In practice, you just put enough "dirty"/"weak" random bits into a good CPRNG, and that is it. As long as your estimate of the entropy contained is not too high, this works and is secure. The problem people having with random-number generation is that they either select a bad (C)PRNG or they mess up the entropy estimation in seeding, e.g. due to overlooking negative effects from virtualization. Some also re-use initial, stored seeds (e.g. by cloning a VM and not redoing the seeds), and that is really a beginner's mistake. These are all engineering errors because the people doing it do not understand what they are doing.
I.e. in situations where you are entropy-starved (so not on regular computers) and getting the, say, 256 bits of entropy to seed a CPRNG is hard to come by. This only applies to really small embedded systems that have no A/D converters or the like. (With an AD converter, just sample some noisy source or even an open input and you get something like > 0.1 bit entropy per sample. Needs careful design though.) Whether such a very small embedded system can the use the math needed here is a different question. In most cases, the right solution is to just sample more noise-source input.
So maybe a "theoretical breakthrough", but likely not very relevant in practice.
True. The French in Algeria may just be the most solid data available. Likely everybody smart in history that could observe torture figured this out sooner or later.
Well, you have a point. To me the kernel boot-process ends when "init" is called and the OS boot ends when all the init-scripts have been executed. As this is a story about the kernel, that definition still makes sense even today.
Of course, in pop-culture "booting" always meant "when it is ready so I can do my stuff". In that sense, you can "boot" into X, but if it does not come up, the kernel may not actually be the place to search for the error as quite a few other things need to happen after it has done its thing.
A good source seems to be the war in Algeria 50 years ago. Both the French and the natives tortured excessively, wit little beneficial effect to them, if at all.
As any halfway sanely organized secret organization has both compartmentalized information and knows what to change and move when somebody is captured, torture is never effective, unless the information would have been easily available by other means. This means the torturers and those that are ordering it are deep in the evil spectrum, because they did not bother to find out about the things they are doing beforehand.
Indeed. People under torture lie to make it stop temporarily. They make up stuff when the interrogators insist they must know something, but do not.
On the moral side, the only sane way to deal with torturers and those ordering it is to either execute them or lock them up permanently in a closed mental institution. These people have no place in human society.
It is about destroying the victim and about spreading fear. It has been known to be completely ineffective as a means of getting information at the very least since the French tortured vigorously in Algeria about 50 years ago. They never found out anything worthwhile they did not already know and they really tried to go to the limits possible.
So, yes, sadism, desire to dehumanize and eradicate the victim, and to terrorize others. Noting even remotely compatible with civilized society. However something religious fanatics of all colors really like, just think of the "eternal torment in hell" used to scare people. They really enjoy giving people a preview of that.
You have misunderstood how terrorism works. The way to do it is to do relatively little direct damage, but cause maximum fear. Then let your adversary do it to themselves. Even 9/11 works that way, and the initial damage was relatively insignificant compared to what the US has done to themselves as a result since.
Terrorism is a bit like a blasting cap: Do not supply the explosives yourself, just set it off. Prime targets are hidden insecurities often under hidden unjustified arrogance ("They would never dare to hit us at home!", "Foreign terrorist cannot hurt us!" and similar mistaken feel-good propaganda used to cover fear by boasting) that are easy to bring to the surface. After hit, the target will then do whatever they can think of to established what they mistakenly believe think was the status-quo before. That cannot work, as the original, hidden problems with that are all still in place. But the target usually does an inordinate amount of damage to themselves in this futile quest. A core booster to this negative effect are law-and-order authoritarians that will grasp at any and all straws to "do something" and will try to establish a police-state or even total fascism as "defensive" measure. Now, history shows us that this makes everybody actually significantly less secure because of the massive negative effects on freedoms, rational use of funds, education and innovation, etc. these efforts have. A society too much focused on "defense" implodes, because it forgets to do the things that make it work and worthwhile as a society.
The totally fascinating thing about this is that a target that is rational and at-ease with itself cannot really be hurt by terrorism to any significant degree.
Good list. However the systemd-fanatics cannot give you that explanation, they must attack what came before (and worked well), because they have no valid arguments why their way is better. That is why they universally use this tactics.
You are free to use an inferior solution any time you want to. Just do not try to push me to do so. The technical aspects have been discussed to death and that you even ask means you are posting in bad faith.
I do non-module kernels with only the needed stuff for all production machines, because there have been module-loading related vulnerabilities in the past and this way I am not dependent on the user-space or an initrd for booting. I also do not use an initrd, and with systemd-free Debian that works still well and has for the last 15 years. Just do not do a kernel-package, but make it directly and boot it yourself. Reduces complexity considerably.
I do not understand it either. Unfortunately, some of our customers require a document they can edit, so I have to use MS Office for that. For other things we have a LeTeX Template that produces documents that look almost the same.
