LUMIERA.clone/tests/core/steam/engine/node-devel-test.cpp
Ichthyostega aac9ee2934 Bugfix: uninitialised stack-memory used in test
...this bug was spotted as the test failed reproducibly
when built with `-fstack-protector-strong` — which adds
additional "canary" markers to some kinds of problematic
stack storage; this seemed to have the effect that now
the second test uses exactly the same location as the
preceding test, and thus finds valid data in the newly
created work buffers.

There is no reason for using uninitialised storage in this
test (I can recall that I wanted to build a generic helper
and intended to use that as ''virtual overlay'' over existing
memory — but that usage never took place, and the `struct Buffer`
is not a general-purpose tool now, but only made for this
specific test. Thus memory can be easily zero-initialised.
2025-11-15 02:00:11 +01:00

503 lines
19 KiB
C++
Raw Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

/*
NodeDevel(Test) - Render Node development and test support
Copyright (C)
2024, Hermann Vosseler <Ichthyostega@web.de>
  **Lumiera** is free software; you can redistribute it and/or modify it
  under the terms of the GNU General Public License as published by the
  Free Software Foundation; either version 2 of the License, or (at your
  option) any later version. See the file COPYING for further details.
* *****************************************************************/
/** @file node-devel-test.cpp
** Unit test \ref NodeDevel_test verifies helpers for testing of render nodes.
*/
#include "lib/test/run.hpp"
#include "lib/hash-combine.hpp"
#include "lib/test/test-helper.hpp"
#include "steam/engine/node-builder.hpp"
#include "steam/engine/test-rand-ontology.hpp"
#include "steam/engine/diagnostic-buffer-provider.hpp"
#include "lib/iter-zip.hpp"
#include "lib/random.hpp"
#include <vector>
#include <array>
using lib::zip;
using lib::izip;
using std::array;
using std::vector;
using std::make_tuple;
using lib::test::showType;
namespace steam {
namespace engine{
namespace test {
namespace {
/** uninitialised local storage that can be passed
* as working buffer and accessed as TestFrame */
struct Buffer
: util::NonCopyable
{
alignas(TestFrame)
array<std::byte, sizeof(TestFrame)> storage{}; //zero-init
operator TestFrame* () { return std::launder (reinterpret_cast<TestFrame* > (&storage)); }
TestFrame* operator->() { return std::launder (reinterpret_cast<TestFrame* > (&storage)); }
TestFrame& operator* () { return * std::launder (reinterpret_cast<TestFrame* > (&storage)); }
TestFrame&
buildData (uint seq=0, uint family=0)
{
return * new(&storage) TestFrame{seq,family};
}
};
}
/***************************************************************//**
* @test verify support for developing Render Node functionality.
* - raw processing functions to generate and manipulate
* \ref TestFrame data, including hash chaining.
* - a »TestRand-Ontology«, which is a test helper framework,
* and mimics a real _Domain Ontology_ (as would be accessible
* through the adapter plug-in of a specific media handling library.
* - some convenience shortcuts to build test-nodes
*/
class NodeDevel_test : public Test
{
virtual void
run (Arg)
{
seedRand();
TestFrame::reseed();
processing_generateFrame();
processing_generateMultichan();
processing_duplicateMultichan();
processing_manipulateMultichan();
processing_manipulateFrame();
processing_combineFrames();
testRand_simpleUsage();
testRand_buildFilterNode();
testRand_buildMixNode();
}
/** @test function to generate random test data frames
*/
void
processing_generateFrame()
{
size_t frameNr = defaultGen.u64();
uint flavour = defaultGen.u64();
Buffer buff;
CHECK (not buff->isSane());
ont::generateFrame (buff, frameNr, flavour);
CHECK ( buff->isSane());
CHECK ( buff->isPristine());
CHECK (*buff == TestFrame(frameNr,flavour));
}
/** @test function to generate an array of random test data frames
* for consecutive channels
*/
void
processing_generateMultichan()
{
size_t frameNr = defaultGen.u64();
uint flavour = defaultGen.u64();
uint channels = 1 + rani(50);
CHECK (1 <= channels and channels <= 50);
Buffer buff[50];
for (uint i=0; i<channels; ++i)
CHECK (not buff[i]->isSane());
ont::generateMultichan (buff[0], channels, frameNr, flavour);
for (uint i=0; i<channels; ++i)
{
CHECK (buff[i]->isPristine());
CHECK (*(buff[i]) == TestFrame(frameNr,flavour+i));
}
}
/** @test clone copy of multichannel test data */
void
processing_duplicateMultichan()
{
size_t frameNr = defaultGen.u64();
uint flavour = defaultGen.u64();
uint channels = 1 + rani(50);
Buffer srcBuff[50];
ont::generateMultichan (srcBuff[0], channels, frameNr, flavour);
Buffer clone[50];
for (uint i=0; i<channels; ++i)
CHECK (not clone[i]->isSane());
ont::duplicateMultichan (clone[0],srcBuff[0], channels);
for (uint i=0; i<channels; ++i)
{
CHECK (clone[i]->isPristine());
CHECK (*(clone[i]) == *(srcBuff[i]));
}
}
/** @test multichannel data hash-chain manipulation
* - use multichannel pseudo random input data
* - store away a clone copy before manipulation
* - the #manipulateMultichan() operates in-place in the buffers
* - each buffer has been marked with a new checksum afterwards
* - and each buffer now differs from original state
* - verify that corresponding data points over all channels
* have been linked by a hashcode-chain, seeded with the `param`
* and then consecutively hashing in data from each channel.
