]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
17 years ago- Add handling of MXGEFW_CMD_UNKNOWN in mxge_send_cmd().
gallatin [Tue, 8 May 2007 18:45:43 +0000 (18:45 +0000)]
- Add handling of MXGEFW_CMD_UNKNOWN in mxge_send_cmd().
- Convert mxge_send_cmd result handling to a switch rather
  than adding a new elseif for MXGEFW_CMD_UNKNOWN

Sponsored by: Myricom Inc.

17 years ago- Copyright change, cisco's silly tool wants it to say:
rrs [Tue, 8 May 2007 17:01:12 +0000 (17:01 +0000)]
-  Copyright change, cisco's silly tool wants it to say:
       "Copyright (c) 2001-2007, by Cisco Systems,"
   instead of
       *Copyright (c) 2001-2007, Cisco Systems,"

-  Also fix a few straglers that were still in 2006.

17 years ago- Get rid of the sctp_inpcb_free() "magic numbers", now they
rrs [Tue, 8 May 2007 15:53:03 +0000 (15:53 +0000)]
-  Get rid of the sctp_inpcb_free() "magic numbers", now they
   are sensible defines that tell what you are directing
   the function to do.

17 years agoSet the program return value non-zero if we enounter
kientzle [Tue, 8 May 2007 15:22:21 +0000 (15:22 +0000)]
Set the program return value non-zero if we enounter
an error while skipping the body of an entry during
"tar -t" processing.

Thanks to: Spencer Minear
PR: misc/112521

17 years ago- Static analyisis fixes for cisco's commit (this is equivilant
rrs [Tue, 8 May 2007 14:32:53 +0000 (14:32 +0000)]
-  Static analyisis fixes for cisco's commit (this is equivilant
   to the coverity tool.. may even be the same one.. not sure).
-  A bug in the way sctp_abort() and friends were
   setting the IP_CLOSE flag.. and NOT passing the
   last argument as a (,1)... so that things would
   get freed..

17 years agoSet the debug.acpi.acpi_ca_version sysctl even if ACPI support is not
marks [Tue, 8 May 2007 14:31:49 +0000 (14:31 +0000)]
Set the debug.acpi.acpi_ca_version sysctl even if ACPI support is not
available.

17 years agoFirmware update & improvements to firmware selection:
gallatin [Tue, 8 May 2007 14:19:43 +0000 (14:19 +0000)]
Firmware update & improvements to firmware selection:

- Update to latest (1.4.17) firmware.

- Use the new MXGEFW_CMD_UNALIGNED_TEST (added in firmare 1.4.16) to
  have the firmware tell us if the PCIe chipset supports aligned PCIe
  completions.

- Hard to maintain, and frequently out of date whitelist of PCIe
  chipsets known to produce aligned completions removed, as it has been
  replaced in its role of selecting the correct firmware to run by the
  use of MXGEFW_CMD_UNALIGNED_TEST.

- Break the dma test out of mxge_reset() and into its own function
  (mxge_dma_test()) so it can be used by both the normal DMA test, and
  to run the unaligned test.

- Improved support for enabling ECRCs

Sponsored by: Myricom Inc.

17 years agoAdd missing socket buffer unlock before returning to userland.
yongari [Tue, 8 May 2007 12:34:14 +0000 (12:34 +0000)]
Add missing socket buffer unlock before returning to userland.

Reviewed by:    rwatson

17 years agoAdd .Li mark-up to Hesiod, NIS, and nsswitch keywords so that they stand out
yar [Tue, 8 May 2007 11:00:07 +0000 (11:00 +0000)]
Add .Li mark-up to Hesiod, NIS, and nsswitch keywords so that they stand out
when the manpage is read via an advanced output device.

17 years agoAdd nsswitch.conf(5) xref to SEE ALSO as it's mentioned in the text.
yar [Tue, 8 May 2007 10:48:01 +0000 (10:48 +0000)]
Add nsswitch.conf(5) xref to SEE ALSO as it's mentioned in the text.

17 years agoMark up text consistently.
yar [Tue, 8 May 2007 10:45:28 +0000 (10:45 +0000)]
Mark up text consistently.

17 years agoDocument the current practice of locking and disabling accounts
yar [Tue, 8 May 2007 10:22:45 +0000 (10:22 +0000)]
Document the current practice of locking and disabling accounts
through passwd(5).

