Community Packages

Please read this before reporting a bug:
https://wiki.archlinux.org/title/Bug_reporting_guidelines

Do NOT report bugs when a package is just outdated, or it is in the AUR. Use the 'flag out of date' link on the package page, or the Mailing List.

REPEAT: Do NOT report bugs for outdated packages!
Tasklist

FS#38124 - [opensmtpd] crash on startup; needs _FORTIFY_SOURCE=1 instead

Attached to Project: Community Packages
Opened by Mantas Mikulėnas (grawity) - Friday, 13 December 2013, 09:59 GMT
Last edited by Sébastien Luttringer (seblu) - Monday, 16 December 2013, 18:32 GMT
Task Type Bug Report
Category Packages
Status Closed
Assigned To Sébastien Luttringer (seblu)
Architecture x86_64
Severity High
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 0
Private No

Details

community/opensmtpd 5.4.1p1 crashes almost every time when processing the offline queue on startup, on two different machines.

Developer has said that _FORTIFY_SOURCE=2 is causing this:

(__gilles) _FORTIFY_SOURCE=1 doesn't cause issues
(__gilles) i suggest you dont use FORTIFY_SOURCE=2 with opensmtpd, we had a
discussion with chl and fateyev from fedora about this recently
(__gilles) we used to build with FORTIFY_SOURCE=2, we had no issue with it, but i
suggested it was removed becaue the documentation stated it could lead to
incorrect code
(__gilles) and we can't really spend time troubleshooting this to figure out what
goes wrong when that option is enabled
(__gilles) https://github.com/poolpOrg/OpenSMTPD/issues/355
(__gilles) the only way i can think of to see if this is a legit issue is to start
analyzing the resulting .o and compare it with one without FORTIFY=2
(__gilles) this is very time consuming and since no one else experiences this crash
and it goes away with FORTIFY=1, I don't really feel the urge to spend
hours and hours on this :-)
(__gilles) to quote doc:
(__gilles) "With _FORTIFY_SOURCE set to 2 some more checking is added, but some
conforming programs might fail."

So the package should explicitly set CPPFLAGS="-D_FORTIFY_SOURCE=1" or something like that.
This task depends upon

Closed by  Sébastien Luttringer (seblu)
Monday, 16 December 2013, 18:32 GMT
Reason for closing:  Fixed
Comment by Mantas Mikulėnas (grawity) - Friday, 13 December 2013, 10:38 GMT
Update: Even _FORTIFY_SOURCE=1 is too unstable. Should probably be disabled completely.
Comment by Sébastien Luttringer (seblu) - Saturday, 14 December 2013, 15:58 GMT
I cannot confirm, my version doesn't crash (even when processing offline queue). What do you make think that is caused by FORTIFY?
Did you build a version without FORTIFY which fix your issue?

Can you show your config?

Comment by Mantas Mikulėnas (grawity) - Saturday, 14 December 2013, 16:26 GMT
> I cannot confirm, my version doesn't crash (even when processing offline queue).

And I confirmed this on two different systems. Both x86_64, however.

> What do you make think that is caused by FORTIFY?

Because a developer (__gilles on #opensmtpd) has said so, and because disabling FORTIFY gets rid of all crashes I had.

> Did you build a version without FORTIFY which fix your issue?

Yes.

> Can you show your config?

listen on ::
table aliases file:/etc/smtpd/aliases
accept from any for domain "nullroute.eu.org" relay backup
accept for local alias <aliases> deliver to mbox
accept for any relay
Comment by Mantas Mikulėnas (grawity) - Saturday, 14 December 2013, 16:28 GMT
Here's a log of one such crash.
Comment by Sébastien Luttringer (seblu) - Monday, 16 December 2013, 01:53 GMT
I read my IRC backlog of your discussion, what I should have done from the beginning.

Anyway, I pushed a new version in testing. Can you confirm that it fixes the problem for you.
Comment by Mantas Mikulėnas (grawity) - Monday, 16 December 2013, 18:24 GMT
Yes, it is fixed now.

Loading...