Profile

Cover photo
Christian Bornträger
Works at IBM
157 followers|9,971 views
AboutPostsPhotosVideos

Stream

Christian Bornträger

Shared publicly  - 
 
 
Früher hat man sich noch angeguckt. Und miteinander geredet.
 ·  Translate
6
Carsten Otte's profile photo
 
Hach früher war alles besser, als die Welt noch Schwarz-Weiß war und die Nachrichten von gestern auf toten Bäumen waren :-).
 ·  Translate
Add a comment...

Christian Bornträger

Shared publicly  - 
 
Platte für 5. Lecker
 ·  Translate
3
Christian Bornträger's profile photoJens Remus's profile photo
2 comments
 
Ja war eigentlich immer super.
Allerding war es die letzten Wochen irgendwie immer zu. Keine Ahnung ob die zu gemacht haben.
 ·  Translate
Add a comment...

Christian Bornträger

Shared publicly  - 
 
Cool, könnte was für meinen Sohn (2 1/4) sein. Bagger kommen immer gut.
 ·  Translate
2
Add a comment...

Christian Bornträger

Shared publicly  - 
 
Picture taken this year in Italy Barberino Val d'Elsa. Local bank
3
Carsten Otte's profile photo
 
How to fix it: If it's too tight use Caramba. If it's too lose use duct tape.
Add a comment...

Christian Bornträger

Shared publicly  - 
 
 
Beware of vertical videos
3
2
Add a comment...

Christian Bornträger

Shared publicly  - 
 
spot on analysis on capital based economy
 
How does the economy nowadays work? I think it's quite easy:
a) capitalist economies strive for efficiency. This rise in efficiency obviously leads to less workforce needed to produce the same amount of service.
b) to avoid massive unemployment the economy has to grow for at least the size of the rise in efficiency. This is probably around 1-2% annually for developed countries. For economies with more potential (like Brazil, India or China) this figure is probably much higher.
c) since unemployment is not good for politics, policy is to work for economic growth.
d) globally, the monetary amount of sold goods is identical to the amount of purchased goods. the complete sum of these is the total size of the global economy.
e) growth means that tomorrow more money is spent (and earned) than today.
f) How can that be accomplished? Only in two ways
1: increase trade frequency
2: create more money by giving credit to the buyers.
Credit was made for exactly that: If someone needs money to expand his business or invest to cut cost, he can pay for these investments with lent money. The idea being, that after the expansion is in effect, it generates more benefit that the credit costs.
For a global economy however, which needs to grow also next year, the lending never stops.

Both can be seen in the financial sector: Massive credit giving without hard backup as well as high-speed-trading
I am not sure about high-speed-trading, I just have a bad feeling about it. But the concept of paying back debt is basically destroying the benefit of global debt. This is what we are seeing around the (western) world currently: A massive debt crisis.

The funny thing is that reducing credit lending will not solve anything. It will just end in economic stagnation, unemployment and financial deflation. This is what Japan is facing since decades now.

So how do we get out of this mess? I don't know. I guess limited annual lending at 0% credit (donating or giving as present) might do the trick.
On the other hand, getting money for nothing (and chicks for free?) might become a problem for competition that undoubtedly is the engine for technical and economic advancement.

So the bug is found, but how to fix it?
1
Andreas Pinkert's profile photoIngo Molnar's profile photo
2 comments
 
The problem with that is that "injecting money"normally  means: Take credit. With the price tag attached to this, it will worsen the financial situation of the state. Finally the state is near bankruptcy My idea of 0% credit still stands. If you mean that, I agree with you. My point is that it would need to be backed up by a money press and not by credit.
Add a comment...
In his circles
232 people
Have him in circles
157 people
Thorsten Scharpenberg's profile photo
Marc Schöchlin's profile photo
Hannes Reinecke's profile photo
Dirk Herrendoerfer's profile photo
Carsten Otte's profile photo

