본문 바로가기

카테고리 없음

Anzwix Chromium Add Posix Shared Memory Support For Mac

(1 reply) There have been some recent discussions on Shared Memory problems in Darwin on the darwin-kernel list at lists.apple.com. The subject is: shm_open and mmap: Invalid argument? Tain't clear to me if this is relevant to the MacOS X and Postgresql shared memory problem that I see mentioned here from time to time. ( some of the discussion centers on releasing the segments).

I recently asked a about the differences between capabilities and mandatory access controls. Among the answers I got the point was made that systems like SE Linux in targeted mode are not a typical MAC system, as the concern is not about restricting information flow. Then, in a framework like SE Linux targeted mode where the focus would seem to be on minimizing harm in the event of unauthorized access, how different is this in practice from a capability based system?

Note, I am only asking about the practical different, not the theoretical differences. Are different attacks possible on SELinux in targeted mode than on a capability based operating system? Can they both prevent access to the same granular level? I think a lot of the problem with capability-based systems and our understanding of them is that there are few in the wild. However, you're in luck - FreeBSD have just added support to their 9.0 release. Capsicum is a research project to implement a lightweight practical set of capabilities1. This work involved people at the Google Security Research team and includes an experimental chrome browser build designed to use the system.

  • My use case is a big shared memory database, typically at a 1 GiB order, but we usually reserve 3 GiB shared memory space just in case it grows. This data is constantly being updated by a writer (a process), and must be made available to readers (other processes).
  • Base: Introduce SharedMemoryTracker for POSIX (but not macOS) This CL introduces SharedMemoryTracker to measure memory usage of SharedMemoy correctly and report them to memory.

What might be most of interest to you are the list of features:. capabilities - refined file descriptors with fine-grained rights Practically, will operate like SELinux, with obvious administration differences. Has analogies to type labels. process descriptors - capability-centric process ID replacement Has analogies to SELinux domains. From a user perspective, may appear to be no different. Capabilities are different from Permissions, which is what SELinux calls capabilities (doh).

As an example, capabilities can be handed off to other programs, possibly attenuated. This is in contrast to the ACL based models where the access controls are managed separately from the programs. Think of capabilities as unforgeable references to objects, or possibly a subset of the methods on objects, that can be used at runtime by a program. So if I want someone to have update a report to me, I can pass them a capability (reference to the rw interface of the file), which they can provide to a program acting on their behalf. In turn, they can delegate the update process to someone else, by passing on the capability. I don't have to open up directories to wider access.

This leads to a programming style where the security/trust relationships are considered at design time and come as a fairly natural extension of the usecase: eg if I hand over my camera to someone to use, I am also giving them permission to use it.

Inb4 250 replies of people shaking their fists in the air. I'm thinking about creating a systemd thread bingo so we can all have fun in these reply thread, because no matter what side you're on, you know it's always the same replies over and over again. Seriously, there's always the guy who 'just migrated' to FreeBSD and feels the need to share this new on an unrelated new article, you have the usual lies, misconceptions and FUD shout out by both sides of the argument, you have that other guy who compares Poettering to hitler or systemd to big brother.

Don't forget the guys who actually know about systemd and reply seriously with a wall of test to every troll. And the timid users who find systemd to be 'not that bad'. You've got the slowpoke who is just now realizing that systemd is used by everyone and requests infos on distros not using it yet. You know it, I know it, these reply threads are always the same and this will be no exception. Enjoy your stay on Soylentnews and remember that the systemd ride never ends. Systemd is a rapidly-growing binary platform alternative on most standard Linux distributions, that is gradually absorbing all use cases once the domain of POSIX deamon and userspace functions.

A reasonable analogy is Google's Android - which runs on the Linux kernel, but cannot be properly understood as a Linux system. Android uses Linux for primitive device bootstrapping and operation of low-level memory, device primitives, I/O and servicing of a JVM - with the functional use cases entirely delivered in the java machine. Systemd will similarly become the 'real' platform and API target for proposed development on Linux - abandoning Unix/POSIX design principles and and portability.

