Commit Graph

97359 Commits

Author SHA1 Message Date
Greg Lehey
80d23b9669 free_plex, free_volume: Don't try to destroy the underlying device if
it doesn't exist.  This can occur under
                        certain failure situations.
2003-12-15 00:44:05 +00:00
Greg Lehey
684873a192 vinumclose: Fix day-one bug. Subdisks and plexes were not being
marked closed.  This made it impossible to stop Vinum
	    after explicitly opening a plex or subdisk.
2003-12-15 00:42:03 +00:00
Bill Paul
a417109f1f Silence irritating watchdog timeout messages: if we call
ndis_send_packets() but there's no link yet, we get an immediate
callback to ndis_txeof(), which clears if_timer. But ndis_start()
sets if_timer right after the call to ndis_send_packets(). Set
if_timer before calling ndis_send_packets().

Also fix mutex locking to prevent ndis_txeof() from running in
the middle of ndis_start().
2003-12-14 22:47:01 +00:00
Bill Paul
232d6b73a0 Whups... remove some debug code that accidentally crept in. 2003-12-14 21:33:07 +00:00
Bill Paul
53947bb732 Rework mbuf<->ndis_packet/ndis_packet<->mbuf translation a little to
make it more robust. This should fix problems with crashes under
heavy traffic loads that have been reported. Also add a 'query done'
callback handler to satisfy the e100bex.sys sample Intel driver.
2003-12-14 21:31:32 +00:00
Joseph Koshy
08d4b26dd8 Document the effect of sysctl tunables controlling p_candebug(9) on
ktrace(2).
2003-12-14 14:54:53 +00:00
Sean Chittenden
12335de1f5 Backout ataraid rcng script. I must have missed ar0 in my scan of /dev
after my first reboot because sure enough, I'm seeing it there now and
ata(4) is doing the right thing(TM).

Pointed out by:	des
2003-12-14 13:36:36 +00:00
Sean Chittenden
7e6b2a1467 Add a script that allows software RAID sets to be created before file
systems are mounted.  An example set of entries for /etc/rc.conf:

ataraid_enable="YES"
ataraid_devices="ar0"
ataraid_ar0_set="ad2 ad3"
ataraid_ar0_type="RAID1"

Because there is no "correct" way of doing ATA raid (ie, geom vs.
atacontrol vs. vinum) that is bikeshed proof, this rcng script stays within
the bounds of atacontrol and assumes that other RAID solutions for GEOM or
vinum will end up in a different rcNG script.

Reviewed by:	green
2003-12-14 09:42:02 +00:00
Jeff Roberson
86e1c22aa4 - Assign the ke_cpu field in kseq_notify() so that all of our callers do not
have to do it.
 - Set the ke_runq to NULL in sched_add() before calling kseq_notify().
   Otherwise we may panic in sched_add() if INVARIANTS is on.
2003-12-14 02:06:29 +00:00
Marcel Moolenaar
a8434283fc In set_mcontext(), take into account that kse_switchin(2) will
eventually be passed an async. context as well as a syscall
context.
While here, fix a serious bug in that if the trapframe is a
syscall frame, but we're restoring an async context, we need
to clear the FRAME_SYSCALL flag so that we leave the kernel
via exception_restore.
2003-12-14 01:59:31 +00:00
Bill Paul
ba102043bc Implement a few new NDIS routines: NdisInitAnsiString(),
NdisAnsiStringToUnicodeString(), NdisWriteConfiguration().

Also add stubs for NdisMGetDeviceProperty(), NdisTerminateWrapper(),
NdisOpenConfigurationKeyByName(), NdisOpenConfigurationKeyByIndex()
and NdisMGetDeviceProperty().
2003-12-14 00:43:40 +00:00
Robert Watson
ce41f4bf59 Comment on initgroups.c:1.8: it can prevent users from logging in if
a user has over 16 groups defined.

