Cover photo
Theodore Ts'o
Works at Google
Attended MIT
Lives in Medford, MA
12,227 followers|2,071,984 views
Have him in circles
12,227 people
Ricardo Quaresma Silva Andrade's profile photo
Manoj Srivastava's profile photo
Johannes Holzhäuer's profile photo
James Angelo's profile photo
Illarion Bykov's profile photo
Franz Ambros's profile photo
Michael Kleber's profile photo
Wade Chatam's profile photo
KRISHNA SHAH's profile photo
Linux Kernel Hacker
  • Google
    Staff Engineer, 2010 - present
  • IBM
    Senior Technical Staff Member, 2001 - 2010
  • VA Linux Systems
    Principal Engineer, 1999 - 2001
  • MIT
    Senior Systems Programmer, 1990 - 1999
Map of the places this user has livedMap of the places this user has livedMap of the places this user has lived
Medford, MA
Chicago, IL - Long Island, NY
I am the first North American Linux Kernel Developer, having started working with Linux in September, 1991.  I have also served as the tech lead for the MIT Kerberos V5 development team, and was the architect at IBM in charge of bringing bring real-time Linux in support of real-time Java to the US Navy.   I previously served as CTO for the Linux Foundation, and am currently employed at Google.
  • MIT
    Computer Science, 1986 - 1990
Basic Information


Theodore Ts'o

Shared publicly  - 
Not only is Taylor Wilson not brown, and not Muslim, he doesn't go to both high school in Texas and his teachers aren't idiots. #IStandWithAhmed
Homeland Security and the Department of Energy got wind of what Taylor was up to and invited him to Washington so he could submit a grant proposal to develop the weapons detector.

But if you're 14 and build a clock while brown and Muslim you can expect to get arrested!
In his quest to better the world,Taylor Wilson captured the interest of Homeland Security and ended up with radioactive pants.
View original post
Albino B Neto (bino)'s profile photoKen Koster's profile photoEric Lacombe's profile photoJosé Almeida's profile photo

Theodore Ts'o

Shared publicly  - 
+AT&T: The same company that cooperated with the NSA is now playing man-in-the-middle and manipulating your http connections to inject advertisements.
John Randolph's profile photoErnesto Crespo's profile photoThomas Brandtstaetter's profile photoDavid Ford (FirefighterBlu3)'s profile photo
"I think that's an overly paranoid perspective" -- I doubt they would have been obvious about how they did it, but we know from the Snowden releases that they deliberately sabotaged IETF work in the past. You do acknowledge that, yes?
Add a comment...

Theodore Ts'o

Shared publicly  - 
Dear lazy G+ --- has anyone managed to compile coreutils for aarch64-android (i..e, arm64 for Android's bionic libc)?    I'm aware of Sonelli's docker repo[1], but it only supports 32-bit arm, and it's not at all obvious how the patches in that repo work around the truly awe-inspiring abstraction violations committed by coreutils into the C library's stdio implementation (among other non-portable hacks perpetrated by the coreutils sources).

If you have, patches and instructions for how to set up the cross compilation environment (preferably using the Android NDK, but I'm not picky) would be gratefully appreciated.


EDIT: Wasn't as bad as I feared.   Using coreutils 8.24 fixed most of the problems, and the rest weren't that nasty to work around, mercifully.
android-coreutils - A docker image that builds the GNU coreutils binaries for Android
Theodore Ts'o's profile photo
Wasn't as bad as I feared.   Using coreutils 8.24 fixed most of the problems, and the rest weren't that nasty to work around, mercifully.
Add a comment...

Theodore Ts'o

Shared publicly  - 
Very interesting... some vegatables (including organic veggies, so paying $$$ extra may not help; in fact, it might make it worse) may be using fertilizer which has a lot of heavy metals.    
A molecular biologist is finding what could be dangerous levels of heavy metals in plants like kale. And they've shown up the most in organic varieties.
Matt Proud's profile photoCaleb Cushing's profile photoARCHANGELUS LEGION ACTIVISTS's profile photoHari Krishna Dara's profile photo
The main problem within this article is it should have been splitted/divided to 2 or 3 articles.

Measuring heavy metals ( +Colin McMillen​ ) does not need double blind studies. But the duplicate containers need to be used and the other one filled with destilled water etc. so that possible contamination is eliminated.

Chelation threapies are something which might work or then not, or are just placebo. BUT spending $$$s and feeling good is therapy too if it helps, or spending $$$s to alcohol if it helps. But even mentioning this should have been in a separate article.

Eating too much cruciferous is not healthy eating because those are poisonous. So constantly eating salads (plant) or making kale smoothies is not healthy. This has been known issue for decades.
These plants are fighting against insects etc. and do generate all kinds of poisonous substaces which can easily be detected in original wild plants.

Once a week cruciferous plants. There are also other edible vegetables ... nuts .. mushrooms ...

(a lacto-vegetarian since 1974)
Add a comment...

Theodore Ts'o

Shared publicly  - 
So it looks like Chrome beta finally has High DPI support for Linux.   Excellent!

(I'm currently running 43.0.2357.18 but I see 43.0.2357.37 has been downloaded and I should get that as soon as I restart my browser.)
Theodore Ts'o's profile photoDaniel Yount's profile photoGreg S's profile photoARCHANGELUS LEGION ACTIVISTS's profile photo
Greg S
So it seems you have to manually specify the dimensions for the things like the tabs and the height of the navigation bar and menus. So I did:

$ xrandr --dpi 144
$ google-chrome-beta  --force-device-scale-factor=1.5 &

And I get a consistent rendering that's a reasonable size. I'm not sure why the Chrome UX people want to have comically large tabs and menus but scaling things down (the correct dpi for the Pixel is 192 and "scale factor" 2) makes them quite reasonable.
Add a comment...

