLogical
* 1.4: asm/dis tests for PtrEqual, PtrNotEqual, PtrDiff
* Basic asm/Dis test for OpCopyMemory
* Test asm/dis OpCopyMemory with 2-memory access
Add asm/dis tests for OpCopyMemorySized
Requires grammar update to add second optional memory access operand
to OpCopyMemory and OpCopyMemorySized
* Validate one or two memory accesses on OpCopyMemory*
* Check av/vis on CopyMemory source and target memory access
This is a proposed rule. See
https://gitlab.khronos.org/spirv/SPIR-V/issues/413
* Validate operation for OpSpecConstantOp
* Validate NonWritable decoration
Also permit NonWritable on members of UBO and SSBO.
* SPIR-V 1.4: NonWrtiable can decorate Function and Private vars
* Update optimizer CLI tests for SPIR-V 1.4
* Testing tools: Give expected SPIR-V version in message
* SPIR-V 1.4 validation for entry point interfaces
* Allow only unique interfaces
* Allow all global variables
* Check that all statically used global variables are listed
* new tests
* Add validation fixture CompileFailure
* Add 1.4 validation for pointer comparisons
* New tests
* Validate with image operands SignExtend, ZeroExtend
Since we don't actually know the image texel format, we can't fully
validate. We need more context.
But we can make sure we allow the new image operands in known-good
cases.
* Validate OpCopyLogical
* Recursively checks subtypes
* new tests
* Add SPIR-V 1.4 tests for NoSignedWrap, NoUnsignedWrap
* Allow scalar conditions in 1.4 with OpSelect
* Allows scalar conditions with vector operands
* new tests
* Validate uniform id scope as an execution scope
* Validate the values of memory and execution scopes are valid scope
values
* new test
* Remove SPIR-V 1.4 Vulkan 1.0 environment
* SPIR-V 1.4 requires Vulkan 1.1
* FIX: include string for spvLog
* FIX: validate nonwritable
* FIX: test case suite for member decorate string
* FIX: test case for hlsl functionality1
* Validation test fixture: ease debugging
* Use binary version for SPIR-V 1.4 specific features
* Switch checks based on the SPIR-V version from the target environment
to instead use the version from the binary
* Moved header parsing into the ValidationState_t constructor (where
version based features are set)
* Added new versions of tests that assemble a 1.3 binary and validate a
1.4 environment
* Fix test for update to SPIR-V 1.4 headers
* Fix formatting
* Ext inst lookup: Add Vulkan 1.1 env with SPIR-V 1.4
* Update spirv-val help
* Operand version checks should use module version
Use the module version instead of the target environment version.
* Fix comment about two-access form of OpCopyMemory
Add functionality to fix-storage-class so that it can fix up mismatched
data types for pointers as well.
Fixes bugs in when fixing up storage class.
Move GenerateCopy to the Pass class to be reused.
The spirv-opt change for #2535.
* Change implementation of post order CFG traversal
It seems like the recursion is going very deep, and causing some problem
is particular situations. I've reimplemented the CFG post order
traversal to not use recursion.
Fixes#2539.
WebGPU requires certain variables to be initialized, whereas there are
known issues with using initializers in Vulkan. This PR is the first
of three implementing a pass to decompose initialized variables into
a variable declaration followed by a store. This has been broken up
into multiple PRs, because there 3 distinct cases that need to be
handled, which require separate implementations.
This first PR implements the basic infrastructure that is needed, and
handling of Function storage class variables. Private and Output will
be handled in future PRs.
This is part of resolving #2388
In WebGPU, the component operand 0xFFFFFFFF is forbidden, but in
Vulkan it is used to indicate a value is undefined. When converting to
WebGPU, 0xFFFFFFFF needs to converted to a legal value, though the
specific one does not matter, since it was used to indicate an
undefined entry in the original code. Choosing to use 0, since the
operands are required to be on [0, N-1], so 0 is guaranteed to always
be valid.
Fixes#2349
Renames the existing flag '--webgpu-mode' to '--vulkan-to-webgpu' for
the Vulkan->WebGPU operation, and adds a new flag '--webgpu-to-vulkan'
for the WebGPU->Vulkan operation.
Currently '--webgpu-to-vulkan' doesn't have any passes associated with
it yet, but further patches will implement them.
Fixes#2495
This pass tries to fix validation error due to a mismatch of storage classes
in instructions. There is no guarantee that all such error will be fixed,
and it is possible that in fixing these errors, it could lead to other
errors.
Fixes#2430.
Fixes#2452
Swaps priority of handling unreachable merge and continues so that the
back-edge is retained in the case a block is both a loop continue and
loop merge
* Check var pointer capability in ADCE.
* Check var ptr capability for common uniform.
* Check var ptr capability in access chain convert.
Since we want this pass to run even if there are variable pointer on
storage buffers, we had to remove asserts that assumed there were no
variable pointers. The functions with the asserts will now work, it
becomes the responsibility of the callers to deal with the output as
appropriate.
* Single block elimination and variable pointers.
It seems like the code in local single block elimination is able to
handle cases with variable pointers already. This is because the
function `HasOnlySupportedRefs` ensures that variables that feed a
variable pointer are not candidates.
* Single store elimination and variable pointers.
It seems like the code in local single stroe elimination is able to
handle cases with variable pointers already. This is because the
function `FindSingleStoreAndCheckUses` ensures that variables that feed
a variable pointer are not candidates.
* SSA rewriter and variable pointers.
It seems like the code in the two passes that call the SSA rewriter are
able to handle cases with variable pointers already. This is because the
function `HasOnlySupportedRefs` ensures that variables that feed
a variable pointer are not candidates.
Fixes#2458.
Fixes#2456
* When eliminating a structured construct that has an unreachable merge,
replace that unreachable terminator with an appropriate return
* New tests
Fixes#2453
* Enable addition of OpPhi instructions when the loop has multiple
predecessors of the merge due to a break
* This can result in some values no longer dominating their uses
* Track return blocks in structured flow to produce OpPhis that have
multiple undef and non-undef arguments
* New tests to catch the bug
* When a block is predicated, mark the new body as a return if the old
block as already a return
If SPV_EXT_descriptor_indexing is enabled, add check that for a
descriptor-based reference, the descriptor is initialized. Initialization
data is stored in the debug input buffer, added to the length information
already there. This feature must be seperately enabled on the pass
creation routine. NOTE: Currently just supports image references; buffer
references are still TODO.
