Beet mode looks for the beet pseudo header after the outer IP header,
which is wrong since that is followed by the ESP header. Additionally
it needs to adjust the packet length after removing the pseudo header
and point the data pointer to the real data location.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Beet mode decapsulation fails to properly set up the skb pointers, which
only works by accident in combination with CONFIG_NETFILTER, since in that
case the skb is fixed up in xfrm4_input before passing it to the netfilter
hooks.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
draft-nikander-esp-beet-mode-07.txt states "The padding MUST be filled
with NOP options as defined in Internet Protocol [1] section 3.1
Internet header format.", so do that.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Beet mode calculates an incorrect value for the transport header location
when IP options are present, resulting in encapsulation errors.
The correct location is 4 or 8 bytes before the end of the original IP
header, depending on whether the pseudo header is padded.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Up until this point we've accepted replay window settings greater than
32 but our bit mask can only accomodate 32 packets. Thus any packet
with a sequence number within the window but outside the bit mask would
be accepted.
This patch causes those packets to be rejected instead.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Incoming trancated packets are counted as not only InTruncatedPkts but
also InHdrErrors. They should be counted as InTruncatedPkts only.
Signed-off-by: Mitsuru Chinen <mitch@linux.vnet.ibm.com>
Acked-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
When we receive an AppleTalk frame shorter than what its header says,
we still attempt to verify its checksum, and trip on the BUG_ON() at
the end of function atalk_sum_skb() because of the length mismatch.
This has security implications because this can be triggered by simply
sending a specially crafted ethernet frame to a target victim,
effectively crashing that host. Thus this qualifies, I think, as a
remote DoS. Here is the frame I used to trigger the crash, in npg
format:
<Appletalk Killer>
{
# Ethernet header -----
XX XX XX XX XX XX # Destination MAC
00 00 00 00 00 00 # Source MAC
00 1D # Length
# LLC header -----
AA AA 03
08 00 07 80 9B # Appletalk
# Appletalk header -----
00 1B # Packet length (invalid)
00 01 # Fake checksum
00 00 00 00 # Destination and source networks
00 00 00 00 # Destination and source nodes and ports
# Payload -----
0C 0D 0E 0F 10 11 12 13
14
}
The destination MAC address must be set to those of the victim.
The severity is mitigated by two requirements:
* The target host must have the appletalk kernel module loaded. I
suspect this isn't so frequent.
* AppleTalk frames are non-IP, thus I guess they can only travel on
local networks. I am no network expert though, maybe it is possible
to somehow encapsulate AppleTalk packets over IP.
The bug has been reported back in June 2004:
http://bugzilla.kernel.org/show_bug.cgi?id=2979
But it wasn't investigated, and was closed in July 2006 as both
reporters had vanished meanwhile.
This code was new in kernel 2.6.0-test5:
http://git.kernel.org/?p=linux/kernel/git/tglx/history.git;a=commitdiff;h=7ab442d7e0a76402c12553ee256f756097cae2d2
And not modified since then, so we can assume that vanilla kernels
2.6.0-test5 and later, and distribution kernels based thereon, are
affected.
Note that I still do not know for sure what triggered the bug in the
real-world cases. The frame could have been corrupted by the kernel if
we have a bug hiding somewhere. But more likely, we are receiving the
faulty frame from the network.
Signed-off-by: Jean Delvare <jdelvare@suse.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Just a one-byter for an ia64 thinko/typo - already fixed for i386 and x86_64.
Acked-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
drivers/tc/zs.c:73:24: error: asm/dec/tc.h: No such file or directory
Signed-off-by: Yoichi Yuasa <yoichi_yuasa@tripeaks.co.jp>
Acked-by: Maciej W. Rozycki <macro@linux-mips.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In debugging a problem w/ the -rt tree, I noticed that on systems that mark
the tsc as unstable before it is registered, the TSC would still be
selected and used for a short period of time. Digging in it looks to be a
result of the mix of the clocksource list changes and my clocksource
initialization changes.