Theodore Ts'o

Shared publicly  - 
Actually, I like using Mutt.  It's fast and supports real mail threading (unlike, say, GMail), and so it can deal with high volume mailing lists like LKML.  The fact that it's more secure is just a bonus.  (One way it's more secure --- I can much more quickly determine which e-mails are phishing attempts.)
Command-line email clients are more secure simply by being simpler.
Andrew Schott's profile photoDoorToDoorGeek “Stephen McLaughlin”'s profile photoIke Devolder's profile photoHoang Tran's profile photo
Ah +Theodore Ts'o your mailbox is tiny! my mutt is cruising along with 100,000 messages :)
Add a comment...

Theodore Ts'o

Shared publicly  - 
Very interesting.   If systemd developers go too insane[1], I wonder how hard it would be to port these daemons from OpenBSD to say, Debian.  :-)

[1] In other news, they've recently reimplemented "su"
Chiang Lin Ng's profile photoAlbino B Neto (bino)'s profile photoStephen Kellat's profile photoMartin Kock's profile photo
+Julian Andres Klode Bah. Thanks. Fixed.
Add a comment...

Theodore Ts'o

Shared publicly  - 
I just wanted to give a shout out to +SendGrid.  Even though I'm very clearly outside their primary target market as far as paying customers are concerned, they quickly provisioned me with a free tier account which I was able to use with gce-xfstests so that someone can get e-mail notifications from a GCE VM running the file system regression tests.

Setting it up was relatively painless:

If you're using gce-xfstests, you will need to get your own sendgrid account (just ask to have a free tier account provisioned), and then configure GCE_SG_USER and GCE_SG_PASS in ~/.config/kvm-xfstests.

Thanks, +SendGrid!
Mike G's profile photoThomas Brandtstaetter's profile photo
Add a comment...

Theodore Ts'o

Shared publicly  - 
I have xfstests running under Google Compute Engine!   It uses kexec to launch the kernel to be tested, and using a n1-highcpu-4 machine with 100GB of SSD provisioned Persistent Disk, and given that I have 10 test configs with each test config taking 40 minutes to run the auto test group for a total of test runtime of around 7 hours, I estimate that it costs roughly $1.25 to do a complete regression test run for ext4.  (BTW, this used to take 24 hours to run on my laptop or desktop)
Prateek Sharma's profile photoSt. Louis Unix Users Group (SLUUG)'s profile photoAlbino B Neto (bino)'s profile photoBjorn Helgaas's profile photo
+Marc-Antoine Ruel  If you can point me at the relevant code that I might be able to borrow, that would be great!
Add a comment...

Theodore Ts'o

Shared publicly  - 
To people making commits to the Linux kernel.  It would save me a bunch of work (and for others are trying to gather statistics for things like "who writes the Linux kernel") if you use an editor macro to add the "Signed-off-by" and "Acked-by" lines.   This avoids noise like this:

1 Eric Dumaze <>
441 Eric Dumazet <>
1 Eric Dumazet  <>
1 Eric DUmazet <>
1 Eric Dumazet <>
1 Eric Dumazet <>

I also see variation in the git Author and Committer fields, although admittedly less since this tends to be automated; however there are times when it's obvious that some developers don't have a .gitconfig file on all of their systems.

Steve Conklin's profile photoMykola Krachkovsky's profile photoJohn Chronister's profile photoAlbino B Neto (bino)'s profile photo
The problem is old and goes all the way to the top of the organization:

commit 8c6b065b792061c2e471d530127f2348fd9d243d
Author: Alan Cox <>
Date:   Fri Jul 20 15:36:31 2007 +0100

    pata_cs5520: Fix probe bug regression introduced in 2.6.22

    Signed-off-by: Alan Cox <>
    Signed-off-by: Linux Torvalds <>
Add a comment...

Theodore Ts'o

Shared publicly  - 
Trying to port features to the Qualcom msm kernel makes me very grumpy.   It would be awfully nice if Qualcomm engineers would (a) not break building the kernel under x86, and (b) use randconfig to make sure the kernel builds with something other than a very specific .config file.    Sigh....
Theodore Ts'o's profile photoMarc Herbert's profile photoWilfried Klaebe (Wonka)'s profile photoHoang Tran's profile photo
28 comments "Obstacles to contribution in embedded Linux"
Add a comment...

Theodore Ts'o

Shared publicly  - 
If you are a voter in the US, please consider signing this.

If you aren't, don't be too smug. Pretty much all other countries have less controls on their spooks compared to the US; some countries might be less competent at implementing a panopticon, but GHCQ (for example) has admitted that they can put any British subject under surveillance if they feel like it, without needing to suspect that the person has committed a crime, and without asking a judge first. 
Read our letter to U.S. House and Senate leadership on how our surveillance laws should be changed. Then add your name to show your support. #United4NSAReform
Dorian Santner's profile photoMatt Dragon's profile photoAndrew Schott's profile photoLee Nelson's profile photo
Given the metadata retention laws the Australian govt just pushed through, "smug" is not where I'm at. Was amusing have a Greens senator using his speech during the debate telling people to go and find a VPN provider though :)
Add a comment...
Yummy corn fritters and pecan rolls are included free at the beginning of the meal! I liked the riverside view.
Public - 3 years ago
reviewed 3 years ago
5 reviews
Public - 4 years ago
reviewed 4 years ago
Public - 5 years ago
reviewed 5 years ago