Add {} braces so that the code conforms to the indentation.
Fortunately, I don't think doing the assignment of cap->tsomax
unconditionally causes any problem.
For an NFSv4 mount with the "nocto" option, don't get the
up to date file attributes upon close. This reduces the
Getattr RPC count by about 65% for software builds.
An execute-only fd (opened with O_EXEC) allows neither read() nor write()
and is therefore incompatible with all stdio modes. Therefore, the [EINVAL]
error applies.
Also adjust the similar check in freopen() with a NULL path, even though
this checks an fd which is already from a FILE.
andrew [Mon, 21 Apr 2014 09:43:22 +0000 (09:43 +0000)]
Rename the fp{get,set}* files so they no longer conflict with the softfloat
version of these files. Keep them within this directory so they can be used
to implement the armv6 version of these functions.
adrian [Mon, 21 Apr 2014 06:07:08 +0000 (06:07 +0000)]
Rewrite the cleanup code to, well, actually work right.
The existing cleanup code was based on the Atheros reference driver
from way back and stuff that was in Linux ath9k. It turned out to be ..
rather silly.
Specifically:
* The whole method of determining whether there's hardware-queued frames
was fragile and the BAW would never quite work right afterwards.
* The cleanup path wouldn't correctly pull apart aggregate frames in the
queue, so frames would not be freed and the BAW wouldn't be correctly
updated.
So to implement this:
* Pull the aggregate frames apart correctly and handle each separately;
* Make the atid->incomp counter just track the number of hardware queued
frames rather than try to figure it out from the BAW;
* Modify the aggregate completion path to handle it as a single frame
(atid->incomp tracks the one frame now, not the subframes) and
remove the frames from the BAW before completing them as normal frames;
* Make sure bf->bf_next is NULled out correctly;
* Make both aggregate session and non-aggregate path frames now be
handled via the incompletion path.
TODO:
* kill atid->incomp; the driver tracks the hardware queued frames
for each TID and so we can just use that.
This is a stability fix that should be merged back to stable/10.
adrian [Mon, 21 Apr 2014 02:09:14 +0000 (02:09 +0000)]
* Modify the debugging output from pause/resume to note the TID and STA
MAC
* Now that the paused < 0 bugs have been identified, make the DPRINTF()
a device_printf() again. Anything else that shows up here needs to be
fixed immediately.
adrian [Mon, 21 Apr 2014 01:02:49 +0000 (01:02 +0000)]
Fix a cleanup hang if cleanup gets called _during_ an active cleanup.
During power save testing I noticed that the cleanup code is being
called during a RUN->RUN state transition. It's because the net80211
stack is treating that (for reasons I don't quitey know yet) as a
reassociation and this calls the node cleanup code. The reason it's
seeing a RUN->RUN transition is because during active power save
stuff it's possible that the RUN->SLEEP and SLEEP->RUN transitions
happen so quickly that the deferred net80211 vap state code
"loses" a transition, namely the intermediary SLEEP transition.
So, this was causing the node reassociation code to sometimes be called
twice in quick succession and this would result in ath_tx_tid_cleanup()
to be called again. The code calling it would always call pause, and
then only call resume if the TID didn't have "cleanup_inprogress" set.
Unfortunately it didn't check if it was already set on entry, so it
would pause but not call resume. Thus, paused would be called more
than once (once before each entry into ath-tx_tid_cleanup()) but resume
would only be called once when the cleanup state was finished.
This doesn't entirely fix all of the issues seen in the cleanup path
but it's a necessary first step.
Since this is a stability fix, it should be merged to stable/10 at some
point.
Modify the NFSv4 client create/mkdir RPC so that it acquires
post-create/mkdir directory attributes. This allows the RPC to
name cache the newly created directory and reduces the lookup RPC
count for applications creating a lot of directories.
Apply a slightly different color hammer to arm/release.sh:
- gcc(1) fails to build usr.bin/dtc
- lack of WITH_GNUCXX=1 causes cc1plus(1) calls to fail
- u-boot fails to build with clang (hard-coded gcc(1) calls)
Implement the proper incantation of WITH_/WITHOUT_ knobs
to get arm snapshot builds working again.
