dragonfly.git
11 years agokernel: Nuke ISA sound card support.
Sascha Wildner [Sun, 10 Feb 2013 21:05:25 +0000 (22:05 +0100)]
kernel: Nuke ISA sound card support.

This removes the following modules:

snd_ad1816.ko
snd_ess.ko
snd_mss.ko
snd_sb8.ko
snd_sb16.ko
snd_sbc.ko

11 years agolibcsu: Assign environment when environ is NULL
John Marino [Sun, 10 Feb 2013 11:32:58 +0000 (12:32 +0100)]
libcsu: Assign environment when environ is NULL

Preloaded libraries can change the environment.  The current logic will
unconditionally reassign the environment, undoing any previous
customization.  This change preserves the previous customization.

Taken-from: FreeBSD SVN 245133 (07 Jan 2013)

11 years agoUpdate files for OpenSSL-1.0.1d import.
Peter Avalos [Sun, 10 Feb 2013 10:04:14 +0000 (02:04 -0800)]
Update files for OpenSSL-1.0.1d import.

11 years agoMerge branch 'vendor/OPENSSL'
Peter Avalos [Sun, 10 Feb 2013 09:34:31 +0000 (01:34 -0800)]
Merge branch 'vendor/OPENSSL'

11 years agoImport OpenSSL-1.0.1d.
Peter Avalos [Sun, 10 Feb 2013 09:23:52 +0000 (01:23 -0800)]
Import OpenSSL-1.0.1d.

      o Fix renegotiation in TLS 1.1, 1.2 by using the correct TLS version.
      o Fix OCSP bad key DoS attack CVE-2013-0166
      o Fix for SSL/TLS/DTLS CBC plaintext recovery attack CVE-2013-0169
      o Fix for TLS AESNI record handling flaw CVE-2012-2686

11 years agokernel: Properly lock dounmount()
François Tigeot [Wed, 6 Feb 2013 08:11:04 +0000 (09:11 +0100)]
kernel: Properly lock dounmount()

* dounmount() wasn't properly locked, make it grab the mntvnode_token

* It is necessary to synchronize against insmntque() clearing the vnode
  linkage

* This commit fixes issue #2506

Thanks to vsrinivas for directions

11 years agopf: Fix set-but-unused gcc47 warning
Sepherosa Ziehau [Sat, 9 Feb 2013 05:19:32 +0000 (13:19 +0800)]
pf: Fix set-but-unused gcc47 warning

11 years agoip6/mroute: Partly fix set-but-unused gcc47 warning
Sepherosa Ziehau [Sat, 9 Feb 2013 03:44:33 +0000 (11:44 +0800)]
ip6/mroute: Partly fix set-but-unused gcc47 warning

11 years agowlan: Partly fix set-but-unused gcc47 warning
Sepherosa Ziehau [Sat, 9 Feb 2013 02:48:08 +0000 (10:48 +0800)]
wlan: Partly fix set-but-unused gcc47 warning

11 years agoacpica: Fix set-but-unused gcc47 warning
Sepherosa Ziehau [Sat, 9 Feb 2013 02:29:00 +0000 (10:29 +0800)]
acpica: Fix set-but-unused gcc47 warning

11 years agoath: Partly fix set-but-unused gcc47 warning
Sepherosa Ziehau [Sat, 9 Feb 2013 02:12:46 +0000 (10:12 +0800)]
ath: Partly fix set-but-unused gcc47 warning

11 years agoiwn: Fix set-but-unused gcc47 warning
Sepherosa Ziehau [Sat, 9 Feb 2013 01:34:03 +0000 (09:34 +0800)]
iwn: Fix set-but-unused gcc47 warning

11 years agoral: Fix set-but-unused gcc47 warning
Sepherosa Ziehau [Sat, 9 Feb 2013 01:31:30 +0000 (09:31 +0800)]
ral: Fix set-but-unused gcc47 warning

11 years agosln: Fix set-but-unused gcc47 warning
Sepherosa Ziehau [Sat, 9 Feb 2013 01:30:05 +0000 (09:30 +0800)]
sln: Fix set-but-unused gcc47 warning

11 years agokernel/firewire: Remove #ifdef __DragonFly__ jungle.
Sascha Wildner [Fri, 8 Feb 2013 18:34:36 +0000 (19:34 +0100)]
kernel/firewire: Remove #ifdef __DragonFly__ jungle.

11 years agokernel/virtio: Constify some struct members to silence warnings.
Sascha Wildner [Fri, 8 Feb 2013 17:59:09 +0000 (18:59 +0100)]
kernel/virtio: Constify some struct members to silence warnings.

