| Summary: | Buffer overflow in mail-filter/opendkim-1.2.1 | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Tilman Giese <gentoo> |
| Component: | Current packages | Assignee: | Daniel Black (RETIRED) <dragonheart> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | ||
| Priority: | High | ||
| Version: | unspecified | ||
| Hardware: | AMD64 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: | Patch that fixes a negative length for sprintf | ||
|
Description
Tilman Giese
2010-01-19 08:59:31 UTC
ok I just added opendkim-1.2.2 though its changelog doesn't mention anything that could explicitly be the cause of this. If you're willing to test this one I'd be grateful. I'm willing to take your personal email for testing. I should be able to identify a cause there. At most I'll share it with Murray S. Kucherawy, opendkim's lead developer. Can you attach/email your configuration file too. No need for private keys however and idea how you have it configured would be great. Cheers, Daniel Black OpenDKIM maintainer and upstream developer It turns out that version 1.2.2 has the same problem. If I send exactly the same message, I get the buffer overflow again. However, looking at the CFLAGS might be helpful. Mine were set to CFLAGS="-march=athlon64 -O3 -pipe -fforce-addr". If I change -O3 to -O2 the problem goes away and the message passes successfully. My configuration is really simple: Domain abfallfabrik.de,ahoh.de KeyFile /etc/opendkim/luthien.private Selector luthien Socket local:/var/run/opendkim/opendkim.sock UserID milter UMask 002 Statistics /var/lib/opendkim/stats.db I will send you the personal email. Created attachment 217012 [details, diff]
Patch that fixes a negative length for sprintf
Murray S. Kucherawy finally found the problem. "It's an sprintf() that is accidentally passed a negative length. The target buffer actually has plenty of room, but the computed length in the loop is subtracted too quickly." I attached the patch that is already committed and will be part of the next release. (In reply to comment #4) > Murray S. Kucherawy finally found the problem. Thank you both. opendkim-1.2.2-r1 added with this patch. Thanks for your bug report and involvement in getting it fixed Tim. Much appreciated. |