Since the cc(1) binary is no longer expected to be clang(1),
remove the chroot(8) post-install cc(1) overwrite.
MFC After: 3 days
X-MFC-With: r264518,r264697,r264698
Tested on: stable/10@r264677 RPI-B
Sponsored by: The FreeBSD Foundation
ian [Sun, 20 Apr 2014 18:21:05 +0000 (18:21 +0000)]
Remove uncessary cache and TLB maintenance ops.
- These were needed on armv4/5 (VIVT cache), not needed on armv6.
- The wbinv_all call can't be used on SMP systems; cache operations by
set/way are not broadcast to other cores.
- The TLB maintenance operations needed for pmap_growkernel() happen in
pmap_grow_l2_bucket(), so there's no need to flush all TLB entries at
the end.
- There may not be any need for the TLB flush at the beginning of
pmap_release(), but it's left in for now pending more investigation.
Pointed out by: Svatopluk Kraus <onwahe@gmail.com>
Discussed with: cognet@
sys/capsicum.h: Remove not very useful visibility guards.
Since <sys/capsicum.h> does not comply to any standards, standards-compliant
applications do not include it and it is not useful to restrict what is
exposed depending on the requested standard.
In addition, the use of types like u_int makes the header fail in strict
standards compliance modes anyway.
Modify the NFSv4 client open/create RPC so that it acquires
post-open/create directory attributes. This allows the RPC to
name cache the newly created file and reduces the lookup RPC
count by about 10% for software builds.
add a TON of missing event handlers with quite possibly bogus
descrptions... If you're one of the many people who forgot to document
their work and don't like my description, well, do your job and update
it...
For 10 years, the "DIALUP <tty>, <user>" message has required having a
hostname (-h) instead of not having a hostname; therefore, it is never
logged. Given that dialup is obsolete and this has not been fixed, remove
the log message.
Note that LOGALL, which is defined by default, logs a message for all
logins, including dialup logins.
Modify the Lookup RPC for NFSv4 so that it acquires directory
attributes. This allows the client to cache directory names
when they are looked up, reducing the Lookup RPC count by
about 40% for software builds.
Allow setting of MK_FOO by anybody. There are too many edge cases that
we're preventing now with this policy. However, these edge cases
should be rare and all that set MK_FOO directly.
WITH_FOO and WITHOUT_FOO both being defined now result in the
non-default behavior happening silently. Users needing determinism
here fall into the edge case exception for MK_FOO setting.
Copy bsd.own.mk to bsd.opts.mk and separate out the option setting
code from the rest. Include bsd.opts.mk in bsd.own.mk to preserve
current behavior. Future revisions will replace the inclusion of
bsd.own.mk elsewhere with bsd.opts.mk or a more appropriate new
file that's still being finalized.
Separate out enabling building clang and/or gcc for the system and
building clang and/or gcc as the bootstrap compiler. Normally, the
default compiler is used. WITH_CLANG_BOOTSTRAP and/or
WITH_GCC_BOOTSTRAP will enable building these compilers as part
bootstrap phase. WITH/WITHOUT_CLANG_IS_CC controls which compiler is
used by default for the bootstrap phase, as well as which compiler is
installed as cc. buildworld now successfully completes building the
cross compiler with WITHOUT_CLANG=t and WITHOUT_GCC=t and produces a
built system with neither of these included.
Similarlly, MK_BINUTILS_BOOTSTRAP controls whether binutils is built
during this phase.
WITHOUT_CROSS_COMPILER will now force MK_BINUTILS_BOOTSTRAP=no,
MK_CLANG_BOOTSTRAP=no and MK_GCC_BOOTSTRAP=no.
BOOTSTRAP_COMPILER was considered, but rejected, since pc98 needs both
clang and gcc to bootstrap still. It should be revisisted in the
future if this requirement goes away. Values should be gcc, clang or
none. It could also be a list.
The odd interaction with Xfoo cross/external tools needs work, but
is beyond the scope of this change as well.
