1 SENDMAIL RELEASE NOTES
| 1 SENDMAIL RELEASE NOTES
|
2 $Id: RELEASE_NOTES,v 8.1991 2011/05/15 04:28:16 ca Exp $
| 2 $Id: RELEASE_NOTES,v 8.2011 2012/12/21 18:42:16 ca Exp $
|
3 4 5This listing shows the version of the sendmail binary, the version 6of the sendmail configuration files, the date of release, and a 7summary of the changes in that release. 8
| 3 4 5This listing shows the version of the sendmail binary, the version 6of the sendmail configuration files, the date of release, and a 7summary of the changes in that release. 8
|
| 98.14.6/8.14.6 2012/12/23 10 Fix a regression introduced in 8.14.5: if a server offers 11 two AUTH lines, the MTA would not read them after 12 STARTTLS has been used and hence SMTP AUTH for 13 the client side would fail. Problem noted by Lena. 14 Do not cache hostnames internally in a non case sensitive way 15 as that may cause addresses to change from lower case 16 to upper case or vice versa. These header modifications 17 can cause problems with milters that rely on receiving 18 headers in the same way as they are being sent out such 19 as a DKIM signing milter. 20 If MaxQueueChildren is set then it was possible that new queue 21 runners could not be started anymore because an 22 internal counter was subject to a race condition. 23 If a milter decreases the timeout it waits for a communication 24 with the MTA, the MTA might experience a write() timeout. 25 In some situations, the resulting error might have been 26 ignored. Problem noted by Werner Wiethege. 27 Note: decreasing the communication timeout in a milter 28 should not be done without considering the potential 29 problems. 30 smfi_setsymlist() now properly sets the list of macros for 31 the milter which invoked it, instead of a global 32 list for all milters. Problem reported by 33 David Shrimpton of the University of Queensland. 34 If Timeout.resolver.retrans is set to a value larger than 20, 35 then resolver.retry was temporarily set to 0 for 36 gethostbyaddr() lookups. Now it is set to 1 instead. 37 Patch from Peter. 38 If sendmail could not lock the statistics file due to a system 39 error, and sendmail later sends a DSN for a mail that 40 triggered such an error, then sendmail tried to access 41 memory that was freed before (causing a crash on some 42 systems). Problem reported by Ryan Stone. 43 Do not log negative values for size= nor pri= to avoid confusing 44 log parsers, instead limit the values to LONG_MAX. 45 Account for an API change in newer versions of Cyrus-SASL. 46 Patch from Hajimu UMEMOTO from FreeBSD. 47 Do not try to resolve link-local addresses for IPv4 (just as it 48 is done for IPv6). Patch from John Beck of Oracle. 49 Improve logging of client and server STARTTLS connection failures 50 that may be due to incompatible cipher lists by including 51 the reason for the failure in a single log line. Suggested 52 by James Carey of Boeing. 53 Portability: 54 Add support for Darwin 11.x and 12.x (Mac OS X 10.7 and 10.8). 55 Add support for SunOS 5.12 (aka Solaris 12). Patch from 56 John Beck of Oracle. 57
|
98.14.5/8.14.5 2011/05/17 10 Do not cache SMTP extensions across connections as the cache 11 is based on hostname which may not be a unique identifier 12 for a server, i.e., different machines may have the 13 same hostname but provide different SMTP extensions. 14 Problem noted by Jim Hermann. 15 Avoid an out-of-bounds access in case a resolver reply for a DNS 16 map lookup returns a size larger than 1K. Based on a
--- 10429 unchanged lines hidden --- | 588.14.5/8.14.5 2011/05/17 59 Do not cache SMTP extensions across connections as the cache 60 is based on hostname which may not be a unique identifier 61 for a server, i.e., different machines may have the 62 same hostname but provide different SMTP extensions. 63 Problem noted by Jim Hermann. 64 Avoid an out-of-bounds access in case a resolver reply for a DNS 65 map lookup returns a size larger than 1K. Based on a
--- 10429 unchanged lines hidden --- |