IRC Logs for #crux Monday, 2015-09-28

*** dxlr8r has joined #crux00:01
*** lostcoffee has joined #crux00:28
*** lostcoffee has quit IRC00:45
*** dbrooke_ has joined #crux01:34
*** kokakoda1 has joined #crux01:37
*** dbrooke has quit IRC01:38
*** kokakoda has quit IRC01:38
*** chris2 has quit IRC01:38
*** blueness has quit IRC01:38
*** blueness_ has joined #crux01:39
*** chris2 has joined #crux01:40
tired890Hi. Say I tried to install an app with lots of depends (Filezilla), in the end decided will switch to a simpler one (gftp)01:40
tired890now filezilla installed a lot which I will likely never use, is there a way to tell prt-get to remove all apps installed today ?01:41
jaegerthe tools don't keep track of that sort of thing but you could check for leaf ports01:49
tired890jaeger, as in orphans?02:16
jaegernot quite the same thing02:16
jaegerleaf ports are ports that are installed but have no upstream dependencies, basically02:16
jaeger is a quick and dirty script I use02:16
tired890jaeger, cheers02:17
tired890Hi frinnst, there is an md5 mismatch for package gdk-pixbuf02:27
*** ________mavric has quit IRC02:50
*** ________mavric has joined #crux02:54
Romsterlooks fine to me tired89002:58
*** tilman_ has joined #crux04:03
*** tilman_ has quit IRC04:03
*** tilman_ has joined #crux04:03
Workstertired890, wish you would state what file had the sum mismatch04:17
Workstertired890, fixed04:19
tired890Workster, cheers04:26
*** Workster has quit IRC04:38
*** Workster has joined #crux04:46
*** Workster has quit IRC04:46
*** Workster has joined #crux04:46
Worksterwindows froze -_-04:47
tired890want to do a setup like that05:00
tired890this has device passthrough, as in pass the entire vga to the virtual machine and get native performance05:00
Worksterjaeger, was doing that.05:06
Worksteri was gonna do it but my mobo does not support it05:06
tired890what feature is your mobo lacking ?05:10
tired890iommu? or hw virtualization features?05:11
*** SiFuh has quit IRC05:13
*** penguinfan has joined #crux05:17
*** SiFuh has joined #crux05:23
*** penguinfan has quit IRC05:53
*** |narcos| has joined #crux05:55
*** Pingax has joined #crux06:44
*** BitPuffin|osx has quit IRC07:00
*** hhhhhhhh has joined #crux07:44
*** hbekel has joined #crux08:06
*** ivs_ has joined #crux08:42
*** onodera has joined #crux09:04
*** xeirrr has joined #crux09:15
*** xeirrr has quit IRC09:16
*** xeirrr has joined #crux09:16
*** xeirrr has quit IRC09:43
*** toriso has quit IRC09:58
*** Pingax has quit IRC10:00
*** Pingax has joined #crux10:00
*** Pingax has quit IRC10:00
*** Pingax has joined #crux10:00
frinnstThe connection to the server was reset while the page was loading.10:00
frinnstlove hp.com10:00
*** toriso has joined #crux10:00
frinnsthow the fuck can they suck so bad?10:02
*** hhhhhhhh has quit IRC10:08
darthlukanfrinnst: Practice!10:25
frinnstthey have been practicing for atleast 15 years now10:26
frinnsti guess they have it down10:26
darthlukanfrinnst: Yep10:34
*** nullmark has quit IRC11:06
*** nullmark has joined #crux11:11
*** pejman has quit IRC11:56
*** pejman has joined #crux11:56
rmullfrinnst: Love it12:13
*** user__ has joined #crux12:32
*** ivs_ has quit IRC12:35
*** BitPuffin has joined #crux13:10
*** |narcos| has quit IRC13:23
*** dougiel has joined #crux14:09
*** hbekel has quit IRC14:14
*** teK__ has quit IRC14:24
*** heroux has quit IRC14:25
*** heroux has joined #crux14:25
*** lostcoffee has joined #crux14:27
*** |narcos| has joined #crux14:36
*** Wildefyr has joined #crux14:41
WildefyrAny ideas why label -f has the UUID column blank?14:41
jaegerwhat's label?14:47
Wildefyralso blank14:47
Wildefyrfstype too14:48
jaegerI mean, what is the "label" command you're using? What package provides that?14:48
WildefyrSorry, phone autocorrected lsblk to label14:48