*/
void
processing_manipulateMultichan()
{
size_t frameNr = defaultGen.u64();
uint flavour = defaultGen.u64();
uint channels = 1 + rani(50);
Buffer buff[50], refData[50];
ont::generateMultichan (buff[0], channels, frameNr, flavour);
// stash away a copy of the test data for verification
ont::duplicateMultichan(refData[0],buff[0], channels);
for (uint c=0; c<channels; ++c)
CHECK (buff[c]->isPristine());
uint64_t param = defaultGen.u64();
ont::manipulateMultichan(buff[0], channels, param);
const uint SIZ = buff[0]->data64().size();
vector<uint64_t> xlink(SIZ, param); // temporary storage for verifying the hash-chain
for (uint c=0; c<channels; ++c)
{
CHECK (buff[c]->isSane()); // checksum matches
CHECK (not buff[c]->isPristine()); // data was indeed changed
CHECK (*(buff[c]) != *(refData[c]));
for (auto& [i, link] : izip(xlink))
{
auto const& refPoint = refData[c]->data64()[i];
lib::hash::combine (link, refPoint);
CHECK (link != refPoint);
CHECK (link == buff[c]->data64()[i]);
}
}
}
/** @test function to apply a numeric computation to test data frames;
* @remark here basically the same hash-chaining is used as for #manipulateMultichan,
* but only one hash-chain per data point is used and output is written to a different buffer.
*/
void
processing_manipulateFrame()
{
size_t frameNr = defaultGen.u64();
uint flavour = defaultGen.u64();
Buffer iBuff, oBuff;
iBuff.buildData(frameNr,flavour);
oBuff.buildData(frameNr,flavour);
CHECK (iBuff->isPristine());
CHECK (oBuff->isPristine());
uint64_t param = defaultGen.u64();
ont::manipulateFrame (oBuff, iBuff, param);
CHECK ( oBuff->isValid());
CHECK (not oBuff->isPristine());
CHECK ( iBuff->isPristine());
for (auto [iDat,oDat] : zip (iBuff->data64()
,oBuff->data64()))
{
CHECK (oDat != iDat);
uint64_t feed = param;
lib::hash::combine (feed, iDat);
CHECK (feed != param);
CHECK (feed != iDat);
CHECK (feed == oDat);
}
// can also process in-place
ont::manipulateFrame (iBuff, iBuff, param);
CHECK (not iBuff->isPristine());
CHECK ( iBuff->isValid());
CHECK (*iBuff == *oBuff); // second invocation exactly reproduced data from first invocation
}
/** @test function to mix two test data frames
*/
void
processing_combineFrames()
{
size_t frameNr = defaultGen.u64();
uint flavour = defaultGen.u64();
Buffer i1Buff, i2Buff, oBuff;
i1Buff.buildData(frameNr,flavour+0);
i2Buff.buildData(frameNr,flavour+1);
oBuff.buildData();
CHECK (i1Buff->isPristine());
CHECK (i2Buff->isPristine());
CHECK (oBuff->isPristine());
double mix = defaultGen.uni();
ont::combineFrames (oBuff, i1Buff, i2Buff, mix);
CHECK ( oBuff->isValid());
CHECK (not oBuff->isPristine());
CHECK ( i1Buff->isPristine());
CHECK ( i2Buff->isPristine());
for (auto [oDat,i1Dat,i2Dat] : zip (oBuff->data()
,i1Buff->data()
,i2Buff->data()))
CHECK (oDat == std::lround((1-mix)*i1Dat + mix*i2Dat));
// can also process in-place
ont::combineFrames (i1Buff, i1Buff, i2Buff, mix);
CHECK (not i1Buff->isPristine());
CHECK ( i1Buff->isValid());
CHECK (*i1Buff == *oBuff); // second invocation exactly reproduced data from first invocation
}
/** @test demonstrate simple usage of test-render setup
* - access the TestRandOntology as singleton
* - create a Spec record
* - retrieve a functor bound suitably to invoke
* data processing code from the TestRandOntology
*/
void
testRand_simpleUsage()
{
auto spec = testRand().setupGenerator();
CHECK (spec.PROTO == "generate-TestFrame"_expect);
// generate a binding as processing-functor
auto procFun = spec.makeFun();
using Sig = lib::meta::_Fun<decltype(procFun)>::Sig;
CHECK (showType<Sig>() == "void (tuple<ulong, uint>, engine::test::TestFrame*)"_expect);
// Behaves identical to processing_generateFrame() — see above...
