Re: linux-next 20180515 - ACPI disabled..

[Al Stone]

On 05/15/2018 03:23 PM, Andy Shevchenko wrote:
> +Cc: Rafael
> 
> On Wed, May 16, 2018 at 12:18 AM,  <valdis.kletnieks@xxxxxx> wrote:
>> Seeing this at boot with linux-next 20180415.  ACPI gets disabled, hilarity and hijinks
>> result - everything from a lot of stuff can't find an IRQ to the dual-core w/ HT CPU
>> coming up as just 1 core no HT. 20180430 works just fine...
>>
>> [    0.000000] ACPI: Early table checksum verification disabled
>> [    0.000000] ACPI: RSDP 0x00000000CB7F1000 000024 (v02 DELL  )
>> [    0.000000] ACPI: XSDT 0x00000000CB7F1088 000094 (v01 DELL   CBX3     01072009 AMI  00010013)
>> [    0.000000] ACPI: FACP 0x00000000CB7FB720 00010C (v05 DELL   CBX3     01072009 AMI  00010013)
>> [    0.000000] ACPI: DSDT 0x00000000CB7F11B0 00A56F (v02 DELL   CBX3     00000021 INTL 20091112)
>> [    0.000000] ACPI: FACS 0x00000000CCFED080 000040
>> [    0.000000] ACPI: APIC 0x00000000CB7FB830 000072 (v03 DELL   CBX3     01072009 AMI  00010013)
>> [    0.000000] ACPI: FPDT 0x00000000CB7FB8A8 000044 (v01 DELL   CBX3     01072009 AMI  00010013)
>> [    0.000000] ACPI: TCPA 0x00000000CB7FB8F0 000032 (v02 APTIO4 NAPAASF  00000001 MSFT 01000013)
>> [    0.000000] ACPI: MCFG 0x00000000CB7FB928 00003C (v01 DELL   CBX3     01072009 MSFT 00000097)
>> [    0.000000] ACPI: HPET 0x00000000CB7FB968 000038 (v01 DELL   CBX3     01072009 AMI. 00000005)
>> [    0.000000] ACPI: SSDT 0x00000000CB7FB9A0 000415 (v01 SataRe SataTabl 00001000 INTL 20091112)
>> [    0.000000] ACPI: SSDT 0x00000000CB7FBDB8 0009B9 (v01 PmRef  Cpu0Ist  00003000 INTL 20051117)
>> [    0.000000] ACPI: SSDT 0x00000000CB7FC778 000B22 (v01 PmRef  CpuPm    00003000 INTL 20051117)
>> [    0.000000] ACPI: DMAR 0x00000000CB7FD2A0 000080 (v01 INTEL  SNB      00000001 INTL 00000001)
>> [    0.000000] ACPI: ASF! 0x00000000CB7FD320 0000A5 (v32 INTEL   HCG     00000001 TFSM 000F4240)
>> [    0.000000] ACPI: SSDT 0x00000000CB7FD3C8 000579 (v01 AMITCG PROC     00000001 INTL 20051117)
>> [    0.000000] ACPI: BGRT 0x00000000CB7FD948 000038 (v00 ??              01072009 AMI  00010013)
>> [    0.000000] ACPI: SSDT 0x00000000CB7FD980 00198A (v01 NvdRef NvdTabl  00001000 INTL 20091112)
>> [    0.000000] ACPI: Local APIC address 0xfee00000
>> [    0.000000] ACPI: [APIC:0x05] Invalid zero length
>> [    0.000000] ACPI: Error parsing LAPIC address override entry
>> [    0.000000] ACPI: Invalid BIOS MADT, disabling ACPI
>> [    0.000000] tsc: Fast TSC calibration using PIT
>>
>> On next-20180430, the boot goes like this, and continues properly:
>>
>> [    0.000000] ACPI: SSDT 0x00000000CB7FD980 00198A (v01 NvdRef NvdTabl  00001000 INTL 20091112)
>> [    0.000000] ACPI: Local APIC address 0xfee00000
>> [    0.000000] tsc: Fast TSC calibration using PIT
>>
>> Not seeing any commits to drivers/acpi/tables.c other than Al Stone's.
>>
>> This ringing any bells before I go bisecting?
>>
> 
> 
> 

Not off-hand.  Could you please send me a copy of /sys/firmware/acpi/tables/APIC
on this machine?  The commit cd8c65a6442b that I wrote looks like it got pulled
in on 20180430, which if I'm understanding correctly, seems to have fixed the
problem.  Did this work before 20180415?  I assume it did.

What puzzles me is that this message:

   ACPI: [APIC:0x05] Invalid zero length

should only have shown up if the MADT has a broken subtable, and I think that
bit of code has been that way for quite some time (git blame says somewhere
around 2012 when the test for this condition was put in).

-- 
ciao,
al
-----------------------------------
Al Stone
Software Engineer
Red Hat, Inc.
ahs3@xxxxxxxxxx
-----------------------------------