virtualbox.org

VB 5.2.0_BETA1 r117406 on Xubuntu64 16.04.3 or Slackware64 Current : Booting e.g. the Ubuntu64 16.04.3 live ISO in EFI mode results in a medium not found message (see attachment). Something comparable happens when the Fedora64 26-1.5 live ISO, or with a Slackware64 Current live ISO.
With VB 5.1.26-117224 booting these ISOs in EFI mode is successful.

burdi01 wrote:the Ubuntu64 16.04.3 live ISO in EFI mode

Which one? Be specific please: http://releases.ubuntu.com/16.04/

MOD EDIT:
Deleted a 120 KB image and replaced it with 34 B of actual, searchable, copyable text.

---

ubuntu-16.04.3-desktop-amd64.iso

The highlighted one (scroll down if necessary):
Screenshot at 2017-08-05 18-10-53.png

[burdi01] Edit Aug 6, 2017: Hint noted.

Thanks, I will try it.

Next time, please avoid unnecessary attachments. You could have simply typed the filename, no? Save a (virtual) tree or two (or five)...

Confirmed only if using the ICH9 chipset. With the default PIIX3, booting works just fine.
The ISO boots just fine with 5.1.26, and it fails with 5.2.0_BETA1 r117406 (Qt5.6.2). The error message (which seems really weird) is:

BusyBox v1.22.1 (Ubuntu 1:1.22.0-15ubuntu1) built-in shell (ash)
Enter 'help' for a list of built-in commands.

(initramfs) Unable to find a medium containing a live file system

• Host: OSX 10.11.6 (15G1611).
• Guest: ubuntu-16.04.3-desktop-amd64.iso
• VirtualBox: Version "5.2.0_BETA1 r117406 (Qt5.6.2)" versus "5.1.26 r117224 (Qt5.6.2)".

Two logs attached (one zip) from a successful 5.1.26 vs. an unsuccessful 5.2.0b0 run.

Edit: Updated to state that this happens only if using the ICH9 chipset. With the default PIIX3, booting works just fine.

Does this for the original poster also happen only with ICH9 chipset + EFI? For me it does, and it makes me wonder why someone out there routinely uses the ICH9 chipset when we rather clearly say "[...] is experimental and not recommended [...]". Also, please provide VBox.log for failures as the problem reporting instructions state. Otherwise we end up easily wasting a lot of time, trying a different config than you use.

klaus wrote:it makes me wonder why someone out there routinely uses the ICH9 chipset

I have a VM called "_LiveEFI" just for testing EFI booting. Chances are that I had set the ICH9 because I wanted to test an OSX guest at some point and it got left enabled. Sorry about that...

I did test again the same VM but with PIIX3 enabled instead of the ICH9. Booting works just fine. My "confirming" post has been edited to reflect that fact.

For complete disclosure, ICH9+EFI with the same VM does work in 5.1.26, just not with 5.2.0b1. But it's not a surprise since it is listed as an experimental feature after all...

Original poster here:
Yes, I have the ICH9 chipset selected. Setting it to PIIX3 resolved the problem.
I cannot even remember when I selected ICH9, it must be years ago and I never looked back.
Chances are that I chose the setting directly and never saw the hoover text.

Thanks for confirming. Of course we still want the issue with ICH9 to be fixed (otherwise it will remain experimental forever). Our investigations so far have shown that this issue only happens when booting off a DVD drive attached to an ATA controller. SATA works just fine for the same setup. Still mysterious what change caused this regression...

Indeed I have the live ISO connected to the IDE controller.
I can confirm that with the ICH9 chipset and the live ISO connected to the SATA controller the VM boots OK.

Beta 2 contains a fix which should allow booting from IDE even with EFI+ICH9.

Beta 2 contains a fix which should allow booting from IDE even with EFI+ICH9.

Confirmed - thks.