Re: [PATCH 1/1] mdadm: enable Intel Alderlake RST VMD configuration

[Kinga Tanska <kinga.tanska@linux.intel.com>]

On Thu, 18 Aug 2022 17:21:20 +0200
Oldřich Jedlička <oldium.pro@gmail.com> wrote:

> čt 18. 8. 2022 v 16:53 odesílatel Oldřich Jedlička
> <oldium.pro@gmail.com> napsal:
> >
> > čt 18. 8. 2022 v 16:12 odesílatel Kinga Tanska
> > <kinga.tanska@linux.intel.com> napsal:  
> > >
> > > On Fri, 5 Aug 2022 14:50:36 +0200
> > > Oldřich Jedlička <oldium.pro@gmail.com> wrote:
> > >  
> > > > pá 5. 8. 2022 v 13:56 odesílatel Kinga Tanska
> > > > <kinga.tanska@linux.intel.com> napsal:  
> > > > >
> > > > > On Fri,  5 Aug 2022 12:05:45 +0200
> > > > > Oldřich Jedlička <oldium.pro@gmail.com> wrote:
> > > > >  
> > > > > > Alderlake changed UEFI variable name to 'RstVmdV' also and
> > > > > > for VMD devices, so check the updated name for VMD devices
> > > > > > like it is done in the SATA case.
> > > > > >
> > > > > > Signed-off-by: Oldřich Jedlička <oldium.pro@gmail.com>
> > > > > > ---
> > > > > >  platform-intel.c | 19 ++++++++++++-------
> > > > > >  1 file changed, 12 insertions(+), 7 deletions(-)
> > > > > >
> > > > > > diff --git a/platform-intel.c b/platform-intel.c
> > > > > > index a4d55a3..2f8e6af 100644
> > > > > > --- a/platform-intel.c
> > > > > > +++ b/platform-intel.c
> > > > > > @@ -512,8 +512,8 @@ static const struct imsm_orom
> > > > > > *find_imsm_hba_orom(struct sys_dev *hba) #define AHCI_PROP
> > > > > > "RstSataV" #define AHCI_SSATA_PROP "RstsSatV"
> > > > > >  #define AHCI_TSATA_PROP "RsttSatV"
> > > > > > -#define AHCI_RST_PROP "RstVmdV"
> > > > > > -#define VMD_PROP "RstUefiV"
> > > > > > +#define RST_VMD_PROP "RstVmdV"
> > > > > > +#define RST_UEFI_PROP "RstUefiV"
> > > > > >
> > > > > >  #define VENDOR_GUID \
> > > > > >       EFI_GUID(0x193dfefa, 0xa445, 0x4302, 0x99, 0xd8,
> > > > > > 0xef, 0x3a, 0xad, 0x1a, 0x04, 0xc6) @@ -607,7 +607,8 @@
> > > > > > const struct imsm_orom *find_imsm_efi(struct sys_dev *hba)
> > > > > > struct orom_entry *ret; static const char * const
> > > > > > sata_efivars[] = {AHCI_PROP, AHCI_SSATA_PROP,
> > > > > > AHCI_TSATA_PROP,
> > > > > > -
> > > > > > AHCI_RST_PROP};
> > > > > > +
> > > > > > RST_VMD_PROP};
> > > > > > +     static const char * const vmd_efivars[] =
> > > > > > {RST_UEFI_PROP, RST_VMD_PROP}; unsigned long i;
> > > > > >
> > > > > >       if (check_env("IMSM_TEST_AHCI_EFI") ||
> > > > > > check_env("IMSM_TEST_SCU_EFI")) @@ -640,10 +641,14 @@ const
> > > > > > struct imsm_orom *find_imsm_efi(struct sys_dev *hba)
> > > > > >               break;
> > > > > >       case SYS_DEV_VMD:
> > > > > > -             if (!read_efi_variable(&orom, sizeof(orom),
> > > > > > VMD_PROP,
> > > > > > -                                    VENDOR_GUID))
> > > > > > -                     break;
> > > > > > -             return NULL;
> > > > > > +             for (i = 0; i < ARRAY_SIZE(vmd_efivars); i++)
> > > > > > {
> > > > > > +                     if (!read_efi_variable(&orom,
> > > > > > sizeof(orom),
> > > > > > +
> > > > > > vmd_efivars[i], VENDOR_GUID))
> > > > > > +                             break;
> > > > > > +             }
> > > > > > +             if (i == ARRAY_SIZE(vmd_efivars))
> > > > > > +                     return NULL;
> > > > > > +             break;
> > > > > >       default:
> > > > > >               return NULL;
> > > > > >       }  
> > > > >
> > > > > Hi,
> > > > >
> > > > > please have a look at the following mail:
> > > > > https://marc.info/?l=linux-raid&m=165969352101643&w=2  
> > > >
> > > > Sorry for double-posting, I received rejection emails regarding
> > > > HTML content. Gmail switched to HTML.
