]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
17 years agoadd my mentees and comentor relation
sem [Mon, 2 Apr 2007 18:16:34 +0000 (18:16 +0000)]
add my mentees and comentor relation

17 years agoAdd another fd leak test for accept() I used to test the fix in 1.234 of
jhb [Mon, 2 Apr 2007 16:02:50 +0000 (16:02 +0000)]
Add another fd leak test for accept() I used to test the fix in 1.234 of
sys/kern/uipc_syscall.c.

17 years agoInstead of directly sourcing the firewall script, run it in a separate shell.
mtm [Mon, 2 Apr 2007 15:38:53 +0000 (15:38 +0000)]
Instead of directly sourcing the firewall script, run it in a separate shell.
If the firewall script is sourced directly from the script, then any
exit statements in it will also terminate the rc.d script prematurely.

PR: conf/78762
MFC-After: 2 weeks

17 years agoAdd myself to the tree of life^WFreeBSD
wes [Mon, 2 Apr 2007 15:36:07 +0000 (15:36 +0000)]
Add myself to the tree of life^WFreeBSD

17 years agojlemon was my mentor and I got my commit bit in 2001.
brooks [Mon, 2 Apr 2007 14:09:08 +0000 (14:09 +0000)]
jlemon was my mentor and I got my commit bit in 2001.

17 years agoIn rc.firewall, make it clear that this is the setup for IPFW(4), and not
rwatson [Mon, 2 Apr 2007 14:02:06 +0000 (14:02 +0000)]
In rc.firewall, make it clear that this is the setup for IPFW(4), and not
for the sundry other firewalls in the system.

MFC after: 3 days
Submitted by: Richard dot Clayton at cl dot cam dot ac dot uk

17 years ago- Split out the part of SYSCALL_MODULE_HELPER() that builds a 'struct
jhb [Mon, 2 Apr 2007 13:53:26 +0000 (13:53 +0000)]
- Split out the part of SYSCALL_MODULE_HELPER() that builds a 'struct
  sysent' for a new system call into a new MAKE_SYSENT() macro.
- Use MAKE_SYSENT() to build a full sysent for the nfssvc system call in
  the NFS server and use syscall_register() and syscall_deregister() to
  manage the nfssvc system call entry instead of manually frobbing the
  sysent[] array.

17 years agoDon't go to a whole lot of extra work to handle the race where the new
jhb [Mon, 2 Apr 2007 13:40:38 +0000 (13:40 +0000)]
Don't go to a whole lot of extra work to handle the race where the new
file descriptor is closed out from under us in kern_open().  This race
is already handled and the file will be closed when kern_open() does an
fdrop just before returning.

17 years ago- Add some comments what these files are for and how one can generate the
gabor [Mon, 2 Apr 2007 11:31:18 +0000 (11:31 +0000)]
- Add some comments what these files are for and how one can generate the
  graphical output

Reviewed by: flz

17 years agoOops, sort properly.
ru [Mon, 2 Apr 2007 10:38:51 +0000 (10:38 +0000)]
Oops, sort properly.

17 years agoRevert busy refcount back to int. As a side note, multiple open
ariff [Mon, 2 Apr 2007 10:24:15 +0000 (10:24 +0000)]
Revert busy refcount back to int. As a side note, multiple open
is still (and always) possible and does not change previous behaviour.

Requested by: netchild

17 years agoAdd my relashionships.
ru [Mon, 2 Apr 2007 10:23:27 +0000 (10:23 +0000)]
Add my relashionships.

17 years agoAdd myself
ache [Mon, 2 Apr 2007 09:59:55 +0000 (09:59 +0000)]
Add myself

17 years agoAdd myself
ache [Mon, 2 Apr 2007 09:40:02 +0000 (09:40 +0000)]
Add myself

17 years agoLogically separate the complex `hanoi' and `math' tests from basic tests.
yar [Mon, 2 Apr 2007 08:20:43 +0000 (08:20 +0000)]
Logically separate the complex `hanoi' and `math' tests from basic tests.

17 years agoDon't forget to close the range if we branched over its end
yar [Mon, 2 Apr 2007 08:14:46 +0000 (08:14 +0000)]
Don't forget to close the range if we branched over its end
and had no chance to match it by the 2nd address precisely.
Otherwise the unclosed range would bogusly extend to the end
of stream.

