Shim 16.1-8: Invalid call dxe_update_mem_attrs, addr is not page aligned

[olifre]

Upgrading from shim 15.8 to 16.1_p8 (named like this on Gentoo Linux, but that just uses the version 16.1-8 from Fedora, i.e. https://kojipkgs.fedoraproject.org/packages/shim/16.1/8/x86_64/ ) yields the following error during boot:

Invalid call to dxe_update_mem_attrs(addr:0x35985680-0x3598667F, size:0x1000, +r, -wx)
  addr is not page aligned

on an Alienware 18 Area-51 AA18250 with latest UEFI available from vendor (version 2.3.1).

The system continues to boot (seemingly) fine after this, this error was not present with shim 15.8.

Is there any further diagnostic information I can provide?

Summary of reports in this issue:

Device	UEFI version	Used loader	Observed problem
Alienware 18 Area-51 AA18250	2.3.1, 2.4.1	rEFInd	dxe_update_mem_attrs, addr is not page aligned, booting still works
Alienware 18 Area-51 AA18250	2.5.0	rEFInd	dxe_update_mem_attrs, addr is not page aligned, system stuck on boot / erratic behaviour
Dell Pro 14 Premium PA14250	2.13.4	rEFInd	works fine
Dell Pro 14 Premium PA14250	2.14.4	rEFInd	dxe_update_mem_attrs, addr is not page aligned, system stuck in rEFInd
Lenovo ThinkPad X390 Yoga	?	rEFInd	several dxe_update_mem_attrs, addr is not page aligned,booting still works
PC	?	Grub2, rEFInd	dxe_update_mem_attrs, addr is not page aligned, boot stops at that point
Gigabyte x870 Aorus Elite AMD	F12	rEFInd	dxe_update_mem_attrs, addr is not page aligned, boot stops at that point

[DominoLeChat]

Hi,
Same problem on a Lenovo ThinkPad X390 Yoga.

[olifre]

I can reproduce the same issue now on a:
Dell Pro 14 Premium PA14250
after UEFI/BIOS updates.

Interestingly, on that machine, I only see the addr is not page aligned issue for one address (similar to the other machine), but only when chainloading refind from shim .
Chainloading a more lightweight loader, i.e. systemd-boot or just an UKI, I do not get this message.

Potentially, it is only triggered if the EFI application is larger / does more allocations?

[olifre]

@DominoLeChat Which loader do you use via shim?
As my additional experience with the Dell Pro 14 Premium PA14250 shows, at least on that machine, it seems the issue is only triggered with more complex loaders such as rEFInd, which might help others to reproduce this more easily.

[DominoLeChat]

Hi,

I'm indeed using rEFInd on a Lenovo Thinkpad X390 Yoga and that’s when the error message appeared.

In my case, with Secure Boot turned on, that memory address error shows up for a moment during boot. It isn't actually blocking, everything eventually loads fine, but it slows down the startup process quite a bit.

[olifre]

@DominoLeChat Thanks for confirming!
Yes, indeed it was not blocking for me at first, too, but after a BIOS update, I am now observing (on two systems already) misbehaviour of rEFInd with Shim 16.1 (see #810 ). So I am anxious that the warning indicates an issue marking memory pages (potentially due to a firmware bug, or a shim bug), which might lead to undefined behaviour in loader applications, and the effect could change with BIOS updates etc.

As I am not an UEFI / shim developer myself, this is just guesswork from the existing data points / reports of course, but it seems wise to be cautious given the observations, and potentially the collected observations can help the developers.
So thanks for confirming that in your case, it's also with rEFInd, so my hunch that it only happens with a loader doing a larger number of allocations seems still likely.

[gnu-like]

I downgraded these two packages unfortunately Shim 16.1-8 didn't work

shim-signed_1.47+15.8-1_amd64.deb
shim-signed-common_1.47+15.8-1_all.deb

[olifre]

@gnu-like Thanks, could you provide some more information on "didn't work" so the Shim developers can hopefully address the problem?
For example:

• Which loader are you using (Grub2, rEFInd,...)?
• Which device / UEFI firmware version?
• Do you also see the problems this issue is about (Invalid call to dxe_update_mem_attrs) or is this an unrelated problem?
• What exactly does didn't work mean, do you get warning when booting, does it get stuck etc.?

I hope this helps to disentangle whether the issue you observe is related or potentially a different issue. Thanks in advance!

[olifre]

Note: I have updated the opening post with a table collecting all information currently reported here, and will try to keep it updated, hoping that this will help developers.

[gnu-like]

@olifre You're welcome , I'm using PC , Grub2, rEFInd
secure boot enable So after upgrading OS (debian) shim packages upgraded too.
after rebooting I get error same as you

Invalid call to dxe_update_mem_attrs(addr:0x35985680-0x3598667F, size:0x1000, +r, -wx)
  addr is not page aligned

[olifre]

@gnu-like Thanks! Just to confirm:
Do you only get the warning message, or does boot stop at that point?
Does that happen both with Grub2 and rEFInd?

[gnu-like]

@olifre Don't mention it
boot stop show that error
This error from the shim bootloader Shim is trying to call it with an address that is not aligned

[Pbryanw]

Hi, another rEFInd user with the same error message (an internet search of my screen capture led me to this thread).

I've been able to workaround this issue by replacing the shimx64.efi file in my refind directory with version 15.8 from here:
https://kojipkgs.fedoraproject.org/packages/shim/15.8/5/x86_64/

You can extract the rpm using Ark. The shimx64.efi file can be found in this directory:
/usr/lib/efi/shim/15.8-5/EFI/fedora/

The shimx64.efi file can then be copied to the refind directory located at /boot/efi/EFI/ where it will overwrite the existing file.
I'm not sure what the security implications are though?

I'm running rEFInd in Fedora 44 on a Gigabyte x870 Aorus Elite AMD motherboard with the F12 BIOS. I don't get this error with Grub2.

[olifre]

I'm running rEFInd in Fedora 44 on a Gigabyte x870 Aorus Elite AMD motherboard with the F12 BIOS. I don't get this error with Grub2.

Thanks for the report, I have added this to the table in the first post.
Just to make sure, you get the error message and then boot stops? Or does rEFInd still work / hang after showing itself / the logo?

[Pbryanw]

@olifre - Hi, thanks for your reply, and for adding my info to your original post.
Following the error message, the boot hangs on this text, and then a power-off using the PC's power button is required.