user1160711's questions

The monitor I have is old and has often acted flaky, so I'm sure the ultimate blame resides with it, however setting the drm.edid_firmware= parameter on the kernel command line is in theory supposed to override whatever nonsense kernel modesetting, X11, and Wayland come up with, yet it doesn't seem to work. Do I have to specify a connector even though the docs say leaving it off will override the setting for all connectors?

Symptoms: When booting, it reaches the point where kernel modesetting happens and the text size it switches to makes it obvious it has picked the wrong resolution. When X comes up the wrong resolution is obvious. Under X, I can turn off the monitor and turn it back on, and when it comes back on, it always has the right resolution. Under Wayland if I do that it almost always is either a black screen or a jumbled mess of wiggly lines.

This is ubuntu 20.04 with amdgpu on a display port connecting to a UHD monitor.

I have no idea where to find Waylang logs, but I see gibberish about falling back to old probe method in the Xorg.log. What is it probing if I've given it an EDID file?

[    12.737] xf86EnableIOPorts: failed to set IOPL for I/O (Operation not permitted)
[    12.739] (WW) Falling back to old probe method for modesetting
[    12.740] (WW) Falling back to old probe method for fbdev

I have a Kensington Expert Wireless Trackball Mouse, it is very close to impossible to use if I can't configure one of the (4) buttons to implement drag lock. I was just trying wayland on 20.04 and there does not appear to be any way to map buttons and enable one of the buttons as a drag lock button. Am I missing something, or has this been utterly ignored for all the years wayland has been in development? I'd be happy to buy some trackballs for wayland developers if it will increase their motivation to get this working :-).

I've been installing stuff on a new ubuntu 20.04 computer to run my 3D printer. I have a crossover cable between the smoothieboard in my printer and an ethernet connector on my ubuntu system. I've configured the ethernet to manually be 192.168.2.1/24 and configured the smoothie to manually connect to 192.168.2.2/24 with 192.168.2.1 as a gateway. For some wild and crazy reason, if I go to 192.168.2.2 in firefox, I get an apache2 success screen. I know the smoothieboard doesn't have a copy of apache2 on it. If I telnet to 192.168.2.2 I do indeed get a smoothieboard telnet prompt. What lunatic setting is redirecting my web access attempt back to the local system? Is this some haproxy insanity from installing OctoPrint?

I installed 20.04 on a new machine using an old monitor connected to HDMI. After it worked, I moved it to a different location and plugged a UHD resolution monitor into the display port connection on this machine. When it boots, I get a BIOS message about how to enter the boot menu or setup on the monitor, then it just sits there till the graphical interface comes up (which does display fine on the new display port monitor). No ubuntu logo or anything during the boot, just the text that the BIOS put on the screen the whole time. Is there some way to convince it to show boot messages on the display port? Grub options to change or something?

Machine: ASRock AMD Barebone 4X4 BOX-4800U with Ryzen cpu. (UEFI only)

Install media: ubuntu-20.04-desktop-amd64.iso

If I install from scratch on a new UEFI machine, can I just copy everything except /boot from an older machine being replaced? Or is there BIOS and kernel related stuff somewhere outside of /boot that I'd want to avoid clobbering? (both are running 20.4). (Perhaps /etc/default or someplace like that?)

Every once and a while, out of the corner of my eye I see the Ubuntu 20 system I'm using to run my 3D printer flash a notification at the top of the screen, but it always disappears before I can see what it says. Is there an app to record notification history I could run so I could review these things and find out what is going on? I subspect it is a momentary problem with wifi, but can't prove it.

Just got the 5.8.0-34 kernel update, and wireless doesn't work with the BCM43224 wireless in this old DELL zino HD mini-pc. It worked fine when I first installed 20.04 with no magic required, and it works fine if I reboot into the older 5.4.0-59 kernel I had previously. Has ubuntu dropped support for this, or is it a kernel/build bug?

