Troubleshooters.Com®, Linux Library, and T.C Void Linux Subsite Present:

EFI Dual Booting Void Linux With Windows On a Dell Inspiron 15 5570

Contents:

Introduction

In mid July, 2018, I bought a Windows 10 equipped Dell Inspiron 15 5570 laptop from Costco. I buy my laptops from Costco because if I don't like them, I have 90 days to return them. Plenty of time to make sure they run Linux.

I did two things the day I bought it: I charged the battery, and I made sure it was possible to turn off the monopolistic "Secure Boot". Had that been impossible, I'd have returned it that night.

Some might question why I bought a Windows computer at Costco instead of getting a Linux-preloaded computer from a specialty shop. Did I mention this 12GB RAM computer cost $399.95, including 2 year warranty? Don't try that at System 76 or Penquin Computing.

One downside of not buying pre-loaded from a specialty shop is that Windows and its diagnostic utilities must run on the computer in order that your warranty be honored. This installation had to be a dual boot.

Although my Dell has a "legacy mode", I decided to do the entire boot in UEFI mode. Launching Windows from Grub2 isn't something I'd look forward to. This was my first Windows/Linux dual boot installation this decade, my first detailed UEFI boot installation dualboot ever. In hindsight the process was fairly straightforward, but at the time there seemed so many alternatives that it took me more than a day of work to complete the installation. Once I even had to use the Dell restore to factory state because I borked the C: drive partition and maybe another one.

By the way, my first impression of this Dell is it's a pretty good machine, and that I'll be buying Dell more often. They seem less wedded to Microsoft than others I've seen. According to the instructions that came with my computer, there's even a Ubuntu version of it available.

First, Turn Off Secure Boot

My first move was to turn off secure boot. It was pretty easy to do: About 25 minutes: 15 to discover that the F12 was the key that invoked the "BIOS Setup", and 10 minutes exploring the BIOS Setup to find Secure Boot. It was clearly marked, and easy to set to "off".

In my world, having an off switch for Secure Boot is a must. Yes, I know, I know, certain distributions have bought secure boot keys from Microsoft and now work with Secure Boot. Umm, no thanks, if I can't have a DIY, my choice OS, I'll just get a Mac.

And there are Secure Boot hacks and shims that work just fine if you get every one of a hundred variables just right. No thanks. Fortunately, it appears manufacturers are still in at least some cases equipping their computers with Secure Boot off switches.

There's an urban legend that you can't boot Windows on a computer that has Secure Boot turned off. That wasn't my experience with this installation.

Observe the Ask Me Anything Field

In Windows 10, there's a field called "Ask me anything" on the bottom bar,left of everything except the Start Button. When you type text into this field, you get a list of information and programs interspersed. This field makes Windows 10 much more discoverable than its chicklet hunt predecessor, Windows 8.1. Discover how it works. Get used to it. Use it often.

Remove All Timeouts

Installing a dual boot Linux isn't the time for going green. If you go to bed while a process is running, you want to see the result in the morning, not a post-hibernate black screen.

So go into the Windows bottom bar's "Ask me anything" input field and type in power and sleep, and view the list until you find the method to adjust the screen blank period and the sleep period. Once there, for situations running on AC power, tell the computer never to black the screen or sleep

Once your installation is complete, you can readjust this to fit your needs.

Make Restore Plans

At least to the extent you can. I'd read something about if Windows fails to boot, Dell gives you a restore program, based on two restore partitions. The first thing to know is that you *never* mess with any partitions except the Windows C drive partition. If you alter or erase one of the restore partitions, it's likely unrecoverable if you bork Windows. The only safe location of your new Linux is in an area freed up from shrinking the C: drive partition.

If you do nothing to damage the restore partitions, and maybe the UEFI partition, you should be able to use the Dell's original factory restore to restore to factory state. But there are no guarantees. This is the route I followed: I just made it a point not to mess with partitions other than C:

Theoretically safety dictates you back up the partition table, any UEFI partitions, and any restore partitions. If you do this, lay down the backed up front of the disk, and are able to put the restore partitions exactly where they were, you can restore.

