PyPi is responsible for distributing malware

Installing packages from PyPi is a dangerous proposition. In July 2020, the request package (note the typo: request and not requests) was downloaded over 10,000 times. It was a malware that installed a daemon in .bashrc, providing the attacker with a remote shell on the machine. PyPi has always been vulnerable to typosquatting attacks but does not display any warning to their users. A system administrator with a healthy discipline with regard to security hosted the “request” malware during 20 days in August 2020 and discovered it by chance. How many machines are still compromised? Why is there no information on PyPi about this malware and all others? Even though there is every reason to suspect such attacks happen on a weekly basis and target hundreds of thousands of users, is is almost never discussed in public.

When contacted about a malicious package the python security team or the people responsible for PyPi only answer is that you can’t trust anything on PyPI. There however is no such warning on https://pypi.org and no documentation on how to verify that a package can be trusted. If PyPi did not already know about the malicious package, they will remove it. But there will be no security warning about which malicious package was published or when. For instance a dozen malicious packages were discovered in 2018 and removed from PyPi, but there is no way to know when they were published and for how long.

Consequently as a PyPi user who wants to avoid malware, I’m responsible:

  • to not make typos when running pip install somepackage, even for test purposes
  • to verify it does not contain a malware (there is no documentation on that topic)
  • to keep track of all packages installed for forensic analysis (because there is no public database of known malware distributed via PyPi)

This is a very heavy burden.

The Debian GNU/Linux distribution maintains a software repository which is similar to PyPi in the sense that it contains packages. With one major difference: it is efficiently protected against malware. All packages are signed by their authors and the signature can be verified with a web of trust when they are installed.

As a Debian GNU/Linux user, my responsibility to avoid malware is:

  • nothing

Ideally PyPi will provide a similar security feature in the future but there currently is no work in this direction (note that PIP 458 Secure PyPI downloads with signed repository metadata is being worked on but is not about package authors signing their packages). As a result downloading a package from PyPi is either very time consuming (because it needs to be carefully analyzed) or a security risk because it can contain a malware for which no information is ever going to be released on PyPi, even after it is discovered.

This is a large scale problem because PyPi is implicitly trusted by hundreds of thousands of users who are not informed about the risks they take. Over the past decade people working on PyPi and organizations financing their work repeatedly made a conscious choice to work on something other than protecting users from malware. It may be OK to do so occasionally but after such an extended period of time it becomes a pattern. PyPi is responsible for distributing malware and urgently needs to re-adjust its priorities to resolve the problem.

How the cancel culture was leveraged against RMS

Starting in August 2019, MIT was being put on the spot for accepting money from Jeffrey Epstein and accusations against the late Marvin Minsky for having sex with one of his victims resurfaced. In a call for protest, someone even wrote that Marvin Minsky assaulted her. In reaction Richard Stallman replied in an email that “the word “assaulting” presumes that [Marvin Minsky] applied force or violence, in some unspecified way, but the article itself says no such thing”. The rise of the “cancel culture” makes it very hazardous to publicly say or write anything regarding sexism, prostitution or racism. Despite this, Richard Stallman has published a number of blog entries on those topics over the years, ignoring people urging him to stop or become a target. But this time he caught the attention of someone.

After reading Richard Stallman email, Selam G. published an article and commented: “[Richard Stallman] says that an enslaved child could, somehow, be “entirely willing””. What Richard Stallman actually wrote is: “We can imagine many scenarios, but the most plausible scenario is that she presented herself to him as entirely willing. Assuming she was being coerced by Epstein, he would have had every reason to tell her to conceal that from most of his associates”. More tabloid articles followed, further mischaracterising Richard Stallman statements. A photo of Richard Stallman office door at MIT labeled “Richard Stallman: Knight for Justice (also: hot ladies)” was also included in Selam G.’s article as a proof against Richard Stallman.