17 years agoDo not set address early. It fix problems when SET_ADDR_FAILED happens
kevlo [Tue, 8 May 2007 03:25:05 +0000 (03:25 +0000)]
Do not set address early. It fix problems when SET_ADDR_FAILED happens
on the hubs.

Reviewed by: imp

17 years agoWe should never zero-pad INF or NaN (yielding silly strings like "00inf")
das [Tue, 8 May 2007 03:08:28 +0000 (03:08 +0000)]
We should never zero-pad INF or NaN (yielding silly strings like "00inf")
even if the programmer asks for zero padding.

17 years agoFix a rounding error.
das [Tue, 8 May 2007 02:59:37 +0000 (02:59 +0000)]
Fix a rounding error.

Reported by: Bruno Haible

17 years ago- More macros for OS compatabilty
rrs [Tue, 8 May 2007 00:21:05 +0000 (00:21 +0000)]
-  More macros for OS compatabilty
-  PR-SCTP would ignore FWD-TSN's above a rwnd's worth
   of TSN's (1 byte msgs).. this left the peer hopelessly
   out of sync.. or an attacker. So now we abort the assoc.
-  New IFN hash, also rename hashes to match addr/ifn now
   that the vrf has multiple.
-  Do not enable SCTP_PCB_FLAGS_RECVDATAIOEVNT per default
   as defined in the Socket API ID.
-  Export MTU information via sysctl.
-  Vrf's need table id's. This is default for
   BSD, but may be other things later when BSD
   fully supports VRFs.
-  Additional stream reset bug (caught by cisco dev-test).
-  Additional validations for the address in sending a message (socket api).
-------- and -----
-  Fix association notifications not to give the active open
   side false notifications.
-  Fix so sendfile and SENDALL will work properly (missing
   flag to say socket sender is done).
-  Fix Bug that prevented COOKIES from being retransmitted.
-  Break out connectx into helper sub-models so that iox routines can
   reuse the helpers.
-  When an address is added during system init (non-dynamic mode) make
   sure that the "defer use" flag is not set.
** its compiling on XR now :-D **

Reviewed by: gnn

17 years agoFix a typo in a bootverbose printf.
jhb [Mon, 7 May 2007 18:29:37 +0000 (18:29 +0000)]
Fix a typo in a bootverbose printf.

MFC after: 3 days
Submitted by: yongari

17 years agoRather than selectively zeroing fields in the tcp_debug structure
rwatson [Mon, 7 May 2007 14:05:23 +0000 (14:05 +0000)]
Rather than selectively zeroing fields in the tcp_debug structure
throughout tcp_trace(), zero the entire structure up front.

Minor style fixes.

17 years agoSince udp_peeraddr() and udp_sockaddr() directly wrap in_setpeeraddr()
rwatson [Mon, 7 May 2007 13:51:24 +0000 (13:51 +0000)]
Since udp_peeraddr() and udp_sockaddr() directly wrap in_setpeeraddr()
and in_setsockaddr(), containing only stale comments on why they
exist, remove them and initialize the protosw for UDP to directly
reference in_setpeeraddr() and in_setsockaddr().

17 years agoMinor style tweaks.
rwatson [Mon, 7 May 2007 13:47:39 +0000 (13:47 +0000)]
Minor style tweaks.

17 years agoWhen setting up timewait state for a TCP connection, don't hold the
rwatson [Mon, 7 May 2007 13:04:25 +0000 (13:04 +0000)]
When setting up timewait state for a TCP connection, don't hold the
socket lock over a crhold() of so_cred: so_cred is constant after
socket creation, so doesn't require locking to read.

17 years agoCast to intmax_t to print tv_sec in struct timeval.
dwmalone [Mon, 7 May 2007 12:23:23 +0000 (12:23 +0000)]
Cast to intmax_t to print tv_sec in struct timeval.
Avoid shadowing a global variable with a function argument.
Set WARNS to 6.

17 years agoUse [u]intmax_t to printf unknown integer types.
dwmalone [Mon, 7 May 2007 12:10:06 +0000 (12:10 +0000)]
Use [u]intmax_t to printf unknown integer types.
Make timeprt always allocate memory, to make things a little simpler.
Remove an unused variable.

17 years agoMake logmsg take const char * arguments.
dwmalone [Mon, 7 May 2007 11:28:01 +0000 (11:28 +0000)]
Make logmsg take const char * arguments.
Set WARNS to 6.

