mir/postfix
2
0
Fork 0
mirror of https://github.com/vdukhovni/postfix synced 2025-08-30 13:48:06 +00:00
Code Issues Releases Wiki Activity

postfix-3.4-20181125

Browse Source
This commit is contained in:
Wietse Venema 2018-11-25 00:00:00 -05:00 committed by Viktor Dukhovni
parent 3525c001ea
commit d5f04c8f9e
20 changed files with 236 additions and 164 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
postfix/HISTORY
View File

@ -23842,3 +23842,15 @@ Apologies for any names omitted.
tls_client_init request specifies an unexpected client
identity, and the test for that condition is now moved to
the right place. File: tlsproxy/tlsproxy.c.
20181124
Documentation: clarified the behavior of whitespace within
"{}". Files: proto/DATABASE_README.html, proto/postconf.proto,
pipe/pipe.c, postconf/postconf.c,
20181125
Cleanup: dict_file_to_xxx() takes a list of file names
separated by CHARS_COMMA_SP. Shoe-horned into the existing
API, make it nicer when there is time. File: util/dict_file.c.

12
postfix/README_FILES/DATABASE_README
View File

@ -215,9 +215,11 @@ To find out what database types your Postfix system supports, use the "ppooss
iinnlliinnee (read-only)
A non-shared, in-memory lookup table. Example: "inline:{ key=value,
{ key = text with whitespace or comma }}". Key-value pairs are
separated by whitespace or comma; whitespace after "{" and before "}"
is ignored. Inline tables eliminate the need to create a database file
for just a few fixed elements. See also the static: map type.
separated by whitespace or comma; with a key-value pair inside "{}",
whitespace is ignored after the opening "{", around the "=" between key
and value, and before the closing "}". Inline tables eliminate the need
to create a database file for just a few fixed elements. See also the
static: map type.
iinntteerrnnaall
A non-shared, in-memory hash table. Its content are lost when a process
terminates.
@ -289,8 +291,8 @@ To find out what database types your Postfix system supports, use the "ppooss
A table that always returns its name as the lookup result. For example,
"static:foobar" always returns the string "foobar" as lookup result.
Specify "static:{ text with whitespace }" when the result contains
whitespace; this form ignores whitespace after "{" and before "}". See
also the inline: map type.
whitespace; this form ignores whitespace after the opening "{" and
before the closing "}". See also the inline: map type.
ttccpp
TCP/IP client. The protocol is described in tcp_table(5). The lookup
table name is "tcp:host:port" where "host" specifies a symbolic

7
postfix/WISHLIST
View File

@ -1,7 +1,10 @@
Wish list:
With DICT_FLAG_RHS_IS_FILE, RHS is list of files (insert
newline for robustness).
Declumsify code that calls vstream_fread() to read data
into VSTRING buffer. This needs vstream_fread_buf(VSTREAM *,
VSTRING *, ssize_t) which calls VSTRING_SPACE(), appends
data to the buffer, and calls VSTRING_AT_OFFSET(). Current
code is error-prone.
With DICT_FLAG_RHS_IS_FILE, should dict_update() open a
file? base64-encode the value?

9
postfix/html/DATABASE_README.html
View File

@ -317,8 +317,10 @@ name as used in "<a href="DATABASE_README.html#types">hash</a>:table" is the dat
<dd> A non-shared, in-memory lookup table. Example: "<a href="DATABASE_README.html#types">inline</a>:{
<i>key=value</i>, { <i>key = text with whitespace or comma</i> }}".
Key-value pairs are separated by whitespace or comma; whitespace
after "{" and before "}" is ignored. Inline tables eliminate the
Key-value pairs are separated by whitespace or comma; with a key-value
pair inside "{}", whitespace is ignored after the opening "{",
around the "=" between key and value, and before the closing "}".
Inline tables eliminate the
need to create a database file for just a few fixed elements. See
also the <a href="DATABASE_README.html#types">static</a>: map type. </dd>
@ -436,7 +438,8 @@ server. See <a href="socketmap_table.5.html">socketmap_table(5)</a> for details.
For example, "<a href="DATABASE_README.html#types">static</a>:foobar" always returns the string "foobar" as
lookup result. Specify "<a href="DATABASE_README.html#types">static</a>:{ <i>text with whitespace</i> }"
when the result contains whitespace; this form ignores whitespace
after "{" and before "}". See also the <a href="DATABASE_README.html#types">inline</a>: map type. </dd>
after the opening "{" and before the closing "}". See also the
<a href="DATABASE_README.html#types">inline</a>: map type. </dd>
<dt> <b>tcp</b> </dt>