Adds an optimization pass to remove usages of AtomicCounterMemory
bit. This bit is ignored in Vulkan environments and outright forbidden
in WebGPU ones.
Fixes#2242
In constant propagation, decoration are transfered from the original
expression to the constant that will replace it. This can be wrong
because there are no decorations that apply to constants. We choose to
simply delete the decorations.
Fixes#2441
* Handle back edges better in dead branch elim.
Loop header must have exactly one back edge. Sometimes the branch
with the back edge can be folded. However, it should not be folded
if it removes the back edge.
The code to check this simply avoids folding the branch in the
continue block. That needs to be changed to not fold the back edge,
wherever it is.
At the same time, the branch can be folded if it folds to a branch to
the header, because the back edge will still exist.
Fixes#2391.
* Fix OpDot folding of half float vectors.
The code that folds OpDot does not handle half floats correctly. After
trying to multiple the first components, we get a nullptr because we
don't fold half float values. This nullptr gets passed to the code that
does the addition, and causes an assert.
Fixes#2405.
The types of input and output variables must match for the pipeline. We
cannot see the uses in all of the shader, so dead member
elimination cannot safely change the type of input and output variables.
Add a pass that looks for members of structs whose values do not affects
the output of the shader. Those members are then removed and just
treated like padding in the struct.
* Fixes#2338. Added check for phi node before merging blocks.
* Added functionality to merge blocks A and B even when B starts with OpPhi instructions, by replacing uses of the OpPhi results with the definitions coming from A. Added some tests for this.
* Fixed assertion.
* Remove use of deprecated googletest macro
INSTANTIATE_TEST_CASE_P has been deprecated. We need to use
INSTANTIATE_TEST_SUITE_P instead.
* Remove extra commas from test suites.
When looking at the uses of the result of an instruction, code sinking
assumes that all uses are in a basic block. However, this is not true
if there is a decoration or name for the result of that insturction.
This commit checks for this.
Fixes https://crbug.com/923243.
It is legal, but not generated by any SPIR-V producer: an OpCompositeExtract
with no indexes. This is essentially just a copy of the object, so we
treat them that way. We simply propagate the live variables of the
result to the operand.
Fixes https://crbug.com/919181.
It is legal, but not generated by any SPIR-V producer: an OpAccessChain
with no indexes. This is essentially just a copy of the pointer.
I have decided to treat it like an OpCopyObject. In CheckUses, we
return that it is not okay.
When looking at this I realized that we had code in GetUsedComponents
that cannot be reached. If there is a use in an OpCopyObject the it
will not call GetUsedComponents. I removed that dead code.
Fixes https://crbug.com/918311.
During unrolling a new loop is created, but its ownership is not clear
as it gets passed through the code. Changed something to unique_ptr to
make that clearer.
Fixes#2299.
Fixing other memory leaks at the same time.
Fixes#2296Fixes#2297
In C++, a bit shift of the same size as the type is undefined, but it is
defined in spir-v. When folding those cases, we have to be careful. We
cannot simply do the shift in C++.
Fixes https://crbug.com/917697.
Fixes#2138
* Modf and frexp are upgraded to use the struct version of the
instruction and generate an explicit store whose flags can be upgraded
separately
* Fixed major bug where availability and visibility were reversed for
non-copy memory instructions
* Fixed bug where availability and visibility scope operands were reversed for copy memory
* Upgraded all opt tests to use SPV_ENV_UNIVERSAL_1_3
* Upgrade tests moved into unified tests and removed standalone test
* Handle CompositeInsert with no indices in VDCE
In the spec, there it nothing that forces an OpCompositeInsert to have
an index, but VDCE assumes there is at least 1 in a couple places.
This commit updates VDCE to handle these cases.
* Added additional changes for the new AccelerationStructureNV type.
* Added additional changes for the new AccelerationStructureNV type. Change tabs to space...
* Added additional changes for the new accelerationStructureNV type -- add proper type name.
Fix TypeManager.TypeStrings test:
[----------] 29 tests from TypeManager
[ RUN ] TypeManager.TypeStrings
[ OK ] TypeManager.TypeStrings (7 ms)
When we are predicating the continue target for a loop, it can no longer
be the continue target because it will have a branch that exits the loop
and is not the bach edge. The continue target will have to be the
target of that branch that is still in the loop.
Fixes#2211.
The function `UpdatePhiNodes` was being called inconsistently. In one
case, the cfg had already been updated to include the new edge, and in
another place the cfg was not updated. This caused the function to
miss flagging a block as needing new phi nodes. I picked that the cfg
should not be updated before making the call. I documented it, and
change the call sites to match.
Fixes#2207.
We initially assumed that if the type manager returned the correct id
for the pointee type, that we would get the correct pointer type back,
but that is not true. See the unit test added with this commit. We
need to fall back to the linear search any time we are looking for a
pointer to a type that may not be unique.
At the same time, SROA considered an OpName on a variable to be a use of
the entire variable. That has been fixed.
Fixes#2209.
We currently place the load instructions at the start of the basic block
that dominates all of the loads. If that basic block contains OpPhi
instructions, then this will generate invalid code. We just need to
search for a location that comes after all of the OpPhi instructions.
Fixes#2204.
* Don't fold specialized branchs in loop unswitch
Folding branches can have a lot of special cases, and can be a little
error prone. So I only want it in one place. That will be in dead
branch elimination. I will change loop unswitching to set the branches
that were being folded to have a constant condition. Then subsequent
pass of dead branch elimination will be able to remove the code.
At the same time, I added a check that loop unswitching will not
unswitch a branch with a constant condition. It is not useful to do it
because dead branch elimination will simple fold the branch anyway.
Also it avoid an infinite loop that would other wise be introduced by my
first change.
Fixes#2203.
Loop unswitching is unswitching the conditional branch that creates the
back-edge. In the version of the loop, where the bachedge is not taken,
there is no back-edge. This is what causes the validator to complain.
The solution I will go with will be to now unswitch a condition with a
back-edge. At this time we do not now if loop unswitching is used. We do
not include it in the optimization sets provided, nor is it used in
glslang's set. When there are opportunities and no breaks from the loop,
the loop with either be a single iteration loop, or an infinite loop.