17 years agoANSIfy function definitions.
dwmalone [Mon, 7 May 2007 11:18:30 +0000 (11:18 +0000)]
ANSIfy function definitions.
Move some extern declarations to a header file.
Remove an unneeded extern declaration for optarg.

17 years agoFix some warnings by making things const, adding missing headers,
dwmalone [Mon, 7 May 2007 11:01:36 +0000 (11:01 +0000)]
Fix some warnings by making things const, adding missing headers,
removing some unused variables and making a variable unsigned.

MFC after: 3 weeks

17 years agoFix kldloading if_em by compiling missing source file.
le [Mon, 7 May 2007 09:55:25 +0000 (09:55 +0000)]
Fix kldloading if_em by compiling missing source file.

17 years ago - Correctly check if lp_ioctl is null
thompsa [Mon, 7 May 2007 09:53:02 +0000 (09:53 +0000)]
 - Correctly check if lp_ioctl is null
 - Remove lagg_ether_purgemulti as its no longer needed
 - Mark the interface as up if any ports are active rather than just the primary

17 years agoTell that nve(4) has got support for VLAN_MTU.
yar [Mon, 7 May 2007 09:52:39 +0000 (09:52 +0000)]
Tell that nve(4) has got support for VLAN_MTU.
Touch Dd.

17 years agoFinish VLAN_MTU support in nve(4). The interface appears to be able
yar [Mon, 7 May 2007 09:45:31 +0000 (09:45 +0000)]
Finish VLAN_MTU support in nve(4).  The interface appears to be able
to handle long frames in its default mode, so setting the respective
bit in if_capenable is enough.

Tested by: yongari

17 years agobuf_addr should be bus_addr_t rather than u_int32_t.
ariff [Mon, 7 May 2007 02:46:48 +0000 (02:46 +0000)]
buf_addr should be bus_addr_t rather than u_int32_t.

17 years agoAdd rum(4)
kevlo [Mon, 7 May 2007 02:06:03 +0000 (02:06 +0000)]
Add rum(4)

17 years agoThe purgemulti call is not needed since all the ports have already been detached.
thompsa [Mon, 7 May 2007 00:52:26 +0000 (00:52 +0000)]
The purgemulti call is not needed since all the ports have already been detached.

17 years agoCall if_setlladdr() on the aggregation port from a taskqueue so the softc lock
thompsa [Mon, 7 May 2007 00:35:15 +0000 (00:35 +0000)]
Call if_setlladdr() on the aggregation port from a taskqueue so the softc lock
is not held. The short delay between aggregating the port and setting the MAC
address is fine.

17 years agoAvoid touching various unsafe parts if the interface is disappearing.
thompsa [Mon, 7 May 2007 00:28:55 +0000 (00:28 +0000)]
Avoid touching various unsafe parts if the interface is disappearing.

17 years agoChange from using if_delmulti() to if_delmulti_ifma() as it simplifies the code
thompsa [Mon, 7 May 2007 00:18:56 +0000 (00:18 +0000)]
Change from using if_delmulti() to if_delmulti_ifma() as it simplifies the code
and is safe to use if the ifp has disappeared.

Suggested by: bms

17 years agoUnbreak the build on 64-bit architectures as well.
des [Sun, 6 May 2007 21:06:08 +0000 (21:06 +0000)]
Unbreak the build on 64-bit architectures as well.

17 years ago- Add missing lock destruction and remove duplicate initializations.
pjd [Sun, 6 May 2007 19:05:37 +0000 (19:05 +0000)]
- Add missing lock destruction and remove duplicate initializations.
  With this change it is possible to unload zfs.ko module from
  WITNESS-enabled kernel.
- Remove bogus comment.

17 years agoAdd Apple HFS+ (0xAF) MBR partition type identificator.
andre [Sun, 6 May 2007 18:48:30 +0000 (18:48 +0000)]
Add Apple HFS+ (0xAF) MBR partition type identificator.

Submitted by: rpaulo (SoC2007 student)

17 years agoDocument the additional fields, si_drv1 and si_drv2 that are available
jmg [Sun, 6 May 2007 18:20:16 +0000 (18:20 +0000)]
Document the additional fields, si_drv1 and si_drv2 that are available
for the user to use...

MFC after: 1 week

17 years agoMake interrupt filtering support compilable.
piso [Sun, 6 May 2007 17:04:34 +0000 (17:04 +0000)]
Make interrupt filtering support compilable.

