Fix documentation accuracy: checklist counts, phantom macro, missing patterns

- Fix "3-location checklist" heading to "4-location" (parameter-system.md)
- Fix "all 3 locations" to "all 4 locations" in PR checklist (common-pitfalls.md)
- Remove non-existent $:END_GPU_LOOP() from block macro example (gpu-and-mpi.md)
- Add d0 literals, double precision, dcos/dsin/dtan to forbidden patterns list
- Reference toolchain/bootstrap/precheck.sh for full forbidden pattern list
- Fix Frontier system slug: OpenACC/OpenMP, not just OpenACC
- Clarify stp as "field data arrays and I/O", not just "I/O"
- Document @:PROHIBIT, @:ASSERT, @:LOG error-checking macros
- Document @:PREFER_GPU unified memory macro
- Document m_checker*.fpp Fortran-side runtime validation

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: sbryngelson

.claude/rules/common-pitfalls.md

@@ -36,7 +36,7 @@ Before submitting a PR:
 - [ ] `./mfc.sh precheck -j 8` (5 CI lint checks)
 - [ ] `./mfc.sh build -j 8` (compiles)
 - [ ] `./mfc.sh test --only <relevant> -j 8` (tests pass)
-- [ ] If adding parameters: all 3 locations updated
+- [ ] If adding parameters: all 4 locations updated
 - [ ] If modifying `src/common/`: all three targets tested
 - [ ] If changing output: golden files regenerated for affected tests
 - [ ] One logical change per commit

.claude/rules/fortran-conventions.md

@@ -24,17 +24,26 @@ Every Fortran module follows this pattern:
 ## Forbidden Patterns
 
 Caught by `./mfc.sh precheck` (source lint step 4/5):
-- `dsqrt`, `dexp`, `dlog`, `dble`, `dabs` → use `sqrt`, `exp`, `log`, `real(..., wp)`
+- `dsqrt`, `dexp`, `dlog`, `dble`, `dabs`, `dcos`, `dsin`, `dtan`, etc. → use generic intrinsics
+- `1.0d0`, `2.5d-3` (Fortran `d` exponent literals) → use `1.0_wp`, `2.5e-3_wp`
+- `double precision` → use `real(wp)` or `real(stp)`
 - `real(8)`, `real(4)` → use `wp` or `stp` kind parameters
 - Raw `!$acc` or `!$omp` directives → use Fypp GPU_* macros from `parallel_macros.fpp`
+- Full list of forbidden patterns: `toolchain/bootstrap/precheck.sh`
 
 Enforced by convention/code review (not automated):
 - `goto`, `COMMON` blocks, global `save` variables
-- `stop`, `error stop` → use `call s_mpi_abort()`
+- `stop`, `error stop` → use `call s_mpi_abort()` or `@:PROHIBIT()`/`@:ASSERT()`
+
+## Error Checking Macros (from macros.fpp)
+- `@:PROHIBIT(condition, message)` — Runtime constraint check; aborts with file/line info
+- `@:ASSERT(predicate, message)` — Invariant assertion; aborts if predicate is false
+- `@:LOG(expr)` — Debug logging, active only in `MFC_DEBUG` builds
+- Fortran-side runtime validation also exists in `m_checker*.fpp` files using `@:PROHIBIT`
 
 ## Precision Types
 - `wp` (working precision): used for computation. Double by default.
-- `stp` (storage precision): used for I/O. Double by default.
+- `stp` (storage precision): used for field data arrays and I/O. Double by default.
 - In single-precision mode (`--single`): both become single.
 - In mixed-precision mode (`--mixed`): wp=double, stp=half.
 - MPI type matching: `mpi_p` must match `wp`, `mpi_io_p` must match `stp`.

