I have an Acer laptop about eight years old. Its UEFI implementation
leaves a bit to be desired, it implements NewxtBook but not
DefaultBoot, and it resets the UEFI boot order if I change it.

The history:

It started life as Win10 in a 32GB eMMC, which isn't really big enough.
The laptop does have a slot for a real drive, which the advertising
curiously didn't mention, so it quickly gained a 240GB SSD as well.

I didn't expect to need Windows, so I clean-installed Stretch. The
installer gave me the option of dual booting, so I accepted, installed
Stretch on the SSD, and everything worked.

Later upgraded to Buster, which killed the grub boot, it would now only
cold boot to the Windows Boot Manager. After a great deal of fiddling
with the /boot/efi contents, I found that NextBoot worked, so I added a
Debian boot script to run efibootmgr to set it. I used the installer
rescue function to boot to grub after using Windows, which was not
often.

Time passed, Windows outgrew its 32GB and by its nature would not put
system stuff on anything but C:. It had only a few months of life left,
so I clean installed Bookworm over it on the eMMC. No problems, cold
booting fine, no NextBoot bodge needed.

Last week I got around to upgrading to Trixie: laptop now a brick.
Powering up actually gave me the blue firmware setup screens. I've had
many boot failures before, going back to the days of lilo and tomsrtbt,
but I've never before been kicked back to the hardware setup.

The Trixie install disk got me to the reinstall grub stage, after which
cold boot gave me a grub prompt. Oddly, it was grub 2.06xxx whereas
when I finally got a Trixie grub menu, it's 2.12xxx. Maybe a clue?

Anyway, I'm back to where I was with Buster, NextBoot will give me
Trixie, but without it, a cold boot gets me that old grub prompt.

So, clean installs of Stretch and Bookworm were able to boot from cold,
Buster and Trixie are not.

Now, I've installed grub to partition 2 of the eMMC, and the EFI stuff
on partition 1, /boot/efi, is enough to boot properly if pointed to by NextBoot. There are also mmcblk0boot0 and mmcblk0boot1. These are 4MB
read-only partitions, apparently in the eMMC hardware. As far as I can
find out, the boot0 can be used for booting but doesn't have to be. The
boot1 is a backup, which is not automatically used if boot0 is
corrupted. The Acer UEFI, as I said, isn't great, and seems almost
hard-wired for Windows. Is it possible this mysterious boot0 must
contain something grub-related for a proper boot, which Trixie's
installer isn't putting there but Bookworm's did? I gather the
read-only aspect is actually imposed by Linux, and can be disabled.

With the NextBoot bodge, I'm running OK, but I'd like to know what's
going on, at least well enough to submit a useful bug report. I can't
believe it's just a matter of luck whether Debian runs on eMMCs, of
which there must be many about in cheaper computers.

--
Joe

--- PyGate Linux v1.5.13
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

On 10/04/2026 8:13 pm, Joe wrote:

I have an Acer laptop about eight years old. Its UEFI implementation
leaves a bit to be desired, it implements NewxtBook but not
DefaultBoot, and it resets the UEFI boot order if I change it.

The history:

Frankly speaking, I am lost in the description of previous
configurations. My impression is that UEFI implementation may work with shim+grub.

Can you boot at least live image from a USB drive?

My suggestion is to look into (and maybe post output of)

efibootmgr -v
lsblk --fs

and inspect list of files on every partition that resembles EFI System Partition. grub.cfg there may contain a hint where the "wrong" version resides.

Is there an option in the firmware to select EFI file to boot from?

--- PyGate Linux v1.5.13
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

On Sat, 11 Apr 2026 10:16:30 +0700
Max Nikulin <manikulin@gmail.com> wrote:

On 10/04/2026 8:13 pm, Joe wrote:

I have an Acer laptop about eight years old. Its UEFI implementation
leaves a bit to be desired, it implements NewxtBook but not
DefaultBoot, and it resets the UEFI boot order if I change it.

The history:

Frankly speaking, I am lost in the description of previous
configurations. My impression is that UEFI implementation may work
with shim+grub.

When the upgrade from Stretch to Buster didn't boot, I spent a while
copying and renaming stuff in /boot/efi according to various Google
results, but nothing worked then.

Can you boot at least live image from a USB drive?

Yes, USB boots anything with no problem. I've always been able to boot
the installer to try to fix things, and the installer at the moment
lives in a Ventoy micro SD card. My first investigative step was to run
Knoppix 9.1, which is a few years old but so is the laptop.

