]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
14 years agoPrevent NULL deference by checking return value of
delphij [Tue, 2 Feb 2010 22:25:22 +0000 (22:25 +0000)]
Prevent NULL deference by checking return value of
gctl_get_asciiparam.

MFC after: 2 weeks

14 years agoSome of the existing ppp and vpn related scripts create and set
qingli [Tue, 2 Feb 2010 20:38:30 +0000 (20:38 +0000)]
Some of the existing ppp and vpn related scripts create and set
the IP addresses of the tunnel end points to the same value. In
these cases the loopback route is not installed for the local
end.

Verified by: avg
MFC after: 5 days

14 years agoBump .Dd, forgotten in r203393
gavin [Tue, 2 Feb 2010 19:44:51 +0000 (19:44 +0000)]
Bump .Dd, forgotten in r203393

Approved by: ed (mentor, implicit)
MFC after: 1 week

14 years agoThe multiplicand a = 0x5deece66d = 25214903917, not 0xfdeece66d.
gavin [Tue, 2 Feb 2010 19:28:01 +0000 (19:28 +0000)]
The multiplicand a = 0x5deece66d = 25214903917, not 0xfdeece66d.
This bug in the man page has gone unnoticed for over 15 years!

PR: docs/143461
Submitted by: Jeremy Huddleston  jeremyhu apple.com
Approved by: ed (mentor, implicit)
MFC after: 1 week

14 years agoImplement strndup(3) using strnlen(3).
ed [Tue, 2 Feb 2010 19:02:08 +0000 (19:02 +0000)]
Implement strndup(3) using strnlen(3).

This makes the implementation a bit more consistent with strdup(3),
which uses strlen(3).

14 years agoExclude the interfaces which IPv6 and/or accepting RA is disabled
ume [Tue, 2 Feb 2010 18:38:17 +0000 (18:38 +0000)]
Exclude the interfaces which IPv6 and/or accepting RA is disabled
from the auto probed interface list.

MFC after: 1 week

14 years agoChange the way in which fake async events generated. Do not use
mav [Tue, 2 Feb 2010 18:07:16 +0000 (18:07 +0000)]
Change the way in which fake async events generated. Do not use
taskqueue for lock decoupling, as it causes unwanted races.

14 years ago- Use separate buffer for identify data fetching. We can't use main buffer
mav [Tue, 2 Feb 2010 18:03:21 +0000 (18:03 +0000)]
- Use separate buffer for identify data fetching. We can't use main buffer
here if device already running, as data need to be formatted before use.
- Remove some saved_ccb variables. They are unused now.

14 years agoSome missed parts for r203376.
mav [Tue, 2 Feb 2010 17:56:54 +0000 (17:56 +0000)]
Some missed parts for r203376.

14 years agoMake -a option actually work.
ume [Tue, 2 Feb 2010 16:02:09 +0000 (16:02 +0000)]
Make -a option actually work.

MFC after: 1 week

14 years agoRespect passwordtime from login.conf if set.
des [Tue, 2 Feb 2010 13:47:18 +0000 (13:47 +0000)]
Respect passwordtime from login.conf if set.

PR: bin/93473
Submitted by: Björn König <bkoenig@cs.tu-berlin.de>
MFC after: 1 week

14 years ago- Give ATA/SATA SIMs info about ATAPI packet size, supported by device.
mav [Tue, 2 Feb 2010 11:09:28 +0000 (11:09 +0000)]
- Give ATA/SATA SIMs info about ATAPI packet size, supported by device.
- Make ATA XPT to reject longer SCSI CDBs then supported by device, or
any SCSI CDBs, if device doesn't support ATAPI.

14 years ago- Add myself
gabor [Tue, 2 Feb 2010 10:00:23 +0000 (10:00 +0000)]
- Add myself

Approved by: delphij (mentor)

14 years ago- Add support for CVSup authentication mechanisms to csup.
lulf [Tue, 2 Feb 2010 05:57:42 +0000 (05:57 +0000)]
- Add support for CVSup authentication mechanisms to csup.
- Include a cpasswd script performing the same mechanisms as the cvpasswd
  utility from CVSup.