The entire code is wrapperd in #ifdef ... #endif so it won't harm
the actual implementation, but developers are encouraged to test it.
For arm, ia64, ppc, sparc64 and sun4v some work is still
needed, thus arch maintainers are encouraged to bring their arch on par
with respect to i386 and amd64.

Approved by: re (implicit?)

17 years agoBring in the reminaing bits to make interrupt filtering work:
piso [Sun, 6 May 2007 17:02:50 +0000 (17:02 +0000)]
Bring in the reminaing bits to make interrupt filtering work:

o push much of the i386 and amd64 MD interrupt handling code
  (intr_machdep.c::intr_execute_handlers()) into MI code
  (kern_intr.c::ithread_loop())
o move filter handling to kern_intr.c::intr_filter_loop()
o factor out the code necessary to mask and ack an interrupt event
  (intr_machdep.c::intr_eoi_src() and intr_machdep.c::intr_disab_eoi_src()),
  and make them part of 'struct intr_event', passing them as arguments to
  kern_intr.c::intr_event_create().
o spawn a private ithread per handler (struct intr_handler::ih_thread)
  with filter and ithread functions.

Approved by: re (implicit?)

17 years agoPCMMKMINOR() bit twiddling, so we can have:
ariff [Sun, 6 May 2007 16:46:23 +0000 (16:46 +0000)]
PCMMKMINOR() bit twiddling, so we can have:
  - 2048 unique sound cards (dsp0 -> dsp2047)
    (used to be 15)
  - 32 unique device types  (dspW, audio, mixer, sndstat, .. 32th)
    (used to be 15)
  -  256 unique cloneable devices (dsp%d.0 -> dsp%d.255)
    (unchanged)

17 years agoRemove unused requested_s_scale from struct tcpcb.
andre [Sun, 6 May 2007 16:04:36 +0000 (16:04 +0000)]
Remove unused requested_s_scale from struct tcpcb.

17 years agoUse existing TF_SACK_PERMIT flag in struct tcpcb t_flags field instead of
andre [Sun, 6 May 2007 15:56:31 +0000 (15:56 +0000)]
Use existing TF_SACK_PERMIT flag in struct tcpcb t_flags field instead of
a decdicated sack_enable int for this bool.  Change all users accordingly.

17 years ago o Remove redundant tcp reassembly check in header prediction code
andre [Sun, 6 May 2007 15:41:06 +0000 (15:41 +0000)]
 o Remove redundant tcp reassembly check in header prediction code
 o Rearrange code to make intent in TCPS_SYN_SENT case more clear
 o Assorted style cleanup
 o Comment clarification for tcp_dropwithreset()

17 years agoReorder the TCP header prediction test to check for the most volatile
andre [Sun, 6 May 2007 15:23:51 +0000 (15:23 +0000)]
Reorder the TCP header prediction test to check for the most volatile
values first to spend less time on a fallback to normal processing.

17 years agoRemove the defunct remains of the TCPS_TIME_WAIT cases from tcp_do_segment
andre [Sun, 6 May 2007 15:16:05 +0000 (15:16 +0000)]
Remove the defunct remains of the TCPS_TIME_WAIT cases from tcp_do_segment
and change it to a void function.

We use a compressed structure for TCPS_TIME_WAIT to save memory.  Any late
late segments arriving for such a connection is handled directly in the TW
code.

17 years agoWhen deleting key, flush write cache after each overwrite, so we don't
pjd [Sun, 6 May 2007 14:56:03 +0000 (14:56 +0000)]
When deleting key, flush write cache after each overwrite, so we don't
overwrite data N times in cache and only once on disk.

17 years agoCorrect some typos.
pjd [Sun, 6 May 2007 14:54:41 +0000 (14:54 +0000)]
Correct some typos.

17 years agoAdd a regression test to detect if waiting on the I/O serialization lock
rwatson [Sun, 6 May 2007 14:28:49 +0000 (14:28 +0000)]
Add a regression test to detect if waiting on the I/O serialization lock
on socket buffers is interruptible or not, which detacts the regression I
introduced recently in 7-CURRENT (spotted by alfred).  This test passes
in older -CURRENT, and with the as-yet uncommitted sx_xlock_sig and
sblock fix patches.