Perhaps the easiest way to guarantee restoreability is to boot the machine on a System Rescue CD disc, plug in a USB3 external drive to a USB3 port, and use ddrescue to copy the entire drive device to a file on the external drive. Obviously, the external drive must be significantly bigger than the one in the computer.

Another alternative, if you have the mechanical skills and if opening the computer doesn't void your warranty, is to remove the original drive after testing that Windows does indeed work, and keep it in a safe place. Then put in the new drive, hopefully a fast SSD, and install Linux on the new drive alone, without dual booting. Easy, lots of room, lightning fast performance, with the ability to restore the computer back to its original state by taking out the SSD and putting back the original, so that your warranty remains in place.

A slight modification of the preceding is to buy a bigger SSD than the original drive, and ddrescue the original over the SSD. Now you can dual boot to the SSD, with the original as a long term restore alternative.

Whatever you do, don't blindly alter or move partitions.

Record the Original Partition Table

Run a Linux live CD like System Rescue CD containing the fdisk program, perform the following command, as root, and output it to a mounted thumb drive:

fdisk -l /dev/sda

Now perform the following command, as root (to get the UUID info) and output it to a file on the mounted thumb drive:

lsblk -l -o +UUID /dev/sda

Make sure to transfer those files to another computer for safekeeping, and if possible, print them (landscape is preferable).

When you're in the thick of it, creating partitions, then realizing you're wrong and re-deleting them, and watching as your free areas seem to slide around like oysters, and even having device names change, you can use these two files to determine which partition is which, because if you delete the wrong one, it will be a bad day indeed.

When creating partitions, try very hard to create them significantly bigger or smaller than the existing ones, so you can identify by size. But if you can't identify by size, read your partition table backups to figure out exactly what's going on.

If your laptop happens to contain two disks, you'll need to do the same for /dev/sdb. Before going further, make sure this file is backed up on

Shrink the C: Partition

As mentioned, on most laptops your only opportunity to find space for a Linux partition is in free space left behind when you shrink the C: drive partition.

Your first step is to defrag the C: drive. Using Windows 10, what I did was type defrag into the "Ask me anything" text field on the bottom bar, and how to defragment came up near the top. First you analyze, then you defragment (optimize). The defragment part could easily take a couple hours or more.

Once the C: drive partition is defragmented, you need to shrink it. In the "Ask me anything" field type computer management, and choose the "Computer Management" program. From there, do the following:

  1. In the leftmost compartment of the Computer Management Window, click "Disk Management.
  2. Wait several seconds for the computer to respond.
  3. In the "OS(C:)" colum of the disk zero row, left click for menu.
  4. From the menu, click "Shrink Volume". After a few seconds you'll be brought to a the "Shrink C:" window, containing four fields:
    1. The "Total size" display-only field, containing the total number of bytes on the C: partition.
    2. The "Amount available for shrink" display-only field, containing the maximum number of bytes by which it can be shrunk.
    3. The "Amount to shrink" input field, where you can put how much you want it to be shrunk, up to the amount in the "Amount available for shrink" field.
    4. The "Total size remaining" display-only field, containing what the partition's size would be after subtracting the amount in the "Amount to shrink" input field. Thus, you can experiment with "Available for shrink" until you get an amount that's big enough for your anticipated Windows use. Don't forget you'll almost certainly install a browser besides Internet Exploer on your Windows system.
  5. After you've got your "Amount to shrink" to the right value, or to the highest value possible if you'd really like to shrink it more, click the "Shrink" button to accomplish the task. Almost immediately the job gets done, and you see the area on the "Disk0" row's "C: column" decrease. You'll also notice "available space following it.

