- User Since
- Jun 4 2014, 6:42 AM (180 w, 4 d)
Fri, Nov 17
Thu, Nov 16
I agree with Allan's suggestion. I suspect that ENOMEM could come as a signal to one of earlier ioctls to increase the buffer size for an output nvlist.
E.g. look at lzc_ioctl.
So, if the problem lies in first setting errno but not clearing it after a successful call, then the problem should be fixed at its source.
Sun, Nov 12
I've just had an accident and got some log messages that demonstrates how the patch formats an event with a nested nvlist:
Nov 11 07:10:21 trant ZFS: checksum mismatch, zpool=pond path=/dev/ada2p3 offset=1465774328832 size=12288
Thu, Nov 9
Wed, Nov 8
Tue, Nov 7
Thu, Nov 2
Wed, Nov 1
Tue, Oct 31
Mon, Oct 30
Based on the very good points in the feedback I now see that I haven't thought through any of the tough points of supporting those three additional types.
Given that currently they are not passed to the fm compatibility code, it probably makes sense to avoid them in the future rather than struggling to support them.
Further simplify based on suggestions from mav; no need for nopen.
Sun, Oct 29
Track opens instead of depending on providergone, per mav's suggestion.
@mav thank you for the suggestion! Just to clarify, do you mean adding tracking of the open provider count into g_slice_access ?
Sat, Oct 28
Fri, Oct 27
Thu, Oct 26
LGTM. Thank you!