4
postfix/html/pipe.8.html
View File

@ -192,8 +192,8 @@ PIPE(8) PIPE(8)
interpreter.
Specify "{" and "}" around command arguments that contain white-
space (Postfix 3.0 and later). Whitespace after "{" and before
"}" is ignored.
space (Postfix 3.0 and later). Whitespace after the opening "{"
and before the closing "}" is ignored.
In the command argument vector, the following macros are recog-
nized and replaced with corresponding information from the Post-

13
postfix/html/postconf.1.html
View File

@ -243,10 +243,11 @@ POSTCONF(1) POSTCONF(1)
A non-shared, in-memory lookup table. Example: "<b><a href="DATABASE_README.html#types">inline</a>:{</b>
<i>key</i><b>=</b><i>value</i><b>, {</b> <i>key</i> <b>=</b> <i>text with whitespace or comma</i> <b>}}</b>".
Key-value pairs are separated by whitespace or comma;
whitespace after "<b>{</b>" and before "<b>}</b>" is ignored. Inline
tables eliminate the need to create a database file for
just a few fixed elements. See also the <i><a href="DATABASE_README.html#types">static</a>:</i> map
type.
with a key-value pair inside "<b>{}</b>", whitespace is ignored
after the opening "<b>{</b>", around the "<b>=</b>" between key and
value, and before the closing "<b>}</b>". Inline tables elimi-
nate the need to create a database file for just a few
fixed elements. See also the <i><a href="DATABASE_README.html#types">static</a>:</i> map type.
This feature is available with Postfix 3.0 and later.
@ -340,8 +341,8 @@ POSTCONF(1) POSTCONF(1)
For example, <b><a href="DATABASE_README.html#types">static</a>:foobar</b> always returns the string <b>foo-</b>
<b>bar</b> as lookup result. Specify "<b><a href="DATABASE_README.html#types">static</a>:{</b> <i>text with white-</i>
<i>space</i> <b>}</b>" when the result contains whitespace; this form
ignores whitespace after "<b>{</b>" and before "<b>}</b>". See also the
<i><a href="DATABASE_README.html#types">inline</a>:</i> map.
ignores whitespace after the opening "<b>{</b>" and before the
closing "<b>}</b>". See also the <i><a href="DATABASE_README.html#types">inline</a>:</i> map.
The form "<b><a href="DATABASE_README.html#types">static</a>:{</b><i>text</i><b>}</b> is available with Postfix 3.0 and
later.

6
postfix/html/postconf.5.html
View File

@ -3308,7 +3308,8 @@ time keeping on System-V-ish systems.
<p>
Specify a list of names and/or name=value pairs, separated by
whitespace or comma. Specify "{ name=value }" to protect whitespace
or comma in parameter values (whitespace after "{" and before "}"
or comma in parameter values (whitespace after the opening "{" and
before the closing "}"
is ignored). The form name=value is supported with Postfix version
2.1 and later; the use of {} is supported with Postfix 3.0 and
later. </p>
@ -3885,7 +3886,8 @@ environment. Examples of relevant parameters: </p>
<p> Specify a list of names and/or name=value pairs, separated by
whitespace or comma. Specify "{ name=value }" to protect whitespace
or comma in parameter values (whitespace after "{" and before "}"
or comma in parameter values (whitespace after the opening "{" and
before the closing "}"
is ignored). The form name=value is supported with Postfix version
2.1 and later; the use of {} is supported with Postfix 3.0 and
later. </p>