At least fundamentalist conservative societies have far less of them or ones of far lower quality. Independent or analytic thought is not encouraged in those societies. It is absolutely critical in any STEM professional, regardless of gender however. (Even if some of the commentators here give the opposite impression...)
Well, so what if? We really have outgrown the need to produce as many offspring as possible by now. This dirt-ball is overcrowded, and individual freedoms and growth rightfully is viewed much more important in progressive societies than producing the next generation.
Compile your own? Putting in 4.6 instead of 4.4 should be pretty painless. And Ubuntu, like any halfway usable distro, should have a clean way to do this via creation of a custom kernel package.
Systemd is not required by the kernel and, for example, Debian still runs well without it, and I expect it will stay that way. The push-back if they tried to make it unusable without systemd would probably fracture the project and making it the default init-system was hopefully a bad enough experience for the systemd-mafia. It is really advisable to not use systemd on anything needing stability or security and I doubt it will ever mature to the degree needed to fix that. The design is just not clean enough and its main designers do not even begin to understand why KISS is critical in solid engineering.
What stops me is that I do not want insecure MS Office to run on Linux. I would rather jail it in a Windows VM. Maybe I could jail it in a Linux VM (running under Linux) with Wine though. Has anybody tried that? Should work in principle just as well as Linux native.
Currently, the only thing that makes me maintain a Win7 installation is gaming. For MS Office (needed for part of my work), I have a laptop from work. Everything else (coding, data-processing, etc.) is and has been Linux for me since 1995. I do hope that the gaming-situation on Linux will get much better with Vulcan, but it is going to take some time.
In principle, most modern games are cross-platform, and as soon as there is solid Vulcan support in the mainstream-engines, making an additional Linux-release is not really a cost factor anymore. I think the chances are good that by end-of-support for Win7, I can finally become MS-free again for my own computing.
Compile your own kernel, have only in there what you want. Be a luser (yes, this is the correct spelling) and have everything in there. As the kernel in in practice very modular, having this stuff as an option does not slow the rest down or cause problems. So far, this model has proven far superior to what MS does with their user-installed drivers.
Slashdot Mirror
Good summary. This really is _theoretical_ work.
In practice, you just put enough "dirty"/"weak" random bits into a good CPRNG, and that is it. As long as your estimate of the entropy contained is not too high, this works and is secure. The problem people having with random-number generation is that they either select a bad (C)PRNG or they mess up the entropy estimation in seeding, e.g. due to overlooking negative effects from virtualization. Some also re-use initial, stored seeds (e.g. by cloning a VM and not redoing the seeds), and that is really a beginner's mistake. These are all engineering errors because the people doing it do not understand what they are doing.
I.e. in situations where you are entropy-starved (so not on regular computers) and getting the, say, 256 bits of entropy to seed a CPRNG is hard to come by. This only applies to really small embedded systems that have no A/D converters or the like. (With an AD converter, just sample some noisy source or even an open input and you get something like > 0.1 bit entropy per sample. Needs careful design though.) Whether such a very small embedded system can the use the math needed here is a different question. In most cases, the right solution is to just sample more noise-source input.
So maybe a "theoretical breakthrough", but likely not very relevant in practice.
True. The French in Algeria may just be the most solid data available. Likely everybody smart in history that could observe torture figured this out sooner or later.
This is an interesting link, in any case. Thanks!
Well, you have a point. To me the kernel boot-process ends when "init" is called and the OS boot ends when all the init-scripts have been executed. As this is a story about the kernel, that definition still makes sense even today.
Of course, in pop-culture "booting" always meant "when it is ready so I can do my stuff". In that sense, you can "boot" into X, but if it does not come up, the kernel may not actually be the place to search for the error as quite a few other things need to happen after it has done its thing.
Sorry for the misunderstanding.
A good source seems to be the war in Algeria 50 years ago. Both the French and the natives tortured excessively, wit little beneficial effect to them, if at all.
As any halfway sanely organized secret organization has both compartmentalized information and knows what to change and move when somebody is captured, torture is never effective, unless the information would have been easily available by other means. This means the torturers and those that are ordering it are deep in the evil spectrum, because they did not bother to find out about the things they are doing beforehand.
Indeed. People under torture lie to make it stop temporarily. They make up stuff when the interrogators insist they must know something, but do not.
On the moral side, the only sane way to deal with torturers and those ordering it is to either execute them or lock them up permanently in a closed mental institution. These people have no place in human society.
It is about destroying the victim and about spreading fear. It has been known to be completely ineffective as a means of getting information at the very least since the French tortured vigorously in Algeria about 50 years ago. They never found out anything worthwhile they did not already know and they really tried to go to the limits possible.
So, yes, sadism, desire to dehumanize and eradicate the victim, and to terrorize others. Noting even remotely compatible with civilized society. However something religious fanatics of all colors really like, just think of the "eternal torment in hell" used to scare people. They really enjoy giving people a preview of that.