Add a basic regression test for the bug fixed.  (This change
also fixes the more complex case 5.3 from `multitest.t'.)

Compared with: SUN and GNU seds
Tested by: regression tests
MFC after: 1 week

17 years agoToday SUN and GNU seds fully agree on test 5.3 and behave
yar [Mon, 2 Apr 2007 07:50:10 +0000 (07:50 +0000)]
Today SUN and GNU seds fully agree on test 5.3 and behave
in a more reasonable way than BSD sed does: they properly
close the range even if we branched over its end.  No doubt,
the range `1,5' should not match lines from 9 through 14.

17 years agoadd myself
ijliao [Mon, 2 Apr 2007 07:09:03 +0000 (07:09 +0000)]
add myself

17 years agoPrevent foot-shooting in advance: Put the MATCH() macro's value
yar [Mon, 2 Apr 2007 06:47:48 +0000 (06:47 +0000)]
Prevent foot-shooting in advance:  Put the MATCH() macro's value
in parentheses.  The ?: operator has a remarkably low precedence, so
expressions like (MATCH(foo) && bar) would have an unexpected meaning
w/o the parentheses around MATCH().

Tested with: md5(1)

17 years agoDisable seq_modevent(). The implementation is incomplete, and causing
ariff [Mon, 2 Apr 2007 06:03:47 +0000 (06:03 +0000)]
Disable seq_modevent(). The implementation is incomplete, and causing
memory leak during unload.

17 years agoUse our own timer for watchdog instead of if_watchdog/if_timer
yongari [Mon, 2 Apr 2007 04:43:41 +0000 (04:43 +0000)]
Use our own timer for watchdog instead of if_watchdog/if_timer
interface.

17 years agoNow that there is always a compression-layer skip function available,
cperciva [Mon, 2 Apr 2007 04:21:22 +0000 (04:21 +0000)]
Now that there is always a compression-layer skip function available,
skip over the end-of-entry padding instead of reading and discarding
it.

Considering that tar files normally have a block size of 10kB, this
isn't likely to avoid reading any data, but at least it makes the code
simpler and clearer.

17 years ago- Fix typo in ntpd(8).
chinsan [Mon, 2 Apr 2007 04:18:49 +0000 (04:18 +0000)]
- Fix typo in ntpd(8).

Noticed by: Ben Kaduk (minimarmot _at_ gmail.com)
Approved by: delphij@ (mentor)

17 years agoNo need to track every closing instance, and put busy counter to rest
ariff [Mon, 2 Apr 2007 03:46:25 +0000 (03:46 +0000)]
No need to track every closing instance, and put busy counter to rest
in its single bit coffin.

17 years agoFreeze the simq, not the devq, if we run out of command slots. This fixes
scottl [Mon, 2 Apr 2007 03:31:37 +0000 (03:31 +0000)]
Freeze the simq, not the devq, if we run out of command slots.  This fixes
the last round of reported instability in the rev 13/14 driver.

Approved by: Erich Chen

17 years agoProvide hint / tunable for possible asynchronous USB execution. Async
ariff [Mon, 2 Apr 2007 03:25:39 +0000 (03:25 +0000)]
Provide hint / tunable for possible asynchronous USB execution. Async
execution should help us avoiding potential deadlock and illegal locking
while sleeping in various mixer -> usb calls. To enable it, use
hint.uaudio.%d.async="1" or sysctl dev.uaudio.%d.async=1. Default is
disable, to remain compatible with old behaviour (with slight risk of
potential deadlock).

17 years ago- Don't wakeup() unnecessarily, so the behavior of dead interrupt or
ariff [Mon, 2 Apr 2007 03:03:06 +0000 (03:03 +0000)]
- Don't wakeup() unnecessarily, so the behavior of dead interrupt or
  stalled DMA engine can be observed and predicted.
- Minor sysctl/tunable cleanup.

17 years agoTemporarily desupport simultaneous target and initiator mode.
mjacob [Mon, 2 Apr 2007 01:04:20 +0000 (01:04 +0000)]
Temporarily desupport simultaneous target and initiator mode.

When the linux port changes were imported which split the
target command list to be separate from the initiator command
list and the handle format changed to encode a type in the handle
the implications to the function isp_handle_index (which only
the NetBSD/OpenBSD/FreeBSD ports use) were overlooked.