With the -rt tree, using a bad TSC, even for a short period of time can
results in a hang at boot. I was not able to reproduce this hang w/
mainline, but I'm not completely certain that someone won't trip on it.
This patch resolves the issue by initializing the jiffies clocksource
earlier so a bad TSC won't get selected just because nothing else is yet
registered.
Signed-off-by: John Stultz <johnstul@us.ibm.com>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This patch adds initialization of drv->cylinders back into the failing case in
cciss_geometry_inquiry. I inadvertently removed it in one my 2TB updates.
Signed-off-by: Mike Miller <mike.miller@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Revert all this. It can cause device-mapper to receive a different major from
earlier kernels and it turns out that the Amanda backup program (via GNU tar,
apparently) checks major numbers on files when performing incremental backups.
Which is a bit broken of Amanda (or tar), but this feature isn't important
enough to justify the churn.
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Fix a bug in the swsusp's memory shrinker that causes some systems using
highmem to refuse to suspend to disk if image_size is set above 1/2 of
available RAM.
Special thanks to Jiri Slaby for reporting the problem and assistance in
debugging it.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Cc: Jiri Slaby <jirislaby@gmail.com>
Cc: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The return value of kernel_recvmsg() should be assigned to "err", not
compared with the random value of a never initialized "err" (and the "< 0"
check wrongly always returned false since == comparisons never have a
result < 0).
Spotted by the Coverity checker.
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Acked-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
A device can be removed from an md array via e.g.
echo remove > /sys/block/md3/md/dev-sde/state
This will try to remove the 'dev-sde' subtree which will deadlock
since
commit e7b0d26a86
With this patch we run the kobject_del via schedule_work so as to
avoid the deadlock.
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This patch adds 2 missing symbol exports: jiffies_to_timeval() and
timeval_to_jiffies(). The (not yet merged) dm-raid4-5 module will need
them, and they used to be indirectly exported by virtue of being inline
functions.
Commit 8b9365d753 ("[PATCH] Uninline
jiffies.h functions") uninlined them, and thus modules now need them
explicitly exported to use them.
Signed-off-by: Thomas Bittermann <t.bittermann@online.de>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: john stultz <johnstul@us.ibm.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'for-linus' of git://git390.osdl.marist.edu/pub/scm/linux-2.6:
[S390] cio: Fix handling of interrupt for csch().
[S390] page_mkclean data corruption.
The generic networking code ensures that no two networking devices
have the same name, so there is no time except when sysfs has
implementation bugs that device_rename when called from
dev_change_name will fail.
The current error handling for errors from device_rename in
dev_change_name is wrong and results in an unusable and unrecoverable
network device if device_rename is happens to return an error.
This patch removes the buggy error handling. Which confines the mess
when device_rename hits a problem to sysfs, instead of propagating it
the rest of the network stack. Making linux a little more robust.
Without this patch you can observe what happens when sysfs has a bug
when CONFIG_SYSFS_DEPRECATED is not set and you attempt to rename
a real network device to a name like (broken_parity_status, device,
modalias, power, resource2, subsystem_vendor, class, driver, irq,
msi_bus, resource, subsystem, uevent, config, enable, local_cpus,
numa_node, resource0, subsystem_device, vendor)
Greg has a patch that fixes the sysfs bugs but he doesn't trust it
for a 2.6.21 timeframe. This patch which just ignores errors should
be safe and it keeps the system from going completely wacky.
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Mention the slab name when listing corrupt objects. Although the function
that released the memory is mentioned, that is frequently ambiguous as such
functions often release several pieces of memory.
