[gentoo-user] problems with performance when booted using systemd
Hi. I am having some strange performance problems when booted under systemd. These problems happened a little bit under openrc, but are much more pronounced with systemd. I am using just virtual consoles, no gui whatsoever at the moment. I also use tmux with 4 windows in one of the vcs. My system is an i7 processor, quod core and 16g of ram and 2g of swap space which appears not to be used. I am using uvesafb for the console, so I get 64x160 screens. The first problem is that if I don't press any keystrokes for several minutes and then want to move to another vc, it takes about 3 or 4 seconds after the alt-left arrow or alt-right arrow command to take effect. Even within the same vt, if I don't do anything for several minutes, it takes several seconds till the keystroke echoes and something happens. Once I have done this, things act normally, but its kind of annoying. Also, my load average seems to always be 1. I have looked at top and things seem to be OK, except that my cpu usage is like this: Tasks: 934 total, 2 running, 931 sleeping, 0 stopped, 1 zombie %Cpu(s): 12.5 us, 1.2 sy, 0.0 ni, 86.0 id, 0.2 wa, 0.0 hi, 0.0 si, 0.0 st KiB Mem: 16450248 total, 9678656 used, 6771592 free, 1084088 buffers KiB Swap: 2097148 total,4 used, 2097144 free. 1147688 cached Mem PID USER PR NIVIRTRESSHR S %CPU %MEM TIME+ COMMAND 9969 root 20 0 708 16 0 R 100.0 0.0 1549:10 v86d 579 root 30 10 0 0 0 S 9.1 0.0 16:09.93 speakup 11789 root 20 0 22524 2388 1116 R 0.7 0.0 0:00.03 top 7 root 0 -20 0 0 0 S 0.3 0.0 0:10.41 kworker/u:0H and onward ... This is an awful lot of tasks, I have never seen so many! Anyone have any ideas? Thanks much. -- Your life is like a penny. You're going to lose it. The question is: How do you spend it? John Covici cov...@ccs.covici.com
Re: [gentoo-user] systemd not shutting down all the way
On June 6, 2014 01:27:51 cov...@ccs.covici.com wrote: Hi. Whenever I issue the command shutdown -h now to shutdown the system, having booted under systemd, the computer never actually shuts down. It does stop some services, but eventually just sits there -- about the last one I see is something about sound card state and that is all, it just waits there forever. How can I trouble shoot this problem? There is no gui running before I issue the shutdown command. Thanks in advance for any suggestions. First, try 'systemctl halt' or 'systemctl poweroff', see if that works any better. I recall getting some warning on a previous systemd installation (no longer exists) that the way I was shutting down might not be working with systemd properly, though it wasn't a problem for me (I vaguely I recall I was using poweroff and switched to shutdown). Otherwise, you might want to try some of the debugging choices here: http://freedesktop.org/wiki/Software/systemd/Debugging/#diagnosingshutdownproblems Cheers, Bryan
Re: [gentoo-user] systemd not shutting down all the way
Bryan Gardiner b...@khumba.net wrote: On June 6, 2014 01:27:51 cov...@ccs.covici.com wrote: Hi. Whenever I issue the command shutdown -h now to shutdown the system, having booted under systemd, the computer never actually shuts down. It does stop some services, but eventually just sits there -- about the last one I see is something about sound card state and that is all, it just waits there forever. How can I trouble shoot this problem? There is no gui running before I issue the shutdown command. Thanks in advance for any suggestions. First, try 'systemctl halt' or 'systemctl poweroff', see if that works any better. I recall getting some warning on a previous systemd installation (no longer exists) that the way I was shutting down might not be working with systemd properly, though it wasn't a problem for me (I vaguely I recall I was using poweroff and switched to shutdown). Otherwise, you might want to try some of the debugging choices here: http://freedesktop.org/wiki/Software/systemd/Debugging/#diagnosingshutdownproblems Thanks, I will try those and see what I get. -- Your life is like a penny. You're going to lose it. The question is: How do you spend it? John Covici cov...@ccs.covici.com
Re: [gentoo-user] problems with performance when booted using systemd
On Friday, June 06, 2014 01:59:18 AM cov...@ccs.covici.com wrote: Hi. I am having some strange performance problems when booted under systemd. These problems happened a little bit under openrc, but are much more pronounced with systemd. I don't think it's necessarily systemd itself, just a setting that systemd does differently then openrc. See below for more. I am using just virtual consoles, no gui whatsoever at the moment. I also use tmux with 4 windows in one of the vcs. My system is an i7 processor, quod core and 16g of ram and 2g of swap space which appears not to be used. I am using uvesafb for the console, so I get 64x160 screens. Sounds similar to my laptop, except I run KDE and got 16g of swap (for hibernate) The first problem is that if I don't press any keystrokes for several minutes and then want to move to another vc, it takes about 3 or 4 seconds after the alt-left arrow or alt-right arrow command to take effect. Even within the same vt, if I don't do anything for several minutes, it takes several seconds till the keystroke echoes and something happens. Once I have done this, things act normally, but its kind of annoying. Sounds like a powersave setting. I used to get the same on my old laptop with spinning rust. SSDs tend to spin-up a lot quicker. Also, my load average seems to always be 1. I have looked at top and things seem to be OK, except that my cpu usage is like this: Tasks: 934 total, 2 running, 931 sleeping, 0 stopped, 1 zombie %Cpu(s): 12.5 us, 1.2 sy, 0.0 ni, 86.0 id, 0.2 wa, 0.0 hi, 0.0 si, 0.0 st KiB Mem: 16450248 total, 9678656 used, 6771592 free, 1084088 buffers KiB Swap: 2097148 total,4 used, 2097144 free. 1147688 cached Mem PID USER PR NIVIRTRESSHR S %CPU %MEM TIME+ COMMAND 9969 root 20 0 708 16 0 R 100.0 0.0 1549:10 v86d 579 root 30 10 0 0 0 S 9.1 0.0 16:09.93 speakup 11789 root 20 0 22524 2388 1116 R 0.7 0.0 0:00.03 top 7 root 0 -20 0 0 0 S 0.3 0.0 0:10.41 kworker/u:0H and onward ... This is an awful lot of tasks, I have never seen so many! That is a lot, I am currently running KDE, firefox and a citrix remote desktop thing. (oh, and skype and kopete and a few other items) KDE is installed with semantic-desktop, but the nepomuk stuff is disabled in system-settings. I have 200 tasks (yes, nice round figure) Anyone have any ideas? Thanks much. For the amount of tasks, check that you are not starting too many unneeded services. For the load-average of 1, shouldn't be too much of an issue, had similar in the past with a lot of stuff running and slow disks. For the freezing, I would suggest checking all the powersave options, especially the ones for the harddrives. Is there anything in the logs when this happens? Eg. check the logs right after the system becomes responsible again, maybe there is a hint there what is causing this. -- Joost
Re: [gentoo-user] problems with performance when booted using systemd
