[dm-crypt] [PATCH] When reading no single byte for the key abort.

[dm-crypt] [PATCH] When reading no single byte for the key abort.

[Ludwig Nussel]

Allows keyscripts that pipe into cryptsetup to quit due to e.g. a
timeout or user pressing ^D.

Signed-off-by: Ludwig Nussel <ludwig.nussel@suse.de>
---
 lib/utils.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/lib/utils.c b/lib/utils.c
index 05d33cd..25f7714 100644
--- a/lib/utils.c
+++ b/lib/utils.c
@@ -490,6 +490,8 @@ void get_key(char *prompt, char **key, unsigned int *passLen, int key_size,
 		}
 		if(key_file)
 			close(fd);
+		if(!i) // we didn't read anything, user pressed ^D?
+			goto out_err;
 		pass[i] = 0;
 		*key = pass;
 		*passLen = i;
-- 
1.6.2.1

Re: [dm-crypt] [PATCH] When reading no single byte for the key abort.

[Milan Broz]

Ludwig Nussel wrote:
> Allows keyscripts that pipe into cryptsetup to quit due to e.g. a
> timeout or user pressing ^D.

> @@ -490,6 +490,8 @@ void get_key(char *prompt, char **key, unsigned int *passLen, int key_size,
>  		}
>  		if(key_file)
>  			close(fd);
> +		if(!i) // we didn't read anything, user pressed ^D?
> +			goto out_err;

there are two situations which with this patch fails now:

if the key-file is empty (but regular) file, it should not fail here, I mean:
	touch empty
	cryptsetup luksOpen ... --key-file=empty

You probably mean to fix only this situation (it should fail if piped input is closed
and not try empty passphrase)
	$(sleep 1) | cryptsetup luksOpen ...

I committed modified patch (see svn), please verify that it fixes reported problem.

Thanks,
Milan
--
mbroz@redhat.com

Re: [dm-crypt] [PATCH] When reading no single byte for the key abort.

[Ludwig Nussel]

Milan Broz wrote:
> Ludwig Nussel wrote:
> > Allows keyscripts that pipe into cryptsetup to quit due to e.g. a
> > timeout or user pressing ^D.
> 
> > @@ -490,6 +490,8 @@ void get_key(char *prompt, char **key, unsigned int *passLen, int key_size,
> >  		}
> >  		if(key_file)
> >  			close(fd);
> > +		if(!i) // we didn't read anything, user pressed ^D?
> > +			goto out_err;
> 
> there are two situations which with this patch fails now:
> 
> if the key-file is empty (but regular) file, it should not fail here, I mean:
> 	touch empty
> 	cryptsetup luksOpen ... --key-file=empty

Hmm, does an empty key file have any use in practice?

cu
Ludwig

-- 
 (o_   Ludwig Nussel
 //\   
 V_/_  http://www.suse.de/
SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nuernberg)

Re: [dm-crypt] [PATCH] When reading no single byte for the key abort.

[anton ivanov]

Hey,

what are you talking about? I tried with empty file as passphrase,
surely not working :)

On Mon, Sep 28, 2009 at 9:51 PM, Milan Broz <mbroz@redhat.com> wrote:
> Ludwig Nussel wrote:
>> Allows keyscripts that pipe into cryptsetup to quit due to e.g. a
>> timeout or user pressing ^D.
>
>> @@ -490,6 +490,8 @@ void get_key(char *prompt, char **key, unsigned int *passLen, int key_size,
>>               }
>>               if(key_file)
>>                       close(fd);
>> +             if(!i) // we didn't read anything, user pressed ^D?
>> +                     goto out_err;
>
> there are two situations which with this patch fails now:
>
> if the key-file is empty (but regular) file, it should not fail here, I mean:
>        touch empty
>        cryptsetup luksOpen ... --key-file=empty
>
> You probably mean to fix only this situation (it should fail if piped input is closed
> and not try empty passphrase)
>        $(sleep 1) | cryptsetup luksOpen ...
>
> I committed modified patch (see svn), please verify that it fixes reported problem.
>
> Thanks,
> Milan
> --
> mbroz@redhat.com
>
> _______________________________________________
> dm-crypt mailing list
> dm-crypt@saout.de
> http://www.saout.de/mailman/listinfo/dm-crypt
>



-- 
ai.

http://biwwy.com/
last.fm: littlewizard

Re: [dm-crypt] [PATCH] When reading no single byte for the key abort.

[anton ivanov]

Oh, sorry, not get used for mailing list :)

On Tue, Sep 29, 2009 at 1:03 PM, anton ivanov
<run.into.flowers@gmail.com> wrote:
> Hey,
>
> what are you talking about? I tried with empty file as passphrase,
> surely not working :)
>
> On Mon, Sep 28, 2009 at 9:51 PM, Milan Broz <mbroz@redhat.com> wrote:
>> Ludwig Nussel wrote:
>>> Allows keyscripts that pipe into cryptsetup to quit due to e.g. a
>>> timeout or user pressing ^D.
>>
>>> @@ -490,6 +490,8 @@ void get_key(char *prompt, char **key, unsigned int *passLen, int key_size,
>>>               }
>>>               if(key_file)
>>>                       close(fd);
>>> +             if(!i) // we didn't read anything, user pressed ^D?
>>> +                     goto out_err;
>>
>> there are two situations which with this patch fails now:
>>
>> if the key-file is empty (but regular) file, it should not fail here, I mean:
>>        touch empty
>>        cryptsetup luksOpen ... --key-file=empty
>>
>> You probably mean to fix only this situation (it should fail if piped input is closed
>> and not try empty passphrase)
>>        $(sleep 1) | cryptsetup luksOpen ...
>>
>> I committed modified patch (see svn), please verify that it fixes reported problem.
>>
>> Thanks,
>> Milan
>> --
>> mbroz@redhat.com
>>
>> _______________________________________________
>> dm-crypt mailing list
>> dm-crypt@saout.de
>> http://www.saout.de/mailman/listinfo/dm-crypt
>>
>
>
>
> --
> ai.
>
> http://biwwy.com/
> last.fm: littlewizard
>



-- 
ai.

http://biwwy.com/
last.fm: littlewizard

Re: [dm-crypt] [PATCH] When reading no single byte for the key abort.

[Milan Broz]

anton ivanov wrote:
> what are you talking about? I tried with empty file as passphrase,
> surely not working :)

I didn't say that empty keyfile can be used as real passphrase (in luksFormat),
but the verification (luksOpen) will fail with "No key available with this passphrase"
and with the patch it fails with different error - error during reading key.

I can imagine that some scripts can depend on this behaviour and
there is no real need to change it.
When you enter empty string from terminal, it does the same.

Milan