If It Didn't Shrink Enough

Shrink amount is determined by the the amount of unused space at the end of the partition, and constrained by the unmoveable byte closest to that end. It's possible, therefore, that even though you have many more gigabytes of free space, you can't shed them because of an unmoveable byte. If that happens to you, you can repeat the whole process again:

  1. Power cycle the computer
  2. Analyze
  3. Defrag/optimize
  4. Shrink

If even the preceding didn't bring you down where you want to be, and remember, you don't want to starve your windows installation because it's vital for your warranty coverage, you can do it again and again.

Common Wisdom Windows Adjustments

A lot of the documentation I read said that, to dual boot Linux with Windows in a UEFI system, you should turn off "Fast Startup". So I did, as follows:

  1. Type control panel into the "Ask me anything" field in order to run Control Panel.
  2. Choose "Power Options"
  3. Choose "What the power button does"
  4. Click "Change settings that are currently unavailable"
  5. Uncheck "Fast Startup"
I don't know how necessary this is, but I did it and things worked out for me.

Congratulations

You've finished the Windows portion of the job. Unless you mess up the partitions, the rest of your install will be done with Linux and/or System Rescue CD, which is Linux.

Re-record the Partition Table

You need to record this new, adjusted partition table for future use. Follow the same instructions as you did in the Record the Original Partition Table section earlier in this document. Store on another computer and print as before. Make absolutely sure you clearly mark/name all, so there's no getting them mixed up. The two partition table printouts after the shrink are even more important than the ones before the shrink.

Create Linux Partitions

Place a modern System Rescue CD CD in the CD drive (or a modern and known good System Rescue CD thumb drive in the USB3 port), power down, turn back on, immediately press F12, and when asked where to boot from, choose the DVD drive. System Rescue CD runs.

Using the partition table printouts from just after the shrink, very slowly and deliberately decide where you're going to put partitions and how big you'll make them. I did them minimal: a 12GB swap partition followed by a single ext4 partition consuming all the useable remaining area shrunk from the C: drive.

Notice I do not need to create an EFI partition, because the original Windows installation included an EFI partition that should be used for all OSes installed on that drive.

Use sfdisk or something like it. Keeping one eye on sfdisk and another on your post-shrink printouts, slowly and carefully match things by size, begin and end location, etc. In sfdisk, create your swap partition at the beginning of the free space left by the shrink. Try to make your swap partition's size differ from all other partitions on the disk by at least 15%, to enable the size to be used for identification. Be alert: Things might move around, all over the place. This is why you have your printouts.

Making sure everything's how you think it is, and remembering the carpenter's addage "Measure twice, cut once", create your Ext4 partition from the remainder of the space.

If you find yourself having to delete a partition, be very careful you're really deleting a partition you just created. Take all the time you need. Because if you delete any of the restore partitions, you can't restore to factory. And if you delete the UEFI partition, I've heard of that kind of thing permanently bricking a computer's hardware (yeah, that's right).

Now create your Ext4 partition, once again trying to make it at least 15% different size from any other partitions, for identification purposes. I personally marked my Ext4 as bootable: I don't know whether that's helpful or just superstition, but I did it. When you're satisfied, after considerable consideration, write to disk.

Now do yet another set of partition printouts, making sure you've recorded the UUID of the new swap and Ext4 partitions.

Deal with Void Wifi

For the purposes of installing Void, things will be better and easier if you can use a network cable plugged into your computer's Ethernet port. If you can easily do that, do it and skip down to the Post Install Wifi subsection.

If you're reading this, I assume it's going to be somewhere between a hassle and an impossibility to use a network cable. So we'll make several suggestions: Perhaps one will work...

The first suggestion is to try installing via Wifi. Maybe that will work. The older your computer, the more likely it is to work. If your computer's Wifi module can run with older drivers, perhaps that will work. So by all means, try installing via Wifi, and if it works, you can skip the rest of this section and go to the Run Void Installer section.

