]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
19 years agoBring the prototype for alq_open() in line with the code.
Joseph Koshy [Sun, 30 Jan 2005 10:21:04 +0000 (10:21 +0000)]
Bring the prototype for alq_open() in line with the code.

Add a section on locking.

19 years agoStart gettys on ttyu0 and ttyu1 instead of ttya and ttyz0 now that
Marcel Moolenaar [Sun, 30 Jan 2005 09:31:22 +0000 (09:31 +0000)]
Start gettys on ttyu0 and ttyu1 instead of ttya and ttyz0 now that
uart(4) is the default driver.

MFC after: 2 weeks

19 years agoo Enable puc(4) and uart(4).
Marcel Moolenaar [Sun, 30 Jan 2005 09:27:49 +0000 (09:27 +0000)]
o  Enable puc(4) and uart(4).
o  Disable ofw_console(4), sab(4) and zs(4).

sab(4) and zs(4) are disabled because the hardware controlled by
them is handled by uart(4)+puc(4) and the latter combination is
functionally complete and up to date.

ofw_console(4) is disabled because it doesn't claim the device it
controls (through OFW) and thus interferes with puc(4)+uart(4),
which has sufficient knowledge to extract the necessary information
from OFW to setup the console. Put differently, ofw_console(4) is
not a proper device driver and can only do harm. Its functionality
is completely handled by uart(4).

This commit makes uart(4) the default driver for serial ports.

MFC after: 2 weeks

19 years agoo Fix the various interrupt related problems caused by reverse
Marcel Moolenaar [Sun, 30 Jan 2005 09:00:50 +0000 (09:00 +0000)]
o  Fix the various interrupt related problems caused by reverse
   engineering the pending interrupt sources from the current
   state of the controller. For channel A we can always read the
   interrupt pending register (RR3). For channel B we can read
   the interrupt vector register (RR2) because it contains the
   modified vector and thus includes the interrupt source.
   Since we currently need puc(4) for the Z8530, we know that
   the interrupt handler for both channels will be called and
   thus that RR3 will always be read at least once, even if ch A
   has no pending interrupt.
   NOTE: The modified interrupt vector has no value that represent
   a lack of pending interrupt for channel B. That is, the
   value read when no interrupts are pending is the same as the
   value for the special receive condition. Fortunately, we don't
   actually have to depend on that interrupt source. This does
   mean that we need to properly handle the overflow condition,
   when we read received character from the chip.
o  The DSR signal is represented by the SYNC bit in the external
   status register (RR0). We now properly track DSR.
o  It's save to enable the external/status interrupt source. We
   now get interrupts when line signals (DSR, DCD or CTS) change.

Problems fixes:
o  interrupt storms.
o  blocked open(2).
o  lack of (hardware) flow control.
o  unable to report DSR.

MFC after: 5 days

19 years agoBoot away another stackgap (one of the lest ones in linuxlator/i386) by
Maxim Sobolev [Sun, 30 Jan 2005 08:12:37 +0000 (08:12 +0000)]
Boot away another stackgap (one of the lest ones in linuxlator/i386) by
providing special version of CDIOCREADSUBCHANNEL ioctl(), which assumes that
result has to be placed into kernel space not user space. In the long run
more generic solution has to be designed WRT emulating various ioctl()s
that operate on userspace buffers, but right now there is only one such
ioctl() is emulated, so that it makes little sense.

MFC after: 2 weeks

19 years agoAdd text about jail root directory as well.
Poul-Henning Kamp [Sun, 30 Jan 2005 08:03:20 +0000 (08:03 +0000)]
Add text about jail root directory as well.

Submitted by: "Mark W. Krentel" <krentel@dreamscape.com>

19 years agoExtend kern_sendit() to take another enum uio_seg argument, which specifies
Maxim Sobolev [Sun, 30 Jan 2005 07:20:36 +0000 (07:20 +0000)]
Extend kern_sendit() to take another enum uio_seg argument, which specifies
where the buffer to send lies and use it to eliminate yet another stackgap
in linuxlator.

MFC after: 2 weeks

19 years agoFix build on AMD64 (and probably other arches where size_t != int).
Maxim Sobolev [Sun, 30 Jan 2005 06:43:17 +0000 (06:43 +0000)]
Fix build on AMD64 (and probably other arches where size_t != int).

