[08:02:37] <peluse> just noticed we have 5 log levels defined but only 3 logging macros defined in log.h.  Has someone just not needed the other 3 yet or are they old news?
[08:05:38] <sethhowe> I added the centos6 machine into the pool this morning. Please make sure that your changes are rebased on https://review.gerrithub.io/#/c/368228/ to avoid failing on that machine.
[08:06:21] <peluse> sethhowe, cool
[10:12:51] <jimharris> peluse: i think we just haven't needed the other 3 yet
[10:12:58] <jimharris> drv: can you take a look at https://review.gerrithub.io/#/c/368264/
[10:13:29] <peluse> jimharris, OK, that's what I figured but before I used one of them wanted to ask...
[10:13:34] <peluse> thx
[10:25:36] <jimharris> drv, bwalker: can you take a look at Ziye's GPT patch?  I'd really like to get this one in, since it gives us a real vbdev example in the tree (which will be helpful as I do some of this other vbdev related work)
[10:26:16] <drv> taking a look now
[10:38:45] <drv> jimharris: I'm OK with merging it as-is, but I'll probably have some cleanup patches afterwards
[10:40:58] <jimharris> sounds good - after we merge could you forward your cleanup requests to ziyeyang so he can implement them?
[12:31:55] <jimharris> bwalker: looking first at the write privilege enforcement - i'm thinking we change the bdev parameter to the various bdev I/O functions to a descriptor instead, and keeping the channel parameter
[12:32:33] <jimharris> currently we pass a descriptor to spdk_bdev_get_io_channel() - basically enforcing that you have to have the bdev open to able to get a channel on it
[12:33:50] <jimharris> i'm not seeing an easy way to combine the channel and the descriptor without requiring multiple io_devices per bdev (which I'd like to avoid)
[13:44:18] <bwalker> I agree - swap the bdev argument for the bdev descriptor
[13:44:27] <bwalker> channels and bdev descriptors can't be combined
[13:44:48] <bwalker> make the bdev descriptor const maybe