The fault is twofold: first, the index into the DMA maps
in  isp_pci is wrong because a target command handle with
the type bit left in place caused a bad index (and panic)
into dma map. Secondly, the assumption of the array
of DMA maps in either PCS or SBUS attachment structures is
that there is a linear mapping between handle index and
DMA map index. This can no longer be true if there are
overlapping index spaces for initiator mode and target
mode commands.

These changes bandaid around the problem by forcing us
to not have simultaneous dual roles and doing the appropriate
masking to make sure things are indexed correctly. A longer
term fix is being devloped.

17 years agoFix type-punned pointer, minor style fixes.
kientzle [Mon, 2 Apr 2007 00:41:37 +0000 (00:41 +0000)]
Fix type-punned pointer, minor style fixes.
Thanks to: Joerg Sonnenberger

17 years agoAdd myself
yongari [Mon, 2 Apr 2007 00:39:29 +0000 (00:39 +0000)]
Add myself

17 years agoRemove unused variable; use consistent types.
kientzle [Mon, 2 Apr 2007 00:34:36 +0000 (00:34 +0000)]
Remove unused variable; use consistent types.
Thanks to: Joerg Sonnenberger

17 years agoBe consistent: file flags are unsigned bitmaps.
kientzle [Mon, 2 Apr 2007 00:32:03 +0000 (00:32 +0000)]
Be consistent: file flags are unsigned bitmaps.
Thanks to: Joerg Sonnenberger

17 years agoDon't compare a signed char to 0xFF.
kientzle [Mon, 2 Apr 2007 00:29:52 +0000 (00:29 +0000)]
Don't compare a signed char to 0xFF.
Thanks to: Joerg Sonnenberger

17 years agoAvoid a potential overflow when 'skip' is larger than a pointer.
kientzle [Mon, 2 Apr 2007 00:25:11 +0000 (00:25 +0000)]
Avoid a potential overflow when 'skip' is larger than a pointer.
Thanks to: Joerg Sonnenberger

17 years agoStyle fix: Use the correct type for 'bytes_to_write'.
kientzle [Mon, 2 Apr 2007 00:21:46 +0000 (00:21 +0000)]
Style fix:  Use the correct type for 'bytes_to_write'.

Thanks to: Joerg Sonnenberger

17 years agoStyle: bare "unsigned" is deprecated, use "unsigned int" instead.
kientzle [Mon, 2 Apr 2007 00:15:45 +0000 (00:15 +0000)]
Style:  bare "unsigned" is deprecated, use "unsigned int" instead.

Thanks to: Joerg Sonnenberger

17 years agoRemove some unused fields from archive_read internal structure.
kientzle [Mon, 2 Apr 2007 00:11:54 +0000 (00:11 +0000)]
Remove some unused fields from archive_read internal structure.
(Left over from when read and write used to share this structure.)

17 years agoAdd -m (megabytes) and -g (gigabytes) options. I'm tired of being told
markm [Sun, 1 Apr 2007 20:28:37 +0000 (20:28 +0000)]
Add -m (megabytes) and -g (gigabytes) options. I'm tired of being told
I can't do this.

MFC: 1 month

17 years agoFill in what's left about myself. Noted eik, jesusr, jcamou.
trhodes [Sun, 1 Apr 2007 18:40:26 +0000 (18:40 +0000)]
Fill in what's left about myself.  Noted eik, jesusr, jcamou.

17 years agoNote that the old firmware modules need to be removed.
mlaier [Sun, 1 Apr 2007 17:49:27 +0000 (17:49 +0000)]
Note that the old firmware modules need to be removed.

Reported by: Jeremie Le Hen
Suggested by: Stefan Ehmann

17 years agoHandle errors from bus_setup_intr().
netchild [Sun, 1 Apr 2007 16:55:31 +0000 (16:55 +0000)]
Handle errors from bus_setup_intr().

Found by: Coverity Prevent (tm)
CID: 1066

17 years agoTell the user when the setup of the interrupt handler failed and return
netchild [Sun, 1 Apr 2007 16:52:54 +0000 (16:52 +0000)]
Tell the user when the setup of the interrupt handler failed and return
an error.

Found by: Coverity Prevent (tm)
CID: 71-78

17 years agong_node and ng_worklist locks both migrated from being spinning locks to
wkoszek [Sun, 1 Apr 2007 15:48:10 +0000 (15:48 +0000)]
ng_node and ng_worklist locks both migrated from being spinning locks to
adaptive mutexes. Let witness(4) calm down and bring proper types of those
locks to the lock order database.

