[exim-dev] Re: is this right?

Αρχική Σελίδα
Delete this message
Reply to this message
Συντάκτης: Jasen Betts
Ημερομηνία:  
Προς: exim-dev
Αντικείμενο: [exim-dev] Re: is this right?
On 2024-07-05, Jeremy Harris via Exim-dev <exim-dev@???> wrote:
> On 03/07/2024 02:03, Jasen Betts via Exim-dev wrote:
>> src/src/transports/smtp.c areound line 4500 says:
>> 
>> 
>> /* If the peer closed the TCP connection after end-of-data, but before
>> we could send QUIT, do TLS close, etc - call it a message error.
>> Otherwise, if all the recipients have been dealt with, call a close no
>> error at all; each address_item should have a suitable result already
>> (2xx: PENDING_OK, 4xx: DEFER, 5xx: FAIL) */
>> 
>> if (!(message_error = Ustrncmp(smtp_command,"end ",4) == 0))
>>      
>> Is that test supposed to have both "== 0" on the end and "!" on the front?
>> It does not seem to match the comment.
>
> My read of it, just now, is "yes it is (supposed to...)".
>
> We're in a case choice for an errno setting that is used to flag a peer-close detection.
> The strncmp==0 establishes that we have reached end-of-data.
> We set the boolean variable "message_error" based on that.
>
> If it is not a message_error here (so, was not at end-of-data in the SMTP conversation)
> we do stuff corresponding to the "Otherwise" of the comment: "call it (the peer close)
> not an error" IF all address_item has a suitable result (which they should).  In other
> words, don't bitch about a non-conforming peer that fails to properly close off the
> SMTP conversation.
>
> Should the "all have a suitable result" not hold for some reason, carry on in the
> error processing path (though we've decided it's not a "message_error").


I'm seeing it run the true branch of that if for "Remote host closed connection in
response to DATA" resulting in discarded (instead of retried and delivered) email
messages.

Might need to also add "DATA" and "BDAT" checks on smtp_command ?



--
Jasen.
🇺🇦 Слава Україні

--
## subscription configuration (requires account):
## https://lists.exim.org/mailman3/postorius/lists/exim-dev.lists.exim.org/
## unsubscribe (doesn't require an account):
## exim-dev-unsubscribe@???
## Exim details at http://www.exim.org/
## Please use the Wiki with this list - http://wiki.exim.org/