size_t frameNr = defaultGen.u64();
uint flavour = defaultGen.u64();
Buffer buff;
CHECK (not buff->isSane());
procFun (make_tuple (frameNr,flavour), buff);
CHECK ( buff->isSane());
CHECK ( buff->isPristine());
CHECK (*buff == TestFrame(frameNr,flavour));
// Build a node using this processing-functor...
ProcNode node{prepareNode(spec.nodeID())
.preparePort()
.invoke(spec.procID(), procFun)
.setParam(frameNr,flavour)
.completePort()
.build()};
CHECK (watch(node).isSrc());
CHECK (watch(node).getNodeSpec() == "Test:generate-◎"_expect);
CHECK (watch(node).getPortSpec(0) == "generate(TestFrame)"_expect);
BufferProvider& provider = DiagnosticBufferProvider::build();
BuffHandle buffHandle = provider.lockBuffer (provider.getDescriptorFor(sizeof(TestFrame)));
uint port{0};
CHECK (not buffHandle.accessAs<TestFrame>().isSane());
// Trigger Node invocation...
buffHandle = node.pull (port, buffHandle, Time::ZERO, ProcessKey{0});
TestFrame& result = buffHandle.accessAs<TestFrame>();
CHECK (result.isSane());
CHECK (result.isPristine());
CHECK (result == *buff);
buffHandle.release();
}
/** shortcut to simplify the following test cases */
static ProcNode
makeSrcNode (ont::FraNo frameNr, ont::Flavr flavour)
{
auto spec = testRand().setupGenerator();
return prepareNode(spec.nodeID())
.preparePort()
.invoke(spec.procID(), spec.makeFun())
.setParam(frameNr,flavour)
.completePort()
.build();
}
/** @test use the »TestRand«-framework to setup a filter node
* - implementation is backed by the ont::manipulateFrame() function
* - it thus operates on \ref TestFrame data and results can be verified
* - the generated spec-recod provides a processing-functor binding and node-spec
* - can build and wire a Node processing chain with a source node and a »filter«
* node based on this data manipulation, which exactly reproduces the data
* content generated by the stand-alone invocation.
* @remark such a test-setup thus not only allows to prove that the function was invoked,
* but also the order in which the processing took place, due to hash-chaining
* applied to every single data word in the `TestFrame` buffer.
* @see NodeLink_test::trigger_node_port_invocation()
*/
void
testRand_buildFilterNode()
{
auto spec = testRand().setupManipulator();
CHECK (spec.PROTO == "manipulate-TestFrame"_expect);
// generate a binding as processing-functor
auto procFun = spec.makeFun();
using Sig = lib::meta::_Fun<decltype(procFun)>::Sig;
CHECK (showType<Sig>() == "void (ulong, engine::test::TestFrame const*, engine::test::TestFrame*)"_expect);
// Results can be verified by ont::manipulateFrame() — see above
size_t frameNr = defaultGen.u64();
uint flavour = defaultGen.u64();
uint64_t param = defaultGen.u64();
Buffer buff;
buff.buildData(frameNr,flavour);
CHECK (buff->isPristine());
// Invoke the processing-functor directly
procFun (param, buff,buff);
CHECK ( buff->isValid());
CHECK (not buff->isPristine());
HashVal checksum = buff->markChecksum();
// reproduce the same checksum...
buff.buildData(frameNr,flavour);
CHECK (buff->isPristine());
CHECK (checksum != buff->getChecksum());
ont::manipulateFrame (buff, buff, param);
CHECK (checksum == buff->getChecksum());
// Build a node using this processing-functor...