In response, Richard Stallman commented on his blog: “headlines say that I defended Epstein. Nothing could be further from the truth. I’ve called him a “serial rapist”, and said he deserved to be imprisoned.” As for the label on Richard Stallman office door, it was added by an unknown person, over a decade ago. In a documentary shot in 2017/2018, it only reads “Richard Stallman: Knight for Justice” (at 6:22) because Richard Stallman removed the ending: (also: hot ladies). The content and the tone of Selam G.’s article is archetypal of the cancel culture and she was joined by people sharing the same faith to demand the resignation of Richard Stallman.

Any effort to establish the truth was a loss of time: Richard Stallman participating in a discussion about rape and prostitution was blasphemy. He was not attacked for what he wrote, he was attacked for writing on a forbidden topic. Another similar and well documented instance is what happened to Bret W. in 2017 at the Evergreen College when he wrote “one’s right to speak –or to be– must never be based on skin color” in an email. In both cases there were many witness who did not intervene for fear of being the next target. I also stayed silent and horrified while Richard Stallman was shamed publicly. I wanted to believe it would blow over after a few days. I trusted the Free Software community to be rational and eventually disregard this episode. I was sorely mistaken.

Two days after Selam G. article was published, Software Freedom Conservancy who is run by some of the most influential people in the Free Software movement, published a statement demanding Richard Stallman’s resignation from the FSF. Neil McGovern, executive director of GNOME, also asked for Richard Stallman resignation and links to a tabloid article grossly mischaracterising Richard Stallman. They endorsed the attacks and legitimized them in the eye of the Free Software community. The next day Richard Stallman resigned from his position as president of the FSF.

It all happened very quickly but the roots of Richard Stallman’s dispute with other members of the Free Software community are deep. Software Freedom Conservancy states that “when considered with other reprehensible comments [Richard Stallman] has published over the years, these incidents form a pattern of behavior that is incompatible with the goals of the free software movement” and Neil McGovern wrote something similar. Although they did not reference any specific event, there are some publicly available records of past disputes. For instance, in September 2018, Richard Stallman wrote a blog entry very similar to the defense of Marvin Minsky. He rewrote it in the following weeks because, in his own words: “some people reading earlier versions of this note seem to have got the idea that I condone enslavement of prostitutes”.

I can only assume the respected members of the Free Software community who leveraged the cancel culture and tabloid journalism felt justified because of years of frustrating disputes with Richard Stallman. But they made a terrible mistake. Richard Stallman was unjustly and severely punished. He no longer has a position at the FSF, which he founded 34 years ago, he also lost his position at MIT, his scheduled talks could be canceled and he lost his home.

History will remember that in 2019 Richard Stallman took a major hit from the cancel culture and tabloid journalists. And also that respected members of the Free Software community endorsed the cancel culture and tabloid journalists. They sent an ominous message to all their opponents: we fight dirty. They negated the benefit of dialog, any sense of rational thinking and threw away years of debate with Richard Stallman. Leveraging someone who had no clue about the past disputes to get rid of their opponent proved to be much more efficient. We should now expect that some Free Software activists will use the cancel culture as a mean to an end. But what they should do instead is to ignore the furor, wait until it passes and resume a discussion based on truth and logic. It takes longer but it is the only way to build a safe and inclusive community.

recovering from lost partition table: a use case

This happened today and may be of use to other Ubuntu or Debian users with a similar configuration. The disk of a laptop lost its partition table because it was mistaken for a USB key and modified. There was no way to recover the old GPT partition table from the backup because it was not corrupted. The new partition table was both in the GPT partition table and its backup.

The laptop was installed with Ubuntu 16.04 about two years ago and it was assumed to:

  • Have an EFI partition
  • Have a separate boot partition

The partition table is rebuilt to start with a 512MB partition for EFI (type 1) followed by another parition for the rest of the disk. When the second partition table is created using fdisk, it will issue a warning about an ext2 signature: it must be preserved.