J. Roeleveld jo...@antarean.org wrote: On Friday, June 06, 2014 01:59:18 AM cov...@ccs.covici.com wrote: Hi. I am having some strange performance problems when booted under systemd. These problems happened a little bit under openrc, but are much more pronounced with systemd. I don't think it's necessarily systemd itself, just a setting that systemd does differently then openrc. See below for more. I am using just virtual consoles, no gui whatsoever at the moment. I also use tmux with 4 windows in one of the vcs. My system is an i7 processor, quod core and 16g of ram and 2g of swap space which appears not to be used. I am using uvesafb for the console, so I get 64x160 screens. Sounds similar to my laptop, except I run KDE and got 16g of swap (for hibernate) The first problem is that if I don't press any keystrokes for several minutes and then want to move to another vc, it takes about 3 or 4 seconds after the alt-left arrow or alt-right arrow command to take effect. Even within the same vt, if I don't do anything for several minutes, it takes several seconds till the keystroke echoes and something happens. Once I have done this, things act normally, but its kind of annoying. Sounds like a powersave setting. I used to get the same on my old laptop with spinning rust. SSDs tend to spin-up a lot quicker. Also, my load average seems to always be 1. I have looked at top and things seem to be OK, except that my cpu usage is like this: Tasks: 934 total, 2 running, 931 sleeping, 0 stopped, 1 zombie %Cpu(s): 12.5 us, 1.2 sy, 0.0 ni, 86.0 id, 0.2 wa, 0.0 hi, 0.0 si, 0.0 st KiB Mem: 16450248 total, 9678656 used, 6771592 free, 1084088 buffers KiB Swap: 2097148 total,4 used, 2097144 free. 1147688 cached Mem PID USER PR NIVIRTRESSHR S %CPU %MEM TIME+ COMMAND 9969 root 20 0 708 16 0 R 100.0 0.0 1549:10 v86d 579 root 30 10 0 0 0 S 9.1 0.0 16:09.93 speakup 11789 root 20 0 22524 2388 1116 R 0.7 0.0 0:00.03 top 7 root 0 -20 0 0 0 S 0.3 0.0 0:10.41 kworker/u:0H and onward ... This is an awful lot of tasks, I have never seen so many! That is a lot, I am currently running KDE, firefox and a citrix remote desktop thing. (oh, and skype and kopete and a few other items) KDE is installed with semantic-desktop, but the nepomuk stuff is disabled in system-settings. I have 200 tasks (yes, nice round figure) Anyone have any ideas? Thanks much. For the amount of tasks, check that you are not starting too many unneeded services. For the load-average of 1, shouldn't be too much of an issue, had similar in the past with a lot of stuff running and slow disks. For the freezing, I would suggest checking all the powersave options, especially the ones for the harddrives. Is there anything in the logs when this happens? Eg. check the logs right after the system becomes responsible again, maybe there is a hint there what is causing this. Unless systemd is setting some powersave options, I certainly never set anything like that, this is a desktop machine, not even a laptop. Next time this happens I will check the logs. Does systemd set some powersave options by default? Thanks. -- Your life is like a penny. You're going to lose it. The question is: How do you spend it? John Covici cov...@ccs.covici.com
Re: [gentoo-user] problems with performance when booted using systemd
On Friday, June 06, 2014 03:45:17 AM cov...@ccs.covici.com wrote: J. Roeleveld jo...@antarean.org wrote: On Friday, June 06, 2014 01:59:18 AM cov...@ccs.covici.com wrote: Hi. I am having some strange performance problems when booted under systemd. These problems happened a little bit under openrc, but are much more pronounced with systemd. I don't think it's necessarily systemd itself, just a setting that systemd does differently then openrc. See below for more. I am using just virtual consoles, no gui whatsoever at the moment. I also use tmux with 4 windows in one of the vcs. My system is an i7 processor, quod core and 16g of ram and 2g of swap space which appears not to be used. I am using uvesafb for the console, so I get 64x160 screens. Sounds similar to my laptop, except I run KDE and got 16g of swap (for hibernate) The first problem is that if I don't press any keystrokes for several minutes and then want to move to another vc, it takes about 3 or 4 seconds after the alt-left arrow or alt-right arrow command to take effect. Even within the same vt, if I don't do anything for several minutes, it takes several seconds till the keystroke echoes and something happens. Once I have done this, things act normally, but its kind of annoying. Sounds like a powersave setting. I used to get the same on my old laptop with spinning rust. SSDs tend to spin-up a lot quicker. Also, my load average seems to always be 1. I have looked at top and things seem to be OK, except that my cpu usage is like this: Tasks: 934 total, 2 running, 931 sleeping, 0 stopped, 1 zombie %Cpu(s): 12.5 us, 1.2 sy, 0.0 ni, 86.0 id, 0.2 wa, 0.0 hi, 0.0 si, 0.0 st KiB Mem: 16450248 total, 9678656 used, 6771592 free, 1084088 buffers KiB Swap: 2097148 total,4 used, 2097144 free. 1147688 cached Mem PID USER PR NIVIRTRESSHR S %CPU %MEM TIME+ COMMAND 9969 root 20 0 708 16 0 R 100.0 0.0 1549:10 v86d 579 root 30 10 0 0 0 S 9.1 0.0 16:09.93 speakup 11789 root 20 0 22524 2388 1116 R 0.7 0.0 0:00.03 top 7 root 0 -20 0 0 0 S 0.3 0.0 0:10.41 kworker/u:0H and onward ... This is an awful lot of tasks, I have never seen so many! That is a lot, I am currently running KDE, firefox and a citrix remote desktop thing. (oh, and skype and kopete and a few other items) KDE is installed with semantic-desktop, but the nepomuk stuff is disabled in system-settings. I have 200 tasks (yes, nice round figure) Anyone have any ideas? Thanks much. For the amount of tasks, check that you are not starting too many unneeded services. For the load-average of 1, shouldn't be too much of an issue, had similar in the past with a lot of stuff running and slow disks. For the freezing, I would suggest checking all the powersave options, especially the ones for the harddrives. Is there anything in the logs when this happens? Eg. check the logs right after the system becomes responsible again, maybe there is a hint there what is causing this. Unless systemd is setting some powersave options, I certainly never set anything like that, this is a desktop machine, not even a laptop. Next time this happens I will check the logs. Does systemd set some powersave options by default? I do not know that for sure, best wait for more knowledgable systemd users to answer that. If it doesn't, then systemd itself is causing more freezes (as per your experience) then openrc. I would guess it does or at least with the default configuration. What you describe makes me think the disks are switched to powersave sooner with systemd. Can you provide the output of the following command: # hdparm -B /dev/sda to get the APM settings of the disk. (If you have multiple disks, please run it for the others as well. Question for others as well, how do you get the current setting for the spindown timeout set with hdparm -S value device ? I couldn't find it. I am happy with openrc and have no intention on switching to systemd as I haven't heard of a single feature that would actually make my life easier. -- Joost
Re: [gentoo-user] problems with performance when booted using systemd
