> ... which means you fall back to the ordinary routing lookups, but only after
> you have wasted cycles to compute a hash and found out that it doesn't match
> anything in your cache -> in this case I would expect only a degradation in
> performance, not an improvement.
If your normal operating c
Author: yongari
Date: Mon Apr 20 07:13:04 2009
New Revision: 191301
URL: http://svn.freebsd.org/changeset/base/191301
Log:
For RTL8139C+ controllers, have controller handle padding short
checksum offload frames. Software workaround used for broken
controllers(RTL8169, RTL8168, RTL8168B) seem
On Monday 20 April 2009 09:01:25 Kip Macy wrote:
...
> > But it seems to me that CAM lookups are pretty resilient against DoSing
> > by throwing malicious synthetic flows on them, whereas flow caches will
> > melt down easily.
>
> Actually a CAM is a hardware implementation of a hash table. It has
On Sun, 19 Apr 2009, Robert Watson wrote:
Log:
Explicitly include machine/param.h for CACHE_LINE_SIZE.
MFC after: 2 weeks
Modified:
head/libexec/rtld-elf/rtld_lock.c
It is a style bug (unwarranted chumminess with the implementation) to
include (or otherwise refer to) machine/param.h d
On Mon, Apr 20, 2009 at 12:29 AM, Marko Zec wrote:
> On Monday 20 April 2009 09:01:25 Kip Macy wrote:
> ...
>> > But it seems to me that CAM lookups are pretty resilient against DoSing
>> > by throwing malicious synthetic flows on them, whereas flow caches will
>> > melt down easily.
>>
>> Actuall
[snip]
People, you are missing a few points here:
- if you want to run a DFZ router or do not like flowtables, turn
them off. There is a kernel option for that.
- if you use it and it doesn't perform but you have a valid setup,
tune it.
- if you are under a (D)DoS your least problem m
On Sun, 19 Apr 2009, Robert Watson wrote:
Log:
For each architecture, define CACHE_LINE_SHIFT and a derived
CACHE_LINE_SIZE constant. These constants are intended to
over-estimate the cache line size, and be used at compile-time
when a run-time tuning alternative isn't appropriate or
avail
On Monday 20 April 2009 09:47:37 Kip Macy wrote:
> On Mon, Apr 20, 2009 at 12:29 AM, Marko Zec wrote:
> > On Monday 20 April 2009 09:01:25 Kip Macy wrote:
> > ...
> >
> >> > But it seems to me that CAM lookups are pretty resilient against
> >> > DoSing by throwing malicious synthetic flows on them
Author: roberto
Date: Mon Apr 20 09:59:08 2009
New Revision: 191302
URL: http://svn.freebsd.org/changeset/base/191302
Log:
Merge r191298 into HEAD.
Prevent a buffer overflow in ntpq. Patch taken from the PR database
after being committed to the official ntp tree and present in 4.2.4p7-rc
On Mon, 20 Apr 2009, Marko Zec wrote:
... which makes a flow cache a perfect DoS target in any environment, be
it a DFZ or enterprise router or an end host or whatever.
Uhm, assuming that you don't put a limit on the number of flows allocated -
which I do. When you hit the zone limit for flow
On Mon, 20 Apr 2009, Kip Macy wrote:
... which means you fall back to the ordinary routing lookups, but only
after you have wasted cycles to compute a hash and found out that it
doesn't match anything in your cache -> in this case I would expect only a
degradation in performance, not an improv
Author: rwatson
Date: Mon Apr 20 10:35:50 2009
New Revision: 191303
URL: http://svn.freebsd.org/changeset/base/191303
Log:
Prefer to for the definition of
CACHE_LINE_SIZE.
Submitted by: bde
MFC after:2 weeks
Modified:
head/libexec/rtld-elf/rtld_lock.c
Modified: head/libexec/rt
On Mon, 20 Apr 2009, Bruce Evans wrote:
Explicitly include machine/param.h for CACHE_LINE_SIZE.
