is the preparation step for moving pmap storage out of vmspace proper.
Reviewed by: Alan Cox <alc@cs.rice.edu>
Matthew Dillion <dillon@apollo.backplane.com>
in the first place.
Use 3sec timeout as recommended.
Reorder some debug messages.
Label som of the 0x%x in debug messages
Make sppp_print_bytes() use %*D and handle zero length.
If we don't have MAGIC numbers, don't yell loopback if 0 == 0
before they got changed. This can help eliminate much of the
gymnastics drivers do in their ioctl routines to figure this out.
Remove commented out IFF_NOTRAILERS
put it, just like on the Alpha. It was wrong to load it at the
fixed address 0x08000000. That should only be done if the dynamic
linker is an executable (not a shared object) with a specific load
address encoded in the object file itself.
This fixes the recent breakage in the Linux emulator.
be in progress at any given moment.
Add two swap tuneables to sysctl:
vm.swap_async_max: 4
vm.swap_cluster_max: 16
Recommended values are a cluster size of 8 or 16 pages. async_max is
about right for 1-4 swap devices. Reduce to 2 if swap is eating too much
bandwidth, or even 1 if swap is both eating too much bandwidth and sitting
on a slow network (10BaseT).
The defaults work well across a broad range of configurations and should
normally be left alone.
Sync rates like 4.032MHz were getting printed as 4.32MHz.
Also, add a quirk entry for the 18G Quantum Atlas III. Like most other
recent Quantum drives, it bogusly reports queue full. Thanks to Andre
Albsmeier <andre.albsmeier@mchp.siemens.de> for the Atlas III inquiry
information.
Reviewed by: gibbs
complaints about ps_refcnt greater than two when we try to fork() a
kthread from proc0 with RFSIGSHARE flag set.
Noticed by: Tor Egge <tegge@fast.no>
Reviewed by: Richard Seaman, Jr. <dick@tar.com>
Unlock vnode before messing with map to avoid deadlock between map and
vnode ( e.g. with exec_map and underlying program binary vnode ). Solves
a deadlock that most often occurs during a large -j# buildworld reported
by three people.
This makes it possible to change the sysctl tree at runtime.
* Change KLD to find and register any sysctl nodes contained in the loaded
file and to unregister them when the file is unloaded.
Reviewed by: Archie Cobbs <archie@whistle.com>,
Peter Wemm <peter@netplex.com.au> (well they looked at it anyway)
Seems to solve a problem with a mouse not responding to movements in the
X direction. Problem description is still rather vague and solution is
not exactly clear. Problem might be a compiler optimisation.
been made but the code has been reorganized and documented to make
it more readable, reduce the size of the code, and optimize the branch
path caching capabilities that most modern processors have.
free swap space out from under a busy page. This is not legal because
the swap may be reallocated and I/O issued while I/O is still in
progress on the same swap page from the madvise()'d object. This bug
could only occur under extreme paging conditions but might not cause
an error until much later. As a side-benefit, madvise() is now even
smaller.
This will trigger inserted()) to be called twice which confuses pccardd.
Add code to not process pcicitr())'s when in the middle of a resume
process. The real insertion of cards and the emulated one in the suspend/resume
code really do not mix up.
driver was thinking irq was enabled although it wasn't.
This case was particular to a no-interrupt static configuration.
Reported by: "Norman C. Rice" <nrice@emu.sourcee.com>
programs, including msdos, set PSL_NT in probes for old cpu types,
although PSL_NT doesn't do anything useful in vm86 or real mode.
PSL_NT is even less useful in the signal handlers. It just causes
T_TSSFLT faults on return from syscalls made by the handlers.
These faults are fixed up lazily so that Xsyscall() doesn't have
to be slowed down to prevent them. The fault handler recently
started complaining about these faults occurring "with interrupts
disabled". It should not have, but the complaints pointed to this
bug.
PR: 9211
the API for freeing up cnp's. This cleanup should not effect nominal
operation one way or the other since NFS VOPs just happen to be called
with flags that match what it actually does to the NAMEI components it
gets. Still, if an NFS error occured, there was probably some memory
leakage of NAMEI components with certain NFS VOP ops.
possible without actually unmapping it from the process.
As of now, I declare madvise() on OBJT_DEFAULT/OBJT_SWAP objects to be
'working and complete'.
