Commit Graph

42499 Commits

Author SHA1 Message Date
Sam Leffler
12394d06d8 place some kernel-specific data structures under #ifdef _KERNEL
Sponsored by:	FreeBSD Foundation
2003-10-03 20:58:56 +00:00
Alan Cox
1b2d9f0653 Make PAGE_SIZE and related quantities signed on sparc64. (They are signed
quantities on every other architecture.)  This change is required in order
to move pmap_prefault() out of the pmap and into the machine-independent
layer.
2003-10-03 19:49:08 +00:00
Robert Watson
5fd04e380f When direct dispatching an netisr (net.isr.enable=1), if there are already
any queued packets for the isr, process those packets before the newly
submitted packet, maintaining ordering of all packets being delivered
to the netisr.  Remove the bypass counter since we don't bypass anymore.
Leave the comment about possible problems and options since later
performance optimization may change the strategy for addressing ordering
problems here.

Specifically, this maintains the strong isr ordering guarantee; additional
parallelism and lower latency may be possible by moving to weaker
guarantees (per-interface, for example).  We will probably at some point
also want to remove the one instance netisr dispatch limit currently
enforced by a mutex, but it's not clear that's 100% safe yet, even in
the netperf branch.

Reviewed by:	sam, others
2003-10-03 18:27:24 +00:00
Sam Leffler
aea8b30f12 trivial locking rtsock_cb
Sponsored by:	FreeBSD Foundation
2003-10-03 18:27:02 +00:00
Sam Leffler
becc44d76c cleanups prior to adding locking (and in some cases to eliminate locking):
o move route_cb to be private to rtsock.c
o replace global static route_proto by locals
o eliminate global #define shorthands for info references
o remove some register decls
o ansi-fy function decls
o move items to be close in scope to their usage
o add rt_dispatch function for dispatching the actual message
o cleanup tangled logic for doing all-but-me msg send

Support by:	FreeBSD Foundation
2003-10-03 18:15:54 +00:00
Bruce M Simpson
c3b52d6499 Shorten 'bad gateway' AF_LINK message.
Submitted by:	green
2003-10-03 17:22:14 +00:00
Bruce M Simpson
beb2ced8ac Make arp_rtrequest()'s 'bad gateway' messages slightly more informative,
to aid me in tracking down LLINFO inconsistencies in the routing table.

Discussed with:	fenner
2003-10-03 17:21:17 +00:00
Peter Wemm
cd349af148 Add #include "opt_pmap.h" so locore picks up DISABLE_PSE etc options. 2003-10-03 14:33:00 +00:00
Bruce M Simpson
b75bead1f2 Only delete the route if arplookup() tried to create it. Do not delete
RTF_STATIC routes. Do not check for RTF_HOST so as to avoid being DoSed
when an RTF_GENMASK route exists in the table.

Add a more verbose comment about exactly what this code does.

Submitted by:	ru
2003-10-03 09:19:23 +00:00
Marcel Moolenaar
5bf2d2b6b4 Swap the syscall caller frame info (i.e. the return pointer and
frame marker) and the syscall stub frame info in the trap frame.
Previously we stored the stub frame info in (rp,pfs) and the
caller frame info in (iip,cfm). This ends up being suboptimal
for the following reasons:
1. When we create a new context, such as for an execve(2), we had
   to set the (rp,pfs) pair for the entry point when using the
   syscall path out of the kernel but we need to set the (iip,cfm)
   pair when we take the interrupt way out. This is mostly just
   an inconsistency from the kernel's point of view, but an ugly
   irregularity from gdb(1)'s point of view.
2. The getcontext(2) and setcontext(2) syscalls had to swap the
   (rp,pfs) and (iip,cfm) pairs to make the context compatible
   with one created purely in userland.

Swapping the (rp,pfs) and (iip,cfm) pairs is visible to signal
handlers that actually peek at the mcontext_t and to gdb(1).
Since this change is made for gdb(1) and we don't care about
signal handlers that peek at the mcontext_t because we're still
a tier 2 platform, this ABI breakage is academic at this moment
in time.