Glanced at by: rwatson

17 years agoMore style nits.
pjd [Sun, 1 Apr 2007 15:40:56 +0000 (15:40 +0000)]
More style nits.

17 years agoAdd myself
ache [Sun, 1 Apr 2007 14:52:51 +0000 (14:52 +0000)]
Add myself

17 years agoTell a statistic checker that not checking the return value of the probing
netchild [Sun, 1 Apr 2007 14:15:26 +0000 (14:15 +0000)]
Tell a statistic checker that not checking the return value of the probing
of the mii phy is intended for this chip.

Found by: Coverity Prevent (tm)
CID: 43

17 years agoMake it obvious that we don't care about the return value of
netchild [Sun, 1 Apr 2007 13:46:39 +0000 (13:46 +0000)]
Make it obvious that we don't care about the return value of
usbd_endpoint_count(), the failure case is handled implicit in the
following code.

Found by: Coverity Prevent (tm)
CID: 56

17 years agoStyle nit.
pjd [Sun, 1 Apr 2007 13:41:10 +0000 (13:41 +0000)]
Style nit.

17 years agoThis trivial change should fix at least 3 similar bugs. All of
yar [Sun, 1 Apr 2007 13:25:03 +0000 (13:25 +0000)]
This trivial change should fix at least 3 similar bugs.  All of
them are related to the `c' function's need to know if we are at
the actual end of the address range.  (It must print the text not
earlier than the whole pattern space was deleted.)  It appears the
only sed function with this requirement.

There is `lastaddr' set by applies(), which is to notify the `c'
function, but it can't always help because it's false when we are
hitting the end of file early.  There is also a bug in applies()
due to which `lastaddr' isn't set to true on degenerate ranges such
as `$,$' or `N,$' if N appears the last line number.

Handling early EOF condition in applies() could look more logical,
but it would effectively revert sed to the unreasonable behaviour
rev. 1.26 of main.c fought against, as it would require lastline()
be called for each line within each address range.  So it's better
to call lastline() only if needed by the `c' function.

Together with this change to sed go regression tests for the bugs
fixed (c1-c3).  A basic test of `c' (c0) is also added as it helped
me to spot my own error.

Discussed with: dds
Tested by: the regression tests
MFC after: 1 week

17 years agoAdd my first (and only until now) mentee Dryice Dong Liu (dryice).
itetcu [Sun, 1 Apr 2007 13:14:38 +0000 (13:14 +0000)]
Add my first (and only until now) mentee Dryice Dong Liu (dryice).

17 years agoAdd myself.
itetcu [Sun, 1 Apr 2007 13:11:50 +0000 (13:11 +0000)]
Add myself.

17 years agoAdd my mentor Sam Lawrance (my other mentor Tom McLaughlin is already in the
itetcu [Sun, 1 Apr 2007 13:08:57 +0000 (13:08 +0000)]
Add my mentor Sam Lawrance (my other mentor Tom McLaughlin is already in the
file).

17 years agoI think the code I'm removing here is completely bogus.
pjd [Sun, 1 Apr 2007 13:08:05 +0000 (13:08 +0000)]
I think the code I'm removing here is completely bogus.
vfs_flags field is used for VFCF_* flags which are given at file system
driver creation time (via VFS_SET(9)) macro.

What this code did was bascially this:

If file system registers itself with VFCF_UNICODE flag (stores file names
as Unicode), it will gain MNT_SOFTDEP flag (UFS soft-updates).

If file system registers itself with VFCF_LOOPBACK flag (aliases some other
mounted FS), it will gain MNT_SUIDDIR flag (special handling of SUID on
dirs).

The latter will be quite dangerous, but those flags are reset later in
vfs_domount().

MFC after: 1 month

17 years agoChange #include <machine/pcpu.h> to #include <sys/pcpu.h>
rodrigc [Sun, 1 Apr 2007 12:48:10 +0000 (12:48 +0000)]
Change #include <machine/pcpu.h> to #include <sys/pcpu.h>
to get definition of curthread, required by <sys/sx.h>.