Now should be able to report speed for cards using NatSemi PHY.
(if you have one please let me know if it works as I
only have the Intel version)
Reviewed by: David Greenman <dg@root.com>
OBJ_ONEMAPPING in the case where an object is extended by an
additional vm_map_entry must be allocated.
In vm_object_madvise(), remove calll to vm_page_cache() in MADV_FREE
case in order to avoid a page fault on page reuse. However, we still
mark the page as clean and destroy any swap backing store.
Submitted by: Alan Cox <alc@cs.rice.edu>
3c905B, the RX and TX reset commands also reset the cyclone chip's internal
PHY, which causes it to restart its autonegotiation session. This takes a
second or two to complete, which makes the interface seem to stop responding
for a few seconds every time you do something that reinitializes it.
Issuing the RX and TX resets on the older 3c905 boomerang adapters doesn't
cause any delay because the boomerang chip requires an external PHY.
This should fix the problem where people doing network installs via 3c905B
cards experience a delay after the interface is first initialized, among
other things.
Submitted by Roger Hardiman.
Added ioctl TVTUNER_GETCHANSET to discover which regions the bktr driver
supports. Submitted by Vsevolod Lobko <seva@alex-ua.com>
Added BT848_GPIO_SET_EN,BT848_GPIO_SET_DATA (and GETs) to allow user land
control of the GPIO pins. This allows a Radio module on the GPIO port
to be controlled. Submitted by Vsevolod Lobko <seva@alex-ua.com>
The kernel option BKTR_GPIO_ACCESS must be used to enable the GPIO ioctls.
Submitted by: Roger Hardiman and Vsevolod Lobko <seva@alex-ua.com>
when the process starts, and having it nonzero causes statically-linked
Linux binaries to fail.
PR: i386/10015
Submitted by: Marcel Moolenaar <marcel@scc.nl>
the same name. Silently return EEXIST if this happens.
vinum_scandisk: Collect drive numbers, not pointers, to avoid problems
of relocated drives.
Tripped-over-by: Bernd Walter <ticso@cicely.de>
time out on an operation. Under these circumstances, vinum(8) will
automatically start another daemon. Add a pid for the daemon, so that
an overtaken daemon will discover that it's no longer in favour, and
will crawl into a corner and die.
peripheral drivers can determine where in the devstat(9) list they are
inserted.
This requires recompilation of libdevstat, systat, vmstat, rpc.rstatd, and
any ports that depend on the devstat code, since the size of the devstat
structure has changed. The devstat version number has been incremented as
well to reflect the change.
This sorts devices in the devstat list in "more interesting" to "less
interesting" order. So, for instance, da devices are now more important
than floppy drives, and so will appear before floppy drives in the default
output from systat, iostat, vmstat, etc.
The order of devices is, for now, kept in a central table in devicestat.h.
If individual drivers were able to make a meaningful decision on what
priority they should be at attach time, we could consider splitting the
priority information out into the various drivers. For now, though, they
have no way of knowing that, so it's easier to put them in an easy to find
table.
Also, move the checkversion() call in vmstat(8) to a more logical place.
Thanks to Bruce and David O'Brien for suggestions, for reviewing this, and
for putting up with the long time it has taken me to commit it. Bruce did
object somewhat to the central priority table (he would rather the
priorities be distributed in each driver), so his objection is duly noted
here.
Reviewed by: bde, obrien
convince myself that nothing will break if we permit IP input while
interface addresses are unconfigured. (At worst, they will hit some
ULP's PCB scan and fail if nobody is listening.) So, remove the restriction
that addresses must be configured before packets can be input. Assume
that any unicast packet we receive while unconfigured is potentially ours.
because there was a concensus on current in regards to leaving bss r+w+x
instead of r+w. This is in order to maintain reasonable compatibility
with existing JIT compilers (e.g. kaffe) and possibly other programs.
define. Add a new config flag param (ISP_CFG_NONVRAM) whose intent it is
to cause NVRAM to be ignored. Add ISPASYNC_LOOP_DOWN and ISPASYNC_LOOP_UP
isp_async enums.
Amazingly enough, I did all my scsi_sa work recently without realizing
that I had a broken isp card whose (unchangeable- it's an old old old
isp1020) NVRAM has sync mode enabled, but disconnect/reconnect disabled-
the ISP_CFG_NONVRAM is definitely warranted when you want to bloody well
ignore the NVRAM and set something sensible.