Wow, you are _really_ fucked in the head.
Larry Ellison gone would probably boost the economy. Just talk to anybody that has to use Oracle's bad and expensive products.
You have misunderstood how terrorism works. The way to do it is to do relatively little direct damage, but cause maximum fear. Then let your adversary do it to themselves. Even 9/11 works that way, and the initial damage was relatively insignificant compared to what the US has done to themselves as a result since.
Terrorism is a bit like a blasting cap: Do not supply the explosives yourself, just set it off. Prime targets are hidden insecurities often under hidden unjustified arrogance ("They would never dare to hit us at home!", "Foreign terrorist cannot hurt us!" and similar mistaken feel-good propaganda used to cover fear by boasting) that are easy to bring to the surface. After hit, the target will then do whatever they can think of to established what they mistakenly believe think was the status-quo before. That cannot work, as the original, hidden problems with that are all still in place. But the target usually does an inordinate amount of damage to themselves in this futile quest. A core booster to this negative effect are law-and-order authoritarians that will grasp at any and all straws to "do something" and will try to establish a police-state or even total fascism as "defensive" measure. Now, history shows us that this makes everybody actually significantly less secure because of the massive negative effects on freedoms, rational use of funds, education and innovation, etc. these efforts have. A society too much focused on "defense" implodes, because it forgets to do the things that make it work and worthwhile as a society.
The totally fascinating thing about this is that a target that is rational and at-ease with itself cannot really be hurt by terrorism to any significant degree.
Good list. However the systemd-fanatics cannot give you that explanation, they must attack what came before (and worked well), because they have no valid arguments why their way is better. That is why they universally use this tactics.
You are free to use an inferior solution any time you want to. Just do not try to push me to do so. The technical aspects have been discussed to death and that you even ask means you are posting in bad faith.
I do non-module kernels with only the needed stuff for all production machines, because there have been module-loading related vulnerabilities in the past and this way I am not dependent on the user-space or an initrd for booting. I also do not use an initrd, and with systemd-free Debian that works still well and has for the last 15 years. Just do not do a kernel-package, but make it directly and boot it yourself. Reduces complexity considerably.
I do not understand it either. Unfortunately, some of our customers require a document they can edit, so I have to use MS Office for that. For other things we have a LeTeX Template that produces documents that look almost the same.
And then there are people that just refer to the definition of what "booting an OS" means, instead of doing silly games.
Are sure sure did not? USSR mathematicians were world-class and leading the field in quote a few areas. Of course, that is in the past.
At least fundamentalist conservative societies have far less of them or ones of far lower quality. Independent or analytic thought is not encouraged in those societies. It is absolutely critical in any STEM professional, regardless of gender however. (Even if some of the commentators here give the opposite impression...)
Well, so what if? We really have outgrown the need to produce as many offspring as possible by now. This dirt-ball is overcrowded, and individual freedoms and growth rightfully is viewed much more important in progressive societies than producing the next generation.
Sounds like badly designed/documented hardware on Intel's part. Not that much of a surprise.
Incidentally, you cannot boot into X, that is just some userspace-stuff your distro is doing to fake it. Boot is long over at that time.
Compile your own? Putting in 4.6 instead of 4.4 should be pretty painless. And Ubuntu, like any halfway usable distro, should have a clean way to do this via creation of a custom kernel package.
Systemd is not required by the kernel and, for example, Debian still runs well without it, and I expect it will stay that way. The push-back if they tried to make it unusable without systemd would probably fracture the project and making it the default init-system was hopefully a bad enough experience for the systemd-mafia. It is really advisable to not use systemd on anything needing stability or security and I doubt it will ever mature to the degree needed to fix that. The design is just not clean enough and its main designers do not even begin to understand why KISS is critical in solid engineering.
What stops me is that I do not want insecure MS Office to run on Linux. I would rather jail it in a Windows VM. Maybe I could jail it in a Linux VM (running under Linux) with Wine though. Has anybody tried that? Should work in principle just as well as Linux native.
Currently, the only thing that makes me maintain a Win7 installation is gaming. For MS Office (needed for part of my work), I have a laptop from work. Everything else (coding, data-processing, etc.) is and has been Linux for me since 1995. I do hope that the gaming-situation on Linux will get much better with Vulcan, but it is going to take some time.
In principle, most modern games are cross-platform, and as soon as there is solid Vulcan support in the mainstream-engines, making an additional Linux-release is not really a cost factor anymore. I think the chances are good that by end-of-support for Win7, I can finally become MS-free again for my own computing.
Compile your own kernel, have only in there what you want. Be a luser (yes, this is the correct spelling) and have everything in there. As the kernel in in practice very modular, having this stuff as an option does not slow the rest down or cause problems. So far, this model has proven far superior to what MS does with their user-installed drivers.