Note that there was no real reason to save the caller frame info
in (iip,cfm) and the stub frame info in (rp,pfs).
2003-10-03 03:50:29 +00:00
Ruslan Ermilov
bf17fc53ac Revert rev. 1.86, I've fixed make(1) (make/dir.c,v 1.32).
PR:		bin/34062
2003-10-02 21:34:06 +00:00
Eric Anholt
7efb14ccd3 Make the i810 AGP device create a "drmsub" child device. This will be attached
to by the DRM for i8xx devices.

Submitted by:	Keith Whitwell <keith@tungstengraphics.com>
2003-10-02 19:49:15 +00:00
Jacques Vidrine
8b7358ca43 Introduce a uiomove_frombuf helper routine that handles computing and
validating the offset within a given memory buffer before handing the
real work off to uiomove(9).

Use uiomove_frombuf in procfs to correct several issues with
integer arithmetic that could result in underflows/overflows.  As a
side-effect, the code is significantly simplified.

Add additional sanity checks when computing a memory allocation size
in pfs_read.

Submitted by:	rwatson  (original uiomove_frombuf -- bugs are mine :-)
Reported by:	Joost Pol <joost@pine.nl>  (integer underflows/overflows)
2003-10-02 15:00:55 +00:00
Hidetoshi Shimokawa
c44a734c56 Remove wrong usage of BUF_REFCNT().
Spotted by: phk
2003-10-02 14:33:30 +00:00
Hidetoshi Shimokawa
9eb73373e7 Be kind to 64bit architectures. 2003-10-02 08:56:14 +00:00
Nate Lawson
8525ffefc2 If requested to Sleep for less than our hz granularity (e.g., 10 ms),
use DELAY instead of tsleep.

Submitted by:	peter
2003-10-02 05:09:37 +00:00
Alan Cox
81b460c5eb Reimplement pagezero() using "movnti". 2003-10-02 05:08:13 +00:00
Hidetoshi Shimokawa
c4778b5d1c MFp4: Change struct fw_xfer to reduce memory copy in fwmem_*() functions.
And many changes.

* all
	- Major change of struct fw_xfer.
		o {send,recv}.buf is splitted into hdr and payload.
		o Remove unnecessary fields.
		o spd is moved under send and recv.
	- Remove unnecessary 'volatile' keyword.
	- Add definition of rtcode and extcode.

* firewire.c
	- Ignore FWDEVINVAL devices in fw_noderesolve_nodeid().
	- Check the existance of the bind before call STAILQ_REMOVE().
	- Fix bug in the fw_bindadd().
	- Change element of struct fw_bind for simplicity.
	- Check rtcode of response packet.
	- Reduce split transaction timeout to 200 msec.
		(100msec is the default value in the spec.)
	- Set watchdog timer cycle to 10 Hz.
	- Set xfer->tv just before calling fw_get_tlabel().

* fwohci.c
	- Simplifies fwohci_get_plen().

* sbp.c
	- Fix byte order of multibyte scsi_status informations.
	- Split sbp.c and sbp.h.
	- Unit number is not necessary for FIFO¤ address.
	- Reduce LOGIN_DELAY and SCAN_DELAY to 1 sec.
	- Add some constants defineded in SBP-2 spec.

* fwmem.c
	- Introduce fwmem_strategy() and reduce memory copy.
2003-10-02 04:06:56 +00:00
Robert Watson
c142b0fcfe Remove the global variable 'cmask', which was used to initialize the
fd_cmask field in the file descriptor structure for the first process
indirectly from CMASK, and when an fd structure is initialized before
being filled in, and instead just use CMASK.  This appears to be an
artifact left over from the initial integration of quotas into BSD.

Suggested by:	peter
2003-10-02 03:57:59 +00:00
Robert Watson
6b66d5bb4b Use __BEGIN_DECLS and __END_DECLS around userland function prototypes
so that mac.h may be more safely included in userland C++ applications.