Those of us with long memories see this as a Trojan horse - subverting the best of comprehensible, open systems and re-injecting the failed monolith of VMS/Vax style computing for those more interested in markets than values. I do not consider myself a GNU/Linux expert however I have been using and maintaining my own desktop machines on one version of Debian or Ubuntu or another since '97 and have worked in IT and IT related roles since '94. I personally do not like the mission creep, bloat, or monolithic nature of systemd which is why my desktop machines all now run Slackware and i have just wiped one ready for a little BSD experimentation. I may also get a machine ready for Devuan when it's available. I realise that this is only my personal opinion but i would prefer that an init system like systemd should be an option rather than a default, perhaps allowing users to choose which init system they would prefer during installation. My personal experience of numerous installs over the years has shown me how easily this could be achieved.

I would suggest a selector section to be added to GUI style installers similar to that used for hard disk partitioning whereby the installer asks whether the person performing the install wishes for a guided install, (easy method,) or a manual install, (custom/expert method,) where someone who does not want systemd can have the GNU/Linux distro of their choice, the way they want it. I understand that there are many people coming to GNU/Linux who don't have experience with the way thing were before all the additional bells and whistles were added.

The best example I can offer as to the affect that making distros easier for non-expert computer users is that I have migrated both of my parents, who are in their late sixties, away from Wind'ohs in favour of GNU/Linux, (they really don't need to be spending money on new laptops,) after a couple of days with their machines they were just as happy using them as they had been with Win XP. So, no angry hair pulling, no ad hominem attacks, no straw men, simply an appeal for common sense from an admitted idiot. You forgot Devuan. Somebody always mentions Devuan. And you forgot Slackware. Someone (sometimes me) always mentions Slackware.

Of course, with KDE's KDM about to depend on systemd's logind, one of two things may soon happen: Either Pat will drop KDE (doubtful), or he (or maybe Eric H) will shim around it for as long as possible. Personally, I say chuck KDE to the curb and focus on Xfce and Razor-QT, or maybe Openbox; I've had a deep seated loathing for KDE since the 3.x days, though I'll admit it's gotten much better lately. Still, to me KDE and systemd seem like peas in a pod: Bloated, full of politics and controversy, and completely unnecessary to a properly working GNU/Linux or BSD OS. Bring on the torches, KDE fans!

You're not wrong for liking and using KDE, KDE is just wrong for me.:-D. Even if it isn't a trojan, I think systemd should still be considered a form of malware/crapware. I don't want it on my Linux systems. Yet if I install pretty much any practical (sorry, that excludes you, Slackware and Gentoo) Linux distro, systemd will be installed without me wanting it to be there. In some cases it's virtually impossible to get rid of without ruining the installation. That meets all of the criteria of malware/crapware, as far as I'm concerned: it doesn't do anything useful, it's unwanted, and it's damn hard to remove without damaging the OS installation.

For those wondering, slackware is about as modern as any other distro these days, and better put together than most. Not to mention stable as hell. The only crashes I've ever had on Slackware were either faulty hardware or faulty third-party software (Chromium browser, I'm looking at you). In fact, it's pretty damn boring from an OS hobbyist point of view. When I want to watch things break and figure out why they broke, I'll spin up Haiku, or Ubuntu, or Arch Linux. When I want to get shit done and not worry about my OS, I switch to my main workstation with Slackware.

Nice to see I'm not the only one waving a red flag here.I mean doesn't it strike ANYBODY else as odd that right after Snowden showed how bad the spying really was and all these 'protect yourself using Linux LiveCDs focused on privacy' articles hit suddenly Red Hat, a company that gets something like 85% of its income from TLAs, says 'Hey this system that nobody was really bitching about MUST BE REPLACED by this big sprawling monster that is connecting to more and more critical systems by the day!' I'm sorry but everything I've read frankly comes off more like MSFT pushing Metro or Apple pushing the latest iStuff, it doesn't read anything at all like the tech heavy merit focused way I've always seen Linux development done in the past. Surely I'm not the only one who is reading the posts from devs and feeling like I stepped into bizarro-land? It's not just you. Somewhere around 50% half of the Linux userbase would appear to agree with some or all of that analysis.

Although admittedly, the number is probably still dropping as some of them migrate to BSD. Well lets put ourselves in the shoes of a guy at a TLA. Lets see.we have red Hat by the balls, because 85% of their living comes from OUR checks so they will do what they are told.and what's that?