11 years agokernel/virtio: Remove some unneeded inclusions of <sys/cdefs.h>.
Sascha Wildner [Fri, 8 Feb 2013 17:45:03 +0000 (18:45 +0100)]
kernel/virtio: Remove some unneeded inclusions of <sys/cdefs.h>.

11 years agokernel/virtio: Remove some unused variables.
Sascha Wildner [Fri, 8 Feb 2013 17:43:08 +0000 (18:43 +0100)]
kernel/virtio: Remove some unused variables.

11 years agoRevert "Revert "i386/files: isa/clock.c is clean enough to compile w/ -Werror""
Sascha Wildner [Fri, 8 Feb 2013 17:21:45 +0000 (18:21 +0100)]
Revert "Revert "i386/files: isa/clock.c is clean enough to compile w/ -Werror""

This reverts commit 9b40fbdac1da8214991bb70ab4767862203f2375.

Now that we no longer have gcc41, it compiles without warnings on both of
our compilers.

11 years agotip(1): Fix some -Wmissing-parameter-type warnings.
Sascha Wildner [Fri, 8 Feb 2013 17:14:31 +0000 (18:14 +0100)]
tip(1): Fix some -Wmissing-parameter-type warnings.

It still needs more cleanup.

11 years agomxge: Fix set-but-unused gcc47 warning
Sepherosa Ziehau [Fri, 8 Feb 2013 14:18:44 +0000 (22:18 +0800)]
mxge: Fix set-but-unused gcc47 warning

11 years agotools/fw: Fix set-but-unused gcc47 warning in generated C file
Sepherosa Ziehau [Fri, 8 Feb 2013 14:17:18 +0000 (22:17 +0800)]
tools/fw: Fix set-but-unused gcc47 warning in generated C file

Only generate parent and the assignment code, if there are more
then 1 input firmware files.

11 years agobce: Fix set-but-unused gcc47 warning
Sepherosa Ziehau [Fri, 8 Feb 2013 13:35:27 +0000 (21:35 +0800)]
bce: Fix set-but-unused gcc47 warning

11 years agobfe: Fix set-but-unused gcc47 warning
Sepherosa Ziehau [Fri, 8 Feb 2013 13:34:47 +0000 (21:34 +0800)]
bfe: Fix set-but-unused gcc47 warning

11 years agobge: Fix set-but-unused gcc47 warning
Sepherosa Ziehau [Fri, 8 Feb 2013 13:33:59 +0000 (21:33 +0800)]
bge: Fix set-but-unused gcc47 warning

11 years agotl: Fix set-bu-unused gcc47 warning
Sepherosa Ziehau [Fri, 8 Feb 2013 13:31:57 +0000 (21:31 +0800)]
tl: Fix set-bu-unused gcc47 warning

11 years agokue: Fix set-but-unused gcc47 warning
Sepherosa Ziehau [Fri, 8 Feb 2013 13:27:33 +0000 (21:27 +0800)]
kue: Fix set-but-unused gcc47 warning

11 years agobnx: Fix set-but-unused gcc47 warning
Sepherosa Ziehau [Fri, 8 Feb 2013 13:23:20 +0000 (21:23 +0800)]
bnx: Fix set-but-unused gcc47 warning

11 years agobridge: Fix set-but-unused gcc47 warning
Sepherosa Ziehau [Fri, 8 Feb 2013 10:58:41 +0000 (18:58 +0800)]
bridge: Fix set-but-unused gcc47 warning

11 years agode: Fix set-but-unused gcc47 warning
Sepherosa Ziehau [Fri, 8 Feb 2013 10:58:14 +0000 (18:58 +0800)]
de: Fix set-but-unused gcc47 warning

11 years agoppp: Fix set-but-unused gcc47 warning
Sepherosa Ziehau [Fri, 8 Feb 2013 10:57:34 +0000 (18:57 +0800)]
ppp: Fix set-but-unused gcc47 warning

11 years agoemx: Implement multiple hardware TX queue support; disabled by default
Sepherosa Ziehau [Thu, 7 Feb 2013 06:38:04 +0000 (14:38 +0800)]
emx: Implement multiple hardware TX queue support; disabled by default

Two notes about the multiple hardware TX queue support:

- It could only be enabled on 82571, 82572 and es2lan.  Enabling it on
  82574, whose datasheet claims multiple hardware TX queue support,
  could cause watchdog timeout when heavy TCP receiving test are
  conducted.