MFC after: 2 weeks
Modified:
head/libexec/rtld-elf/rtld_lock.c
It is a style bug (unwarranted chumminess with the implementation) to
include (or otherwise refer to) machine/param.h directl
Author: ed
Date: Mon Apr 20 10:40:42 2009
New Revision: 191304
URL: http://svn.freebsd.org/changeset/base/191304
Log:
Get rid of the device index number stored in the sa(4) unit number.
The device index number stored in the unit number of sa(4) devices is
only used to print debug messages
Author: bz
Date: Mon Apr 20 11:22:51 2009
New Revision: 191305
URL: http://svn.freebsd.org/changeset/base/191305
Log:
Back out r186308:
in case of AF_LINK, which the kernel still returns for an RTAX_GATEWAY
as an empty sockaddr_dl in the classic tun case.
Copying the address into the me
Author: bz
Date: Mon Apr 20 11:51:35 2009
New Revision: 191306
URL: http://svn.freebsd.org/changeset/base/191306
Log:
While we currently still seem to have a gateway address, for future
stability, check for gw to be set before adding the flag and the address
to the routing message.
r186
Author: raj
Date: Mon Apr 20 12:40:28 2009
New Revision: 191307
URL: http://svn.freebsd.org/changeset/base/191307
Log:
Provide locking for PowerPC interrupt sources config.
Reviewed by: attilio
Modified:
head/sys/powerpc/powerpc/intr_machdep.c
Modified: head/sys/powerpc/powerpc/intr_ma
Author: stas
Date: Mon Apr 20 12:43:07 2009
New Revision: 191308
URL: http://svn.freebsd.org/changeset/base/191308
Log:
- Properly unlock mutex on failure in channel_trigger. Before this
the function just returned with the mutex held.
MFC after:1 week
Modified:
head/sys/dev/sound
On Sunday 19 April 2009 7:02:50 pm Robert Watson wrote:
> Author: rwatson
> Date: Sun Apr 19 23:02:50 2009
> New Revision: 191291
> URL: http://svn.freebsd.org/changeset/base/191291
>
> Log:
> Now that the kernel defines CACHE_LINE_SIZE in machine/param.h, use
> that definition in the custom l
Author: rwatson
Date: Mon Apr 20 12:59:23 2009
New Revision: 191309
URL: http://svn.freebsd.org/changeset/base/191309
Log:
Don't conditionally define CACHE_LINE_SHIFT, as we anticipate sizing
a fair number of static data structures, making this an unlikely
option to try to change without als
On Mon, 20 Apr 2009, John Baldwin wrote:
Now that the kernel defines CACHE_LINE_SIZE in machine/param.h, use
that definition in the custom locking code for the run-time linker
rather than local definitions.
Perhaps CACHE_LINE_SIZE for ia64 and sparc64 should be 128 since that is
what r
Author: stas
Date: Mon Apr 20 13:15:25 2009
New Revision: 191310
URL: http://svn.freebsd.org/changeset/base/191310
Log:
- Unlock softc mutex on failure.
MFC after:1 week
Modified:
head/sys/dev/sound/pci/envy24ht.c
Modified: head/sys/dev/sound/pci/envy24ht.c
=
On Mon, 20 Apr 2009, Bruce Evans wrote:
+#ifndef CACHE_LINE_SHIFT
+#defineCACHE_LINE_SHIFT6
+#endif
+#defineCACHE_LINE_SIZE (1 << CACHE_LINE_SHIFT)
This still has the bogus ifdef. Overriding the definition is difficult to
do consistently even in a kernel with
Author: rwatson
Date: Mon Apr 20 13:45:39 2009
New Revision: 191311
URL: http://svn.freebsd.org/changeset/base/191311
Log:
In icmp_reflect(), acquire the inteface address list lock when
searching for a source address to use.
MFC after:2 weeks
Reviewed by: bz
Modified:
head/sys/n
Author: trasz
Date: Mon Apr 20 13:53:40 2009
New Revision: 191312
URL: http://svn.freebsd.org/changeset/base/191312
Log:
Adjust route(4) manual page to remove rtm_use field, which was removed
from the route.h some time ago.
