My first post so a brief introduction: I began using OI in 2015. Prior to that I used FreeBSD and Linux for years. I am a Software Engineer but not a C/C++ programmer.
On 20160831 I dry-run tested the Demo 7 from Toomas Soome on two different kinds of PC's plus a notebook. (It took me this long to respond because I didn't want to register publicly to post issues at illumos.org!) But my development PC cannot currently use any illumos kernel distro. I tested the USB live booting experimentals, except GNOME, on my two PC's and a borrowed Intel i5 32bit CPU ASUS touch screen notebook with 1 GiB RAM. All worked for the notebook and my PC #1, but all failed for my PC #2. Recently I re-tested more in-depth on my PC's. ======================================== (from Debian v8.5 amd64 on PC #2) $ sudo dd bs=1M sync if=OI_Loader_D7_Text_X86.usb of=/dev/sdd (Note: I deliberately did not allow internet connection during testing.) PC #1: (from illumos) $ dmesg | less cpu0,1,2,3: AMD FX-4130 Quad-Core Processor cpu0,1,2,3: x86 (chipid 0x0 AuthenticAMD 600F12 family 21 model 1 step 2 clock 3817 MHz) usba: USB 2.0 device 4 GiB RAM DDR-3 1333 Nvidia GeForce 430, GF108 HDMI Audio Two minor problems that bother but don't prevent my usage: 1) WHY does any OI installer try to silently connect to the Internet from the Install menu without my permission? It is bad policy for a home PC user. I would prefer being informed first and giving consent or else me doing a manual download of what is needed and providing it for the installer to find later. At the Demo 7 installer welcome menu for shell prompt it attempts to enable DHCP service silently and only alerts me if there is no rge device for ethernet or nwamd wireless access to internet. Seconds later I am alerted it tried silently but failed to connect. These messages spam after the "Welcome to the OpenIndiana illumos-gate installation menu ... Please enter a number [1]:" on the same line as the prompt. I can assume they are system log messages but a novice has left confused. It displayed this first impression: "Please enter a number [1]: Sep 25 02:28:16 openindiana nwamd[1002]: 9: start_dhcp: ipadm_create_addr failed for rge0: Address object already exists" (from illumos) $ dmesg | less rge: NOTICE: rge0: Using MSI internet type mac: NOTICE: rge0: registered nwamd[1002]: nwamd_set_unset_link_properties: dladm_set_linkprop failed: operation not supported nwamd[1002]: nwamd_down_interface: ipadm_delete_addr failed on rge0: Could not communicate with dhcpagent. in.routed[1822]: rge0 has a bad address 0.0.0.0 nwamd[1002]: start_dhcp: ipadm create_addr failed for rge0: Address object already exists 2) Specifically for the Demo 7 of btxForth it has slow screen updating for the default terminal type sun-color (same result for the dtterm, xterm, vt100 types). But that is acceptable. Personally I prefer white text on black background and the btxForth lets me change it via tem variables but I cannot find an edit command or cat to create a file such as my own /boot/loader.conf.local to override the default settings. I really appreciate the "ok vesa set 1280x800x8" capability. This loader is much better than grub legacy. (Minor point: the currently required fast transition from black background to white at the startup prevents me from reading any of the btxForth startup information. White background is blinding to me anyhow.) (Minor point: a confusing superfluous comma typo is shown after "more," when "ok help more".) (Minor point: Forth word draw-beastie could be renamed to draw-logo or draw-os because that is what is intended for illumos. There is another illumos kernel picture I prefer which lacks the white background phoenix.) In fact there is a possible bug lurking that I witnessed by accident. I was testing the shell command line option. I changed directory to / from /root. Then I typed 'du' command. Combined with the slow screen updating this took five to ten minutes to complete. Frustrated, I tried ctrl-c to stop the process but that did not work. So I tried ctrl-break, shift-break, alt-break, escape and finally ctrl-z which (temporarily) stopped the job. However I was not allowed to then exit the shell to go back to the menu by typing 'exit'. 