17 years agoIf nooption SMP on powerpc, also nooption ADAPTIVE_SX, which depends on
rwatson [Sun, 1 Apr 2007 11:10:16 +0000 (11:10 +0000)]
If nooption SMP on powerpc, also nooption ADAPTIVE_SX, which depends on
SMP and is now in the global NOTES.

17 years agoNow that the vdropl() function is public, assert that the vnode interlock
pjd [Sun, 1 Apr 2007 10:45:32 +0000 (10:45 +0000)]
Now that the vdropl() function is public, assert that the vnode interlock
is held.

17 years agoo Document vdropl(9) [1].
maxim [Sun, 1 Apr 2007 09:48:59 +0000 (09:48 +0000)]
o Document vdropl(9) [1].
o Add an MLINK for vdropl().

Reviewed by: des [1]
Obtained from: wording from vgone(9)

17 years agoAdd Inoue-san and Shiozaki-san.
ume [Sun, 1 Apr 2007 08:48:00 +0000 (08:48 +0000)]
Add Inoue-san and Shiozaki-san.

17 years agoAdd bge(4).
marcel [Sun, 1 Apr 2007 06:24:19 +0000 (06:24 +0000)]
Add bge(4).
Fix a white-space nit while I'm here.

17 years agoWhen writing to PCI configuration registers, don't immediately
marcel [Sun, 1 Apr 2007 06:15:53 +0000 (06:15 +0000)]
When writing to PCI configuration registers, don't immediately
read the same register back. It can cause hangs or machine
checks in certain cases. One particular case is with bge(4)
when a reset is initiated for the controller.

MFC after: 1 month

17 years agoAdd Suzuki-san and myself.
ume [Sun, 1 Apr 2007 05:53:52 +0000 (05:53 +0000)]
Add Suzuki-san and myself.

17 years agoAdd myself and grehan.
benno [Sun, 1 Apr 2007 01:11:50 +0000 (01:11 +0000)]
Add myself and grehan.

17 years agoFix typo
imp [Sun, 1 Apr 2007 00:47:08 +0000 (00:47 +0000)]
Fix typo

17 years agoRemove unused file.
marcel [Sun, 1 Apr 2007 00:41:01 +0000 (00:41 +0000)]
Remove unused file.

17 years agoMake vdropl() public; zfs needs it. There is also plenty of existing
des [Sat, 31 Mar 2007 23:57:17 +0000 (23:57 +0000)]
Make vdropl() public; zfs needs it.  There is also plenty of existing
file system code (mostly *_reclaim()) which look like this:

    VOP_LOCK(vp);
    /* examine vp */
    VOP_UNLOCK(vp);
    vdrop(vp);

This can now be rewritten to:

    VOP_LOCK(vp);
    /* examine vp */
    vdropl(vp); /* will unlock vp */

MFC after: 1 week

17 years agoOptimize sx locks to use simple atomic operations for the common cases of
jhb [Sat, 31 Mar 2007 23:23:42 +0000 (23:23 +0000)]
Optimize sx locks to use simple atomic operations for the common cases of
obtaining and releasing shared and exclusive locks.  The algorithms for
manipulating the lock cookie are very similar to that rwlocks.  This patch
also adds support for exclusive locks using the same algorithm as mutexes.

A new sx_init_flags() function has been added so that optional flags can be
specified to alter a given locks behavior.  The flags include SX_DUPOK,
SX_NOWITNESS, SX_NOPROFILE, and SX_QUITE which are all identical in nature
to the similar flags for mutexes.

Adaptive spinning on select locks may be enabled by enabling the
ADAPTIVE_SX kernel option.  Only locks initialized with the SX_ADAPTIVESPIN
flag via sx_init_flags() will adaptively spin.

The common cases for sx_slock(), sx_sunlock(), sx_xlock(), and sx_xunlock()
are now performed inline in non-debug kernels.  As a result, <sys/sx.h> now
requires <sys/lock.h> to be included prior to <sys/sx.h>.

The new kernel option SX_NOINLINE can be used to disable the aforementioned
inlining in non-debug kernels.

The size of struct sx has changed, so the kernel ABI is probably greatly
disturbed.