There is no performance advantage to performing loop unswitching in
either of those cases. If there is a break, maintaining structured
control flow will be tricky. Unless we see a clear advantage to handling
these case, I would go with the safer simpler solution.
Fixes#2201.
If there are multiple edges to a basic block, then the ssa rewriter will
create OpPhi instructions with duplicate entries. This is invalid, and
it is fixed in this commit.
Fixes#2202.
* Invalidate the decoration manager at the start of ADCE.
If the decoration manager is kept live the the contex will try to keep
it up to date. ADCE deals with group decorations by changing the
operands in |OpGroupDecorate| instructions directly without informing
the decoration manager. This puts it in an invalid state, which will
cause an error when the context tries to update it. To Avoid this
problem, we will invalidate the decoration manager upfront.
At the same time, the decoration manager is now considered when checking
the consistency of the decoration manager.
* Fix invalid OpPhi generated by merge-return.
When we create a new phi node for a value say %10, we have to replace
all of the uses of %10 that are no longer dominated by the def of %10
by the result id of the new phi. However, if the use is in a phi node,
it is possible that the bb contains the use is not dominated by either.
In this case, needs to be handled differently.
* Split loop headers before add a new branch to them.
In merge return, Phi node in loop header that are also merges for loop
do not get updated correctly. Those cases do not fit in with our
current analysis. Doing this will simplify the code by reducing the
number of cases that have to be handled.
Added documentation to the ir context to indicates that TakeNextId()
returns 0 when the max id is reached. TODOs were added to each call
sight so that we know where we have to start to handle this case.
Handle id overflow in |SplitLoopHeader|.
Handle id overflow in |GetOrCreatePreHeaderBlock|.
Handle failure to create preheader in LICM.
Part of https://github.com/KhronosGroup/SPIRV-Tools/issues/1841.
We currently simulate all shift operations when the two operand are
constants. The problem is that if the shift amount is larger than
32, the result is undefined.
I'm changing the folder to return 0 if the shift value is too high.
That way, we will have defined behaviour.
https://crbug.com/910937.
Upgrade to VulkanKHR memory model
* Converts Logical GLSL450 memory model to Logical VulkanKHR
* Adds extension and capability
* Removes deprecated decorations and replaces them with appropriate
flags on downstream instructions
* Support for Workgroup upgrades
* Support for copy memory
* Adding support for image functions
* Adding barrier upgrades and tests
* Use QueueFamilyKHR scope instead of device
* Move ProcessFunction* function from pass to the context.
There are a few functions that are used to traverse the call tree.
They currently live in the Pass class, but they have nothing to do with
a pass, and may be needed outside of a pass. They would be better in
the ir context, or in a specific call tree class if we ever have a need
for it.
* Don't inline recursive functions.
Inlining does not check if a function is recursive or not. This has
been fine as long as the shader was a Vulkan shader, which forbid
recursive functions. However, not all shaders are vulkan, so either
we limit inlining to Vulkan shaders or we teach it to look for recursive
functions.
I prefer to keep the passes as general as is reasonable. The change
does not require much new code in inlining and gives a reason to refactor
some other code.
The changes are to add a member function to the Function class that
checks if that function is recursive or not.
Then this is used in inlining to not inlining a function call if it calls
a recursive function.
* Add id to function analysis
There are a few places that build a map from ids to Function whose
result is that id. I decided to add an analysis to the context for this
to reduce that code, and simplify some of the functions.
* Add missing file.
Fixes#2104
* Checks the rules for logical addressing and variable pointers
* Has an out for relaxed logical pointers
* Updated PassFixture to expose validator options
* enabled relaxed logical pointers for some tests
* New validator tests
These are bookend passes designed to help preserve line information
across passes which delete, move and clone instructions. The propagation
pass attaches a debug line instruction to every instruction based on
SPIR-V line propagation rules. It should be performed before optimization.
The redundant line elimination pass eliminates all line instructions
which match the previous line instruction. This pass should be performed
at the end of optimization to reduce physical SPIR-V file size.
Fixes#2027.
That function currently only handled OpPtrAccessChain if it was in the
middle of the chain, but not at the start. Fixing that up.
Fixes crbug.com/905271.
* Add base and core bindless validation instrumentation classes
* Fix formatting.
* Few more formatting fixes
* Fix build failure
* More build fixes
* Need to call non-const functions in order.
Specifically, these are functions which call TakeNextId(). These need to
be called in a specific order to guarantee that tests which do exact
compares will work across all platforms. c++ pretty much does not
guarantee order of evaluation of operands, so any such functions need to
be called separately in individual statements to guarantee order.
* More ordering.
* And more ordering.
* And more formatting.
* Attempt to fix NDK build
* Another attempt to address NDK build problem.
* One more attempt at NDK build failure
* Add instrument.hpp to BUILD.gn
* Some name improvement in instrument.hpp
* Change all types in instrument.hpp to int.
* Improve documentation in instrument.hpp
* Format fixes
* Comment clean up in instrument.hpp
* imageInst -> image_inst
* Fix GetLabel() issue.
If there is only 1 return and it is in a loop, then the function cannot be inlined.
Fix condition when inlined code needs one-trip loop wrapper. The dummy loop is needed when there is a return inside a selection construct. Even if there is only 1 return.
This CL takes the various opt unit tests and makes a single executable
instead of one per test. This reduces the number of build targets by
~125 when building with ninja.
When looking for a break from a selection construct, we do not realize
that a jump to the continue target of a loop containing the selection
is a break. This causes and infinit loop, or possibly other failures.
Fixes#2004.
When looking for a break from a selection construct, we do not need to
look inside nested constructs. However, if a loop header has an
unconditional branch, then we enter the loop. Entering the loop causes
an infinite loop because we keep going through the loop.
The solution is to look for a merge block, if one exsits, even for block
terminated by an OpBranch.
Fixes#1979.
ADCE liveness algorithm should treat OpUnreachable at least like other
branch instructions. It was being treated as always live which was
preventing useless structured constructs from being eliminated.
OpUnreachable is generated by dead branch elimination which is now
being required by merge return, so this fix should accompany that
change.
We currently run merge-return on all functions, but
dead-branch-elimination only runs on function reachable from an entry
point or exported function. Since dead-branch-elimination is needed for
merge-return, they have to match.