PR: bin/114129
Submitted by: Petar Zhivkov Petrov <pesho.petrov -at- gmail.com>
MFC after: 1 month

14 years agoEnable MTRR on all VIA CPUs that claim support (amd64).
rnoland [Tue, 2 Feb 2010 01:20:33 +0000 (01:20 +0000)]
Enable MTRR on all VIA CPUs that claim support (amd64).

This is the amd64 part of r203289.

Noticed by: jhb
MFC after: 2 weeks

14 years agoUpdate the version number.
delphij [Tue, 2 Feb 2010 00:30:44 +0000 (00:30 +0000)]
Update the version number.

14 years agoGPIB overhaul, part #2: make the tnt4882 driver work with the newer
joerg [Mon, 1 Feb 2010 21:21:10 +0000 (21:21 +0000)]
GPIB overhaul, part #2: make the tnt4882 driver work with the newer
TNT5004 IC.  This involved a major rewrite of a number of things, as
this chip no longer supports the NAT7210 legacy mode but requires the
host to use the (more modern) FIFO mode.

In theory, this also ought to work on the older TNT4882C chip.  I'll
probably add this as optional support (perhaps by a device.hints flag)
later on.  By now, FIFO mode is *only* activates iff a TNT5004 chip
has been detected (where the old code didn't work at all), while
everything else is supposed to use the old code.

MFC after: 2 weeks

14 years agoAdd xrefs to gpib(3), now that it exists.
joerg [Mon, 1 Feb 2010 21:01:20 +0000 (21:01 +0000)]
Add xrefs to gpib(3), now that it exists.

MFC after: 3 days

14 years agoPCI express device status register has W1C feature. Writing 0 has
yongari [Mon, 1 Feb 2010 20:58:45 +0000 (20:58 +0000)]
PCI express device status register has W1C feature. Writing 0 has
no effect. Make sure to clear error bits by writing 1. [1]
While I'm here use predefined value instead of hardcodig magic
vlaue.

Submitted by: msaitoh at NetBSD [1]

14 years agoAdd a libgpib to the list of known .Lb names.
joerg [Mon, 1 Feb 2010 20:54:45 +0000 (20:54 +0000)]
Add a libgpib to the list of known .Lb names.

MFC after: 3 days

14 years agoFinally, document libgpib.
joerg [Mon, 1 Feb 2010 20:53:55 +0000 (20:53 +0000)]
Finally, document libgpib.

MFC after: 3 days

14 years agoAdd more bit definitions to PCI express device control and device
yongari [Mon, 1 Feb 2010 20:50:49 +0000 (20:50 +0000)]
Add more bit definitions to PCI express device control and device
status register.

Reviewed by: jhb

14 years agoA few minor changes: add altq option header, add missing conditional
jfv [Mon, 1 Feb 2010 19:28:43 +0000 (19:28 +0000)]
A few minor changes:  add altq option header, add missing conditional
around a buf_ring call that will break 7.3, and thanks to Fabien Thomas
add POLLING support for igb and a minor related fix in the em driver.

14 years agoMake PCI Express host controllers functional, by:
marcel [Mon, 1 Feb 2010 17:51:01 +0000 (17:51 +0000)]
Make PCI Express host controllers functional, by:
1.  checking whether there's a link before initializing devices
    on the bus. When there's no link any access onto the bus
    will wedge the CPU.
2.  synthesizing the class & subclass so that the host controller
    appears as a standard PCI bridge, rather than a PowerPC CPU.

14 years agoChange the default value for the flag enabling superpage mapping and
alc [Mon, 1 Feb 2010 17:36:48 +0000 (17:36 +0000)]
Change the default value for the flag enabling superpage mapping and
promotion to "on".

14 years agoUse the capability pointer to indicate whether the host controller is
marcel [Mon, 1 Feb 2010 17:26:58 +0000 (17:26 +0000)]
Use the capability pointer to indicate whether the host controller is
PCI Express, rather than a bit-field (boolean). Saving the capability
pointer this way makes access to capability-specific configuration
registers easy and efficient.

14 years agoConform style.Makefile(5) and replace = and += with ?=, which lets you to
wkoszek [Mon, 1 Feb 2010 16:13:56 +0000 (16:13 +0000)]
Conform style.Makefile(5) and replace = and += with ?=, which lets you to
override WARNS.