9
postfix/man/man1/postconf.1
View File

@ -274,8 +274,10 @@ with support for Berkeley DB databases.
A non\-shared, in\-memory lookup table. Example: "\fBinline:{
\fIkey\fB=\fIvalue\fB, { \fIkey\fB = \fItext with whitespace
or comma\fB }}\fR". Key\-value pairs are separated by
whitespace or comma; whitespace after "\fB{\fR" and before "\fB}\fR"
is ignored. Inline tables eliminate the need to create a
whitespace or comma; with a key\-value pair inside "\fB{}\fR",
whitespace is ignored after the opening "\fB{\fR", around
the "\fB=\fR" between key and value, and before the closing
"\fB}\fR". Inline tables eliminate the need to create a
database file for just a few fixed elements. See also the
\fIstatic:\fR map type.
@ -359,7 +361,8 @@ A table that always returns its name as lookup result. For
example, \fBstatic:foobar\fR always returns the string
\fBfoobar\fR as lookup result. Specify "\fBstatic:{ \fItext
with whitespace\fB }\fR" when the result contains whitespace;
this form ignores whitespace after "\fB{\fR" and before
this form ignores whitespace after the opening "\fB{\fR"
and before the closing
"\fB}\fR". See also the \fIinline:\fR map.
The form "\fBstatic:{\fItext\fB}\fR is available with Postfix

6
postfix/man/man5/postconf.5
View File

@ -2071,7 +2071,8 @@ time keeping on System\-V\-ish systems.
.PP
Specify a list of names and/or name=value pairs, separated by
whitespace or comma. Specify "{ name=value }" to protect whitespace
or comma in parameter values (whitespace after "{" and before "}"
or comma in parameter values (whitespace after the opening "{" and
before the closing "}"
is ignored). The form name=value is supported with Postfix version
2.1 and later; the use of {} is supported with Postfix 3.0 and
later.
@ -2409,7 +2410,8 @@ Needed to make "\fBpostfix \-c\fR" work.
.PP
Specify a list of names and/or name=value pairs, separated by
whitespace or comma. Specify "{ name=value }" to protect whitespace
or comma in parameter values (whitespace after "{" and before "}"
or comma in parameter values (whitespace after the opening "{" and
before the closing "}"
is ignored). The form name=value is supported with Postfix version
2.1 and later; the use of {} is supported with Postfix 3.0 and
later.

2
postfix/man/man8/pipe.8
View File

@ -201,7 +201,7 @@ shell meta characters by a shell command interpreter.
.sp
Specify "{" and "}" around command arguments that contain
whitespace (Postfix 3.0 and later). Whitespace
after "{" and before "}" is ignored.
after the opening "{" and before the closing "}" is ignored.
.sp
In the command argument vector, the following macros are recognized
and replaced with corresponding information from the Postfix queue

9
postfix/proto/DATABASE_README.html
View File

@ -317,8 +317,10 @@ name as used in "hash:table" is the database file name without the
<dd> A non-shared, in-memory lookup table. Example: "inline:{
<i>key=value</i>, { <i>key = text with whitespace or comma</i> }}".
Key-value pairs are separated by whitespace or comma; whitespace
after "{" and before "}" is ignored. Inline tables eliminate the
Key-value pairs are separated by whitespace or comma; with a key-value
pair inside "{}", whitespace is ignored after the opening "{",
around the "=" between key and value, and before the closing "}".
Inline tables eliminate the
need to create a database file for just a few fixed elements. See
also the static: map type. </dd>
@ -436,7 +438,8 @@ server. See socketmap_table(5) for details. </dd>
For example, "static:foobar" always returns the string "foobar" as
lookup result. Specify "static:{ <i>text with whitespace</i> }"
when the result contains whitespace; this form ignores whitespace
after "{" and before "}". See also the inline: map type. </dd>
after the opening "{" and before the closing "}". See also the
inline: map type. </dd>
<dt> <b>tcp</b> </dt>

