On Sun, Feb 07, 2021 at 08:13:13PM +0200, Shai Malin wrote:
> Queue Initialization:
> =
> The nvme-tcp-offload ULP module shall register with the existing
> nvmf_transport_ops (.name = "tcp_offload"), nvme_ctrl_ops and blk_mq_ops.
> The nvme-tcp-offload vendor driver shall regi
8021q match the
conditional inclusion in netdevice.h
Signed-off-by: Chris Leech
---
net/8021q/vlan_dev.c | 28
1 file changed, 16 insertions(+), 12 deletions(-)
diff --git a/net/8021q/vlan_dev.c b/net/8021q/vlan_dev.c
index 15293c2a5dd8..8d77b6ee4477 100644
--- a/net
fixes.
Signed-off-by: Chris Leech
---
drivers/scsi/scsi_transport_iscsi.c | 29 ++---
1 file changed, 18 insertions(+), 11 deletions(-)
diff --git a/drivers/scsi/scsi_transport_iscsi.c
b/drivers/scsi/scsi_transport_iscsi.c
index f4b52b44b966..65f6c94f2e9b 100644
--- a/dr
On Tue, Nov 21, 2017 at 11:26:09AM +, David Laight wrote:
> From: Chris Leech
> > Sent: 15 November 2017 00:25
> > To: David Laight
> > Cc: netdev@vger.kernel.org; contain...@lists.linux-foundation.org
> > Subject: Re: [PATCH 0/9] use network namespace for iSCSI con
On Wed, Nov 08, 2017 at 10:31:04AM +, David Laight wrote:
> From: Chris Leech
> > Sent: 07 November 2017 22:45
> >
> > I've posted these changes to allow iSCSI management within a container
> > using a network namespace to the SCSI and Open-iSCSI lists, but se
Make use of the per-net netlink sockets. Responses are sent back on the
same socket/namespace the request was received on. Async events are
reported on the socket/namespace stored in the iscsi_cls_host associated
with the event.
Signed-off-by: Chris Leech
---
drivers/scsi
e iSCSI netlink family and
sysfs objects from the iSCSI transport class.
Thank you,
Chris Leech
---
This series of changes makes the iSCSI netlink and sysfs control
interfaces filtered by network namespace. This is required to run
iscsid in any network namespace other than the initial default one.
This lets iscsi_tcp operate in multiple namespaces. It uses current
during session creation to find the net namespace, but it might be
better to manage to pass it along from the iscsi netlink socket.
Signed-off-by: Chris Leech
---
drivers/scsi/iscsi_tcp.c| 7 +++
drivers/scsi
This makes the iscsi_host, iscsi_session, iscsi_connection, iscsi_iface,
and iscsi_endpoint transport class devices only visible in sysfs under a
matching network namespace. The network namespace for all of these
objects is tracked in the iscsi_cls_host structure.
Signed-off-by: Chris Leech
cleanups after the bus to class conversion
Signed-off-by: Chris Leech
---
drivers/scsi/qla4xxx/ql4_os.c | 52 +-
drivers/scsi/scsi_transport_iscsi.c | 102 ++--
include/scsi/scsi_transport_iscsi.h | 48 +
3 files changed
All internal lookups of iSCSI transport objects need to be filtered by
net namespace.
Signed-off-by: Chris Leech
---
drivers/infiniband/ulp/iser/iscsi_iser.c | 5 +-
drivers/scsi/be2iscsi/be_iscsi.c | 4 +-
drivers/scsi/bnx2i/bnx2i_iscsi.c | 4 +-
drivers/scsi/cxgbi
The flashnode session and connection devices should be filtered by net
namespace along with the iscsi_host, but we can't do that with a bus
device. As these don't use any of the bus matching functionality, they
make more sense as a class device anyway.
Signed-off-by: Chris Leech
--
Prepare iSCSI netlink to operate in multiple namespaces.