J. Roeleveld jo...@antarean.org wrote: On Friday, June 06, 2014 01:59:18 AM cov...@ccs.covici.com wrote: Hi. I am having some strange performance problems when booted under systemd. These problems happened a little bit under openrc, but are much more pronounced with systemd. I don't think it's necessarily systemd itself, just a setting that systemd does differently then openrc. See below for more. I am using just virtual consoles, no gui whatsoever at the moment. I also use tmux with 4 windows in one of the vcs. My system is an i7 processor, quod core and 16g of ram and 2g of swap space which appears not to be used. I am using uvesafb for the console, so I get 64x160 screens. Sounds similar to my laptop, except I run KDE and got 16g of swap (for hibernate) The first problem is that if I don't press any keystrokes for several minutes and then want to move to another vc, it takes about 3 or 4 seconds after the alt-left arrow or alt-right arrow command to take effect. Even within the same vt, if I don't do anything for several minutes, it takes several seconds till the keystroke echoes and something happens. Once I have done this, things act normally, but its kind of annoying. Sounds like a powersave setting. I used to get the same on my old laptop with spinning rust. SSDs tend to spin-up a lot quicker. Also, my load average seems to always be 1. I have looked at top and things seem to be OK, except that my cpu usage is like this: Tasks: 934 total, 2 running, 931 sleeping, 0 stopped, 1 zombie %Cpu(s): 12.5 us, 1.2 sy, 0.0 ni, 86.0 id, 0.2 wa, 0.0 hi, 0.0 si, 0.0 st KiB Mem: 16450248 total, 9678656 used, 6771592 free, 1084088 buffers KiB Swap: 2097148 total,4 used, 2097144 free. 1147688 cached Mem PID USER PR NIVIRTRESSHR S %CPU %MEM TIME+ COMMAND 9969 root 20 0 708 16 0 R 100.0 0.0 1549:10 v86d 579 root 30 10 0 0 0 S 9.1 0.0 16:09.93 speakup 11789 root 20 0 22524 2388 1116 R 0.7 0.0 0:00.03 top 7 root 0 -20 0 0 0 S 0.3 0.0 0:10.41 kworker/u:0H and onward ... This is an awful lot of tasks, I have never seen so many! That is a lot, I am currently running KDE, firefox and a citrix remote desktop thing. (oh, and skype and kopete and a few other items) KDE is installed with semantic-desktop, but the nepomuk stuff is disabled in system-settings. I have 200 tasks (yes, nice round figure) Anyone have any ideas? Thanks much. For the amount of tasks, check that you are not starting too many unneeded services. For the load-average of 1, shouldn't be too much of an issue, had similar in the past with a lot of stuff running and slow disks. For the freezing, I would suggest checking all the powersave options, especially the ones for the harddrives. Is there anything in the logs when this happens? Eg. check the logs right after the system becomes responsible again, maybe there is a hint there what is causing this. Well, not a peep out of the logs -- I did journaldctl -rb right after one of those pauses and not an entry -- in fact it looked like mail was being received, etc right along. -- Your life is like a penny. You're going to lose it. The question is: How do you spend it? John Covici cov...@ccs.covici.com
Re: [gentoo-user] problems with performance when booted using systemd
J. Roeleveld jo...@antarean.org wrote: On Friday, June 06, 2014 03:45:17 AM cov...@ccs.covici.com wrote: J. Roeleveld jo...@antarean.org wrote: On Friday, June 06, 2014 01:59:18 AM cov...@ccs.covici.com wrote: Hi. I am having some strange performance problems when booted under systemd. These problems happened a little bit under openrc, but are much more pronounced with systemd. I don't think it's necessarily systemd itself, just a setting that systemd does differently then openrc. See below for more. I am using just virtual consoles, no gui whatsoever at the moment. I also use tmux with 4 windows in one of the vcs. My system is an i7 processor, quod core and 16g of ram and 2g of swap space which appears not to be used. I am using uvesafb for the console, so I get 64x160 screens. Sounds similar to my laptop, except I run KDE and got 16g of swap (for hibernate) The first problem is that if I don't press any keystrokes for several minutes and then want to move to another vc, it takes about 3 or 4 seconds after the alt-left arrow or alt-right arrow command to take effect. Even within the same vt, if I don't do anything for several minutes, it takes several seconds till the keystroke echoes and something happens. Once I have done this, things act normally, but its kind of annoying. Sounds like a powersave setting. I used to get the same on my old laptop with spinning rust. SSDs tend to spin-up a lot quicker. Also, my load average seems to always be 1. I have looked at top and things seem to be OK, except that my cpu usage is like this: Tasks: 934 total, 2 running, 931 sleeping, 0 stopped, 1 zombie %Cpu(s): 12.5 us, 1.2 sy, 0.0 ni, 86.0 id, 0.2 wa, 0.0 hi, 0.0 si, 0.0 st KiB Mem: 16450248 total, 9678656 used, 6771592 free, 1084088 buffers KiB Swap: 2097148 total,4 used, 2097144 free. 1147688 cached Mem PID USER PR NIVIRTRESSHR S %CPU %MEM TIME+ COMMAND 9969 root 20 0 708 16 0 R 100.0 0.0 1549:10 v86d 579 root 30 10 0 0 0 S 9.1 0.0 16:09.93 speakup 11789 root 20 0 22524 2388 1116 R 0.7 0.0 0:00.03 top 7 root 0 -20 0 0 0 S 0.3 0.0 0:10.41 kworker/u:0H and onward ... This is an awful lot of tasks, I have never seen so many! That is a lot, I am currently running KDE, firefox and a citrix remote desktop thing. (oh, and skype and kopete and a few other items) KDE is installed with semantic-desktop, but the nepomuk stuff is disabled in system-settings. I have 200 tasks (yes, nice round figure) Anyone have any ideas? Thanks much. For the amount of tasks, check that you are not starting too many unneeded services. For the load-average of 1, shouldn't be too much of an issue, had similar in the past with a lot of stuff running and slow disks. For the freezing, I would suggest checking all the powersave options, especially the ones for the harddrives. Is there anything in the logs when this happens? Eg. check the logs right after the system becomes responsible again, maybe there is a hint there what is causing this. Unless systemd is setting some powersave options, I certainly never set anything like that, this is a desktop machine, not even a laptop. Next time this happens I will check the logs. Does systemd set some powersave options by default? I do not know that for sure, best wait for more knowledgable systemd users to answer that. If it doesn't, then systemd itself is causing more freezes (as per your experience) then openrc. I would guess it does or at least with the default configuration. What you describe makes me think the disks are switched to powersave sooner with systemd. Can you provide the output of the following command: # hdparm -B /dev/sda to get the APM settings of the disk. (If you have multiple disks, please run it for the others as well. Question for others as well, how do you get the current setting for the spindown timeout set with hdparm -S value device ? I couldn't find it. I am happy with openrc and have no intention on switching to systemd as I haven't heard of a single feature that would actually make my life easier. I don't have hdparm on the system, is it only for older disks? If memory serves, it did not work at all when I tried it as my disks are all /dev/sda, etc, but that may be wrong. -- Your life is like a penny. You're going to lose it. The question is: How do you spend it? John Covici cov...@ccs.covici.com
Re: [gentoo-user] problems with performance when booted using systemd
On Friday 06 Jun 2014 06:59:18 cov...@ccs.covici.com wrote: I am using uvesafb for the console, so I get 64x160 screens. Why don't you use KMS? I am asking in the off-chance that uvesa is not working happily with your video card and the native kernel drive performs better. This of course would not explain the high number of tasks; are these only evident under systemd? -- Regards, Mick signature.asc Description: This is a digitally signed message part.
