Messages in this thread | | | Subject | Re: USB Attached SCSI breakage due no udev involvement | From | Dio Putra <> | Date | Sun, 10 May 2020 20:15:08 +0700 |
| |
On 5/10/20 4:48 PM, Dio Putra wrote: > On 5/10/20 3:48 PM, Greg KH wrote: >> On Sun, May 10, 2020 at 03:35:34PM +0700, Dio Putra wrote: >>> On 5/10/20 2:32 PM, Greg KH wrote: >>>> On Sun, May 10, 2020 at 02:10:04PM +0700, Dio Putra wrote: >>>>> On 5/10/20 1:54 PM, Greg KH wrote: >>>>>> On Sun, May 10, 2020 at 01:48:24PM +0700, Dio Putra wrote: >>>>>>> On 5/10/20 12:47 PM, Greg KH wrote: >>>>>>>> On Sun, May 10, 2020 at 09:55:57AM +0700, Dio Putra wrote: >>>>>>>>> Hi, it's first time for me to report user-space breakage in here, so >>>>>>>>> i'm begging your pardon. >>>>>>>>> >>>>>>>>> I want to report that Linux 5.4 breaking my USB mount workflow due >>>>>>>>> udevadm monitor report here (I'm using vanilla kernel 5.4.39 on >>>>>>>>> Slackware64 Current and vanilla kernel 4.4.221 on Slackware64 14.2): >>>>>>>> >>>>>>>> <snip> >>>>>>>> >>>>>>>> Sorry, but what actually changed that you can see in the logs? >>>>>>> Sorry, what do you mean? The dmesg log or the kernel changelogs? >>>>>> >>>>>> Either, your message made them pretty impossible to compare with all of >>>>>> the line-wrapping :( >>>>>> >>>>> I'm so sorry for first message mess, because that message has been sent by >>>>> Gmail Website. Can I send my logs as attachment? I try to convenient everyone >>>>> here. ( FYI, I just switched to Thunderbird with these settings: >>>>> https://www.kernel.org/doc/html/v4.12/process/email-clients.html#thunderbird-gui ) >>>> >>>> Sure, attachments work, but better yet, if you can show the difference >>>> in a few lines that is much nicer than having to dig through large >>>> numbers of log files. >>>> >>> Okay, I'll attach long messages and trim it as far as I can. >> >> Again, I really do not understand what exactly is "not working". >> >> Please explain that when you send the new log messages. >> > Okay, here's compilation of "dmesg | grep -i udev && udevadm > monitor && lsblk" on linux-4.4.221 vs linux-5.4.39. You can check > my attachment here because I can't explain it by words (I'm not > english native speaker anyway, which makes it so difficult to > explain it for me). > >>>>>>>> What functionality broke? What used to work that no longer does work? >>>>>>>> >>>>>>> Yes, it supposed that just work and kernel could talk with udev, not just handled by the kernel. >>>>>> >>>>>> I don't understand, what functionality changed? What exactly used to >>>>>> work that no longer does? >>>>> linux-5.4 has been never called the udev dependencies whereas >>>>> linux-4.4 will call any udev dependencies if necessary, that's the problem. >>>> >>>> I do not understand what exactly you mean by "call udev dependencies". >>>> >>>> udev is used to create symlinks and set user/group permissions on device >>>> nodes in /dev/ which is created by devtmpfs. What exactly is not >>>> happening in your /dev/ with the move to a newer kernel? >>>> >>> Would I send my dmesg log with "udev.log-priority=debug" as attachment then? >>> >>>>>> Did you change anything else other than the kernel on your system? Did >>>>>> you change to a newer version of udev/systemd or anything else? >>>>>> >>>>> I'm using eudev-master from their official mirror github: >>>>> https://github.com/gentoo/eudev >>>> >>>> Have you contacted the eudev developers to see if something different >>>> needs to be set in your kernel when moving 4 years in kernel development >>>> forward? Are you sure you have all the correct config options enabled? >>>> >>> It's my bad not to contact the eudev developers first. However I'm not quite >>> sure to contact the eudev developers would solve the problem, but CMIIW. >>> >>>> Why such a huge leap forward all at once, how about going from 4.4.y to >>>> 4.9.y and then 4.14.y and then 5.4.y? That might help narrow things >>>> down a bit easier. >>>> >>> Unfortunately I need to think twice due almost ran out of electricity >>> here every time I power on my laptop for long time. So maybe I can't. >>> >>> But if these steps are necessary, I'll think solution later. >> >> Why would it take a long time to do this type of change? >> > It's a weak machine, it took appropriately 2 hours to just compiling > linux kernel. I need to think twice too because electricity is expensive here. > After deep digging into /dev/ folder, indeed some important file has been deleted after restart on Slackware.
On behalf from myself for linux-kernel community, I'm really sorry for causing these noise. Thank you Greg and all for the hints. >> thanks, >> >> greg k-h >> >
| |