Reported by:	"Klaus-J. Wolf" <yanestra@web.de>
2003-12-14 00:42:00 +00:00
Poul-Henning Kamp
0a93720637 Fix a locking problem with MD_ROOT_SIZE.
Retire md(4)'s static major number.
2003-12-13 18:12:58 +00:00
Tom Rhodes
a27f75251a Provide a way to deal with rc.conf which may already be populated in certain
cases.
2003-12-13 16:12:55 +00:00
Hidetoshi Shimokawa
c54d1fe2bd Fix panic.
- Don't call sbp_targ_status_FIFO() twice for LOGIN error.
- Don't access login if it's NULL.
2003-12-13 15:33:45 +00:00
Dag-Erling Smørgrav
4358431405 Install a SIGCHLD handler so select(2) will be interrupted when a child
terminates.  Without this patch, 'make -j1 buildworld' takes about 30%
longer than 'make -B buildworld' on my 2.4 GHz P4; the difference is
probably even larger on faster systems.  With this patch, there is no
perceptible difference in wall time between the two.

Submitted by:	bde
MFC after:	3 days
2003-12-13 15:26:27 +00:00
Bill Paul
fc2ada1918 Correct the implementation of NDIS_BUFFER_TO_SPAN_PAGES(). 2003-12-13 09:07:35 +00:00
Bill Paul
153a70182b Support yet another variant of WinXP .INF files. 2003-12-13 08:48:35 +00:00
Paul Saab
591d10119e White space cleanup 2003-12-13 07:54:07 +00:00
Bill Paul
d3eb09f083 subr_ndis.c:
- fix ndis_time() so that it returns a time based on the proper
  epoch (wacky though it may be)
- implement NdisInitializeString() and NdisFreeString(), and add
  stub for NdisMRemoveMiniport()

ntoskrnl_var.h:
- add missing member to the general_lookaside struct (gl_listentry)

subr_ntoskrnl.c:
- Fix arguments to the interlocked push/pop routines: 'head' is an
  slist_header *, not an slist_entry *
- Kludge up _fastcall support for the push/pop routines. The _fastcall
  convention is similar to _stdcall, except the first two available
  DWORD-sized arguments are passed in %ecx and %edx, respectively.
  One kludge for this __attribute__ ((regparm(3))), however this
  isn't entirely right, as it assumes %eax, %ecx and %edx will be
  used (regparm(2) assumes %eax and %edx). Another kludge is to
  declare the two fastcall-ed args as local register variables and
  explicitly assign them to %ecx and %edx, but experimentation showed
  that gcc would not guard %ecx and %edx against being clobbered.
  Thus, I came up with a 3rd kludge, which is to use some inline
  assembly of the form:

	void		*arg1;
	void		*arg2;

	__asm__("movl %%ecx, %%ecx" : "=c" (arg1));
	__asm__("movl %%edx, %%edx" : "=d" (arg2));

  This lets gcc know that we're going to reference %ecx and %edx and
  that it should make an effort not to let it get trampled. This wastes
  an instruction (movl %reg, %reg is a no-op) but insures proper
  behavior. It's possible there's a better way to do this though:
  this is the first time I've used inline assembler in this fashion.

The above fixes to ntoskrnl_var.h an subr_ntoskrnl.c make lookaside
lists work for the two drivers I have that use them, one of which
is an NDIS 5.0 miniport and another which is 5.1.
2003-12-13 07:41:12 +00:00
Bill Paul
7359dfc0b1 Implement some more NDIS and ntoskrnl API calls:
subr_ndis.c: NdisGetCurrentSystemTime() which, according to the
Microsoft documentation returns "the number of 100 nanosecond
intervals since January 1, 1601." I have no idea what's so special
about that epoch or why they chose 100 nanosecond ticks. I don't
know the proper offset to convert nanotime() from the UNIX epoch
to January 1, 1601, so for now I'm just doing the unit convertion
to 100s of nanoseconds.

subr_ntoskrnl.c: memcpy(), memset(), ExInterlockedPopEntrySList(),
ExInterlockedPushEntrySList().