Using dumpe2fs on the second partition the actual size of the partition is displayed, for instance:

$ dumpe2fs /dev/nvme0n1p2
...
Block count:              249856
...
Block size:               1024

The partition table is adjusted accordingly, taking into account that filesystem blocks are 1024 bytes and fdisk sectors are 512 bytes (i.e. 249856 * 2 == 499712):

$ fdisk -l /dev/nvme0n1
Device           Start        End   Sectors   Size Type
/dev/nvme0n1p1    2048    1050623   1048576   512M EFI System
/dev/nvme0n1p2 1050624    1550335    499712   244M Linux filesystem

Note: for some reason fdisk adds 1 to the specified number of sectors in some cases. Care must be taken to verify the actual number of sectors allocated to the partition and adjust them accordingly.

The signature of the third partition gives a hint regarding its content. If it is crypto_LUKS, it can be open with cryptsetup luksOpen /dev/nvme0n1p3 root and the device will show under /dev/mapper.

Why I deleted my GitHub account

Today I permanently deleted the GitHub account that was in my name. I used it to contribute to Free Software projects that only use this proprietary platform. I started using GitHub a few years ago and constantly tried (and failed) to convince projects to self-host using Free Software instead. As more and more Free Software projects I care about moved to GitHub, I ended up using it daily and it made me sad.

The recent acquisition of GitHub by Microsoft turned it into a service that belongs to a company that is actively working against the Free Software movement by:

  • Lobbying for Software Patents and being one of the largest Software Patents holder in the world
  • Lobbying for DRMs and implementing them in their software products
  • Convincing people they must not fight in court when their rights under copyleft licenses are violated

By using Microsoft services daily, I would also implicitly support them, even if indirectly. I feel better knowing I no longer depend on Microsoft in any way.

Running tests on a Nextcloud app

The sample Nextcloud app is generated with restrictions that won’t allow it to run on version 13 or above. This can be fixed by updating apps/foobar/appinfo/info.xml with:

<nextcloud min-version="12" max-version="14"/>

The Docker test container can be modified to run on /opt/nextcloud/server, a clone of the Nextcloud server including the generated app.

version: '3'

services:
  db:
    image: postgres
    restart: always
    volumes:
      - db:/var/lib/postgresql/data
    env_file:
      - db.env

  app:  
    image: nextcloud:fpm
    restart: always
    volumes:
      - /opt/nextcloud/server:/var/www/html
    environment:
      - POSTGRES_HOST=db
    env_file:
      - db.env
    depends_on:
      - db

  web:
    build: ./web
    restart: always
    ports:
      - 8080:80
    volumes:
      - /opt/nextcloud/server:/var/www/html:ro
    depends_on:
      - app

volumes:
  db:
#  nextcloud:

The phpunit-5.7 script is then installed with:

  • cd /usr/local/bin
  • wget -O phpunit https://phar.phpunit.de/phpunit-5.phar
  • chmod +x phpunit

and run with

  • cd /var/www/html
  • chown -R www-data .
  • cd apps/foobar
  • phpunit
PHPUnit 5.7.27 by Sebastian Bergmann and contributors.
.                                                                   1 / 1 (100%)
Time: 241 ms, Memory: 8.00MB
OK (1 test, 2 assertions)
  • phpunit –debug -c phpunit.integration.xml
PHPUnit 5.7.27 by Sebastian Bergmann and contributors.
Starting test 'OCA\FooBar\Tests\Integration\Controller\AppTest::testAppInstalled'.
.                                                                   1 / 1 (100%)
Time: 250 ms, Memory: 8.00MB
OK (1 test, 1 assertion)

From a SecureDrop talk to the first edition of the privacy devroom at FOSDEM


Only this year did I fully understand what motivates so many people to devote their free time to making FOSDEM a success, year after year. As surprising as it sounds, organizing the privacy devroom and standing at the booth was better than just visiting, chatting and attending talks. A big part of that good feeling certainly comes from the excellent organization. Accidents happened, of course, but they were resolved quickly. And ancient Free Software was used with clunky interfaces but nothing got in the way and people used to it help the newcomers. As a Free Software developer, this sets a fine example of the feeling we would like to provide to every contributor of our code base. When participating is more enjoyable than passively using, we’re one step closer to contributopia.