More properly account for free/reserved segments to avoid deadlock or
worse when filling up a device and then trying to erase files to make
space. Without enough space, you can't do that. Also, ensure that the
metadata writes don't generate ENOSPC. They will be retried later
since the buffers are still dirty...
Add -x to the removal of trees to prevent crossing mount points to
limit the damage from errored out runs and failure to unmount loopback
mounts properly.
Add support for reading the PIT Counter 2 output signal via the NMI
Status and Control register at port 0x61.
Be more conservative about "catching up" callouts that were supposed
to fire in the past by skipping an interrupt if it was
scheduled too far in the past.
Restore the PIT ACPI DSDT entries and add an entry for NMISC too.
For NFS mounts using rsize,wsize=65536 over TSO enabled
network interfaces limited to 32 transmit segments, there
are two known issues.
The more serious one is that for an I/O of slightly less than 64K,
the net device driver prepends an ethernet header, resulting in a
TSO segment slightly larger than 64K. Since m_defrag() copies this
into 33 mbuf clusters, the transmit fails with EFBIG.
A tester indicated observing a similar failure using iSCSI.
The second less critical problem is that the network
device driver must copy the mbuf chain via m_defrag()
(m_collapse() is not sufficient), resulting in measurable overhead.
This patch reduces the default size of if_hw_tsomax
slightly, so that the first issue is avoided.
Fixing the second issue will require a way for the
network device driver to inform tcp_output() that it
is limited to 32 transmit segments.
Reported and tested by: csforgeron@gmail.com, markus.gebert@hostpoint.ch
MFC after: 2 weeks
fcntl.h: Make visible various POSIX.1-2008 features.
Also, remove #if __BSD_VISIBLE where it is redundant. When __BSD_VISIBLE is
defined to 1, __POSIX_VISIBLE, __XSI_VISIBLE and __ISO_C_VISIBLE are also
defined to the newest supported version.
Fix releasing the lock in the parent atrun process after the queue
directory has been processed. Otherwise, a long-running child process
caused other atrun invocations to stall unnecessarily.
Based on xlp_machdep.c and completed the list of options based on
boot/mips/beri/loader/metadata.c allow FDT configuration to set
command line options.
This leads to an interesting quesiton of future interactions with loader.
However for configurations without loader this allows bootverbose or boot
single user to be set by compiling a new kernel, which is good enough for
testing and debugging.
when watchdogd is asked to exit nicely (via SIGTERM) it will
stop timer. since watchdogd rc.d script is marked as 'shutdown'
it will exit (on shutdown) and stop timer. if system happens to
hung after watchdogd exited, manual reset is required. when one
operates in "lights-out" type of environments and without
readily available "remote hands" it could create a problem.
this provides ability to override "stop signal" for watchdogd.
default behavior is preserved, i.e. watchdogd will still be killed
via SIGTERM and timer will be stopped. in order to activate new
feature, one needs to put
watchdogd_sig_stop="KILL"
into /etc/rc.conf and also make sure watchdogd timeout is set
to long enough value allowing system to come back online before
timeout fires.
Fix VIRTUAL and PROF interval timers for short intervals, broken at r247903.
Due to the way those timers are implemented, we can't handle very short
intervals. In addition to that mentioned patch caused math overflows
for short intervals. To avoid that round those intervals to 1 tick.
Make it possible for the initiator side to operate in both proxy
and normal mode; this makes it possible to compile with the former
by default, but use it only when neccessary. That's especially
important for the userland part.
When switching variables to flags in r243185 a few cases were missed.
After r263152 this leaves unused variables if route(8) is compiled
without INET support.
Switch the remaining variable accesses to flags and remove now obsolete
variables.
Rework the way we enable CTL iSCSI port. Previously conf_apply()
needed it to be already enabled, because listening in proxy mode
requires it; however, it's conf_apply() that opens pidfiles,
so it resulted in port being enabled before pidfile was opened.
This was not so bad, but it was also disabled when pidfile couldn't
be opened due to ctld already running; this means that starting
second ctld instance screwed up the first.