The latter two are different from InterlockedPopEntrySList()
and InterlockedPushEntrySList() in that they accept a spinlock to
hold while executing, whereas the non-Ex routines use a lock
internal to ntoskrnl. I also modified ExInitializePagedLookasideList()
and ExInitializeNPagedLookasideList() to initialize mutex locks
within the lookaside structures. It seems that in NDIS 5.0,
the lookaside allocate/free routines ExInterlockedPopEntrySList()
and ExInterlockedPushEntrySList(), which require the use of the
per-lookaside spinlock, whereas in NDIS 5.1, the per-lookaside
spinlock is deprecated. We need to support both cases.

Note that I appear to be doing something wrong with
ExInterlockedPopEntrySList() and ExInterlockedPushEntrySList():
they don't appear to obtain proper pointers to their arguments,
so I'm probably doing something wrong in terms of their calling
convention (they're declared to be FASTCALL in Widnows, and I'm
not sure what that means for gcc). It happens that in my stub
lookaside implementation, they don't need to do any work anyway,
so for now I've hacked them to always return NULL, which avoids
corrupting the stack. I need to do this right though.
2003-12-12 22:35:13 +00:00
Bruce A. Mah
946bee360e New release notes: boot0sio, ndis(4), resolver(5) supports timeout:
and attempts:, ipfw(8) -b, ACPI-CA 20031203.
2003-12-12 21:58:13 +00:00
Robert Watson
09a4a69c1d Although sometimes to the uninitiated, it may seem like goup, KSEGOUP
is actually spelt KSEGROUP.  Go figure.

Reported by:	samy@kerneled.com
2003-12-12 21:25:56 +00:00
Jesper Skriver
8ea562a029 Fix typo, I forgot daily_ in front of the status_ata_raid_enable 2003-12-12 21:23:27 +00:00
Robert Watson
43ad78e751 Spell btkr as bktr.
Pointy hat to:	obrien
Reported by:	Martin <nakal@web.de>
2003-12-12 21:18:04 +00:00
Nate Lawson
b279c35a4f Fix throttling to use the proper mask. The bug resulted in only two
throttling values being available regardless of the CPU's capabilities.
This has been broken since rev 1.1.  Also clarify a comment.

Submitted by:	Taku YAMAMATO <taku@cent.saitama-u.ac.jp>
2003-12-12 19:42:16 +00:00
Luigi Rizzo
ac6cec512b Add a -b flag to /sbin/ipfw to print only action and comment for each
rule, thus omitting the entire body.
This makes the output a lot more readable for complex rulesets
(provided, of course, you have annotated your ruleset appropriately!)

MFC after: 3 days
2003-12-12 16:14:28 +00:00
Poul-Henning Kamp
8a1a55ec29 DUH!
Write 100 times for tomorrow:
	I will never again free(9) a modified pointer.

Pointy Hat:	yeah, yeah, yeah, can you just put it in the pile over there...
2003-12-12 12:17:28 +00:00
Dag-Erling Smørgrav
3b64ed3a59 Revert the last two commits. It seems that compat mode is actually faster
than non-compat (using -j1), though reading the make(1) sources suggests
that the opposite is true.
2003-12-12 12:03:59 +00:00
Peter Pentchev
a61f15043c Fix a couple of typos: hexidecimal -> hexadecimal.
Reviewied by:	scottl
2003-12-12 11:08:51 +00:00
Poul-Henning Kamp
48529e81e4 Attempt to get the short cable fix to work better on the if_sis:
Only do short-cable on revisions that need it.

Move generic initialization before short-cable fix, in order to not
clobber short cable fix register setting.
2003-12-12 10:15:39 +00:00
Peter Grehan
2abd35bc59 - accept device_type of "block", which is how psim/gdb6.0 defines
disks. continue to accept "disk" for psim/gdb5.x users.
- remove unneeded ofwd_identify
2003-12-12 09:54:39 +00:00
Mike Makonnen
099fe19901 Doh! Lock the thread passed in by the caller, not the current thread. 2003-12-12 09:51:39 +00:00
Bill Paul
8f0387a278 Correct the behavior of ndis_adjust_buflen(): the NDIS spec says
it's an error to set the buffer bytecount to anything larger than
the buffer's original allocation size, but anything less than that
is ok.