Signed-off-by: Chris Leech
---
drivers/scsi/scsi_transport_iscsi.c | 67 +++--
1 file changed, 57 insertions(+), 10 deletions(-)
diff --git a/drivers/scsi/scsi_transport_iscsi.c
b/drivers/scsi
Finished the net namespace support for flashnode sysfs devices
Signed-off-by: Chris Leech
---
drivers/scsi/scsi_transport_iscsi.c | 33 +
1 file changed, 33 insertions(+)
diff --git a/drivers/scsi/scsi_transport_iscsi.c
b/drivers/scsi/scsi_transport_iscsi.c
Right now the iscsi_endpoint is only linked to a connection once that
connection has been established. For net namespace filtering of the
sysfs objects, associate an endpoint with the host that it was
allocated for when it is created.
Signed-off-by: Chris Leech
---
drivers/infiniband/ulp/iser
> situations.
>
> Signed-off-by: Elena Reshetova
> Signed-off-by: Hans Liljestrand
> Signed-off-by: Kees Cook
> Signed-off-by: David Windsor
This looks OK to me.
Acked-by: Chris Leech
> ---
> drivers/scsi/libiscsi.c| 8
> drivers/scsi/qedi/qedi_is
On Wed, May 20, 2015 at 11:45:43AM -0700, Andy Grover wrote:
> On 05/13/2015 03:12 PM, Chris Leech wrote:
> >This is only about the structures and functionality involved in maintaining
> >the
> >iSCSI session, the SCSI host along with it's discovered targets and devices
etwork namespaces.
These patches are functional, but not complete. There's no isolation enforced
in the kernel just yet, so it relies on well behaved userspace. I plan on
fixing that, but wanted some feedback on the idea and approach so far.
Thanks,
Chris
Chris Leech (4):
iscsi:
This lets iscsi_tcp operate in multiple namespaces. It uses current
during session creation to find the net namespace, but it might be
better to manage to pass it along from the iscsi netlink socket.
---
drivers/scsi/iscsi_tcp.c| 7 +++
drivers/scsi/scsi_transport_iscsi.c | 7
g Ethernet driver supports multiple unicast addresses as
well.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
include/linux/netdevice.h |4 ++
net/8021q/vlan_dev.c |7 ++-
net/core/dev.c| 96 +
net/core/dev_mca
Christoph Hellwig wrote:
I just did a very quick glance over the tree. Some extremly highlevel
comments to start with before actually starting the source review:
Thanks for taking a look Christoph
- why do you need your own libcrc? lib/crc32.c has a crc32_le
We shouldn't, but we may want
On 9/21/07, jamal <[EMAIL PROTECTED]> wrote:
> On Fri, 2007-21-09 at 14:34 -0700, Kok, Auke wrote:
>
> > I never saw any bugreports about e1000 not being able to accept vlan packets
> > because of this, so I'm quite certain it works OK, feel free to find me a
> > case
> > where this isn't so :)
>
On 9/21/07, jamal <[EMAIL PROTECTED]> wrote:
> On Fri, 2007-21-09 at 08:43 -0700, Ben Greear wrote:
>
> > I just re-read the spec, and a bridge *may* pad up to 68, but it is not
> > required.
> > On page 166, it says equipment must be able to handle 64 byte minimums.
> >
> > See page 22 (section 7.
On 6/19/07, Stephen Hemminger <[EMAIL PROTECTED]> wrote:
Looks good. I have some changes to allow devices with multiple MAC addresses
(never finished). This device could use that.
Stephen,
Is this patch available somewhere? I'd be interested in taking a look at it.
- Chris
-
To unsubscribe
On Wed, 2007-05-02 at 15:44 -0700, David Miller wrote:
>
> Chrstopher, I really really would like you to post these patches early
> and often to netdev@vger.kernel.org especially because you are
> touching the TCP code.
You're right, I should have sent this to netdev as well. I'm Sorry.
As for
On 4/18/07, David Miller <[EMAIL PROTECTED]> wrote:
Ok, I'll try to remember to high-priority reviewing PJ's patch
on my next rebase of the net-2.6.22 tree which should be
tonight or tomorrow sometime.
