by yamt@.
- Introduce SOBJ_SLEEPQ_LIFO, and use for LWPs sleeping via _lwp_park.
libpthread enqueues most waiters in LIFO order to try and wake LWPs that
ran recently, since their working set is more likely to be in cache.
Matching the order of insertion reduces the time spent searching queues
in the kernel.
- Do not boost the priority of LWPs sleeping in _lwp_park, just let them
sleep at their user priority level. LWPs waiting for some I/O event in
the kernel still wait with kernel priority and get woken more quickly.
This needs more evaluation and is to be revisited, but the effect on a
variety of benchmarks is positive.
- When waking LWPs, do not send an IPI to remote CPUs or arrange for the
current LWP to be preempted unless (a) the thread being awoken has kernel
priority and has higher priority than the currently running thread or (b)
the remote CPU is idle.
Also, error case might happen before proc_vmspace_getref() (hi <ad>!).
Thanks CID 4551 and 4552. This is serious, pullup will be requested.
OK by <wrstuden>.
and dom_sa_len members from struct domain. Pools of fixed-size
objects are too rigid for sockaddr_dls, whose size can vary over
a wide range.
Return sockaddr_dl to its "historical" size. Now that I'm using
malloc(9) instead of pool(9) to allocate sockaddr_dl, I can create
a sockaddr_dl of any size in the kernel, so expanding sockaddr_dl
is useless.
Avoid using sizeof(struct sockaddr_dl) in the kernel.
Introduce sockaddr_dl_alloc() for allocating & initializing an
arbitrary sockaddr_dl on the heap.
Add an argument, the sockaddr length, to sockaddr_alloc(),
sockaddr_copy(), and sockaddr_dl_setaddr().
Constify: LLADDR() -> CLLADDR().
Where the kernel overwrites LLADDR(), use sockaddr_dl_setaddr(),
instead. Used properly, sockaddr_dl_setaddr() will not overrun
the end of the sockaddr.
identify sockaddr_dl abuse that remains in the kernel, especially
the potential for overwriting memory past the end of a sockaddr_dl
with, e.g., memcpy(LLADDR(), ...).
Use sockaddr_dl_setaddr() in a few places.
(semu + seminfo.semmnu) is wrong, because the type of semu is int*.
You could fix the offset ((char *)semu + seminfo.semusz), but simply
putting the condvars first is more clear.
lwp_exit() before suspending. In such case, LWP might be already freed after
cv_wait_sig() and checking the list of LWPs via lwp_find() is necessary.
Possible problem catched by Andrew Doran.
checking for a readonly lookup. This shouldn't make a difference
now, though, as the only RDONLY lookup is done by getcwd(), and
that a) doesn't create files b) calls LOOKUP directly anyway.
Also, fix comment I managed to miss in the previous commit (I didn't
expect the same comment to be there twice).
it possible to get the pid, euid and egid of the process at the remote
end at the time it did bind() or connect().
Add a new libc function, getpeereid() to easily get at the euid and egid.
As a consequence, bump libc's minor number.
Document the LOCAL_PEEREID socket option in unix(4).
Based on contribution by Arne H. Juul, minor modifications by myself.
introduce vrele2(), which allows to release vnodes the way lfs
sometimes wants it:
+ without calling inactive
+ inserting the vnode at the head of the freelist (this is a very
questionable optimization that isn't even enabled by default,
but I went along with the same semantics for now)
1 microsecond into the future, the thread could enter an untimed sleep.
- Change the signature of _lwp_park() to accept an lwpid_t and second
hint pointer, but do so in a way that remains compatible with older
pthread libraries. This can be used to wake another thread before the
calling thread goes asleep, saving at least one syscall + involuntary
context switch. This turns out to be a fairly large win on the condvar
benchmarks that I have tried.
- Mark some more syscalls MP safe.
into sdl_data[].
Move the macro satocsdl() to net/if_dl.h, and introduce satosdl().
Add some helpers for initializing sockaddr_dl (sockaddr_dl_init),
for finding out the length to put in a sockaddr_dl's sdl_len member
(sockaddr_dl_measure), and for setting the link-layer address in
a sockaddr_dl to a new value (sockaddr_dl_setaddr).
Make sockaddr_copy() panic if the caller tries to copy a sockaddr
to a destination where it will not fit.
- Make structures CPU-cache friendly, as suggested and explained
by Andrew Doran. CACHE_LINE_SIZE definition is invented.
- Use current CPU if NULL is passed to the workqueue_enqueue().
- Implemented MI CPU index, which could be used as an index of array.
Removed linked-lists usage for work queues.
The roundup2() function avoids division, but works only with power of 2.
Reviewed by: <ad>, <yamt>, <tech-kern>
do not leak file descriptors when sending a datagram with SCM_RIGHTS
fails. Patch from Gary Thorpe, based on changes in FreeBSD and work
from Christian Biere.
knew what it was supposed to be used for and wrstuden gave a go-ahead
* while rototilling, convert file systems which went easily to
use VFS_PROTOS() instead of manually prototyping the methods
conjunction with LK_DRAIN. This has the same effect as LK_DRAIN
except it atomically does NOT mark the lock as drained. This
guarantees that when we got the lock, we were the last one currently
waiting for the lock.
Use LK_DRAIN|LK_RESURRECT in vclean() to make sure there are no
waiters for the lock. This should fix behaviour theoretized to be
caused by vfs_subr.c 1.289 which caused vclean() to run into
completion and free the vnode before all lock-waiters had been
processed. Should therefore fix the "simple_lock: unitialized lock"
problems seen recently.
thanks to Juergen Hannken-Illjes for some analysis of the problem
and Erik Bertelsen for testing
need to understand the locking around that field. Instead of setting
B_ERROR, set b_error instead. b_error is 'owned' by whoever completes
the I/O request.
setting vnode sizes, is handled elsewhere: file system vnode creation
or spec_open() for regular files or block special files, respectively.
Add a call to VOP_MMAP() to the pagedvn exec path, since the vnode
is being memory mapped.
reviewed by tech-kern & wrstuden