Re: ❌ FAIL: Test report for for-kernelci (6.13.0-rc3, upstream-arm-next, 06a57361)

[Konrad Dybcio]

On 20.12.2024 2:43 PM, Will Deacon wrote:
> On Fri, Dec 20, 2024 at 01:25:50PM +0000, Will Deacon wrote:
>> On Fri, Dec 20, 2024 at 02:06:21PM +0100, Ard Biesheuvel wrote:
>>> On Fri, 20 Dec 2024 at 13:49, Ard Biesheuvel <ardb at kernel.org> wrote:
>>>>
>>>> On Fri, 20 Dec 2024 at 13:35, Will Deacon <will at kernel.org> wrote:
>>>>>
>>>>> On Thu, Dec 19, 2024 at 11:12:08PM -0000, cki-project at redhat.com wrote:
>>>>>> Hi, we tested your kernel and here are the results:
>>>>>>
>>>>>>     Overall result: FAILED
>>>>>>              Merge: OK
>>>>>>            Compile: OK
>>>>>>               Test: FAILED (Did not boot)
>>>>>>
>>>>>>
>>>>>> Kernel information:
>>>>>>     Commit message: Merge branch 'for-next/core' into for-kernelci
>>>>>>
>>>>>> You can find all the details about the test run at
>>>>>>     https://datawarehouse.cki-project.org/kcidb/checkouts/redhat:1596629923
>>>>>>
>>>>>> One or more kernel builds failed:
>>>>>>     Builds that are failing to boot:
>>>>>>          aarch64 - https://datawarehouse.cki-project.org/kcidb/builds/redhat:1596629923-aarch64-kernel
>>>>>
>>>>> Hmm, I can't find any console logs here. Did the machines really fail that
>>>>> early?
>>>>>
>>>>
>>>> This appears to be my fault: commit 32d053d6f5e92efd
>>>>
>>>>   arm64/mm: Drop configurable 48-bit physical address space limit
>>>>
>>>> inadvertently sets CONFIG_ARM64_LPA2 for all VA space sizes. The
>>>> following hunk needs to be applied to address this
>>>>
>>>> --- a/arch/arm64/Kconfig
>>>> +++ b/arch/arm64/Kconfig
>>>> @@ -1418,7 +1418,7 @@ config ARM64_VA_BITS
>>>>
>>>>  config ARM64_LPA2
>>>>         def_bool y
>>>> -       depends on !ARM64_64K_PAGES
>>>> +       depends on ARM64_VA_BITS_52 && !ARM64_64K_PAGES
>>>>
>>>>  choice
>>>>         prompt "Endianness"
>>>
>>> On second thought, let's just omit commit 32d053d6f5e92efd  for now -
>>> it's at the tip of the for-next/mm branch so it can just be dropped.
>>>
>>> I'll revisit this later - I have to double check that non-LPA2 configs
>>> (such as the one this report is about) work correctly on LPA2 capable
>>> hardware, and don't deviate from LPA2 configs booting with arm64.nolva
>>
>> Sure thing, I'll drop it for now. I've managed to repro the failure
>> under qemu using the CKI .config, so I'll check that things are happy
>> again before I pushed the updated branch.
> 
> Ok, now pushed!

(resending as thunderbird prioritizes replying to the list instead of
all correspondents, sorry)

FWIW I ran a bisect and this breaks X1E80100 / Oryon as well..

4k pages, CONFIG_ARM64_LPA2=y, id_aa64mmfr0 = 0x2100122100001024

Konrad