Thanks Dave, PJ is offline this week so I'm trying to keep an eye out
for discussions relate
On 4/18/07, David Miller <[EMAIL PROTECTED]> wrote:
Ok, it will give you one level of decapsulation.
What do we tell people who want 2 devices previous? :-)
I can tell you that the intent of PJs patch was to provide the ifindex
of the physical interface that a packet entered the system on,
reg
> This sounds like something that will always be wrong -- or in other
> words, always be right for only the latest CPUs. Can this be made
> dynamic, based on some timing factor?
In fact I think this has been tweaked twice in the vanilla tree
already.
This is actually just the same tweak you re
net/ipv4/tcp.c: In function 'tcp_recvmsg':
net/ipv4/tcp.c:: warning: unused variable 'available'
Signed-off-by: Andrew Morton <[EMAIL PROTECTED]>
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
net/ipv4/tcp.c | 26 --
1 fil
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
Documentation/networking/ip-sysctl.txt |6 ++
1 files changed, 6 insertions(+), 0 deletions(-)
diff --git a/Documentation/networking/ip-sysctl.txt
b/Documentation/networking/ip-sysctl.txt
index d3aae1f..9541691 100644
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
MAINTAINERS | 12
1 files changed, 12 insertions(+), 0 deletions(-)
diff --git a/MAINTAINERS b/MAINTAINERS
index 1dfba85..2dd5d23 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -1156,6 +1156,12 @@ M: [EMAIL PROTECT
The performance wins come with having the DMA copy engine doing the copies
in parallel with the context switch. If there is enough data ready on the
socket at recv time just use a regular copy.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
net/ipv4/tcp.c | 10 +++---
1 files c
Every 20 descriptors turns out to be to few append commands with
newer/faster CPUs. Pushing every 4 still cuts down on MMIO writes to an
acceptable level without letting the DMA engine run out of work.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/ioatdma.c |4 +
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/ioatdma.c| 60 +++
drivers/dma/ioatdma_io.h | 118 --
2 files changed, 28 insertions(+), 150 deletions(-)
diff --git a/drivers/dma/ioatdma.c b/drive
From: Jeff Garzik <[EMAIL PROTECTED]>
Signed-off-by: Jeff Garzik <[EMAIL PROTECTED]>
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/dmaengine.c | 22 --
1 files changed, 20 insertions(+), 2 deletions(-)
diff --git a/drivers/dma/dmaengi
t;[EMAIL PROTECTED]>
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
Signed-off-by: Andrew Morton <[EMAIL PROTECTED]>
---
drivers/dma/ioatdma.c | 13 +
1 files changed, 13 insertions(+), 0 deletions(-)
diff --git a/drivers/dma/ioatdma.c b/drivers/dma/ioatdma.c
index cbf93c
There's only one now anyway, and it's not in a performance path,
so make it behave the same on 32-bit and 64-bit CPUs.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/ioatdma.c | 10 --
1 files changed, 4 insertions(+), 6 deletions(-)
diff --git a/drive
Andrew Morton (1):
I/OAT: warning fix
Chris Leech (6):
ioatdma: Push pending transactions to hardware more frequently
ioatdma: Remove the wrappers around read(bwl)/write(bwl) in ioatdma
ioatdma: Remove the use of writeq from the ioatdma driver
I/OAT: Add documentation for
On 2/5/07, Olaf Kirch <[EMAIL PROTECTED]> wrote:
Nowhere in the dma_async_*complete functions can I see any code
that would sleep if the DMA is not yet complete. Am I missing something,
or are we really busy-waiting on the DMA engine? Wouldn't this kind of
defeat the purpose of freeing up the CP
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/ioatdma.c| 60 +++
drivers/dma/ioatdma_io.h | 118 --
2 files changed, 28 insertions(+), 150 deletions(-)
diff --git a/drivers/dma/ioatdma.c b/drive
There's only one now anyway, and it's not in a performance path,
so make it behave the same on 32-bit and 64-bit CPUs.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/ioatdma.c | 10 --
1 files changed, 4 insertions(+), 6 deletions(-)
diff --git a/drive
From: Jeff Garzik <[EMAIL PROTECTED]>
Signed-off-by: Jeff Garzik <[EMAIL PROTECTED]>
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/dmaengine.c | 22 --
1 files changed, 20 insertions(+), 2 deletions(-)
diff --git a/drivers/dma/dmaengi