> > > >
> > > > Hi, the described issue applies specifically in the
> > > > SYS_DEV_SATA (SATA configuration) case, so it should not apply
> > > > to SYS_DEV_VMD (VMD configuration) one.
> > > >
> > > > For me, the platform output looks reasonable (I have RAID 0
> > > > active): 
> > > > #> sudo mdadm --detail-platform  
> > > >        Platform : Intel(R) Rapid Storage Technology
> > > >         Version : 19.0.7.5579
> > > >     RAID Levels : raid0 raid1 raid10 raid5
> > > >     Chunk Sizes : 4k 8k 16k 32k 64k 128k
> > > >     2TB volumes : supported
> > > >       2TB disks : supported
> > > >       Max Disks : 32
> > > >     Max Volumes : 2 per array, 4 per controller
> > > >  3rd party NVMe : supported
> > > >  I/O Controller : /sys/devices/pci0000:00/0000:00:0e.0 (VMD)
> > > >  NVMe under VMD : /dev/nvme0n1 (S6P1NS0T318266R)
> > > >  NVMe under VMD : /dev/nvme1n1 (S6P1NS0T318223V)
> > > >
> > > > Without the patch the platform isn't even recognized. Common to
> > > > both changes is the usage of the new UEFI variable 'RstVmdV',
> > > > not the changes to the controller.
> > > >
> > > > Regards,
> > > > Oldrich.
> > > >  
> > > > >
> > > > > Regards,
> > > > > Kinga Tanska  
> > >
> > > Hello,
> > >
> > > I've done test to check if your patch doesn't change way of
> > > recognizing controllers. I've got two SATA controllers - first
> > > one with AHCI mode and the other one with RAID MODE enabled.
> > > Command "mdadm --detail-platform" should display info only about
> > > controller with RAID MODE:
> > >
> > > mdadm: imsm capabilities not found for controller:
> > > /sys/devices/pci0000:00/0000:00:17.0 (type SATA) Platform :
> > > Intel(R) Rapid Storage Technology enterprise Version : 5.3.0.1052
> > >     RAID Levels : raid0 raid1 raid10 raid5
> > >     Chunk Sizes : 4k 8k 16k 32k 64k 128k
> > >     2TB volumes : supported
> > >       2TB disks : supported
> > >       Max Disks : 8
> > >     Max Volumes : 2 per array, 8 per controller
> > >  I/O Controller : /sys/devices/pci0000:00/0000:00:11.5 (SATA)
> > >           Port4 : - non-disk device (TEAC DV-W28S-B) -
> > >           Port0 : - no device attached -
> > >           Port1 : - no device attached -
> > >           Port2 : - no device attached -
> > >           Port3 : - no device attached -
> > >           Port5 : - no device attached -
> > >
> > >        Platform : Intel(R) Rapid Storage Technology enterprise
> > >         Version : 5.3.0.1052
> > >     RAID Levels : raid0 raid1 raid10 raid5
> > >     Chunk Sizes : 4k 8k 16k 32k 64k 128k
> > >     2TB volumes : supported
> > >       2TB disks : supported
> > >       Max Disks : 24
> > >     Max Volumes : 2 per array, 24 per controller
> > >  3rd party NVMe : supported
> > >  I/O Controller : /sys/devices/pci0000:17/0000:17:05.5 (VMD)
> > >  I/O Controller : /sys/devices/pci0000:d7/0000:d7:05.5 (VMD)
> > >  NVMe under VMD : /dev/nvme3n1 (PHLJ915000201P0FGN)
> > >  NVMe under VMD : /dev/nvme2n1 (PHLJ915003201P0FGN)
> > >  I/O Controller : /sys/devices/pci0000:85/0000:85:05.5 (VMD)
> > >  I/O Controller : /sys/devices/pci0000:ae/0000:ae:05.5 (VMD)
> > >  I/O Controller : /sys/devices/pci0000:5d/0000:5d:05.5 (VMD)
> > >  NVMe under VMD : /dev/nvme0n1 (PHFT536600GT400GGN)
> > >  NVMe under VMD : /dev/nvme1n1 (CVFT523100122P0KGN)
> > >
> > > But with your patch this command returns info about all
> > > controllers which is not correct:
> > >
> > > Platform : Intel(R) Rapid Storage Technology enterprise
> > >         Version : 5.3.0.1052
> > >     RAID Levels : raid0 raid1 raid10 raid5
> > >     Chunk Sizes : 4k 8k 16k 32k 64k 128k
> > >     2TB volumes : supported
