Do not reserve KVA for paging bufs in vm_ksubmap_init(), since now
Fix compilation failures on different arches that have vm_machdep.c not
style(9): break long line.
Remove harmless leftover from code that cycles over zone's kegs. Just use +
This was missed in r343051: make uz_allocs, uz_frees and uz_fails counter(9).
Only do uz_items accounting for zones that have a limit set in uz_max_items.
Make uz_allocs, uz_frees and uz_fails counter(9). This removes some
Fix compilation on 32-bit.
Remove pbuf(9) manual page that is now nearly 100% outdated.
Allocate pager bufs from UMA instead of 80-ish mutex protected linked list.
Add flag LK_NEW for lockinit() that is converted to LO_NEW and passed
o Move zone limit from keg level up to zone level. This means that now
Remove recursive NET_EPOCH_ENTER() from sysctl_ifmalist(), missed in r342872.
Remove support for FreeBSD 9 kernel, which used to change byte order
Remove unnecessary ifdef. With INVARIANTS all KASSERTs are empty statements,
Bring the comment up to date.
Simplify sosetopt() so that function has single return point. No
Timer interval is correctly counted in ticks (1/100 s) in
Mechanical cleanup of epoch(9) usage in network stack.
Remove part of comment that doesn't match reality.
Remove patch that I missed to add to commit index in r489641.
Upgrade minidlna to current snapshot in my repo.
Fix regression in r331368, that broke dumping of UMA startup pages
Use mbufq instead of ifqueue to queue mbufs.
I would like to see a generic code in netgraph that marks hooks with DLT. So, no special messages to be needed, nodes will autosense what's connected to them. However, this is just a wish not a blocker for this change.
As said, I'm not seeking for a review & approval so that I commit it. I'm not a ports committer. I'm seeking for a person who can finalize this ugly patch and commit. Thanks!
Fix yet another edge case in uma_startup_count(). If zone size fits into
For not offpage zones the slab is placed at the end of page. Keg's uk_pgoff
P.S. In pf itself, the dtrace probing should just substitute existing printf(9) debugging.
Merge remote-tracking branch 'FreeBSD/master' into pbuf to take
in and resolve with recent D16666 check-in.
Initialize compatibility epoch tracker for thread0. Fixes
Fix build on some architectures after r340413. On amd64 epoch.h
style(9), mostly adjusting overly long lines.
With epoch not inlined, there is no point in using _lite KPI. While here,
Plug build break after r340411.
The dualism between epoch_tracker and epoch_thread is fragile and
For compatibility KPI functions like if_addr_rlock() that used to have
Uninline epoch(9) entrance and exit. There is no proof that modern
Remove unnecessary include from libstand.
Define QMD_SAVELINK() only for QUEUE_MACRO_DEBUG_TRASH case. Otherwise
My current patchset for UMA has something similar. I also have zone_put_bucket() although named differently, to keep accounting of all items for a zone in buckets. So this change goes in line with my changes. At Netflix we also run the vm_page_cache zone with certain bucket limit, instead of no buckets at all.
I'm not absolutely sure that all possible races are fixed. There still could be dangling ifnet pointers. But that's up to your justification. If you are sure everything is covered, feel free to remove.
Doesn't make sense to remove it from services(5) database. It isn't a database of protocols FreeBSD supports. It is just database of port numbers that are well known.
This is extremely hacky. The only proper fix is to find situations when rcvif isn't cleared for a packet that is reused for transmit and fix them.
Add a note that epoch(9) may change, to untie our hands for any future MFCs.
Remove code that is dead since r287197. Today wlan(4) interfaces aren't
Fix ipw_start(), where logic was reverted in r287197.
If we lost race or were migrated during bucket allocation for the per-CPU
Fix exiting an epoch(9) we never entered. May happen only with MAC.
syncache_chkrst() isn't a public function, so no worries on KBI.
Plug sendfile(2) on a listening socket with proper error code.
Avoid OOB reads in memmem(3).
In PR 227259, a user is reporting that they have code which is using
Since struct inpcb is no longer exported to the userland, the spare field can be removed.
Unbreak nVidia drivers on recent -CURRENT, after change to
Either "free" or "allocated" is misleading here, since an item
Fix comment. The actual meaning of ub_cnt is the opposite.
Unbreak nVidia drivers on recent -CURRENT, after change to
Now that after r335979 the kernel addresses in API structures are
Use if_tunnel_check_nesting() for ng_iface(4).
Check the inp_flags under inp lock. Looks like the race was hidden
Correct r335242. Use unsigned cast instead of abs(). Using abs() gives
Revert r334843, and partially revert r335180.
Since 'ticks' is an int, it may wrap around and cr_ticks at a certain
TCPOUTFLAGS no longer exists since r334843.
UMA memory debugging enabled with INVARIANTS consists of two things:
Add new functionality and syntax to cron(1) to allow to run jobs at a