Re: [gentoo-user] problems with performance when booted using systemd
Mick michaelkintz...@gmail.com wrote: On Friday 06 Jun 2014 06:59:18 cov...@ccs.covici.com wrote: I am using uvesafb for the console, so I get 64x160 screens. Why don't you use KMS? I am asking in the off-chance that uvesa is not working happily with your video card and the native kernel drive performs better. This of course would not explain the high number of tasks; are these only evident under systemd? Under openrc, I get much fewer tasks and even when I first booted systemd, it was fewer, but it was maybe be 400 whereas openrc had 200 or less. I am using uvesafb, because when I tried the kernel driver, nvidia closed source driver was not happy when X was started. -- Your life is like a penny. You're going to lose it. The question is: How do you spend it? John Covici cov...@ccs.covici.com
Re: [gentoo-user] Use Flags and Updating
On Thu, May 22, 2014 at 3:54 AM, Marc Joliet mar...@gmx.de wrote: I think nowadays one would prefer --keep-going, which automatically resumes on failure (and recomputes the dependency tree!), and prints a list of failed packages when it's finished. However its output is more verbose than just ok and failed (it'll print the build.log if it's only one package, IIRC). Hmm, after using this script for some time I found a problem with this approach. If you use --buildpkgonly and ---keep-going then emerge won't build a single thing if anything in the list is missing a build-time dependency. The script I posted will try to emerge everything individually so at least some of the packages will be compiled. That seems like a bug in --buildpkgonly. If you use it with --keep-going it should at least compile the packages that aren't missing build-time options. I'll file that as a bug if it isn't already there... Rich
Re: [gentoo-user] problems with performance when booted using systemd
On Friday, June 06, 2014 04:46:35 AM cov...@ccs.covici.com wrote: J. Roeleveld jo...@antarean.org wrote: On Friday, June 06, 2014 03:45:17 AM cov...@ccs.covici.com wrote: J. Roeleveld jo...@antarean.org wrote: On Friday, June 06, 2014 01:59:18 AM cov...@ccs.covici.com wrote: Hi. I am having some strange performance problems when booted under systemd. These problems happened a little bit under openrc, but are much more pronounced with systemd. I don't think it's necessarily systemd itself, just a setting that systemd does differently then openrc. See below for more. I am using just virtual consoles, no gui whatsoever at the moment. I also use tmux with 4 windows in one of the vcs. My system is an i7 processor, quod core and 16g of ram and 2g of swap space which appears not to be used. I am using uvesafb for the console, so I get 64x160 screens. Sounds similar to my laptop, except I run KDE and got 16g of swap (for hibernate) The first problem is that if I don't press any keystrokes for several minutes and then want to move to another vc, it takes about 3 or 4 seconds after the alt-left arrow or alt-right arrow command to take effect. Even within the same vt, if I don't do anything for several minutes, it takes several seconds till the keystroke echoes and something happens. Once I have done this, things act normally, but its kind of annoying. Sounds like a powersave setting. I used to get the same on my old laptop with spinning rust. SSDs tend to spin-up a lot quicker. Also, my load average seems to always be 1. I have looked at top and things seem to be OK, except that my cpu usage is like this: Tasks: 934 total, 2 running, 931 sleeping, 0 stopped, 1 zombie %Cpu(s): 12.5 us, 1.2 sy, 0.0 ni, 86.0 id, 0.2 wa, 0.0 hi, 0.0 si, 0.0 st KiB Mem: 16450248 total, 9678656 used, 6771592 free, 1084088 buffers KiB Swap: 2097148 total,4 used, 2097144 free. 1147688 cached Mem PID USER PR NIVIRTRESSHR S %CPU %MEM TIME+ COMMAND 9969 root 20 0 708 16 0 R 100.0 0.0 1549:10 v86d 579 root 30 10 0 0 0 S 9.1 0.0 16:09.93 speakup 11789 root 20 0 22524 2388 1116 R 0.7 0.0 0:00.03 top 7 root 0 -20 0 0 0 S 0.3 0.0 0:10.41 kworker/u:0H and onward ... This is an awful lot of tasks, I have never seen so many! That is a lot, I am currently running KDE, firefox and a citrix remote desktop thing. (oh, and skype and kopete and a few other items) KDE is installed with semantic-desktop, but the nepomuk stuff is disabled in system-settings. I have 200 tasks (yes, nice round figure) Anyone have any ideas? Thanks much. For the amount of tasks, check that you are not starting too many unneeded services. For the load-average of 1, shouldn't be too much of an issue, had similar in the past with a lot of stuff running and slow disks. For the freezing, I would suggest checking all the powersave options, especially the ones for the harddrives. Is there anything in the logs when this happens? Eg. check the logs right after the system becomes responsible again, maybe there is a hint there what is causing this. Unless systemd is setting some powersave options, I certainly never set anything like that, this is a desktop machine, not even a laptop. Next time this happens I will check the logs. Does systemd set some powersave options by default? I do not know that for sure, best wait for more knowledgable systemd users to answer that. If it doesn't, then systemd itself is causing more freezes (as per your experience) then openrc. I would guess it does or at least with the default configuration. What you describe makes me think the disks are switched to powersave sooner with systemd. Can you provide the output of the following command: # hdparm -B /dev/sda to get the APM settings of the disk. (If you have multiple disks, please run it for the others as well. Question for others as well, how do you get the current setting for the spindown timeout set with hdparm -S value device ? I couldn't find it. I am happy with openrc and have no intention on switching to systemd as I haven't heard of a single feature that would actually make my life easier. I don't have hdparm on the system, is it only for older disks? If memory serves, it did not work at all when I tried it as my disks are all /dev/sda, etc, but that may be wrong. It also works on new SATA drives and SSDs: # smartctl -a /dev/sda
Re: [gentoo-user] Re: Systemd upower
On Fri, Jun 6, 2014 at 1:46 AM, Mick michaelkintz...@gmail.com wrote: On Friday 06 Jun 2014 00:15:02 Peter Humphrey wrote: I bet you have quite a lot of systemd components lurking in the background though, ready to take over the world the next time you aren't looking :-) Ha! I can already see this one: 338 ?Ss 0:00 /lib/systemd/systemd-udevd --daemon I have set USE=-systemd, but if/when Gentoo migrates to systemd as the default startup I will probably have to remove it and then learn how to use systemd. That would be udev. It has been around long before systemd, and you must have missed the huge flamewar when they renamed it to systemd-udevd. Maybe we'll see java renamed to java-by-oracle-with-ask-toolbar next. :) If you ever migrate to systemd you really just need to set USE=systemd and install systemd. Portage will swap out your udev in the process, though nothing there will really change as systemd and udev install the same udev components. There is a guide for installing systemd that you should follow which gets into all the details. Rich
Re: [gentoo-user] problems with performance when booted using systemd
On Fri, Jun 06, 2014 at 05:19:58AM -0400, cov...@ccs.covici.com wrote I am using uvesafb, because when I tried the kernel driver, nvidia closed source driver was not happy when X was started. Have you tried the nouveau open-source drivers for Nvidia cards? You don't have to rebuild/upgrade the video driver every time you upgrade your kernel. -- Walter Dnes waltd...@waltdnes.org I don't run desktop environments; I run useful applications
Re: [gentoo-user] problems with performance when booted using systemd
Walter Dnes waltd...@waltdnes.org wrote: On Fri, Jun 06, 2014 at 05:19:58AM -0400, cov...@ccs.covici.com wrote I am using uvesafb, because when I tried the kernel driver, nvidia closed source driver was not happy when X was started. Have you tried the nouveau open-source drivers for Nvidia cards? You don't have to rebuild/upgrade the video driver every time you upgrade your kernel. I did try those, but if X was started, nvidia crapped out. Also, I was getting a lot less screen real estate with those drivers than with uvesafb. -- Your life is like a penny. You're going to lose it. The question is: How do you spend it? John Covici cov...@ccs.covici.com
[gentoo-user] upower suddenly demands systemd
After today's emerge-webrsync, I have found out that usual # emerge --update --deep --with-bdeps=y --newuse --ask world does not work, trying but being unable to emerge systemd. As I have found out, the reason for it was that upower suddenly decided that it needs systemd but, for some (lucky :) reason, it could not be emerged as my system uses OpenRC. The latest news said the following: UPower discontinued hibernate and suspend support in favor of systemd. Because of this, we have created a compability package at sys-power/upower-pm-utils which will give you the old UPower with sys-power/pm-utils support back. Some desktops have integrated the sys-power/pm-utils support directly to their code, like Xfce, and as a result, they work also with the new UPower as expected. All non-systemd users are recommended to choose between: # emerge --oneshot --noreplace 'sys-power/upower-pm-utils' or # emerge --oneshot --noreplace '=sys-power/upower-0.99.0' However, all systemd users are recommended to stay with sys-power/upower. However, that news did *not* say that without # emerge --oneshot --noreplace 'sys-power/upower-pm-utils' it is impossible to update the system even if you use xfce. Only after executing the last command, which installed upower-pm-utils and unistalled upower, the # emerge --update --deep --with-bdeps=y --newuse --ask world worked as desired. upower was not in my world file, so I think it's a some kind of a bug.