Submitted by: Tinderbox
MFC after: 2 weeks

19 years agoFix spelling of integer in a comment.
Robert Watson [Sun, 30 Jan 2005 00:31:19 +0000 (00:31 +0000)]
Fix spelling of integer in a comment.

Beady eyes: ceri

19 years agoGrrr, this committer needs to have a sleep. Remove lines from the previous
Maxim Sobolev [Sat, 29 Jan 2005 23:51:05 +0000 (23:51 +0000)]
Grrr, this committer needs to have a sleep. Remove lines from the previous
delta not intended for public consumption.

MFC after: 2 weeks

19 years agoFix small non-conformance introduced in the previous commit: execve() is
Maxim Sobolev [Sat, 29 Jan 2005 23:47:36 +0000 (23:47 +0000)]
Fix small non-conformance introduced in the previous commit: execve() is
expected to return ENAMETOOLONG, not E2BIG if first argument doesn't
fit into {PATH_MAX} bytes.

MFC after: 2 weeks

19 years agoRemove local hack which cowardly slipped into previous commit.
Maxim Sobolev [Sat, 29 Jan 2005 23:21:00 +0000 (23:21 +0000)]
Remove local hack which cowardly slipped into previous commit.

MFC after: 2 weeks

19 years agoStyle cleanup for O_DIRECT sysctl comment introduced in nfs_vnops.c:1.242.
Robert Watson [Sat, 29 Jan 2005 23:19:08 +0000 (23:19 +0000)]
Style cleanup for O_DIRECT sysctl comment introduced in nfs_vnops.c:1.242.

19 years agoDisable ethernet flow control in if_fxp by default, in order to prevent
Robert Watson [Sat, 29 Jan 2005 23:13:20 +0000 (23:13 +0000)]
Disable ethernet flow control in if_fxp by default, in order to prevent
unexpected surprises when a system panics or is left in the debugger.

Requested by: kris
MFC after: 3 days

19 years agoo Split out kernel part of execve(2) syscall into two parts: one that
Maxim Sobolev [Sat, 29 Jan 2005 23:12:00 +0000 (23:12 +0000)]
o Split out kernel part of execve(2) syscall into two parts: one that
  copies arguments into the kernel space and one that operates
  completely in the kernel space;

o use kernel-only version of execve(2) to kill another stackgap in
  linuxlator/i386.

Obtained from:  DragonFlyBSD (partially)
MFC after:      2 weeks

19 years agoCorrect a minr whitespace inconsistency introduced in revision 1.159:
Robert Watson [Sat, 29 Jan 2005 22:04:30 +0000 (22:04 +0000)]
Correct a minr whitespace inconsistency introduced in revision 1.159:
add a tab between #define and DF_REBID instead of a space.

19 years agoHopefully unbreak modules build.
Ruslan Ermilov [Sat, 29 Jan 2005 21:43:34 +0000 (21:43 +0000)]
Hopefully unbreak modules build.

19 years agoStart sentences from a capital letter.
Ruslan Ermilov [Sat, 29 Jan 2005 20:05:07 +0000 (20:05 +0000)]
Start sentences from a capital letter.

Submitted by: Joel Dahl

19 years agoFix typo.
Nate Lawson [Sat, 29 Jan 2005 19:45:31 +0000 (19:45 +0000)]
Fix typo.

19 years agoCleanup debugging code and put it under bootverbose
Bjoern A. Zeeb [Sat, 29 Jan 2005 19:26:53 +0000 (19:26 +0000)]
Cleanup debugging code and put it under bootverbose
(includes minor style polishing).

Approved by: rwatson (mentor)

19 years agoUse MAXMINOR
Poul-Henning Kamp [Sat, 29 Jan 2005 16:50:04 +0000 (16:50 +0000)]
Use MAXMINOR

19 years agoWhen dumping to a unpartitioned disk, make sure to chop the
Poul-Henning Kamp [Sat, 29 Jan 2005 16:49:43 +0000 (16:49 +0000)]
When dumping to a unpartitioned disk, make sure to chop the
length of the dump area accordingly.

Run into by: scottl

19 years agoDescribe 'u' and 'U' in sequence mode.
Poul-Henning Kamp [Sat, 29 Jan 2005 16:34:10 +0000 (16:34 +0000)]
Describe 'u' and 'U' in sequence mode.