- It is disabled by default.  Since 82571, 82572 and es2lan only have
  one global context for TX hardware checksum and TSO, the driver will
  have to setup TX context descriptor for each packet.  Using TX context
  descriptor for each packet could cause serious transmission performance
  regression for tiny packets, see comment of following commit:
  51e6819fc2ad7cefdf830dcc9f20bcbe2af2aa52

  However, multiple hardware TX queues still could be enabled depending
  on the workload (*):
  o  TSO probably are not affected by setting up a TX context descriptor
     for each TSO packet.
  o  Stuffs do not using hardware TX checksum, e.g. IP fast forwarding.

  Multiple hardware TX queues could be enabled globally by tunable:
      hw.emx.txr=2
  Multiple hardware TX queues could be enabled on individual interface by
  tunable:
      hw.emxY.txr=2

(*) Another thought is to add per-interface metric, which indicates the
size of packets that worth hardware checksum; if the packets' size is
below that metric, software checksum is used.

11 years agokether_ntoa.9: Fix prototypes.
Sascha Wildner [Thu, 7 Feb 2013 21:55:21 +0000 (22:55 +0100)]
kether_ntoa.9: Fix prototypes.

11 years agogames/atc: Fix a logic bug.
Sascha Wildner [Thu, 7 Feb 2013 19:05:52 +0000 (20:05 +0100)]
games/atc: Fix a logic bug.

The original logic should have really been:

if ((l < 'a' && l > 'z') || (l < 'A' && l > 'Z'))

But using islower()/isupper() is simpler and more readable.

Taken-from: NetBSD

11 years agokether_ntoa.9: Mention %D.
Sascha Wildner [Thu, 7 Feb 2013 17:38:23 +0000 (18:38 +0100)]
kether_ntoa.9: Mention %D.

Also, perform some minor cleanup.

11 years agohexncpy.9: Some little cleanup.
Sascha Wildner [Thu, 7 Feb 2013 17:37:31 +0000 (18:37 +0100)]
hexncpy.9: Some little cleanup.

11 years agoem/emx: Clear bit 28 of TARC1 for 82571/82572/es2lan
Sepherosa Ziehau [Thu, 7 Feb 2013 01:43:48 +0000 (09:43 +0800)]
em/emx: Clear bit 28 of TARC1 for 82571/82572/es2lan

Clearing bit 28 of TARC1 is done in HAL's device initialization, however,
it happens before the MULR bit is turned on in TX unit initialization;
MULR is not necessarily on by default, since its value is loaded from
EEPROM according to the datasheet.

11 years agodoc - Add a note about %D removal in porting_drivers.txt document.
Antonio Huete Jimenez [Thu, 7 Feb 2013 01:07:30 +0000 (02:07 +0100)]
doc - Add a note about %D removal in porting_drivers.txt document.

11 years agoAdd a manpage for hexncpy function.
Antonio Huete Jimenez [Thu, 7 Feb 2013 00:47:25 +0000 (01:47 +0100)]
Add a manpage for hexncpy function.

Also fix .Dd in kether_ntoa(9).

11 years agoAdd manual pages for kether_ntoa and kether_aton functions.
Antonio Huete Jimenez [Wed, 6 Feb 2013 23:03:28 +0000 (00:03 +0100)]
Add manual pages for kether_ntoa and kether_aton functions.

11 years agoigb: Sync w/ intel igb-2.3.7
Sepherosa Ziehau [Wed, 6 Feb 2013 08:50:52 +0000 (16:50 +0800)]
igb: Sync w/ intel igb-2.3.7

- Support I210 and I211
- EEE settings are only applied to copper chips
- Explicitly disable DMA coalesce, if not requested

11 years agoem: Sync w/ intel em-7.3.4
Sepherosa Ziehau [Wed, 6 Feb 2013 04:34:35 +0000 (12:34 +0800)]
em: Sync w/ intel em-7.3.4

- Move PBA configuration into em_reset(); mainly to put hardware into
  stable state after the driver attached.
- Add workaround for PCH2 chips during resuming.
- Fixup flow control parameters for PCH, ICH9, ICH10 and ES2LAN chips.
- Remove no longer needed ERT setting for PCH2, ICH9 and ICH10 chips.

11 years agoemx: Sync w/ intel em-7.3.4
Sepherosa Ziehau [Wed, 6 Feb 2013 04:32:16 +0000 (12:32 +0800)]
emx: Sync w/ intel em-7.3.4

Move PBA configuration into emx_reset(); mainly to put hardware into
stable state after the driver attached.