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/libata-dev:
libata: Limit ATAPI DMA to R/W commands only for TORiSAN DVD drives (take 3)
libata: Limit max sector to 128 for TORiSAN DVD drives (take 3)
libata: Clear tf before doing request sense (take 3)
libata: reorder HSM_ST_FIRST for easier decoding (take 3)
libata bugfix: preserve LBA bit for HDIO_DRIVE_TASK
2.6.21 fix lba48 bug in libata fill_result_tf()
This adds some NCQ blacklist entries taken from the Silicon Image 3124/3132
Windows driver .inf files. There are some confirming reports of problems
with these drives under Linux (for example http://lkml.org/lkml/2007/3/4/178)
so let's disable NCQ on these drives.
[ I'm personally starting to wonder whether we shouldn't disable NCQ by
default, and perhaps have a white-list. There seems to be a *lot* of
drives that do this wrong.. - Linus ]
Signed-off-by: Robert Hancock <hancockr@shaw.ca>
Acked-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6:
r8169: fix suspend/resume for down interface
r8169: issue request_irq after the private data are completely initialized
b44: fix IFF_ALLMULTI handling of CAM slots
cxgb3 - Firwmare update
cxgb3 - Tighten xgmac workaround
cxgb3 - detect NIC only adapters
cxgb3 - Safeguard TCAM size usage
Wipe internal irb if the clear function bit is set before accumulating
bits from the irb in order to follow hardware behaviour.
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
The git commit c2fda5fed8 which
added the page_test_and_clear_dirty call to page_mkclean and the
git commit 7658cc2892 which fixes
the "nasty and subtle race in shared mmap'ed page writeback"
problem in clear_page_dirty_for_io cause data corruption on s390.
The effect of the two changes is that for every call to
clear_page_dirty_for_io a page_test_and_clear_dirty is done. If
the per page dirty bit is set set_page_dirty is called. Strangly
clear_page_dirty_for_io is called for not-uptodate pages, e.g.
over this call-chain:
[<000000000007c0f2>] clear_page_dirty_for_io+0x12a/0x130
[<000000000007c494>] generic_writepages+0x258/0x3e0
[<000000000007c692>] do_writepages+0x76/0x7c
[<00000000000c7a26>] __writeback_single_inode+0xba/0x3e4
[<00000000000c831a>] sync_sb_inodes+0x23e/0x398
[<00000000000c8802>] writeback_inodes+0x12e/0x140
[<000000000007b9ee>] wb_kupdate+0xd2/0x178
[<000000000007cca2>] pdflush+0x162/0x23c
The bad news now is that page_test_and_clear_dirty might claim
that a not-uptodate page is dirty since SetPageUptodate which
resets the per page dirty bit has not yet been called. The page
writeback that follows clobbers the data on disk.
The simplest solution to this problem is to move the call to
page_test_and_clear_dirty under the "if (page_mapped(page))".
If a file backed page is mapped it is uptodate.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
patch 4/4:
Limit ATAPI DMA to R/W commands only for TORiSAN DRD-N216 DVD-ROM drives
(http://bugzilla.kernel.org/show_bug.cgi?id=6710)
Signed-off-by: Albert Lee <albertcc@tw.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
patch 3/4:
The TORiSAN drive locks up when max sector == 256.
Limit max sector to 128 for the TORiSAN DRD-N216 drives.
(http://bugzilla.kernel.org/show_bug.cgi?id=6710)
Signed-off-by: Albert Lee <albertcc@tw.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
patch 2/4:
Clear tf before doing request sense.
This fixes the AOpen 56X/AKH timeout problem.
(http://bugzilla.kernel.org/show_bug.cgi?id=8244)
Signed-off-by: Albert Lee <albertcc@tw.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
patch 1/4:
Reorder HSM_ST_FIRST, such that the task state transition is easier decoded with human eyes.
Signed-off-by: Albert Lee <albertcc@tw.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Preserve the LBA bit in the DevSel/Head register for HDIO_DRIVE_TASK.
Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Current 2.6.21 libata does the following:
void ata_tf_read(struct ata_port *ap, struct ata_taskfile *tf)
{
struct ata_ioports *ioaddr = &ap->ioaddr;
tf->command = ata_check_status(ap);
...
if (tf->flags & ATA_TFLAG_LBA48) {
iowrite8(tf->ctl | ATA_HOB, ioaddr->ctl_addr);
tf->hob_feature = ioread8(ioaddr->error_addr);
...
}
}
...
static void fill_result_tf(struct ata_queued_cmd *qc)
{
struct ata_port *ap = qc->ap;
ap->ops->tf_read(ap, &qc->result_tf);
qc->result_tf.flags = qc->tf.flags;
}
Based on this, those last two statements fill_result_tf()
appear to me to be in the wrong order, in that the tf->flags
are uninitialized at the point where tf_read() is invoked.
So for lba48 commands, tf_read() won't be reading back the
full lba48 register contents..
Correct?
This patch corrects fill_result_tf() so that the flags
get copied to result_tf before they are used by tf_read().
Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
The PM hooks are no-op if the r8169 interface is down (i.e. !IFF_UP).
However, as the chipset is enabled, the device will not work after a
suspend/resume cycle. The patch always issue the required PCI suspend
sequence and removes the module unload/reload workaround.
Signed-off-by: Arnaud Patard <apatard@mandriva.com>
Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
The irq handler schedules a NAPI poll request unconditionally as soon as
the status register is not clean. It has been there - and wrong - for
ages but a recent timing change made it apparently easier to trigger.
Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Cc: Jay Cliburn <jacliburn@bellsouth.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
If you set the IFF_ALLMULTI flag on a b44 device, or if you join more than
B44_MCAST_TABLE_SIZE multicast groups, the device will stop receiving unicast
messages. This is because the __b44_set_mac_addr call sets the zeroth CAM
entry to the MAC address of the device, and then the loop at line 1722
proceeds to overwrite it unless the value of i is set by the __b44_load_mcast
call. However, when IFF_ALLMULTI is set, that call is bypassed, leaving i set
to zero.
Fixed by starting the loop at 1 to make it skip the CAM entry for the MAC
address.
Signed-off-by: Bill Helfinstine <bhelf@flitterfly.whirpon.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Run the watchdog task when the link is up.
Flush the XGMAC Tx FIFO when the link drops.
Also remove a statistics update that should have gone
in the previous modification of xgmac.c.
Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Differentiate NIC only adapters from RNICs.
Initialize offload capabilities for RNICs only.
Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
This is a simplified and actually more comprehensive form of a bug
fix from Mitch Williams <mitch.a.williams@intel.com>.
When we mask or unmask a msi-x irqs the writes may be posted because
we are writing to memory mapped region. This means the mask and
unmask don't happen immediately but at some unspecified time in the
future. Which is out of sync with how the mask/unmask logic work
for ioapic irqs.
The practical result is that we get very subtle and hard to track down
irq migration bugs.
This patch performs a read flush after writes to the MSI-X table for mask
and unmask operations. Since the SMP affinity is set while the interrupt
is masked, and since it's unmasked immediately after, no additional flushes
are required in the various affinity setting routines.
The testing by Mitch Williams on his especially problematic system should
still be valid as I have only simplified the code, not changed the
functionality.
We currently have 7 drivers: cciss, mthca, cxgb3, forceth, s2io,
pcie/portdrv_core, and qla2xxx in 2.6.21 that are affected by this
problem when the hardware they driver is plugged into the right slot.
Given the difficulty of reproducing this bug and tracing it down to
anything that even remotely resembles a cause, even if people are
being affected we aren't likely to see many meaningful bug reports, and
the people who see this bug aren't likely to be able to reproduce this
bug in a timely fashion. So it is best to get this problem fixed
as soon as we can so people don't have problems.
Then if people do have a kernel message stating "No irq for vector" we
will know it is yet another novel cause that needs a complete new
investigation.
