Bug#928299: installation-reports: network-console doesnt set password hash, ssh not possible

[Frank Gard]

Dear Cyril,

Am 02.05.19 um 10:32 schrieb Cyril Brulebois:
> Control: tag -1 patch pending

that sounds good…

> Control: reassign -1 network-console
>
> […]
>
> It seems crypt() is no longer declared as it used to, and we get
> implicit function declaration warnings from gcc, which later explains
> the segfault at runtime.
>
> The declaration is guarded by __USE_MISC but #define'ing it doesn't seem
> to be sufficient. I've decided to add a  include, which seems
> to do the trick: gen-crypt no longer segfaults, and seems to return
> suitable results on my amd64 machine:
>
> kibi@armor:~/debian-installer/packages/network-console$ ./gen-crypt foo
> $1$g.dM$EJhR11EsVD5jY1Smnomsc0A
>
> Source uploaded to the archive (1.81), then it'll get picked up by the
> buildds[1] and later made available in daily builds. It'd be nice if you
> could double check that the runtime within d-i is also fine.
>
>  1. https://buildd.debian.org/status/package.php?p=network-console&suite=sid
>
> Thanks again for the report & follow-ups (und vielen Dank zu Marc for
> the initial forward from the user list).

Many, many thanks to all involved people for your immediate help. I'm looking 
very forward to one of the next of Buster's weekly-builds :) .

> Cheers,

Have a nice day,
Frank.

eXirius IT Dienstleistungen GmbH
Juchem-Straße 24 | 66571 Eppelborn
T +49 6881 5 0 | i...@exirius.de
http://www.exirius.de

Amtsgericht Saarbrücken HRB 12124
Geschäftsführer: Dipl.-Math. oec. Michael Royar

Bug#928299: installation-reports: network-console doesnt set password hash, ssh not possible

[Cyril Brulebois]

Control: tag -1 patch pending
Control: reassign -1 network-console

Hi,

Frank Gard  (2019-05-02):
> Am 02.05.19 um 09:43 schrieb Cyril Brulebois:
> > Hi again,
> >
> > […]
> > OK, thanks. That confirms at least the echo instruction I quoted is
> > executed, but gen-crypt didn't do its job…
> > […]
> > … which is understandable given the segfault.
> > […]
> >
> > Please mention at the very least your architecture and the target
> > device, in case the segfault is hardware/architecture-dependent. This
> > could be some trivial bug, but let's see what you're running d-i on.
> 
> Please find attached the complete output of dmesg. I hope there's all
> information you need. If not, please let me know.

That was enough, thanks. I tend to think about arm* devices whenever
network-console is involved; seeing that affect amd64 means I could try
and run some commands locally without having to deal with arch-specific
environments.

It seems crypt() is no longer declared as it used to, and we get
implicit function declaration warnings from gcc, which later explains
the segfault at runtime.

The declaration is guarded by __USE_MISC but #define'ing it doesn't seem
to be sufficient. I've decided to add a  include, which seems
to do the trick: gen-crypt no longer segfaults, and seems to return
suitable results on my amd64 machine:

kibi@armor:~/debian-installer/packages/network-console$ ./gen-crypt foo
$1$g.dM$EJhR11EsVD5jY1Smnomsc0A

Source uploaded to the archive (1.81), then it'll get picked up by the
buildds[1] and later made available in daily builds. It'd be nice if you
could double check that the runtime within d-i is also fine.

 1. https://buildd.debian.org/status/package.php?p=network-console&suite=sid

Thanks again for the report & follow-ups (und vielen Dank zu Marc for
the initial forward from the user list).


Cheers,
-- 
Cyril Brulebois (k...@debian.org)
D-I release manager -- Release team member -- Freelance Consultant


signature.asc
Description: PGP signature

Bug#928299: installation-reports: network-console doesnt set password hash, ssh not possible

[Frank Gard]

Hi

Am 02.05.19 um 09:43 schrieb Cyril Brulebois:
> Hi again,
>
> […]
> OK, thanks. That confirms at least the echo instruction I quoted is
> executed, but gen-crypt didn't do its job…
> […]
> … which is understandable given the segfault.
> […]
>
> Please mention at the very least your architecture and the target
> device, in case the segfault is hardware/architecture-dependent. This
> could be some trivial bug, but let's see what you're running d-i on.
Please find attached the complete output of dmesg. I hope there's all 
information you need. If not, please let me know.
>
> Cheers,
Thanks and cheers.

eXirius IT Dienstleistungen GmbH

Juchem-Straße 24 | 66571 Eppelborn

T +49 6881 5 0 | i...@exirius.de

http://www.exirius.de



Amtsgericht Saarbrücken HRB 12124

Geschäftsführer: Dipl.-Math. oec. Michael Royar



dmesg.txt.bz2
Description: application/bzip

Bug#928299: installation-reports: network-console doesnt set password hash, ssh not possible

[Cyril Brulebois]

Hi again,

Frank Gard  (2019-05-02):
> On Thu, 2 May 2019 08:55:54 +0200 Cyril Brulebois  wrote:
> > OK. What does /etc/shadow look like?
> 
> Its content is:
> 
> installer::1:0:9:7:::

