[libvirt] [PATCH 1/1] Add SCSI pool support.

Dave Allan dallan at redhat.com
Wed Apr 1 14:21:29 UTC 2009

Daniel Veillard wrote:
> On Tue, Mar 31, 2009 at 05:49:47PM -0400, Dave Allan wrote:
>> Dave Allan wrote:
>> Attached is what I think is a final version of the scsi host pool code.  
>> It's the set of patches we've been discussing rolled up into a single  
>> patch, so it should look ok, but let me know if you have additional  
>> comments.  I allowed both disk and rom and took out the XML enhancement  
>> for device type.
>> Dave
>   Patch looks fine to me, but I admit I didn't tested it

Can anybody volunteer to try it out?  (Dan has been shouldering the 
entire burden so far.)  I've done a fair amount of testing over the 
course of writing it, but at least one other person should try it out 
before we commit it.  In particular, I haven't tried it with media in a 
CD or DVD drive (my devel system is remote), so I don't know if a volume 
actually gets created properly.  All the debug output is correct, so I 
am reasonably confident, but it should be run once before committing. 
I'm happy to advise on that effort.

For testing, you'll need a few DMXs and fully redundant fabrics.  ;)

I'm kidding, of course...but seriously, if anybody else that has FC 
infrastructure wants to try it out, that would be great feedback.

> [...]
>> --- a/src/storage_conf.c
>> +++ b/src/storage_conf.c
>> @@ -187,6 +187,14 @@ static virStoragePoolTypeInfo poolTypeInfo[] = {
>>              .formatToString = virStoragePoolFormatDiskTypeToString,
>>          }
>>      },
>> +    { .poolType = VIR_STORAGE_POOL_SCSI,
>> +      .poolOptions = {
>> +            .flags = (VIR_STORAGE_POOL_SOURCE_ADAPTER),
>> +        },
>> +      .volOptions = {
>> +            .formatToString = virStoragePoolFormatDiskTypeToString,
>> +        }
>> +    },
>>      { .poolType = VIR_STORAGE_POOL_DISK,
>>        .poolOptions = {
>>              .flags = (VIR_STORAGE_POOL_SOURCE_DEVICE),
>   I think at some point the C99 initialization should be turned back
>   into plain old initialization, but that's something we should probably
>   do more globally independantly.

Agreed, since it sounds from the other thread like old-style is what 
we're standardizing on.  I don't feel strongly which is better, but we 
should standardize.  I'll try to put together a patch for that sooner 
rather than later.


More information about the libvir-list mailing list