richlowe: very possible

hrm...

[illumos-gate] 18106 Update AMD microcode to 20251202 -- Andy Fiddaman <illumos⊙fn>

[illumos-gate] 18108 libtopo reports the wrong errno on enumeration failure -- Andy Fiddaman <illumos⊙fn>

we've run into some systems that really want _smaller_ I/Os while zfs does some amount of work to do the opposite...

though there are some SCSI VPD pages that advertise preferred I/O sizes, we don't use them, but even if sd does, there'd still be no way (currently) to bubble that up to zfs

(just to avoid zfs aggregating I/Os for just sd to split them up again)

ioctl() query to allow zfs to recognize this?:)

yeah, for some of this, maybe a new dkio(4I) ioctl

(I don't see a way to add it into an existing one w/o breaking things)

i'll need to look at the nvme specs, but I vaguely recall there's might be something analogous there as well..

granted, i've found the mroe recent versions a lot more of a pain to find things than the older versions

Found a bug in the ISC DHCP server over the weekend with a one-line fix: OpenIndiana/oi-userland #26576 ; impact is that some DHCP server replies may get broadcast to irrelevant interfaces. distros might want to look at that.

(only matters if you have devices that set the "please broadcast your reply" flag in requests.)

[illumos-gate] 18103 ena: validate device-supplied register offsets and descriptor indices -- Andy Fiddaman <illumos⊙fn>