My suggestion is to look into (and maybe post output of)

efibootmgr -v
lsblk --fs

This is from a clean install of Trixie to /dev/mmcblk0p2.

efibootmgr:

BootNext: 0002
BootCurrent: 0002
Timeout: 0 seconds
BootOrder: 0000,2001,2002,2003
Boot0000* Windows Boot Manager HD(1,GPT,0b9f6d3b-b061-4f42-940c-a51fd53fc9c1,0x800,0x32000)/File(\EFI\Micr osoft\Boot\bootmgfw.efi)RC
Boot0002* debian HD(1,GPT,0b9f6d3b-b061-4f42-940c-a51fd53fc9c1,0x800,0x32000)/File(\EFI\debi an\shimx64.efi)
Boot2001* EFI USB Device RC
Boot2002* EFI DVD/CDROM RC
Boot2003* EFI Network RC

I have previously added 0002 to the boot order list several times but it doesn't seem to stay there. Wretched Acer.

efibootmgr -v:

BootNext: 0002
BootCurrent: 0002
Timeout: 0 seconds
BootOrder: 0000,2001,2002,2003
Boot0000* Windows Boot Manager HD(1,GPT,0b9f6d3b-b061-4f42-940c-a51fd53fc9c1,0x800,0x32000)/File(\EFI\Micr osoft\Boot\bootmgfw.efi)RC
dp: 04 01 2a 00 01 00 00 00 00 08 00 00 00 00 00 00 00 20 03 00 00 00
00 00 3b 6d 9f 0b 61 b0 42 4f 94 0c a5 1f d5 3f c9 c1 02 02 / 04 04 46
00 5c 00 45 00 46 00 49 00 5c 00 4d 00 69 00 63 00 72 00 6f 00 73 00 6f
00 66 00 74 00 5c 00 42 00 6f 00 6f 00 74 00 5c 00 62 00 6f 00 6f 00 74
00 6d 00 67 00 66 00 77 00 2e 00 65 00 66 00 69 00 00 00 / 7f ff 04 00
data: 52 43 Boot0002* debian HD(1,GPT,0b9f6d3b-b061-4f42-940c-a51fd53fc9c1,0x800,0x32000)/File(\EFI\debi an\shimx64.efi)
dp: 04 01 2a 00 01 00 00 00 00 08 00 00 00 00 00 00 00 20 03 00 00 00
00 00 3b 6d 9f 0b 61 b0 42 4f 94 0c a5 1f d5 3f c9 c1 02 02 / 04 04 34
00 5c 00 45 00 46 00 49 00 5c 00 64 00 65 00 62 00 69 00 61 00 6e 00 5c
00 73 00 68 00 69 00 6d 00 78 00 36 00 34 00 2e 00 65 00 66 00 69 00 00
00 / 7f ff 04 00
Boot2001* EFI USB Device RC dp: 7f ff 04 00
data: 52 43
Boot2002* EFI DVD/CDROM RC dp: 7f ff 04 00
data: 52 43
Boot2003* EFI Network RC dp: 7f ff 04 00
data: 52 43

lsblk (sdb is a USB stick):

NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINTS
sda 8:0 0 223.6G 0 disk
ÃÄsda1 8:1 0 97.7G 0 part
ÀÄsda2 8:2 0 125.9G 0 part
sdb 8:16 1 7.2G 0 disk
ÃÄsdb1 8:17 1 4.7G 0 part /media/joe/37720218-ecdd-4e63-a839-f66190c8de43
ÀÄsdb2 8:18 1 2.5G 0 part
mmcblk0 179:0 0 29.1G 0 disk
ÃÄmmcblk0p1 179:1 0 100M 0 part /boot/efi
ÀÄmmcblk0p2 179:2 0 29G 0 part /
mmcblk0boot0 179:256 0 4M 1 disk
mmcblk0boot1 179:512 0 4M 1 disk

and inspect list of files on every partition that resembles EFI
System Partition. grub.cfg there may contain a hint where the "wrong"
version resides.

When I have time, I was intending to do that. Just experimented with
removing NextBoot, the grub version is grub 2.06-3-deb11u6.

Is there an option in the firmware to select EFI file to boot from?

Yes, but it only shows the Windows Boot Manager from a cold start. I'm
assuming that with NextBoot set, the menu must contain the Debian entry
or it wouldn't know where to find it. After the upgrade attempt, when I
was dropped to the hardware setup, the EFI boot menu was completely
empty.

I'm assuming the Windows Boot Manager no longer exists, and it
certainly never produces its normal menu. When I change the boot order
with efibootmgr, the next boot always resets the first entry to the
Windows one and removes the Debian one. I won't be buying another Acer computer, but I suppose most brands do that kind of thing.

--
Joe

--- PyGate Linux v1.5.13
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

On 4/11/26 7:23 AM, Joe wrote:

but it only shows the Windows Boot Manager from a cold start.

I take it the Windows Boot Manager entry probably is still stored in the
efi firmware and no longer exists on the eMMC?

What is the chance of going into the efi/bios settings (I'm talking
firmware, not the os) and deleting that Windows entry? I have done that
on my pc. I'm not sure all models support that. Some delete the entry automatically if you unplug the ssd and boot something else. But your
eMMC is likely soldered on, so can't be unplugged.

