History of FuguIta/BBS/10 (No. 15)

• History
• View the diff.
• View the diff current.
• View the source.
• Go to FuguIta/BBS/10.
  • 1 (2020-07-22 (Wed) 15:39:07)
  • 2 (2020-07-22 (Wed) 15:45:43)
  • 3 (2020-07-23 (Thu) 00:02:58)
  • 4 (2020-07-23 (Thu) 12:54:59)
  • 5 (2020-07-23 (Thu) 23:05:59)
  • 6 (2020-07-24 (Fri) 11:28:18)
  • 7 (2020-08-01 (Sat) 03:37:52)
  • 8 (2020-09-08 (Tue) 14:03:49)
  • 9 (2020-09-08 (Tue) 15:05:37)
  • 10 (2020-09-08 (Tue) 15:05:37)
  • 11 (2020-09-24 (Thu) 22:14:55)
  • 12 (2020-09-24 (Thu) 22:14:55)
  • 13 (2020-10-05 (Mon) 10:58:51)
  • 14 (2020-10-05 (Mon) 10:58:51)
  • 15 (2020-10-16 (Fri) 00:40:21)
  • 16 (2020-10-24 (Sat) 00:30:55)
  • 17 (2020-10-24 (Sat) 22:05:09)
  • 18 (2020-10-25 (Sun) 15:49:03)
  • 19 (2020-10-26 (Mon) 10:54:11)
  • 20 (2020-10-26 (Mon) 19:31:47)
  • 21 (2020-11-03 (Tue) 23:37:38)
  • 22 (2021-01-09 (Sat) 23:14:21)
  • 23 (2021-01-09 (Sat) 23:14:21)
  • 24 (2021-01-30 (Sat) 07:41:08)
  • 25 (2021-02-01 (Mon) 11:30:26)
  • 26 (2021-02-01 (Mon) 11:30:26)
  • 27 (2021-02-16 (Tue) 19:03:34)
  • 28 (2021-03-23 (Tue) 18:01:06)
  • 29 (2021-04-24 (Sat) 18:32:10)
  • 30 (2021-04-25 (Sun) 15:02:05)
  • 31 (2021-04-27 (Tue) 15:35:19)
  • 32 (2021-04-28 (Wed) 07:42:25)
  • 33 (2021-05-01 (Sat) 09:54:24)
  • 34 (2021-05-02 (Sun) 07:41:16)
  • 35 (2021-05-02 (Sun) 10:47:10)
  • 36 (2021-05-11 (Tue) 14:30:43)
  • 37 (2021-05-12 (Wed) 11:12:57)
  • 38 (2021-05-14 (Fri) 12:22:21)
  • 39 (2021-05-16 (Sun) 01:36:06)
  • 40 (2021-05-16 (Sun) 06:42:36)
  • 41 (2021-05-16 (Sun) 06:42:36)
  • 42 (2021-05-16 (Sun) 14:09:30)
  • 43 (2021-05-21 (Fri) 04:40:06)
  • 44 (2021-05-21 (Fri) 13:40:06)
  • 45 (2021-05-21 (Fri) 04:40:06)

---

Your message?

Notice: To avoid posts by spam, a message body without the word '#FuguIta' is rejected.
Please include the word in your message text.

---

Name:
Subject:

Live Update for FuguIta LiveUSB†

kaw (2020-09-24 (Thu) 22:13:14)

I'm currently working on a Live Update for #FuguIta LiveUSB.
This is in beta testing.

To update with this utility, place the gzipped ISO image (not the *.img.gz file!) and the MD5 checksum file in the current directory.
Then run the fiupdate command.

# cat /usr/fuguita/version
6.7-amd64-202008261
# ls -l
total 606528
-rw-r--r--  1 root  wheel  310528773 Sep 24 18:47 FuguIta-6.7-amd64-202009041.iso.gz
-rw-r--r--  1 root  wheel      11660 Sep 24 18:47 MD5
# fiupdate 202009041

fiupdate - Live Updater for FuguIta LiveUSB
  Version/Arch: 6.7/amd64  (FuguIta-6.7-amd64-202008261)

Note:
This sorftware is now under beta test.
Please use this at YOUR OWN RISK.

We recommend that you execute this command with fresh boot (boot mode 0 or 1).

Or you should quit all application softwares and save all your data
before you update this FuguIta device.

Are you sure? [y/N] -> y

Checking:
     environment: ok
	  mounts: ok
  existing files: ok
	checksum: (MD5) FuguIta-6.7-amd64-202009041.iso.gz: OK
all done, OK.

extracting FuguIta-6.7-amd64-202009041.iso.gz...
 296MiB 0:00:48 [6.12MiB/s] [================================>] 100% ETA 0:00:00

Now ready to update FuguIta-6.7-amd64-202008261 to FuguIta-6.7-amd64-202009041.

This machine will reboot immediately after update completed.