6
postfix/proto/postconf.proto
View File

@ -1447,7 +1447,8 @@ time keeping on System-V-ish systems.
<p>
Specify a list of names and/or name=value pairs, separated by
whitespace or comma. Specify "{ name=value }" to protect whitespace
or comma in parameter values (whitespace after "{" and before "}"
or comma in parameter values (whitespace after the opening "{" and
before the closing "}"
is ignored). The form name=value is supported with Postfix version
2.1 and later; the use of {} is supported with Postfix 3.0 and
later. </p>
@ -1918,7 +1919,8 @@ environment. Examples of relevant parameters: </p>
<p> Specify a list of names and/or name=value pairs, separated by
whitespace or comma. Specify "{ name=value }" to protect whitespace
or comma in parameter values (whitespace after "{" and before "}"
or comma in parameter values (whitespace after the opening "{" and
before the closing "}"
is ignored). The form name=value is supported with Postfix version
2.1 and later; the use of {} is supported with Postfix 3.0 and
later. </p>

2
postfix/src/global/mail_version.h
View File

@ -20,7 +20,7 @@
* Patches change both the patchlevel and the release date. Snapshots have no
* patchlevel; they change the release date only.
*/
#define MAIL_RELEASE_DATE "20181123"
#define MAIL_RELEASE_DATE "20181125"
#define MAIL_VERSION_NUMBER "3.4"
#ifdef SNAPSHOT

2
postfix/src/pipe/pipe.c
View File

@ -191,7 +191,7 @@
/* .sp
/* Specify "{" and "}" around command arguments that contain
/* whitespace (Postfix 3.0 and later). Whitespace
/* after "{" and before "}" is ignored.
/* after the opening "{" and before the closing "}" is ignored.
/* .sp
/* In the command argument vector, the following macros are recognized
/* and replaced with corresponding information from the Postfix queue

9
postfix/src/postconf/postconf.c
View File

@ -268,8 +268,10 @@
/* A non-shared, in-memory lookup table. Example: "\fBinline:{
/* \fIkey\fB=\fIvalue\fB, { \fIkey\fB = \fItext with whitespace
/* or comma\fB }}\fR". Key-value pairs are separated by
/* whitespace or comma; whitespace after "\fB{\fR" and before "\fB}\fR"
/* is ignored. Inline tables eliminate the need to create a
/* whitespace or comma; with a key-value pair inside "\fB{}\fR",
/* whitespace is ignored after the opening "\fB{\fR", around
/* the "\fB=\fR" between key and value, and before the closing
/* "\fB}\fR". Inline tables eliminate the need to create a
/* database file for just a few fixed elements. See also the
/* \fIstatic:\fR map type.
/*
@ -353,7 +355,8 @@
/* example, \fBstatic:foobar\fR always returns the string
/* \fBfoobar\fR as lookup result. Specify "\fBstatic:{ \fItext
/* with whitespace\fB }\fR" when the result contains whitespace;
/* this form ignores whitespace after "\fB{\fR" and before
/* this form ignores whitespace after the opening "\fB{\fR"
/* and before the closing
/* "\fB}\fR". See also the \fIinline:\fR map.
/*
/* The form "\fBstatic:{\fItext\fB}\fR is available with Postfix