Reviewed by: rwatson
Modified:
head/share/man/man4/route.4
M
Author: kib
Date: Mon Apr 20 14:34:55 2009
New Revision: 191313
URL: http://svn.freebsd.org/changeset/base/191313
Log:
On the exit of the child process which parent either set SA_NOCLDWAIT
or ignored SIGCHLD, unconditionally wake up the parent instead of doing
this only when the child is a l
Author: rwatson
Date: Mon Apr 20 14:35:42 2009
New Revision: 191314
URL: http://svn.freebsd.org/changeset/base/191314
Log:
In ip_input(), cache the received mbuf's network interface in a local
variable. Acquire the interface address list lock when iterating over
the interface address list s
Author: kib
Date: Mon Apr 20 14:36:01 2009
New Revision: 191315
URL: http://svn.freebsd.org/changeset/base/191315
Log:
In ufs_checkpath(), recheck that '..' still points to the inode with
the same inode number after VFS_VGET() and relock of the vp. If '..'
changed, redo the lookup. To reduce
Author: bz
Date: Mon Apr 20 14:38:48 2009
New Revision: 191316
URL: http://svn.freebsd.org/changeset/base/191316
Log:
Conditionally add the interface name and address if available
so that a ppp running in `receiver' (server) mode can properly
update routes, for example to update the MTU.
Author: rwatson
Date: Mon Apr 20 14:41:23 2009
New Revision: 191317
URL: http://svn.freebsd.org/changeset/base/191317
Log:
Lock interface address lists before iterating over them in nd6.
MFC after:2 weeks
Modified:
head/sys/netinet6/icmp6.c
Modified: head/sys/netinet6/icmp6.c
==
Author: antoine
Date: Mon Apr 20 15:01:45 2009
New Revision: 191318
URL: http://svn.freebsd.org/changeset/base/191318
Log:
vlan(4) no longer depends on miibus(4).
Reviewed by: jhb@
MFC after:1 month
Modified:
head/share/man/man4/vlan.4
head/sys/conf/NOTES
Modified: head/share/m
Author: kib
Date: Mon Apr 20 15:10:03 2009
New Revision: 191319
URL: http://svn.freebsd.org/changeset/base/191319
Log:
Fix typo.
Noted by: jhb
MFC after:2 weeks
Modified:
head/sys/kern/kern_exit.c
Modified: head/sys/kern/kern_exit.c
Author: ed
Date: Mon Apr 20 15:15:24 2009
New Revision: 191320
URL: http://svn.freebsd.org/changeset/base/191320
Log:
Make mse(4) use si_drv1, instead of using unit numbers.
Discussed with: imp
Modified:
head/sys/dev/mse/mse.c
Modified: head/sys/dev/mse/mse.c
=
Author: imp
Date: Mon Apr 20 15:19:54 2009
New Revision: 191321
URL: http://svn.freebsd.org/changeset/base/191321
Log:
Sometimes we can call ed_detach() before the mtx has been initialized.
Avoid it if it hasn't been initialized.
Modified:
head/sys/dev/ed/if_ed.c
Modified: head/sys/dev/ed/
On Mon, Apr 20, 2009 at 02:00:56PM +0100, Robert Watson wrote:
>
> On Mon, 20 Apr 2009, John Baldwin wrote:
>
> >> Now that the kernel defines CACHE_LINE_SIZE in machine/param.h, use
> >> that definition in the custom locking code for the run-time linker
> >> rather than local definitions.
> >
> Hmm, such a scheme raises suspicion that in your particular case very few flow
> cache lookups could be serviced from CPU caches. 16MB + 80MB sounds to be in
> range with memory footprint of a DFZ table stored in our normal radix tree -
> so where's the benefit of the flow cache?
>
Well its a co
Author: stas
Date: Mon Apr 20 15:47:06 2009
New Revision: 191322
URL: http://svn.freebsd.org/changeset/base/191322
Log:
- Give a warning and start the oscillator if it was not previously
runned.
- Rename ds1672 -> rtc to follow the other drivers.
- Refactor/simplify the code a bit.
Author: rwatson
Date: Mon Apr 20 16:05:16 2009
New Revision: 191323
URL: http://svn.freebsd.org/changeset/base/191323
Log:
Close some but not all writer-writer races when maintaining IPv6
interface address lists by locking the interface address list lock.