OK, thanks. That confirms at least the echo instruction I quoted is
executed, but gen-crypt didn't do its job…

> > […]
> > echo "installer:$(gen-crypt $PASSWORD):1:0:9:7:::" >> /etc/shadow
> >
> > with gen-crypt reading some tiny amount of data from /dev/urandom. I'd
> > be happy to see a syslog of an affected system (please compress it when
> > attaching it to the bug report).
> >
> 
> I did a "dmesg", and the two very last lines are quite interesting, I think:
> […]
> [    1.069070] sd 2:0:0:0: [sda] Attached SCSI disk
> [    1.583116] random: crng init done
> [   15.103655] usbcore: registered new interface driver usb-storage
> [   16.247345] ISO 9660 Extensions: Microsoft Joliet Level 3
> [   16.248981] ISO 9660 Extensions: RRIP_1991A
> [   55.531147] virtio_net virtio3 ens18: renamed from eth0
> [  102.052912] gen-crypt[3815]: segfault at 88e3d260 ip 
> 7f36154b2206 sp 7ffcd3b4bbe8 error 5 in libc.so.6[7f361543c000+148000]
> [  102.052917] Code: 0f 1f 40 00 66 0f ef c0 66 0f ef c9 66 0f ef d2 66 0f ef 
> db 48 89 f8 48 89 f9 48 81 e1 ff 0f 00 00 48 81 f9 cf 0f 00 00 77 6a  0f 
> 6f 20 66 0f 74 e0
> 66 0f d7 d4 85 d2 74 04 0f bc c2 c3 48 83

… which is understandable given the segfault.

> Let me know, if you need the full log or anything else.

Please mention at the very least your architecture and the target
device, in case the segfault is hardware/architecture-dependent. This
could be some trivial bug, but let's see what you're running d-i on.


Cheers,
-- 
Cyril Brulebois (k...@debian.org)
D-I release manager -- Release team member -- Freelance Consultant


signature.asc
Description: PGP signature

Bug#928299: installation-reports: network-console doesnt set password hash, ssh not possible

[Cyril Brulebois]

Frank Gard  (2019-05-02):
> Hi all,

If you only mail n...@bugs.debian.org, only the maintainers will see you
reply, please use reply-all.

> I'm terribly sorry, but no. All I can say is that there are no
> problems at all when using stable ("stretch").

OK, thanks anyway.

> But it was the very first time I tried a buster installation (shame on
> me), so I was really "surprised" that there is such a "basic" (IMHO)
> problem there.

There are many features that people find “basic”, which is probably true
when that's a recurrent usecase for you. The thing is: we cannot test
all installation scenarios on all hardware in the world.

> Hopefully somebody else can tell you more details about the point in
> time since when the bug exists

This is not a strict requirement. That can be some useful clue though,
hence my asking that upfront.


Cheers,
-- 
Cyril Brulebois (k...@debian.org)
D-I release manager -- Release team member -- Freelance Consultant


signature.asc
Description: PGP signature

Bug#928299: installation-reports: network-console doesnt set password hash, ssh not possible

[Frank Gard]

Dear Cyril,

On Thu, 2 May 2019 08:55:54 +0200 Cyril Brulebois  wrote:
> […]
>
> OK. What does /etc/shadow look like?

Its content is:

installer::1:0:9:7:::

> […]
> echo "installer:$(gen-crypt $PASSWORD):1:0:9:7:::" >> /etc/shadow
>
> with gen-crypt reading some tiny amount of data from /dev/urandom. I'd
> be happy to see a syslog of an affected system (please compress it when
> attaching it to the bug report).
>

I did a "dmesg", and the two very last lines are quite interesting, I think:
[…]
[    1.069070] sd 2:0:0:0: [sda] Attached SCSI disk
[    1.583116] random: crng init done
[   15.103655] usbcore: registered new interface driver usb-storage
[   16.247345] ISO 9660 Extensions: Microsoft Joliet Level 3
[   16.248981] ISO 9660 Extensions: RRIP_1991A
[   55.531147] virtio_net virtio3 ens18: renamed from eth0
[  102.052912] gen-crypt[3815]: segfault at 88e3d260 ip 
7f36154b2206 sp 7ffcd3b4bbe8 error 5 in libc.so.6[7f361543c000+148000]
[  102.052917] Code: 0f 1f 40 00 66 0f ef c0 66 0f ef c9 66 0f ef d2 66 0f ef 
db 48 89 f8 48 89 f9 48 81 e1 ff 0f 00 00 48 81 f9 cf 0f 00 00 77 6a  0f 6f 
20 66 0f 74 e0
66 0f d7 d4 85 d2 74 04 0f bc c2 c3 48 83

Let me know, if you need the full log or anything else.

Cheers.