27
postfix/src/tlsproxy/tlsproxy.c
View File

@ -1026,10 +1026,10 @@ static int tlsp_client_init(TLS_APPL_STATE **client_appl_state,
/*
* Use one TLS_APPL_STATE object for all requests that specify the same
* TLS_CLIENT_INIT_PROPS. Each TLS_APPL_STATE owns an SSL_CTX, which is
* expensive.
* expensive to create.
*
* First, compute the TLS_APPL_STATE cache lookup key. Save a copy of the
* key that corresponds to the pre-jail internal call, which uses the
* key that corresponds to the pre-jail internal request, which uses the
* tlsproxy_client_* settings.
*/
buf = vstring_alloc(100);
@ -1042,9 +1042,9 @@ static int tlsp_client_init(TLS_APPL_STATE **client_appl_state,
/*
* Log a warning if a post-jail request differs from the tlsproxy_client_*
* settings AND the request specifies file or directory arguments. Those
* are problematic after chroot (pathname resolution) and after dropping
* privileges (key files must be root read-only).
* settings AND the request specifies file/directory pathname arguments.
* Those are problematic after chroot (pathname resolution) and after
* dropping privileges (key files must be root read-only).
*
* We can eliminate this complication by adding code that opens a cert/key
* lookup table at pre-jail time, and by reading cert/key info on-the-fly
@ -1410,17 +1410,20 @@ static void pre_jail_init(char *unused_name, char **unused_argv)
* for the minority of sites that want to use TLS connection caching with
* multiple TLS client identities. To alert the operator, tlsproxy will
* log a warning when a TLS_CLIENT_INIT message specifies a different
* configuration with cert or key pathnames. The workaround is to have
* one tlsproxy process per TLS client identity.
* configuration than the tlsproxy pre-jail client configuration, and
* that different configuration specifies file/directory pathname
* arguments. The workaround is to have one tlsproxy process per TLS
* client identity.
*
* The general solution for single-identity or multi-identity clients is to
* stop loading certs and keys from individual files. Instead, have a
* cert/key map, indexed by client identity, read-only by root. After
* opening the map at pre-jail time, tlsproxy can read certs/keys
* on-the-fly at post-jail time. This is the approach that was already
* proposed for server-side SNI support, and it could be reused here. It
* would also end the proliferation of RSA cert/key parameters, DSA
* cert/key parameters, EC cert/key parameters, and so on.
* opening the map as root at pre-jail time, tlsproxy can read certs/keys
* on-the-fly as an unprivileged process at post-jail time. This is the
* approach that was already proposed for server-side SNI support, and it
* could be reused here. It would also end the proliferation of RSA
* cert/key parameters, DSA cert/key parameters, EC cert/key parameters,
* and so on.
*
* Horror: In order to create the same pre-jail TLS client context as the
* one used in the Postfix SMTP client, we have to duplicate intricate