11 years agoig_hal: Merge intel em-7.3.4 and igb-2.3.7 hardware abstraction layer
Sepherosa Ziehau [Tue, 5 Feb 2013 09:05:57 +0000 (17:05 +0800)]
ig_hal: Merge intel em-7.3.4 and igb-2.3.7 hardware abstraction layer

11 years agoEnable -Werror for world building with gcc47.
Sascha Wildner [Sun, 20 Jan 2013 02:25:15 +0000 (03:25 +0100)]
Enable -Werror for world building with gcc47.

The kernel still has warnings to be reviewed.

11 years agoRemove xlint. It's outdated and there are much better checkers around.
Sascha Wildner [Tue, 5 Feb 2013 20:23:52 +0000 (21:23 +0100)]
Remove xlint. It's outdated and there are much better checkers around.

clang static analyzer, cppcheck, coccinelle et al.

So there's no need to maintain it in base.

11 years agolibm: Exclude -Woverflow from -Werror in s_csqrtl.c for gcc47/i386.
Sascha Wildner [Sun, 20 Jan 2013 02:15:59 +0000 (03:15 +0100)]
libm: Exclude -Woverflow from -Werror in s_csqrtl.c for gcc47/i386.

The issue is that a long double in gcc47 on i386 doesn't seem to be
able to hold LDBL_MAX. At least a warning is generated. Cause
unknown, might be a bug in gcc (or our gcc's configuration).

gcc47 on FreeBSD/i386 does not show this behavior (no warning
generated at least).

Until we have a better clue, don't turn this warning into an error,
even if -Werror is specified, so that building can go on.

FreeBSD-testing: tuxillo

11 years agoFix some more things related to the switch to gcc47.
Sascha Wildner [Tue, 5 Feb 2013 19:54:06 +0000 (20:54 +0100)]
Fix some more things related to the switch to gcc47.

* Adjust make.conf and its manual page.

* Create gcc47 MLINKS properly (clean up the Makefile a bit, too).

* Exclude gcc44 in the qemu setup.

11 years agoUPDATING: Adjust heading.
Sascha Wildner [Sun, 3 Feb 2013 11:53:24 +0000 (12:53 +0100)]
UPDATING: Adjust heading.

11 years agokernel -- ffs: Softdep lock fix (lk unheld in process_worklist_item)
Venkatesh Srinivas [Wed, 30 Jan 2013 18:21:10 +0000 (10:21 -0800)]
kernel -- ffs: Softdep lock fix (lk unheld in process_worklist_item)

In process_worklist_item, from newdirrem/request_cleanup, the softdep
lock was not held before removing workitems. Make newdirrem take the
softdep lock around request_cleanup.

Reported-by: swildner@
Bug: 2501

11 years agopktgen: Support multiple TX queues
Sepherosa Ziehau [Mon, 4 Feb 2013 09:59:04 +0000 (17:59 +0800)]
pktgen: Support multiple TX queues

- Dispatch start operation to subqueue owner CPU's netisr.
- Add PKTGENMQSTART to utilize multiple TX queues

11 years agopktgen: Factor out pktgen_ref
Sepherosa Ziehau [Mon, 4 Feb 2013 09:33:06 +0000 (17:33 +0800)]
pktgen: Factor out pktgen_ref

Will be needed for multiple TX queue support

11 years agopktgen: Move buflist and stop callout into per-cpu data struct
Sepherosa Ziehau [Mon, 4 Feb 2013 09:29:56 +0000 (17:29 +0800)]
pktgen: Move buflist and stop callout into per-cpu data struct

Prepare for the multiple TX queue support.

11 years agoUnbreak world with NO_GCC44
François Tigeot [Sun, 3 Feb 2013 12:32:01 +0000 (13:32 +0100)]
Unbreak world with NO_GCC44

gcc44 in now optional, do not require its use during buildworld
anymore.

11 years agoPromote GCC 4.7 to be the default compiler
John Marino [Sun, 3 Feb 2013 09:12:27 +0000 (10:12 +0100)]
Promote GCC 4.7 to be the default compiler

This commit causes world and kernel to be built by GCC 4.7 if the compiler
is not specified.  As a side effect, the make variable "NO_GCC47" no longer
functions.  GCC 4.7 will always be built.  In addition to changing the
default of WORLD_CCVER, the CCVER variable also defaults to "gcc47" now.

The variable "NO_GCC44" has been resurrected to prevent GCC 4.4 from
being built.  As noted in the UPDATING file, using this option is highly
discouraged.  GCC 4.4 is very useful in building some packages in pkgsrc
that fail to build with GCC 4.7's stricter standards.  The new DPorts
system is designed to use GCC 4.4 by default, regardless of the base's
default compiler setting.  If at all possible, build *both* compilers
for the maximum flexibility to build packages.