Use fast memory timing NVRAM parameter. Clean up and fix establishment
of default target parameters. Don't use NVRAM if are flagged as not to
do so (I had a busted NVRAM setup which I couldn't edit that enabled SYNC
mode but disabled disconnect/reconnect and wide!!). Fix delays after
resets. BUS resets not done in isp_init anymore- relegated to OS
specific outer layers. Fix a buglet where you can get in a loop for
a NULL xs in the completion list in isp_intr. Add in some defines that
can disable fast posting. Add in code for Loop Up/Loop Down events that
call into the outer layers as to what to do.
no major operational changes were made. The three core object->memq loops
were moved into a single inline procedure and various operational
characteristics of the collapse function were documented.
Do not modify m_len before passing mbuf chains to bpf.
Don't forget to pass packets to bpf when running in Crynwr mode (LINK0).
Partially based on a patch by Bill Fenner <fenner@freebsd.org>.
PR: bin/7241
Improved MSP34xx reset for bt848 Hauppauge boards.
Added detection for Bt848a.
Vsevolod Lobko<seva@sevasoft.alex-ua.com> added more XUSSR channels.
Submitted by: parts from Vsevolod Lobko<seva@sevasoft.alex-ua.com>
Obtained from: parts from OpenBSD
Divert was not feeding clean data to ifa_ifwithaddr() so it was
giving bad results.
Submitted by: kseel <kseel@utcorp.com>, Ruslan Ermilov <ru@ucb.crimea.ua>
PQ_FREE. There is little operational difference other then the kernel
being a few kilobytes smaller and the code being more readable.
* vm_page_select_free() has been *greatly* simplified.
* The PQ_ZERO page queue and supporting structures have been removed
* vm_page_zero_idle() revamped (see below)
PG_ZERO setting and clearing has been migrated from vm_page_alloc()
to vm_page_free[_zero]() and will eventually be guarenteed to remain
tracked throughout a page's life ( if it isn't already ).
When a page is freed, PG_ZERO pages are appended to the appropriate
tailq in the PQ_FREE queue while non-PG_ZERO pages are prepended.
When locating a new free page, PG_ZERO selection operates from within
vm_page_list_find() ( get page from end of queue instead of beginning
of queue ) and then only occurs in the nominal critical path case. If
the nominal case misses, both normal and zero-page allocation devolves
into the same _vm_page_list_find() select code without any specific
zero-page optimizations.
Additionally, vm_page_zero_idle() has been revamped. Hysteresis has been
added and zero-page tracking adjusted to conform with the other changes.
Currently hysteresis is set at 1/3 (lo) and 1/2 (hi) the number of free
pages. We may wish to increase both parameters as time permits. The
hysteresis is designed to avoid silly zeroing in borderline allocation/free
situations.
to an architecture-specific value defined in <machine/elf.h>. This
solves problems on large-memory systems that have a high value for
MAXDSIZ.
The load address is controlled by a new macro ELF_RTLD_ADDR(vmspace).
On the i386 it is hard-wired to 0x08000000, which is the standard
SVR4 location for the dynamic linker.
On the Alpha, the dynamic linker is loaded MAXDSIZ bytes beyond
the start of the program's data segment. This is the same place
a userland mmap(0, ...) call would put it, so it ends up just below
all the shared libraries. The rationale behind the calculation is
that it allows room for the data segment to grow to its maximum
possible size.
These changes have been tested on the i386 for several months
without problems. They have been tested on the Alpha as well,
though not for nearly as long. I would like to merge the changes
into 3.1 within a week if no problems have surfaced as a result of
them.
attempt to optimize forks but were essentially given-up on due to
problems and replaced with an explicit dup of the vm_map_entry structure.
Prior to the removal, they were entirely unused.
(SVR4 does have it so that;s probably the cause of this bug)
2) Add a wrapper function for translation between ibcs2_ipc_perm and
ipc_perm as I think we screwed up when defining the ipc_perm struct and
mixed up 'normal' and creator [ug]id's
3) Fix IBCS2_IPC_STAT semctl. The FreeBSD version needs a union semun
whereas the IBCS2 version gives a struct ibcs2_semid_ds.
Apparently this is all fixed in the SVR4 compatibility code.
PR: 7729
and the variable doesn't even exist (though it is referenced elsewhere).