Also, in ndis_ptom(), use the same logic: if the bytecount is
larger than the allocation size, consider the bytecount invalid
and the allocation size as the packet fragment length (m_len)
instead of the bytecount.

This corrects a consistency problem between the Broadcom wireless
driver and some of the ethernet drivers I've tested: the ethernet
drivers all report the packet frag sizes in buf->nb_bytecount, but
the Broadcom wireless driver reports them in buf->nb_size. This
seems like a bug to me, but it clearly must work in Windows, so
we have to deal with it here too.
2003-12-12 08:54:48 +00:00
Jeff Roberson
cac77d0422 - Now that we have kseq groups, balance them seperately.
- The new sched_balance_groups() function does intra-group balancing while
   sched_balance() balances the available groups.
 - Pick a random time between 0 ticks and hz * 2 ticks to restart each
   balancing process.  Each balancer has its own timeout.
 - Pick a random place in the list of groups to start the search for lowest
   and highest group loads.  This prevents us from prefering a group based on
   numeric position.
 - Use a nasty hack to stop us from preferring cpu 0.  The problem is that
   softclock always runs on cpu 0, so it always has a little extra load.  We
   ignore this load in the balancer for now.  In the future softclock should
   run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
Bill Paul
d7e92f7b85 In NDIS 5.1 miniport drivers, the shutdown handler function pointer
is provided to NDIS via the the miniport characteristics structure
supplied in the call to NdisMRegisterMiniport(). But in NDIS 5.0
and earlier, you had to call NdisMRegisterAdapterShutdownHandler()
and supply both a function pointer and context pointer.

We try to handle both cases in ndis_shutdown_nic(). If the
driver registered a shutdown routine and a context,then used
that context, otherwise pass it the adapter context from
NdisMSetAttributesEx().

This fixes a panic on shutdown with the sample Intel 82559 e100bex.sys
driver from the Windows DDK.
function pointer
2003-12-12 05:27:58 +00:00
Bill Paul
c42411d56f Turn off build of if_ndis.ko by default, since it depends on an
autogenerated file (ndis_driver_data.h) which by definition can't
be available unless the user creates it.
2003-12-12 04:45:15 +00:00
Bill Paul
c7253f4fea Attempt to handle WinXP .INF files. Also, in dump_regvals(), handle
the case where there's an entry in the manufacturer's device list but
no corresponding installation section (and hence no AddReg assignments),
i.e. if dev = find_assign(sname, "AddReg"); returns NULL, then
don't try to dereference dev.

There is a fundamental problem with the handling of .INF files that
contain definitions for multiple devices: right now we dump all the
AddReg sections that we find, but don't distinguish what device they
belong to. This often results in duplicate keys.
2003-12-12 03:40:05 +00:00
Peter Wemm
6143d8ba5f Fix dynamic linking a bit more.. enough that mozilla-firebird works if you
dig up the patches for amd64 support for it.

Note to self: do not put a 64 bit value in a 32 bit space.
2003-12-12 01:12:41 +00:00
Bill Paul
da1ddac9f4 Fix handling of .INF files with more than one PCI ID defined. 2003-12-11 23:37:56 +00:00
Bill Paul
895ac9675f Fix getopt() string so -o works. 2003-12-11 23:30:36 +00:00
Bill Paul
87feaafc5b Grrr. Put the right .PATH statements in the right Makefiles. 2003-12-11 23:22:44 +00:00
Bill Paul
940311dc21 Ack! Only build if_ndis.ko and ndis.ko if arch == i386. 2003-12-11 23:06:54 +00:00
John Baldwin
acbf8a8d4a - Change the lookup() function to report success or failure using the carry
flag rather than explicitly halting if a lookup failed.
- Add a loop around the call to lookup() to traverse an array of
  nul-terminated strings for possible paths to the boot loader.  A double
  nul character denotes the end of the list.
- Add a new message to say that the boot failed if all of the path lookups
  for a boot loader file failed.
- Add '/boot/loader' as a second boot path.  If you build an ISO using
  risky options to mkisofs such as -U then the loader will be called
  '/boot/loader' rather than '/BOOT/LOADER;0'.  This allows cdboot to work
  with such risky ISO images.
- Bump version to 1.2 to denote added functionality.