Some efi bioses are so poor it is sad. Like a cheap mini-pc I run from
time to time. Setting boot order seems to arrange the boot list via some random shuffle. It says + or - to move menu items up or down. I select
the item I wish to move and hit + or - and the entire list gets randomly shuffled. So I hit + multiple times until the list is generally in an
order I can live with.

I'm
assuming that with NextBoot set, the menu must contain the Debian entry
or it wouldn't know where to find it. After the upgrade attempt, when I
was dropped to the hardware setup, the EFI boot menu was completely
empty.

I'm assuming the Windows Boot Manager no longer exists, and it
certainly never produces its normal menu. When I change the boot order
with efibootmgr, the next boot always resets the first entry to the
Windows one and removes the Debian one. I won't be buying another Acer computer, but I suppose most brands do that kind of thing.

--- PyGate Linux v1.5.13
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

On Sat, 11 Apr 2026 11:40:52 -0500
nwe <nwe@gitcoding.net> wrote:

On 4/11/26 7:23 AM, Joe wrote:

but it only shows the Windows Boot Manager from a cold start.

I take it the Windows Boot Manager entry probably is still stored in
the efi firmware and no longer exists on the eMMC?

It must be stored somewhere, but it may be in the firmware itself,
hardcoded. I wonder if it is in the mmcblk0boot0 area. There doesn't
seem to be much about this on the net. It's probably nothing, but when
there's something mysterious laying around, there is a tendency to
assume it must be to blame for any problems.

What is the chance of going into the efi/bios settings (I'm talking firmware, not the os) and deleting that Windows entry? I have done
that on my pc. I'm not sure all models support that. Some delete the
entry automatically if you unplug the ssd and boot something else.
But your eMMC is likely soldered on, so can't be unplugged.

I haven't looked, but I'd assume so.

When I have more time, I'll have a play around with what's in
/boot/efi. It still contains some Microsoft stuff, but I'm not 100%
sure that can be removed without upsetting the firmware.

--
Joe

--- PyGate Linux v1.5.13
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

It must be stored somewhere, but it may be in the firmware itself,
hardcoded. I wonder if it is in the mmcblk0boot0 area. There doesn't
seem to be much about this on the net. It's probably nothing, but when there's something mysterious laying around, there is a tendency to
assume it must be to blame for any problems.

AFAIK `mmcblk0boot0` and `mmcblk0boot1` are "normal" in that what you
read is really what is there, there is no magic to them.
The only unusual part is that they are "hardware" partitions, not
something you can change with `fdisk` and friends.

Most GNU/Linux thingies never touch them, AFAICT, so for example on my
NanoPi I get:

# od /dev/mmcblk2boot0
0000000 000000 000000 000000 000000 000000 000000 000000 000000
*
20000000
#

If that command also shows you that it's just full of zeroes as it is
for me, then clearly there isn't "something mysterious laying around".


=== Stefan

--- PyGate Linux v1.5.13
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

On Fri, Apr 10, 2026 at 9:14?AM Joe <joe@jretrading.com> wrote:

I have an Acer laptop about eight years old. Its UEFI implementation
leaves a bit to be desired, it implements NewxtBook but not
DefaultBoot, and it resets the UEFI boot order if I change it.

Update the BIOS/UEFI. It is just another buggy program that needs
updating on occasion.

Once the software is up to date, then start working on problems.

Jeff

--- PyGate Linux v1.5.13
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

On 11/04/2026 7:23 pm, Joe wrote:

efibootmgr -v:

Assuming that firmware really overrides your attempts to change Boot* variables, I see nothing suspicious.

lsblk (sdb is a USB stick):

NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINTS
sda 8:0 0 223.6G 0 disk
ÃÄsda1 8:1 0 97.7G 0 part
ÀÄsda2 8:2 0 125.9G 0 part

Guessing from partition sizes, there is no ESP on SSD ("lsblk --fs"
would add FS type).

mmcblk0 179:0 0 29.1G 0 disk
ÃÄmmcblk0p1 179:1 0 100M 0 part /boot/efi

Is there EFI/BOOT/BOOTX64.EFI on this partition? It may get priority.
Grub package has a configuration parameter to install a copy of shim or
grub to "removable" path. It might explain why some Debian releases
could boot.

You may try to copy EFI/debian to EFI/BOOT and to rename shimx64.efi to bootx64.efi.

On Sat, 11 Apr 2026 10:16:30 +0700 Max Nikulin wrote:>> Is there an option in the firmware to select EFI file to boot from?

Yes, but it only shows the Windows Boot Manager from a cold start.

At least HP firmware, besides configured Boot* entries, allowed to pick
an arbitrary .efi file using a kind of file browser. There was even an
option to create a custom boot entry by typing device and file type.