Just make sure it produces error messages when Mike get back to it.
PR: kern/9934
Submitted by: Adrian Filipi-Martin
alpha to John Polstra yesterday and it was checked in. Then there
was a bunch of CVS activity, and it ended up with the *broken*
prototype being reinserted. All I see in the history is 'style nits'
as a comment, and I certainly agree with the latter of those two words.
with -aout. Added translation back to elf names in asnames.h as
usual. The elf names were inconsistent in the aout case even
internally because a macro adds an underscore to just one of them.
Removed commented out code for a previous life of `svr4_esigcode'.
Didn't add an underscore to `svr4_esigcode' since it is correct for
aout although wrong for elf, like most internal names in assembler
files. These names should be in a different namespace so that gprof
can ignore them.
Fixed some disorder in asnames.h.
to fix the problem w/ NFSV3 whereby a make installworld would get into
high-network-bandwidth situations continuously trying to retry nfs writes
that fail with a 'stale file handle' error.
world breakage (mainly for cross-world cases). The world Makefile
attempts to build tools static so that nonexistent or wrong shared
libraries and interpreters don't get used. This is broken anyway
since the world Makefile doesn't know about svr4_genassym.
Force building svr4_genassym static. This is part of "fixing"
aout-to-elf-build breakage. aout-to-elf-build abuses NOTOOLS to
avoid rebuilding all the aout tools. This saves time and avoids
some complications. However, it breaks all the internal tools --
they get linked to target libraries which might not work. Cases
where the host can run the target's static libraries are "fixed"
by encrufting all Makefiles that build internal tools to build the
tools static.
Don't add .depend to CLEANFILES -- it just breaks the separation of
`make cleandepend' from `make clean'.
Removed some superflous explicit dependencies.
This was missed in the 4.4-Lite2 merge.
Noticed by: Mohan Parthasarathy <Mohan.Parthasarathy@eng.Sun.COM> and
jayanth@loc201.tandem.com (vijayaraghavan_jayanth)
on the tcp-impl mailing list.
the screen width.
- Store the current video mode information in the `video_adapter' struct.
- The size of the `v_offscreensize' field in the VESA mode information
block is u_int16, not u_int8.
rather then VM_PROT_ALL. obreak, on the otherhand, uses VM_PROT_ALL.
This prevents vm_map_insert() from being able to coalesce the heap and
creates an extra map entry. Since current architectures ignore
VM_PROT_EXECUTE anyway, and since not having VM_PROT_EXECUTE on data/bss
may provide protection in the future, obreak now uses read+write rather
then all (r+w+x).
This is an optimization, not a bug fix.
Submitted by: Alan Cox <alc@cs.rice.edu>
of the minor). Establish and use a control mode open. Control
mode opens may open the device without locking, but are prohibited
from all but some ioctls. MTIOCGET always works. MTIOCERRSTAT
works, but the clearing of latched error status is contingent
upon whether another application has the device open, in which
case an interruptible perip acquire is done. MTSETBSIZ, MTSETDNSTY
and MTCOMP also require a periph aquire.
Relative fileno and blkno are tracked. Note that just about any
error will make these undefined, and if you space to EOD or use
hardware block positioning, these are also lost until the next
UNLOAD or REWIND.
Driver state is also tracked and recorded in the unit softc
to be passed back in mt_dsreg for a MTIOCGET call.
Thanks to Dan Strick for suggesting this.
Reintroduce 2 filemarks at EOD for all but QIC devices. I
really think it's wrong, but there is a lot of 3rd party
software that depends upon this (not the least of which is
tcopy). Introduce a SA_QUIRK_1FM to ensure that some devices
can be marked as only being able to do 1 FM at EOD.
At samount time force a load to BOT if we aren't mounted. If the
LOAD command fails, use the REWIND command (e.g., for the IBM 3590
which for some gawdawful reason doesn't support the LOAD (to BOT)
command).
Also at samount time, if you don't know fixed or variable, try to
*set* to one of the known fixed (or variable, for special case)
density codes. We only have to do this once per boot, so it's not
that painful. This is another way to try and figure out the wierd
QIC devices without having to quirk everything in the universe.
A substantial amount of cleanup as to what operations can and what
operations cannot be retried. Don't retry space operations if they
fail- it'll just lead to lossage.