Fixes#1976.
Was removing control structures which didn't have data dependency
with enclosed live loop and otherwise did not contain live code.
An example is a counting loop around a live loop.
Fixes#1967.
Merge return assumes that the only unreachable blocks are those needed
to keep the structured cfg valid. Even those must be essentially empty
blocks.
If this is not the case, we get unpredictable behaviour. This commit
add a check in merge return, and emits an error if it is not the case.
Added a pass of dead branch elimination before merge return in both the
performance and size passes. It is a precondition of merge return.
Fixes#1962.
The current implementation in the folder when seeing a division by zero
is to assert. In the release build, the compiler will attempt to
compute the value, which causes its own problems.
The solution I will go with is to fold the division, and just give it
the value of 0. The same goes for remainder and mod operations.
Fixes#1961.
The HlslCounterBufferGOOGLE that was introduced changed the OpDecorateId
so that is can now reference an id other than the target. If that other
id is used only in the decoration, then the definition of the id will be
removed because decoration do not count as real uses.
However, if the target of the decoration is still live the decoration
will not be removed. This leaves a reference to an id that is not
defined.
There are two solutions to consider. The first is that is the decoration
is kept, then the definition of the id should be kept live. Implementing
this change would be involved because the way ADCE handles decorations
will have to be reimplemented.
The other solution is to remove the decoration the id is otherwise dead.
This works for this specific case. Also this is the more desirable
behaviour in this case. The id will always be the id of a variable that
belongs to a descriptor set. If that variable is not bound and we do
not remove it, the driver will complain.
I chose to implement the second solution. The first will be left to when
a case for it comes up.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/1885.
* Check rules from Execution Mode tables, 2.16.2 and the Vulkan
environment spec
* Allows MeshNV execution model with the following execution modes
* LocalSize, LocalSizeId, OutputPoints and OutputVertices
* Done to not break their validation
There are a few spots where copy propagate arrays is trying
to go from a Type to an id, but the type is not unique. When generating
code this pass needs specific ids, otherwise we get type mismatches.
However, the ambigous types means we can sometimes get the wrong type
and generate invalid code.
That code has been rewritten to not rely on the type manager, and just
look at the instructions instead.
I have opened https://github.com/KhronosGroup/SPIRV-Tools/issues/1939 to
try to get a way to make this more robust.
In DecorationManager::RemoveDecorationsFrom, we do not remove the id
from a decoration group if the group has no decorations. This causes
problems because KillNamesAndDecorates is suppose to remove all
references to the id, but in this case, there is still a reference.
This is fixed by adding a special case.
Also, there is the possibility of a double free because
RemoveDecorationsFrom will delete the instructions defining |id| when
|id| is a decoration group. Later, KillInst would later write to memory
that has been deleted when trying to turn it into a Nop. To fix this,
we will only remove the decorations that use |id| and not its definition
in RemoveDecorationsFrom.
It seems like the current implementation of KillNameAndDecorates does
not handle group decorations correctly. The id being removed is not
removed from the OpGroupDecorate instructions. Even worst, any
decorations that apply to that group are removed.
The solution is to use the function in the decoration manager that will
remove the decorations and update the instructions instead of doing the
work itself.
Adds unrolling to the legalization passes.
After enabling unrolling I found a bug when there is a self-referencing
phi node. That has been fixed.
The test that checks for that the order of optimizations is correct also
needed to be updated.
The current implementation of merge return can create bad, but correct,
code. When it is not in a loop construct, it will insert a lot of
extra branch around code. The potentially large number of branches are
bad. At the same time, it can separate code store to variables from
its uses hiding the fact that the store dominates the load.
This hurts the later analysis because the compiler thinks that multiple
values can reach a load, when there is really only 1. This poorer
analysis leads to missed optimizations.
The solution is to create a dummy loop around the entire body of the
function, then we can break from that loop with a single branch. Also
only new merge nodes would be those at the end of loops meaning that
most analysies will not be hurt.
Remove dead code for cases that are no longer possible.
It seems like some drivers expect there the be an OpSelectionMerge
before conditional branches, even if they are not strictly needed.
So we add them.
* Create structed cfg analysis.
There are lots of optimization that have to traverse the CFG in a
structured order just because it wants to know which constructs a
basic block in contained in. This adds extra complexity to these
optimizations, for causes too much refactoring of older optimizations.
To help with this problem, I have written an analysis that can give this
information.
* Identify branches breaking from loops.
Dead branch elimination does a search for a conditional branch to the
end of the current selection construct. This search assumes that the
only way to leave the construct is through the merge node. But that is
not true. The code can jump to the merge node of a loop that contains
the construct.
The search needs to take this into consideration.
In merge blocks, we do not allow the merging of two blocks with merge
instructions. This is because if the two block are merged only 1 of
those instructions can exists. However, if the successor block is the
merge block of the predecessor, then we can delete the merge instruction
in the predecessor. In this case, we are able to merge the blocks.
* Create a new entry point for the optimizer
Creates a new struct to hold the options for the optimizer, and creates
an entry point that take the optimizer options as a parameter.
The old entry point that takes validator options are now deprecated.
The validator options will be one of the optimizer options.
Part of the optimizer options will also be the upper bound on the id bound.
* Add a command line option to set the max value for the id bound. The default is 0x3FFFFF.
* Modify `TakeNextIdBound` to return 0 when the limit is reached.
* Copy decorations when creating new ids.
When creating a new value based on an old value, we need to copy the
decorations to the new id. This change does this in 3 places:
1) The variable holding the return value of the function generated by
merge return should get decorations from the function.
2) The results of the OpPhi instructions should get decorations from the
variable they are replacing in the ssa writer.
3) In local access chain convert the intermediate struct (result of
OpCompositeInsert) generated for the store replacement should get its
decorations from the variable being stored to.
Fixes#1787.
If seems like at least 1 driver does not like a condition jump to the end
of a selection construct. We are generating these in the merge return
pass. This change stops merge return from generating this sequence.
Part of #1861.
When doing predicate blocks, we need to traverse every block in
structured order in order to keep track of which construct a block is
contained in. The standard way of traversing code in structured order
is to create a list with all of the nodes in order. However, when
predicating blocks, new blocks are created, and those blocks are missed.
This causes branches that go too far.
