On Sat, Jul 08, 2023 at 07:08:35PM -0400, Joseph Koshakow wrote:
> On Sat, Jul 8, 2023 at 6:09 PM Nathan Bossart <nathandbossart@gmail.com>
> wrote:
>
>>> I think the issue here is that if a session loses the ability to set
>>> their session authorization in the middle of a transaction, then
>>> rolling back the transaction may fail and cause the server to panic.
>>> That's probably what the deleted comment mean when it said:
>>>
>>>> * It's OK because the check does not require catalog access and can't
>>>> * fail during an end-of-transaction GUC reversion
>>
>> Yeah. IIUC the ERROR longjmps to a block that calls AbortTransaction(),
>> which ERRORs again when resetting the session authorization, which causes
>> us to call AbortTransaction() again, etc., etc.
src/backend/utils/misc/README has the following relevant text:
Note that there is no provision for a failure result code. assign_hooks
should never fail except under the most dire circumstances, since a failure
may for example result in GUC settings not being rolled back properly during
transaction abort. In general, try to do anything that could conceivably
fail in a check_hook instead, and pass along the results in an "extra"
struct, so that the assign hook has little to do beyond copying the data to
someplace. This applies particularly to catalog lookups: any required
lookups must be done in the check_hook, since the assign_hook may be
executed during transaction rollback when lookups will be unsafe.
> Everything seems to work fine if the privilege check is moved to
> check_session_authorization. Which is maybe what the comment meant
> instead of assign_session_authorization.
Ah, that does make more sense.
I think we should split this into two patches: one to move the permission
check to check_session_authorization() and another for the behavior change.
I've attached an attempt at the first one (that borrows heavily from your
latest patch). AFAICT the only reason that the permission check lives in
SetSessionAuthorization() is because AuthenticatedUserIsSuperuser is static
to miscinit.c and doesn't have an accessor function. I added one, but it
would probably just be removed by the following patch. WDYT?
--
Nathan Bossart
Amazon Web Services: https://aws.amazon.com