11 years agoUnbreak world (gcc47 plugin)
John Marino [Sat, 2 Feb 2013 22:35:27 +0000 (23:35 +0100)]
Unbreak world (gcc47 plugin)

The installincludes target is run before "depend" target, so gtype.state
hasn't been created by the time "installincludes" is executed in stage 4.
Luckily there is a similar functionality for FILES that is executed at
the install phase, which will unbreak world (broken on previous commit).

Other headers are generated as well, so swap INCSGROUPS with FILESGROUPS
for everything to avoid having to figure which ones.  The headers are
needed at runtime, not buildtime, so this is okay.

11 years agogcc47: Fix plugin mechanism
John Marino [Sat, 2 Feb 2013 15:29:40 +0000 (16:29 +0100)]
gcc47: Fix plugin mechanism

There were two problems with gcc 4.7 LTO plugin:
1) It needed a number of object files from libiberty.
   Unfortunately the previous object files couldn't be used as they need
   to be built with -fPIC and libiberty is only an internal (non-pic) lib.
2) The plugin header directory was missing completely, as seen with
   gcc -print-file-name=plugin only returning "plugin".

11 years agopktgen: Use atomic for pktgen and module reference count
Sepherosa Ziehau [Sat, 2 Feb 2013 08:23:57 +0000 (16:23 +0800)]
pktgen: Use atomic for pktgen and module reference count

Module's token is no longer needed in many places.  Ease the upcoming
changes to support multiple hardware TX queues.

11 years agokernel - Fix kernel panic caused by rename race
Matthew Dillon [Fri, 1 Feb 2013 21:47:37 +0000 (13:47 -0800)]
kernel - Fix kernel panic caused by rename race

* kern_rename() must temporarily unlock the source ncp when resolving the
  target ncp, and will sometimes have to temporarily unlock the target ncp
  when relocking both.  This can race against a topological removal of one
  or both ncp's, resulting in incorrect operation or a panic due to a
  NULL vp.

* Detect the case and return EAGAIN if the ncp's have been marked NCF_DESTROYED
  or if fromncp loses its vp resolution.  Callers detect the EAGAIN and retry
  the operation.

Reported-by: ftigeot
11 years agokernel - cycle vnode in cache_unlink()
Matthew Dillon [Fri, 1 Feb 2013 17:57:55 +0000 (09:57 -0800)]
kernel - cycle vnode in cache_unlink()

* cache_unlink() and cache_rename() are called by VFS's to update the
  namecache after a file deletion or rename.  cache_rename() also calls
  cache_unlink().

* Add code to this function which cycles any vnode attached to the
  namecache entry for the case where its refs are 0.  This will
  trigger VOP_INACTIVE and allow the VFS to detect file deletions.

  Generally speaking this means that VFS's no longer have to detect
  deleted files in code paths outside of VOP_INACTIVE w/regards to
  forcing a reclaim on the vnode.

* HAMMER2 now depends on this behavior.  The change in behavior allows
  HAMMER2 to do away with inode backpointers from chain structures,
  greatly simplifying the code and removing a chicken-egg dependency.

* Other VFSs (HAMMER, TMPFS, UFS) may be adjusted in the future but for
  now their file deletion code outside of VOP_INACTIVE is being left
  intact.

11 years agokernel - Fix memory leak on iocom terminaion
Matthew Dillon [Fri, 1 Feb 2013 17:54:43 +0000 (09:54 -0800)]
kernel - Fix memory leak on iocom terminaion

* freerd_state was being cleaned up by the reader thread, but during
  decomissioning the writer thread can simulate incoming packets
  to cleanly shutdown open transactions, and this can reload
  freerd_state.

* Move the cleanup for freerd_state and freewr_state out of the
  reader/writer threads and into kdmsg_iocom_uninit() proper.

* Fixes dangling 112 bytes of memory left allocated after uninit.

11 years agohammer2 - serialized flush work part 4
Matthew Dillon [Fri, 1 Feb 2013 17:19:41 +0000 (09:19 -0800)]
hammer2 - serialized flush work part 4

* Hardlink helper functions now take and return hammer2_chain pointers
  instead of messing with hammer2_inode's, which simplifies matters
  greatly.

* Major cleanup of the hardlink handling code.

* Get hardlink creation/deletion mostly working as expected again.
  There is still an issue with hammer_inode aliasing (see TODO),
  and the rename-directory-to-some-parent hardlink issue has not been
  addressed yet.