.claude/rules/gpu-and-mpi.md

@@ -49,7 +49,6 @@ Block macro usage:
   do k = 0, n; do j = 0, m
     ! loop body
   end do; end do
-  $:END_GPU_LOOP()
 #:endcall GPU_PARALLEL
 ```
 
@@ -59,6 +58,7 @@ The precheck source lint will catch raw directives and fail.
 ### Memory Management Macros (from macros.fpp)
 - `@:ALLOCATE(var1, var2, ...)` — Fortran allocate + `GPU_ENTER_DATA(create=...)`
 - `@:DEALLOCATE(var1, var2, ...)` — `GPU_EXIT_DATA(delete=...)` + Fortran deallocate
+- `@:PREFER_GPU(var1, var2, ...)` — NVIDIA unified memory page placement hint
 - Every `@:ALLOCATE` MUST have a matching `@:DEALLOCATE` in finalization
 - Conditional allocation MUST have conditional deallocation
 

.claude/rules/parameter-system.md

@@ -23,7 +23,7 @@ MFC has ~3,300 simulation parameters defined in Python and read by Fortran via n
    - Reads `&user_inputs` namelist
    - Each parameter must be declared in the namelist statement
 
-## Adding a New Parameter (3-location checklist)
+## Adding a New Parameter (4-location checklist)
 
 YOU MUST update all 4 locations. Missing any causes silent failures or compile errors.
 
@@ -43,6 +43,12 @@ YOU MUST update all 4 locations. Missing any causes silent failures or compile e
 - Create new cases with `./mfc.sh new <name>`
 - Search parameters with `./mfc.sh params <query>`
 
+## Fortran-Side Runtime Validation
+Each target has `m_checker*.fpp` files (e.g., `src/simulation/m_checker.fpp`,
+`src/common/m_checker_common.fpp`) containing runtime parameter validation using
+`@:PROHIBIT(condition, message)`. When adding parameters with physics constraints,
+add Fortran-side checks here in addition to `case_validator.py`.
+
 ## Analytical Initial Conditions
 String expressions in parameters become Fortran code via `case.py.__get_analytic_ic_fpp()`.
 These are compiled into the binary, so syntax errors cause build failures, not runtime errors.

CLAUDE.md

@@ -75,7 +75,7 @@ Supported systems and their slugs (full list in `toolchain/modules`):
 | Slug | System | GPU Backend | Example |
 |------|--------|-------------|---------|
 | `p` | GT Phoenix | OpenACC (nvfortran) | `source ./mfc.sh load -c p -m g` |
-| `f` | OLCF Frontier | OpenACC (Cray ftn) | `source ./mfc.sh load -c f -m g` |
+| `f` | OLCF Frontier | OpenACC/OpenMP (Cray ftn) | `source ./mfc.sh load -c f -m g` |
 | `tuo` | LLNL Tuolumne | OpenMP (Cray ftn) | `source ./mfc.sh load -c tuo -m g` |
 | `d` | NCSA Delta | OpenACC (nvfortran) | `source ./mfc.sh load -c d -m g` |
 | `b` | PSC Bridges2 | OpenACC (nvfortran) | `source ./mfc.sh load -c b -m g` |
@@ -132,7 +132,8 @@ NEVER use raw OpenACC/OpenMP pragmas (`!$acc`, `!$omp`). Use `GPU_*` Fypp macros
   Raw `#ifdef`/`#ifndef` preprocessor guards for feature/compiler/library gating ARE normal.
 NEVER use double-precision intrinsics: `dsqrt`, `dexp`, `dlog`, `dble`, `dabs`, `real(8)`, `real(4)`.
   Use generic intrinsics (`sqrt`, `exp`, `log`) and precision types (`wp`, `stp`).
-NEVER use `stop` or `error stop`. Use `call s_mpi_abort()`.
+NEVER use `d` exponent literals (`1.0d0`). Use `1.0_wp` instead.
+NEVER use `stop` or `error stop`. Use `call s_mpi_abort()` or `@:PROHIBIT()`/`@:ASSERT()`.
 NEVER use `goto`, `COMMON` blocks, or global `save` variables.
 
 Every `@:ALLOCATE(...)` MUST have a matching `@:DEALLOCATE(...)`.
@@ -153,7 +154,7 @@ Changes to `src/common/` affect ALL three executables. Test comprehensively.
 
 ## Precision System
 
-- `wp` = working precision (computation). `stp` = storage precision (I/O).
+- `wp` = working precision (computation). `stp` = storage precision (field data arrays and I/O).
 - Default: both double. Single mode: both single. Mixed: wp=double, stp=half.
 - MPI types must match: `mpi_p` ↔ `wp`, `mpi_io_p` ↔ `stp`.