19 years agoBetter locking.
Poul-Henning Kamp [Sat, 29 Jan 2005 16:33:51 +0000 (16:33 +0000)]
Better locking.

Add 'u' and 'U' "wait for next UTC second" in sequence mode.

19 years agoUnbreak a few filesystems for which vnode_create_vobject() wasn't being
Peter Edwards [Sat, 29 Jan 2005 16:23:39 +0000 (16:23 +0000)]
Unbreak a few filesystems for which vnode_create_vobject() wasn't being
called in "open", causing mmap() to fail.

Where possible, pass size of file to vnode_create_vobject() rather
than having it find it out the hard way via VOP_LOOKUP

Reviewed by: phk

19 years agoTypo.
Poul-Henning Kamp [Sat, 29 Jan 2005 15:10:30 +0000 (15:10 +0000)]
Typo.

19 years agoAdd MAXMINOR #define, we should have had this long time ago.
Poul-Henning Kamp [Sat, 29 Jan 2005 15:07:13 +0000 (15:07 +0000)]
Add MAXMINOR #define, we should have had this long time ago.

Add minor2unit() in addition to dev2unit() and unit2minor().

If it wasn't such a hazzle we should redefine minor numbers in
the kernel without the gap for the major number, but it's not worth
the bother (yet).

19 years agoTell vnode_create_vobject() how big an object to create, rather
Peter Edwards [Sat, 29 Jan 2005 14:23:09 +0000 (14:23 +0000)]
Tell vnode_create_vobject() how big an object to create, rather
than having it work it out via the more expensive VOP_GETATTR

Reviewed by: phk@

19 years agoIn 1.276 of kern/subr_trap.c I introduced a mechanism for delaying
Poul-Henning Kamp [Sat, 29 Jan 2005 14:03:41 +0000 (14:03 +0000)]
In 1.276 of kern/subr_trap.c I introduced a mechanism for delaying
a process return to userspace if it had pending GEOM events.

We need to have the same check in the exit pass to catch the case
where a GEOM related filedescriptor is not explicitly closed by
the process.

Bumped into by: people using dd(1) to build releases, nanobsd etc.

19 years agoBecause the argument to strvalid() is a size_t, use a size_t to iterate
Robert Watson [Sat, 29 Jan 2005 13:34:22 +0000 (13:34 +0000)]
Because the argument to strvalid() is a size_t, use a size_t to iterate
over the array.

Submitted by: stefanf

19 years agoPer permission from McAfee, remove clause 3 of the BSD license on
Robert Watson [Sat, 29 Jan 2005 13:04:04 +0000 (13:04 +0000)]
Per permission from McAfee, remove clause 3 of the BSD license on
strvalid.c.

19 years agoReport jail directory if set.
Poul-Henning Kamp [Sat, 29 Jan 2005 11:38:00 +0000 (11:38 +0000)]
Report jail directory if set.

Sanity-check fd_lastfile.

PR: 62699
Patch by: "Mark W. Krentel" <krentel@dreamscape.com>

19 years agoThe Olicom OC2220 has an ID of 0x101, not 0x2220 as I previously
Warner Losh [Sat, 29 Jan 2005 04:20:01 +0000 (04:20 +0000)]
The Olicom OC2220 has an ID of 0x101, not 0x2220 as I previously
thought.  I'm unsure why I thought this was the case, but it
definitely isn't for this card.  If another card with the other ID
makes an appearance, then we'll add a second entry for it.

# With this change my Olicom OC2220 is now working again, since I make
# this commit with that device. :-)

19 years agoRoll firmware to the latest version. There are a bunch of features
Matt Jacob [Sat, 29 Jan 2005 01:12:37 +0000 (01:12 +0000)]
Roll firmware to the latest version. There are a bunch of features
in alternate f/w versions that will be pursued at some points.

MFC after: 1 month

19 years agoFixed whitespace at EOL and double word.
Ruslan Ermilov [Fri, 28 Jan 2005 21:18:14 +0000 (21:18 +0000)]
Fixed whitespace at EOL and double word.
Uncompact enumaration to make it look better.

19 years agoTypo.
Ruslan Ermilov [Fri, 28 Jan 2005 21:14:16 +0000 (21:14 +0000)]
Typo.

19 years agoProperly terminate sentence.
Ruslan Ermilov [Fri, 28 Jan 2005 21:13:34 +0000 (21:13 +0000)]
Properly terminate sentence.

