With a the result of trace_is_enabled() cached in p->p_trace_enabled the
cost of 2 correctly predicted branches is insignificant.
Reorder prologue slightly, should reduce memory stalls.
can save the result of trace_is_enabled() so that it can be efficiently
determined on every system call without having 2 separate syscall functions.
The death of syscall_fancy() looms.
change from an adaptive to spin mutex @ IPL_TTY. This seems to mitigate
or eliminate a lot of locking-related issues (such as a hang when running
glxgears under AIGLX). Tested by me and Vincent <10.50@free.fr>.
In this code, we have three locks: irq, dma, and device. If anyone has
any thoughts on how to improve this situation, e.g. by using different
lock types/priorities/soft interrupts/etc. for different locks, please
share them with me.
NOTE: I've had the best results thusfar using Mesa 7.0.2-rc1 (r300 users
could even try the r300 driver code from the master branch) and xorg-server
1.4+patches from GIT.
- make sure to set HD_ROW1_ADDR before putting characters at the first line
in hd44780_ddram_redraw()
- use memmove(9) rather than bcopy(9) for possibly overlapped copies
group block buffer busy. If filesystem has any active snapshots, bawrite
can come back trying to allocate new snapshot data block from the same
cylinder group and cause deadlock.
From FreeBSD Rev. 1.117
The latter actually function (due to luck) for calls with one argument,
but will fail badly if more than one is required.
Noticed as an error in the ktrace outut by Nicolas Joly, reported on
tech-kern.
GENERIC_LAPTOP, which is not derived from that and misses a lot of
drivers and options. The NOACPI kernel doesn't have this issue, so make
it more useful by including apm and also the PCIBIOS fixup options
needed on many non-ACPI system. Tested by freza@ on an older system
where APM and ne@PCMCIA had issues in the past and which works fine with
-current.
With stackgap and CHECK_ALT_xxx removal, some linux32 and netbsd32
syscalls are now identical.
To avoid code duplication, remove the linux32 definition and use the
netbsd32 one (no functional change).
locking and makes it possible to run file systems which create
threads. It also makes rump file system behaviour better match
file system behaviour in the kernel.
I don't like doing this, but there's too much pain to get this file
to compile clean due to how SPINLOCK_{BACKOFF,SPIN}_HOOK and
mb_write() are spread out in weird weird places throughout MD code.
code to retry on SIOCGIFCONF not returning all interfaces is broken,
this unreasonably large buffer should handle the case of a machine
with lots of (usually virtual) interfaces. dhcpd 4 uses a different
method to get interface configuration, so this kludge should go away
on upgrade.
Fix based on understanding of the code, but has been compile-tested
only, to address failure report received from Michael Graff.
to avoid possible race condition between CPU writeback vs DMA write
- remove unnecessary FS and LS bits from RxD->xd0
(sys/dev/ic/tulipvar.h sets them but maybe they are ignored?)
get our versions if any of the functions aren't present on the host system.
Still assumes if the functions are all there, they work like ours, which
may be a problem too.