Not yet done is invalidating mounts correctly after errors. ENOTIME.
to run Solaris executables (or executables from any other ELF system)
directly off the CD-ROM without having to waste megabytes of disk
by copying them to another filesystem just to brand them.
give the same behaviour produced before today. If sysadmin sets it
to a valid ELF brand, ELF image activator will attempt to run unbranded
ELF exectutables as if they were branded with that value.
Suggested by: Dima Ruban <dima@best.net>
process to sneak in and write to or close the pipe. The read code
enters a 'piperd' state after doing the lock operation without
checking to see if the state changed, which can cause the process
to wait forever.
The code has also been documented more.
completes, change if() to KASSERT(). This is not a bug, we are
simplify clarifying and optimizing the code.
In if/else in vfs_object_create(), the failure of both conditionals
will lead to a NULL object. Exit gracefully if this case occurs.
( this case does not normally occur, but needed to be handled ).
Obtained from: Eivind Eklund <eivind@FreeBSD.org>
Since paging is in progress, page scan in vm_page_qcollapse() must be
protected at atleast splbio() to prevent pages from being ripped out from
under the scan.
Add freebsd.fr, for FreeBSD specific forth source. Add $ and %, to
replace the lost @ and - functionality of include. $ has the opposite
behavior of @ though, since the default behavior was inverted.
Change include() so it will be able to load files with forth code,
instead of just builtins. Remove #@- from the include section of the
help file, since they don't work in the new version of include, unless
BOOT_FORTH is not defined.
Change bf_run() so it will return the result returned by ficlExec(). Also,
make bf_run() push "interpret" to be executed by ficlExec(), since ficlExec()
doesn't do it by itself. (Things worked previously because nothing
recursed through ficlExec() by the way of bf_run()).
Change/extend comments on builtin behavior.
Search for "interpret" at the end of bf_init(), so /boot/boot.4th can
provide it's own version.
Remove dead code.
error cases, but the replacement should be doing everything they
did, except what did shouldn't be doing, and might do a little more
they ought to be doing.
structures" but since tty structs aren't malloced it is actually
mainly for tty-level (clist) buffers. It was slightly misused
here for com structs, and the previous commit completely misused
it for device buffers.
Fixed some bugs in nearby pccard code:
- memory leak when pccards go away (broken in previous commit).
- bogus bzeroing of the com struct before freeing it.
- style bugs.
the input speed, so that it can work at speeds larger than 115200
bps without being flow controlled. The buffer is twice as large
as before at 115200 bps and half as large as before at low speeds
Use a single interrupt-level buffer instead of ping-pong buffers
because the simplifications provided by ping-pong buffers became
complications.
This change is over-engineered. Statically configured buffering
was simpler and faster, and increasing the buffer size to support
1.5Mbps would cost about 1 US cent's worth of RAM per port, but I
was interested in the buffer switching mechanism.
not to vmThrow errors. This is not what the comments say it does, and
it doesn't work when there is no ficlExec environment (like it's only
use in sys/boot/common/interp_forth.c).
PR: bin/9772
allocated was not big enough, but it ended up to being used where it
was supposed to be used. The person who did that ought to be shot, but
since I'm a good person, I'll forgive myself...
PR: bin/9743
panic during boot on machines with >=2GB of RAM. Also changed some
incorrect printf conversion specifiers from %d to %u (signed to unsigned).
This fixes bugs when printing the amount of memory on machines with >=2GB
of RAM.
The vm_map_insert()/vm_object_coalesce() optimization has been extended
to include OBJT_SWAP objects as well as OBJT_DEFAULT objects. This is
possible because it costs nothing to extend an OBJT_SWAP object with
the new swapper. We can't do this with the old swapper. The old swapper
used a linear array that would have had to have been reallocated, costing
time as well as a potential low-memory deadlock.
/dev/urandom takes about 38 seconds on a P5/133. It is useful
to be able to kill such reads almost immediately. Processes
doing such reads are now scheduled so their denial of service
is no worse than that of processes looping in user mode.
code when examining their fix, which caused my code (in rev 1.52) to:
- panic("soaccept: !NOFDREF")
- fatal trap 12, with tracebacks going thru soclose and soaccept
Drivers should be updated if they get flagged by this message.
(The reason this is important is because we do not have a way
to catch this mistake for interfaces added after ifinit() runs.)
