tech-kern archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: On softints, softnet_lock and sleeping (aka ipv6 vs USB network interfaces)



>>> That aside, can softints even interrupt softints, or are the
>>> priorities only about who goes first if two softints are scheduled
>>> `simultaneously' (as far as softint_dispatch can discern)?
>> The latter, iiuc.
> I checked on arm, and it looks like higher-priority softints can
> interrupt lower-priority softints, [...]

They certainly used to be able to.  When I first saw soft interrupts
(back around BSD4.3, I think), they looked a whole lot like the VAX
SIRR/SISR facilities, which _are_ a "higher priority softints can
interrupt lower priority ones" system.  (This would not be surprising,
given how influential the VAX was at the time.)  There were even, IIRC,
comments about how _this_ softint was higher priority than _that_
softint because its subsystem should have better response time.

I've seen some softint implementations which map all the softint levels
to something like one or two hardware priority levels.  I've always
seen this as a case of "the hardware doesn't make it easy to do this
right and doing it right in software is more work than I want to get
into now".

Not that am any kind of authority on NetBSD's softint design.  But
then, I'm not convinced anyone truly designed it, rather than it just
growing by accretion since softints were created (by UCB?).

/~\ The ASCII				  Mouse
\ / Ribbon Campaign
 X  Against HTML		mouse%rodents-montreal.org@localhost
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B


Home | Main Index | Thread Index | Old Index