[Fix] (Pipe, State): look the resource handle up on the content paths instead of minting it, publish the bind mask from the two draw-time emitters, pair the destroy with the create's own latch, and state the record-lifetime rule the applier now holds

This commit is contained in:
2026-09-08 04:47:51 -04:00
parent f11e78b0a4
commit 0c55560510
7 changed files with 273 additions and 38 deletions
+2 -1
View File
@@ -1571,7 +1571,8 @@ namespace {
// And every piece the walk produced has to be encodable by the record builder -
// a piece the box refuses is a record the applier's bounds gate would abort on.
MGPSubData record{};
EXPECT_TRUE(MGPipeBuildSubDataRecord(MGPipeHandle{1, 1}, piece.first, piece.second, record))
EXPECT_TRUE(MGPipeBuildSubDataRecord(MGPipeHandle{1, 1}, piece.first, piece.second, record,
/*verbatimShadow=*/true))
<< "a piece the splitter produced does not fit one record";
EXPECT_EQ(MGPipeSubDataBufferOffset(record), piece.first);
EXPECT_EQ(MGPipeSubDataBufferSize(record), piece.second);