The solution is to update the order as new blocks are created. Since
we are using an std::list, we do not have to worry about invalidation of
iterators when changing the list.
* Refactor PredicateBlocks
Refactor PredicateBlocks so that we know which constructs a return
is contained in. Will be used later.
* Have PredicateBlocks jump the existing merge blocks.
In PredicateBlocks, we currently skip instructions with side effects,
but it still follows the same control flow (sort-of). This causes a
problem, when we are trying to predicate code in a loop. We skip all
of the code with side effects (IV increment), but still follow the
same control flow (jump back the start of the loop). This creates an
infinite loop because the code will keep jumping back to the start of
the loop without changing the values that effect the exit condition.
This is a large change to merge-return. When predicating a block that
is in a loop or merge construct, it will jump to the merge block of the
construct. Once out of all constructs we will generate code as we did
before.
* Handle breaks from structured-ifs in DCE.
dead code elimination assumes that are conditional branches except for
breaks and continues in loops will have an OpSelectionMerge before them.
That is not true when breaking out of a selection construct.
The fix is to look for breaks in selection constructs in the same place
we look for breaks and continues for loops.
When dead-branch-elim folds a conditional branch, it also deletes the
OpSelectionMerge instruction. If that construct contains a
conditional branch to the merge node, it will not have its own
OpSelectionMerge. When the headers merge instruction is deleted, the
the inner conditional branch will no longer be legal. It will be a
selection to a node that is not a merge node.
We fix this up by moving the OpSelectionMerge to a new location if it is
still needed.
In local-access-chain-convert, we replace loads by load the entire
variable, then doing the extract. The extract will have the same value
as the load. However, if the load has a decoration on it, the
decoration is lost because we do not copy any them to the new id.
This is fixed by rewritting the load into the extract and keeping the
same result id.
This change has the effect that we do not call DCEInst on the loads
because the load is not being deleted, but replaced. This could leave
OpAccessChain instructions around that are not used. This is not a
problem for -O and -Os. They run local_single_*_elim passes and then
dead code elimination. The dce will remove the unused access chains,
and the load elimination passes work even if there are unused access
chains. I have added test to them to ensure they will not loss
opportunities.
Fixes#1787.
The code in source/message was only used in a single set of tests to
format the output results. This CL changes the test to verify the
message instead of all the error values and removes the source/message
code.
* Run the validator in the optimization fuzzers.
The optimizers assumes that the input to the optimizer is valid. Since
the fuzzers do not check that the input is valid before passing the
spir-v to the optimizer, we are getting a few errors.
The solution is to run the validator in the optimizer to validate the
input.
For the legalization passes, we need to add an extra option to the
validator to accept certain types of variable pointers, even if the
capability is not given. At the same time, we changed the option
"--legalize-hlsl" to relax the validator in the same way instead of
turning it off.
In the merge return pass, we will split a block, but not update the phi
instructions that reference the block. Since the branch in the original
block is now part of the block with the new id, the phi nodes must be
updated.
This commit will change this.
I have also considered other places where an id of a basic block could
be referenced, and I don't think any of them need to change.
1) Branch and merge instructions: These jump to the start of the
original block, and so we want them to jump to the block that uses the
original id. Nothing needs to change.
2) Names and decorations: I don't think it matters with block keeps the
name, and there are no decorations that apply to basic blocks.
Fixes#1736.
When creating a new phi for a value in the function, merge return will
rewrite all uses of an id that are no longer dominated by its
definition. Uses that are not in a basic block, like OpName or
decorations, are not dominated, but they should not be replaced.
Fixes#1736.
* Combines OpAccessChain, OpInBoundsAccessChain, OpPtrAccessChain and
OpInBoundsPtrAccessChain
* New folding rule to fold add with 0 for integers
* Converts to a bitcast if the result type does not match the operand
type
V
This re-implements the -Oconfig=<file> flag to use a new API that takes
a list of command-line flags representing optimization passes.
This moves the processing of flags that create new optimization passes
out of spirv-opt and into the library API. Useful for other tools that
want to incorporate a facility similar to -Oconfig.
The main changes are:
1- Add a new public function Optimizer::RegisterPassesFromFlags. This
takes a vector of strings. Each string is assumed to have the form
'--pass_name[=pass_args]'. It creates and registers into the pass
manager all the passes specified in the vector. Each pass is
validated internally. Failure to create a pass instance causes the
function to return false and a diagnostic is emitted to the
registered message consumer.
2- Re-implements -Oconfig in spirv-opt to use the new API.
Fixes#1731
* Updated folding rules related to vector shuffle to account for the
undef literal value:
* FoldVectorShuffleFeedingShuffle
* FoldVectorShuffleFeedingExtract
* FoldVectorShuffleWithConstants
* These rules would commit memory violations due to treating the undef
literal value as an accessible composite component
Fixes#1727
* If the pass finds any dead branches it can optimize then at the end of
the pass it reorders basic blocks to ensure they satisfy block ordering
requirements
* Added some new tests
* While investigating this issue, found and fixed a non-deterministic
ordering of dominators
* Now the edges used to construct the dominator tree are sorted
according to posorder traversal indices
With current implementation, the constant manager does not keep around
two constant with the same value but different types when the types
hash to the same value. So when you start looking for that constant you
will get a constant with the wrong type back.
I've made a few changes to the constant manager to fix this. First off,
I have changed the map from constant to ids to be an std::multimap.
This way a single constant can be mapped to mutiple ids each
representing a different type.
Then when asking for an id of a constant, we can search all of the ids
associated with that constant in order to find the one with the correct
type.
When folding an OpVectorShuffle where the first operand is defined by
an OpVectorShuffle, is unused, and is equal to the second, we end up
with an infinite loop. This is because we think we change the
instruction, but it does not actually change. So we keep trying to
folding the same instruction.
This commit fixes up that specific issue. When the operand is unused,
we replace it with Null.
When folding a vector shuffle that feeds another vector shuffle causes
the size of the first operand to change, when other indices have to be
adjusted reletive to the new size.
The function class provides a {Set|Get}Parent call in order to provide
the context to the LoopDescriptor methods. This CL removes the module
from Function and provides the needed context directly to LoopDescriptor
on creation.
This CL removes the context() method from opt::Function. In the places
where the context() was used we can retrieve, or provide, the context in
another fashion.