jaegerAh, ok14:48
jaegerthe label can be blank, it's not required. Odd that the UUID would be.14:49
WildefyrI used fdisk to create a new gpt table and partitioned from there14:52
jaegerno idea there, sorry. Is it working fine even without a UUID?14:54
jaegerdoes "blkid" show the same?14:54
WildefyrThat shows the UUIDs fine14:56
*** lostcoffee has quit IRC15:01
*** dougiel has quit IRC15:33
*** frinnst has quit IRC15:42
*** newbie__ has quit IRC15:42
*** dougiel has joined #crux15:43
*** frinnst has joined #crux15:47
*** nogagplz_ has joined #crux16:04
*** nogagplz has quit IRC16:06
*** nogagplz__ has joined #crux16:08
*** nogagplz_ has quit IRC16:10
onoderathis is pretty cool16:20
onodera( )16:20
jaegervery nice16:22
joacimck4up works, but can be a pain to setup sometimes16:23
*** newbie__ has joined #crux16:25
frinnststupid noisy fan17:33
*** frinnst has quit IRC17:33
*** newbie__ has quit IRC17:33
*** dougiel has quit IRC17:42
*** frinnst has joined #crux17:45
frinnstthere we go, blissful silence17:45
tilmanthought you were refering to the silent irc channel for a bit17:46
frinnst40mm fans are so quiet17:46
frinnstwhen I want quiet I just open a query window with myself and stare at it17:46
Wildefyrjaeger, worked out why it wasn't showing, I was doing it from inside the chroot! but blkid worked with no issues from inside it17:50
jaegergood to know but still kinda odd17:53
*** SiFuh has quit IRC18:00
*** SiFuh has joined #crux18:01
*** onodera has quit IRC18:02
*** onodera has joined #crux18:02
rmullI want to shift off the least significant bits of a uint16_t until it is less than or equal to a given bitmask. Is there a bit twiddling hack that can do this without resorting to a loop?18:06
*** Pingax has quit IRC18:22
tilmanrmull: is this what you have now?18:55
tilmani'm a bit confused because you said "bitmask" when what you described isn't a masking operation18:55
rmulltilman: Yeah, that's what I have now18:56
rmullIt's not a masking operation, I guess I could have called it a "number"18:56
rmullThe thing I called a mask is a PWM period register, and the input is a pwm duty cycle value18:57
rmullAnyway, it's a corner case and not a big deal18:58
rmullBut then I started thinking that I can't have been the only person to ever want to do an operation like this18:59
tilmanhacker's delight says:19:03
tilmana function mapping words to words can be impl'd with word-parallel add, sub, and, or, and not insns iff each bit of the result depends only on bits at and to the right of each input op19:03
rmullI think my problem qualifies, then, right? I'm not dependent on bits to the left, though I don't quite understand what they mean by "at and to the right" if my input op is a 16-bit word19:06
DaViruzwhat a tricky problem, i've gone about it several ways now, and it always ends up needing to be iterative19:08
rmullIt "feels" like there should be a hack available, but I'm just gonna use the loop19:15
DaViruzi think i can do it using log219:26
DaViruza loop is probably more efficient then that19:27
*** teK__ has joined #crux19:28
DaViruzseems there is a way to calculate log2 for a 16bit number using 4 oR and 4 bitshifts19:29
*** BitPuffin has quit IRC19:37
*** lostcoffee has joined #crux19:40
DaViruzit can probably be improved a fair bit19:47
DaViruzbut it seems to work fine19:47
tilmanthese divisions might hurt though19:50
tilman(sorry i've been distracted, would love to jump in)19:50
DaViruzyeah depending on architecture19:51
DaViruzbeing an mcu i'm guessing it does single tic integer division19:51
tilmangood point!19:51
DaViruzappearantly some architectures have dedicated instructions for counting leading unset bits. that would be useful19:54
DaViruzthere probably is a way to do it without divisions though.. but i need to go to sleep19:57