Reported by: <marius@nuenneri.ch> and uqs@

14 years agoNetCell is a PCI hardware RAID without cable and mode setting.
mav [Mon, 1 Feb 2010 15:22:22 +0000 (15:22 +0000)]
NetCell is a PCI hardware RAID without cable and mode setting.

14 years agouse u_char instead of u_int for short bitfields.
luigi [Mon, 1 Feb 2010 14:13:44 +0000 (14:13 +0000)]
use u_char instead of u_int for short bitfields.

For our compiler the two constructs are completely equivalent, but
some compilers (including MSC and tcc) use the base type for alignment,
which in the cases touched here result in aligning the bitfields
to 32 bit instead of the 8 bit that is meant here.

Note that almost all other headers where small bitfields
are used have u_int8_t instead of u_int.

MFC after: 3 days

14 years agoManageSieve has been added as port 4190:
ume [Mon, 1 Feb 2010 13:30:06 +0000 (13:30 +0000)]
ManageSieve has been added as port 4190:
http://www.iana.org/assignments/port-numbers

Obtained from: http://www.iana.org/assignments/port-numbers
MFC after: 3 day

14 years agoMerge r178860 from sparc64:
marius [Mon, 1 Feb 2010 12:43:04 +0000 (12:43 +0000)]
Merge r178860 from sparc64:

- Remove the BUS_HANDLE_MIN checking in the __BUS_DEBUG_ACCESS macro;
  for UPA it should have fulfilled its purpose by now and Fireplane-
  and JBus-based machines are way to messy in organization to implement
  something equivalent.
- Fix a bunch of style(9) bugs.

14 years agoMerge r177565 from sparc64:
marius [Mon, 1 Feb 2010 10:18:00 +0000 (10:18 +0000)]
Merge r177565 from sparc64:

- Const'ify the bus_stream_asi and bus_type_asi arrays.
- Replace hard-coded functions names missed in bus_machdep.c with __func__.
- Break some long lines.

14 years agoMerge r157224 from sparc64:
marius [Mon, 1 Feb 2010 09:57:43 +0000 (09:57 +0000)]
Merge r157224 from sparc64:

Sync with the other archs and declare the memory location referenced by
the address argument of the bus_space_write_multi_*() familiy as const.

14 years agoFix bugs:
jasone [Sun, 31 Jan 2010 23:16:10 +0000 (23:16 +0000)]
Fix bugs:

  * Fix a race in chunk_dealloc_dss().

  * Check for allocation failure before zeroing memory in base_calloc().

Merge enhancements from a divergent version of jemalloc:

  * Convert thread-specific caching from magazines to an algorithm that is
    more tunable, and implement incremental GC.

  * Add support for medium size classes, [4KiB..32KiB], 2KiB apart by
    default.

  * Add dirty page tracking for pages within active small/medium object
    runs.  This allows malloc to track precisely which pages are in active
    use, which makes dirty page purging more effective.

  * Base maximum dirty page count on proportion of active memory.

  * Use optional zeroing in arena_chunk_alloc() to avoid needless zeroing
    of chunks.  This is useful in the context of DSS allocation, since a
    long-lived application may commonly recycle chunks.

  * Increase the default chunk size from 1MiB to 4MiB.

Remove feature:

  * Remove the dynamic rebalancing code, since thread caching reduces its
    utility.

14 years agoMake sure we convert audit records that were produced as the result of the
csjp [Sun, 31 Jan 2010 22:31:01 +0000 (22:31 +0000)]
Make sure we convert audit records that were produced as the result of the
closefrom(2) syscall.

14 years agoRemove extra semicolon.
brucec [Sun, 31 Jan 2010 21:47:39 +0000 (21:47 +0000)]
Remove extra semicolon.

Approved by: rrs (mentor)

14 years agoFix typo of ENOTCONN.
brucec [Sun, 31 Jan 2010 21:45:14 +0000 (21:45 +0000)]
Fix typo of ENOTCONN.
Add missing RETURN VALUES section in sctp_opt_info(3).

Approved by: rrs (mentor)

14 years agoHook up the siba_bwn module to the build.
weongyo [Sun, 31 Jan 2010 21:22:45 +0000 (21:22 +0000)]
Hook up the siba_bwn module to the build.

14 years agoAdds siba_bwn module which is used with bwn(4). Main purpose of this
weongyo [Sun, 31 Jan 2010 21:18:22 +0000 (21:18 +0000)]
Adds siba_bwn module which is used with bwn(4).  Main purpose of this
module is to distinguish parts of Silicon Backplane and of Broadcom
Wireless.