MFC after:2 weeks
Modified:
Author: kmacy
Date: Mon Apr 20 16:16:43 2009
New Revision: 191324
URL: http://svn.freebsd.org/changeset/base/191324
Log:
simplify code by removing bit_fns and replacing with the use of a temporary
mask
Modified:
head/sys/net/flowtable.c
Modified: head/sys/net/flowtable.c
===
Author: emax
Date: Mon Apr 20 16:55:48 2009
New Revision: 191325
URL: http://svn.freebsd.org/changeset/base/191325
Log:
MFC r191164
Prevent atkbd(4) interrupt handler from calling keyboard callback function
when polled mode is enabled. This should help with duplicated/missing
characters
Author: emax
Date: Mon Apr 20 16:57:43 2009
New Revision: 191326
URL: http://svn.freebsd.org/changeset/base/191326
Log:
MFC r191232
Fix bug in event timeout handling
Submitted by: mav
Approved by: re (kib)
Modified:
stable/7/usr.sbin/bluetooth/btpand/ (props changed)
stable/7
Author: emax
Date: Mon Apr 20 17:09:53 2009
New Revision: 191327
URL: http://svn.freebsd.org/changeset/base/191327
Log:
MFC r191164
Prevent atkbd(4) interrupt handler from calling keyboard callback function
when polled mode is enabled. This should help with duplicated/missing
characters
Author: emax
Date: Mon Apr 20 17:11:21 2009
New Revision: 191328
URL: http://svn.freebsd.org/changeset/base/191328
Log:
MFC r191232
Fix bug in event timeout handling
Submitted by: mav
Approved by: re (kib)
Modified:
releng/7.2/usr.sbin/bluetooth/btpand/ (props changed)
releng
Author: rdivacky
Date: Mon Apr 20 18:19:38 2009
New Revision: 191330
URL: http://svn.freebsd.org/changeset/base/191330
Log:
Implement highlighting of today in month view of cal/ncal just like
gnu cal does. This is currently disabled for year view because of hard
coded padding in that case. T
Robert Watson wrote:
On Mon, 20 Apr 2009, Bruce Evans wrote:
+#ifndef CACHE_LINE_SHIFT
+#defineCACHE_LINE_SHIFT6
+#endif
+#defineCACHE_LINE_SIZE(1 << CACHE_LINE_SHIFT)
This still has the bogus ifdef. Overriding the definition is
difficult to do consistently even in a ke
Author: rwatson
Date: Mon Apr 20 20:09:55 2009
New Revision: 191335
URL: http://svn.freebsd.org/changeset/base/191335
Log:
Prefer if_addrhead (FreeBSD) to if_addrlist (BSD compat) naming for the
interface address list in if_stf.c.
Acquire interface address list locks around address list a
Author: rwatson
Date: Mon Apr 20 21:05:37 2009
New Revision: 191336
URL: http://svn.freebsd.org/changeset/base/191336
Log:
Use TAILQ_FOREACH() and TAILQ_FOREACH_SAFE() rather than manually
accessing queue(9) structure fields for if_addrhead.
Prefer FreeBSD field name if_addrhead to compat
On Mon, Apr 20, 2009 at 06:19:38PM +, Roman Divacky wrote:
> Author: rdivacky
> Date: Mon Apr 20 18:19:38 2009
> New Revision: 191330
> URL: http://svn.freebsd.org/changeset/base/191330
>
> Log:
> Implement highlighting of today in month view of cal/ncal just like
> gnu cal does. This is c
On Mon, Apr 20, 2009 at 09:16:08PM +, Alexey Dokuchaev wrote:
> On Mon, Apr 20, 2009 at 06:19:38PM +, Roman Divacky wrote:
> > Author: rdivacky
> > Date: Mon Apr 20 18:19:38 2009
> > New Revision: 191330
> > URL: http://svn.freebsd.org/changeset/base/191330
> >
> > Log:
> > Implement hig
Author: rwatson
Date: Mon Apr 20 21:37:46 2009
New Revision: 191337
URL: http://svn.freebsd.org/changeset/base/191337
Log:
Acquire interface address list lock around access to if_addrhead,
closing several writer-writer races, and some read-write races.