19 years agouse strdup().
Hajimu UMEMOTO [Fri, 28 Jan 2005 19:35:42 +0000 (19:35 +0000)]
use strdup().

Obtained from: KAME
MFC after: 1 week

19 years ago - Don't drop the wref on the bufobj until after bufdone() has completed.
Jeff Roberson [Fri, 28 Jan 2005 17:48:58 +0000 (17:48 +0000)]
 - Don't drop the wref on the bufobj until after bufdone() has completed.
   Without this, threads waiting in bufobj_wwait() may wakeup prior to
   bufdone() completing.

Sponsored by: Isilon Systems, Inc.

19 years agoClean up after utilizing FILES.
Ruslan Ermilov [Fri, 28 Jan 2005 17:13:39 +0000 (17:13 +0000)]
Clean up after utilizing FILES.

19 years agoClean up makefile.
Ruslan Ermilov [Fri, 28 Jan 2005 17:01:54 +0000 (17:01 +0000)]
Clean up makefile.

- Remove all:.  It's redundant, and ${LIB} in it is just a bug.
- Remove .ORDER:.  *.mgc files can safely be built in parallel.
- Remove PITA.  The mkmagic tool is smart to put the binary file
  into the current directory (${.OBJDIR}) even if the source file
  lives somewhere else, which is just what we need.

19 years agoUtilize bsd.files.mk.
Ruslan Ermilov [Fri, 28 Jan 2005 16:22:46 +0000 (16:22 +0000)]
Utilize bsd.files.mk.

19 years agoLet bsd.prog.mk set SRCS and MAN to their default values.
Ruslan Ermilov [Fri, 28 Jan 2005 16:08:11 +0000 (16:08 +0000)]
Let bsd.prog.mk set SRCS and MAN to their default values.

19 years ago - If mpsafevfs is off, acquire giant around all calls to bufdone().
Jeff Roberson [Fri, 28 Jan 2005 16:04:44 +0000 (16:04 +0000)]
 - If mpsafevfs is off, acquire giant around all calls to bufdone().

Sponsored by:   Isilon Systems, Inc.

19 years agoMake filesystems get rid of their own vnodes vnode_pager object in
Poul-Henning Kamp [Fri, 28 Jan 2005 14:42:17 +0000 (14:42 +0000)]
Make filesystems get rid of their own vnodes vnode_pager object in
VOP_RECLAIM().

19 years agoRemove unused argument to vrecycle()
Poul-Henning Kamp [Fri, 28 Jan 2005 13:08:21 +0000 (13:08 +0000)]
Remove unused argument to vrecycle()

19 years agoIntegrate vclean() into vgonel().
Poul-Henning Kamp [Fri, 28 Jan 2005 13:00:03 +0000 (13:00 +0000)]
Integrate vclean() into vgonel().

Various associated polishing.

19 years agoRemove register keyword
Poul-Henning Kamp [Fri, 28 Jan 2005 12:39:10 +0000 (12:39 +0000)]
Remove register keyword

19 years agoUpdate kdump manual page with information on the dump format.
George V. Neville-Neil [Fri, 28 Jan 2005 12:17:33 +0000 (12:17 +0000)]
Update kdump manual page with information on the dump format.
Add a table with the different types of operations traced.

Reviewed by: Ruslan Ermilov
Approved by: Robert Watson (mentor)

19 years agoMark lists as in the rest of section 9 manpages.
Ruslan Ermilov [Fri, 28 Jan 2005 10:48:52 +0000 (10:48 +0000)]
Mark lists as in the rest of section 9 manpages.

19 years agoMark lists as in the rest of section 9 manpages.
Ruslan Ermilov [Fri, 28 Jan 2005 10:43:05 +0000 (10:43 +0000)]
Mark lists as in the rest of section 9 manpages.

Submitted by: Joel Dahl (mostly)

19 years agoMove the contents of vop_stddestroyvobject() to the new vnode_pager
Poul-Henning Kamp [Fri, 28 Jan 2005 08:56:48 +0000 (08:56 +0000)]
Move the contents of vop_stddestroyvobject() to the new vnode_pager
function vnode_destroy_vobject().

Make the new function zero the vp->v_object pointer so we can tell
if a call is missing.