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
Documentation/networking/ip-sysctl.txt |6 ++
1 files changed, 6 insertions(+), 0 deletions(-)
diff --git a/Documentation/networking/ip-sysctl.txt
b/Documentation/networking/ip-sysctl.txt
index fd3c0c0..e9ee102 100644
//lost.foo-projects.org/~cleech/linux-2.6 master
--
Chris Leech <[EMAIL PROTECTED]>
I/O Acceleration Technology Software Development
LAN Access Division / Digital Enterprise Group
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECT
The performance wins come with having the DMA copy engine doing the copies
in parallel with the context switch. If there is enough data ready on the
socket at recv time just use a regular copy.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
net/ipv4/tcp.c | 10 +++---
1 files c
Every 20 descriptors turns out to be to few append commands with
newer/faster CPUs. Pushing every 4 still cuts down on MMIO writes to an
acceptable level without letting the DMA engine run out of work.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/ioatdma.c |4 +
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
MAINTAINERS | 12
1 files changed, 12 insertions(+), 0 deletions(-)
diff --git a/MAINTAINERS b/MAINTAINERS
index 5305dd6..533adbe 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -915,6 +915,12 @@ M: [EMAIL PROTECTED]
L:
On 8/18/06, Pavel Machek <[EMAIL PROTECTED]> wrote:
Huh, two version bumps for... ONE ONE-LINER :-).
Could we get rid of embedded version? It helps no one.
Version numbers for drivers that can be built as modules are very
helpful for anyone wanting to upgrade a driver on top of a
distribution
I thought that support statement sounded familiar, large portions of
the source code and documentation are modified from an older release
of e1000. Nothing wrong with that as it's released under the GPL,
except that the copyright statements have mostly just been switched
from Intel to Attansic.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
MAINTAINERS | 12
1 files changed, 12 insertions(+), 0 deletions(-)
diff --git a/MAINTAINERS b/MAINTAINERS
index 21116cc..2d484aa 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -881,6 +881,12 @@ M: [EMAIL PROTECTED]
L:
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
Documentation/networking/ip-sysctl.txt |6 ++
1 files changed, 6 insertions(+), 0 deletions(-)
diff --git a/Documentation/networking/ip-sysctl.txt
b/Documentation/networking/ip-sysctl.txt
index d46338a..841d61e 100644
There's only one now anyway, and it's not in a performance path,
so make it behave the same on 32-bit and 64-bit CPUs.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/ioatdma.c | 10 --
1 files changed, 4 insertions(+), 6 deletions(-)
diff --git a/drive
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
MAINTAINERS | 10 ++
1 files changed, 10 insertions(+), 0 deletions(-)
diff --git a/MAINTAINERS b/MAINTAINERS
index 21116cc..9ae73c9 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -881,6 +881,11 @@ M: [EMAIL PROTECTED]
L:
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/ioatdma.c| 60 +++
drivers/dma/ioatdma_io.h | 118 --
2 files changed, 28 insertions(+), 150 deletions(-)
diff --git a/drivers/dma/ioatdma.c b/drive
The performance wins come with having the DMA copy engine doing the copies
in parallel with the context switch. If there is enough data ready on the
socket at recv time just use a regular copy.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
net/ipv4/tcp.c | 10 +++---
1 files c
Local traffic (loopback) is generally in cache anyway, and the overhead
cost of offloading the copy is worse than just doing it with the CPU.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
net/ipv4/tcp.c |4 +++-
1 files changed, 3 insertions(+), 1 deletions(-)
diff --git a/ne
Every 20 descriptors turns out to be to few append commands with
newer/faster CPUs. Pushing every 4 still cuts down on MMIO writes to an
acceptable level without letting the DMA engine run out of work.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/ioatdma.c |4 +
On 8/3/06, Evgeniy Polyakov <[EMAIL PROTECTED]> wrote:
On Fri, Aug 04, 2006 at 03:59:37PM +1000, Herbert Xu ([EMAIL PROTECTED]) wrote:
> Interesting. Could you guys post figures on alloc_page speed vs. kmalloc?
