Bug#879792: nvidia-detect: misses nvidia tesla p100 (GP100GL)

[Luca Boccassi]

On Thu, 2017-10-26 at 12:22 +1100, Vincent McIntyre wrote:
> Package: nvidia-detect
> Version: 375.82-6
> Severity: normal
> Tags: patch
> thanks
> 
> Also found in
> 375.82-1~deb9u1
> 375.82-4~bpo9+1
> 384.90-1
> 
> I have a system with a nvidia tesla p100 GPU:
> # lspci -v -s 04:00
> 
> 04:00.0 3D controller: NVIDIA Corporation GP100GL (rev a1)
>         Subsystem: NVIDIA Corporation GP100GL
>         Flags: fast devsel, IRQ 113, NUMA node 0
>         Memory at 91000000 (32-bit, non-prefetchable) [size=16M]
>         Memory at 3b800000000 (64-bit, prefetchable) [size=16G]
>         Memory at 3bc00000000 (64-bit, prefetchable) [size=32M]
>         Capabilities: [60] Power Management version 3
>         Capabilities: [68] MSI: Enable+ Count=1/1 Maskable- 64bit+
>         Capabilities: [78] Express Endpoint, MSI 00
>         Capabilities: [100] Virtual Channel
>         Capabilities: [250] Latency Tolerance Reporting
>         Capabilities: [258] L1 PM Substates
>         Capabilities: [128] Power Budgeting <?>
>         Capabilities: [420] Advanced Error Reporting
>         Capabilities: [600] Vendor Specific Information: ID=0001
> Rev=1 Len=024 <?>
>         Capabilities: [900] #19
>         Kernel modules: nouveau
> 
> # lspci -n -s 04:00
> 04:00.0 0302: 10de:15f8 (rev a1)
> 
> The nvidia-detect script does not detect there is support for this
> unless I supply the PCI ID, which is inconvenient for automation.
> 
> # nvidia-detect
> No NVIDIA GPU detected.
> 
> # nvidia-detect 10de:15f8
> Checking driver support for PCI ID [10de:15f8]
> Your card is supported by the default drivers.
> It is recommended to install the
>     nvidia-driver
> package.
> 
> The fix looks straightforward, allow the processing of lspci -mn to
> match the appropriate device class:
> 
> # diff -u /usr/bin/nvidia-detect{,.new}
> --- /usr/bin/nvidia-detect      2017-10-26 08:43:02.568875942 +0800
> +++ /usr/bin/nvidia-detect.new  2017-10-26 08:42:42.792830644 +0800
> @@ -215,7 +215,7 @@
>                 exit 1
>         fi
> 
> -       NV_DEVICES=$(lspci -mn | awk '{ gsub("\"",""); if ($2 ==
> "0300" && ($3 == "10de" || $3 == "12d2")) { print $1 } }')
> +       NV_DEVICES=$(lspci -mn | sed -e's/"//g'| awk '$2 ~ /^030[0-
> 2]$/{ if ( ($3 == "10de" || $3 == "12d2")) { print $1 } }')
> 
>         if [ -z "$NV_DEVICES" ]; then
>                 echo "No NVIDIA GPU detected."
> 
> #  /usr/bin/nvidia-detect.new
> Detected NVIDIA GPUs:
> 04:00.0 3D controller [0302]: NVIDIA Corporation GP100GL [10de:15f8]
> (rev a1)
> 82:00.0 3D controller [0302]: NVIDIA Corporation GP100GL [10de:15f8]
> (rev a1)
> 
> Checking card:  NVIDIA Corporation GP100GL (rev a1)
> Your card is supported by the default drivers.
> It is recommended to install the
>     nvidia-driver
> package.
> 
> Checking card:  NVIDIA Corporation GP100GL (rev a1)
> Your card is supported by the default drivers.
> It is recommended to install the
>     nvidia-driver
> package.
> 
> 
> 
> The bit I am unsure about is which classes should be matched.
> https://pci-ids.ucw.cz/read/PD/03 lists these subclasses:
>   00	VGA compatible controller	
>   01	XGA compatible controller	
>   02	3D controller	
>   80	Display controller
> 
> So I think my suggested change is ok but there may be other
> considerations
> I am not aware of.
> 
> Comments?
> 
> Kind regards
> Vince

Hi,

Is the Tesla series compatible with the Geforce series drivers that we
package?

They are sorted separately on Nvidia's website at least:

http://www.nvidia.co.uk/download/driverResults.aspx/124885/en-uk

But I've never used them so I'm not sure.

-- 
Kind regards,
Luca Boccassi
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: This is a digitally signed message part
URL: <http://lists.alioth.debian.org/pipermail/pkg-nvidia-devel/attachments/20171026/65fe9690/attachment-0001.sig>