[01:35] hi all [01:36] wb [01:38] thanks. [01:38] Does `make bindeb-pkg` make the modules? I don't have any .deb that has the word "modules" in it. [01:40] Sorry I’m not at my computer right now to confirm but I think maybe the bindeb-pkg target might just roll them all up into the *images*.deb [01:40] If you run “dpkg -x | grep nvme” the modules might be listed [01:40] ok, thanks. let me check [01:41] sorry for being so high-maintenance [01:41] Wanting to learn something new is hardly high maintenance :-) if anyone tells you that then they’re wrong [01:42] Er sorry “dpkg -x *image*.deb | grep nvme” [01:42] I don't see anything else in the make help output that looks like it'd generate modules debs [01:42] seth@hydrogen:~/kernel$ dpkg -c linux-image-5.3.10+_5.3.10+-1_amd64.deb | grep -i nvme [01:42] drwxr-xr-x root/root 0 2019-12-16 15:39 ./lib/modules/5.3.10+/kernel/drivers/nvme/ [01:42] drwxr-xr-x root/root 0 2019-12-16 15:39 ./lib/modules/5.3.10+/kernel/drivers/nvme/host/ [01:42] -rw-r--r-- root/root 163713 2019-12-16 15:39 ./lib/modules/5.3.10+/kernel/drivers/nvme/host/nvme-core.ko [01:42] that's dpkg -c [01:42] Okay cool, looks like they’re all rolled up into *image*.deb [01:42] does that mean the modules are loaded? [01:43] ok [01:43] is that the right image deb? [01:43] I don't want dbg, right? [01:43] I think the *modules*.deb is purely an ubuntu artifact with how the scripts are set up [01:43] Yes that’s the right deb [01:43] linux-headers-5.3.10+_5.3.10+-1_amd64.deb linux-image-5.3.10+_5.3.10+-1_amd64.deb [01:43] linux-image-5.3.10+-dbg_5.3.10+-1_amd64.deb linux-libc-dev_5.3.10+-1_amd64.deb [01:43] Dbg will only be helpful if you want to run that kernel under a debugger [01:43] those are my choices [01:43] second one, I guess. [01:43] Yeah [01:44] ok, now a dpkg -i on the HV, right? [01:44] wish me luck. [01:44] Good luck! [01:45] W: Possible missing firmware /lib/firmware/amdgpu/vega20_ta.bin for module amdgpu [01:45] W: Possible missing firmware /lib/firmware/amdgpu/navi10_mes.bin for module amdgpu [01:45] I seem to recall having that earlier too [01:45] so I'm going from 5.3.0 to 5.3.10. [01:45] will be back after reboot, hopefully. :) [01:45] * aberrant waves [01:47] o/ [01:50] w000000 [01:50] seth@elemental:~$ sudo fstrim -v / [01:50] seth@elemental:~$ sudo fstrim -v / [01:50] damnit, that's not pasting [01:50] suffice to say it works [01:51] we need to get this patch into the kernel :) [01:51] TIL there's a -v to fstrim :) [01:51] /: 758.3 GiB (814159003648 bytes) trimmed [01:51] there [01:52] I'll update the bug report. [01:52] hmm which irc client do you have? I thuoght most? all? had paste detection to handle a leading / in a paste correctly [01:53] that might be worth a bug report for your irc client, heh [01:53] Nice job aberrant!! [01:55] sarnold: I'm using irssi [01:55] connor_k: I couldn't have done it without you. Thanks so much! [01:55] hm. now I've got to try.. :) [01:55] What’s the bug number, aberrant? [01:55] 1856603 [01:56] # fstrim -v / [01:56] /: 6.7 GiB (7215800320 bytes) trimmed [01:57] irssi 0.8.19-1ubuntu1.9 on this machine [01:58] sarnold: from what I understand the fstrim issue only manifests in certain nvme drives that don't follow the spec precisely. [01:58] 758.3 GiB trimmed seems like an awful lot though. [01:59] aberrant: heh, I was more just interested in the / leading char thing, the actual issue you and connor_k are discussing I haven't followed at all, hehe [01:59] I don't think I've gone past 20% utilization on the drive. [01:59] hahahaha [01:59] Cool! I wrote it down and can look at it tomorrow for SRU’ing the patch into the kernel [01:59] aberrant: how long has it been since you've done an fstrim on this system? what kind of write churn does it have? do you have the weekly fstrim timers running correctly? [01:59] CTCP VERSION reply from aberrant: irssi v1.2.1-1ubuntu2 [02:00] sarnold: the machine is less than a week old. I've never done an fstrim on it. [02:00] 700G trimmed *does* feel like a huge trim if you're getting it done weekly already.. [02:00] sheeeeeesh [02:00] I set up some vms on it, and transferred a bunch of data. [02:00] but currently I'm at 19% utilization [02:01] (1TB) [02:01] I wonder if the first fstrim on a brand new disk is somehow large. The subsequent one I did was 0 GiB. [02:01] oh, look [02:03] that's roughly the size of the total free space [02:03] yeah, any more and I'd have been very worried :) [02:03] so maybe it had to do a complete fstrim on the entire free list. [02:03] it's possible that the 'first' fstrim on the filesystem does have to start from scratch and trim it all, whether or not it'd been used [02:05] 250ish gigs, though, the average data would only need to be overwritten three times.. between write amplifications on small writes, OS updates on host and guests.. maybe 700-ish gigs of dirtied but no longer needed data isn't crazy for a 250-ish gig used system [02:05] I shudder to think about eg atime updates on all those bloody files. relatime is a good thing, but even then.. [02:06] I'm mounting relatime [02:07] let's see what happens next sunday :) [02:07] :) [02:07] this channel is fantastic. Thanks for all the support here. [02:07] hmm does the timer use fstrim -v? or fstrim? or .. just run the ioctls raw? [02:07] as it turns out I have to RMA my motherboard. [02:07] it won't take a BIOS update. It's stuck on the penultimate version. [02:10] here's the logs from three of my machines http://paste.ubuntu.com/p/GGJHQKFDJj/ [02:11] that's frustrating :( [02:15] yeah, so I confirmed: if I add the trimmed GiB to the used GiB I get exactly the capacity of the disk [02:15] so the first trim probably trims the entire free space [02:16] 758.3 + 158 [02:17] well, not *exactly* but close enough for government work [02:20] ok. That's enough stress for one day. thanks again. Have a great evening, all. [02:20] * aberrant waves. === mamarley_ is now known as mamarley === kerneltoast_ is now known as kerneltoast === acheronuk is now known as RikMills === lan3y is now known as Laney === cjwatson_ is now known as cjwatson === stgraber_ is now known as stgraber [18:23] morning all :) [18:24] after that patch everything is wonderful. [20:34] hi there. may i have a question regd. mainline kernels here or is there a better channel to ask? [20:36] this may be fine, but there's also a #kernelnewbies on irc.oftc.net that's more about kernel programming. maybe ask and see what happens :) [20:37] cool thanks [20:42] so, all builds on kernel.ubuntu.com/~kernel-ppa/mainline/ from v5.3.7 failed for i386. Is that intentional? I mean, i386 is kind of dead these days but if there is a build recipe i think it should either build successfully or it should go away... If there's something i could help with..? [20:49] heh that's indeed one that wouldn't be useful for #kernelnewbies :) [20:51] :)