ProcNode nSrc = makeSrcNode (frameNr,flavour);
ProcNode nFilt{prepareNode(spec.nodeID())
.preparePort()
.invoke(spec.procID(), procFun)
.setParam(param)
.connectLead(nSrc)
.completePort()
.build()};
CHECK (watch(nSrc).isSrc());
CHECK (not watch(nFilt).isSrc());
CHECK (watch(nSrc).getNodeSpec() == "Test:generate-◎"_expect );
CHECK (watch(nFilt).getNodeSpec() == "Test:manipulate◁—Test:generate-◎"_expect );
CHECK (watch(nFilt).getPortSpec(0) == "manipulate(TestFrame)(TestFrame)"_expect );
// prepare to invoke this Node chain...
BufferProvider& provider = DiagnosticBufferProvider::build();
BuffHandle buffHandle = provider.lockBuffer (provider.getDescriptorFor(sizeof(TestFrame)));
uint port{0};
CHECK (not buffHandle.accessAs<TestFrame>().isValid());
// Trigger Node invocation...
buffHandle = nFilt.pull (port, buffHandle, Time::ZERO, ProcessKey{0});
TestFrame& result = buffHandle.accessAs<TestFrame>();
CHECK ( result.isValid());
CHECK (not result.isPristine());
CHECK (result == *buff);
buffHandle.release();
}
/** @test use the »TestRand«-framework to setup a two-chain mixer node
* - demonstrate convenience setup to package the ont::combineFrames() as »mix« Node
* - this time, we need two source chains, both generating \ref TestFrame data
* - complete processing with all steps can be verified by performing similar
* computations directly and comparing the result checksum.
*/
void
testRand_buildMixNode()
{
auto spec = testRand().setupCombinator();
CHECK (spec.PROTO == "combine-TestFrame"_expect);
// generate a binding as processing-functor
auto procFun = spec.makeFun();
using Sig = lib::meta::_Fun<decltype(procFun)>::Sig;
CHECK (showType<Sig>() == "void (double, array<engine::test::TestFrame const*, 2ul>, "
"engine::test::TestFrame*)"_expect); //^^/////////////////TICKET #1391 needlessly rendered as `long`
size_t frameNr = defaultGen.u64();
uint flavour = defaultGen.u64();
double mix = defaultGen.uni();
// Build node graph to combine two chains
ProcNode nS1 = makeSrcNode (frameNr,flavour+0);
ProcNode nS2 = makeSrcNode (frameNr,flavour+1);
ProcNode nMix{prepareNode(spec.nodeID())
.preparePort()
.invoke(spec.procID(), procFun)
.setParam(mix)
.connectLead(nS1)
.connectLead(nS2)
.completePort()
.build()};
CHECK (not watch(nMix).isSrc());
CHECK (watch(nS1).getNodeSpec() == "Test:generate-◎"_expect );
CHECK (watch(nS2).getNodeSpec() == "Test:generate-◎"_expect );
CHECK (watch(nMix).getNodeSpec() == "Test:combine┉┉{Test:generate}"_expect );
CHECK (watch(nMix).getPortSpec(0) == "combine(TestFrame/2)(TestFrame)"_expect );
// prepare to invoke this Node chain...
BufferProvider& provider = DiagnosticBufferProvider::build();
BuffHandle buffHandle = provider.lockBuffer (provider.getDescriptorFor(sizeof(TestFrame)));
CHECK (not buffHandle.accessAs<TestFrame>().isValid());
uint port{0};
// Trigger Node invocation...
buffHandle = nMix.pull (port, buffHandle, Time::ZERO, ProcessKey{0});
CHECK (buffHandle.accessAs<TestFrame>().isValid());
HashVal checksum = buffHandle.accessAs<TestFrame>().getChecksum();
buffHandle.release();
// verify the result data by reproducing it through direct computation
Buffer bu1, bu2;
bu1.buildData(frameNr,flavour+0);
bu2.buildData(frameNr,flavour+1);
ont::combineFrames (bu1, bu1, bu2, mix);
CHECK (bu1->getChecksum() == checksum);
}
};
/** Register this test class... */
LAUNCHER (NodeDevel_test, "unit node");
}}} // namespace steam::engine::test