Write through the device address table into guest memory too, so the GPU's writes there are no longer lost.

The table's records gain a twin: a pool with a VRAM mirror names the mirror first and its imported buffer, guest
memory, second; any other record names one buffer twice. Shaders read through the first address and store and run
atomics through both (translator patch 0034); a page without a twin sends the second write to a 4 KiB scratch buffer.
The GPU's writes through the table were not recorded as GPU writes, so they stayed in the mirror: the CPU never saw
them, and a sync of a page the CPU wrote copied the stale guest page over them. Trees drawn from records lost that
way stretched into columns reaching the sky. With NX2NAVI_MEMORY=verify the Hyrule Field walk found 16 such mirror
pages before and none or one after; GPU time per frame 5.92 ms before, 6.17 ms after.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
chrg6
2026-11-07 20:24:41 -05:00
parent 8102b6a222
commit 4cb8188491
4 changed files with 168 additions and 14 deletions

View File

@@ -99,8 +99,8 @@ def prepare_sources() -> Path:
for junk in list(src.glob("src/*/obj")) + list(src.glob("src/*/bin")):
shutil.rmtree(junk)
patches = sorted((XLATE / "patches").glob("*.patch"))
if len(patches) != 32:
sys.exit(f"expected 32 patches, found {len(patches)}")
if len(patches) != 33:
sys.exit(f"expected 33 patches, found {len(patches)}")
for patch in patches:
# git apply works outside a repository when given a directory; --check first so a failure names the patch.
for extra in (["--check"], []):