Currently the IRContext is passed into the Pass::Process method. It is
then up to the individual pass to store the context into the context_
variable. This CL changes the Run method to store the context before
calling Process which no-longer receives the context as a parameter.
For the instructions which execute after the IdPass check we can provide
the Instruction instead of the spv_parsed_instruction_t. This
Instruction class provides a bit more context (like the source line)
that is not available from spv_parsed_instruction_t.
This CL moves the files in opt/ to consistenly be under the opt::
namespace. This frees up the ir:: namespace so it can be used to make a
shared ir represenation.
The folding routines are currently global functions. They also rely on
data in an std::map that holds the folding rules for each opcode. This
causes that map to not have a clear owner, and therefore never gets
deleted.
There has been a request to delete this map. To implement this, we will
create a InstructionFolder class that owns the maps. The IRContext will
own the InstructionFolder instance. Then the global functions will
become public memeber functions of the InstructionFolder.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/1659.
There are a few locations where we need to handle duplicate types. We
cannot merge them because they may be needed for reflection. When this
happens we need do some extra lookups in the type manager.
The specific fixes are:
1) When generating a constant through `GetDefiningInstruction` accept
and use an id for the desired type of the constant. This will make sure
you get the type that is needed.
2) In Private-to-local, make sure we to update the def-use chains when a
new pointer type is created.
3) In the type manager, make sure that `FindPointerToType` returns a
pointer that points to the given type and not a duplicate type.
4) In scalar replacment, make sure the null constants that are created
are the correct type.
Revert "Don't merge types of resources"
This reverts commit f393b0e480, but leaves
the tests that were added. Added new test. These test are the so that,
if someone tries the same change I made, they will see the test that
they need to handle.
Don't run remove duplicates in -O and -Os
Romve duplicates was run to help reduce compile time when looking for
types in the type manager. I've run compile time test on three sets
of shaders, and the compile time does not seem to change.
It should be safe to remove it.
The Compact IDs pass can corrupt the CFG, but first the CFG has to
be setup. To do this, a test that builds the CFG, then performs the
compact IDs pass, then checks context consistency.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/1648
When doing reflection users care about the names of the variable, the
name of the type, and the name of the members. Remove duplicates breaks
this because it removes the names one of the types when merging.
To fix this we have to keep the different types around for each
resource. This commit adds code to remove duplicates to look for the
types uses to describe resources, and make sure they do not get merged.
However, allow merging of a type used in a resource with something not
used in a resource. Was done when the non resource type came second.
This could have a negative effect on compile time, but it was not
expected to be much.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/1372.
Add SPV_ENV_WEBGPU_0 for work-in-progress WebGPU.
val: Disallow OpUndef in WebGPU env
Silence unused variable warnings when !defined(SPIRV_EFFCE)
Limit visibility of validate_instruction.cpp's symbols
Only InstructionPass needs to be visible so all other functions are put
in an anonymous namespace inside the libspirv namespace.
Fixes#1120
Checks that all static uses of the Input and Output variables are listed
as interfaces in each corresponding entry point declaration.
* Changed validation state to track interface lists
* updated many tests
* Modified validation state to store entry point names
* Combined with interface list and called EntryPointDescription
* Updated uses
* Changed interface validation error messages to output entry point name
in addtion to ID
Try to reduce the amount of disk space used by especially by debug builds,
which may be contributing to AppVeyor failures.
Collapses tests in categories:
- validator
- loop optimizations
- dominator analysis
- linker
Contributes to #1615
Fixes#1281
* New structured cfg check: all non-construct header blocks'
predecessors must come from within the construct
* New function to calculate blocks in a construct
* Fixed a bug in BasicBlock type bitset
Relaxing check to not consider unreachable predecessors
* Fixing broken common uniform elim test
By using forward pointers, we are able to define a struct that has a
pointer to itself. This could be directly or indirectly. The current
implementation of the type manager did not handle this case. There are
three changes that are made in this commit inorder to handle this case:
1) Change the handling of OpTypeForwardPointer
The current handling of OpTypeForwardsPointer is broken if there is a
reference to the pointer before the real definition. When build the
type that contain the forward delared pointer, the type manager will ask
for the type for that ID, and will get a nullptr because it does not
exists. This nullptr is not handleded very well.
The change is to keep track of the incomplete types the first time
through all of the types. An incomplete type is a ForwardPointer or any
type that references an incomplete type.
Then we implement a second pass through the incomplete types that will
complete them.
2) Hashing types.
When hashing a type, we want to uses all of the subtypes as part of the
hash. However, with types that reference them selves, this creates an
infinite recursion. To get around this, we keep track of which types
have been seen on the path from the root type. If we have see the
current type already then we can stop the recursion.
3) Comparing types.
In order to check if two types are the same, we must check that all of
their subtypes are the same as well. This also causes an infinit
recursion. The solution is to stop comparing the subtypes if we are
trying to compare two pointer types that we are already in the middle of
comparing. The ideas is that if the two pointer are different, then in
progress compare will return false itself.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/1578.
We add a new rule to the folding rules to fold an FMix feeding an
extract when the alpha value for the element being extracted is either
0 or 1. In those case, we can simple extract from one of the operands
to the FMix.
With that change the simplification pass completely subsumes the
insert-extract elimination pass. So we remove the insert-extract
elimination passes and replce them with calls to the simplification
pass.
In a follow up PR, we should delete the insert-extract elimination pass.
Contributes to https://github.com/KhronosGroup/SPIRV-Tools/issues/1570.
Removes the limit on scalar replacement for the lagalization passes.
This is done by adding an option to the pass (and command line option)
to set the limit on maximum size of the composite that scalar
replacement is willing to divide.
Fixes#1494.
ADCE does not treat OpCopyMemory as an instruction that references
memory. Because of that stores are removed that should not be.
This change teaches ADCE that OpCopyMemory and OpCopyMemorySize both
loads from and stores to memory. This will keep other stores live when
needed, and will allows ADCE to remove OpCopyMemory instructions as
well.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/1556.
Currently in scalar replacement, we create a new variable for every
memeber of the composite being divided. It is often overkill, because
not all of those members will be used. This change will check which
elements are used and only create variable for the members that are
used.
This reduces the compile time for one set of shader from 248s to 165s.
Part of https://github.com/KhronosGroup/SPIRV-Tools/issues/1494.