'man du' did not help. I tried 'top' which is crippled (can't scroll down to see more processes) compared to Linux GNU top. I hit 'k' to attempt to kill the du process given the pid number shown but that repeatedly did nothing. So I quit top then hit ctrl-z to unstop the job and waited until completed. There was a white block cursor at the end of it all which I assumed meant I could hit enter to get the command prompt back. I was mistaken due to some strange buggy behavior. It immediately went into the default install [1] procedure which I never selected. During the end game of scrolling from du I had noticed the install menu was displayed again but it scrolled up and away. I suspect the keyboard buffer is not flushed when exiting the shell process. But I had not commanded it to exit the shell again after my only attempt had failed during the stopped job. I had only hit the enter key once during all the scrolling BS and that was at the very end. Maybe the "enter" key is a hotkey by mistake here. Long story short: I did not go any further to install the Demo 7 to erase my only working OI from 2015. It was a dry-run. I will try again later. I really need "OI text minimal" to safely install to my PC #2. I am very interested in building an OI minimal for USB pen so I can customize it with my choice of system tools and userland. Unfortunately I don't know what is in the OI minimal experimental yet nor what to remove from the kernel for targeting home PC user class. Any good advice would be appreciated here. Unlike commercial users, home users don't usually have mirrored images and backups on separate disks. ZFS can help on this matter if ZFS can fit in a 1 GiB RAM home PC and not allow itself to be easily fooled into erasing everything. Due to my unfamiliarity with using and trusting ZFS, I am worried about ZFS too easily silently sending a copy of my disk image to a remote user. As a home PC user I never want a remote user to be able to access my PC. I would prefer Unix on a FAT32 file system on USB pen so the user does not need to weird format nor accidentally erase unrecognized weird Unix format from Windows. Thus at least the user data zone should be recognized as FAT32 FUSE to alert the user not to erase the USB pen. ========================================= PC #2: All OpenIndiana releases and experimentals tested since 2015 fail to install. (from Debian v8.5 amd64) $ dmesg | less DMI: MICRO-STAR INTERNATIONAL CO., LTD MS-7576/785G-E53 (MS-7576), BIOS V3.11 05/24/2011 smpboot: CPU0: AMD Athlon II X4 640 Processor (fam:10, model: 05, stepping: 03) [drm] Initialized drm 1.1.0 20060810 [drm] radeon kernel modesetting enabled. [drm] initializing kernel modesetting (RS880 0x1002:0x9710 0x1462:0x7576). [drm] detected VRAM RAM=64M, BAR=64M [drm] RAM width 32 bits DDR [drm] radeon: 64M of VRAM memory ready [drm] radeon: 512M of GTT memory ready. [drm] Loading RS780 Microcode radeon 0000:01:05.0: firmware: direct-loading firmware radeon/RS780_pfp.bin, RS780_me.bin, R600_rlc.bin [drm] Initialized radeon 2.39.0 20080528 for fb0:radeondrmfb frame buffer device 0000:01:05.0 on minor 0 8 GiB RAM DDR-3 1333 Actual failure point during any OI installer attempt is right after it first displays the uname -a and Oracle copyright but before the hostname line can be shown. For example with the OI experimentals 2016.04 (actually released last month): GRUB v0.97 1) unix 2) unix -B livemode=vesa 3) unix -B livemode=text All three hang at the block cursor after displaying "SunoOS Release 5.11 Version illumos-2f7f7a6 64-bit" "Copyright (c) 1983, 2010, Oracle and/or its affiliates. All rights reserved." There is no way for me to know what any OI experimental is trying to do next here. Please, can we always get a notice in the boot menu to press a special key to force displaying what the installer is about to attempt - in order to help us diagnose an issue? I would suggest using the Scroll Lock key because of the LED feedback and otherwise useless purpose for a home PC user. At a minimum I would like to know what module or driver it might try to install in case it is causing the hang. (Drivers are not an issue with Demo 7 which has a nice drivers step during the text install at the welcome menu on my PC #1.) But the hang is present at the same place for all OI that I tested. It is the illumos kernel and not the distribution which is at fault. ================================================= Special note about the Windows 7 on the notebook. After I had finished all dry-run testing of OI experimentals and D7 on the ASUS i5 then I simply removed the USB pen during a normal reboot. Win7 got stuck in a boot loop that it could never recover from. I checked the disk and there were no errors and no changes made by me. Win7 simply detected that it was not previously in control and refused to load itself from that moment on. The owner I borrowed the notebook from was not happy but understood. She gave it to her son who had to erase Win7 and put some other software on it then gave it as a toy to his girls. Unfortunately she bought another notebook with a Windows OS on it but I am hesitant to ask to borrow it again for dry-run testing illumos. The implication is that I screwed it up even tho I know it was deliberate from Microsoft. ------------------------------------------- illumos-discuss Archives: https://www.listbox.com/member/archive/182180/=now RSS Feed: https://www.listbox.com/member/archive/rss/182180/21175430-2e6923be Modify Your Subscription: https://www.listbox.com/member/?member_id=21175430&id_secret=21175430-6a77cda4 Powered by Listbox: http://www.listbox.com