Here is the relevant lshw from the 5.8 kernel:

  *-network UNCLAIMED
   description: Network controller
   product: BCM43224 802.11a/b/g/n
   vendor: Broadcom Inc. and subsidiaries
   physical id: 0
   bus info: pci@0000:02:00.0
   version: 01
   width: 64 bits
   clock: 33MHz
   capabilities: pm msi pciexpress bus_master cap_list
   configuration: latency=0
   resources: memory:fe8fc000-fe8fffff

In the 5.4 kernel I see this:

  *-network
   description: Wireless interface
   product: BCM43224 802.11a/b/g/n
   vendor: Broadcom Inc. and subsidiaries
   physical id: 0
   bus info: pci@0000:02:00.0
   logical name: wlp2s0
   version: 01
   serial: c0:cb:38:27:23:c0
   width: 64 bits
   clock: 33MHz
   capabilities: pm msi pciexpress bus_master cap_list ethernet physical wireless
   configuration: broadcast=yes driver=wl0 driverversion=6.30.223.271 (r587334) latency=0 multicast=yes wireless=IEEE 802.11
   resources: irq:18 memory:fe8fc000-fe8fffff

So it looks like the driver is gone, but I didn't do anything to install that driver when I installed 20.04, it just worked out of the box.

Setting up keys for a local repository on a new ubuntu 20.10 virtual machine, I got a message that apt-key add was deprecated and I should read the apt-key(8) man page. The apt-key(8) man page is a collection of words strung together, but if it contains information I can't winkle it out. Can anyone tell me what, exactly, I should type on my terminal instead of:

apt-key add name-of-file

The command does apparently still work after honking at me, so I was able to proceed, but would like to know what I'll need to do in the future.

I have a program that calls mmap() on /dev/zero to make some new private pages. It works fine if I use PROT_READ|PROT_WRITE but if I use PROT_EXEC instead of write, it always gets operation not permitted (errno 1). This has worked forever, it works on even newer kernel versions in fedora, but in ubuntu 20.04 with kernel 5.4.0-26-generic (x86_64) is always gives the error.

There must be some way to mmap read/execute pages since the dynamic loader still works.

Anyone know what magic new thing I need to do to make the map work again?

I have tried many suggestions, and none of them work. I don't want my Ubuntu system to touch the network in any way shape or form to check for updates until I explicitly run an update check.

I've set the check to "Never" in the GUI app with the repos. I've disabled every systemd timer that says something about update. Still I get update-notifier notifications popping up from time to time.

What on earth do I have to do to stamp out this scourge on Ubuntu 19.04?

Note that I don't want to just disable the update-notifier popup; I want to disable all network traffic doing the update check.

More details on what I have already done:

systemctl disable apt-daily-upgrade.timer apt-daily.timer

Run Software & Updates, go to Updates tab set "Automatically check for updates" to "Never".

There is information which only appears in a notification for an app I'm using (specifically, the open port number in the private internet access app). Super+V will show me the notification from when the app started, so the information has to be laying around somewhere. Anyone know of a way to extract the notification text to use in a script? Some dbus message I can send to get the text perhaps? Some file the desktop stores the notifications in? Or a way to replace the standard notification app with one that could log the info somewhere?

I have just been painfully discovering that gcc apparently generates -fpic code by default and links with -fPIE by default (On ubuntu 17.04). This completely screws up thousands of tests I run with scripts and makefiles used by lots of different linux distros. Is there any global or per-user way to turn off these defaults and make the compiler backward compatible with the behavior it has had for decades? I'm not interested in tracking down every compilation in thousands of scripts that knows the default is not -fpic, etc. An environment variable or two perhaps?

I have a perfectly wonderful multi-boot setup with a stand alone grub2 partition booting from the MBR and from that grub2 menu I can boot other operating systems on other partitions (using, for example, the configfile command).

This has been working perfectly till an ubuntu update came along today and overwrote the MBR, thus disabling my wonderful stand alone grub.

I got it back, but what I want is to never have ubuntu do that again. How do I make sure ubuntu keeps its grubby fingers off my MBR?