Hi Justin,

I searched for the ovs in mikelangelo-project repo, and I found the 
virtio-ivshmem driver. I used diff and ported the patch to the latest ovs 
master branch on my local computer. 
The console prints:
OSv v0.24-510-gc604860
...
[E/42 virtio]: Wrong virtio dev id 1110
Assertion failed: status == true (drivers/virtio.cc: virtio_driver: 35)

I see that in drivers/virtio.cc:112 the dev_id can only sit between 0x1000 
and 0x103f, but ivshmem dev_id is 0x1110. I commented this check and 
encountered another problem. I was wondering if the virtio-ivshmem has been 
tested or not.

Maxul


在 2018年4月15日星期日 UTC+8下午6:10:58,lmy20...@gmail.com写道:
>
> I wish to allow inter "OSv"s to communicate efficiently via ivshmem 
> device. Has anyone finished the driver for OSv?
>
> Thanks,
> Maxul
>
> 在 2016年8月29日星期一 UTC+8下午7:52:40,Justin Cinkelj写道:
>>
>> I'm trying to evaluate if ivshmem cuold be helpful for Open MPI app. 
>> ivshmem is presented as PCI device iside VM guest. 
>> In linux, after initial discovery showing that data is at 0xfe000000 
>> address, 1 MB of ivshmem is made accesible via 
>>              char* pch = ioremap(0xfe000000, 1024*1024); 
>>              if (!pch) return 0; 
>>              snprintf((char*)pch, 100, "some-test\n"); 
>> And changed data is visible on host, so this works. 
>>
>> What would be equivalent to ioremap on OSv? 
>>
>> Justin 
>>
>

-- 
You received this message because you are subscribed to the Google Groups "OSv 
Development" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to osv-dev+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to