17 years agoUnbreak the build.
des [Sun, 6 May 2007 14:25:52 +0000 (14:25 +0000)]
Unbreak the build.

17 years agoFix two comments.
andre [Sun, 6 May 2007 13:38:25 +0000 (13:38 +0000)]
Fix two comments.

17 years agoAdd missing 'bp', since it should be passed as g_print_bio() argument.
wkoszek [Sun, 6 May 2007 12:48:39 +0000 (12:48 +0000)]
Add missing 'bp', since it should be passed as g_print_bio() argument.

Oked by: pjd

17 years agoDon't acquire Giant unconditionally.
wkoszek [Sun, 6 May 2007 12:00:38 +0000 (12:00 +0000)]
Don't acquire Giant unconditionally.

Reviewed by: rwatson

17 years agoAdd support for Ralink Technology RT2501USB/RT2601USB devices.
kevlo [Sun, 6 May 2007 10:07:21 +0000 (10:07 +0000)]
Add support for Ralink Technology RT2501USB/RT2601USB devices.

Reviewed by: sam, sephe
Obtained from: OpenBSD

17 years agoRemove parameter names from prototypes to avoid namespace issues.
des [Sun, 6 May 2007 10:00:27 +0000 (10:00 +0000)]
Remove parameter names from prototypes to avoid namespace issues.

17 years agoUse provider's ident to handle situations when disks are moved around
pjd [Sun, 6 May 2007 01:39:39 +0000 (01:39 +0000)]
Use provider's ident to handle situations when disks are moved around
and show up with different names: first try to open provider using
remembered name and compare its ident, if equal, this is our provider,
if not equal or there is no provider with such name, find provider with
remembered ident and don't care about the name.

17 years agoMFp4: We don't need to cover vnode_pager_setsize() with the z_map_lock.
pjd [Sun, 6 May 2007 01:27:54 +0000 (01:27 +0000)]
MFp4: We don't need to cover vnode_pager_setsize() with the z_map_lock.

17 years agoClean up a little.
pjd [Sun, 6 May 2007 01:23:03 +0000 (01:23 +0000)]
Clean up a little.

Obtained from: OpenSolaris

17 years agoFile d_ident field with disk's serial number.
pjd [Sun, 6 May 2007 01:20:06 +0000 (01:20 +0000)]
File d_ident field with disk's serial number.

No answer from: sos

17 years agoImplement and document some utility functions that can be used to communicate
pjd [Sun, 6 May 2007 01:17:46 +0000 (01:17 +0000)]
Implement and document some utility functions that can be used to communicate
with GEOM providers.

OK'ed by: phk

17 years agoFix mdoc warnings.
pjd [Sun, 6 May 2007 00:31:38 +0000 (00:31 +0000)]
Fix mdoc warnings.

17 years agoPrint provider's ident when in verbose mode.
pjd [Sun, 6 May 2007 00:25:21 +0000 (00:25 +0000)]
Print provider's ident when in verbose mode.

17 years agoTwo bugs:
rrs [Sun, 6 May 2007 00:01:17 +0000 (00:01 +0000)]
Two bugs:
  - Locks were not being unlocked when an invalid size chunk is
    sent in.
  - When a notification comes in, we cannot use it to look up
    the fragment interleave stream information since its not
    on a stream.

17 years agoMake this driver MP safe and still be a multi-release driver.
mjacob [Sat, 5 May 2007 20:18:24 +0000 (20:18 +0000)]
Make this driver MP safe and still be a multi-release driver.

Obtained from: 99% of the work done by Scott Long.
MFC after: 3 days

17 years agoMake this an MP safe driver but also still be multi-release.
mjacob [Sat, 5 May 2007 20:17:23 +0000 (20:17 +0000)]
Make this an MP safe driver but also still be multi-release.
Seems to work on RELENG_4 through -current and also on sparc64
now. There may still be some issues with the auto attach/detach
code to sort out.