14 years agoNote sendmail 8.14.4 upgrade
gshapiro [Sun, 31 Jan 2010 19:56:37 +0000 (19:56 +0000)]
Note sendmail 8.14.4 upgrade

14 years agoThe EFI vendor UUID for FreeBSD variables is called EFI_FREEBSD_VARIABLE
marcel [Sun, 31 Jan 2010 19:47:38 +0000 (19:47 +0000)]
The EFI vendor UUID for FreeBSD variables is called EFI_FREEBSD_VARIABLE
and not EFI_FREEBSD_VENDOR (cf. EFI_GLOBAL_VARIABLE).

14 years agoImplement the "-i" option to sysctl(8), to ignore failures while
gavin [Sun, 31 Jan 2010 19:33:25 +0000 (19:33 +0000)]
Implement the "-i" option to sysctl(8), to ignore failures while
retrieving individual OIDs.  This allows the same list of OIDs to be
passed to sysctl(8) across different systems where particular OIDs may not
exist, and still get as much information as possible from them.

PR: bin/123644
Submitted by: dhw
Approved by: ed (mentor)
MFC after: 2 weeks

14 years agoPatch the experimental NFS client so that there is a timeout
rmacklem [Sun, 31 Jan 2010 19:12:24 +0000 (19:12 +0000)]
Patch the experimental NFS client so that there is a timeout
for negative name cache entries in a manner analogous to
r202767 for the regular NFS client. Also, make the code in
nfs_lookup() compatible with that of the regular client
and replace the sysctl variable that enabled negative name
caching with the mount point option.

MFC after: 2 weeks

14 years agoProperly use dev_refl()/dev_rel() in kern.devname.
ed [Sun, 31 Jan 2010 15:19:16 +0000 (15:19 +0000)]
Properly use dev_refl()/dev_rel() in kern.devname.

While there, perform some clean-up fixes. Update some stale comments on
struct cdev * instead of dev_t and devfs_random(). Also add some missing
whitespace.

MFC after: 1 week

14 years agoPerform some cleanups to devname(3).
ed [Sun, 31 Jan 2010 14:51:04 +0000 (14:51 +0000)]
Perform some cleanups to devname(3).

- Make sure the mode argument is either a character or a block device.
- Use S_IS*() instead of checking S_IF*-flags by hand.
- Don't use kern.devname when the argument is already NODEV.
- Always call snprintf with the proper amount of arguments corresponding
  with the format.
- Perform some whitespace fixes. Tabs instead of 4 spaces, missing space
  for return statement.
- Remove unneeded includes.

14 years agoEnable MTRR on all VIA CPUs that claim support.
rnoland [Sun, 31 Jan 2010 14:35:49 +0000 (14:35 +0000)]
Enable MTRR on all VIA CPUs that claim support.

This may not be entirely correct either, but the existing check is
bogus.  I have both a C3 and a C7 that fail this check, but work fine.

MFC after: 2 weeks

14 years agoWelcome drm support for VIA unichrome chips.
rnoland [Sun, 31 Jan 2010 14:30:39 +0000 (14:30 +0000)]
Welcome drm support for VIA unichrome chips.

MFC after: 2 weeks

14 years agoImport simple drm memory manager.
rnoland [Sun, 31 Jan 2010 14:25:29 +0000 (14:25 +0000)]
Import simple drm memory manager.

This is required for the VIA driver and at least some parts are needed
for GEM.

MFC after: 2 weeks

14 years agoImprove previous commit and add missing ar9280.c
rpaulo [Sun, 31 Jan 2010 14:25:09 +0000 (14:25 +0000)]
Improve previous commit and add missing ar9280.c

14 years agoBuild 928x files too.
rpaulo [Sun, 31 Jan 2010 13:39:08 +0000 (13:39 +0000)]
Build 928x files too.

14 years agoAllow mknod(8) to be only invoked with a pathname as an argument.
ed [Sun, 31 Jan 2010 11:48:25 +0000 (11:48 +0000)]
Allow mknod(8) to be only invoked with a pathname as an argument.

In 99% of the cases people just want to recreate device nodes they
removed from /dev. There is no reason to pass the additional "c 0 0"
anymore.