I won't be buying another Acer
computer, but I suppose most brands do that kind of thing.

Maybe firmware has been improved over 8 years. At least comparing HP
firmware versions, I should say that 2021 variant is more sane that one
from 2014. Discussions of specific models may be more relevant.

My experience is that boot configuration in firmware setup menu may
override variables available to efibootmgr in some nontrivial way.
Firmware may have an options whether to boot from specific "devices". I
would not be surprised if it affects whether EFI/Boot removable paths
are considered.

You may try to inspect variables from grub command line

echo $prefix
echo $cmdpath
echo $fw_path

Last one added by a patch for 2.12.

I have seen messages suggesting that installing shim (or grub) as EFI/Microsoft/Boot/bootmgfw.efi is the only workaround for some laptops.

--- PyGate Linux v1.5.13
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

On Sun, 12 Apr 2026 09:55:16 +0700
Max Nikulin <manikulin@gmail.com> wrote:

On 11/04/2026 7:23 pm, Joe wrote:

efibootmgr -v:

Assuming that firmware really overrides your attempts to change Boot* variables, I see nothing suspicious.

lsblk (sdb is a USB stick):

NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINTS
sda 8:0 0 223.6G 0 disk
ÃÄsda1 8:1 0 97.7G 0 part
ÀÄsda2 8:2 0 125.9G 0 part

Guessing from partition sizes, there is no ESP on SSD ("lsblk --fs"
would add FS type).

No, it used to be Stretch then Buster, but when I put Bookworm on the
eMMC I cleared the SSD for data usage. But all Debians have used the
eMMC FAT partition for /boot/efi.

mmcblk0 179:0 0 29.1G 0 disk
ÃÄmmcblk0p1 179:1 0 100M 0 part /boot/efi

Is there EFI/BOOT/BOOTX64.EFI on this partition? It may get priority.
Grub package has a configuration parameter to install a copy of shim
or grub to "removable" path. It might explain why some Debian
releases could boot.

No, I've never used the removable option when installing grub. Stretch
(SSD) and Bookworm (eMMC) Just Worked.

You may try to copy EFI/debian to EFI/BOOT and to rename shimx64.efi
to bootx64.efi.

Well, at the time of the Buster upgrade, I unsuccessfully copied and
renamed loads of stuff, and left it all in place until the right answer
was found. In the past, when two fixes have been needed, I've tried
one, removed it, and tried the other, both unsuccessfully of course, so
I don't take out an attempted fix until it seems safe to do so. But
nothing worked then, so it's all still there, including at least one
alleged Microsoft file being really a grub one.

Possibly an approach would be to backup then empty the /boot/efi from
inside Trixie, then grub-install. It may be that grub isn't installing
properly because it's finding things already there.

On Sat, 11 Apr 2026 10:16:30 +0700 Max Nikulin wrote:>> Is there an
option in the firmware to select EFI file to boot from?

Yes, but it only shows the Windows Boot Manager from a cold start.

At least HP firmware, besides configured Boot* entries, allowed to
pick an arbitrary .efi file using a kind of file browser. There was
even an option to create a custom boot entry by typing device and
file type.

When everything is working, yes, the F12 menu does contain all bootable
EFI entries. If I have an installer USB plugged in, it finds and lists
that. But it never contained any tools, which I think is because
Microsoft strongly discourages messing with the EFI variables, and it's
a pain to try to edit them from inside Windows. You certainly have to
reboot to Safe Mode to begin doing that.

I won't be buying another Acer
computer, but I suppose most brands do that kind of thing.

Maybe firmware has been improved over 8 years. At least comparing HP firmware versions, I should say that 2021 variant is more sane that
one from 2014. Discussions of specific models may be more relevant.

I did try at the time of the Buster upgrade to get an updated UEFI
firmware, but there wasn't one, I already had the only released version
for that model. I'll have another look, but the model was already
discontinued when I looked before.

My experience is that boot configuration in firmware setup menu may
override variables available to efibootmgr in some nontrivial way.
Firmware may have an options whether to boot from specific "devices".
I would not be surprised if it affects whether EFI/Boot removable
paths are considered.

You may try to inspect variables from grub command line

echo $prefix
echo $cmdpath
echo $fw_path

Last one added by a patch for 2.12.

I have seen messages suggesting that installing shim (or grub) as EFI/Microsoft/Boot/bootmgfw.efi is the only workaround for some
laptops.

Thank you for your help.

The bottom line is that an upgrade from a working clean install
of Bookworm to Trixie seemed to go perfectly, but did not boot
afterwards, despite Trixie being able to boot and run on this machine
when kicked hard enough. That's a bug, and I was hoping to work out
what caused it. I'll have another go when I have some time.

--
Joe

--- PyGate Linux v1.5.13
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)