Re: [Regression in 7.0] OLPC-XO-1.5 will not boot after commit 157266edcc

Re: [Regression in 7.0] OLPC-XO-1.5 will not boot after commit 157266edcc

[Thorsten Leemhuis]

Lo! Looks like nothing happened since the report (please correct me if
I'm wrong, it's easy to miss something). Adding a few more people and
lists to the recipients of this reply, maybe that will things rolling.

FWIW, the culprit seems to be 157266edcc5671 ("x86/boot/e820: Simplify
append_e820_table() and remove restriction on single-entry tables")
[v7.0-rc1] from Ingo.

Therefore adding David Gow, who submitted a fix for the culprit -- but
that didn't make it anywhere afaics, unless I'm missing something. For
details see:
https://lore.kernel.org/all/20260416065746.1896647-1-david@davidgow.net/

Ciao, Thorsten

On 4/19/26 22:08, An wrote:
> On 19/04/2026 10.46, An wrote:
>> [please keep me in CC, I'm not subscribed]
>>
>> Hi Ingo,
>>
>> my OLPC XO-1.5 stopped booting when switching from 6.19 to 7.0
>>
>> I bisected this to
>>
>> commit 157266edcc56715323de1bd60e49194b3b66a174
>> Author: Ingo Molnar <mingo@kernel.org>
>> Date:   Thu May 15 14:05:41 2025 +0200
>>
>>      x86/boot/e820: Simplify append_e820_table() and remove
>> restriction on single-entry tables
>>
>> The XO boots from a forth environment and is rather quirky.
>> If I can help with any logs or tests, please let me know.
>>
>> I'm compiling a 7.0 kernel with the patch reverted,
>> but it smells like a plain revert isn't the best idea.
>>
>> Thank you,
>> An Müller
>>
> 
> I've tested both 7.0 and current mainline: with this patch reverted,
> they boot fine, the revert applies cleanly in both cases.
> 
> Thanks,
> An

Re: [Regression in 7.0] OLPC-XO-1.5 will not boot after commit 157266edcc

[An]

On 06/05/2026 10.07, Thorsten Leemhuis wrote:
> Lo! Looks like nothing happened since the report (please correct me if
> I'm wrong, it's easy to miss something). Adding a few more people and
> lists to the recipients of this reply, maybe that will things rolling
 > > FWIW, the culprit seems to be 157266edcc5671 ("x86/boot/e820: Simplify
> append_e820_table() and remove restriction on single-entry tables")
> [v7.0-rc1] from Ingo.
> 
> Therefore adding David Gow, who submitted a fix for the culprit -- but
> that didn't make it anywhere afaics, unless I'm missing something. For
> details see:
> https://lore.kernel.org/all/20260416065746.1896647-1-david@davidgow.net/
> 
> Ciao, Thorsten

Thanks Thorsten!

No, nothing happened here.
I have not been able to grok the program flow.
Now reading David's patch, the "initially no entry at all"
bit escaped me. Oh well.

I have compiled and booted mainline (today's git) with
David's patch applied just now. It works for me too.
Thank you David!

And thanks all,
An


> On 4/19/26 22:08, An wrote:
>> On 19/04/2026 10.46, An wrote:
>>> [please keep me in CC, I'm not subscribed]
>>>
>>> Hi Ingo,
>>>
>>> my OLPC XO-1.5 stopped booting when switching from 6.19 to 7.0
>>>
>>> I bisected this to
>>>
>>> commit 157266edcc56715323de1bd60e49194b3b66a174
>>> Author: Ingo Molnar <mingo@kernel.org>
>>> Date:   Thu May 15 14:05:41 2025 +0200
>>>
>>>       x86/boot/e820: Simplify append_e820_table() and remove
>>> restriction on single-entry tables
>>>
>>> The XO boots from a forth environment and is rather quirky.
>>> If I can help with any logs or tests, please let me know.
>>>
>>> I'm compiling a 7.0 kernel with the patch reverted,
>>> but it smells like a plain revert isn't the best idea.
>>>
>>> Thank you,
>>> An Müller
>>>
>>
>> I've tested both 7.0 and current mainline: with this patch reverted,
>> they boot fine, the revert applies cleanly in both cases.
>>
>> Thanks,
>> An
> 

Re: [Regression in 7.0] OLPC-XO-1.5 will not boot after commit 157266edcc

[Ingo Molnar]

* Thorsten Leemhuis <regressions@leemhuis.info> wrote:

> Lo! Looks like nothing happened since the report (please correct me if
> I'm wrong, it's easy to miss something). Adding a few more people and
> lists to the recipients of this reply, maybe that will things rolling.
>
> FWIW, the culprit seems to be 157266edcc5671 ("x86/boot/e820: Simplify
> append_e820_table() and remove restriction on single-entry tables")
> [v7.0-rc1] from Ingo.

Yeah, I missed the empty table case. :-/

> Therefore adding David Gow, who submitted a fix for the culprit -- but
> that didn't make it anywhere afaics, unless I'm missing something. For
> details see:
> https://lore.kernel.org/all/20260416065746.1896647-1-david@davidgow.net/

This fix is now queued up and should go upstream later this week:

  5772f6535227 ("x86/boot/e820: Re-enable BIOS fallback if e820 table is empty")

Thanks,

	Ingo