11 years agopktgen: Unbreak compile
Sepherosa Ziehau [Fri, 1 Feb 2013 08:33:21 +0000 (16:33 +0800)]
pktgen: Unbreak compile

11 years agohammer2 - serialized flush work part 3
Matthew Dillon [Fri, 1 Feb 2013 00:03:59 +0000 (16:03 -0800)]
hammer2 - serialized flush work part 3

* Clean up hammer2_inode vs hammer2_chain.  Remove chain->u.ip (the
  backpointer from the chain to the inode), which simplifies a lot
  of things.  This was only used to detect when a vnode might have to
  be recycled but we solve that a different way by adjusting the
  kernel's cache_unlink() code.

* Revamp hammer2_inode_get().  This routine now always kmalloc()'s a new
  hammer2_inode.  In HAMMER2, the hammer2_inode structure is actually
  an indirect pointer to an underlying chain so it does not have to be
  unique.  There can multiple ip's pointing to the same chain.

* Revamp hammer2_inode_put().  The ref on ip->chain is dropped via ip->chain,
  and the (passed_chain) argument is treated separately and does not
  necessarily have to match ip->chain.  This makes hammer2_inode_put()'s API
  comparable to hammer2_inode_unlock_ex(), reducing confusion.

* Depend on adjustments to cache_unlink() and cache_rename() to detect
  vnodes which should be recycled.  This code (in the proper kernel, not
  here) now triggers a deactivation if the vnode related to the namecache
  entry has no refs, allowing HAMMER2 to detect any deletions and vrecycle()
  the vnode immediately.

* Numerous ref/lock fixes.

11 years agoMerge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Fri, 1 Feb 2013 00:03:21 +0000 (16:03 -0800)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly

11 years agoRemove legacy lex source files
John Marino [Thu, 31 Jan 2013 19:18:54 +0000 (20:18 +0100)]
Remove legacy lex source files

The ancient lex was replaced by flex recently.  The bootstrapping issue
seems to be solved, so let's go ahead and purge lex from base now.

11 years agolibstdc++47: Fix iostream bug 2478
John Marino [Thu, 31 Jan 2013 15:58:59 +0000 (16:58 +0100)]
libstdc++47: Fix iostream bug 2478

c++ iostreams were segfaulting -- apparently _M_table isn't defined in all
cases.  Fix inspired from FreeBSD configuration files.

https://bugs.dragonflybsd.org/issues/2478

11 years agohammer2 - serialized flush work part 2
Matthew Dillon [Thu, 31 Jan 2013 06:41:52 +0000 (22:41 -0800)]
hammer2 - serialized flush work part 2

* Adjust the hammer2_inode_lock*() API to separate the inode lock from
  the chain lock (it gets both).  The locked chain associated with the
  inode is now returned and can change during search iterations, eventually
  to be returned via hammer2_inode_unlock*().

* Fixes issues related to threads obtaining more than 1 shared lock on the
  same chain, which can deadlock deeper procedures which might need to
  temporarily upgrade the lock.

* Continue working through bugs.

11 years agohammer2 - serialized flush work part 1
Matthew Dillon [Thu, 31 Jan 2013 04:53:27 +0000 (20:53 -0800)]
hammer2 - serialized flush work part 1

This is preliminary work required to support chain structure replication
for the purposes of recording modifications which are then separated by
serialization points (by transaction id).  Ultimately this will allow the
flush code to flush to an exact serialization point and in the process
operate completely asynchronously from any further modifications being
made on the frontend after that serialization point.

* Separate hammer2_inode from hammer2_chain.

* Split the locking APIs for inodes and chains into their own functions.

* Move ip_data into chain->data->ipdata (kmalloc'd), instead of embedding
  it in hammer2_inode.  This allows the inode structure to disconnect from
  the chain.

11 years agousb4bsd: Add missing parameter to ue_start().
Sascha Wildner [Wed, 30 Jan 2013 22:49:35 +0000 (23:49 +0100)]
usb4bsd: Add missing parameter to ue_start().

Submitted-by: Yellow Rabbit <yrabbit@sdf.lonestar.org>
Dragonfly-bug: <http://bugs.dragonflybsd.org/issues/2503>

11 years agohammer2 - Start working on the freemap (note: freemap not yet operational)
Matthew Dillon [Wed, 30 Jan 2013 07:27:01 +0000 (23:27 -0800)]
hammer2 - Start working on the freemap (note: freemap not yet operational)

* hammer2 show now tries to dump all four volume headers

* Minimum storage allocation chunk size is 1K, increase MIN_RADIX and
  MIN_ALLOC appropriately.