MFC after: 3 days

17 years agoDefine every architecture as either VM_PHYSSEG_DENSE or
alc [Sat, 5 May 2007 19:50:28 +0000 (19:50 +0000)]
Define every architecture as either VM_PHYSSEG_DENSE or
VM_PHYSSEG_SPARSE depending on whether the physical address space is
densely or sparsely populated with memory.  The effect of this
definition is to determine which of two implementations of
vm_page_array and PHYS_TO_VM_PAGE() is used.  The legacy
implementation is obtained by defining VM_PHYSSEG_DENSE, and a new
implementation that trades off time for space is obtained by defining
VM_PHYSSEG_SPARSE.  For now, all architectures except for ia64 and
sparc64 define VM_PHYSSEG_DENSE.  Defining VM_PHYSSEG_SPARSE on ia64
allows the entirety of my Itanium 2's memory to be used.  Previously,
only the first 1 GB could be used.  Defining VM_PHYSSEG_SPARSE on
sparc64 allows USIIIi-based systems to boot without crashing.

This change is a combination of Nathan Whitehorn's patch and my own
work in perforce.

Discussed with: kmacy, marius, Nathan Whitehorn
PR: 112194

17 years agoAllow to use ':' in d_ident, which is quite handy character.
pjd [Sat, 5 May 2007 18:09:17 +0000 (18:09 +0000)]
Allow to use ':' in d_ident, which is quite handy character.

17 years agoHandle GEOM::ident attribute by attaching 'sX' string at the end of ident
pjd [Sat, 5 May 2007 17:52:22 +0000 (17:52 +0000)]
Handle GEOM::ident attribute by attaching 'sX' string at the end of ident
received from the underlying provider, where X is pp->index value.

OK'ed by: phk

17 years agoBecause there are many strange hardware out there, allow to use only
pjd [Sat, 5 May 2007 17:47:20 +0000 (17:47 +0000)]
Because there are many strange hardware out there, allow to use only
[a-zA-Z0-9-_@#%.] characters in d_ident field.

17 years agofixup talk of kern.maxswzone... It's been 32MB for almost 5 years now...
jmg [Sat, 5 May 2007 17:36:42 +0000 (17:36 +0000)]
fixup talk of kern.maxswzone...  It's been 32MB for almost 5 years now...
and only supports just over 7GB of swap...

Sound a bit more professional..

Inspired by: Marc G. Fournier
MFC After: 3 days

17 years ago- Extend disk structure to allow to store disk's serial number, which can be
pjd [Sat, 5 May 2007 17:12:15 +0000 (17:12 +0000)]
- Extend disk structure to allow to store disk's serial number, which can be
  retrieved via GEOM::ident attribute.
- Bump disk(9) ABI version.

OK'ed by: phk

17 years agoImplement three new ioctls that can be used with GEOM provider:
pjd [Sat, 5 May 2007 17:02:19 +0000 (17:02 +0000)]
Implement three new ioctls that can be used with GEOM provider:

DIOCGFLUSH - Flush write cache (sends BIO_FLUSH).

DIOCGDELETE - Delete data (mark as unused) (sends BIO_DELETE).

DIOCGIDENT - Get provider's uniqe and fixed identifier (asks for
GEOM::ident attribute).

First two are self-explanatory, but the last one might not be. Here are
properties of provider's ident:

- ident value is preserved between reboots,
- provider can be detached/attached and ident is preserved,
- provider's name can change - ident can't,
- ident value should not be based on on-disk metadata; in other words
  copying whole data from one disk to another should not yield the same
  ident for the other disk,
- there could be more than one provider with the same ident, but only if
  they point at exactly the same physical storage, this is the case for
  multipathing for example,
- GEOM classes that consumes single providers and provide single providers,
  like geli, gbde, should just attach class name to the ident of the
  underlying provider,
- ident is an ASCII string (is printable),
- ident is optional and applications can't relay on its presence.

The main purpose for this is that application and remember provider's ident
and once it tries to open provider by its name again, it may compare idents
to be sure this is the right provider. If it is not (idents don't match),
then it can open provider by its ident.

OK'ed by: phk

17 years agoImplement g_delete_data() similar to g_read_data() and g_write_data().
pjd [Sat, 5 May 2007 16:35:22 +0000 (16:35 +0000)]
Implement g_delete_data() similar to g_read_data() and g_write_data().

OK'ed by: phk

17 years ago- Implement helper g_handleattr_str() function for string attributes
pjd [Sat, 5 May 2007 16:33:44 +0000 (16:33 +0000)]
- Implement helper g_handleattr_str() function for string attributes
  handling.
- Extend g_handleattr() to treat attribute as string when len=0.

OK'ed by: phk

17 years ago- Nuke unnecessary header.
sephe [Sat, 5 May 2007 11:07:52 +0000 (11:07 +0000)]
- Nuke unnecessary header.
- Make wlan_amrr depend on wlan, so that it can find various symbols in
  wlan module if wlan is not compiled into kernel.