*** lostcoffee has quit IRC20:13
*** dougiel has joined #crux20:28
*** xor29ah has quit IRC20:44
tilmanDaViruz, rmull: something like    num >> (nlz(cmp) - nlz16(num) + 1)  but not quite20:45
tilmanrmull: do you have a clz/nlz and/or popcount insn?20:45
*** xor29ah has joined #crux20:52
*** nullmark has quit IRC20:55
*** nullmark has joined #crux20:59
tilmanDaViruz, rmull: actually what i just proposed might be what you're after. DaViruz' formula seems to break for the input pair (12037,5221)21:01
DaViruzoh dear, i tested it in some random cases and some i thought might break it21:01
DaViruz(sleep is going well)21:02
DaViruzmaybe some argv would be in order so i don't have to recompile for every test21:03
tilman this is what i have21:03
DaViruzindeed it broke :(21:04
tilmanrmull: you need to report back on this! hope you're using modern arm, which does indeed have a CLZ instruction :)21:07
DaViruzpop16 seems to be iterative though?21:14
rmulltilman: This is meant to be portable code. ARM would be one target, but it would probably see most use on 16-bit PIC2421:27
*** user__ has quit IRC21:27
rmullDivides kind of suck on the PIC21:29
rmull19 cycles21:30
rmullSo the while-loop-comparison is still going to be the fastest solution here, I think21:30
rmulland of secondary importance, the most readable21:31
*** |narcos| has quit IRC21:33
DaViruzisnt compares pretty cycle intense too?21:34
rmullI'll check21:35
DaViruzi'm probably mixing it up with something else21:36
*** tierd891 has joined #crux21:38
*** BitPuffin|osx has joined #crux21:39
rmullSeems like it's typically 4 cycles per instruction on the PIC, excepting DIV21:41
rmullAnd the while loop disassembles into 8 instructions21:41
rmullso... 32 cycles each time through the loop, compared to the 19-cycle DIV and some other cruft - it's not necessarily a winner like I had assumed21:42
DaViruzyep looping seems to be beneficial. probably way less program space too21:43
DaViruzi think my method can be made to work without division though.21:44
DaViruz..and be made to indeed work at all since it seems to give wrong results in some cases21:44
rmullYes, less program space for the loop, but not necessarily faster than the DIV depending on loop iteration count21:50
rmullI haven't tried to compile your example though, so I don't know what the disassembly looks like21:50
rmullGotta run though. Thanks for the insights.21:51
*** tierd891 has quit IRC21:53
*** teK__ has quit IRC22:10
*** teK__ has joined #crux22:12
*** teK__ has quit IRC22:15
*** gerardlally has joined #crux22:21
gerardlallyhi again everyone - long time no see22:22
gerardlallyjust wondering if anyone has a guide to creating initrd on crux22:22
*** teK__ has joined #crux22:22
*** toriso has quit IRC22:25
*** onodera has quit IRC22:39
*** gerardlally has quit IRC23:08
*** tired890 has joined #crux23:13
tired890hey guys23:19
tired890I'm trying to recompile kernel 4.2.1 with kvm support, but when I enable virtualization in menuconfig I get only two items inside, no kvm or amd_kvm and such23:20
tired890do I need to enable something else to get the full menu ?23:20
tired890oh well, I enabled it, re-ran menuconfig and its there!23:44
tired890so u need to enablle, save .config, quit.. then re-run. Rinse and repeat for all "missing" options23:45
jaegerNormally that isn't required, for reference23:48
tired890jaeger, yes correct.. but for some reason on my box it didn't appear the first time (namely KVM menu/submenus)..23:49
tired890now it does though and I'm compiling 4.2.1 with kvm support.. I got a tiny cheapo gfx card and will try passthrough will see how that goes :)23:50
jaegergood luck :)23:51
tired890cheers! Rebooting into new kernel brb23:52
*** tired890 has quit IRC23:52
*** tired890 has joined #crux23:55

Generated by 2.11.0 by Marius Gedminas - find it at!