Arch Linux

Please read this before reporting a bug:
https://wiki.archlinux.org/title/Bug_reporting_guidelines

Do NOT report bugs when a package is just outdated, or it is in the AUR. Use the 'flag out of date' link on the package page, or the Mailing List.

REPEAT: Do NOT report bugs for outdated packages!
Tasklist

FS#15716 - [initscripts] Reorder/changing RTC device initialization code

Attached to Project: Arch Linux
Opened by Gerardo Exequiel Pozzi (djgera) - Wednesday, 29 July 2009, 07:04 GMT
Last edited by Eric Belanger (Snowman) - Monday, 09 November 2009, 09:13 GMT
Task Type Feature Request
Category Packages: Core
Status Closed
Assigned To Aaron Griffin (phrakture)
Thomas Bächler (brain0)
Roman Kyrylych (Romashka)
Architecture All
Severity Very Low
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 0
Private No

Details

* This changes how RTC device is created at startup. Instead of doing this
manually like now (and only with rtc-cmos), leave the work for udev.
* The hwclock is executed just after all devices are created (udevadm settle),
and all modules are loaded.
* In case of no udevd is in the system, do it manually.
* For the manual step, change the code a bit to be more simple,
and use only bash "read" built-in and /sys
* Also change the test for detect if udev is loaded. Instead of using pidof,
check if the directory /dev/.udev exists.


Additional info:
initscripts-2009.07-3
   0002-Reorder-changing-RTC-dev... (3.5 KiB)
This task depends upon

Closed by  Eric Belanger (Snowman)
Monday, 09 November 2009, 09:13 GMT
Reason for closing:  Deferred
Additional comments about closing:  See last comment.
Comment by Roman Kyrylych (Romashka) - Wednesday, 29 July 2009, 07:10 GMT
IIRC we use manual RTC setup due to some problem (need to dig old reports, maybe that's no longer relevant since hwclock calling was moved too),
so an extra care should be taken here.
Comment by Gerardo Exequiel Pozzi (djgera) - Wednesday, 29 July 2009, 07:19 GMT
True, maybe for some old kernels, when RTC is not registered in /sys, so udev can't detect it then not modules are loaded.
Comment by Roman Kyrylych (Romashka) - Wednesday, 29 July 2009, 07:23 GMT
Found the related bugreport:  FS#9636 . Please read. Also  FS#9649 .
Comment by Thomas Bächler (brain0) - Wednesday, 29 July 2009, 07:31 GMT
I think there was a bug report for udev being started before the time has been set, that's why we enabled it manually before that. Can't remember the details.
Comment by Gerardo Exequiel Pozzi (djgera) - Wednesday, 29 July 2009, 08:09 GMT
@Roman: OK, thanks for the comments and links. Reading a bit I think that the problem is before kernel 2.6.26 (and how is configured, and/or what type of RTC is used). So if this patch is applied and the system is running old kernel, will be a problem. Anyway the current hwclock will fallback to I/O (isa) if not rtc is found.

@Thomas, Thats is weird (udev vs time), here works fine, but only god knows...

Seems that applying it is more prone to problems for users running older kernels (Considering that Arch "support" 2.6.18)
I can change the patch, but is introducing more complexity than simplifying (this not the original idea) Comparing "cons" vs "pros"... Is rasonably to closing this.

Thanks.
Comment by Gerardo Exequiel Pozzi (djgera) - Wednesday, 29 July 2009, 18:22 GMT
  • Field changed: Status (Assigned → Researching)
  • Field changed: Severity (Low → Very Low)
@Thomas, aja! Yes, the commit is "25811ba" and fixes  FS#8665 . Creating a side-effect with "make" when use /dev/null can report warnings about future timestamps.
Comment by Gerardo Exequiel Pozzi (djgera) - Sunday, 08 November 2009, 23:35 GMT
Request closure for now, if any better idea without adding complexity or problems appears, in future I will request reopen.
Aditional note: current udev create some initial needed devices to start like /dev/null if not exists. And maybe this will be superceded or improved by devtmpfs.

Loading...