[pve-devel] [RFC qemu-server] avoid setting lun number for drives when pvscsi controller is used

[Fabian Ebner]

Am 14.06.21 um 11:15 schrieb Stefan Reiter:
> On 6/14/21 8:29 AM, Fabian Ebner wrote:
>> Any feedback for this?
>>
> 
> Looks good to me and works as advertised:
> 
> Reviewed-by: Stefan Reiter <s.reiter at proxmox.com>
> Tested-by: Stefan Reiter <s.reiter at proxmox.com>
> 

Thanks for the review!

> The `$unit = $drive->{index} % maxdev` logic doesn't seem useful here 
> though, I
> think it's a copy/paste from the IDE/SATA code below (it was already 
> used before
> this patch). How about this for a cleanup on top:
>  >
> diff --git a/PVE/QemuServer.pm b/PVE/QemuServer.pm
> index c00bde1..afd47b5 100644
> --- a/PVE/QemuServer.pm
> +++ b/PVE/QemuServer.pm
> @@ -1424,7 +1424,6 @@ sub print_drivedevice_full {
>       } elsif ($drive->{interface} eq 'scsi') {
> 
>       my ($maxdev, $controller, $controller_prefix) = 
> scsihw_infos($conf, $drive);
> -    my $unit = $drive->{index} % $maxdev;


Using LSI 53C895A controller and 9 disks I get

     kvm: -device 
scsi-hd,bus=scsihw1.0,scsi-id=8,drive=drive-scsi8,id=scsi8: bad scsi 
device id: 8

As long as scsihw_infos returns the correct thing, I don't see a problem 
sticking to the modular arithmetic ;)

>       my $devicetype = 'hd';
>       my $path = '';
>       if (drive_is_cdrom($drive)) {
> @@ -1452,7 +1451,7 @@ sub print_drivedevice_full {
>       }
> 
>       if (!$conf->{scsihw} || $conf->{scsihw} =~ m/^lsi/ || 
> $conf->{scsihw} eq 'pvscsi') {
> -       $device = 
> "scsi-$devicetype,bus=$controller_prefix$controller.0,scsi-id=$unit";
> +       $device = 
> "scsi-$devicetype,bus=$controller_prefix$controller.0,scsi-id=$drive->{index}"; 
> 
>       } else {
>          $device = 
> "scsi-$devicetype,bus=$controller_prefix$controller.0,channel=0,scsi-id=0"
>              .",lun=$drive->{index}";
> 
> 
>> Am 13.04.21 um 10:24 schrieb Fabian Ebner:
>>> Reported in the community forum[0].
>>>
>>> In QEMU's hw/scsi/vmw_pvscsi.c in the SCSIBusInfo struct, the max_lun 
>>> property
>>> is set to 0. This means that in our stack, one cannot have multiple 
>>> disks and
>>> use 'scsihw: pvscsi' currently, as kvm would fail with
>>>       bad scsi device lun: 1
>>>
>>> Instead of increasing the lun number, increase the scsi-id, as we 
>>> already do for
>>> lsi.* (in hw/scsi/lsi53c895a.c the max_lun property is also 0).
>>>
>>> [0]: https://forum.proxmox.com/threads/kvm-bad-scsi-device-lun-1.84318/
>>>
>>> Signed-off-by: Fabian Ebner <f.ebner at proxmox.com>
>>> ---
>>>
>>> I'm not experienced in this area, so not at all sure if this is the 
>>> proper
>>> solution/workaround.
>>>
>>>    PVE/QemuServer.pm | 2 +-
>>>    1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/PVE/QemuServer.pm b/PVE/QemuServer.pm
>>> index fdb2ac9..4ce663b 100644
>>> --- a/PVE/QemuServer.pm
>>> +++ b/PVE/QemuServer.pm
>>> @@ -1451,7 +1451,7 @@ sub print_drivedevice_full {
>>>            }
>>>        }
>>> -    if (!$conf->{scsihw} || ($conf->{scsihw} =~ m/^lsi/)){
>>> +    if (!$conf->{scsihw} || $conf->{scsihw} =~ m/^lsi/ || 
>>> $conf->{scsihw} eq 'pvscsi') {
>>>            $device = 
>>> "scsi-$devicetype,bus=$controller_prefix$controller.0,scsi-id=$unit";
>>>        } else {
>>>            $device = 
>>> "scsi-$devicetype,bus=$controller_prefix$controller.0,channel=0,scsi-id=0" 
>>>
>>>
>>
>>
>> _______________________________________________
>> pve-devel mailing list
>> pve-devel at lists.proxmox.com
>> https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
>>
>>