[gentoo-user] dev-ruby/json-1.8.0
Hi all
I am trying to emerge dev-ruby/json-1.8.0 and it keeps failing.
I have tried everything I know to fix it, without any success.
Regards
* Package:dev-ruby/json-1.8.0
* Repository: gentoo
* Maintainer: r...@gentoo.org
* USE:amd64 doc elibc_glibc kernel_linux ruby_targets_ruby19
ruby_targets_ruby20 userland_GNU
* FEATURES: preserve-libs sandbox userpriv usersandbox
Unpacking source...
* Running unpack phase for all ...
* Unpacking .gem file...
...
[ ok ]
* Uncompressing metadata
...
[ ok ]
* Unpacking data.tar.gz
...
[ ok ]
Source unpacked in /var/tmp/portage/dev-ruby/json-1.8.0/work
Preparing source in /var/tmp/portage/dev-ruby/json-1.8.0/work ...
* Running prepare phase for all ...
* Running source copy phase for ruby19 ...
* Running source copy phase for ruby20 ...
Source prepared.
Configuring source in /var/tmp/portage/dev-ruby/json-1.8.0/work ...
Source configured.
Compiling source in /var/tmp/portage/dev-ruby/json-1.8.0/work ...
* Running compile phase for ruby19 ...
GNU Make 3.82
Built for x86_64-pc-linux-gnu
Copyright (C) 2010 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later http://gnu.org/licenses/gpl.html
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
Ragel State Machine Compiler version 6.7 May 2011
Copyright (c) 2001-2009 by Adrian Thurston
Ragel State Machine Compiler version 6.7 May 2011
Copyright (c) 2001-2009 by Adrian Thurston
cd ext/json/ext/parser
/usr/bin/ruby19 extconf.rb
creating Makefile
gmake
compiling parser.c
linking shared-object json/ext/parser.so
cd -
cp ext/json/ext/parser/parser.so ext/json/ext
cd ext/json/ext/generator
/usr/bin/ruby19 extconf.rb
creating Makefile
gmake
compiling generator.c
linking shared-object json/ext/generator.so
cd -
cp ext/json/ext/generator/generator.so ext/json/ext
* Running compile phase for ruby20 ...
GNU Make 3.82
Built for x86_64-pc-linux-gnu
Copyright (C) 2010 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later http://gnu.org/licenses/gpl.html
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
Ragel State Machine Compiler version 6.7 May 2011
Copyright (c) 2001-2009 by Adrian Thurston
Ragel State Machine Compiler version 6.7 May 2011
Copyright (c) 2001-2009 by Adrian Thurston
cd ext/json/ext/parser
/usr/bin/ruby20 extconf.rb
creating Makefile
gmake
compiling parser.c
linking shared-object json/ext/parser.so
cd -
cp ext/json/ext/parser/parser.so ext/json/ext
cd ext/json/ext/generator
/usr/bin/ruby20 extconf.rb
creating Makefile
gmake
compiling generator.c
linking shared-object json/ext/generator.so
cd -
cp ext/json/ext/generator/generator.so ext/json/ext
* Running compile phase for all ...
GNU Make 3.82
Built for x86_64-pc-linux-gnu
Copyright (C) 2010 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later http://gnu.org/licenses/gpl.html
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
Ragel State Machine Compiler version 6.7 May 2011
Copyright (c) 2001-2009 by Adrian Thurston
Ragel State Machine Compiler version 6.7 May 2011
Copyright (c) 2001-2009 by Adrian Thurston
Writing version information for 1.8.0
rdoc -o doc -t 'JSON Implementation for Ruby' -m README.rdoc README.rdoc
lib/json.rb lib/json/add/bigdecimal.rb lib/json/add/complex.rb
lib/json/add/core.rb lib/json/add/date.rb lib/json/add/date_time.rb
lib/json/add/exception.rb lib/json/add/ostruct.rb lib/json/add/range.rb
lib/json/add/rational.rb lib/json/add/regexp.rb lib/json/add/struct.rb
lib/json/add/symbol.rb lib/json/add/time.rb lib/json/common.rb
lib/json/ext.rb lib/json/generic_object.rb lib/json/pure.rb
lib/json/pure/generator.rb lib/json/pure/parser.rb lib/json/version.rb
ext/json/ext/parser/parser.c ext/json/ext/generator/generator.c
sh: rdoc: command not found
rake aborted!
Command failed with status (127): [rdoc -o doc -t 'JSON Implementation for
Ru...]
/var/tmp/portage/dev-ruby/json-1.8.0/work/all/json-1.8.0/Rakefile:346:in
`block in top (required)'
Tasks: TOP = doc
(See full trace by running task with --trace)
* ERROR: dev-ruby/json-1.8.0::gentoo failed (compile phase):
* failed to (re)build documentation
*
* Call stack:
* ebuild.sh, line 93: Called src_compile
* environment, line 4281: Called ruby-ng_src_compile
* environment, line 3939: Called _ruby_invoke_environment 'all'
'all_ruby_compile'
* environment, line 501: Called all_ruby_compile
* environment, line 585: Called all_fakegem_compile
* environment, line 545: Called die
* The specific snippet of code:
* rake ${RUBY_FAKEGEM_TASK_DOC} || die failed to
(re)build documentation
*
* If you need support, post the output of `emerge --info
Re: [gentoo-user] upower suddenly demands systemd
On 06/06/2014 05:34 PM, Gevisz wrote: After today's emerge-webrsync, I have found out that usual # emerge --update --deep --with-bdeps=y --newuse --ask world does not work, trying but being unable to emerge systemd. As I have found out, the reason for it was that upower suddenly decided that it needs systemd but, for some (lucky :) reason, it could not be emerged as my system uses OpenRC. The latest news said the following: UPower discontinued hibernate and suspend support in favor of systemd. Because of this, we have created a compability package at sys-power/upower-pm-utils which will give you the old UPower with sys-power/pm-utils support back. Some desktops have integrated the sys-power/pm-utils support directly to their code, like Xfce, and as a result, they work also with the new UPower as expected. All non-systemd users are recommended to choose between: # emerge --oneshot --noreplace 'sys-power/upower-pm-utils' or # emerge --oneshot --noreplace '=sys-power/upower-0.99.0' However, all systemd users are recommended to stay with sys-power/upower. However, that news did *not* say that without # emerge --oneshot --noreplace 'sys-power/upower-pm-utils' it is impossible to update the system even if you use xfce. Only after executing the last command, which installed upower-pm-utils and unistalled upower, the # emerge --update --deep --with-bdeps=y --newuse --ask world worked as desired. upower was not in my world file, so I think it's a some kind of a bug. The situation you've encountered has been discussed at length on this mailing list over the last few days. Please see the threads with upower and systemd as the Subject matter for further details.