Do you proceed? [y/N] -> y
stopping all daemons...
cron(ok)
ntpd(ok)
pflogd(ok)
slaacd(ok)
smtpd(ok)
sndiod(ok)
sshd(ok)
syslogd(ok)
overwriting uniprocessor kernel...
8.68MiB 0:00:03 [2.31MiB/s] [================================>] 100% ETA 0:00:00
overwriting multiprocessor kernel...
8.71MiB 0:00:03 [2.23MiB/s] [================================>] 100% ETA 0:00:00
overwriting filesystem image...
 894MiB 0:04:25 [3.36MiB/s] [================================>] 100% ETA 0:00:00

update completed. now rebooting...
syncing disks... done
rebooting...

We welcome your trial report.

• revised this script, See https://jp1.dl.fuguita.org/test/ . -- kaw 2020-10-05 (Mon) 10:58:51
• #FuguIta-6.7-*-202010071 has this. (Still can't handle tarball, sorry arm64 user) -- kaw 2020-10-16 (Fri) 00:40:20

Name:

isotop on FuguIta†

kaw (2020-09-08 (Tue) 14:07:35)

isotop is a script to configure and customize vanilla OpenBSD into ready-to-use desktop environment.

I have modified isotop to be able to run on #FuguIta.

To run isotop.sh on FuguIta after download it, apply this patch isotop-665-fi.sh.diff .
Then run install script as a root.

$ ftp https://framagit.org/3hg/isotop/raw/master/src/isotop.sh
$ patch < isotop-665-fi.sh.diff
# sh isotop.sh

Here's a screenshot of FuguIta configured by isotop.

Name:

Archive video of FuguIta Talk on OSC2020 Niigata†

kaw (2020-08-01 (Sat) 03:37:52)

Video of Talk about #FuguIta at OSC2020 Niigata has been published:
https://www.youtube.com/watch?v=4wPSr1WPH3U

And the document is at 河豚板#OSC2020

• To those who do not understand Japanese:
  Auto-translated subtitles have been greatly improved.
  
  Talk about OpenBSD and #FuguIta -- kaw 2020-08-07 (Fri) 14:22:35

Name:

OSC2020新潟オンラインで河豚板の発表を行います。
Talk about FuguIta at OSC2020 Niigata†

kaw (2020-07-22 (Wed) 15:27:51)

Open Source Conference 2020 Online/Niigataが7月25日(土)に開催されます。
今回はオンラインでの開催になります。https://ospn.connpass.com/event/181888/ の開催概要ページよりZoom参加に登録頂くか、YouTube Liveにてご視聴下さい。

Open Source Conference 2020 Online/Niigata will be held on Saturday, July 25th.
This event will be held online and I will have a 15 minute talk about #FuguIta in the name of EBUG (Echigo BSD Users Group).
Please register for Zoom participation from the overview page of https://ospn.connpass.com/event/181888/ , or watch it on YouTube Live.

• YouTube Live Stream will be broadcast at https://www.youtube.com/watch?v=XSKG1rK3nKQ .
  The live stream will begin at 12:30JST (03:30UTC) on 25 Sat, July 2020.
  There will be three BSD-related talks:

  Time (UTC)	Title	Affiliation	Speaker
  06:00-06:15	Introduction of NetBSD	Japan NetBSD Users Group (JNUG)	Jun Ebihara
  06:15-06:30	Introduction of FuguIta	Echigo BSD Users Group (EBUG)	Yoshihiro Kawamata
  06:30-06:45	Introduction of vyos-cli: VyOS CLIs for Ubuntu	Echigo Network Operators Group (ENOG)	Masakazu Asama (also member of EBUG)

  These talks will be available as archives at a later date.

  -- kaw 2020-07-22 (Wed) 23:46:40

Name:

Boot mode difference†

Fugu (2020-07-19 (Sun) 15:51:32)

Hello Kaw:

Is there a difference in boot mode of OpenBSD and #FuguIta? I have a Dell PC where OpenBSD is very slow without disabling acpimadt whereas #FuguIta has no such problems.

Any guidance will be gratefully received.

Thank you.

• The kernel config of #FuguIta is almost same as GENERIC.
  Differences are as follows:

  nimbus9 amd64 # diff -u GENERIC RDROOT
  --- GENERIC    Mon Jul 20 12:43:03 2020
  +++ RDROOT     Mon Jul 20 12:40:08 2020
  @@ -49,7 +49,7 @@
   option                UDF             # UDF (DVD) file system
   option                MSDOSFS         # MS-DOS file system
   option                FIFO            # FIFOs; RECOMMENDED
  -#option               TMPFS           # efficient memory file system
  +option                TMPFS           # efficient memory file system
   option                FUSE            # FUSE
   
   option                SOCKET_SPLICE   # Socket Splicing for TCP and UDP
  @@ -88,7 +88,7 @@
   pseudo-device nmea    1       # NMEA 0183 line discipline
   pseudo-device msts    1       # MSTS line discipline
   pseudo-device endrun  1       # EndRun line discipline
  -pseudo-device vnd     4       # vnode disk devices
  +pseudo-device vnd     6       # vnode disk devices
   pseudo-device ksyms   1       # kernel symbols device
   #pseudo-device        dt              # Dynamic Tracer
   
  @@ -134,7 +134,12 @@
   option                NTFS            # NTFS support
   option                HIBERNATE       # Hibernate support
   
  -config                bsd     swap generic
  +config                bsd root on rd0a swap on wd0b and sd0b
  +option                RAMDISK_HOOKS
  +option                MINIROOTSIZE=3800
  +option                NKPTP=5
  +
  +pseudo-device   rd              1 # ramdisk
   
   mainbus0 at root
   
  nimbus9 amd64 # 

  So I don't know why it behaves differently.
  
