x230-hotp-verification: not waking from suspend

[stmc-droid]

After suspending a x230 when trying to wake it does a hard restart instead. I am using the latest x230-hotp-verification build. Does anyone else have the same issue? I tested with Qubes and Tails.

[lsafd]

Also having the same issue.

I would recommend reading through #608 if you haven't already. These suspend issues may be related to the version of Coreboot used by Heads (This is just secondhand from that issue).

[tlaurion]

Which is funny since I do not have that issue in my old artifact, deployed with the PrivacyBeast as per QubesOS certified firmware. Coreboot version is the same as used here, so not the cause.

I have tested latest build and can confirm the issue.

@MrChromebox I have seen that you have fixed similar issues in the past. You have a troubleshooting path to recommend?

[MrChromebox]

@tlaurion the issue you linked was due to a change in Chrome EC event handling on the coreboot side which broke things on certain ChromeOS devices

[tlaurion]

@MrChromebox @jans23 @osresearch @PatrickRudolph : see comment #721 (comment)

Coreboot 4.12 doesn't fix this issue

[jans23]

@szszszsz Any ideas?

[PatrickRudolph]

Please provide more details.
Which resume doesn't work? S3? S4?
Please provide a log of a regular boot and a failing S3/S4 wake up boot.

The problem can be a invalid MRC cache, changed e820 memory map during regular boot vs wake up boot, touched hardware that shouldn't be touched (Intel IGD), ...

[tlaurion]

Waking up from S3 results in a reset.

Not sure how I could provide a log for failing since the machine reboots as if the power button was a reset button, where the half moon led showing the laptop is under sleep is still on upon resume from suspend (while resetting and actually powering on, and sleep led stays on).

@PatrickRudolph you can flash the referred ROMs for testing. Else provide testing guidance, but I cannot reproduce https://gitlab.com/tlaurion/heads/-/jobs/312958349/artifacts/raw/build/x230-libremkey/coreboot.rom?inline=false?job=build rom. Heads tree got broken for a while, which lead from musl-cross to be replaced to musl-cross-make which lead to #734 and #608

[alex-nitrokey]

@tlaurion are you sure this issue started when changing to musl-cross-make? I am surprised that this did not caught our attention earlier then...

I am not sure how to debug this either. But I was wondering if this might be related to cbfs size as well, as it is happening on x230-hotp-verification build and not the x230, right?

When adapting the cbfs size after failed boots ( #737 ), we just assumed a size that is okay. So maybe most stuff seemed to work, but not everything is okay - like this issue, maybe even #765 (EDIT: no, #765 is not fixed!).

Thus, what I was trying is the following:

• setting CONFIG_CBFS_SIZE=0x700000 in config/coreboot-x230-hotp-verification.config
• disabling CONFIG_DROPBEAR and CONFIG_LINUX_E1000E to save some space

This build resulted in a "suspendable" version of x230-hotp-verification. Would be glad about a confirmation of my observation.

[tlaurion]

Will look into this, thanks for the report.

[tlaurion]

Successfull build per tlaurion@b1886f3
Direct link to ROM to reproduce and comment: https://430-103208611-gh.circle-artifacts.com/0/build/x230-hotp-verification/coreboot.rom

If it works, it means that the maximum space available in CBFS for BIOS is really 0x700000 (I remember I did the calculations somewhere under Heads....)

EDIT: explanation is here

I'll try to test in next test batch @alex-nitrokey

[alex-nitrokey]

Successfull build per tlaurion/heads@b1886f3

I just tested the build. Suspend works fine here!

EDIT: explanation is here

Thanks! I vaguely remembered this post. Good work!

[tlaurion]

Will be fixed by #838, Cherry picking Nitrokey@ad75d8e and referring to #667 (comment)

[tlaurion]

fixed under 30852aa