Also slightly improve the manpage. Remove references to non-existent
device names and platforms.

14 years ago- Check if_type of "addm <interface>" before setting the
hrs [Sun, 31 Jan 2010 08:16:37 +0000 (08:16 +0000)]
- Check if_type of "addm <interface>" before setting the
  interface's MTU to the if_bridge(4) interface.  This fixes a
  bug that MTU value of "addm <interface>" is used even when it
  is invalid for the if_bridge(4) member:

  # ifconfig bridge0 create
  # ifconfig bridge0
  bridge0: flags=8802<BROADCAST,SIMPLEX,MULTICAST> metric 0 mtu 1500
  ...
  # ifconfig bridge0 addm lo0
  ifconfig: BRDGADD lo0: Invalid argument
  # ifconfig bridge0
  bridge0: flags=8802<BROADCAST,SIMPLEX,MULTICAST> metric 0 mtu 16384
  ...

- Do not ignore MTU value of an interface even when if_type == IFT_GIF.
  This fixes MTU mismatch when an if_bridge(4) interface has a
  gif(4) interface and no other interface as the member, and it
  is directly used for L2 communication with EtherIP tunneling
  enabled.

- Implement SIOCSIFMTU ioctl.  Changing the MTU is allowed only
  when all members have the same MTU value.

14 years agoExport the UUID of the partition in the XML. The partition UUID is used
marcel [Sat, 30 Jan 2010 23:13:19 +0000 (23:13 +0000)]
Export the UUID of the partition in the XML. The partition UUID is used
by EFI's device path to identify a partition. In order for FreeBSD to
add EFI boot options, proper device paths need to be constructed.

14 years agoAdd files to remove when MK_GPIB=no.
antoine [Sat, 30 Jan 2010 19:36:55 +0000 (19:36 +0000)]
Add files to remove when MK_GPIB=no.

14 years agoAdd files to remove when MK_GDB=no.
antoine [Sat, 30 Jan 2010 19:33:45 +0000 (19:33 +0000)]
Add files to remove when MK_GDB=no.

14 years agoAdd files to remove when MK_GAMES=no.
antoine [Sat, 30 Jan 2010 19:32:01 +0000 (19:32 +0000)]
Add files to remove when MK_GAMES=no.

14 years agoAdd files to remove when MK_FREEBSD_UPDATE=no.
antoine [Sat, 30 Jan 2010 19:29:02 +0000 (19:29 +0000)]
Add files to remove when MK_FREEBSD_UPDATE=no.

14 years agoAdd files to remove when MK_FLOPPY=no.
antoine [Sat, 30 Jan 2010 19:26:28 +0000 (19:26 +0000)]
Add files to remove when MK_FLOPPY=no.

14 years agoAllow use of -6 option to "server" and "peer" in ntp.conf.
ume [Sat, 30 Jan 2010 16:34:52 +0000 (16:34 +0000)]
Allow use of -6 option to "server" and "peer" in ntp.conf.

MFC after: 1 week

14 years agoImplement handling of the third argument of cpu_switch(). This unbreaks
marius [Sat, 30 Jan 2010 14:04:21 +0000 (14:04 +0000)]
Implement handling of the third argument of cpu_switch(). This unbreaks
sparc64 after r202889.

PR: 143215
MFC after: 1 week

14 years agoSupport -V option to print a dot for each file processed.
kientzle [Sat, 30 Jan 2010 06:49:18 +0000 (06:49 +0000)]
Support -V option to print a dot for each file processed.
Also, change the existing -vi behavior to send the filenames to
stderr rather than stdout, as GNU cpio does.

PR:             bin/141175
Submitted by: Philip Kizer
MFC after: 14 days

14 years agoIntroduce libefi -- a library around EFI runtime services and protocols.
marcel [Sat, 30 Jan 2010 04:24:03 +0000 (04:24 +0000)]
Introduce libefi -- a library around EFI runtime services and protocols.
This first commit brings 3 functions for enumerating, retrieving,
adding, removing and modifying EFI variables. The immediate use of these
include the insertion of a new boot option as part of the installation
process.

This library uses ioctl(2) requests implemented by io(4) to pass the
requests down through the kernel to EFI. These ioctl requests are only
implemented on ia64, so libefi is currently only enabled on ia64. The
interface is generic and io(4) on mad64/i386 can easily be taught to
handle these once EFI support has been added to the kernel there.

