Re: [HACKERS] GetLockConflicts() and thus recovery conflicts seem pretty broken

2015-01-29 Thread Simon Riggs
On 27 January 2015 at 14:27, Andres Freund wrote: > While investigating other bugs I noticed that > ResolveRecoveryConflictWithLock() wasn't really working. Turns out > GetLockConflicts() violates it's API contract which says: > > * The result array is palloc'd and is terminated with an invalid

[HACKERS] GetLockConflicts() and thus recovery conflicts seem pretty broken

2015-01-27 Thread Andres Freund
Hi, While investigating other bugs I noticed that ResolveRecoveryConflictWithLock() wasn't really working. Turns out GetLockConflicts() violates it's API contract which says: * The result array is palloc'd and is terminated with an invalid VXID. Problem 1: We don't actually put the terminator t