They probalby measured kmalloc cache access, which only falls to
alloc_pages when ca
On 8/3/06, Evgeniy Polyakov <[EMAIL PROTECTED]> wrote:
> You're changing the size of the buffer without telling the hardware.
> In the interrupt context e1000 knows the size of what was DMAed into
> the skb, but that's after the fact. So e1000 could detect that memory
> was corrupted, but not pr
Maximum e1000 frame is 16128 bytes, which is enough before being rounded
to 16k to have a space for shared info.
My patch just tricks refilling logic to request to allocate slightly less
than was setup when mtu was changed.
The maximum supported MTU size differs between e1000 devices due to
diff
On 8/3/06, Arnd Hannemann <[EMAIL PROTECTED]> wrote:
Well you say "if a single buffer per frame is going to be used". Well,
if I understood you correctly i could set the MTU to, lets say 4000.
Then the driver would enable the "jumbo frame bit" of the hardware, and
allocate only a 4k rx buffer, ri
On 8/3/06, Evgeniy Polyakov <[EMAIL PROTECTED]> wrote:
> Strange, why this skb_shared_info cannon be added before first alignment?
> And what about smaller frames like 1500, does this driver behave similar
> (first align then add)?
It can be.
Could attached (completely untested) patch help?
Add an extra argument to sk_eat_skb, and make it move early copied packets
to the async_wait_queue instead of freeing them.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
include/net/sock.h | 13 -
net/dccp/proto.c |4 ++--
net/ipv4/tcp.c |8
n
Adds a new ioatdma driver
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/Kconfig |9
drivers/dma/Makefile|1
drivers/dma/ioatdma.c | 839 +++
drivers/dma/ioatdma.h | 126 ++
d
Needed to be able to call tcp_cleanup_rbuf in tcp_input.c for I/OAT
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
include/net/tcp.h |2 ++
net/ipv4/tcp.c| 10 +-
2 files changed, 7 insertions(+), 5 deletions(-)
diff --git a/include/net/tcp.h b/include/net/tcp.h
Locks down user pages and sets up for DMA in tcp_recvmsg, then calls
dma_async_try_early_copy in tcp_v4_do_rcv
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
net/ipv4/tcp.c | 103 --
net/ipv4/tcp_input.c
Any socket recv of less than this ammount will not be offloaded
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
include/linux/sysctl.h |1 +
include/net/tcp.h |1 +
net/core/user_dma.c|4
net/ipv4/sysctl_net_ipv4.c | 10 ++
4 files chang
Provides for pinning user space pages in memory, copying to iovecs,
and copying from sk_buffs including fragmented and chained sk_buffs.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/Makefile |3
drivers/dma/iovlock.c
Adds an async_wait_queue and some additional fields to tcp_sock, and a
dma_cookie_t to sk_buff.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
include/linux/skbuff.h |4
include/linux/tcp.h|8
include/net/sock.h |2 ++
include/net/tcp.h
Attempts to allocate per-CPU DMA channels
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/Kconfig | 12 +
include/linux/netdevice.h |4 ++
include/net/netdma.h | 38
net/core/dev.c
Provides an API for offloading memory copies to DMA devices
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/Kconfig |2
drivers/Makefile |1
drivers/dma/Kconfig | 13 +
drivers/dma/Makefile |1
drivers/dma/dmaengine.c
Add a sysctl to tune the minimum offloaded I/O size for TCP
9) The main TCP receive offload changes
--
Chris Leech <[EMAIL PROTECTED]>
I/O Acceleration Technology Software Development
LAN Access Division / Digital Enterprise Group
-
To unsubscribe from this list: send the line "un
Any idea why 120Mhz is used instead of 133? It doesn't
seem to matter in my performance tests, but I am curious...