>
> Cheers,
> --
> Cyril Brulebois (k...@debian.org) 
> D-I release manager -- Release team member -- Freelance Consultant


eXirius IT Dienstleistungen GmbH
Juchem-Straße 24 | 66571 Eppelborn
T +49 6881 5 0 | i...@exirius.de
http://www.exirius.de

Amtsgericht Saarbrücken HRB 12124
Geschäftsführer: Dipl.-Math. oec. Michael Royar

Bug#928299: installation-reports: network-console doesnt set password hash, ssh not possible

[Frank Gard]

Hi all,

On Thu, 2 May 2019 08:07:46 +0200 Marc Haber  
wrote:
> On Thu, May 02, 2019 at 08:00:52AM +0200, Cyril Brulebois wrote:
> > […]
> > Any chance you'd know when it last worked for you on sid or buster?

I'm terribly sorry, but no. All I can say is that there are no problems at all 
when using stable ("stretch"). But it was the very first time I tried a buster 
installation
(shame on me), so I was really "surprised" that there is such a "basic" (IMHO) 
problem there.

Hopefully somebody else can tell you more details about the point in time since 
when the bug exists, and –of course– that your're able to fix it quite soon.

> […]
>
> Greetings
> Marc

Many thanks and greetings, too
Frank.

> […]


eXirius IT Dienstleistungen GmbH
Juchem-Straße 24 | 66571 Eppelborn
T +49 6881 5 0 | i...@exirius.de
http://www.exirius.de

Amtsgericht Saarbrücken HRB 12124
Geschäftsführer: Dipl.-Math. oec. Michael Royar

Bug#928299: installation-reports: network-console doesnt set password hash, ssh not possible

[Cyril Brulebois]

(Adding Frank; please use reply-all on the Debian BTS.)

Marc Haber  (2019-05-02):
> On Thu, May 02, 2019 at 08:00:52AM +0200, Cyril Brulebois wrote:
> > Marc Haber  (2019-05-01):
> > > at least today's daily installer snapshot has a totally broken
> > > network-console. It asks for a password and confirmation and then
> > > prompts to continue isntallation with ssh installer@.
> > > 
> > > However, no password hash is written to /etc/shadow, hence login into
> > > the installer is not possible. After changing to a different virtual
> > > console of the installing system and putting a valid password hash into
> > > /etc/shadow allowed installation to continue.
> > 
> > Any chance you'd know when it last worked for you on sid or buster?
> 
> Years ago. I seldomly use the installer myself, I was just investigating
> a report sent by a user to debian-user-german. I have asked the original
> reporter.

OK. What does /etc/shadow look like?

network-console does that in its postinst script:

echo "installer:$(gen-crypt $PASSWORD):1:0:9:7:::" >> /etc/shadow

with gen-crypt reading some tiny amount of data from /dev/urandom. I'd
be happy to see a syslog of an affected system (please compress it when
attaching it to the bug report).


Cheers,
-- 
Cyril Brulebois (k...@debian.org)
D-I release manager -- Release team member -- Freelance Consultant


signature.asc
Description: PGP signature

Bug#928299: installation-reports: network-console doesnt set password hash, ssh not possible

[Marc Haber]

On Thu, May 02, 2019 at 08:00:52AM +0200, Cyril Brulebois wrote:
> Marc Haber  (2019-05-01):
> > at least today's daily installer snapshot has a totally broken
> > network-console. It asks for a password and confirmation and then
> > prompts to continue isntallation with ssh installer@.
> > 
> > However, no password hash is written to /etc/shadow, hence login into
> > the installer is not possible. After changing to a different virtual
> > console of the installing system and putting a valid password hash into
> > /etc/shadow allowed installation to continue.
> 
> Any chance you'd know when it last worked for you on sid or buster?

Years ago. I seldomly use the installer myself, I was just investigating
a report sent by a user to debian-user-german. I have asked the original
reporter.

Greetings
Marc

-- 
-
Marc Haber | "I don't trust Computers. They | Mailadresse im Header
Leimen, Germany|  lose things."Winona Ryder | Fon: *49 6224 1600402
Nordisch by Nature |  How to make an American Quilt | Fax: *49 6224 1600421

Bug#928299: installation-reports: network-console doesnt set password hash, ssh not possible

[Cyril Brulebois]

Control: tag -1 - d-i

Hi,

Marc Haber  (2019-05-01):
> at least today's daily installer snapshot has a totally broken
> network-console. It asks for a password and confirmation and then
> prompts to continue isntallation with ssh installer@.
> 
> However, no password hash is written to /etc/shadow, hence login into
> the installer is not possible. After changing to a different virtual
> console of the installing system and putting a valid password hash into
> /etc/shadow allowed installation to continue.

Any chance you'd know when it last worked for you on sid or buster?


Cheers,
-- 
Cyril Brulebois (k...@debian.org)
D-I release manager -- Release team member -- Freelance Consultant


signature.asc
Description: PGP signature

Bug#928299: installation-reports: network-console doesnt set password hash, ssh not possible

[Marc Haber]

Package: installation-reports
Severity: important
Tags: d-i

Dear Maintainer,

at least today's daily installer snapshot has a totally broken
network-console. It asks for a password and confirmation and then
prompts to continue isntallation with ssh installer@.

However, no password hash is written to /etc/shadow, hence login into
the installer is not possible. After changing to a different virtual
console of the installing system and putting a valid password hash into
/etc/shadow allowed installation to continue.

Greetings
Marc