[Pkg-xen-devel] Bug#963607: Bug#963607: xen-hypervisor-4.11-amd64: Xen Hypervisor kernel fails to load arcmsr module with "arcmsr0: dma_alloc_coherent got error" message.

[debianbugs at red-sand.com]

On 1/07/2020 22:48, Hans van Kranenburg wrote:
> On 7/1/20 5:09 PM, Alex Sanderson wrote:
>> On 1/07/2020 02:05, Hans van Kranenburg wrote:
>>> To do a bug report upstream, we'll need that information.
>> Sorry, completely misunderstood.   Here is the output from the serial
>> terminal as Xen started.
> Thanks! Yes. That's the XEN part that appears first.
>
> I attached the files, AFAIK you can just send attachments to a debian
> bug report. So, you can also do that, next time.
Good to know :) Wasn't sure, I have been using debian since the late
90's and so far have always logged bugs upstream.
> The files justlinux-notime.txt and xen-dom0-notime.txt have the
> timestamps removed from the beginning of the linux part of the logging,
> so that we can diff the files.
>
> E.g. colordiff -u justlinux-notime.txt xen-dom0-notime.txt |less -r
>
> Now, here's an interesting part:
>
> ---- 8< ----
>
> @@ -973,20 +1031,15 @@
>  usb usb1: SerialNumber: 0000:00:1a.0
>  hub 1-0:1.0: USB hub found
>  hub 1-0:1.0: 2 ports detected
> -isci 0000:09:00.0: OEM parameter table found in OROM
> -isci 0000:09:00.0: OEM SAS parameters (version: 1.0) loaded (platform)
> +xen: registering gsi 23 triggering 0 polarity 1
> +Already setup the GSI :23
>  ehci-pci 0000:00:1d.0: EHCI Host Controller
>  ehci-pci 0000:00:1d.0: new USB bus registered, assigned bus number 2
>  ehci-pci 0000:00:1d.0: debug port 2
>  ehci-pci 0000:00:1d.0: cache line size of 64 is not supported
>  ehci-pci 0000:00:1d.0: irq 23, io mem 0xcfc22000
> -isci 0000:09:00.0: SCU controller 0: phy 3-0 cables: {short, short,
> short, short}
> -scsi host1: isci
>  Areca RAID Controller0: Model ARC-1882, F/W V1.56 2019-07-30
> -scsi host0: Areca SAS/SATA RAID Controller (RAID6 capable)
> -arcmsr version v1.40.00.09-20180709
> -
> -arcmsr0: msi-x enabled
> +arcmsr0: dma_alloc_coherent got error
>  ehci-pci 0000:00:1d.0: USB 2.0 started, EHCI 1.00
>  usb usb2: New USB device found, idVendor=1d6b, idProduct=0002,
> bcdDevice= 4.19
>  usb usb2: New USB device strings: Mfr=3, Product=2, SerialNumber=1
>
> ---- >8 ----
>
> So, when running Linux as dom0 under Xen, you get the "arcmsr0:
> dma_alloc_coherent got error". Did you already spot this error?
>
> Just pasting it into a search engine leads me to the following page:
>
> http://faq.areca.com.tw/index.php?action=artikel&cat=16&id=387&artlang=en
>
> ...which only displays an error.
>
> There's also:
>
> https://webcache.googleusercontent.com/search?q=cache:wQHp6mpYuckJ:faq.areca.com.tw/index.php%3Faction%3Dartikel%26cat%3D16%26id%3D387%26artlang%3Den+&cd=1&hl=nl&ct=clnk&gl=nl
>
> ---- 8< ----
>
> Q10051103 - We have problems with the new Linux driver
> (arcmsr.1.20.0X.15-100729) for the raid controller ARC-1880. Our system
> is Linux debian 5.07 (lenny) amd64. Build the driver agains the XEN
> kernel (2.6.26-2-xen-amd64) we got following errors on boot time:
> arcmsr0: dma_alloc_coherent got error arcmsr0: arcmsr_alloc_ccb_pool got
> error
>
> you may get this error message when system have no enough memory space
> for driver requested please configure driver host queue to 64 or less to
> see if it helps.
>
> you can configure driver host queue parameter by typing the driver
> parameter as kernel option during system booting.
>
>     arcmsr.host_can_queue=64
>
> ---- >8 ----
>
> So, you have an 1882, which is close to 1880.
>
> I'd say, try it! This arcmsr.host_can_queue=64 should go on the linux
> command line.
>
> You can use GRUB_CMDLINE_LINUX_XEN_REPLACE in
> /etc/default/grub.d/xen.cfg for that, so that it gets applied when
> booting with Xen, and not when booting standalone Linux. There's some
> documentation in that file.

Yes I have seen the error, unfortunately it was the only information I
had on the problem and I have squeezed everything I can from the net
regarding it.  I even looked at the source code.   In my original bug
report I did actually mention testing that kernel argument but it didn't
help.   I tried that arg and every related memory argument to do with
coherent memory allocation but nothing worked.  

I am about to purchase a new SAS HBA card to test as we have a number of
these servers with Areca cards that I imagine will have the same problem
on Xen 4.11.   I am leaning towards mpt3 driver cards but we have had
problems with mpt3 previously so I am hesitating there too.  mpt2 has
been rock solid. 

If you can think of anything else that I could try that would be
excellent.  

Thanks

Alex


>
> Feel free to ask more about it if it's unclear how the different options
> work together. The text in that file is part of our Xen packages, so
> feedback on it is always appreciated. We might be able to improve something.
>
> Thanks, curious to hear about the results, and otherwise we'll continue
> our searching for a solution,
>
> Hans