* Preliminary reserved block mappings for the freemap indirect and bitmap
  blocks.

* Preliminary freemap design will use normal indirect blocks (but with
  a different bref.type name).  The check area of the bref will be used to
  store additional freemap hints.

  The volume header will contain a blockref to the root of the freemap,
  and newfs_hammer2 now initializes this blockref.

* Bring major hammer2_disk.h media structure comments up to snuff.

* Formalize blockref.methods and set proper defaults.  This field contains
  the crc/check and compression methods for a blockref.

11 years agokernel - Fix deadlock when umount races an access on the underlying filesystem
Matthew Dillon [Tue, 29 Jan 2013 23:04:05 +0000 (15:04 -0800)]
kernel - Fix deadlock when umount races an access on the underlying filesystem

* The nlookup code temporarily busies the target mount when diving a
  mount point in cases where the base of the mount is not resolved
  in the namecache.

* Fix a deadlock which can occur between the namecache structural lock
  and the vfs_busy() on the mount structure by reordering the lock.

* Generally only occured if an attempt was made to unmount a filesystem
  on which programs are still doing active operations (verses just
  passively holding references to the filesystem).  For example, a
  umount on the target filesystem occuring while a cpdup using an
  absolute path is running.

Reported-by: ftigeot
11 years agokernel - Attempt to fix NULL pointer dereference during console switch
Matthew Dillon [Tue, 29 Jan 2013 19:19:22 +0000 (11:19 -0800)]
kernel - Attempt to fix NULL pointer dereference during console switch

* Attempt to fix NULL pointer dereference during console switch.
  Bug 2481.

Reported-by: marino
11 years agokernel - Fix tty cool-aid
Matthew Dillon [Tue, 29 Jan 2013 19:11:49 +0000 (11:11 -0800)]
kernel - Fix tty cool-aid

* Fix a tty crash in the ^T handling code.  The code can race LWP
  destruction and blow up.

11 years agosocket: Mark the asynchronous rcvd netmsg dead, when it is dropped
Sepherosa Ziehau [Tue, 29 Jan 2013 08:29:35 +0000 (16:29 +0800)]
socket: Mark the asynchronous rcvd netmsg dead, when it is dropped

This is an attempt to fix the race that asynchronous rcvd netmsg
is queued to netisr after it is dropped in tcp_close().

Reported-by: ftigeot@
11 years agoigb: Enable multiple hardware TX queues
Sepherosa Ziehau [Tue, 29 Jan 2013 06:48:11 +0000 (14:48 +0800)]
igb: Enable multiple hardware TX queues

11 years agoigb: Implement aggregated RX/TX MSI-X support
Sepherosa Ziehau [Tue, 29 Jan 2013 06:27:11 +0000 (14:27 +0800)]
igb: Implement aggregated RX/TX MSI-X support

11 years agoigb: Rework serializer array setup
Sepherosa Ziehau [Tue, 29 Jan 2013 03:32:32 +0000 (11:32 +0800)]
igb: Rework serializer array setup

11 years agoigb: Setup TX rings' intr CPUID properly if MSI/legacy intr is used
Sepherosa Ziehau [Tue, 29 Jan 2013 02:50:14 +0000 (10:50 +0800)]
igb: Setup TX rings' intr CPUID properly if MSI/legacy intr is used

11 years agoigb: Fix up TX rings' MSI-X setup
Sepherosa Ziehau [Tue, 29 Jan 2013 02:42:58 +0000 (10:42 +0800)]
igb: Fix up TX rings' MSI-X setup

11 years agoif: Fix wrong assertion condition in ifq_set_subq_mask
Sepherosa Ziehau [Mon, 28 Jan 2013 13:37:26 +0000 (21:37 +0800)]
if: Fix wrong assertion condition in ifq_set_subq_mask

11 years agoigb: Utilize power of 2 mask based CPUID to subqueue mapping
Sepherosa Ziehau [Mon, 28 Jan 2013 13:24:55 +0000 (21:24 +0800)]
igb: Utilize power of 2 mask based CPUID to subqueue mapping

11 years agoif: Add power of 2 mask based CPUID to subqueue mapping
Sepherosa Ziehau [Mon, 28 Jan 2013 13:24:17 +0000 (21:24 +0800)]
if: Add power of 2 mask based CPUID to subqueue mapping

11 years agoigb: Setup subqueue count properly
Sepherosa Ziehau [Mon, 28 Jan 2013 13:08:16 +0000 (21:08 +0800)]
igb: Setup subqueue count properly