[gentoo-user] Re: quick installs on older/embedded hardware
Alan McKinnon alan.mckinnon at gmail.com writes: So how do other vendors do it? If we look at their workflow, perhaps a useable method will filter up through the wetware I have no idea how to find and convince an appropriate company to share how they setup for multi image testing on various hardware platforms. If you know of such an opportunity, I'd be all ears on what they would be willing to share. Do tell. James
Re: [gentoo-user] OT: Mapping random numbers (PRNG)
Canek Peláez Valdés can...@gmail.com [14-06-06 17:36]:
On Thu, Jun 5, 2014 at 9:56 PM, meino.cra...@gmx.de wrote:
Hi,
I am experimenting with the C code of the ISAAC pseudo random number
generator
(http://burtleburtle.net/bob/rand/isaacafa.html).
Currently the implementation creates (on my embedded linux) 32 bit
hexadecimal output.
So it's a 32 bit integer.
From this I want to create random numbers in the range of [a-Za-z0-9]
*without violating randomness* and (if possible) without throwing
away bits of the output.
You mean *characters* int the range [A-Za-z0-9]?
How can I do this mathemtically (in concern of the quality of output)
correct?
The easiest thing to do would be:
---
#include time.h
#include stdio.h
#include stdlib.h
#define N (26+26+10)
static char S[] = { 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J',
'K', 'L', 'M',
'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W',
'X', 'Y', 'Z',
'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j',
'k', 'l', 'm',
'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w',
'x', 'y', 'z',
'0', '1', '2', '3', '4', '5', '6', '7', '8', '9' };
int
next_character()
{
// Use the correct call for ISAAC instead of rand()
unsigned int idx = rand() % N;
return S[idx];
}
int
main(int argc, char* argv[])
{
// Use the correct call for initializing the ISAAC seed
srand((unsigned int)time(NULL));
for (int i = 0; i 20; i++) // --std=c99
printf(%c\n, next_character());
return 0;
}
---
If the ISAAC RNG has a good distribution, then the next_character()
function will give a good distribution among the set [A-Za-z0-9].
Unless I missunderstood what you meant with create random numbers in
the range of [a-Za-z0-9].
Regards.
--
Canek Peláez Valdés
Profesor de asignatura, Facultad de Ciencias
Universidad Nacional Autónoma de México
Hi,
Thank you very much for the input! :)
I have a question about the algorithm:
Suppose rand() has an equal distribution of numbers and furthermore
one has a count of 2^32 random numbers listed in numerical sort
order.
In this list each number would appear (nearly) with the same count: 1
To get an better imagination of that...suppose the rand() would only
return numbers in the range of 1...12 and the alphabet has only 8
characters (as 2^32 is not devideable by 62)
rand():
1 2 3 4 5 6 7 8 9 10 11 12
rand()%N : rand()%7
1 2 3 4 5 6 7 0 1 2 3 4
or in other words: An even distribution of numbers of rand()
would result in a unevenly distributed sequence of characters...or?
This would break the quality of ISAACs output.
I am sure I did something wrong here...but where is the logic trap?
Thank you very much for any help in advance!
Best regards,
mcc
Re: [gentoo-user] Re: Systemd upower
On Friday 06 Jun 2014 12:18:09 Rich Freeman wrote: That would be udev. It has been around long before systemd, and you must have missed the huge flamewar when they renamed it to systemd-udevd. Maybe we'll see java renamed to java-by-oracle-with-ask-toolbar next. :) TBH I wouldn't be surprised. At least java offers a choice of avoiding it. ;-) If you ever migrate to systemd you really just need to set USE=systemd and install systemd. Portage will swap out your udev in the process, though nothing there will really change as systemd and udev install the same udev components. There is a guide for installing systemd that you should follow which gets into all the details. I didn't miss the flamewar. Actually I recall joining in the fun and posting the odd message about systemd. I know that I could use eudev or systemd-udev (or even mdev as kindly shared in this list by Walter). I am mostly happy with openrc and therefore have no reason to move to the systemd monoculture, unless gentoo falls in line with Debian et al. and leaves me no choice. -- Regards, Mick signature.asc Description: This is a digitally signed message part.
Re: [gentoo-user] Re: Systemd upower
On Fri, Jun 6, 2014 at 3:13 PM, Mick michaelkintz...@gmail.com wrote: I am mostly happy with openrc and therefore have no reason to move to the systemd monoculture, unless gentoo falls in line with Debian et al. and leaves me no choice. I don't really see that happening anytime soon - it will be more likely to become an issue for the more complex desktop environments (Gnome is already going this way - KDE may very well go this way later). Historically they're the first packages to require things like HAL, udev, dbus, pulseaudio, etc (and on Gentoo the maintainers tend to do a good job of minimizing dependencies). I think many will switch to systemd anyway, simply because that is the way the wind is blowing and it does have some benefits depending on your situation (but so do a number of other configurations). I tend to use it by default on new installs, and anytime I find myself tweaking my monit rules I keep bumping up migrating entirely to systemd a little higher on my to-do list. Rich
Re: [gentoo-user] OT: Mapping random numbers (PRNG)
On 06/06/14 20:39, meino.cra...@gmx.de wrote: To get an better imagination of that...suppose the rand() would only return numbers in the range of 1...12 and the alphabet has only 8 characters (as 2^32 is not devideable by 62) rand(): 1 2 3 4 5 6 7 8 9 10 11 12 rand()%N : rand()%7 1 2 3 4 5 6 7 0 1 2 3 4 or in other words: An even distribution of numbers of rand() would result in a unevenly distributed sequence of characters...or? This would break the quality of ISAACs output. I am sure I did something wrong here...but where is the logic trap? You thought on that is totally right. Let's say random numbers have the rande [0..RAND_MAX] Blindly calculating modulus creates a bias towards lower numbers if RAND_MAX+1 is not divisible by N. For most applications this bias is negligible. If you really want the same distribution you have to discard numbers greater than RAND_MAX - ((RAND_MAX+1)%N) (Beware the undefined behaviour in the above line if you try C.) Basically you want one less than the largest by N dividable number in your range. With numbers: rand(): 0 1 2 3 4 5 6 7 8 9 10 11 12 rand()%5 with discarding 12-((12+1)%5)=9: 0 1 2 3 4 0 1 2 3 4 discarded --- null_ptr Please don't follow me.
Re: [gentoo-user] OT: Mapping random numbers (PRNG)
On Fri, Jun 6, 2014 at 1:39 PM, meino.cra...@gmx.de wrote:
Canek Peláez Valdés can...@gmail.com [14-06-06 17:36]:
On Thu, Jun 5, 2014 at 9:56 PM, meino.cra...@gmx.de wrote:
Hi,
I am experimenting with the C code of the ISAAC pseudo random number
generator
(http://burtleburtle.net/bob/rand/isaacafa.html).
Currently the implementation creates (on my embedded linux) 32 bit
hexadecimal output.
So it's a 32 bit integer.
From this I want to create random numbers in the range of [a-Za-z0-9]
*without violating randomness* and (if possible) without throwing
away bits of the output.
You mean *characters* int the range [A-Za-z0-9]?
How can I do this mathemtically (in concern of the quality of output)
correct?
The easiest thing to do would be:
---
#include time.h
#include stdio.h
#include stdlib.h
#define N (26+26+10)
static char S[] = { 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J',
'K', 'L', 'M',
'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W',
'X', 'Y', 'Z',
'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j',
'k', 'l', 'm',
'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w',
'x', 'y', 'z',
'0', '1', '2', '3', '4', '5', '6', '7', '8', '9' };
int
next_character()
{
// Use the correct call for ISAAC instead of rand()
unsigned int idx = rand() % N;
return S[idx];
}
int
main(int argc, char* argv[])
{
// Use the correct call for initializing the ISAAC seed
srand((unsigned int)time(NULL));
for (int i = 0; i 20; i++) // --std=c99
printf(%c\n, next_character());
return 0;
}
---
If the ISAAC RNG has a good distribution, then the next_character()
function will give a good distribution among the set [A-Za-z0-9].