14 years agoProvide access to pcpu structures for SMP kernels.
neel [Sat, 30 Jan 2010 01:54:29 +0000 (01:54 +0000)]
Provide access to pcpu structures for SMP kernels.

The basic idea is to use a the same virtual address as a window onto
distinct physical memory locations - one per processor. The physical
address that you access through this mapping depends on which cpu you
are currently executing on. We can now use the same virtual address
on any processor to access its per-cpu area.

The details are:

- The virtual address for 'struct pcpu *pcpup' is obtained by
  stealing 2 pages worth of KVA in pmap_bootstrap().

- The mapping from the constant virtual address to a distinct
  physical page is done in cpu_pcpu_init() through a wired TLB entry.

- A side-effect of this is that we reserve 2 pages worth of memory
  for the pcpu but in reality it needs much less than that. The unused
  memory is now used as the boot stack for the BSP and APs.

Remove SMP-specific bits from locore.S. The plan is to use a separate
mpboot.S for AP bootstrap.

Discussed on: freebsd-mips

Approved by: imp (mentor)

14 years agoFix for kern/141646: when stacking pseudo drivers like
jfv [Sat, 30 Jan 2010 00:11:44 +0000 (00:11 +0000)]
Fix for kern/141646: when stacking pseudo drivers like
lagg and vlan the vlan attach/detach event is not being
handed down to em, this caused some init code not to run,
and thus VLANs did not work. Ultimately having the event
get propagated would be nice, but for now the solution is
to have HWFILTER off by default, when this is the case
VLANs will work, ifconfig can be used to turn it on and
then get HW tag filtering.

14 years agoDon't check the device ID. Instead, check the class, subclass and
marcel [Fri, 29 Jan 2010 20:37:12 +0000 (20:37 +0000)]
Don't check the device ID. Instead, check the class, subclass and
programming I/F. New SoC designs have different device IDs, but
don't need special treatment. Consequently, we fail to probe and
attach for no other reason than not having added the device ID to
the code.

Bank on Freescale's sense of backward compatibility and assume
that if we find a host controller, we know how work with it.

This fixes detection of the PCI Express host controllers on
Freescale's QorIQ family of processors (P1, P2 and P4).

14 years agoThe MAP_ENTRY_NEEDS_COPY flag belongs to protoeflags, cow variable
kib [Fri, 29 Jan 2010 19:25:45 +0000 (19:25 +0000)]
The MAP_ENTRY_NEEDS_COPY flag belongs to protoeflags, cow variable
uses different namespace.

Reported by: Jonathan Anderson <jonathan.anderson cl cam ac uk>
MFC after: 3 days

14 years ago- Fix some style(9) bugs
gabor [Fri, 29 Jan 2010 18:33:59 +0000 (18:33 +0000)]
- Fix some style(9) bugs

Pointed out by: bde

14 years agoWhen backtracing self, start with the current frame (i.e. the
marcel [Fri, 29 Jan 2010 16:14:35 +0000 (16:14 +0000)]
When backtracing self, start with the current frame (i.e. the
frame of db_trace_self()) and not the caller's frame. The use
of builtin_frame_address(1) to get the caller's frame is not
reliable and can cause panics.

14 years agoBump .Dd for r203164.
jh [Fri, 29 Jan 2010 15:11:50 +0000 (15:11 +0000)]
Bump .Dd for r203164.

Approved by: trasz (mentor)

14 years agoAdd myself and show rrs@ as my mentor.
brucec [Fri, 29 Jan 2010 14:44:05 +0000 (14:44 +0000)]
Add myself and show rrs@ as my mentor.

Approved by: rrs (mentor)

14 years agoAdd my birthday.
brucec [Fri, 29 Jan 2010 14:41:16 +0000 (14:41 +0000)]
Add my birthday.

Approved by: rrs (mentor)

14 years agoReset port on disconnect event, to abort any running requests.
mav [Fri, 29 Jan 2010 12:47:04 +0000 (12:47 +0000)]
Reset port on disconnect event, to abort any running requests.