The code patterns generated by DXC around function calls can cause many
store to be storing the same value that was just loaded from the same
location:
```
%10 = OpLoad %type %var
OpStore %var %10
```
We want to clean these up very early on because they can cause other
transformations to do a lot of work. For the cases I see, they can be
removed during local-single-block-elim.
For one set of shaders the compile time goes from 248s to 182s. A 26%
improvement.
Part of https://github.com/KhronosGroup/SPIRV-Tools/issues/1494.
We have already disabled common uniform elimination because it created
sequences of loads an entire uniform object, then we extract just a
single element. This caused problems in some drivers, and is just
generally slow because it loads more memory than needed.
However, there are other way to get into this situation, so I've added
a pass that looks specifically for this pattern and removes it when only
a portion of the load is used.
Fixes#1547.
An FClamp instruction forces a values to be within a certain interval.
When the upper or lower bound of the FClamp is a constant and the value
being compared with is a constant, then in some case we can fold the
compared because the entire range is say less than the value.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/1549.
If there is a shader with a variable in the workgroup storage class that
is stored to, but not loadeds, then we know nothing will read those
loads. It should be safe to remove them.
This is implemented in ADCE by treating workgroup variables the same
way that private variables are treated.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/1550.
When doing if-conversion, we do not currently move code out of the side
nodes. The reason for this is that it can increase the number of
instructions that get executed because both side nods will have to be
executed now.
In this commit, we add code to move an instruction, and all of the
instructions it depends on, out of a side node and into the header of
the selection construct. However to keep the cost down, we only do it
when the two values in the OpPhi node compute the same value. This way
we have to move only one of the instructions and the other becomes
unused most of the time. So no real extra cost.
Makes the value number table an alalysis in the ir context.
Added more opcodes to list of code motion safe opcodes.
Fixes#1526.
Previously, the loop class used the terms latch and continue block
interchangeably. This patch splits the two and corrects and tests some
uses of the old uses of GetLatchBlock.
This pass will look for adjacent loops that are compatible and legal to
be fused.
Loops are compatible if:
- they both have one induction variable
- they have the same upper and lower bounds
- same initial value
- same condition
- they have the same update step
- they are adjacent
- there are no break/continue in either of them
Fusion is legal if:
- fused loops do not have any dependencies with dependence distance
greater than 0 that did not exist in the original loops.
- there are no function calls in the loops (could have side-effects)
- there are no barriers in the loops
It will fuse all such loops as long as the number of registers used for
the fused loop stays under the threshold defined by
max_registers_per_loop.
Adds support for spliting loops whose register pressure exceeds a user
provided level. This pass will split a loop into two or more loops given
that the loop is a top level loop and that spliting the loop is legal.
Control flow is left intact for dead code elimination to remove.
This pass is enabled with the --loop-fission flag to spirv-opt.
Track live scalars in VDCE as if they were single element vectors.
Handle the extended instructions for GLSL in VDCE.
Handle composite construct instructions in VDCE.
If one of the operands to an OpVectorTimesScalar instruction is zero,
then the result will be the 0 vector. Currently we do not fold the
insturction unless both operands are constants. This change fixes that.
We also allow folding of OpPhi instructions where the incoming values
are either an OpUndef or the OpPhi instruction itself. As with other
cases, this can be simplified to the OpUndef.
Track live scalars in VDCE as if they were single element vectors.
Handle the extended instructions for GLSL in VDCE.
Handle composite construct instructions in VDCE.
Fixes#1511.
Eliminate unused store to variable if followed by store to same
variable in same block.
Most significantly, this cleans up stores made unused by this pass.
These useless stores can inhibit subsequent optimizations, specifically
LocalSingleStoreElim. Eliminating them makes subsequent optimization more
effective.
The main effect of this pass is to simplify the work done by the SSA
rewriter. It catches many local loads/stores that help speeding up the
work done by the main rewriter.
Introduce a pass that does a DCE type analysis for vector elements
instead of the whole vector as a single element.
It will then rewrite instructions that are not used with something else.
For example, an instruction whose value are not used, even though it is
referenced, is replaced with an OpUndef.
For each function, the analysis determine which SSA registers are live
at the beginning of each basic block and which one are killed at
the end of the basic block.
It also includes utilities to simulate the register pressure for loop
fusion and fission.
The implementation is based on the paper "A non-iterative data-flow
algorithm for computing liveness sets in strict ssa programs" from
Boissinot et al.
The local-single-store-elim algorithm is not fundamentally bad.
However, when there are a large number of variables, some of the
maps that are used can become very large. These large data structures
then take a very long time to be destroyed. I've seen cases around 40%
if the time.
I've rewritten that algorithm to not use as much memory. This give a
significant improvement when running a large number of shader through
DXC.
I've also made a small change to local-single-block-elim to delete the
loads that is has replaced. That way local-single-store-elim will not
have to look at those. local-single-store-elim now does the same thing.
The time for one set goes from 309s down to 126s. For another set, the
time goes from 102s down to 88s.
GCD MIV test as described in Chapter 3 of "Optimizing Compilers for
Modern Architectures: A Dependence-Based Approach" by Randy Allen, and
Ken Kennedy.
Delta test as described in Figure 3 of "Practical Dependence Testing" by
Gina Goff, Ken Kennedy, and Chau-Wen Tseng from PLDI '91.
For each loop in a function, the pass walks the loops from inner to outer most loop
and tries to peel loop for which a certain amount of iteration can be done before or after the loop.
To limit code growth, peeling will not happen if the growth in code size goes above a configurable threshold.
Provides functionality to perform ZIV and SIV dependency analysis tests
between a load and store within the same loop.
Dependency tests rely on scalar analysis to prove and disprove dependencies
with regard to the loop being analysed.
Based on the 1990 paper Practical Dependence Testing by Goff, Kennedy, Tseng
Adds support for marking loops in the loop nest as IRRELEVANT.
Loops are marked IRRELEVANT if the analysed instructions contain
no induction variables for the loops, i.e. the loops induction
variable is not relevent to the dependence of the store and load.
Adding three rules to fold OpDot (implemented as two).
- When an OpDot has two constants, then fold to the resulting const.
- When one of the inputs is the 0 vector, then fold to zero.
- When one of the inputs is a single 1 with 0s, then rewrite to an
OpCompositeExtract of the appropriate element. This will help find
even more folding opportunities.
