it's really a possible programming error rather than something actually
happening in wild - since rev. 1.57 in 2012 this condition leaked
locked mutex, and nobody noticed
xennet_tx_complete() is now called only in xennet_softstart() just
before processing Tx queue; xennet_handler() just schedules the
softint for xennet_softstart() iff the Tx ring list is completely empty
no performance changes observed
XXX the detach and suspend code expects xennet_handler() to collect
XXX the Tx responses, the tsleep() loop should still work since the tsleep()
XXX would allow the softint to trigger; untested though
the set-up functions will be called, so it is perfectly acceptable
for a compat code's routine to be called ahead of the code in other
parts of the kernel.
So make sure that the 2nd level sysctl node ``vfs.generic'' exists
before trying to add the 3rd level entries.
XXX Rather than creating the 2nd level node in two places, we could
XXX add the shared ``vfs.generic'' node to sysctl_init_base.c but
XXX this is left for another day.
called (unless pause).
- Opening /dev/audio always initializes pause with false. Therefore
it always starts recording.
- Opening /dev/sound inherites pause from the sticky parameter.
Therefore whether /dev/sound starts recording or not depends on the
sticky pause parameter.
This fixes two problems:
- Opening /dev/audio didn't start recording after merging isaki-audio2.
- Opening /dev/sound didn't start recording regardless of the sticky
pause, probably since long time ago (at least netbsd-7).
The code already handles "snps,dwc3" not being a subordinate of
an SoC-specific node, but the string was missing from the match
routine.
Necessitated by the sun50i-h6 device tree update on Jan 3 2020, which
elimiated the "allwinner,sun50i-h6-dwc3" node and placed all of the
clocks, resets, etc. directly under the "snps,dwc3" node.
result type of strerror() (and strerror_l()). While that once should
really have been present, when strerror() was invented, there was no
"const" qualifier in C to apply, and now the way the code is writtem
(really needs to be because of NLS support) the const is no longer
really appropriate.
Applications still shouldn't attempt to modify the result however.
While here also document (but comment it out since it isn't
available - yet) strerror_lr(). To include that, simply
uncomment the relevant lines, and (twice I think) s/returns/return/
on lines just after currently commented out lines (that is, it
currently says, "A returns" after the comments are returned, we
need it to be "A and B return" - the "and B" appears when the comment
markers are removed, removing the 's' from returns must be done manually.
In addition to adding strerror_l() some additional enhancements were
made to the general strerror() doc.
The restriction that a fd passed to fdatasync(2) must be writable was
added in 2003 in order to comply with POSIX. Since then, POSIX has
removed that requirement, and POSIX-valid programs have been therefore
encountering errors on NetBSD.
Patch by Paul Ripke after discussion on netbsd-users. Issue
discovered with pkgsrc/databases/mongodb3 as used by pkgsrc/net/unifi.
unsigned equivalent of -1.
While here, guarantee, even when !NLS, that nothing here (not even
snprintf deciding to complain about EILSEQ or something) can ever
alter errno (ie: always save and restore it, not only in the NLS case).
The functions here must never alter errno, whatever happens.
