No bump because it wouldn't have been possible to acquire the lock in
VOP_RECLAIM anyway -- instant deadlock because vn_lock waits to
transition out of the RECLAIMING state first. Benefit is that we can
now assert ownership of the lock in any operations called by
VOP_RECLAIM.
Discussed on tech-kern:
https://mail-index.netbsd.org/tech-kern/2017/04/01/msg021751.html
It's a document from June 2015.
DESCRIPTION
The NetBSD kernel provides several synchronization and interrupt control
primitives. This manpage aims at giving an overview of these interfaces
and their proper application. This document includes also basic kernel
thread control primitives and rough overview of the NetBSD kernel design.
Part of interfaces are missing, like new mechanisms for networking SMP,
as this documentation page predates them.
Initial review back in 2015 by Thomas Klausner <wiz>
This example presents a C module with its device (/dev entry) and its
content generator with algorithm defined in Lua state. The Lua state can
be changed dynamically from userland, without interruption of read(2) over
the device.
This is an example how to call Lua code from C.
* restored --logfile support as a few people complained it vanished
The new logging code even makes the overall binary size smaller
on most platforms.
* BPF filter now trims garbage trailing the payload
OK, it's not garbage, but userland doesn't know some drivers append
FCS to it.
* support NetBSD's RO_MSGFILTER socket option to reduce avoid context
switching for route(4) messages that don't interest us.
* Don't open sockets if just sending signals.
* HMAC-MD5 test's now check expectations in code rather than relying
on visual confirmation.
* added eloop-bench to test performance of eloop with available
polling mechanisms.
Some system does not detect ioapic when "boot -1", disable acpi, and so on.
In such cases, msi/msix doesn't work, so disable them.
This patch is implemented by nonaka@n.o, I just commit by proxy, thanks.