They have a sloppy as fuck cowboy coder that wants to stick his paws on every critical subsystem? AND he is the guy that came up with Pulse, which is probably the least stable thing on any Linux desktop? You see why they wouldn't need to stand over him saying 'You WILL put backdoor 23j in this point in the chain' because they have enough blackhats on the payroll that all they really need is to have a really sloppy coder (and considering Poettering has gone on record blogging such wit as 'Can't get systemd working on ARM, shipping it anyway' I'd say he is more than qualified for that task) who keeps sticking more and more critical systems under his control.

Remember in the Windows world its NOT the OS that is usually the first point of attack, you go after a weak component like Flash, Java, IE, and use THAT as the door to the good stuff.and what do we see on today's TFA? Tying fricking MEDIA PLAYERS into the fricking replacement for init?!?! If I were working for a TLA I'd just find a guy like Poettering and let his massive ego and bad coding do the rest. By the time he's done Linux will be nothing more than what DOS was to Win98, a bootstrap that hands over to systemd that then spreads all over the damned place and thanks to it all being tied together in a big fucking mess all it'll take is a weak spot in one to make all vulnerable, weakest link in the chain and all. Lets face it if it isn't a TLA they really should send RH some fruitcake and flowers, because they couldn't have asked for a better break than the sprawling ever expanding oilslick that is systemd.

There is a oddity in the systemd changelog (or should i say systemd-udev?). A patch was offered that would make systemd trigger a system lock whenever a keyboard or mouse was plugged in, requiring a password to be entered to unlock. This arguably to protect against mouse wiggling dongles found in police data seizure kits. Poettering accepted the patch, but only after modifying it so that it only triggered on the USB id of a known dongle. This in effect neutered the patch, as changing the USB id to be anything (a Microsoft mouse perhaps) is relatively straight forward.

Some of what systemd wants to accomplish is good. Some of it less so. But, it is a fact that it is breaking things. I only had systemd on my boxes for a couple weeks (stealth upgrade on Debian Jessie), and found that manual fsck if rootfs has errors was not possible without exploiting a bug in systemd (below), shutdown -rF to run fsck on next reboot was broken, consoled/logind ran commands typed at the root pwd prompt (for the emergency shell) as root!!

My unpriv lxc containers stopped working, Even with systemd purged from my systems, there are still bits of the brokenness left since systemd took over udev etc. (this is from yesterday): root@nim:/etc/udev/rules.d# dmesg grep rename 160 systemd-udevd27418: renamed network interface wlan1 to rename21 yup, add vif to a wireless card post systemd mess, and now systemd-udev renames the interface to 'renamerandom small integer'. I think the issue is that people who only do simple stuff are less likely to run into issues with systemd (both like the above, and also not finding that binary logs are an issue since the current text logs were unusable for them since they don't understand REs).

These simple users love systemd. Folks who do more advanced things find systemd is getting in the way / breaking things, and have less love. A few bits of systemd show promise (if they ever focus on bugfixes instead of adding more crap), but the whole is a pretty scary mess. There seems to be a shism going through IT these days. Where before there was a continuum between 'user' and 'developer' (power-user, admin, etc), now there is only user or developer. And the developer has to be opted into with careful thought, as doing so will mean not bringing over anything from the user side unless you have it stashed away somewhere. Best this AC can tell, it comes down to commercial interests.

The continuum allowed such things as power-users to copy random files around, and Big Media can't have that. And so the user side is locked down 'for your own protection'. And to go developer you either have to forgo anything commercial, or you have to sign a contract with your own blood so they know where to send the lawyers. I'm glad I'm not the only one who saw 'download from the Internet' as a vague threat.

Of course, it appears that most of systemd seems to assume that Linux is used on personal laptops and desktops and not on corporate-owned devices. Does this mean that systemd can download and install updates from anywhere, or is there a well-defined repository? None of the referenced articles seems to have the answer to that question. If I can't see where my updates are coming from, why would I allow this on my (or my company's) servers? You're correct.someone might 'sudo wget shady.site.com/not-virus /usr/bin'.

