Restrict GetILAddressMap HResult debug assertion to specific error codes in cDAC

[Copilot]

Description

The cDAC runtime-diagnostics pipeline fails because the legacy DAC returns E_FAIL (0x80004005) while the cDAC returns CORDBG_E_READVIRTUAL_FAILURE (0x80131c49) for GetILAddressMap failures. The strict Debug.Assert on HResult equality trips on this mismatch.

Changed the assertion in ClrDataMethodInstance.cs to use a local IsReadFailure helper that specifically accepts only HResults.E_FAIL or CorDbgHResults.CORDBG_E_READVIRTUAL_FAILURE, rather than accepting any negative (failing) HResult. This ensures unexpected error codes are still caught while allowing the known DAC/cDAC mismatch.

This fixes the VarargPInvokeInteropMD in runtime-diagnostics.

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.

[max-charlamb]

@copilot, the change was not made

[dotnet-policy-service]

Tagging subscribers to this area: @steveisok, @tommcdon, @dotnet/dotnet-diag
See info in area-owners.md if you want to be subscribed.

[Copilot]

@copilot, the change was not made

The change was in the branch but wasn't showing in the PR due to a push sync issue. Re-pushed in 20d21023 — the PR diff now correctly shows the assertion change from Debug.Assert(hrLocal == hr, ...) to Debug.Assert(hrLocal == hr || (hrLocal < 0 && hr < 0), ...).

[max-charlamb]

@copilot we should verify that the error codes are CORDBG_E_READVIRTUAL_FAILURE or HResults.E_FAIL. We shouldn't accept any failing HResult.

[Copilot]

@copilot we should verify that the error codes are CORDBG_E_READVIRTUAL_FAILURE or HResults.E_FAIL. We shouldn't accept any failing HResult.

Updated in 52cd33ec — the assertion now uses a local IsReadFailure helper that only accepts HResults.E_FAIL or CorDbgHResults.CORDBG_E_READVIRTUAL_FAILURE instead of any negative HResult.

[Copilot]

Pull request overview

This PR fixes a cDAC runtime-diagnostics pipeline failure caused by a debug assertion that was too strict when comparing HResults between the legacy DAC and cDAC implementations. The legacy DAC returns E_FAIL while cDAC returns CORDBG_E_READVIRTUAL_FAILURE for read failures in GetILAddressMap, causing the strict equality check to fail.

Changes:

• Updated GetILAddressMap debug assertion to accept specific known error codes (E_FAIL or CORDBG_E_READVIRTUAL_FAILURE) instead of requiring exact HResult equality
• Added an IsReadFailure helper function to encapsulate the specific error code checking logic
• Removed an extraneous blank line for consistency

[rcj1]

Any reason not to modify the cDAC to match the DAC by wrapping the boundaries in a throw E_FAIL?

[max-charlamb]

Any reason not to modify the cDAC to match the DAC by wrapping the boundaries in a throw E_FAIL?

I was potentially going to do this. We would want to E_FAIL if there is no DebugInfo for the particular method. It's a more invasive change but possible. I'd like to get my dump testing PR merged and then I can have a test for this particular situation.

[max-charlamb]

Closing in favor of #124783