16:56:09 [illumos-gate] 16429 zpool hot spare replacement not grabbing similar/appropriate sized drive -- Andy Fiddaman 17:43:32 thanks andyf 17:47:16 Will that be backportedd to 151046? 17:47:19 You're welcome. It will likely be in next week's OmniOS updates for r46/r48/r50. 17:47:28 great! 17:47:57 The hosts that need it are scheduled for patching on 11 JUN so that's plenty of time. 20:06:22 [illumos-gate] 16524 driver for 38xx HBA in illumos -- Stuart Maybee 20:07:51 Woodstock: I have definitely found raw framebuffer performance to vary from system to system haha 20:08:47 I believe tsoome made some changes so that we only _write_ to the video region now, without ever reading back, so the writes can all be posted -- this helped _a lot_ with, say, systems with a BMC-emulated VGA adapter 20:09:30 https://github.com/oxidecomputer/clock makes direct use (not through wscons) of the framebuffer mapping, essentially, and works perhaps surprisingly well even on a relatively low end system (Dell Wyse 3040) 20:09:52 yep. you never want to read from write combined memory:) 20:10:09 yeeeah 20:10:09 there's something else very odd with that system, installing OI took more than 12h :) 20:10:16 That's excruciating 20:10:35 And yes there must surely be some underlying issue there 20:27:44 it's a thinkcentre m600, intel pentium n3000 1.04ghz :) 20:27:49 but it shouldn't be that slow 20:46:21 is it still slow after the install, or did you give up after 12 hours? 20:55:20 ide cd acting up? or is the shell prompt slow as well? 21:06:58 oh yeah, if it's IDE, that's going to be painful, but shouldn't be that painful 21:07:07 our IDE driver is still PIO 21:07:23 and I doubt anyone's interested in the effort to add DMA support to it 21:08:55 hrm... is mac_soft_ring.h considered 'private' to mac? we don't ship it, but I thought the (very informal) convention was to add _impl for more 'private' things 21:09:30 (i'm at least prototyping out adding the packet disturber stuff to flows, but trying to figure out the 'best' place to share bits for tx/rx 21:10:33 if it works well enough, i'll throw it up for review... 21:11:32 there are though a few differences in behavior compared w/ ipd, but i suspect given the use case for it, it's probably ok 21:12:41 mainly on tx 21:13:47 since you might have a chain of packets... easy enough to just drop or corrupt one packet in the chain, but delay would end up delaying all of the packets in a chain w/o getting complicated 21:43:23 anyone here familiar with fmtopo? I'm trying to set LOCATE on a drive bay and I'm stuck with the -P arugments. 21:50:58 nm, I think I've figured it out.