> > >       2TB disks : supported
> > >       Max Disks : 8
> > >     Max Volumes : 2 per array, 8 per controller
> > >  I/O Controller : /sys/devices/pci0000:00/0000:00:17.0 (SATA)
> > >           Port2 : /dev/sdd (WDEBMLJ2)
> > >           Port3 : /dev/sde (BTPR2300034W120LGN)
> > >           Port0 : /dev/sdb (CVTS5396007S240JGN)
> > >           Port1 : /dev/sdc (ZDE0XM9Z)
> > >           Port4 : - no device attached -
> > >           Port5 : - no device attached -
> > >           Port6 : - no device attached -
> > >           Port7 : - no device attached -
> > >
> > >        Platform : Intel(R) Rapid Storage Technology enterprise
> > >         Version : 5.3.0.1052
> > >     RAID Levels : raid0 raid1 raid10 raid5
> > >     Chunk Sizes : 4k 8k 16k 32k 64k 128k
> > >     2TB volumes : supported
> > >       2TB disks : supported
> > >       Max Disks : 8
> > >     Max Volumes : 2 per array, 8 per controller
> > >  I/O Controller : /sys/devices/pci0000:00/0000:00:11.5 (SATA)
> > >           Port4 : - non-disk device (TEAC DV-W28S-B) -
> > >           Port0 : - no device attached -
> > >           Port1 : - no device attached -
> > >           Port2 : - no device attached -
> > >           Port3 : - no device attached -
> > >           Port5 : - no device attached -
> > >
> > >        Platform : Intel(R) Rapid Storage Technology enterprise
> > >         Version : 5.3.0.1052
> > >     RAID Levels : raid0 raid1 raid10 raid5
> > >     Chunk Sizes : 4k 8k 16k 32k 64k 128k
> > >     2TB volumes : supported
> > >       2TB disks : supported
> > >       Max Disks : 24
> > >     Max Volumes : 2 per array, 24 per controller
> > >  3rd party NVMe : supported
> > >  I/O Controller : /sys/devices/pci0000:17/0000:17:05.5 (VMD)
> > >  I/O Controller : /sys/devices/pci0000:d7/0000:d7:05.5 (VMD)
> > >  NVMe under VMD : /dev/nvme3n1 (PHLJ915000201P0FGN)
> > >  NVMe under VMD : /dev/nvme2n1 (PHLJ915003201P0FGN)
> > >  I/O Controller : /sys/devices/pci0000:85/0000:85:05.5 (VMD)
> > >  I/O Controller : /sys/devices/pci0000:ae/0000:ae:05.5 (VMD)
> > >  I/O Controller : /sys/devices/pci0000:5d/0000:5d:05.5 (VMD)
> > >  NVMe under VMD : /dev/nvme0n1 (PHFT536600GT400GGN)
> > >  NVMe under VMD : /dev/nvme1n1 (CVFT523100122P0KGN)
> > >
> > > Please analyze it.  
> >
> > Hi Kinga Tanska,
> >
> > My patch touches only the VMD controllers. I see that there are no
> > changes to VMD display in your output. My patch is only about VMD.
> >
> > I guess that your issue is identical to what you already wrote to
> > RAID mailing list 5th of August. Correct? I cannot investigate this
> > for you, I am neither the author of the change, nor I have RST
> > enterprise RAID controller, sorry. Please contact the patch
> > submitter/author.  
> 
> I just checked the patch code, I cannot debug this. You are most
> probably affected by patch which you already identified:
> 
>   [PATCH 19/23] mdadm: enable Intel Alderlake RSTe configuration.
> 
> It tries to exclude SATA in AHCI mode (PCI class 0x010601) from
> the changed EFI variable check. What is your `lspci -nn` output?
> It could help in verifying that my guess is correct.
> 
> Cheers,
> Oldrich.
> 
> >
> > Regards,
> > Oldrich.
> >  
> > > Regards,
> > > Kinga Tanska  

Hi Oldrich,

I've noticed that your patch is on top of mdadm-CI for-jes/20220728
branch. This branch contains previous patch for Alderlake RST VMD
support ([PATCH 19/23] mdadm: enable Intel Alderlake RSTe
configuration). This patch causes defect, which I've described. You were
right saying that this patch affected me.
Please move your patch to the HEAD and then I've been able to test it
again.

Thanks,
Kinga