  Since the storage device used is different between regular OpenBSD(internal HDD, SSD...) and FuguIta(DVD, USB Flashdrive, SD card...) there may be something wrong with the built-in disk device.
  
  In any case, I think we need to see what's slowing down regular OpenBSD.
  • disk I/O
  • stray interrupts
  • anything else?

    -- kaw 2020-07-20 (Mon) 12:50:48

• Thank you for the detailed response. It has to do with buggy bios - APIC. Thank you for sharing #FuguIta with us. -- Fugu? 2020-07-21 (Tue) 09:49:27
• With following procedure, you can check whether the internal disk or its corresponding BIOS function causes the problem:
  
  • In internal disk, make a directory named "ISO" at the root of a partition which is FAT, NTFS, Linux ExtFS or OpenBSD FFS.
  • Put a FuguIta's ungziped ISO image into "ISO" directory.
  • Boot FuguIta with LiveDVD or LiveUSB.
  • The internal disk will be listed as follows:

    scanning partitions: cd0a sd0i sd0j
    FuguIta's operating device(s): cd0a sd0i
    Which is FuguIta's operating device? [default: cd0a] -> sd0i  # in case that sd0 is internal disk

  • Select internal device, then continue booting.

If FuguIta runs slow as OpenBSD with this procedure, the problem could lay on internal disk device. -- kaw 2020-07-21 (Tue) 16:28:11

• I had quite a few ideas to test this:
  
  • Chroot install
  • chain load from a different OS
  • install via different system and move hard drive
  • use a different boot manager
  • install #Fuguita and hack via ramdisk

and one other escaping attention now.
To this end, I started reading through the show and list variables in the UKC config and found that #FuguIta has a lot more that get initialized than vanilla OpenBSD. When that did not work out, I once again disabled acpimadt, installed the operating system and waited for it to boot. The scenario now stood at reordering libraries (library_aslr). I let it complete the first time and then disabled the service via rcctl. This made a dramatic impact on reboot.
Presently, after having run sysupgrade (following -current), the performance is acceptable as good as on a ThinkPad.
My speculation is to do with the execute bit. I haven't read the source code to be certain of this but would appreciate your knowledgeable thoughts on it. I will try the above solution if video and browser performance in unacceptable with my continued testing. -- Fugu 2020-07-21 (Tue) 19:37:36

• Summary Update: During upgrade/install, the acpimadt needs to be disabled during runtime only. All things work except for the cpu temperature which is stuck at a very high inordinate value. The only similarity found is when the install medium is flash disk. Fans, hibernate, sleep - power management; userland setup; networking have no problems. Interestingly, #FuguIta also spends some time in kernel reordering but being a ram-mounted filesystem, the time spent is comparatively smaller. This is a two-hard drive machine and OpenBSD live USB is now getting stuck before Install/Upgrade/Interrupt message prompt (custom part of #FuguIta) - seems to be directly correlational to number of partitions. Pressing power option still gives back the prompt. Although the device will not be high workload intensive, lack of temperature monitoring (#FuguIta live also lists the same numbers) is unnerving. -- Fugu 2020-07-22 (Wed) 14:55:26
• Thank you for the detailed information.
  Is vanilla OpenBSD slowing down because the CPU temperature is considered high and the CPU clock frequency is low?
  If let apmd enabled and manually change the CPU clock with the apm command, can you see anything relevant? -- kaw 2020-07-23 (Thu) 12:54:58
• Hello Kaw: What I meant to say is everything works as in a supported machine except the sensor for reading the cpu temperature is broken.
  My guess is that this failure to initialize during boot is what causes the problem. #FuguIta does not display this behavior since the boot process has custom setup at this exact point where OpenBSD seemingly hangs. Just before shell access, there is an info which says XSDT entry 8 is corrupt. Not sure if OpenBSD devs would want this bug report since this is BIOS initiated.
  When I say userland setup, the fans (two of them) scale according to workload. I stress tested by encoding a video and simultaneously running a 4K Blender video. Everything works once the module is disabled during install/upgrade (workaround with sysupgrade -n) - one time activity only.
  I genuinely look forward to seeing you presenting #FuguIta this weekend. The only time I had these issues was with booting lilo on Slackware though FreeBSD works without complaint.
  Best wishes. -- Fugu 2020-07-23 (Thu) 23:05:59
• Thank you Fugu, see you in the Live Stream! #FuguIta -- kaw 2020-07-24 (Fri) 11:28:18

Name:

---

Former articles are at FuguIta/BBS/9.

---

Return to Top