MFC after: 1 month
Submitted by: attilio
Tested by: kris, pjd

17 years agooops, another missed file from crypto api change
sam [Sat, 31 Mar 2007 23:15:11 +0000 (23:15 +0000)]
oops, another missed file from crypto api change

17 years agoProvide a dummy compression-layer skip function which just reads data and
cperciva [Sat, 31 Mar 2007 22:59:43 +0000 (22:59 +0000)]
Provide a dummy compression-layer skip function which just reads data and
discards it, for use when the compression layer code doesn't know how to
skip data (e.g., everything other than the "none" compressor).  This makes
format level code simpler because that code can now assume that the
compression layer always knows how to skip and will always skip exactly
the requested number of bytes.

Discussed with: kientzle (3 months ago)

17 years ago- Add retire dates for alumni where possible
gabor [Sat, 31 Mar 2007 22:55:56 +0000 (22:55 +0000)]
- Add retire dates for alumni where possible

17 years agoAdd myself, mentor and mentee here
mat [Sat, 31 Mar 2007 22:45:48 +0000 (22:45 +0000)]
Add myself, mentor and mentee here

17 years agoMake vfs_mount_destroy() and vfs_freeopts() non-static, I'd like to use them.
pjd [Sat, 31 Mar 2007 22:44:45 +0000 (22:44 +0000)]
Make vfs_mount_destroy() and vfs_freeopts() non-static, I'd like to use them.

17 years ago- Extend incomplete entries for asmodai, bmilekic, brooks, csjp, gibbs,
gabor [Sat, 31 Mar 2007 22:42:27 +0000 (22:42 +0000)]
- Extend incomplete entries for asmodai, bmilekic, brooks, csjp, gibbs,
  jayanth, peter, ps

17 years agoFlesh in some more of my mentees. Add scottl and some other misc bits
rwatson [Sat, 31 Mar 2007 22:28:59 +0000 (22:28 +0000)]
Flesh in some more of my mentees.  Add scottl and some other misc bits
of history.

17 years ago- Fix a localized character. Please only use ASCII here, otherwise dot won't
gabor [Sat, 31 Mar 2007 22:20:47 +0000 (22:20 +0000)]
- Fix a localized character. Please only use ASCII here, otherwise dot won't
  be able to produce the output

17 years ago- Extend the incomplete entries for knu, daichi, maho, nobutaka
gabor [Sat, 31 Mar 2007 22:12:45 +0000 (22:12 +0000)]
- Extend the incomplete entries for knu, daichi, maho, nobutaka

17 years agoAdd tests for "bsdtar -t". These are useful primarily because they test
cperciva [Sat, 31 Mar 2007 22:04:36 +0000 (22:04 +0000)]
Add tests for "bsdtar -t".  These are useful primarily because they test
the archive_read_data_skip code.

17 years agoFix bug in njl's email address.
philip [Sat, 31 Mar 2007 21:56:36 +0000 (21:56 +0000)]
Fix bug in njl's email address.

17 years agoAdd relationship between my mentor and my grand-mentor. While here, add
philip [Sat, 31 Mar 2007 21:50:52 +0000 (21:50 +0000)]
Add relationship between my mentor and my grand-mentor.  While here, add
brother committers marks en sepotvin.

This is fun. :-)

17 years agoAdd constants for the fields in a BAR. Also, add two new macros
jhb [Sat, 31 Mar 2007 21:39:02 +0000 (21:39 +0000)]
Add constants for the fields in a BAR.  Also, add two new macros
PCI_BAR_(IO|MEM)() that return true if the passed in value from a BAR
is for an IO or memory BAR, respectively.

Reviewed by: imp

17 years agoAdd my only ports mentee so far, koitsu.
philip [Sat, 31 Mar 2007 21:13:29 +0000 (21:13 +0000)]
Add my only ports mentee so far, koitsu.

17 years agoFix compilation problem (add a const) for pre-7.0 compiles.
mjacob [Sat, 31 Mar 2007 21:01:35 +0000 (21:01 +0000)]
Fix compilation problem (add a const) for pre-7.0 compiles.

17 years agoForced commit to note that the previous commit log was wrong for this
jhb [Sat, 31 Mar 2007 20:59:11 +0000 (20:59 +0000)]
Forced commit to note that the previous commit log was wrong for this
file.  The previous commit fixed a bug in pci_pending_msix() where it
mis-calculated the word in the PBA to read to see if a message was
pending.  Nothing actually uses pci_pending_msix() yet though, so this
didn't result in any visible problems.

MFC after: 1 week

17 years ago- Add missing constants for subclasses.
jhb [Sat, 31 Mar 2007 20:41:00 +0000 (20:41 +0000)]
- Add missing constants for subclasses.
- Add a few progif constants as well.