PR:		bin/56595
Submitted by:	"KONDOU, Kazuhiro" <kazuhiro@alib.jp>
2003-10-02 03:07:51 +00:00
Robert Watson
e590eca2ad Create a tunable for net.isr.enable so that it may be set from
inception, rather than having to wait for the boot to finish.
2003-10-02 02:54:10 +00:00
Peter Wemm
6ccf265bb0 Commit Bosko's patch to clean up the PSE/PG_G initialization to and
avoid problems with some Pentium 4 cpus and some older PPro/Pentium2
cpus.  There are several problems, some documented in Intel errata.
This patch:
1) moves the kernel to the second page in the PSE case.  There is an
errata that says that you Must Not point a 4MB page at physical
address zero on older cpus.  We avoided bugs here due to sheer luck.
2) sets up PSE page tables right from the start in locore, rather than
trying to switch from 4K to 4M (or 2M) pages part way through the boot
sequence at the same time that we're messing with PG_G.

For some reason, the pmap work over the last 18 months seems to tickle
the problems, and the PAE infrastructure changes disturb the cpu
bugs even more.

A couple of people have reported a problem with APM bios calls during
boot.  I'll work with people to get this resolved.

Obtained from:	bmilekic
2003-10-01 23:46:08 +00:00
Robert Watson
3164565d39 Temporarily turn net.isr.enable back off again until patches to
correct potential nits in packet ordering are resolved.
2003-10-01 22:15:16 +00:00
Hajimu UMEMOTO
5d40536819 add randomtab for ip6_randomflowlabel().
Obtained from:	KAME
2003-10-01 21:45:57 +00:00
Robert Watson
19288f738a Enable net.isr.enable by default, causing "delivery to completion"
(direct dispatch) in interrupt threads when the netisr in question
isn't already active.  If a netisr is already active, or direct
dispatch is already in progress, we queue the packet for later
delivery.  Previously, this option was disabled by default.  I have
measured 20%+ performance improvements in IP packet forwarding with
this enabled.

Please report any problems ASAP, especially relating to stack depth or
out-of-order packet processing.

Discussed with:	jlemon, peter
Sponsored by:	DARPA, Network Associates Laboratories
2003-10-01 21:31:09 +00:00
Hajimu UMEMOTO
b79274ba41 randomize IPv6 flowlabel when RANDOM_IP_ID is defined.
Obtained from:	KAME
2003-10-01 21:24:28 +00:00
Hajimu UMEMOTO
18193b6f63 use arc4random() 2003-10-01 21:10:02 +00:00
Hajimu UMEMOTO
de27a78aca - include opt_random_ip_id.h
- we don't need to obtain microtime when using ip6_randomid.
2003-10-01 20:24:20 +00:00
Hajimu UMEMOTO
8513854d16 we don't need ip6_id when RANDOM_IP_ID is defined. 2003-10-01 18:23:27 +00:00
Hajimu UMEMOTO
01e22dc51b include opt_random_ip_id.h 2003-10-01 17:28:42 +00:00
Hajimu UMEMOTO
672467eb28 Don't compiled ip6_randomid() in if RANDOM_IP_ID is not defined. 2003-10-01 16:22:58 +00:00
Hajimu UMEMOTO
2923494300 Obey RANDOM_IP_ID.
Requested by:	sam
2003-10-01 16:00:12 +00:00
Hajimu UMEMOTO
8373d51d4b randomize IPv6 fragment ID.
Obtained from:	KAME
2003-10-01 15:13:29 +00:00
Bernd Walter
1313b38492 Fix indention
Noticed by:	johan
2003-10-01 14:49:53 +00:00
Bernd Walter
a244477908 Don't call makedev().
Cache result of make_dev() and use it when calling destroy_dev().
2003-10-01 13:53:51 +00:00
Maxime Henrion
a214dd13e8 Remove extra parentheses from the __GNUC_PREREQ__ macro.
Prodded by:	bde
2003-10-01 11:26:20 +00:00
Søren Schmidt
6d92de35c9 Fix the problem with accessing certain FLASH devices. The problem
was that accessing the status reg could occour too fast, confusing
the logic in the flash part. Could not have been located without:

HW donated by: Jonas Bülow <jonas@servicefactory.se>
2003-10-01 09:58:19 +00:00
Jeff Roberson
460c513686 - Add a memory barrier before the sse2_pagezero() function returns. This
code uses write combining which must be committed to memory prior to
   other uses of this page.

