Bug 259427 - Fails to boot on HP ProBook 455 G7: Firmware Error (ACPI): AE_AML_PACKAGE_LIMIT, Index (0x000000005) is beyond end of object (length 0x5) (20201113/exoparg2-569)
Summary: Fails to boot on HP ProBook 455 G7: Firmware Error (ACPI): AE_AML_PACKAGE_LIM...
Status: Open
Alias: None
Product: Base System
Classification: Unclassified
Component: kern (show other bugs)
Version: 12.2-RELEASE
Hardware: amd64 Any
: --- Affects Only Me
Assignee: freebsd-bugs (Nobody)
URL:
Keywords: needs-patch, needs-qa
Depends on:
Blocks:
 
Reported: 2021-10-25 04:28 UTC by Joshua Kinard
Modified: 2021-11-07 02:17 UTC (History)
1 user (show)

See Also:
koobs: mfc-stable13?
koobs: mfc-stable12?
koobs: mfc-stable11-


Attachments
Output of 'acpidump -dt' (72.40 KB, application/x-xz)
2021-10-25 04:28 UTC, Joshua Kinard
no flags Details
Screenshot of 13.0-RELEASE Installer getting overran by ACPI Errors (558.08 KB, image/jpeg)
2021-10-25 04:30 UTC, Joshua Kinard
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Joshua Kinard 2021-10-25 04:28:28 UTC
Created attachment 229011 [details]
Output of 'acpidump -dt'

Attempting to boot the FreeBSD 13.0-RELEASE-p0 installer memory stick on an HP ProBook 455 G7 laptop running the latest firmware/BIOS release (S79 v01.06.00) usually gets to the installer greeting (sometimes reboots in the middle of booting, but works second time around; different bug).  However, within several seconds, the following ACP error messages start spamming the screen once every 2-3 seconds:

Firmware Error (ACPI): AE_AML_PACKAGE_LIMIT, Index (0x000000005) is beyond end of object (length 0x5) (20201113/exoparg2-569)
ACPI Error: Aborting method \_TZ.GTTP due to previous error (AE_AML_PACKAGE_LIMIT) (20201113/psparse-689)
ACPI Error: Aborting method \_TZ.CHGZ due to previous error (AE_AML_PACKAGE_LIMIT) (20201113/psparse-689)

---

I bought the laptop as a custom build in late 2020, but never really did anything with it because of these ACPi errors on FreeBSD 12.2-RELEASE media.  Finally got around to trying 13.0-RELEASE media, and getting the same error.  Doing a Google search for portions of these error strings returns a lot of similar errors on Linux distributions, but no real resolutions, other than complaining to the manufacturer.  I am not going to bother with that because I'd be surprised if they did anything about it.  The system came preloaded with a FreeDOS release.

I gave a quick test of 14.0-CURRENT cut from 9aa29457d55 (r249761), and that doesn't even boot all the way, crashing into the kernel debugger due to a page fault panic on.....it looks like the audio driver.

I am including as attachments the output of 'acpidump -dt | gzip -c9 > foo.asl.gz', just renamed and compressed with XZ.  Also attached a screenshot showing how messy the primary console is with these ACPI errors repeating every few seconds.

If anyone thinks this is fixable, I can try testing things, but will probably have to wait until a later 14.0-CURRENT snapshot, or someone will need to provide modified copies of 13.0-RELEASE GENERIC kernels that I can kludge onto this thumb drive I am booting off of.  Assuming the bug is in the kernel...
Comment 1 Joshua Kinard 2021-10-25 04:30:30 UTC
Created attachment 229012 [details]
Screenshot of 13.0-RELEASE Installer getting overran by ACPI Errors

The text of the errors are these three lines repeated forever:

Firmware Error (ACPI): AE_AML_PACKAGE_LIMIT, Index (0x000000005) is beyond end of object (length 0x5) (20201113/exoparg2-569)

ACPI Error: Aborting method \_TZ.GTTP due to previous error (AE_AML_PACKAGE_LIMIT) (20201113/psparse-689)

ACPI Error: Aborting method \_TZ.CHGZ due to previous error (AE_AML_PACKAGE_LIMIT) (20201113/psparse-689)
Comment 2 Joshua Kinard 2021-10-25 04:33:23 UTC
IMHO, looks like a string parsing bug of some kind where it's failing to handle multiple padding 0's so that it thinks 0x000000005 != 0x5.  At least, that is my take based on what the error messages say.  I tried searching through the acpidump output for anything to do with those values, but did not find anything that fit.
Comment 3 Andriy Gapon freebsd_committer 2021-10-25 08:30:11 UTC
(In reply to Joshua Kinard from comment #2)
That's very unlikely.  It's the same number but printed slightly differently.

The error, just as the message says, is that the (zero-based) index 5 is beyond the length of 5.  Valid indexes are 0 to 4.
Comment 4 Kubilay Kocak freebsd_committer freebsd_triage 2021-10-25 22:44:09 UTC
^Triage: Set version to earliest confirmed reproducible version
Comment 5 georg.lastname 2021-11-06 18:08:34 UTC
I'm afraid only a firmware (aka bios, uefi) update will resolve this.

TZ refers to a thermal zone. GTTP and CHGZ are undocumented functions. They are not described in the acpi specification. One could guess GTTP stands for 'get thermal trip point', but who knows..

FreeBSD has nothing to do with all this (I think).
Comment 6 Joshua Kinard 2021-11-07 02:17:19 UTC
(In reply to georg.lastname from comment #5)

I think the bug on the FreeBSD-side is that these messages spam repeatedly every few seconds and they clobber the main console, making using the boot/install console unusable.  Some searching around, it looks like when Linux encounters these errors, they are at least stuffed away in the dmesg log and don't flood the console.  I can't tell if they repeat forever in dmesg, though.  Haven't actually tried booting Linux on this laptop yet (I have enough systems running Linux to keep me entertained as-is).

If the console could be fixed, I might be able to at least install BSD on it and see how it holds up.  As for fixing the BIOS/UEFI/ACPI issues directly, if anyone has contacts at HP I can go and pester, I'll give it a shot.  But I did not buy the CarePack for this thing, so going through their normal support channels would be a futile effort.

Google did turn up this post on the FreeBSD forums that I hadn't come across before:
https://forums.freebsd.org/threads/panic-when-connecting-or-disconnecting-a-c-power.81550/

Which points at amdgpu being somewhat responsible.  So if this is thermal zone-related, possibly tied to the graphics adapter, and thus, the drm-kmod/gpu-firmware packages?

Push comes to shove, I'll donate it to goodwill and hunt down a non-HP laptop that'll be more....compatible.