MFC after:2 weeks
Modified:
h
On Mon, 20 Apr 2009, Alexey Dokuchaev wrote:
On Mon, Apr 20, 2009 at 06:19:38PM +, Roman Divacky wrote:
Author: rdivacky
Date: Mon Apr 20 18:19:38 2009
New Revision: 191330
URL: http://svn.freebsd.org/changeset/base/191330
Log:
Implement highlighting of today in month view of cal/ncal ju
Author: rwatson
Date: Mon Apr 20 22:40:44 2009
New Revision: 191338
URL: http://svn.freebsd.org/changeset/base/191338
Log:
Prefer actual field names (if_addrhead, ifa_link) to macros aliasing
those field names in FreeBSD code.
MFC after:2 weeks
Modified:
head/sys/netinet/ip_fw2.c
Author: rwatson
Date: Mon Apr 20 22:41:19 2009
New Revision: 191339
URL: http://svn.freebsd.org/changeset/base/191339
Log:
Prefer ifa_link (structure field) to ifa_list (macro alias for it).
MFC after:2 weeks
Modified:
head/sys/net/if_stf.c
Modified: head/sys/net/if_stf.c
==
Author: rwatson
Date: Mon Apr 20 22:45:21 2009
New Revision: 191340
URL: http://svn.freebsd.org/changeset/base/191340
Log:
Prefer structure fields (ifa_link) to macro aliases for them
(ifa_list).
MFC after:2 weeks
Modified:
head/sys/netinet6/icmp6.c
head/sys/netinet6/in6.c
head
Author: rwatson
Date: Mon Apr 20 22:56:34 2009
New Revision: 191341
URL: http://svn.freebsd.org/changeset/base/191341
Log:
Assert the interface address list lock in IFP_TO_IA6(), as it will
iterate the interface address list. Marginally expand IF_ADDR_LOCK()
coverage in mld6.c to make sure
Author: rwatson
Date: Mon Apr 20 23:23:31 2009
New Revision: 191342
URL: http://svn.freebsd.org/changeset/base/191342
Log:
Acquire address list lock before walking an interface's address list to
identify possible jail addresses on it for IPv4 and IPv6.
MFC after:2 weeks
Modified:
h
Author: rwatson
Date: Mon Apr 20 23:25:38 2009
New Revision: 191343
URL: http://svn.freebsd.org/changeset/base/191343
Log:
Acquire interfce address list lock while walking the interface address
list during tun device initialization.
MFC after:2 weeks
Modified:
head/sys/net/if_tun.c
Author: yongari
Date: Tue Apr 21 00:34:31 2009
New Revision: 191344
URL: http://svn.freebsd.org/changeset/base/191344
Log:
Clear IFF_DRV_OACTIVE flag if one of queued packets was transmitted.
Previously it used to clear the flag only when the transmit queue
is empty which may slow down Tx pe
Author: yongari
Date: Tue Apr 21 00:42:11 2009
New Revision: 191345
URL: http://svn.freebsd.org/changeset/base/191345
Log:
To make it easy whether xl(4) missed Tx completion interrupt check
number of queued packets in watchdog timeout handler. If there are
no queued packets just print a info
Are you really sure backing this change out is the right thing to do ??
-- Qing
> -Original Message-
> From: owner-src-committ...@freebsd.org
> [mailto:owner-src-committ...@freebsd.org] On Behalf Of Bjoern A. Zeeb
> Sent: Monday, April 20, 2009 4:23 AM
> To: src-committ...@freebsd.org
Author: takawata
Date: Tue Apr 21 05:06:31 2009
New Revision: 191346
URL: http://svn.freebsd.org/changeset/base/191346
Log:
Add another FTDI serial converter.
Modified:
head/sys/dev/usb/serial/uftdi.c
head/sys/dev/usb/usbdevs
Modified: head/sys/dev/usb/serial/uftdi.c
==
61 matches
Mail list logo
