[dm-crypt] Cryptsetup (>= 1.7.0) luksOpen/luksClose gets stuck after open/close

Guido Trentalancia guido at trentalancia.net
Fri Apr 8 12:28:59 CEST 2016

Hello Milan,

I have checked again and I can now confirm that the problem happens
only when using LVM version

After upgrading to LVM version the problem disappears.

However, it might not depend on the udev rule, as such rule had not
been uninstalled when using LVM version Also, I still
cannot explain the reason why cryptsetup < 1.7.0 works fine even when
LVM2.2.02.148 and the mentioned udev rule are installed.



On Thu, 2016-04-07 at 21:24 +0200, Milan Broz wrote:
> On 04/07/2016 08:47 PM, Guido Trentalancia wrote:
> > 
> > Hello.
> > 
> > My first message contains the debug information (last part of the
> > strace output).
> No, it does not. Please post debug information and NOT strace as
> I described in the first reply...
> Just add --debug to cryptsetup command and you have it (without
> strace).
> It contains a lot of info that is needed to debug almost all kind of
> cryptsetup problem.
> You also missed request to post distro version here.
> Do you compile cryptsetup yourself?
> Anyway, I am almost sure you are missing the udev dmsetup
> udevcomplete
> call in udev, this is very common error.
> > 
> > If you need other information, please let me know. Obviously I
> > cannot
> > post gdb output as the program does not crash, it simply gets
> > stuck.
> Strace or gdb info is not needed if not requested, --debug log is
> usually
> enough to see what is going on.
> And did you try that "dmsetup udevcomplete_all" I already suggested?
> (once it is "stuck", run from another terminal - is should finish
> that command)
> > 
> > The two versions show a different behaviour (the working version
> > reads
> > SELinux configuration, the broken version opens /dev/urandom and
> > gets
> > stuck), so the code must have changed.
> I do not know about any change there.
> Running code depends on compilation parameters, some basic info how
> code
> behaves is in the debug log I am still waiting for...
> Milan

More information about the dm-crypt mailing list