14 years agoAdd "maxfilesize" mount option for tmpfs to allow specifying the
jh [Fri, 29 Jan 2010 12:09:14 +0000 (12:09 +0000)]
Add "maxfilesize" mount option for tmpfs to allow specifying the
maximum file size limit. Default is UINT64_MAX when the option is
not specified. It was useless to set the limit to the total amount of
memory and swap in the system.

Use tmpfs_mem_info() rather than get_swpgtotal() in tmpfs_mount() to
check if there is enough memory available.

Remove now unused get_swpgtotal().

Reviewed by: Gleb Kurtsou
Approved by: trasz (mentor)

14 years agoYou must include fcntl.h (in practice) to be able to do anything useful
rwatson [Fri, 29 Jan 2010 10:32:01 +0000 (10:32 +0000)]
You must include fcntl.h (in practice) to be able to do anything useful
with shm_open(2), as otherwise the O_ flags are undefined.

MFC after: 3 days

14 years agoadd static qualifier to definition of a function already declared static
avg [Fri, 29 Jan 2010 10:20:11 +0000 (10:20 +0000)]
add static qualifier to definition of a function already declared static

This is for improving code readibility only.

MFC after: 1 week

14 years agoAdd support for the AR9285 chipset, which is found on many netbooks
rpaulo [Fri, 29 Jan 2010 10:10:14 +0000 (10:10 +0000)]
Add support for the AR9285 chipset, which is found on many netbooks
available today.

This card is a low power 802.11bgn that only does 11n rates up to MCS 7
(that's 65 Mbps in 20Mhz mode and 135 in 40Mhz mode).
802.11n is  not yet supported, but will be in the future.

The driver still has a problem regarding to the setting of txpower on
the card, so don't expect good performance yet. After fixing this
problem, an MFC is possible.

Special thanks to iXsystems and S Smirnov <tonve at yandex.ru> for help
with the purchase of a netbook with this card.

Sponsored by: iXsystems, Inc.

14 years agoReplace Id keyword with the FreeBSD keyword.
rpaulo [Fri, 29 Jan 2010 10:07:17 +0000 (10:07 +0000)]
Replace Id keyword with the FreeBSD keyword.

14 years ago- Handle short reads when the -P option is used. Short reads must be
jh [Fri, 29 Jan 2010 10:04:00 +0000 (10:04 +0000)]
- Handle short reads when the -P option is used. Short reads must be
  handled when reading from pipes.
- Remove dead code related to the -P option from getvol(). pipein and
  pipecmdin are never set at the same time.

PR: bin/121502
Approved by: trasz (mentor)
MFC after: 2 weeks

14 years agoReplace Id keyword with the FreeBSD keyword.
rpaulo [Fri, 29 Jan 2010 10:02:50 +0000 (10:02 +0000)]
Replace Id keyword with the FreeBSD keyword.

14 years ago- Cast time_t, int64_t and some int32_t values to intmax_t and use "%jd"
jh [Fri, 29 Jan 2010 10:00:42 +0000 (10:00 +0000)]
- Cast time_t, int64_t and some int32_t values to intmax_t and use "%jd"
  in format strings.
- Use (void) instead of (void *) when discarding strcat(3) return value.
- Format string fixes to match variable types.
- Change canon() len parameter and getcmd() size parameter type from
  int to size_t.
- Style Makefile and increase WARNS to 2.

PR: bin/140061
Submitted by: uqs
Approved by: trasz (mentor)

14 years agoFollow Neel's suggestion and switch to using
rrs [Fri, 29 Jan 2010 05:38:41 +0000 (05:38 +0000)]
Follow Neel's suggestion and switch to using
restoreint() in combination with saving off the
old level. That way we don't blast out the old
level.

14 years agoFor our memory re-mapping trick to work
rrs [Fri, 29 Jan 2010 04:07:38 +0000 (04:07 +0000)]
For our memory re-mapping trick to work
interrupts must be disabled through the
page_zero's or copys etc. Note that the
temporary mapping used by panic's may
cause us pain since int's may not be disabled.
When we get dumps working we may have to revist
this. Note that with this fix the build got
much much further.. until it hung on disk IO (I
would imagine thats the rge/msgring driver acting
up).