The basic idea as well as some of the code were provided by the submitter,
but I added some extra code to use a loop rather than hard-code just 2
possible paths.

PR:		misc/43543
Submitted by:	kientzle
MFC after:	1 week
2003-12-11 22:42:50 +00:00
Bill Paul
d934c8b0de Commit the ndiscvt(8) utility too. (Missed it in the last import.) 2003-12-11 22:38:14 +00:00
Bill Paul
c854fc1092 Commit the first cut of Project Evil, also known as the NDISulator.
Yes, it's what you think it is. Yes, you should run away now.

This is a special compatibility module for allowing Windows NDIS
miniport network drivers to be used with FreeBSD/x86. This provides
_binary_ NDIS compatibility (not source): you can run NDIS driver
code, but you can't build it. There are three main parts:

sys/compat/ndis: the NDIS compat API, which provides binary
compatibility functions for many routines in NDIS.SYS, HAL.dll
and ntoskrnl.exe in Windows (these are the three modules that
most NDIS miniport drivers use). The compat module also contains
a small PE relocator/dynalinker which relocates the Windows .SYS
image and then patches in our native routines.

sys/dev/if_ndis: the if_ndis driver wrapper. This module makes
use of the ndis compat API and can be compiled with a specially
prepared binary image file (ndis_driver_data.h) containing the
Windows .SYS image and registry key information parsed out of the
accompanying .INF file. Once if_ndis.ko is built, it can be loaded
and unloaded just like a native FreeBSD kenrel module.

usr.sbin/ndiscvt: a special utility that converts foo.sys and foo.inf
into an ndis_driver_data.h file that can be compiled into if_ndis.o.
Contains an .inf file parser graciously provided by Matt Dodd (and
mercilessly hacked upon by me) that strips out device ID info and
registry key info from a .INF file and packages it up with a binary
image array. The ndiscvt(8) utility also does some manipulation of
the segments within the .sys file to make life easier for the kernel
loader. (Doing the manipulation here saves the kernel code from having
to move things around later, which would waste memory.)

ndiscvt is only built for the i386 arch. Only files.i386 has been
updated, and none of this is turned on in GENERIC. It should probably
work on pc98. I have no idea about amd64 or ia64 at this point.

This is still a work in progress. I estimate it's about %85 done, but
I want it under CVS control so I can track subsequent changes. It has
been tested with exactly three drivers: the LinkSys LNE100TX v4 driver
(Lne100v4.sys), the sample Intel 82559 driver from the Windows DDK
(e100bex.sys) and the Broadcom BCM43xx wireless driver (bcmwl5.sys). It
still needs to have a net80211 stuff added to it. To use it, you would
do something like this:

# cd /sys/modules/ndis
# make; make load
# cd /sys/modules/if_ndis
# ndiscvt -i /path/to/foo.inf -s /path/to/foo.sys -o ndis_driver_data.h
# make; make load
# sysctl -a | grep ndis

All registry keys are mapped to sysctl nodes. Sometimes drivers refer
to registry keys that aren't mentioned in foo.inf. If this happens,
the NDIS API module creates sysctl nodes for these keys on the fly so
you can tweak them.

An example usage of the Broadcom wireless driver would be:

# sysctl hw.ndis0.EnableAutoConnect=1
# sysctl hw.ndis0.SSID="MY_SSID"
# sysctl hw.ndis0.NetworkType=0 (0 for bss, 1 for adhoc)
# ifconfig ndis0 <my ipaddr> netmask 0xffffff00 up

Things to be done:

- get rid of debug messages
- add in ndis80211 support
- defer transmissions until after a status update with
  NDIS_STATUS_CONNECTED occurs
- Create smarter lookaside list support
- Split off if_ndis_pci.c and if_ndis_pccard.c attachments
- Make sure PCMCIA support works
- Fix ndiscvt to properly parse PCMCIA device IDs from INF files
- write ndisapi.9 man page
2003-12-11 22:34:37 +00:00
Bill Paul
0d27d14070 Commit the first cut of Project Evil, also known as the NDISulator.
Yes, it's what you think it is. Yes, you should run away now.