17 years agoLink myself up with my mentors.
bmah [Sat, 31 Mar 2007 20:27:31 +0000 (20:27 +0000)]
Link myself up with my mentors.

17 years agoAdd myself
gshapiro [Sat, 31 Mar 2007 20:18:32 +0000 (20:18 +0000)]
Add myself

17 years agoMove dg back to the active list
imp [Sat, 31 Mar 2007 20:09:44 +0000 (20:09 +0000)]
Move dg back to the active list
Use 'alumni' jkh entry for active jkh (he's stilla round)
move my alumni mentees to that section
sort alumni section

17 years agoA few more chicken-scratchings.
rwatson [Sat, 31 Mar 2007 19:53:08 +0000 (19:53 +0000)]
A few more chicken-scratchings.

17 years agoAdd a pseudo-entry for "Birth of FreeBSD"
phk [Sat, 31 Mar 2007 19:44:27 +0000 (19:44 +0000)]
Add a pseudo-entry for "Birth of FreeBSD"
Add entries for the six persons listed in the first version of
commit.allow in the old FreeBSD 1 CVS repo.
Correct joergs date, and link him to jkh.
Add my own entry.

17 years agoFlesh out some information on my mentees and indirect mentees. Flesh out
rwatson [Sat, 31 Mar 2007 19:36:51 +0000 (19:36 +0000)]
Flesh out some information on my mentees and indirect mentees.  Flesh out
jake's family tree a bit also, since he's inactive these days but
mentored quite a few interesting people.

17 years agoAdd all my mentees, both formal as well as the ones that I may have
imp [Sat, 31 Mar 2007 19:21:43 +0000 (19:21 +0000)]
Add all my mentees, both formal as well as the ones that I may have
just been the trigger man for core@'s approval in access.  Who knew
there were 21 of them...

17 years agoToss in my two bits.
des [Sat, 31 Mar 2007 19:18:22 +0000 (19:18 +0000)]
Toss in my two bits.

17 years agoAdd blackend to the team too, as mentored by me
keramida [Sat, 31 Mar 2007 19:17:24 +0000 (19:17 +0000)]
Add blackend to the team too, as mentored by me

17 years agoAdd myself and my mentor apparent (jkh), although there was no mentoring
imp [Sat, 31 Mar 2007 18:52:29 +0000 (18:52 +0000)]
Add myself and my mentor apparent (jkh), although there was no mentoring
at the time, as such.  I'll add my mentees as soon as I can tease that
information from cvs.

17 years agoAdd my bits.
mjacob [Sat, 31 Mar 2007 18:46:23 +0000 (18:46 +0000)]
Add my bits.

17 years agoFix bug in matteo entry.
rwatson [Sat, 31 Mar 2007 17:32:47 +0000 (17:32 +0000)]
Fix bug in matteo entry.

17 years agoadd myself
wilko [Sat, 31 Mar 2007 17:28:45 +0000 (17:28 +0000)]
add myself

17 years agoFix a typo in rrs's entry.
rwatson [Sat, 31 Mar 2007 17:26:54 +0000 (17:26 +0000)]
Fix a typo in rrs's entry.

Document eivind's grand-mentorships unto the second generation via me, as
well as some related loose ends.

17 years agoRather than ignoring any error return from getnewvnode() in nameiinit(),
rwatson [Sat, 31 Mar 2007 16:08:50 +0000 (16:08 +0000)]
Rather than ignoring any error return from getnewvnode() in nameiinit(),
explicitly test and panic.  This should not ever happen, but if it does,
this is a preferred failure mode to a NULL pointer dereference in kernel.

Coverity CID: 1716
Found with: Coverity Prevent(tm)

17 years agoWe don't need spinning locks here. Change them to the adaptive mutexes. This
wkoszek [Sat, 31 Mar 2007 15:43:06 +0000 (15:43 +0000)]
We don't need spinning locks here. Change them to the adaptive mutexes. This
change should bring no performance decrease, as it did not in my tests.

Reviewed by: julian, glebius
Approved by: cognet (mentor)

17 years agoAdd myself and knu mafia I know:-).
nork [Sat, 31 Mar 2007 15:16:38 +0000 (15:16 +0000)]
Add myself and knu mafia I know:-).