14 years agoIts possible that our RMI box has memory extending
rrs [Fri, 29 Jan 2010 04:05:17 +0000 (04:05 +0000)]
Its possible that our RMI box has memory extending
above 4Gig. If so when we add the base address with
the size we will wrap. So for now we just ignore
such memory and only use what we can. When we
get 64 bit working then we will be much better ;->

14 years agoMove ID up into comment block.. per bsdimp
rrs [Fri, 29 Jan 2010 04:03:36 +0000 (04:03 +0000)]
Move ID up into comment block.. per bsdimp

14 years agoUuencode the rt2870 firmware into ascii like the other firmware blobs.
thompsa [Fri, 29 Jan 2010 03:35:01 +0000 (03:35 +0000)]
Uuencode the rt2870 firmware into ascii like the other firmware blobs.

14 years agoAdd a function to check if the usb devices is still connected.
thompsa [Fri, 29 Jan 2010 02:44:06 +0000 (02:44 +0000)]
Add a function to check if the usb devices is still connected.

Submitted by: Hans Petter Selasky

14 years agoRework cam error handling to fix Mitsumi floppy drives.
thompsa [Fri, 29 Jan 2010 02:38:34 +0000 (02:38 +0000)]
Rework cam error handling to fix Mitsumi floppy drives.

Submitted by: mav

14 years agoSimplify attach for UMASS_PROTO_CBI_I mode and change some switch() returns
thompsa [Fri, 29 Jan 2010 02:35:50 +0000 (02:35 +0000)]
Simplify attach for UMASS_PROTO_CBI_I mode and change some switch() returns
into breaks.

Submitted by: Hans Petter Selesky

14 years agoAdd null check on quirk lookup and add a couple of umass quirks.
thompsa [Fri, 29 Jan 2010 02:33:50 +0000 (02:33 +0000)]
Add null check on quirk lookup and add a couple of umass quirks.

Submitted by: Hans Petter Selesky

14 years agoAdd the Netgear WPN111
thompsa [Fri, 29 Jan 2010 02:32:46 +0000 (02:32 +0000)]
Add the Netgear WPN111

14 years agoSync usb products to perforce.
thompsa [Fri, 29 Jan 2010 02:32:18 +0000 (02:32 +0000)]
Sync usb products to perforce.

14 years agoAttempt to recover on a TX error rather than stopping all transfers.
thompsa [Fri, 29 Jan 2010 02:31:33 +0000 (02:31 +0000)]
Attempt to recover on a TX error rather than stopping all transfers.

Submitted by: Hans Petter Selesky

14 years agoOptimise EHCI ISOC HS done check.
thompsa [Fri, 29 Jan 2010 01:30:13 +0000 (01:30 +0000)]
Optimise EHCI ISOC HS done check.

Submitted by: Hans Petter Selasky

14 years agoAdd device ID.
thompsa [Fri, 29 Jan 2010 01:27:54 +0000 (01:27 +0000)]
Add device ID.

PR: usb/142427

14 years agoUse device_printf rather than printf + device_get_nameunit.
thompsa [Thu, 28 Jan 2010 22:54:01 +0000 (22:54 +0000)]
Use device_printf rather than printf + device_get_nameunit.

14 years agoRelease the firmware after loading to the device.
thompsa [Thu, 28 Jan 2010 22:46:04 +0000 (22:46 +0000)]
Release the firmware after loading to the device.

14 years agoAdd run(4) to NOTES.
thompsa [Thu, 28 Jan 2010 22:28:59 +0000 (22:28 +0000)]
Add run(4) to NOTES.

14 years agoHook run(4) to the build.
thompsa [Thu, 28 Jan 2010 22:28:01 +0000 (22:28 +0000)]
Hook run(4) to the build.

14 years agoAdd run(4), a driver for Ralink RT2700U/RT2800U/RT3000U USB 802.11agn devices.
thompsa [Thu, 28 Jan 2010 22:24:54 +0000 (22:24 +0000)]
Add run(4), a driver for Ralink RT2700U/RT2800U/RT3000U USB 802.11agn devices.

This driver was written for OpenBSD by Damien Bergamini and ported over by
Akinori Furukoshi.

14 years ago- Increase timeouts to 100 milliseconds, 1 millisecond is definitely not
gonzo [Thu, 28 Jan 2010 21:55:56 +0000 (21:55 +0000)]
- Increase timeouts to 100 milliseconds, 1 millisecond is definitely not
    enough for PCI controller to get into shape

Thanks to: adrian@