Cc: Greg KH <greg@kroah.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Acked-by: Mitch Williams <mitch.a.williams@intel.com>
Acked-by: "Siddha, Suresh B" <suresh.b.siddha@intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6:
[SCSI]: Fix scsi_send_eh_cmnd scatterlist handling
[SPARC]: Add unsigned to unused bit field in a.out.h
This fixes a regression caused by commit:
2dc611de5a
The sense buffer code in scsi_send_eh_cmnd was changed to use
alloc_page() and a scatter list, but the sense data copy was not
updated to match so what we actually get in the sense buffer is total
grabage starting with the kernel address of the struct page we got.
Basically the stack frame of scsi_send_eh_cmd() is what ends up
in the sense buffer.
Depending upon how pointers look on a given platform, you can
end up getting sr_ioctl.c errors when you mount a cdrom. If
the CDROM gives a check condition for GPCMD_GET_CONFIGURATION issued
by drivers/cdrom/cdrom.c:cdrom_mmc_profile(), sr_ioctl will
spit out this error message in sr_do_ioctl() with the way pointers
are on sparc64:
default:
printk(KERN_ERR "%s: CDROM (ioctl) error, command: ", cd->cdi.name);
__scsi_print_command(cgc->cmd);
scsi_print_sense_hdr("sr", &sshdr);
err = -EIO;
This is the error Tom Callaway reported in:
http://marc.info/?l=linux-sparc&m=117407453208101&w=2
Anyways, fix this by using page_address(sgl.page) which is OK
because we know this is low-mem due to GFP_ATOMIC.
Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Christoph Hellwig <hch@lst.de>
Add unsigned to unused bit field in a.out.h to make sparse happy.
[ I took care of the sparc64 side as well -DaveM ]
Signed-off-by: Robert Reif <reif@earthlink.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
The nvram dword alignment logic was broken when writing less than 4
bytes on a non-aligned offset. It was missing logic to round the
length to 4 bytes.
The page erase code is also moved so that it is only called when
using non-buffered flash for better code clarity.
Update version to 1.5.7.
Based on initial patch from Tony Cureington <tony.cureington@hp.com>.
Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In article <20070329.142644.70222545.davem@davemloft.net> (at Thu, 29 Mar 2007 14:26:44 -0700 (PDT)), David Miller <davem@davemloft.net> says:
> From: Sridhar Samudrala <sri@us.ibm.com>
> Date: Thu, 29 Mar 2007 14:17:28 -0700
>
> > The check for length in rawv6_sendmsg() is incorrect.
> > As len is an unsigned int, (len < 0) will never be TRUE.
> > I think checking for IPV6_MAXPLEN(65535) is better.
> >
> > Is it possible to send ipv6 jumbo packets using raw
> > sockets? If so, we can remove this check.
>
> I don't see why such a limitation against jumbo would exist,
> does anyone else?
>
> Thanks for catching this Sridhar. A good compiler should simply
> fail to compile "if (x < 0)" when 'x' is an unsigned type, don't
> you think :-)
Dave, we use "int" for returning value,
so we should fix this anyway, IMHO;
we should not allow len > INT_MAX.
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Acked-by: Sridhar Samudrala <sri@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This changes the "not found" error return for the lookup
function to -ESRCH so that it can be distinguished from
the case where a rule or route resulting in -ENETUNREACH
has been found during the search.
It fixes a bug where if DECnet was compiled with routing
support, but no routes were added to the routing table,
it was failing to fall back to endnode routing.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Signed-off-by: Patrick Caulfield <pcaulfie@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
* 'for-linus' of git://one.firstfloor.org/home/andi/git/linux-2.6:
[PATCH] x86: Don't probe for DDC on VBE1.2
[PATCH] x86-64: Increase NMI watchdog probing timeout
[PATCH] x86-64: Let oprofile reserve MSR on all CPUs
[PATCH] x86-64: Disable local APIC timer use on AMD systems with C1E
David S. Miller