This is a special compatibility module for allowing Windows NDIS
miniport network drivers to be used with FreeBSD/x86. This provides
_binary_ NDIS compatibility (not source): you can run NDIS driver
code, but you can't build it. There are three main parts:

sys/compat/ndis: the NDIS compat API, which provides binary
compatibility functions for many routines in NDIS.SYS, HAL.dll
and ntoskrnl.exe in Windows (these are the three modules that
most NDIS miniport drivers use). The compat module also contains
a small PE relocator/dynalinker which relocates the Windows .SYS
image and then patches in our native routines.

sys/dev/if_ndis: the if_ndis driver wrapper. This module makes
use of the ndis compat API and can be compiled with a specially
prepared binary image file (ndis_driver_data.h) containing the
Windows .SYS image and registry key information parsed out of the
accompanying .INF file. Once if_ndis.ko is built, it can be loaded
and unloaded just like a native FreeBSD kenrel module.

usr.sbin/ndiscvt: a special utility that converts foo.sys and foo.inf
into an ndis_driver_data.h file that can be compiled into if_ndis.o.
Contains an .inf file parser graciously provided by Matt Dodd (and
mercilessly hacked upon by me) that strips out device ID info and
registry key info from a .INF file and packages it up with a binary
image array. The ndiscvt(8) utility also does some manipulation of
the segments within the .sys file to make life easier for the kernel
loader. (Doing the manipulation here saves the kernel code from having
to move things around later, which would waste memory.)

ndiscvt is only built for the i386 arch. Only files.i386 has been
updated, and none of this is turned on in GENERIC. It should probably
work on pc98. I have no idea about amd64 or ia64 at this point.

This is still a work in progress. I estimate it's about %85 done, but
I want it under CVS control so I can track subsequent changes. It has
been tested with exactly three drivers: the LinkSys LNE100TX v4 driver
(Lne100v4.sys), the sample Intel 82559 driver from the Windows DDK
(e100bex.sys) and the Broadcom BCM43xx wireless driver (bcmwl5.sys). It
still needs to have a net80211 stuff added to it. To use it, you would
do something like this:

# cd /sys/modules/ndis
# make; make load
# cd /sys/modules/if_ndis
# ndiscvt -i /path/to/foo.inf -s /path/to/foo.sys -o ndis_driver_data.h
# make; make load
# sysctl -a | grep ndis

All registry keys are mapped to sysctl nodes. Sometimes drivers refer
to registry keys that aren't mentioned in foo.inf. If this happens,
the NDIS API module creates sysctl nodes for these keys on the fly so
you can tweak them.

An example usage of the Broadcom wireless driver would be:

# sysctl hw.ndis0.EnableAutoConnect=1
# sysctl hw.ndis0.SSID="MY_SSID"
# sysctl hw.ndis0.NetworkType=0 (0 for bss, 1 for adhoc)
# ifconfig ndis0 <my ipaddr> netmask 0xffffff00 up

Things to be done:

- get rid of debug messages
- add in ndis80211 support
- defer transmissions until after a status update with
  NDIS_STATUS_CONNECTED occurs
- Create smarter lookaside list support
- Split off if_ndis_pci.c and if_ndis_pccard.c attachments
- Make sure PCMCIA support works
- Fix ndiscvt to properly parse PCMCIA device IDs from INF files
- write ndisapi.9 man page

PR:
Submitted by:
Reviewed by:
Approved by:
Obtained from:
MFC after:
2003-12-11 22:34:36 +00:00
Maxim Konovalov
369bf356c9 Grammar.
PR:		docs/60158
Submitted by:	Lukas Ertl
MFC after:	1 week
2003-12-11 21:54:26 +00:00
John Baldwin
bb17adda46 Properly fix a typo that the previous revision made even worse. 2003-12-11 20:40:12 +00:00
Peter Wemm
080f5381b7 Revert last change. ../rtld.c uses CACHE_LINE_SIZE too.
Change it to 64 while here.

Reported by:  ps
2003-12-11 18:42:51 +00:00