Approved by: sam (mentor)
Tested by: kevlo

17 years agoForgot to bump .Dd
joel [Sat, 5 May 2007 10:36:57 +0000 (10:36 +0000)]
Forgot to bump .Dd

17 years agoDocument more quirks and supported codecs.
joel [Sat, 5 May 2007 10:33:28 +0000 (10:33 +0000)]
Document more quirks and supported codecs.

17 years agoFix (or rather workaround) Intel 440MX Errata #36
ariff [Sat, 5 May 2007 09:18:05 +0000 (09:18 +0000)]
Fix (or rather workaround) Intel 440MX Errata #36
- http://www.intel.com/design/chipsets/specupdt/245051.htm

AC97 Soft Audio and Soft Modem Master Abort Errata

Issue:
  Use of either soft audio or soft modem on an IntelĀ® 82443MX PCISet
  based platform running a 100 MHz Processor System Bus and an AC97 codec
  may result in failures. The system continues to function normally while
  the AC97 hardware may not resume and may require a cold-boot to
  recover. As a result of the failure, the Master Abort Status bit will
  be set in the audio or modem function PCI header space.

Workaround:
  Force uncacheable DMA on both BDL and pcm buffers.

Tested by: Emil Holmstr|m <emil@linux.se>

17 years agoMiscellaneous changes and fix:
ariff [Sat, 5 May 2007 09:17:36 +0000 (09:17 +0000)]
Miscellaneous changes and fix:

- Remove explicit call to pmap_change_attr(), since we now have proper
  and functional definition of BUS_DMA_NOCACHE.
- Enable PCI(e) bus snooping for non i386/amd64 as an alternative for
  uncacheable DMA.
- Codecs changes:
  * Analag Device -> Analog Devices, AD1988.
  * New codec: VIA VT1708 and VT1709, Realtek ALC262, ALC861-VD and
    ALC885.
  * Various fixups for Conexant Waikiki, fix recording (read: microphone)
    on various Analog Devices codecs due to vendor BIOS mess, various
    quirks for several ASUS laptops/boards.
- Fix connection list handling, closely following the specification to
  handle range of nids.
- Basic Jack sense polling infrastructure for possible hardwares with
  broken unsolicited response interrupt.

Ideas/Submitted/Tested by: Andriy Gapon <avg@icyb.net.ua>,
                           #freebsd-azalia, many.

17 years agoFix a TLS memory leak.
davidxu [Sat, 5 May 2007 08:44:59 +0000 (08:44 +0000)]
Fix a TLS memory leak.

PR: threads/112297
MFC: 1 week

17 years agoAdd Prologix GPIB-USB Controller (Highly recommended btw)
phk [Sat, 5 May 2007 08:23:05 +0000 (08:23 +0000)]
Add Prologix GPIB-USB Controller (Highly recommended btw)

17 years agoDocument devices that can drivea
takawata [Sat, 5 May 2007 08:10:37 +0000 (08:10 +0000)]
Document devices that can drivea
.
Obtained from: [FreeBSD-users-jp 90628](uftdi.c rev 1.25 also)

17 years agosome minor modification to the previous commit to sys/netinet6/nd6.c and nd6_nbr.c.
suz [Sat, 5 May 2007 04:24:01 +0000 (04:24 +0000)]
some minor modification to the previous commit to sys/netinet6/nd6.c and nd6_nbr.c.
- added some clarification comments
- removed an unnecesary code

Obtained from: KAME
MFC after: 1 week

17 years agoAdd global mutex tcp_debug_mtx, which will protect global TCP debugging
rwatson [Fri, 4 May 2007 23:43:18 +0000 (23:43 +0000)]
Add global mutex tcp_debug_mtx, which will protect global TCP debugging
state tcp_debug, tcp_debx.  Acquire and drop as required in tcp_trace().

Move to ANSI C function header, correct prototype types so that short TCP
state is no longer promoted to int unnecessarily.

Add comments.

MFC after: 3 weeks

17 years agoMFC after: 2 weeks
davidch [Fri, 4 May 2007 23:14:19 +0000 (23:14 +0000)]
MFC after: 2 weeks

Updated copyright date to 2007.

Tested with BCM5706 A3.

Added ID for BCM5708 B2.

Removed unused driver version string.

