Hi, (proposed patch to maildrop/formatmbox.C attached)
This issue was reported to Debian BTS: http://bugs.debian.org/520344 on Mar 2009 and I recently took part in its maintenance. Problem: The maildrop's log file generates funny line when "from:" exists in the DKIM header content like in the attached example mailbox.txt. You can reproduce this by running: /usr/bin/maildrop < mailbox.txt This generate log entry as (space indented with | for readability): |Date: Sun Jun 17 10:29:24 2012 |From: "Barnikow, Katja" <[email protected]>message-id:date:subject:con |Subj: Test |File: /var/mail/osamu I checked this source and I think I found the bug :-) This bug is in maildrop/formatmbox.C where pointer skipping excess whitespaces (but not banklines) at the start of line. There is no need to do this since From: header always start at the first character. Continuation lines in header files have leading space. Patch is for upstream 2.5.5 version. In old 2.0.3 source, this bug caused the From: header content to be overwritten by the second match and not concatenated like now. It is probably because 2.0.3 version was not using "bufp->append(p, l);" but making simple assignment in the last part of this function. So this patch should fix to match only ^From: but if this header exist twice, those contents are concatenated (without space). Osamu
>From [email protected] From: "Barnikow, Katja" <[email protected]> To: "Ralf Hildebrandt" <[email protected]> Subject: Test DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=charite.de; h= from:message-id:date:subject:content-type:mime-version :content-class:received:received:received:x-virus-scanned;s= default; t=1237383677; bh=7Q6qSv4vMWBVhAT79vzOdErsuTcjoLR3jOcVw/ 21upA=; b=IEVP0Ip6pwN8cG2vyAr7Gk2gaEmX7ZXrbKPbzGzip+GCcN9KihZOpo Ho49OItUmaGGCLeGudu+0rukZ95vSDFlvHqL5JBLU2NWuD/V9LS2lYl9upXDsnyp NNI6s7SWf8ZKjPdtPYDC2hc+iyk2AJQCk6XLmAl8QHR38ZuD5W2tI= Body
--- formatmbox.C.orig 2012-06-17 14:42:59.176829893 +0900 +++ formatmbox.C 2012-06-17 14:50:50.135165248 +0900 @@ -85,7 +85,6 @@ const char *p=msglinebuf; Buffer *bufp=0; - while (*p != '\n' && isspace(*p)) p++; if ( tolower(*p) == 'f' && tolower(p[1]) == 'r' && tolower(p[2]) == 'o' && tolower(p[3]) == 'm' && p[4] == ':')
