https://issues.apache.org/bugzilla/show_bug.cgi?id=50335
--- Comment #40 from Joe Orton <[email protected]> --- Using a subpool for the fake request sounds like it is not actually fixing the difficult problem here of getting the lifetimes right. But Max is right that basically everything after that ap_get_brigade() call is potentially dangerous as-is because of dangers with bucket lifetime mismatches. Comparing trunk with 2.2.x I notice there is a further complication in r1035605 In the EOF case, the proxy now reads HEAP buckets from the backend socket, converts them into TRANSIENT buckets so the bucket *structure* lifetime is correct... and then converts those buckets back into HEAP buckets so the bucket structure the bucket *data* lifetime are correct. That seems over-complicated. -- You are receiving this mail because: You are the assignee for the bug. --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