Contributes to #709.
From the test case, the slice of the CFG that is interesting for the bug
is
25
|
v
30
|
v
31<-+
| |
v |
34--+
1. In block 25, we have a Phi candidate for %f with arguments
%47 = Phi[%float_0, %0]. This merges %float_0 and a yet unknown
argument from the external loop backedge.
2. We are now processing block 34:
i. The load %35 = OpLoad %f triggers a Phi candidate to be placed in
block 31.
ii. The Phi candidate %50 = Phi needs two arguments. The one coming
from block 30 is %47. But the one coming from block 34 (which we
are now processing and have marked sealed), finds %50 itself as
the reaching def for %f.
3. This wrongfully marks %50 as a copy-of Phi, which ultimately makes
both %47 and %50 copy-of Phis that get eliminated.
Currently OpImageTexelPointer operations are treat like a use of the
pointer, but it does
not look for the memory being referenced to make sure stores are not
removed.
This change teaches it so identify the memory being accessed, and
treats it as if that memory is loaded.
Fixes to #1445.
OpImageTexelPointer acts like a special kind of load. It is not an
array load, but it also cannot be removed the same way a regular
load can. The type of propagation that needs to be done is similar
to what we do for arrays, so I want to merge that code into that
optmization.
Contributers to #1445.
OpImageTexelPointer acts like a special kind of load. It is still
safe to change the storage class of a variable used in a
OpImageTexalPointer instruction.
Contributes to #1445.
CPPreference.com has this description of digits10:
“The value of std::numeric_limits<T>::digits10 is the number of
base-10 digits that can be represented by the type T without change,
that is, any number with this many significant decimal digits can be
converted to a value of type T and back to decimal form, without
change due to rounding or overflow.”
This means that any number with this many digits can be represented
accurately in the corresponding type. A change in any digit in a
number after that may or may not cause it a different bitwise
representation. Therefore this isn’t necessarily enough precision to
accurately represent the value in text. Instead we need max_digits10
which has the following description:
“The value of std::numeric_limits<T>::max_digits10 is the number of
base-10 digits that are necessary to uniquely represent all distinct
values of the type T, such as necessary for
serialization/deserialization to text.”
The patch includes a test case in hex_float_test which tries to do a
round-robin conversion of a number that requires more than 6 decimal
places to be accurately represented. This would fail without the
patch.
Sadly this also breaks a bunch of other tests. Some of the tests in
hex_float_test use ldexp and then compare it with a value which is not
the same as the one returned by ldexp but instead is the value rounded
to 6 decimals. Others use values that are not evenly representable as
a binary floating fraction but then happened to generate the same
value when rounded to 6 decimals. Where the actual value didn’t seem
to matter these have been changed with different values that can be
represented as a binary fraction.
When the original code copies an entire array or struct one element at a
time, this turns into a series of OpCompositeInsert instruction followed
by a store of the whole array. We currently miss opportunities in copy
propagate arrays because we do not recognize this as a copy.
This commit adds code to copy propagate arrays to identify this code
pattern.
Also updates the performance passed to run array copy propagation.
The first implementation of MemroyObject, which is used in copy
propagate arrays, forced the access chain to be like the access chains
in OpCompositeExtract. This excluded the possibility of the memory
object from representing an array element that was extracted with a
variable index. Looking at the code, that restriction is not
neccessary. I also see some opportunities for doing this in some real
shaders.
Contributes to #1430.
This patch adds support for the analysis of scalars in loops. It works
by traversing the defuse chain to build a DAG of scalar operations and
then simplifies the DAG by folding constants and grouping like terms.
It represents induction variables as recurrent expressions with respect
to a given loop and can simplify DAGs containing recurrent expression by
rewritting the entire DAG to be a recurrent expression with respect to
the same loop.
When we change the type of an object that gets stored, we do not want to
change the type of the memory location being stored to. In order to
still be able to do the rewrite, we will decompose and rebuild the
object so it is the type that can be stored.
Fixes#1416.
The sprir-v generated from HLSL code contain many copyies of very large
arrays. Not only are these time consumming, but they also cause
problems for drivers because they require too much space.
To work around this, we will implement an array copy propagation. Note
that we will not implement a complete array data flow analysis in order
to implement this. We will be looking for very simple cases:
1) The source must never be stored to.
2) The target must be stored to exactly once.
3) The store to the target must be a store to the entire array, and be a
copy of the entire source.
4) All loads of the target must be dominated by the store.
The hard part is keeping all of the types correct. We do not want to
have to do too large a search to update everything, which may not be
possible, do we give up if we see any instruction that might be hard to
update.
Also in types.h, the element decorations are not stored in an std::map.
This change was done so the hashing algorithm for a Struct is
consistent. With the std::unordered_map, the traversal order was
non-deterministic leading to the same type getting hashed to different
values. See |Struct::GetExtraHashWords|.
Contributes to #1416.
This pass replaces the load/store elimination passes. It implements the
SSA re-writing algorithm proposed in
Simple and Efficient Construction of Static Single Assignment Form.
Braun M., Buchwald S., Hack S., Leißa R., Mallon C., Zwinkau A. (2013)
In: Jhala R., De Bosschere K. (eds)
Compiler Construction. CC 2013.
Lecture Notes in Computer Science, vol 7791.
Springer, Berlin, Heidelberg
https://link.springer.com/chapter/10.1007/978-3-642-37051-9_6
In contrast to common eager algorithms based on dominance and dominance
frontier information, this algorithm works backwards from load operations.
When a target variable is loaded, it queries the variable's reaching
definition. If the reaching definition is unknown at the current location,
it searches backwards in the CFG, inserting Phi instructions at join points
in the CFG along the way until it finds the desired store instruction.
The algorithm avoids repeated lookups using memoization.
For reducible CFGs, which are a superset of the structured CFGs in SPIRV,
this algorithm is proven to produce minimal SSA. That is, it inserts the
minimal number of Phi instructions required to ensure the SSA property, but
some Phi instructions may be dead
(https://en.wikipedia.org/wiki/Static_single_assignment_form).
The loop peeler util takes a loop as input and create a new one before.
The iterator of the duplicated loop then set to accommodate the number
of iteration required for the peeling.
The loop peeling pass that decided to do the peeling and profitability
analysis is left for a follow-up PR.