Spotted by:	alc
2003-10-01 08:52:46 +00:00
Ruslan Ermilov
deb62e2887 By popular demand, added the "static ARP" per-interface option. 2003-10-01 08:32:37 +00:00
Hidetoshi Shimokawa
c727011a76 Fix a bug in fwdev_clone().
Spotted by: grog
2003-10-01 06:04:30 +00:00
Jeff Roberson
1419773de3 - Hide more #ifdef logic in a new invlcaddr inline. This function flushes
the full tlb if you're on an I386or does an invlpg otherwise.

Glanced at by:	peter
2003-10-01 05:56:46 +00:00
Jeff Roberson
043407f803 - Define an inline pagezero() to select the appropriate full-page zeroing
function from one of bzero, i686_pagezero, or sse2_pagezero.
 - Use pagezero() in the three pmap functions that need to zero full pages.
2003-10-01 05:42:58 +00:00
Jeff Roberson
977f9ab515 - Add ss2_pagezero() for zeroing pages using the movnti instruction. This
instruction is enabled with SSE2 but does not use SSE registers.  It is a
   "non-temporal" move which bypasses the cache and does not dirty lines.
2003-10-01 05:25:27 +00:00
Peter Wemm
a93020d7a1 Use __register_t instead of register_t, otherwise <sys/types.h> is a
prerequisite for <ucontext.h> on amd64.  Oops.
2003-10-01 01:08:04 +00:00
Jeff Roberson
fb9bde2d2b - Correct a problem with the last commit. The CMAP ptes need to be zeroed
prior to invalidating the TLB to be certain that the processor doesn't
   keep a cached copy.

Discussed with:	pete
Paniced:	tegge
Pointy Hat:	The usual spot
2003-10-01 00:27:09 +00:00
Maxime Henrion
4e513b7a01 Allow the compiler to micro-optimize byte swapping functions by
evaluating them at compile time rather than at run time.  As for x86
and amd64, this requires GCC and it's enabled only if __OPTIMIZE__ is
defined (ie, if at least -O is used).

Reviewed by:	jake
2003-09-30 22:35:27 +00:00
Maxime Henrion
d68d22ba52 Use __predict_false() in the KASSERT() macro. We expect this test
to fail most of the time.
2003-09-30 20:54:12 +00:00
Maxime Henrion
6e478cc056 Introduce new __predict_false(exp) and __predict_true(exp) keywords.
Their purpose is to give explicit hints to the compiler to judge
the likelyhood of a test to succeed or fail.  Not all architectures
have support for such optimizations, but for those who do, it can
give a nice performance improvement in hot loops.

Obviously, this should be used very rarely in very specific code.

Reviewed by:	peter
Obtained from:	OpenBSD
2003-09-30 20:45:14 +00:00
Maxime Henrion
91dc1097d1 Define a new __GNUC_PREREQ__(maj, min) to test if we have a specific
version of GCC or later and use it where appropriate.  It makes this
file more readable.

Obtained from:	OpenBSD
2003-09-30 20:34:57 +00:00
Nate Lawson
ae0781a1cb Revert 1.24. It causes a hang on shutdown for Marcel. This is due to
AcpiEnterSleepState() calling a long AcpiOsStall() with interrupts
disabled.  This fix will instead be added to ACPI-CA.

PR:
Submitted by:
Reviewed by:
Approved by:
Obtained from:
MFC after:
2003-09-30 19:12:37 +00:00