Unless I missunderstood what you meant with create random numbers in
the range of [a-Za-z0-9].
Regards.
--
Canek Peláez Valdés
Profesor de asignatura, Facultad de Ciencias
Universidad Nacional Autónoma de México
Hi,
Thank you very much for the input! :)
I have a question about the algorithm:
Suppose rand() has an equal distribution of numbers and furthermore
one has a count of 2^32 random numbers listed in numerical sort
order.
In this list each number would appear (nearly) with the same count: 1
To get an better imagination of that...suppose the rand() would only
return numbers in the range of 1...12 and the alphabet has only 8
characters (as 2^32 is not devideable by 62)
rand():
1 2 3 4 5 6 7 8 9 10 11 12
rand()%N : rand()%7
1 2 3 4 5 6 7 0 1 2 3 4
or in other words: An even distribution of numbers of rand()
would result in a unevenly distributed sequence of characters...or?
This would break the quality of ISAACs output.
I am sure I did something wrong here...but where is the logic trap?
In theory, it doesn't matter if the number of random characters you
want is not multiple of the size of the set of characters. In
practice, it doesn't matter if the number of random characters you
want is big enough.
Consider the following modification to my program (I changed the order
of the array S so it's sorted and I can do binary search):
--
#include time.h
#include stdio.h
#include stdlib.h
#define N (10+26+26)
static char S[] = {
'0', '1', '2', '3', '4', '5', '6', '7', '8', '9',
'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M',
'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z',
'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm',
'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z'
};
static int
index_of_aux(char c, int a, int b)
{
if (a b)
return -1;
int m = (a + b) / 2;
if (S[m] == c)
return m;
if (S[m] c)
return index_of_aux(c, a, m-1);
return index_of_aux(c, m+1, b);
}
static int
index_of(char c)
{
return index_of_aux(c, 0, N-1);
}
int
next_character()
{
// Use the correct call for isaac instead of rand()
unsigned int idx = rand() % N;
return S[idx];
}
int
main(int argc, char* argv[])
{
srand((unsigned int)time(NULL));
int total = 100;
// Size = 62
int count[] = {
0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
};
for (int i = 0; i total; i++) {
char c = next_character();
count[index_of(c)]++;
}
int min =
Re: [gentoo-user] OT: Mapping random numbers (PRNG)
On Thu, Jun 05, 2014 at 10:58:51PM -0500, Canek Peláez Valdés wrote:
On Thu, Jun 5, 2014 at 9:56 PM, meino.cra...@gmx.de wrote:
Hi,
I am experimenting with the C code of the ISAAC pseudo random number
generator
(http://burtleburtle.net/bob/rand/isaacafa.html).
Currently the implementation creates (on my embedded linux) 32 bit
hexadecimal output.
So it's a 32 bit integer.
From this I want to create random numbers in the range of [a-Za-z0-9]
*without violating randomness* and (if possible) without throwing
away bits of the output.
You mean *characters* int the range [A-Za-z0-9]?
Well this isn't as simple problem as it sounds. A random 32 bit integer
has 32 bits of randomness. If you take a divison reminder of 62 from this
integer you will get only 5,95419631039 bits of randomness
(log(62)/log(2)). So you are wasting 81,4% of your random data. Which is
quite much and usually random data is quite expensive. You can save your
precious random data by taking only 6 bit from your 32 bit integer and
dividing it by 62. Then you will be wasting only 0,8% of random data.
Another problem is alignment, but that is about mathematical correctness.
How can I do this mathemtically (in concern of the quality of output)
correct?
The easiest thing to do would be:
The easiest is not mathematically correct though. Random data will stay
random only if you select and modify it so that randomness is preserved.
If you take devison reminder of 62 from 32 bit integer there are 69 273
667 possibilities of the reminder to be 3 or less. For the reminder to 4
or more the number of possibilities is 69 273 666. In mathematically
ideal case the probability for every index of the list should be same:
1/62 = 1,61290322581%. But the modulo 62 modifies this probability: for
index 0-3 the probability is 69 273 667/2^32 = 1,61290324759%. And for
indexes 4-61 the probability will be 69 273 666/2^32 = 1,6129032243%.
If you wish not to waste those random bits the probabilities will get
worse. With 6 bits of random the probability for index 0-1 will be 2/64
and for 2-63 it will be 1/64. This is a very significant change because
first and second index will appear twice as much as the rest. If you add
2 characters to your list you will perfect alignment and you can take 6
bits of data without it modifying probabilities.
If you are looking a mathematically perfect solution there is a simple
one even if your list is not in the power of 2! Take 6 bits at a time of
the random data. If the result is 62 or 63 you will discard the data and
get the next 6 bits. This selectively modifies the random data but keeps
the probabilities in correct balance. Now the probability for index of
0-61 is 1/62 because the probability to get 62-63 out of 64 if 0.
---
#include time.h
#include stdio.h
#include stdlib.h
#define N (26+26+10)
static char S[] = { 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J',
'K', 'L', 'M',
'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W',
'X', 'Y', 'Z',
'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j',
'k', 'l', 'm',
'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w',
'x', 'y', 'z',
'0', '1', '2', '3', '4', '5', '6', '7', '8', '9' };
int
next_character()
{
// Use the correct call for ISAAC instead of rand()
unsigned int idx = rand() % N;
return S[idx];
}
so modify the next_char function:
char next_character()
{
static unsigned int rand = 0; //(sizeof(int) = 32)
static char bit_avail = 0;
char result = 0;
char move_bits = 0;
char bits_moved = 0;
do {
if (!bits_avail) {
// Use the correct call for ISAAC instead of rand()
rand = rand();
bit_avail = 32;
}
move_bits = bits_avail = 6 ? 6 : bits_avail;
result = move_bits;
result = (result | rand (0xFF (8 - move_bits))) 0x3F;
bits_avail -= move_bits;
bits_moved += move_bits;
rand = move_bits;
} while (bits_moved != 6 result 61);
return result;
}
This function will give perfect distribution of 1/62 probability for
every index. It will waste 6 bits with the probability of 1/32 (2/64).
int
main(int argc, char* argv[])
{
// Use the correct call for initializing the ISAAC seed
srand((unsigned int)time(NULL));
for (int i = 0; i 20; i++) // --std=c99
printf(%c\n, next_character());
return 0;
}
---
If the ISAAC RNG has a good distribution, then the next_character()
function will give a good distribution
Re: [gentoo-user] OT: Mapping random numbers (PRNG)
On Fri, Jun 06, 2014 at 08:39:28PM +0200, meino.cra...@gmx.de wrote:
Canek Peláez Valdés can...@gmail.com [14-06-06 17:36]:
On Thu, Jun 5, 2014 at 9:56 PM, meino.cra...@gmx.de wrote:
Hi,
I am experimenting with the C code of the ISAAC pseudo random number
generator
(http://burtleburtle.net/bob/rand/isaacafa.html).
Currently the implementation creates (on my embedded linux) 32 bit
hexadecimal output.
So it's a 32 bit integer.
From this I want to create random numbers in the range of [a-Za-z0-9]
*without violating randomness* and (if possible) without throwing
away bits of the output.
You mean *characters* int the range [A-Za-z0-9]?
How can I do this mathemtically (in concern of the quality of output)
correct?