19 years agoFix the worst offenders of style(9) with a small style sweep.
Warner Losh [Fri, 28 Jan 2005 06:50:59 +0000 (06:50 +0000)]
Fix the worst offenders of style(9) with a small style sweep.

19 years agoCleanup tabs vs spaces.
Warner Losh [Fri, 28 Jan 2005 06:45:42 +0000 (06:45 +0000)]
Cleanup tabs vs spaces.

19 years agoFor the PC Card implementation of the CS8920M that's in my IBM
Warner Losh [Fri, 28 Jan 2005 06:35:39 +0000 (06:35 +0000)]
For the PC Card implementation of the CS8920M that's in my IBM
EtherJet, the interrupt is selected in the eeprom based on the layout
of the PC Card board.  Since this is encoded into the EEPROM, and has
no relationship to the IRQ that the pccard bridge routes the PC Card's
interrupt pin to.

As such, stop writing to that register.  This gets my EtherJet working.

# The eeprom reading code appears to be totally wrong for my EtherJet
# card.  This causes the card to bogusly detect the media options
# available.

19 years agoSetting hw.cs.recv_delay should set the delay, not the ignore the
Warner Losh [Fri, 28 Jan 2005 06:13:29 +0000 (06:13 +0000)]
Setting hw.cs.recv_delay should set the delay, not the ignore the
eeprom checksum.

19 years agoRemove all of the spl() markers.
Scott Long [Fri, 28 Jan 2005 05:37:51 +0000 (05:37 +0000)]
Remove all of the spl() markers.

19 years agoLock the IPS driver and bring it out from under Giant. Also do some
Scott Long [Fri, 28 Jan 2005 05:02:13 +0000 (05:02 +0000)]
Lock the IPS driver and bring it out from under Giant.  Also do some
significant clean up and optimizations:
- don't call bioq_disksort() on every command, the hardware will do that for
  us.
- remove all of the complicated bio deferral code.  bio's that can't be
  serviced immediately can just wait on the bioq.
- Only reserve one command object for doing control commands to the card.
  This simplifies a lot of code and significantly reduces the size of the
  command struct.
- Allocate commands out of a slab instead of embedding them into the softc.
- Call the command action method directly instead of having ips_get_free_cmd()
  call it indirectly.

MFC After: 1 week

19 years agoWrap calls to memcpy(3) in a function called block_copy(). This way,
Marcel Moolenaar [Fri, 28 Jan 2005 02:58:32 +0000 (02:58 +0000)]
Wrap calls to memcpy(3) in a function called block_copy(). This way,
and as long as we're not compiling with IPA, gcc(1) won't optimize
the call away. The whole purpose of using memcpy(3) is to avoid
misaligned loads and stores when we need to read or write the value
in the unaligned memory location. But if gcc(1) optimizes the call
to memcpy(3) away, it will typically introduce misaligned loads and
stores. In this context that's not a good idea.

19 years agoerror = is needed before ether_ioctl() so that unsupported/unknown
Warner Losh [Fri, 28 Jan 2005 00:28:22 +0000 (00:28 +0000)]
error = is needed before ether_ioctl() so that unsupported/unknown
IOCLTs are properly handled.  This gets the cs driver properly
reporting things via ifconfig.

# my pccard still doesn't work.

19 years agoSkip the register based postinc stores here and don't count them
Marcel Moolenaar [Thu, 27 Jan 2005 23:15:58 +0000 (23:15 +0000)]
Skip the register based postinc stores here and don't count them
in the number of tests. This way we avoid skipped tests for non-
existent memory access combinations. The number of tests dropped
from 60 to 48.

19 years agoMake the local variables global so that the compiler cannot assume
Marcel Moolenaar [Thu, 27 Jan 2005 23:11:55 +0000 (23:11 +0000)]
Make the local variables global so that the compiler cannot assume
too much about them. This prevents certain peephole optimizations
at -O that invalidate the tests.

19 years agoFix the unaligned store with post increment test: The misaligned pointer
Marcel Moolenaar [Thu, 27 Jan 2005 22:46:15 +0000 (22:46 +0000)]
Fix the unaligned store with post increment test: The misaligned pointer
stopped pointing to the value the moment we wrote it due to the post
increment. So, grab the value for comparison out of the data structure
directly.

19 years agoConsistently use pcib for a printf.
Nate Lawson [Thu, 27 Jan 2005 20:49:59 +0000 (20:49 +0000)]
Consistently use pcib for a printf.

