Bug#310327: patch
Aníbal Monsalve Salazar wrote: Upon investigation of this problem I noticed that ssmtp (oldstable and stable) always strips the last line of the input before sending. gluck!joey(pts/4):~ seq 1 10|sendmail [EMAIL PROTECTED] -- 1..9 gluck!joey(pts/4):~ echo seq 1 10|sendmail [EMAIL PROTECTED] -- no lines This is not fixed by the above patch. I've patched ssmtp.c to fix the problem above and to close #310327 in ssmtp 2.61-5. It can be argued that you _must_ separate the data from the mail headers with an empty line. The following works: (echo; seq 1 10)|sendmail [EMAIL PROTECTED] If you want to really fix it, I can send you patches for the oldstable and stable versions of ssmtp. I can also prepare new packages as well. Please let me know what would you like me to do. Oh, I see. If I recall correctly, I've used mailx to build the mail before so that there was a header. Well, then it may not be a bug but a feature and we can leave it as it is. The other problem, the subject of #310327, is fixed in ssmtp 2.61-5 by the patch at: http://bugs.debian.org/cgi-bin/bugreport.cgi/ssmtp_2.61-4_non_blocking_fgets.diff?bug=310327;msg=64;att=2 That patch is not requiered for the versions of ssmtp currently in oldstable and stable as bug #310327 was introduced with ssmtp 2.61-3. Good news. Thanks. Regards, Joey -- All language designers are arrogant. Goes with the territory... -- Larry Wall Please always Cc to me when replying to me on the lists.
Bug#310327: patch
On Fri, Aug 26, 2005 at 10:47:29AM +0200, Martin Schulze wrote: Upon investigation of this problem I noticed that ssmtp (oldstable and stable) always strips the last line of the input before sending. gluck!joey(pts/4):~ seq 1 10|sendmail [EMAIL PROTECTED] -- 1..9 gluck!joey(pts/4):~ echo seq 1 10|sendmail [EMAIL PROTECTED] -- no lines This is not fixed by the above patch. I've patched ssmtp.c to fix the problem above and to close #310327 in ssmtp 2.61-5. It can be argued that you _must_ separate the data from the mail headers with an empty line. The following works: (echo; seq 1 10)|sendmail [EMAIL PROTECTED] If you want to really fix it, I can send you patches for the oldstable and stable versions of ssmtp. I can also prepare new packages as well. Please let me know what would you like me to do. The other problem, the subject of #310327, is fixed in ssmtp 2.61-5 by the patch at: http://bugs.debian.org/cgi-bin/bugreport.cgi/ssmtp_2.61-4_non_blocking_fgets.diff?bug=310327;msg=64;att=2 That patch is not requiered for the versions of ssmtp currently in oldstable and stable as bug #310327 was introduced with ssmtp 2.61-3. Regards, Joey -- WARNING: Do not execute! This call violates patent DE10108564. http://www.elug.de/projekte/patent-party/patente/DE10108564 wget -O patinfo-`date +%Y%m%d`.html http://patinfo.ffii.org/ Please always Cc to me when replying to me on the lists. -- Mit freundlichen Grüßen, Anibal Monsalve Salazar -- .''`. Debian GNU/Linux : :' : Free Operating System `. `' http://debian.org/ `- http://v7w.com/anibal signature.asc Description: Digital signature
Bug#310327: patch fixing the stdin-no-headers scenario
This quick patch should fix the problem of ssmtp eating lines when there are no headers supplied on stdin, as Martin Schulze remarked before. Ssmtp expects to receive 'headers - empty line - body', and when only the body is included, it tries to put all lines in the header and creatively eats some at the end. On its way, MTAs usually add real headers to the message, making the body appear again but incomplete. What this patch does, is checking the first line for a colon and jumping out of the headers routine if the message doesn't appear to have a header. Can somebody verify if this fixes at least that problem? Regards, Wouter --- ssmtp-2.61/ssmtp.c 2005-09-04 18:05:20.0 +0200 +++ ssmtp-2.61-wvh/ssmtp.c 2005-09-04 19:19:53.0 +0200 @@ -727,11 +727,12 @@ /* header_parse() -- Break headers into seperate entries */ -void header_parse(FILE *stream) +char *header_parse(FILE *stream) { size_t size = BUF_SZ, len = 0; char *p = (char *)NULL, *q; bool_t in_header = True; + bool_t on_first_line = True; char l = (char)NULL; int c; @@ -749,7 +750,15 @@ } len++; + /* -wvh- sniff out the first line to see if it really is a header */ + if (on_first_line c == '\n' strchr(p, ':') == (char *)NULL) { + in_header = False; + } + if(l == '\n') { + + on_first_line = False; + switch(c) { case ' ': case '\t': @@ -781,7 +790,14 @@ l = c; } + + if (!in_header on_first_line) { + return (p); + } + (void)free(p); + + return (char *)NULL; } /* @@ -1319,6 +1335,7 @@ struct passwd *pw; int i, sock; uid_t uid; + char *head_body = (char *)NULL; outbytes = 0; @@ -1347,7 +1364,7 @@ ht = headers; rt = rcpt_list; - header_parse(stdin); + head_body = header_parse(stdin); #if 1 /* With FromLineOverride=YES set, try to recover sane MAIL FROM address */ @@ -1528,6 +1545,11 @@ /* End of headers, start body */ outbytes += smtp_write(sock, ); + if ((char *)head_body != (char *)NULL) { + standardise(head_body); + outbytes += smtp_write(sock, %s, head_body); +} + /*prevent blocking on pipes, we really shouldnt be using stdio functions like fgets in the first place */ fcntl(STDIN_FILENO,F_SETFL,O_NONBLOCK);
Bug#310327: patch
Aidas Kasparas wrote: Please find bellow a patch which check EOF condition instead of no input. Without fix for this bug package is virtually not useable (I experienced mysterious attachment cuts, so I can not relay on it at it's present form :-( Please consider importance of this bug as serious at the very least. --- ssmtp.c.R 2005-08-25 19:41:15.0 +0300 +++ ssmtp.c 2005-08-25 19:45:11.0 +0300 @@ -1532,7 +1532,13 @@ stdio functions like fgets in the first place */ fcntl(STDIN_FILENO,F_SETFL,O_NONBLOCK); - while(fgets(buf, sizeof(buf), stdin)) { + while(!feof(stdin)) { + if (!fgets(buf, sizeof(buf), stdin)) { + /* if nothing was received, then no transmission +* over smtp should be done */ + sleep(1); + continue; + } /* Trim off \n, double leading .'s */ standardise(buf); - 8 Upon investigation of this problem I noticed that ssmtp (oldstable and stable) always strips the last line of the input before sending. gluck!joey(pts/4):~ seq 1 10|sendmail [EMAIL PROTECTED] -- 1..9 gluck!joey(pts/4):~ echo seq 1 10|sendmail [EMAIL PROTECTED] -- no lines This is not fixed by the above patch. Regards, Joey -- WARNING: Do not execute! This call violates patent DE10108564. http://www.elug.de/projekte/patent-party/patente/DE10108564 wget -O patinfo-`date +%Y%m%d`.html http://patinfo.ffii.org/ Please always Cc to me when replying to me on the lists. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#310327: patch
Please find bellow a patch which check EOF condition instead of no input. Without fix for this bug package is virtually not useable (I experienced mysterious attachment cuts, so I can not relay on it at it's present form :-( Please consider importance of this bug as serious at the very least. --- ssmtp.c.R 2005-08-25 19:41:15.0 +0300 +++ ssmtp.c 2005-08-25 19:45:11.0 +0300 @@ -1532,7 +1532,13 @@ stdio functions like fgets in the first place */ fcntl(STDIN_FILENO,F_SETFL,O_NONBLOCK); - while(fgets(buf, sizeof(buf), stdin)) { + while(!feof(stdin)) { + if (!fgets(buf, sizeof(buf), stdin)) { + /* if nothing was received, then no transmission +* over smtp should be done */ + sleep(1); + continue; + } /* Trim off \n, double leading .'s */ standardise(buf); - 8 -- Aidas Kasparas IT administrator GM Consult Group, UAB -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]