Continue reading “From a SecureDrop talk to the first edition of the privacy devroom at FOSDEM”

HOWTO Anonymous mobile in Paris

Using a mobile anonymously with encrypted messages and voice is challenging. With Signal text and voice are encrypted but it sends your contacts to Signal which makes me uncomfortable. With Orfox you can browse the web without revealing your IP address but the GSM module tracks your location.

With a small budget you can however buy a second hand mobile and dedicate it to anonymous communications, as long as you remove its battery when you’re not in a public place. You do not leak your contacts to Signal and the GSM module tracks a SIM card that is not associated with you. Here is a detailed description of the preparation of an anonymous phone I did today. This is not rocket science and I’m sure lots of people already know all of that. But I did not find a HOWTO and it took me some time to figure it out.

Continue reading “HOWTO Anonymous mobile in Paris”

gnome3 / libnotify notification for org-mode appointments

Org mode appointments

can be notified 12 minutes before with libnotify

by adding the following to the .emacs:

; Desktop notifications
(setq alert-default-style 'libnotify)
(setq appt-disp-window-function (lambda (min-to-app new-time appt-msg)
                                                         (alert appt-msg)))
(setq appt-delete-window-function (lambda ()))
; Rebuild the reminders everytime the agenda is displayed
(add-hook 'org-agenda-finalize-hook (lambda () (org-agenda-to-appt t)))
; Run once when Emacs starts
(org-agenda-to-appt t)
; Activate appointments so we get notifications
(appt-activate t)

Continue reading “gnome3 / libnotify notification for org-mode appointments”

HOWTO nginx & letsencrypt on Debian GNU/Linux stretch/9

The goal is to configure a nginx server with automatic Let’s Encrypt renewal, assuming a new dedicated virtual machine running a pristine Debian GNU/Linux stretch/9.

Install docker-compose:

sudo apt-get update
sudo apt-get install -y apt-transport-https ca-certificates dirmngr
sudo apt-key adv --keyserver hkp://p80.pool.sks-keyservers.net:80 --recv-keys 58118E89F3A912897C070ADBF76221572C52609D
echo deb https://apt.dockerproject.org/repo debian-stretch main | sudo tee /etc/apt/sources.list.d/docker.list
sudo apt-get update
sudo apt-get install -y docker-engine
sudo bash -c 'curl -L https://github.com/docker/compose/releases/download/1.13.0/docker-compose-`uname -s`-`uname -m` > /usr/local/bin/docker-compose'
sudo chmod +x /usr/local/bin/docker-compose

Assuming the FQDN of the machine is download.securedrop.club and the person responsible can be reached at admin@securedrop.cub, create the docker-compose.yml with:

cat > docker-compose.yml <<EOF
version: '2'
services:
  web:
    image: nginx:1.13.3
    volumes:
      - ./html:/usr/share/nginx/html:ro
    ports:
      - "8080:80"
    environment:
      - VIRTUAL_HOST=download.securedrop.club
      - LETSENCRYPT_HOST=download.securedrop.club
      - LETSENCRYPT_EMAIL=admin@securedrop.club
  proxy:
    image: jwilder/nginx-proxy
    volumes:
      - /var/run/docker.sock:/tmp/docker.sock:ro
      - ./certs:/etc/nginx/certs:ro
      - /etc/nginx/vhost.d
      - /usr/share/nginx/html
    ports:
      - "80:80"
      - "443:443"
    restart: always
    depends_on:
      - web
  letsencrypt:
    image: jrcs/letsencrypt-nginx-proxy-companion
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock:ro
      - ./certs:/etc/nginx/certs:rw
    volumes_from:
      - proxy
EOF

and run docker-compose up in the same directory as the docker-compose.yml file.