IF they were on my sudoers list with sufficient rights. In that case, it would be logged to a text based log file.

Don't think 'my laptop that I tinker around with occasionally' or 'some random VPS that I use to host my blog.' Think 'a server farm that contains information to be governed by HIPPA, SAS-70, ISO-20001, or some other data privacy/security regulation.' Yes, Linux is used in these types of environments. No, unverified and poorly documented system packages are not wanted on them.

'systemd' is a bunch of different things, ranging from an init system and binary (ick) logger to programs that manage login sessions, date and time, etc. It's a blanket moniker for a group of programs that Poettering and co.

Have decided will be the standard components for Linux systems. With that mindset, they're not worrying about those parts depending on each other, because it's supposed to be a standard kit that you use all together. It's sort of like how FreeBSD separates the base system, which updates all together at once, and is separate from the other packages. The naming of the entire suite as 'systemd' causes issues because people sometimes say 'systemd' while meaning 'systemd init', and it's akin to the problem that the KDE people tried to solve by naming the desktop Plasma, the entire KDE suite of apps as 'KDE Software Collection', and using another name for KDE as a community/entity. Most of the new features are additions to the suite, not pieces bolted onto systemd-init in the style of emacs. There's still the question of 'why do I want to use this suite at all?'

However, which is harder to answer. For the distro maintainers it's probably going to make things easier, because adopting systemd as a FreeBSD style userland will simplify testing and maintenance. For the user, unless you happen to like the systemd components, it's either 'I don't touch it and don't care' or 'I don't like my stuff being ripped up and replaced by systemd's stuff', the latter of which either hates all of systemd, or like me, dislikes the excessive integration that makes it hard to cherry-pick the good parts and replace the rest. I, for example, avoid the init and logger, but allowed systemd-logind because it's generally inoffensive and desktop components need it. I think my biggest complaint with this FreeBSD-style approach is that it's essentially making RedHat the gatekeeper of the Linux userland. You see it in the features it's adding here, like the containers.

Anz Wix Chromium Add Posix Shared Memory Support For Mac

It's good if you want homogenisation, sure, but it's essentially turning the other distros into respins of RedHat instead of unique systems with their own strengths. If every distro has the same underlying FreeBSD-esque userland (systemd) and the only difference is what extra software is on top, then the distros become little more than Kubuntu vs. I'm using Debian for a reason, and that reason isn't 'I love how RedHat works'. I've seen people claim over the years that distro fragmentation is what has kept Linux from 'winning' against Windows and OS X, and that what it needs is a consistent base. They may be right, but I'm not looking forward to victory when it requires handing the keys to the kingdom over to RedHat, via systemd.

As the saying goes, the road to hell is paved with good intentions. I've seen people claim over the years that distro fragmentation is what has kept Linux from 'winning' against Windows and OS X, and that what it needs is a consistent base.

They may be right, but I'm not looking forward to victory when it requires handing the keys to the kingdom over to RedHat, via systemd Reminds me of the term wikipedia.org You can spin it as victory all you want, but in reality you are talking about critical failure. There is literally nothing good that could possibly come of this. That would imply that systemd-shim is maintained by the systemd dev.

Not at all the case. If it was, there would be little to no complaining, as people would just use systemd-shim and shut up. No, systemd-shim is maintained independently. And so has to chase the systemd API tail. Every time logind or similar has a change in how it talks to systemd-as-init (.groan.), the shim devs have to track down and implement the changes on their own.

Thus -shim is in pretty much the same situation as Wine, except for having been started earlier. It sounds almost like a standardized linux distro running within the user's own non-standardized one in order to allow everyone to run the same applications without rebuilding them right? Basically, yeah. Its what one of those old fashioned 'operating systems' was supposed to implement.

But the old standards were huge and baroque and interdependent so we'll create a new small sleek standard, which will rapidly metastasize into something as big as the old system with an even bigger wrapper around it, until someone comes up with a new small sleek standard. Repeat infinitely into the future.

IBM VM was supposed to do this. Then microcomputer sized server hardware. Then OS packages. Next, java jarfiles. Next, virtual images.

Next, not-so-partitioned docker containers. I'm sure there will be a new one. It never works out and there's plenty of money to be made selling the dream.