71
postfix/src/util/dict_file.c
View File

@ -8,11 +8,11 @@
/*
/* VSTRING *dict_file_to_buf(
/* DICT *dict,
/* const char *pathname)
/* const char *pathnames)
/*
/* VSTRING *dict_file_to_b64(
/* DICT *dict,
/* const char *pathname)
/* const char *pathnames)
/*
/* VSTRING *dict_file_from_b64(
/* DICT *dict,
@ -24,11 +24,15 @@
/* void dict_file_purge_buffers(
/* DICT *dict)
/* DESCRIPTION
/* dict_file_to_buf() reads the content of the specified file.
/* dict_file_to_buf() reads the content of the specified
/* files, with names separated by CHARS_COMMA_SP, while inserting
/* a gratuitous newline character between files.
/* It returns a pointer to a buffer which is owned by the DICT,
/* or a null pointer in case of error.
/*
/* dict_file_to_b64() reads the content of the specified file,
/* dict_file_to_b64() reads the content of the specified
/* files, with names separated by CHARS_COMMA_SP, while inserting
/* a gratuitous newline character between files,
/* and converts the result to base64.
/* It returns a pointer to a buffer which is owned by the DICT,
/* or a null pointer in case of error.
@ -81,50 +85,71 @@
#define STR(x) vstring_str(x)
#define LEN(x) VSTRING_LEN(x)
/* dict_file_to_buf - read a file into a buffer */
/* dict_file_to_buf - read files into a buffer */
VSTRING *dict_file_to_buf(DICT *dict, const char *pathname)
VSTRING *dict_file_to_buf(DICT *dict, const char *pathnames)
{
struct stat st;
VSTREAM *fp;
VSTREAM *fp = 0;
ARGV *argv;
char **cpp;
/* dict_file_to_buf() postcondition: dict->file_buf exists. */
if (dict->file_buf == 0)
dict->file_buf = vstring_alloc(100);
if ((fp = vstream_fopen(pathname, O_RDONLY, 0)) == 0
|| fstat(vstream_fileno(fp), &st) < 0) {
vstring_sprintf(dict->file_buf, "open %s: %m", pathname);
if (fp)
vstream_fclose(fp);
return (0);
#define DICT_FILE_ERR_RETURN do { \
argv_free(argv); \
if (fp) vstream_fclose(fp); \
return (0); \
} while (0);
argv = argv_split(pathnames, CHARS_COMMA_SP);
if (argv->argc == 0) {
vstring_sprintf(dict->file_buf, "empty pathname list: >>%s<<'",
pathnames);
DICT_FILE_ERR_RETURN;
}
VSTRING_RESET(dict->file_buf);
for (cpp = argv->argv; *cpp; cpp++) {
if ((fp = vstream_fopen(*cpp, O_RDONLY, 0)) == 0
|| fstat(vstream_fileno(fp), &st) < 0) {
vstring_sprintf(dict->file_buf, "open %s: %m", *cpp);
DICT_FILE_ERR_RETURN;
}
if (st.st_size > SSIZE_T_MAX - LEN(dict->file_buf)) {
vstring_sprintf(dict->file_buf, "file too large: %s", pathnames);
DICT_FILE_ERR_RETURN;
}
VSTRING_SPACE(dict->file_buf, st.st_size);
if (vstream_fread(fp, STR(dict->file_buf), st.st_size) != st.st_size) {
vstring_sprintf(dict->file_buf, "read %s: %m", pathname);
vstream_fclose(fp);
return (0);
if (vstream_fread(fp, STR(dict->file_buf) + LEN(dict->file_buf),
st.st_size) != st.st_size) {
vstring_sprintf(dict->file_buf, "read %s: %m", *cpp);
DICT_FILE_ERR_RETURN;
}
(void) vstream_fclose(fp);
VSTRING_AT_OFFSET(dict->file_buf, st.st_size);
VSTRING_AT_OFFSET(dict->file_buf, LEN(dict->file_buf) + st.st_size);
if (cpp[1] != 0)
VSTRING_ADDCH(dict->file_buf, '\n');
}
argv_free(argv);
VSTRING_TERMINATE(dict->file_buf);
return (dict->file_buf);
}
/* dict_file_to_b64 - read a file into a base64-encoded buffer */
/* dict_file_to_b64 - read files into a base64-encoded buffer */
VSTRING *dict_file_to_b64(DICT *dict, const char *pathname)
VSTRING *dict_file_to_b64(DICT *dict, const char *pathnames)
{
ssize_t helper;
if (dict_file_to_buf(dict, pathname) == 0)
if (dict_file_to_buf(dict, pathnames) == 0)
return (0);
if (dict->file_b64 == 0)
dict->file_b64 = vstring_alloc(100);
helper = (VSTRING_LEN(dict->file_buf) + 2) / 3;
helper = (LEN(dict->file_buf) + 2) / 3;
if (helper > SSIZE_T_MAX / 4) {
vstring_sprintf(dict->file_buf, "file too large: %s", pathname);
vstring_sprintf(dict->file_buf, "file too large: %s", pathnames);
return (0);
}
VSTRING_RESET(dict->file_b64);

1
postfix/src/util/dict_pcre_file.in
View File

@ -1,3 +1,4 @@
get file1
get file2
get file3
get files12

3
postfix/src/util/dict_pcre_file.map
View File

@ -1,3 +1,6 @@
/file1/ dict_pcre_file1
/file2/ dict_pcre_file2
/file3/ dict_pcre_file3
/files12/ dict_pcre_file1, dict_pcre_file2
/files13/ dict_pcre_file1, dict_pcre_file3
/file-comma/ ,

4
postfix/src/util/dict_pcre_file.ref
View File

@ -1,4 +1,6 @@
./dict_open: warning: pcre map dict_pcre_file.map, line 3: open dict_pcre_file3: No such file or directory: skipping this rule
./dict_open: warning: pcre map dict_pcre_file.map, line 5: open dict_pcre_file3: No such file or directory: skipping this rule
./dict_open: warning: pcre map dict_pcre_file.map, line 6: empty pathname list: >>,<<': skipping this rule
owner=untrusted (uid=USER)
> get file1
file1=dGhpcy1pcy1maWxlMQo=
@ -6,3 +8,5 @@ file1=dGhpcy1pcy1maWxlMQo=
file2=dGhpcy1pcy1maWxlMgo=
> get file3
file3: not found
> get files12
files12=dGhpcy1pcy1maWxlMQoKdGhpcy1pcy1maWxlMgo=