19 years agosupply a default ic_reset method for drivers; the ioctl code expect this
Sam Leffler [Thu, 27 Jan 2005 17:39:17 +0000 (17:39 +0000)]
supply a default ic_reset method for drivers; the ioctl code expect this
method to always be setup

Submitted by: Tai-hwa Liang

19 years agoChange the ifr_media operation to only get its value and only set
Doug Ambrisko [Thu, 27 Jan 2005 16:40:12 +0000 (16:40 +0000)]
Change the ifr_media operation to only get its value and only set
its value once per ifconfig run.  Use Sam's new callback
operation to set it when everything is done.

The purpose for this is that if you did something like
ifconfig bge0 media 100baseTX mediaopt full-duplex
multiple times it would end up causing the PHY to re-sync
since it would send the IOCTLs:
ifconfig bge0 media 100baseTX -mediaopt full-duplex
ifconfig bge0 media 100baseTX mediaopt full-duplex
This would cause the PHY to be updated twice even though
there really wasn't any change since the check in
sys/net/if_media.c would always fail.

Caveat is that this doesn't fix the case of:
ifconfig bge0 media autoselect
etc. since in sys/net/if_media.c it forces an autoselect to go through
the entire process in ifmedia_ioctl :-( :
          /*
           * If no change, we're done.
           * XXX Automedia may invole software intervention.
           *     Keep going in case the the connected media changed.
           *     Similarly, if best match changed (kernel debugger?).
           */
           if ((IFM_SUBTYPE(newmedia) != IFM_AUTO) &&
                (newmedia == ifm->ifm_media) &&
                (match == ifm->ifm_cur))
                     return 0;

Briefly looked at by: sam

19 years agomake _getipnodebyname_multi() static.
Hajimu UMEMOTO [Thu, 27 Jan 2005 15:01:05 +0000 (15:01 +0000)]
make _getipnodebyname_multi() static.

19 years agoRemove unnecessary SRCS= where could be guessed directly by our
Xin LI [Thu, 27 Jan 2005 14:52:47 +0000 (14:52 +0000)]
Remove unnecessary SRCS= where could be guessed directly by our
bsd.*.mk infrasture.

Obtained from: ru

19 years agoimplement AI_NUMERICSERV (as defined in RFC3493).
Hajimu UMEMOTO [Thu, 27 Jan 2005 14:45:11 +0000 (14:45 +0000)]
implement AI_NUMERICSERV (as defined in RFC3493).

Obtained from: KAME
MFC after: 1 week

19 years agoWARNS?=6 cleanup for [gs]et[fp]mac:
Xin LI [Thu, 27 Jan 2005 14:44:39 +0000 (14:44 +0000)]
WARNS?=6 cleanup for [gs]et[fp]mac:
- Constify structure members that should not be changed
  during process.
- Apply static where needed
- signed/unsigned madness
- Bump WARNS?= levels from 2 to 6

(this is a diff reduction for a subsequent commit against these
Makefile's)

19 years agofill ai_canonname field for numeric hostname, by the given hostname.
Hajimu UMEMOTO [Thu, 27 Jan 2005 14:41:56 +0000 (14:41 +0000)]
fill ai_canonname field for numeric hostname, by the given hostname.
follow new recommendation in RFC3493.

Obtained from: KAME
MFC after: 1 week

19 years agoAdd a radius_Flush() function that waits for the response (or timeout) to
Brian Somers [Thu, 27 Jan 2005 14:09:33 +0000 (14:09 +0000)]
Add a radius_Flush() function that waits for the response (or timeout) to
any pending RADIUS transaction.  Use this before sending RAD_STOP RADIUS
messages so that we definitely ``stop'' the session.

It was discovered that sometimes when the link timed out, we got lucky
enough to have an un-ACK'd RADIUS accounting transaction in progress,
resulting in the RAD_STOP message failing to send.

Original report found on: A russion news group
Text translated by: glebius
Tested by: Alexey Popov llp at iteranet dot com
MFC after: 7 days

19 years agoMake NTFS at least minimally usable after bufobj and GEOM fallout.
Peter Edwards [Thu, 27 Jan 2005 13:50:27 +0000 (13:50 +0000)]
Make NTFS at least minimally usable after bufobj and GEOM fallout.

mmap() on NTFS files was hosed, returning pages offset from the
start of the disk rather than the start of the file. (ie, "cp" of
a 1-block file would get you a copy of the boot sector, not the
data in the file.) The solution isn't ideal, but gives a functioning
filesystem.