This time I'm sure they'll get it right. I suspect uselessd is in a difficult situation: just like Wine and Octave, it's running to catch up with a moving target. (uselessd with systemd, Wine with MS Windows, Octave with Matlab).

Anzwix Chromium Add Posix Shared Memory Support For Mac Free

So people don't appreciate the value of the program, instead they whine: 'why hasn't it implemented feature X yet, that has had for 2 months?' Better to just not step on that caroussel at all.

My sincere hope is that the Devuan project will work good enough to provide a 'safety net' for disillusioned uses dropping out of Debian Jessie. I'll stay with the systemd less distro I use, PClinuxOS, even with KDE looking at systemd in 6 months time there isnt an issue because we'll just change the default UI:) It's not that I'm against systemd it's because it's a buggy piece of shit that lazy distro rollers are using way before it's ready. There isn't one systemd distro that will boot after being installed on this laptop. There is no use case for this. This is an imaginary use case developed by a stupid fuck who thinks he is a computer programmer. I mean, I just installed a fedora system as a dhcp server, for test purposes. By default, systemd couldn't even start dhcpd.

I thought I was going crazy. But, it's an actual thing.

The unit file contained improper crap, and the result, dhcpd would never start automatically. Distro issue or systemd issue? Who gives a fuck. Fact is, no-one from systemd or the distro 'caught' this specific use case. It's time for Lennart and co to fuck off. Their ideas are fucking stupid. ZOMFG LETS IMPORT CONTAINERS!!#@#@#@!oneone!!eleventy.

Anzwix Chromium Add Posix Shared Memory Support For Mac Pro

SharedShared

What the fuck is dhcpd? Who cares, we are building systemd-dhcp. Seriously Lennart and related system developers need to fuck off.

They don't know shit about Linux, and are hell bent on fucking it up because they all have hard ons for Apple, and think systemd will turn linux into os x. 1 Troll for speaking the truth? Fact is, anyone modding this as troll is a systemd appologist. The web is full of them. And I doubt any of them run even one server for a living. No, running a linux box in your mums basement does not count, nor does running Linux on your laptop. I guess I will be seeing FreeBSD sooner rather than later, with fuckwits like lennart and soylentnews calling me a troll because of stupid and avoidable mistakes, that aren't seen as issues, because thats the price we need to pay to live with the greatness that is systemd.

Die in a fire. They run them on 'servers', aka a 3 digit number of instances over at Amazon's E2C service. The biggest complainers, and usually scoffed at for being old farts, are the traditional server admins. One take on it is to look at it as ranch animals vs pets.

The 'cloud' way is like ranches, the VM instances cattle. If one keel over ten more are ready to take its place. Traditional servers are like pets.

Groomed and taken care of to ensure a long and happy coexistence. The big drivers behind systemd are firmly in the server ranch camp. Most of them don't use Linux as anything more than middleware between the VM and their oh so fancy web service, while on the desktop is the macbooks all the way. There is a undercurrent though, and that is government/military systems. It seems the US military has finally figured out that using Windows is not a reliable option. As such they have embraced RH in a massive way.

And apparently NSA is actually using a in-house recreation of Amazon's E2C. Meaning that it is in software terms E2C, but running on NSA hardware within NSA buildings. So what looks like a webserver push at first glance, may also be a sales pitch to the MIC. I wonder how long before our entire viewing log can be forwarded when VLC and/or systemd phones home and/or the RIAA.

Do you think that's crazy talk? VLC already does by default (for media info), and the Debian Ganeti packages phone home (ie.

Google) explicitly to track who's using it if you want another example. Browsers are particularly famous for this kind of thing, but there are plenty of other offenders, and they're multiplying exponentially - never thought I'd see this in the Linux world. 'As systemd is already the preferred init system of most major Linux distributions today,' maybe of the distro's for their own reasons. The users on the other hand, may have a different opinion.Time will tell. I don't really see why the init system needs to be included at all in the 'container system' at all. Since the machine at that point is serving the images which surely can use their own init system in the chrooted environment. Unless of course after an image is downloaded the whole machine needs to be rebooted so both inits can work together.

Once again we see the boundaries being blurred by yet another 'feature'.