The easiest thing to do would be:
---
#include time.h
#include stdio.h
#include stdlib.h
#define N (26+26+10)
static char S[] = { 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J',
'K', 'L', 'M',
'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W',
'X', 'Y', 'Z',
'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j',
'k', 'l', 'm',
'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w',
'x', 'y', 'z',
'0', '1', '2', '3', '4', '5', '6', '7', '8', '9' };
int
next_character()
{
// Use the correct call for ISAAC instead of rand()
unsigned int idx = rand() % N;
return S[idx];
}
int
main(int argc, char* argv[])
{
// Use the correct call for initializing the ISAAC seed
srand((unsigned int)time(NULL));
for (int i = 0; i 20; i++) // --std=c99
printf(%c\n, next_character());
return 0;
}
---
If the ISAAC RNG has a good distribution, then the next_character()
function will give a good distribution among the set [A-Za-z0-9].
Unless I missunderstood what you meant with create random numbers in
the range of [a-Za-z0-9].
Regards.
--
Canek Peláez Valdés
Profesor de asignatura, Facultad de Ciencias
Universidad Nacional Autónoma de México
Hi,
Thank you very much for the input! :)
I have a question about the algorithm:
Suppose rand() has an equal distribution of numbers and furthermore
one has a count of 2^32 random numbers listed in numerical sort
order.
In this list each number would appear (nearly) with the same count: 1
To get an better imagination of that...suppose the rand() would only
return numbers in the range of 1...12 and the alphabet has only 8
characters (as 2^32 is not devideable by 62)
rand():
1 2 3 4 5 6 7 8 9 10 11 12
rand()%N : rand()%7
1 2 3 4 5 6 7 0 1 2 3 4
or in other words: An even distribution of numbers of rand()
would result in a unevenly distributed sequence of characters...or?
This would break the quality of ISAACs output.
I am sure I did something wrong here...but where is the logic trap?
This is the thing I explained in my message.
--
-Matti
[gentoo-user] Re: dev-ruby/json-1.8.0
On 06/06/2014 09:48 AM, Stephen Reynolds wrote: rdoc -o doc -t 'JSON Implementation for Ruby' -m README.rdoc snip ext/json/ext/generator/generator.c sh: rdoc: command not found You apparently have ruby19 and ruby20 installed, is this right? Do you have a version of ruby eselected? eselect ruby show I'm using ruby19, so I have: #ls -l /usr/bin/rdoc /usr/bin/rdoc - rdoc19 That symlink to rdoc19 was created by using eselect ruby set to choose between ruby19 and ruby20. Is all of the above familiar to you? If not, you may need more help with managing multiple ruby versions. I find it a large PITA and I could use more help myself :)
Re: [gentoo-user] upower suddenly demands systemd
On 06/06/2014 16:34, Gevisz wrote: After today's emerge-webrsync, I have found out that usual # emerge --update --deep --with-bdeps=y --newuse --ask world does not work, trying but being unable to emerge systemd. As I have found out, the reason for it was that upower suddenly decided that it needs systemd but, for some (lucky :) reason, it could not be emerged as my system uses OpenRC. The latest news said the following: UPower discontinued hibernate and suspend support in favor of systemd. Because of this, we have created a compability package at sys-power/upower-pm-utils which will give you the old UPower with sys-power/pm-utils support back. Some desktops have integrated the sys-power/pm-utils support directly to their code, like Xfce, and as a result, they work also with the new UPower as expected. All non-systemd users are recommended to choose between: # emerge --oneshot --noreplace 'sys-power/upower-pm-utils' or # emerge --oneshot --noreplace '=sys-power/upower-0.99.0' However, all systemd users are recommended to stay with sys-power/upower. However, that news did *not* say that without # emerge --oneshot --noreplace 'sys-power/upower-pm-utils' it is impossible to update the system even if you use xfce. The news item is a compromise as the problem you ran into is really a very minor one; it's the presence of systemd and two blocking upower packages that cause confusion. Extra points for the magic flame war-inducing trigger-word systemd - see the enormous threads here and on -dev for proof. So the news item clarifies that you are not being forced to use systemd and explains the alternatives in a concise manner. Once you have that done, anything that comes next is simple routine portage blockers which you are expected to know how to deal with, and are not at all worthy of mention in a news item. Only after executing the last command, which installed upower-pm-utils and unistalled upower, the # emerge --update --deep --with-bdeps=y --newuse --ask world worked as desired. upower was not in my world file, so I think it's a some kind of a bug. It's not a bug and upower is not supposed to be in world. It's a simple dep library that portage will pull in if you have any other packages that need it. Keep it and upower-pm-utils out of world so that Samuli's next planned phase of this change in UPower will go smoothly. The reason for all the confusion is due to how portage works internally. Briefly, it sees you have a choice between upower and upower-pm-utils and usually ends up picking upower. It's similar to virtuals where portage doesn't know what you want so just picks the first in the list. If you want the second, then you have to emerge it yourself first, stopping portage from deciding. As I said above, this simple package addition produces the magic flamewar trigger word systemd whcih always gets half the audience very very upset indeed. -- Alan McKinnon alan.mckin...@gmail.com
Re: [gentoo-user] Re: quick installs on older/embedded hardware
On 06/06/2014 19:20, James wrote: Alan McKinnon alan.mckinnon at gmail.com writes: So how do other vendors do it? If we look at their workflow, perhaps a useable method will filter up through the wetware I have no idea how to find and convince an appropriate company to share how they setup for multi image testing on various hardware platforms. If you know of such an opportunity, I'd be all ears on what they would be willing to share. Do tell. Sorry, I have no idea myself. I was hoping that this stuff would be reasonably common knowledge but apparently in the embedded space people guard their stuff jealously. In my space (ISPs) we tend to proudly show off our deployment magic at every chance and every seminar, mostly to get bragging rights and brownie points plus a good healthy dose of one-upmanship. And everyone knows everyone else anyway, half of us have worked at every other ISP at least once - no such thing as keeping your internal systems secret :-) -- Alan McKinnon alan.mckin...@gmail.com
Re: [gentoo-user] Use Flags and Updating
On 06/06/2014 12:44, Rich Freeman wrote: On Thu, May 22, 2014 at 3:54 AM, Marc Joliet mar...@gmx.de wrote: I think nowadays one would prefer --keep-going, which automatically resumes on failure (and recomputes the dependency tree!), and prints a list of failed packages when it's finished. However its output is more verbose than just ok and failed (it'll print the build.log if it's only one package, IIRC). Hmm, after using this script for some time I found a problem with this approach. If you use --buildpkgonly and ---keep-going then emerge won't build a single thing if anything in the list is missing a build-time dependency. The script I posted will try to emerge everything individually so at least some of the packages will be compiled. That seems like a bug in --buildpkgonly. If you use it with --keep-going it should at least compile the packages that aren't missing build-time options. I'll file that as a bug if it isn't already there... I don't think it's a bug, it's more like a difference in interpretation. From the man page: --buildpkgonly (-B) Creates binary packages for all ebuilds processed without actually merging the packages. This comes with the caveat that all build-time dependencies must already be emerged on the system. --keep-going [ y | n ] Continue as much as possible after an error. When an error occurs, dependencies are recalculated for remaining pack‐ ages and any with unsatisfied dependencies are automati‐ cally dropped. Also see the related --skipfirst option. So, decisions about --buildpkgonly are made at the start of an emerge and --keep-going kicks in only when an error occurs at the end, and the former must have higher precedence than the latter. It doesn't make sense to expect portage to change it's behaviour about it's initial decisions just because you also have an entirely unrelated option set that is only a convenience in the event of a build failure. That seems to me too much of an unexpected side effect -- Alan McKinnon alan.mckin...@gmail.com