Alternative Wifi Approaches

If your computer's Wifi didn't work with the Void installation media, well, neither did mine. There are alternatives. Although Void's an up-to-the-minute rolling-release, its installation media are nine months old at this point. If you bought a brand new laptop with brand new Wifi components, you can see the problem: Your Wifi components most likely didn't exist when the installation media was created, so it's impossible for your Wifi's drivers to be on the installation media. So you need to turn to alternatives:

Use a Wifi Dongle

Wifi dongles cost between $5.00 and $50.00, and many are specified to work with Linux. From what I hear, the ones made for Raspberry Pi systems are especially good and reliable. Get a Linux-capable Wifi dongle, plug it into a USB port, and install through it. If you already have such a dongle around the house, better still, because it's older and most likely capable of being supported by the drivers on the install media.

Use a Travel Router

Travel Routers cost between $30 and $200. Many of them have a feature which receives wifi traffic out of the air, and pumps the network packets out an Ethernet port, so if you connect a network cable between that Ethernet port and your computer's Ethernet port, you can receive Wifi. Be careful: Some of them can only do this with unencrypted Wifi: You'll probably need one that handles encryption. For any computer with an Ethernet port, these travel routers are the solution to the Linux Wifi Blues.

Visit a Friend With Wired Internet

Find a friend willing to let you use his wired Internet for a few hours, and install from there using your computer, your Void Installer CD, and your own network cable (bring a long one). Once it's installed, you never have to do that again.

Chroot Install Using System Rescue CD

If the latest System Rescue CD can interoperate with your computer's wifi, you might be able to install by booting System Rescue CD copying the files to a subdirectory, and doing a chroot install. I don't know whether you'd be able to keep your System Rescue CD providing access to your wifi once you chroot into that directory. Perhaps the drives could be made available via bind mounts made before you issue the chroot command. I've done Chroot installs before, but never on OSes not capable of running the network.

Chroot installs involve many details and many opportunities to slip up. It's definitely not a first choice. There are some good Void Linux Chroot install documents on the web.

Don't Use Void

I'll bet you dollars to donuts that Ubuntu install media will have no trouble with your computer's Wifi components. Ubuntu has been created from the bottom up to spare no expense in making the user's life trivially simple, and working with any and all hardware out there. You could use Ubuntu instead of Void. Or perhaps Debian, Devuan, Gentoo or Funtoo.

But before you do those things, consider why you wanted to install Void Linux in the beginning: An almost carefree, latest tested program, logical, simple, DIY, no-systemd operating system. If you take the No Void route, try Void again every six months. Your computer keeps getting older, and Void install media keep getting newer, so sooner or later Void will install on the computer without the slightest effort.

Post Install Wifi

You want to hear about problems? My wifi didn't work even after installation. I'm not alone. Linux Wifi has always been challenging.

Unlike most distros, Void doesn't depend on either network manager or wicd, thank goodness. Those two are so intimately woven into dbus that they seem like a convoluted black boxes.

Instead, Void operates under the principle that, if you have the proper info about a given network in /etc/wpa_supplicant/wpa_supplicant.conf, and you have dhcpcd running as a daemon, you can link to that network. Assuming good hardware and drivers, this approach almost always works.

The following is a pretty much complete network specification in /etc/wpa_supplicant/wpa_supplicant.conf:

network={
	ssid="reseda"
	psk="wouldntYouLikeToKnow"
	proto=RSN
	key_mgmt=WPA-PSK
	pairwise=CCMP
	auth_alg=OPEN
}

Nobody can remember all of that, and a lot of it changes depending on the properties of the network. Luckily, it's usually sufficient to specify the SSID and password, as follows:

network={
	ssid="reseda"
	psk="wouldntYouLikeToKnow"
}

If you're specifying an open, no password network, then the specification looks like the following:

network={
	ssid="Bates_Motel_Wifi"
	key_mgmt=NONE
	priority=5
}

The /etc/wpa_supplicant/wpa_supplicant.conf file is owned by user root, so you need to be root to edit it. Once you've added the new network's information, simply issue the following command, as root:

dhcpcd

The preceding command signals the daemon copy of dhcpcd to reread /etc/wpa_supplicant/wpa_supplicant.conf and restart, after which it recognizes the new network.

wpa_passphrase

If you have trouble remembering the wpa_supplicant.conf syntax, use wpa_passphrase, which asks you for your password, then outputs through stdout the network's wpa_supplicant.conf information. Just copy the info from the screen into wpa_supplicant.conf using an editor.

