RaytracingTier1_1 flag should not be set in module based on subobject #6321
Comments
|
Note, in PR #6207, the computation of RaytracingTier1_1 flag was moved from inside per-function flag computation into per-module (in DxilModule). It turns out that this didn't change the behavior overall, because before #6207, subobjects got stripped from the module just before the per-function RDAT flags were computed, and the module flag still had RaytracingTier1_1 set because they were computed before that point during DxilPreparePasses. |
tex3d
added a commit
that referenced
this issue
Feb 16, 2024
…on (#6320) In validator version 1.7 and below, the RaytracingTier1_1 module flag was set on every function if any StateObjectConfig subobject had the AllowStateObjectAdditions flag set. This was incorrect, as the requirement is validated in the runtime based on the use of the subobject instead. Subobjects are removed from the module and placed in RDAT during container serialization, so the requirement would be lost when recomputing the flags in validation. This didn't break validation because flag validation was completely disabled for libraries! This change fixes this problem, and allows DxilValidation to validate ShaderFlags because they will no longer mismatch due to this issue. Running CollectShaderFlagsForModule is also necessary for collecting per-function shader flags which will be used by call graph validation in a subsequent change, so enabling flag validation unblocks that as well. Fixes #6321
tex3d
added a commit
to tex3d/DirectXShaderCompiler
that referenced
this issue
Feb 16, 2024
…on (microsoft#6320) In validator version 1.7 and below, the RaytracingTier1_1 module flag was set on every function if any StateObjectConfig subobject had the AllowStateObjectAdditions flag set. This was incorrect, as the requirement is validated in the runtime based on the use of the subobject instead. Subobjects are removed from the module and placed in RDAT during container serialization, so the requirement would be lost when recomputing the flags in validation. This didn't break validation because flag validation was completely disabled for libraries! This change fixes this problem, and allows DxilValidation to validate ShaderFlags because they will no longer mismatch due to this issue. Running CollectShaderFlagsForModule is also necessary for collecting per-function shader flags which will be used by call graph validation in a subsequent change, so enabling flag validation unblocks that as well. Fixes microsoft#6321 (cherry picked from commit fafbc42)
tex3d
added a commit
that referenced
this issue
Feb 20, 2024
…on (#6320) (#6322) In validator version 1.7 and below, the RaytracingTier1_1 module flag was set on every function if any StateObjectConfig subobject had the AllowStateObjectAdditions flag set. This was incorrect, as the requirement is validated in the runtime based on the use of the subobject instead. Subobjects are removed from the module and placed in RDAT during container serialization, so the requirement would be lost when recomputing the flags in validation. This didn't break validation because flag validation was completely disabled for libraries! This change fixes this problem, and allows DxilValidation to validate ShaderFlags because they will no longer mismatch due to this issue. Running CollectShaderFlagsForModule is also necessary for collecting per-function shader flags which will be used by call graph validation in a subsequent change, so enabling flag validation unblocks that as well. Fixes #6321 (cherry picked from commit fafbc42)
It turns out this was wrong. The flags were computed before stripping the subobjects. Merging the change based on this assumption causes a failure when testing against the previous released validator. Reactivating to merge a more complete fix which restores this only for validator version 1.7 and below. |
tex3d
added a commit
that referenced
this issue
Feb 21, 2024
It turns out that in the prior validator version, if a subobject required DXR 1.1, the DXR 1.1 flag would be set on each function in RDAT output, as well as the global flags. It didn't appear this was the case through a D3DReflect test because that goes through disassembly to assembly step, where subobjects are lost because they aren't in the llvm IR. The previous change assumed this was not the case when the subobjects were removed, but this removal occurs after the RDATWriter constructor, which does the full RDAT serialization since size if required right away. This restores the detection code and hooks it into DxilModule::ComputeShaderCompatInfo when validator version is in range [1.5, 1.7]. DXR 1.1 was introduced in 1.5, and we no longer tag every function as requiring DXR 1.1 based on subobjects in 1.8. At the moment, there is no way to CHECK the subobject RDAT in D3DReflect because they get stripped from the module (even reflection and debug modules) before serialization, and the test path for D3DReflect goes through a disassemble/re-assemble step between the prior stage and the D3DReflect stage. Fixes #6321 (really).
tex3d
added a commit
to tex3d/DirectXShaderCompiler
that referenced
this issue
Feb 21, 2024
It turns out that in the prior validator version, if a subobject required DXR 1.1, the DXR 1.1 flag would be set on each function in RDAT output, as well as the global flags. It didn't appear this was the case through a D3DReflect test because that goes through disassembly to assembly step, where subobjects are lost because they aren't in the llvm IR. The previous change assumed this was not the case when the subobjects were removed, but this removal occurs after the RDATWriter constructor, which does the full RDAT serialization since size if required right away. This restores the detection code and hooks it into DxilModule::ComputeShaderCompatInfo when validator version is in range [1.5, 1.7]. DXR 1.1 was introduced in 1.5, and we no longer tag every function as requiring DXR 1.1 based on subobjects in 1.8. At the moment, there is no way to CHECK the subobject RDAT in D3DReflect because they get stripped from the module (even reflection and debug modules) before serialization, and the test path for D3DReflect goes through a disassemble/re-assemble step between the prior stage and the D3DReflect stage. Fixes microsoft#6321 (really). (cherry picked from commit 22bb078)
tex3d
added a commit
that referenced
this issue
Feb 21, 2024
It turns out that in the prior validator version, if a subobject required DXR 1.1, the DXR 1.1 flag would be set on each function in RDAT output, as well as the global flags. It didn't appear this was the case through a D3DReflect test because that goes through disassembly to assembly step, where subobjects are lost because they aren't in the llvm IR. The previous change assumed this was not the case when the subobjects were removed, but this removal occurs after the RDATWriter constructor, which does the full RDAT serialization since size if required right away. This restores the detection code and hooks it into DxilModule::ComputeShaderCompatInfo when validator version is in range [1.5, 1.7]. DXR 1.1 was introduced in 1.5, and we no longer tag every function as requiring DXR 1.1 based on subobjects in 1.8. At the moment, there is no way to CHECK the subobject RDAT in D3DReflect because they get stripped from the module (even reflection and debug modules) before serialization, and the test path for D3DReflect goes through a disassemble/re-assemble step between the prior stage and the D3DReflect stage. Fixes #6321 (really). (cherry picked from commit 22bb078)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Description
DxilModule checks
StateObjectConfigin subobjects forAllowStateObjectAdditions, then sets the module flagRaytracingTier1_1based on this, if not already set for some other reason.This is incorrect, as the runtime will enforce feature requirement based on the subobjects used, and the module flag will prevent a library containing the subobject from being created, even if this Tier 1.1 subobject is not used.
On top of this, since subobjects are removed from the module during container serialization, recomputation of flags for validation will not see the requirement. This doesn't fail right now because flag validation is disabled for libraries (which it shouldn't be). This validation will be enabled for 1.8, so it's necessary to fix this issue before DXIL 1.8 release.
Steps to Reproduce
// dxc -T lib_6_5 StateObjectConfig soc = { STATE_OBJECT_FLAGS_ALLOW_LOCAL_DEPENDENCIES_ON_EXTERNAL_DEFINITONS | STATE_OBJECT_FLAG_ALLOW_STATE_OBJECT_ADDITIONS };"Raytracing tier 1.1 features" should not be a required feature for this module.
Actual Behavior
Disassembly output states that "Raytracing tier 1.1 features" is required.
Environment
The text was updated successfully, but these errors were encountered: