IRC Logs for #crux Tuesday, 2018-01-09

leetspete1PCIe 4 peak throughput is a little under 15Gb/s, I imagine 10Gbps requires some hard work to keep the PCIe bus saturated.00:04
leetspete1A little under 16Gb/s, I mean.00:05
john_cephalopodaWell, a PCI card is usually not more than 20 cm away from the CPU, while network cables run for dozens of meters easily in a building.00:08
leetspete1I mean if you're running a 20cm cable even, I suspect it requires a lot more effort to turn edges on copper into bits and ship them off to the PCIe bus quickly enough.00:11
leetspete1There's an order of magnitude more wiggle room the hardware has between 1Gbps and 10Gbps.00:11
*** john_cephalopoda has quit IRC00:17
*** darfo has quit IRC00:28
*** leah2 has quit IRC00:40
*** nogagplz has joined #crux01:24
WorksterA Dual ABI is provided by the library. A new ABI is enabled by default. The old ABI is still supported and can be used by defining the macro _GLIBCXX_USE_CXX11_ABI to 0 before including any C++ standard library headers.03:22
*** _________mavric6 has quit IRC03:51
*** _________mavric6 has joined #crux03:53
*** tilman_ has joined #crux03:55
jaegerFinally took the time to build projectm ports for kodi in my repo, if anyone cares about that04:28
*** amonn has quit IRC04:49
*** amonn has joined #crux04:50
*** elderK has joined #crux05:54
elderKHey guys.05:56
elderKI just received my new laptop today and am, of course, installing Crux on it.05:56
elderKSo, I'm just configuring my kernel.05:56
elderKI've done this many times. But, I was wondering: Is there comprehensive documentation somewhere about the options? "Help" during configuration is handy and all but sometimes they don't tell you whether you should enable an option or not, and sometimes, you don't know yourself.05:57
elderKAny tips or advice? :)05:57
elderKWRT drivers, if I don't know if I need it, I build it as a module. And see if it's loaded.05:58
joacimi usually just go to gentoo and arch documentation06:11
Anselmoalso readling things like lspci and lsusb can be very helpful if you need to figure out what specific hardware you're needing a driver for06:29
tilmanelderK: in the really old days, we used the kernel HOWTO :D06:52
elderKtilman: Does it descrbie all kernel options?07:11
elderKBecause that's really what is key :)07:11
tilmanit was probably last updated in 200107:11
tilmanelderK: wrt drivers, there's "make localconfig" which looks at the hardware present in the system where you build (configure) the kernel07:12
elderKNo rule?07:12
elderKLike, I wonder: How do I determine if my battery is a "smart battery"?07:13
elderKmake localconfig errored out "No rule."07:13
elderKin /usr/src/linux07:13
tilman1.) don't put the kernel sources in /usr/src/linux. that's ancient shit07:14
tilmanput it in ~ somewhere07:14
elderKtilman: I normally don't. This is literally first boot setup07:14
tilman2.) try "make localyesconfig" or "make localmodconfig"07:14
elderKFrom the Cd.07:14
tilmansorry :D07:14
elderKNormally I do it in a much saner way.07:14
elderKBut I'm just bringing teh system up for now :D07:14
elderKI.e. Don't build kernel as root, etc.07:14
elderKDo we still need to enable AGPGART in Video, for instance, when we are using PCIExpress?07:24
elderKSay, the integrated video.07:24
elderKSeems not. I'll try it out :)07:24
*** lyosha[m] has left #crux ("User left")07:47
*** workodera has joined #crux08:11
*** workodera has quit IRC08:14
*** workodera has joined #crux08:14
*** mutin-sa has joined #crux08:25
*** mutin-s has quit IRC08:27
*** chinarulezzz has joined #crux08:30
SiFuhWow $5000 for a hat08:35
*** leah2 has joined #crux08:38
*** abenz has quit IRC08:46
elderKAh, fresh installs :D08:56
elderK:D Hope you guys enjoy the image. I sure did :D09:00
SiFuhworkodera: Faijin is so sad movie, and so different09:10
workoderayeah right09:12
workoderaa romance movie where the leads don't meet, I really like it09:12
*** abenz has joined #crux09:13
SiFuhYeah it is very different09:13
*** abenz has quit IRC09:18
elderKThanks for the help peeps :)09:28
*** abenz has joined #crux10:15
frinnstI need an ascii version of that image. To print when I update all our servers via ansible10:33
joacimcan probably pipe it through caca or something10:58
joacimaaaaaand i start updating all my stuff at home10:59
SiFuhis the best I can find11:05
frinnstyeah I did it already :-)11:05
frinnstbit big tho11:05
joacimi made a port for myself a long time ago11:05
SiFuhhmm i wonder about ascii porn movie ;-)11:06
joacimmight have a lot of ass play11:06
*** SiFuh has quit IRC11:11
*** SiFuh has joined #crux11:11
frinnstamiga stag disks!11:21
*** elderK has quit IRC11:34
*** john_cephalopoda has joined #crux11:43
*** john_cephalopoda has quit IRC11:51
*** workodera has quit IRC12:07
*** workodera has joined #crux12:08
*** john_cephalopoda has joined #crux12:08
*** abenz has quit IRC12:26
*** workodera has quit IRC13:30
*** workodera has joined #crux13:31
*** workodera has quit IRC14:00
*** workodera has joined #crux14:01
*** workodera has quit IRC14:09
*** Guest22874 has joined #crux14:12
*** workodera has joined #crux14:18
*** workodera has quit IRC14:39
*** workodera has joined #crux14:40
*** workodera has quit IRC14:46
*** workodera has joined #crux14:47
*** workodera has quit IRC14:53
*** workodera has joined #crux14:54
*** amonn_ has joined #crux15:34
*** abenz has joined #crux15:36
*** amonn has quit IRC15:37
*** abenz has quit IRC15:43
*** workodera has quit IRC16:36
*** Workster has quit IRC17:04
*** Workster_ has joined #crux17:06
cruxbot[opt.git/3.3]: nvidia: updated to version 384.11117:09
cruxbot[compat-32.git/3.3]: nvidia-32: updated to version 384.11117:09
pedjaMeltdown PoC 'traps: meltdown[18032] trap invalid opcode'17:21
pedjaso, either it doesn't work or my CPU is way too old, as suggested by the PoC docs17:23
pedjaAMD permanently slashing MSRP for most Ryzen and some Threadripper processors is just clearing up shelves before Zen+ in April, I guess17:27
*** onodera has joined #crux17:55
*** heroux has quit IRC17:56
*** heroux has joined #crux17:56
frinnstpedja: where did you find the poc?18:00
tilmani'm guessing18:04
tilmanwritten by the Graz guys it looks like18:05
*** elderK has joined #crux18:07
elderKHey guys, I need some advice.18:07
elderKLong story short: IOMMU support isn't so hot in kernel 4.9.18:07
elderKAnd, I left my machine building stuff all night.18:07
elderKCome back, some things have failed. Don't think much. But check dmesg, and there are a ton of DMA read errors, etc.18:07
elderKSeems that is related to IOMMU.18:08
elderKInvestigate a little more, seems NVMe was the one triggering them.18:08
elderKSo, I am in an uncomfortable situation.18:08
elderKIf there are DMA read errors from nvme - which my HD is on - is there a possibility that stuff on my drive is corrupt in some way?18:08
elderKThat the things I've been installing - some of the files, etc, may not have been written properly?18:08
elderKI'd really like to avoid having to rebuild all the stuff again.18:09
elderKBut I'm not sure I can avoid that?18:09
tilmanIOMMU has been around for a while18:14
tilmanare you sure it's the root cause of your problems?18:14
*** BitPuffin has joined #crux18:14
elderKWell, this is an old kernel.18:15
elderKAnd I've found a great deal of stuff related to this.18:16
elderKI'm not entirely sure.18:16
elderKI just know it's telling me PTE fault, etc, from 04:00.0 (which is the nvme controller)18:16
elderKDMAR, etc.18:16
elderKtilman: Any advice on how I can prove it is IOMMU?18:16
elderKtilman: How concerned should I be about corruption?18:16
tilmani'd start over with the installation/build18:17
tilmanbecause i don't know how you'd proove that the builds are okay18:18
elderKYeah, that's what I was worried about.18:18
elderKThat's 9 hours down the drain :P18:18
tilmanwhat happens when you don't compile in support for IOMMU at all?18:19
tilmanyou'd very probably want it for production use though18:19
elderKI haven't tested that yet. With my last laptop, it had similar problems with IOMMU. I just disabled it and life was fine.18:19
elderKI'd like to use IOMMU for VM and stuff.18:19
elderK(This isn't a server though - just my new laptop.)18:20
elderKI need a way to trigger the error... like, reliably.18:20
j_velderK: can you paste the dmesg?18:20
elderKAfraid not. It's on another machine :P18:20
elderKI'll type one of the errors if it helps:18:21
elderKDMAR: DRHD: handling fault status 218:21
elderKDMAR: [DMA Read] Request device [04:00.0] fault addr fd1690000 [fault reason 06] PTE Read access is not set18:21
elderKdmar_fault: 30 callbacks supressed18:22
elderKtilman: If I am to redo the installation, I think it would be wise to disable IOMMU - not unless I can find a way to trigger the error consistently. Otherwise, I can't be sure that I'm not going to get this error building stuff on my next run. And I have no way of validating that all installed binaries are actually sane.18:23
tilmanthat's what i'd do18:23
tilmanbuild a running system without IOMMU, then figure out what's wrong18:23
elderKI found some recent posts suggesting that Intel NVMe SSDs have firmware issues, sometimes, too.18:24
elderKIt seemed to be fixed and integrated into later kernels.18:24
elderKI'm not if that applies to me or not.18:24
elderKBut it's kinda worrying.18:24
elderKtilman: What benefits does IOMMU bring to the system apart from virtualization?19:01
j_velderK: does your nvme drive match the one listed in that ubuntu bug? if so, looks like you may need a kernel >= 4.12... i don't find any of the fixes mentioned in that launchpad page in kernels earlier than 4.12.*19:04
elderKj_v: No, it doesn't match.19:04
elderKMine seems to be a later revision.19:04
elderKI did a quick Google for my part number, found nothing.19:04
elderKAnyway, for the time being, I've disabled VT-d19:05
elderKI'll re-enable it once I've got a stable system up and running.19:05
elderKAnd with a modern kernel.19:05
elderKBtw guys, is there a way to force EFIFB / KMS to use a specific resolution?19:06
elderK:P Text is so damned small.19:06
*** tsaop has joined #crux19:12
j_velderK: i ran into issues with text being too small using nouveau fb. my workaround is to set console font in /etc/rc.conf (FONT=) to a point size that works for me.19:36
*** BitPuffin has quit IRC19:53
john_cephalopodaj_v: Ask in #nouveau - they will probably offer you a patch within five minutes, as I know them ;)19:53
j_vjohn_cephalopoda: i don't see it as a bug. i set font anyways (i use terminus fonts), so i just pick a point size that works for my gradually failing eyes.19:55
john_cephalopodaAh, ok.19:56
*** Kruppt has joined #crux20:24
*** nwe has quit IRC20:36
*** nwe has joined #crux20:41
*** abenz has joined #crux20:54
j_vWorkster_: thanks for your input. while that info didn't specifically help find a fix, it got me thinking in a different vein and seem to have found a fix after looking up older firefox crashes due to gcc6 builds.21:04
cruxbot[opt.git/3.3]: glib: updated to 2.54.321:09
cruxbot[opt.git/3.3]: gtk: updated to 2.24.3221:09
cruxbot[opt.git/3.3]: freetype: updated to 2.921:09
pedjahm, new freetype. I wonder if it's time to try it without enabling subpixel rendering21:35
*** fun2 has joined #crux21:46
fun2new microcode from intel 2018010821:47
*** onodera has quit IRC21:48
pedjano Changelog, because, by now, everyone knows what it's supposed to fix :)21:54
*** tsaop has quit IRC22:00
*** tsaop has joined #crux22:06
*** tsaop has quit IRC22:09
*** Workster has joined #crux22:20
*** john_cephalopoda has quit IRC22:34
*** john_cephalopoda has joined #crux22:37
*** chinarulezzz has quit IRC23:10
*** Kruppt has quit IRC23:46
*** kori has quit IRC23:47
*** kori has joined #crux23:49

Generated by 2.14.0 by Marius Gedminas - find it at!