[slitt@mydesk images]$ wpa_passphrase wl01
# reading passphrase from stdin
wouldntyouliketoknow
network={
	ssid="wl01"
	#psk="wouldntyouliketoknow"
	psk=859ded664bee3e13d68fd71b00cdf120b2727309b9c8a7c32c0d2afd4f603953
}
[slitt@mydesk images]$

DANGER!

NEVER put the password on the command line, because doing so makes the password available in your command-line history!

Warning: Case Sensitive!

The conversion of string password to 64 digit hex number depends on both string password and case-sensitive SSID, so get the right case on SSID.

Tip:

If copying the output of wpa_passphrase to wpa_supplicant.conf fails to produce a connection, just for the fun of it remove the comment pound sign from the password string, put it in front of the 64 digit Hex representation, save, and perform the dhcpcd command. Perhaps that will produce a connection.

If you don't want to mess with copying off the screen and editing a file, you can do the following:

wpa_passphrase wls >> /etc/wpa_supplicant/wpa_supplicant.conf

DANGER:

In the preceding command, be sure you use a DOUBLE angle bracket. If you were to use a single angle bracket, you'd irretrievably destroy your existing /etc/wpa_supplicant/wpa_supplicant.conf

Tip: No Prompt

The preceding command gives you no prompt to put in your paassword, but instead just sits there doing nothing. Just type in your string password for the network and press the Enter key, and the info will be added to /etc/wpa_supplicant/wpa_supplicant.conf

Tip:

Once again, if the 64 digit hex representation doesn't work, edit /etc/wpa_supplicant/wpa_supplicant.conf, uncomment the string representation, and comment the 64 digit hex.

Troubleshooting and Investigational Tools

Linux Wifi is a big, complex system, involving SSID name, passwords, dhcpcd, wpa_supplicant and several other factors. Without tools to peer inside, it can seem like a big black box. Here are some investigational tools to use when wifi doesn't work. Do them all as root, because most won't work when run by an ordinary user.

Run the ip link command to find the correct device name for your Wifi from a list of network devices. The Wifi device name will almost certainly begin with the letter w.

ip addr

The ip addr command lists all wifi networks and the IP addresses of any that has an IP address.

iwlist <devicename> scanning

In this command, the <ssid> is the device name of your laptop's wifi device. wlo1, wlp2s0, and evenwlp0s22f2u1u3 are examples.

This command uses <devicename> to hunt down the radio signals of all nearby network access points, and list them with all info about each. In 2018, a failure to find any access points means either a failure in your Wifi device or its drivers or connection to drivers, or you're in a tunnel or other structure acting as a Faraday Cage, or you're waaaaaay out in the boondocks.

Use this command to find an access point with which you want to connect.

wpa_cli ping

If wpa_supplicant is running in the background and functioning properly, this command responds with the string PONG, otherwise it prints an error message or nothing at all.

In the Void Linux way of doing things, wpa_supplicant is spawned by dhcpcd, and should always be running. If wpa_supplicant isn't running, all bets are off, and that's the first problem you should solve.

wpa_cli -i <devicename list_n

List all the networks configured in /etc/wpa/wpa_supplicant.conf, giving them a "network number" useful with other wpa_cli commands.

