Re: aggregate crash

Hi, On 2020-01-15 19:16:30 -0800, Andres Freund wrote: > The bug, in a way, exists all the way back, but it's a bit harder to > create NULL values where the datum component isn't 0. > To fix I suggest we, in all branches, do the equivalent of adding > something like: > diff --git i/src/backend/ex

Re: aggregate crash

Hi, On 2020-01-15 12:47:47 -0800, Andres Freund wrote: > FWIW, I'm working on narrowing it down to something small. I can > reliably trigger the bug, and I understand the mechanics, I > think. Interestingly enough the reproducer currently only triggers on > v12, not on v11 and before. That's just

Re: aggregate crash

Hi, On 2020-01-14 23:27:02 -0800, Andres Freund wrote: > On 2020-01-14 17:54:16 -0500, Tom Lane wrote: > > Andres Freund writes: > > > I'm still not sure I actually fully understand the bug. It's obvious how > > > returning the input value again could lead to memory not being freed (so > > > that

Re: aggregate crash

Hi, On 2020-01-14 17:54:16 -0500, Tom Lane wrote: > Andres Freund writes: > > On 2020-01-14 17:01:01 -0500, Tom Lane wrote: > >> But I agree that not checking null-ness > >> explicitly is kind of unsafe. We've never before had any expectation > >> that the Datum value of a null is anything in pa

Re: aggregate crash

Andres Freund writes: > On 2020-01-14 17:01:01 -0500, Tom Lane wrote: >> But I agree that not checking null-ness >> explicitly is kind of unsafe. We've never before had any expectation >> that the Datum value of a null is anything in particular. > I'm still not sure I actually fully understand t

Re: aggregate crash

Hi, On 2020-01-14 17:01:01 -0500, Tom Lane wrote: > Alvaro Herrera writes: > > On 2019-Dec-27, Teodor Sigaev wrote: > >> Found crash on production instance, assert-enabled build crashes in pfree() > >> call, with default config. v11, v12 and head are affected, but, seems, you > >> need to be a bi

Re: aggregate crash

Alvaro Herrera writes: > On 2019-Dec-27, Teodor Sigaev wrote: >> Found crash on production instance, assert-enabled build crashes in pfree() >> call, with default config. v11, v12 and head are affected, but, seems, you >> need to be a bit lucky. > Is this bug being considered for the next set of

Re: aggregate crash

On 2019-Dec-27, Teodor Sigaev wrote: > Hi! > > Found crash on production instance, assert-enabled build crashes in pfree() > call, with default config. v11, v12 and head are affected, but, seems, you > need to be a bit lucky. Is this bug being considered for the next set of minors? -- Álvaro H

Re: aggregate crash

Hi, On 2019-12-27 20:13:26 +0300, Teodor Sigaev wrote: > Found crash on production instance, assert-enabled build crashes in pfree() > call, with default config. v11, v12 and head are affected, but, seems, you > need to be a bit lucky. > > The bug is comparing old and new aggregate pass-by-ref va

aggregate crash

Hi! Found crash on production instance, assert-enabled build crashes in pfree() call, with default config. v11, v12 and head are affected, but, seems, you need to be a bit lucky. The bug is comparing old and new aggregate pass-by-ref values only by pointer value itself, despite on null flag.