Christian Bornträger

Shared publicly  - 
 
I love german wine as well, but I guess that this sticker is not located at the best place for a wine ad
2
Christian Bornträger's profile photoMark K Post's profile photoCarsten Otte's profile photo
3 comments
 
+Carsten Otte I somewhat agree for red wines (but several wine bloggers and experts don't and praise the spaetburgunder or lemberger). Regarding white wines there are several examples of German wines winning the world championship in the last years in a blind tasting competition. In fact, climate change actually challenges German wine makers to keep the alcohol level at a reasonable level due to that, so in some years there is already more than enough sun.
But of course, there is also fantastic wine from Italy France and others. I personally consider the wine maker and the cellar more important than the country.
Add a comment...
 
Trifft den Nagel auf den Kopf
 
Deutschland wählt.

(Karikatur aus der Schweiz.) #btw13   #merkel   #cdu   #spd   #grüne   #fdp  
 ·  Translate
1
Add a comment...

Christian Bornträger

Shared publicly  - 
 
Marktwirtschaft wie sie nicht funktioniert #1:

2.Oktober, Aral Tankstelle. Es gibt nur Sommerzusatz für die Scheibenwaschanlage
 ·  Translate
1
Andreas Pinkert's profile photoIngo Molnar's profile photoCarsten Otte's profile photo
6 comments
 
+Ingo Molnar jedenfalls haben sie das Zeug für die Waschanlage am Start
 ·  Translate
Add a comment...

Christian Bornträger

Shared publicly  - 
 
Just some remarks about my current pet project:
I am currently fixing up valgrind to understand the side effects of some KVM ioctls.
Lots of ioctls are just fine (like KVM_CHECK_EXTENSION) since they only set the return value and expect a simple arg value (e.g. 0). This is valgrinds default assumption for ioctl definitions. I had to add case statement for them as well, since the numerical value doesnt tell you if the definition was based on the _IO macro (== no input/output) or is just one of the historic values without any direction hints. So for valgrind it would have been better to not use 00 for bits 31/30 in case of _IO. Anyway that work is easy, it just needs to be done.

Some s390 ioctls (like KVM_S390_INITIAL_RESET) dont even care about arg. Here it would have been better to explicitly check for arg=0 in the kernel to allow for future extensions.

ioctls with direction hints are easy - valgrind can handle those.

The most interesting part is probably how to deal with KVM_RUN. For memory bits valgrind has an associated bit regarding the definedness. Since valgrind relies on JIT-ing all the code it will not track the state of memory that was set by the guest running under KVM. Currently I see two options:

1. mark the guest memory as undefined: this will trigger false positives, but gives you a list of places were the code or data flow depends on guest memory - useful for targetted review. This is the default if the guest memory
is allocated with qemu_vmalloc.
2. mark the guest memory as defined: This reduces the noise and helps to detect real errors that are in qemu even without touching guest state.

Are there any better ideas? Maybe a modified memcheck tool that is based on the notion of untrusted instead of undefined


PS: another hickup is KVM_SET_SIGNAL_MASK. valgrind implements exit_group via exit calls in each thread to capture all states gracefully. Therefore it defines a usr signal as SIG_VGKILL and opens the signal mask for that signal before doing the client system calls. Than enables valgrind to kick threads out of blocking system calls.
A spinning guest cpus wont react on SIG_VGKILL, though since KVM_SET_SIGNAL_MASK will override the thread signal mask. Seems that I have to add a wrapper for KVM_SET_SIGNAL_MASK as well that takes care of that.
3
Add a comment...
People
In his circles
232 people
Have him in circles
157 people
Thorsten Scharpenberg's profile photo
Marc Schöchlin's profile photo
Hannes Reinecke's profile photo
Dirk Herrendoerfer's profile photo
Carsten Otte's profile photo
Basic Information
Gender
Male
Work
Occupation
Software Engineer
Employment
  • IBM
    Open Source Development for IBM mainframes (s390): gcc, valgrind, kernel, KVM, 2003 - present