This differs from iwlist <devicename> scanning because it names whatever access points are configured in wpa_supplicant.conf, and nothing but those access points, whether those access points are in range or not. On the other hand, the iwlist version reports all nearby access points, whether configured in wpa_supplicant.conf or not.

wpa_cli -i <devicename> status

This command delivers all info for the selected Wifi devicename including connection information if it's connected. If it lists an IPV4 address, that's a strong indicator that you're fully functional. If it indicates that wpa_state=COMPLETED, you're either fully functional or need to just get an IP address assigned. If, on the other hand, wpa_state=DISCONNECTED, you're nowhere near connection.

The "Stop the Madness" Command

Wifi systems occasionally get themselves into a very bad state, which is a bad thing because bringing up a Wifi system is basically a state machine procedure. Sometimes these bad states include infinite loops with error messages going who knows where. If your endeavors in Wifi land just seem to bring more and more unexpected results, consider issuing the "Stop the Madness" command, which also brings down your Wifi. Here's the command:

wpa_cli -i <devicename> select_n 0; \
wpa_cli -i <devicename> disable_n 0

In the preceding, the "select" part disables all other networks than network 0, and the "disable" part disables network 0, so all networks are disabled and you're free to start fresh.

Sometimes it's a good idea to bring down the wifi too:

ip addr set dev <devicename> down

Obviously, you must then bring the Wifi device up so wpa_supplicant and dhcpcd can work with it.

And for some reason, on some computers you can't turn on the wifi until you bring down the wired Ethernet:

ip addr set dev <etherdevname> down

wpa_gui

I make sure to install wpa_gui on every Linux computer I own. Of all the Wifi tools, wpa_gui is most likely to succeed when the others have failed. And once you know it and its idiocyncracies, you can assimilate a new access point in seconds. There's a beginner's guide to wpa_gui in the Set Up Wifi Networking With wpa_supplicant article of the document on Troubleshooters.Com.

The user interface for wpa_gui is surprising to say the least. It writes your changes to wpa_supplicant.conf when you add a network, but not when you delete one. Unless you perform File->Save from the menu bar. Huh? The following are some tips for navigating the unexpected in wpa_gui, to add certainty to the material at wpa_gui:

Run Void Installer

The vast majority of Void x86_64 installations are done with the void-installer program that comes with all the installation CDs. This menu is reproduced below:

void-installer menu

In the preceding menu from Void Installer, the vast majority of items are done just the same, whether you're installing Void alone, or installing Void aside Windows on a laptop with only one hard drive. The following menu items are done the same way in both cases:

Only Bootloader, Partitions and Filesystem require special handling by you.

If you remember from earlier in this document, you already created your swap partition and your one and only Ext4 partition. So when you get to "Partitions", just exit without doing anything: You did it much earlier in this document. The one thing I did in the installer's partition activity was to mark my Ext4 partition as bootable, and the final result worked. I don't know whether it would have worked otherwise: There comes a point when you can't keep flipping variables to see every exact cause and effect.

In "Filesystem" you'll first the existing partitions that you made earlier, for your swap and Ext4. The type for the swap partition is Linux Swap, and the type for the Ext4 is, of course, ext4. The mount point for your Ext4 is, of course, /. Mark both of these as needing formatting.

Only after you've completed the plans for the Ext4 and the swap drive do you turn your attention to the special sauce for your EFI partition. Your disk already has an EFI partition: It's the same one used by Windows. Let's call it /dev/sdaZ. In formatting, you pick your existing EFI partition, tell it to mount at /boot/efi, and tell the installer NOT TO FORMAT IT!

DANGER: DON'T FORMAT THE EFI PARTITION!

DANGER OF PERMANENT HARDWARE DAMAGE: I've seen enough anecdotes in enough places on the web and one from a guy I know personally, to convince me that if you in any way harm (erase, overwrite, whatever) the EFI partition, you can permanently brick your computer. I think it happens very rarely and appears to involve computers with incomplete UEFI systems, but I've heard of it happening.

DON'T MARK YOUR EXISTING EFI PARTITION AS NEEDING FORMATTING!.

After you have your filesystems all set up, it's time to go back and revisit the Bootloader step. The only two choices are to put the bootloader on the disk device /dev/sda, not a partition. This sounds an awful lot like it would disable Windows, doesn't it? The other choice is "none: Manager Bootloader Otherwise". The first time through I chose "none", and got a Windows computer with no access to Linux.

So the second time through, knowing that I had the resources to put the computer's disk back to factory specs, I chose /dev/sda, crossed my fingers, held my breath, and proceeded with the install. It worked perfectly. When I pressed F12 immediately on boot, it brought up a screen where I could pick either Windows or Linux. Whew!

The remaining task, making Linux the default OS, was a matter of a fairly discoverable tweak inside the Boot F12 menu system. Once I did that, the computer booted to Linux unless I pressed F12 and changed the boot order.

Taming the AER Error Message Monster

This happened to me, and an hour of research tells me if you dual boot Linux with a modern Windows on modern hardware, it's likely to happen to you too.

You boot up your computer, get to the login prompt, and error messages start scrolling down your screen at about 10 per second. No problem: You switch to another Virtual Terminal, but the same problem happens. If you manage to log in and go into X mode with all those messages pouring through, a dmesg command shows those same error messages. Mine showed these messages at a rate of 10.2 messages per second. Each message starts with a square-bracketed timestamp, followed by the word "pcieport", followed by "0000:00:1c.5", so the beginning of each looks like the following:

[  127.231954] pcieport 0000:00:1c.5

The following is the four repeating messages, with the preceding beginning stripped away from each:

AER: Corrected error received: id=00e5
PCIe Bus Error: severity=Corrected, type=Physical Layer, id=00e5(Receiver ID)
 device [8086:9d15] error status/mask=00000001/00002000
  [ 0] Receiver Error         (First)

Subjectively, this problem seems to slow the computer appreciably.

Looking over the Internet, the solution that comes up again and again looks something like "add pci=noaer to the Kernel line." One particularly well written solution told me to do the following:

  1. Edit /etc/default/grub
  2. Find the line containing GRUB_CMDLINE_LINUX_DEFAULT
  3. Add pci=noaer to the end of that line.
  4. Save the file.
  5. update-grub
  6. Reboot

So that's what I did. Before, the line looked like the following:

GRUB_CMDLINE_LINUX_DEFAULT="loglevel=4 slub_debug=P page_poison=1"

Afterward, it looked like the following:

GRUB_CMDLINE_LINUX_DEFAULT="loglevel=4 slub_debug=P page_poison=1 pci=noaer"

When I rebooted, the problem was gone.

Summary

In hindsight this dual boot install was both easy and fairly as expected. But in the thick of it, presented with choices at every turn, and knowing that people had reported bricked computers for EFI partition fouls, it was serious stuff.

Watch out when partitioning. With fdisk, sfdisk, cfdisk, and all the rest, it's remarkably easy to forget what's what, and delete something you really needed. Follow the Carpenter's Motto: Measure twice, cut once.

So now I have a very nice dual boot. Void Linux for all my daily activities, and Windows for when the warranty department won't replace my dead battery unless I play their games with a Windows-only diagnostic software. And also, for when I need to test something under Windows or go to one of the (thank goodness) ever rarer "Windows Only" web apps.

Use this document at your own risk. It details a successful dual boot outcome using Void Linux on a Dell Inspiron 15 5570 laptop computer. It may or may not be of more general value, and it's always possible that it might not work again on My Dell 15-5570/Void combination. But I'm pretty happy with it, and will refer to it before every install.

I've never owned a Dell before. Most of my former computers made it as hard as possible to dual boot Linux. This Dell seemed to actually care that I might use a secondary OS. Easy turn-off of Secure Boot without borking Windows, and very straightforward "bios setup" to get the boot situation I wanted, without having to quickly push a button every time I boot the machine. From now on, if a Dell is $50 more than an equivalent HP/Acer/Asus/Whatever, I'll choose the Dell.


[ Training | Troubleshooters.Com | Email Steve Litt ]