On 11/2/20 12:49 AM, Jason Wang wrote:
On 2020/11/2 下午2:36, Jason Wang wrote:
The need to share event/control vqs between devices is a problem though,
and sending lots of ioctls on things like reset is also not that
elegant.
Jason, did you have a good solution in mind?
Nope, I'm not familiar with SCSI so I don't even know sharing evt/cvq
is possible. Consider VHOST_SCSI_MAX_VQ is already 128 per device.
Mike's proposal seems to be better.
Hey, which proposal are you saying was best?
1. Add on to the current scsi mq design where we are doing a single
device and multiple vqs already. So basically just fix what we have and
add in patches 12 - 16 to do a thread per VQ?
2. The proposal I stated to hack up over the weekend to try and support
the current design and then add in support for your multiple device
single vq design:
http://archive.lwn.net:8080/linux-scsi/292879d9-915d-8587-0678-8677a800c613@xxxxxxxxxx/
Thanks
Btw, it looks to me vhost_scsi_do_evt_work() has the assumption of iovec
layout which needs to be fixed.
I wanted to be clear, because I thought you meant #1, but this comment
seems like it would only be for #2.
