Page MenuHomeFreeBSD
Differential D38897

stop looping if it fails, we'll try again in 100ms..
Needs ReviewPublic
Actions

Authored by jmg on Mar 3 2023, 6:44 PM.
Tags
None
Referenced Files
F111515127: D38897.diff
Tue, Mar 4, 5:21 PM
Unknown Object (File)
Mon, Feb 24, 11:33 AM
Unknown Object (File)
Thu, Feb 6, 8:11 AM
Unknown Object (File)
Jan 29 2025, 4:17 AM
Unknown Object (File)
Jan 27 2025, 4:57 PM
Unknown Object (File)
Jan 19 2025, 4:04 AM
Unknown Object (File)
Nov 25 2024, 7:43 AM
Unknown Object (File)
Nov 24 2024, 1:25 AM
View All 85 Files
Subscribers
cem
freebsd_igalic.co
imp
markm
obrien

Details

Reviewers
None
Group Reviewers
csprng

Diff Detail

Repository
rG FreeBSD src repository
Lint
Lint Skipped
Unit
Tests Skipped
Build Status
Buildable 50134
Build 47026: arc lint + arc unit

Event Timeline

jmg created this revision.Mar 3 2023, 6:44 PM
Herald added a subscriber: imp. · View Herald TranscriptMar 3 2023, 6:44 PM
jmg requested review of this revision.Mar 3 2023, 6:44 PM
Harbormaster completed remote builds in B50134: Diff 118278.Mar 3 2023, 6:44 PM
jmg added a reviewer: csprng.Mar 3 2023, 6:45 PM
cem added a subscriber: cem.Mar 3 2023, 6:48 PM

No objection. I'll let someone else stamp it.

jmg mentioned this in D38898: virtio_random: pipeline fetching the data... This hides latencies that reach 500us, where otherwise we are busy looping....Mar 3 2023, 11:43 PM
markm added a subscriber: markm.Mar 6 2023, 9:13 AM

I don't have a major objection to this, but I'm nervous of the lower-numbered pools being favoured. I'm not formally requesting changes yet, as I could be persuaded that I'm wrong.

sys/dev/random/random_harvestq.c
303

This may result in the lower-numbered pools being favoured, which is undesirable. All pools need an equal chance at this entropy. A fix may be to make i static, and simply increment it mod(npools), with the loop terminating at the point where it wraps.

freebsd_igalic.co added a subscriber: freebsd_igalic.co.Apr 16 2023, 10:50 AM

the comment says

We don't perform any other verification of an entropy source (i.e., length is allowed to be anywhere from 1 to sizeof(entropy), quality is unchecked, etc)

is there a way we could do that?

not in a live / production system, unless it's via dtrace

Herald added a subscriber: obrien. · View Herald TranscriptApr 16 2023, 10:50 AM

Revision Contents
Changeset List

PathSize
sys/
dev/
random/
4 lines

Diff 118278

View Options

sys/dev/random/random_harvestq.c

Loading...