11 years agoif: Add function to set subqueue count
Sepherosa Ziehau [Mon, 28 Jan 2013 13:08:00 +0000 (21:08 +0800)]
if: Add function to set subqueue count

11 years agoigb: Add per-TX ring tx_pkt sysctl node, if IGB_TSS_DEBUG is enabled
Sepherosa Ziehau [Mon, 28 Jan 2013 11:51:23 +0000 (19:51 +0800)]
igb: Add per-TX ring tx_pkt sysctl node, if IGB_TSS_DEBUG is enabled

11 years agoigb: Add IGB_TSS_DEBUG kernel option
Sepherosa Ziehau [Mon, 28 Jan 2013 09:53:23 +0000 (17:53 +0800)]
igb: Add IGB_TSS_DEBUG kernel option

11 years agoigb: Add per-TX ring enable flag.
Sepherosa Ziehau [Mon, 28 Jan 2013 09:45:50 +0000 (17:45 +0800)]
igb: Add per-TX ring enable flag.

For TX rings' whose enable flag are not set, which could happen during
polling entering/exiting, its corresponding subqueue will be purged.

11 years agoigb: Move IGB_FLAG_TSO_IPLEN0 into TX rings
Sepherosa Ziehau [Mon, 28 Jan 2013 09:36:36 +0000 (17:36 +0800)]
igb: Move IGB_FLAG_TSO_IPLEN0 into TX rings

Avoid extra access to softc and leave enough space for TX ring specific
flags.

11 years agoigb: Define macros for wreg_nsegs, instead of using magic number
Sepherosa Ziehau [Mon, 28 Jan 2013 09:22:17 +0000 (17:22 +0800)]
igb: Define macros for wreg_nsegs, instead of using magic number

11 years agowlan: ifaltq is no longer ifqueue
Sepherosa Ziehau [Mon, 28 Jan 2013 01:57:04 +0000 (09:57 +0800)]
wlan: ifaltq is no longer ifqueue

11 years agowlan: Remove unused ieee80211_drain_ifq
Sepherosa Ziehau [Mon, 28 Jan 2013 01:43:03 +0000 (09:43 +0800)]
wlan: Remove unused ieee80211_drain_ifq

11 years agoigb: Make tx_intr_nsegs and tx_wreg_nsegs nodes aware of multiple TX queues
Sepherosa Ziehau [Sun, 27 Jan 2013 13:49:51 +0000 (21:49 +0800)]
igb: Make tx_intr_nsegs and tx_wreg_nsegs nodes aware of multiple TX queues

11 years agoigb: Merge per-RX queue rx_wregs into single rx_wreg_nsegs node
Sepherosa Ziehau [Sun, 27 Jan 2013 13:21:22 +0000 (21:21 +0800)]
igb: Merge per-RX queue rx_wregs into single rx_wreg_nsegs node

There is no use case that it needs to be tuned on per-RX queue base.

11 years agokernel/agp: No need to include opt_bus.h here.
Sascha Wildner [Sat, 26 Jan 2013 13:58:32 +0000 (14:58 +0100)]
kernel/agp: No need to include opt_bus.h here.

11 years agokernel/mfi: Sort commands by opcode.
Sascha Wildner [Fri, 25 Jan 2013 22:48:59 +0000 (23:48 +0100)]
kernel/mfi: Sort commands by opcode.

11 years agoigb: Don't assume only one TX queue will be enabled.
Sepherosa Ziehau [Fri, 25 Jan 2013 05:52:59 +0000 (13:52 +0800)]
igb: Don't assume only one TX queue will be enabled.

Prepare multiple TX queue support.

- Add TX queue count when MSI-X is used
- Add currently used TX queue count

Number of TX queues are still limited to 1 as of this commit.

11 years agoigb: Use ifsq_devstart_sched in watchdog and resume methods
Sepherosa Ziehau [Fri, 25 Jan 2013 04:33:14 +0000 (12:33 +0800)]
igb: Use ifsq_devstart_sched in watchdog and resume methods

So these two method won't consume too much time.

In watchdog method, schedule all subqueues instead of the subqueue
causing watchdog timeout.

11 years agoigb: Setup subqueue CPUID before possible calling of igb_init
Sepherosa Ziehau [Fri, 25 Jan 2013 03:34:36 +0000 (11:34 +0800)]
igb: Setup subqueue CPUID before possible calling of igb_init

11 years agoigb: Utilize subqueue watchdog functions
Sepherosa Ziehau [Fri, 25 Jan 2013 03:31:01 +0000 (11:31 +0800)]
igb: Utilize subqueue watchdog functions

Prepare for the multiple TX queue support