Cached vnode lookup was also broken, resulting in vnode haemorrhage.
A lookup on the same file twice would give you two vnodes, and the
resulting cached pages.

Just recently, mmap() was broken due to a lack of a call to
vnode_create_vobject() in ntfs_open().

Discussed with: phk@

19 years agops(1) is WARNS=6 on all Tier-1 platforms as far as I can test,
Xin LI [Thu, 27 Jan 2005 13:42:40 +0000 (13:42 +0000)]
ps(1) is WARNS=6 on all Tier-1 platforms as far as I can test,
so use default WARNS level from bin/Makefile.inc

19 years agoDon't specify SRCS= when it can be obtained from PROGS=
Xin LI [Thu, 27 Jan 2005 13:25:09 +0000 (13:25 +0000)]
Don't specify SRCS= when it can be obtained from PROGS=

19 years agoWARNS?=6 is already in bin/Makefile.inc, so remove this one.
Xin LI [Thu, 27 Jan 2005 13:23:05 +0000 (13:23 +0000)]
WARNS?=6 is already in bin/Makefile.inc, so remove this one.

19 years ago"pst" is not 64-bit clean for reasons specified in sys/amd64/conf/NOTES.
Ruslan Ermilov [Thu, 27 Jan 2005 11:07:13 +0000 (11:07 +0000)]
"pst" is not 64-bit clean for reasons specified in sys/amd64/conf/NOTES.

19 years agoPolish the formatting.
Ruslan Ermilov [Thu, 27 Jan 2005 10:30:53 +0000 (10:30 +0000)]
Polish the formatting.

19 years agoUpdate the information, (and likely break all formatting rules).
Poul-Henning Kamp [Thu, 27 Jan 2005 09:29:52 +0000 (09:29 +0000)]
Update the information, (and likely break all formatting rules).

19 years agoComment out "lnc" on amd64 for reasons specified in sys/amd64/conf/NOTES.
Ruslan Ermilov [Thu, 27 Jan 2005 09:21:23 +0000 (09:21 +0000)]
Comment out "lnc" on amd64 for reasons specified in sys/amd64/conf/NOTES.

19 years agoquery A RR before AAAA RR.
Hajimu UMEMOTO [Thu, 27 Jan 2005 08:03:46 +0000 (08:03 +0000)]
query A RR before AAAA RR.

MFC after: 1 week

19 years agoThe unit test for unaligned loads/stores can be found under ../ia64.
Marcel Moolenaar [Thu, 27 Jan 2005 06:51:45 +0000 (06:51 +0000)]
The unit test for unaligned loads/stores can be found under ../ia64.

19 years agoAdd tests for post increment. This bumps the number of tests up to
Marcel Moolenaar [Thu, 27 Jan 2005 06:50:05 +0000 (06:50 +0000)]
Add tests for post increment. This bumps the number of tests up to
60. The postinc store tests currently fail (value mismatch). Hence
the score as of this commit is 48 out of 60. Either the kernel or
the tests need to be fixed.

19 years agoFix handling of post increment: Either the first or second operand
Marcel Moolenaar [Thu, 27 Jan 2005 06:01:44 +0000 (06:01 +0000)]
Fix handling of post increment: Either the first or second operand
is the register with the memory address, and it's that register's
value we need to increment or decrement.

MFC after: 3 days

19 years ago- Move the functions presently described in in ieee(3) to their own
David Schultz [Thu, 27 Jan 2005 05:46:17 +0000 (05:46 +0000)]
- Move the functions presently described in in ieee(3) to their own
  manpages.  They are not very related, so separating them makes it
  easier to add meaningful cross-references and extend some of the
  descriptions.
- Move the part of math(3) that discusses IEEE 754 to the ieee(3)
  manpage.

19 years agoDon't compile the gdtoa package's strtoIg.c into libc.
David Schultz [Thu, 27 Jan 2005 05:44:16 +0000 (05:44 +0000)]
Don't compile the gdtoa package's strtoIg.c into libc.
I never got around to making use of it.

