Repository navigation
Commit 9bedf7e
lane-u5a-2359
test: give the waiting writer more attempts than the parent can hold for
Thread 4544685126 claims a child can still exhaust its retries before the parent
releases the lock. Measured at head 2464d70 the
parent is the holder in both waiting rows, acquiring at lines 1568 and 1602 and
releasing at lines 1586 and 1618, and the two barrier polls in between run to
15000 ms each at lines 1576, 1579, 1610 and 1614, so the ceiling on the hold is
their sum, 30000 ms. 100 attempts at 250 ms gave the child 25000 ms, 5000 ms short
of that ceiling, and 121 attempts give 30250 ms, 250 ms above it.
The comment now names the sum of the two barrier polls, which is the quantity that
binds the child. The identifier keeps the name of the sizing this replaces, and
that name is a residue rather than a description.
Nothing here is verification. No Stryker leg ran for this change, so no mutation
digest exists for it, and the hosted windows runner has never been read on this
machine. the mutation leg is unmeasured, a PASS over an empty digest is not
evidence in either direction, and CI at the first pushed unit is the only proof of
the gate.1 parent 2464d70 commit 9bedf7e
1 file changed
Lines changed: 6 additions & 4 deletions
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
1523 | 1523 | | |
1524 | 1524 | | |
1525 | 1525 | | |
1526 | | - | |
1527 | | - | |
1528 | | - | |
1529 | | - | |
| 1526 | + | |
| 1527 | + | |
| 1528 | + | |
| 1529 | + | |
| 1530 | + | |
| 1531 | + | |
1530 | 1532 | | |
1531 | 1533 | | |
1532 | 1534 | | |
| |||
0 commit comments