Modified BCE_PRINTF macro to automatically fill-in the sc pointer.

Fixed a kernel panic when the driver was loaded as a module from the
command-line because the MII bus pointer was null (i.e. the MII bus
hadn't been enumerated yet).

Added fix proposed by Vladimir Ivanov <wawa@yandex-team.ru> to prevent
driver state corruption when releasing the lock during the ISR in
bce_rx_intr() to send packets up the stack.

Added new TX chain and register read sysctl interfaces for debugging.

Cleaned up formatting for various other debug routines.

Added a new statistic maintained by firmware which tracks the number
of received packets dropped because no receive buffers are available.

17 years agoSince if_gem is being touted as one of our more architecturally
phk [Fri, 4 May 2007 19:15:28 +0000 (19:15 +0000)]
Since if_gem is being touted as one of our more architecturally
correct network drivers with respect to busmaster DMA, go over it
with at duster to make other aspects of it a role model:

Eliminate the pci specific softc, it serves no rational purpose.

Use convenience resource allocation/deallocation functions to save
code and errorhandling.

Switch from bus_space_{read|write}_%u() to bus_{read|write}_%u()
functions and forget about tags and handles, the resource will know
about those, should they be needed.  This also eliminates a number
of inconsistently named local variables.

17 years agoTweak comment at end of tcp_input() when calling into tcp_do_segment(): the
rwatson [Fri, 4 May 2007 17:45:52 +0000 (17:45 +0000)]
Tweak comment at end of tcp_input() when calling into tcp_do_segment(): the
pcbinfo lock will be released as well, not just the pcb lock.

17 years agoAvoid extra rc4_init() when ng_mppc_updatekey() going to do it anyway.
mav [Fri, 4 May 2007 16:20:47 +0000 (16:20 +0000)]
Avoid extra rc4_init() when ng_mppc_updatekey() going to do it anyway.

Approved by: glebius (mentor)

17 years agoCompact code a bit
mav [Fri, 4 May 2007 16:12:54 +0000 (16:12 +0000)]
Compact code a bit

Approved by: glebius (mentor)

17 years agoMake coherency counter 12bit as it should
mav [Fri, 4 May 2007 16:05:58 +0000 (16:05 +0000)]
Make coherency counter 12bit as it should

Approved by: glebius (mentor)

17 years ago- Fix typo
gabor [Fri, 4 May 2007 16:01:07 +0000 (16:01 +0000)]
- Fix typo

Approved by: keramida (mentor)
MFC after: 3 days

17 years agoFix small mistake (sizeof(pad2) instead of sizeof(pad1))
mav [Fri, 4 May 2007 15:44:22 +0000 (15:44 +0000)]
Fix small mistake (sizeof(pad2) instead of sizeof(pad1))

Approved by: glebius (mentor)

17 years ago- s/jail id/jail ID/, acronyms should be in uppercase in general. Also,
rafan [Fri, 4 May 2007 15:42:58 +0000 (15:42 +0000)]
- s/jail id/jail ID/, acronyms should be in uppercase in general. Also,
  it is written this way in jail(8).

Suggested by: brueffer
Approved by: delphij (mentor, implicit)

17 years agoRemove unneded bzero().
mav [Fri, 4 May 2007 15:41:49 +0000 (15:41 +0000)]
Remove unneded bzero().
SHA1Final() does not require clean buffer.

Approved by: glebius (mentor)

17 years ago- Add top '-j' switch
rafan [Fri, 4 May 2007 15:25:05 +0000 (15:25 +0000)]
- Add top '-j' switch

Reviewed by: brueffer
Approved by: delphij (mentor, implicit), brueffer

17 years agoFixes a missing unlock in the one-2-one hash table, if
rrs [Fri, 4 May 2007 15:19:10 +0000 (15:19 +0000)]
Fixes a missing unlock in the one-2-one hash table, if
it was full and a collision occured, then we would leave
a inp locked. Also fixes a missing inp unlock if IPSEC was
on and it failed during the attach. Bug found by Weongyo Jeong.

17 years agoCorrect osa@ full name.
fjoe [Fri, 4 May 2007 14:51:56 +0000 (14:51 +0000)]
Correct osa@ full name.

17 years agoAdd jdp@ and myself.
fjoe [Fri, 4 May 2007 14:51:39 +0000 (14:51 +0000)]
Add jdp@ and myself.