19 years agoWrite cs_detach() and use it. This resolves the twin problems of the
Warner Losh [Thu, 27 Jan 2005 04:51:44 +0000 (04:51 +0000)]
Write cs_detach() and use it.  This resolves the twin problems of the
cs1 interface linger on card eject, as well as the warnings about the
card still using resources.  Ooops.

19 years agoAdd back support for D-LINK DMR-650TX, and all the other OEMd versions
Warner Losh [Thu, 27 Jan 2005 02:37:39 +0000 (02:37 +0000)]
Add back support for D-LINK DMR-650TX, and all the other OEMd versions
of this card (evidentally MultiMobile also sold this card as
MT5634ZLXI/E).

Reported by: Bastian Brinkman

19 years agoMove 143 back to its rightful owner: Grey Cell systems, which OEMd enet cards
Warner Losh [Thu, 27 Jan 2005 02:33:03 +0000 (02:33 +0000)]
Move 143 back to its rightful owner: Grey Cell systems, which OEMd enet cards

19 years ago64-bit clean fixes: Use %zx in preference to %x to print size_t items.
Warner Losh [Thu, 27 Jan 2005 01:49:23 +0000 (01:49 +0000)]
64-bit clean fixes: Use %zx in preference to %x to print size_t items.
Cast a byte to uint8_t before printing.

19 years agoar and sr (and their netgraph cousins) don't appear to be 64-bit clean, so
Warner Losh [Thu, 27 Jan 2005 01:45:15 +0000 (01:45 +0000)]
ar and sr (and their netgraph cousins) don't appear to be 64-bit clean, so
disable them on all but i386.

19 years agoFix a few printf problems on ia64 (and other 64-bit platforms).
Warner Losh [Thu, 27 Jan 2005 01:40:12 +0000 (01:40 +0000)]
Fix a few printf problems on ia64 (and other 64-bit platforms).

19 years agoAdd PCI ID for Dell RAC IV/ERA Virtual UART (PowerEdge 1850).
Bruce M Simpson [Thu, 27 Jan 2005 01:17:32 +0000 (01:17 +0000)]
Add PCI ID for Dell RAC IV/ERA Virtual UART (PowerEdge 1850).
Trim name of existing Dell RAC devices. Trim comments.

With help from: dpk at dpk dot net
MFC after: 1 week

19 years agoThe ar driver appears to do naughty things with pointers and integers
Warner Losh [Thu, 27 Jan 2005 01:12:58 +0000 (01:12 +0000)]
The ar driver appears to do naughty things with pointers and integers
and so appears to not be 64-bit clean.  disable it on ia64 for the moment.

19 years agopcic is gone
Warner Losh [Thu, 27 Jan 2005 00:33:17 +0000 (00:33 +0000)]
pcic is gone

19 years agoRemove policy references to mpo_check_vnode_mprotect(), which is
Robert Watson [Wed, 26 Jan 2005 23:43:32 +0000 (23:43 +0000)]
Remove policy references to mpo_check_vnode_mprotect(), which is
currently unimplemented.

Update copyrights.

Pointed out by: csjp

19 years agoIf CDSR_OFLOW (stty dsrflow) is enabled on one or both sides of a
Poul-Henning Kamp [Wed, 26 Jan 2005 23:42:18 +0000 (23:42 +0000)]
If CDSR_OFLOW (stty dsrflow) is enabled on one or both sides of a
null-modem tty device emulate the speed settings faithfully.

The speed is emulated independently for the two directions, using
the slower of the local sides ispeed and the remote sides ospeed.

The emulated speed takes settings of bits/char, parity and stopbit
into account.

Inspired by: The BSD-DK Editor Celebrity Deathmatch Contest

19 years agoAdd required ing
Warner Losh [Wed, 26 Jan 2005 23:27:12 +0000 (23:27 +0000)]
Add required ing

19 years agoAdd -fno-strict-alias whenever someone is compiling with -O2,
Warner Losh [Wed, 26 Jan 2005 21:35:55 +0000 (21:35 +0000)]
Add -fno-strict-alias whenever someone is compiling with -O2,
unconditionally.

19 years agoAdd cs module. It has built in my tree for ages, and it just never
Warner Losh [Wed, 26 Jan 2005 21:27:26 +0000 (21:27 +0000)]
Add cs module.  It has built in my tree for ages, and it just never
made it into FreeBSD's tree.