I think Rick is right, the bus between the bridge on the card and the
e1000s is running at 120Mhz.
- Chris
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
t
Adds an async_wait_queue and some additional fields to tcp_sock, and a
dma_cookie_t to sk_buff.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
include/linux/skbuff.h |4
include/linux/tcp.h|8
include/net/sock.h |2 ++
include/net/tcp.h
Provides an API for offloading memory copies to DMA devices
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/Kconfig |2
drivers/Makefile |1
drivers/dma/Kconfig | 13 +
drivers/dma/Makefile |1
drivers/dma/dmaengine.c
Provides for pinning user space pages in memory, copying to iovecs,
and copying from sk_buffs including fragmented and chained sk_buffs.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/Makefile |3
drivers/dma/iovlock.c
Needed to be able to call tcp_cleanup_rbuf in tcp_input.c for I/OAT
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
include/net/tcp.h |2 ++
net/ipv4/tcp.c| 10 +-
2 files changed, 7 insertions(+), 5 deletions(-)
diff --git a/include/net/tcp.h b/include/net/tcp.h
Add an extra argument to sk_eat_skb, and make it move early copied packets
to the async_wait_queue instead of freeing them.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
include/net/sock.h | 13 -
net/dccp/proto.c |4 ++--
net/ipv4/tcp.c |8
n
Locks down user pages and sets up for DMA in tcp_recvmsg, then calls
dma_async_try_early_copy in tcp_v4_do_rcv
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
net/ipv4/tcp.c | 103 --
net/ipv4/tcp_input.c
Attempts to allocate per-CPU DMA channels
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/Kconfig | 12 +
include/linux/netdevice.h |4 ++
include/net/netdma.h | 38
net/core/dev.c
Any socket recv of less than this ammount will not be offloaded
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
include/linux/sysctl.h |1 +
include/net/tcp.h |1 +
net/core/user_dma.c|4
net/ipv4/sysctl_net_ipv4.c | 10 ++
4 files chang
[I/OAT] Driver for the Intel(R) I/OAT DMA engine
From: Chris Leech <[EMAIL PROTECTED]>
Adds a new ioatdma driver
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/Kconfig |9
drivers/dma/Makefile|1
drivers/dma/ioatdma.c
cture changes needed for TCP receive offload
6) Rename cleanup_rbuf to tcp_cleanup_rbuf
7) Make sk_eat_skb aware of early copied packets
8) Add a sysctl to tune the minimum offloaded I/O size for TCP
9) The main TCP receive offload changes
--
Chris Leech <[EMAIL P
On 5/3/06, Ben Greear <[EMAIL PROTECTED]> wrote:
So, as of 2.6.16.13, is the hardware stripping (SERC) enabled? Could
you also let me know where this bit is defined in case I want to twiddle
it myself (a quick grep for SERC in 2.6.16.13 yields nothing.)
You missed a C, it's SECRC (Strip Ethern
> Netperf2 TOT now accesses the buffer that was just recv()'d rather than
> the one that is about to be recv()'d.
We've posted netperf2 results with I/OAT enabled/disabled and the data
access option on/off at
http://kernel.org/pub/linux/kernel/people/grover/ioat/netperf-icb-1.5-postscaling-both.pd
On 4/20/06, David S. Miller <[EMAIL PROTECTED]> wrote:
> Yes, and it means that the memory bandwidth costs are equivalent
> between I/O AT and cpu copy.
The following is a response from the I/OAT architects. I only point
out that this is not coming directly from me because I have not seen
the dat
> Could you please describe how struct ioat_dma_chan channels are freed?
Sorry, I got distracted by other issues and never ended up following
up on this. You're right, and it's just sloppiness on my part for
missing it, those structs are being leaked on module unload. I'll fix
it. Thanks.
-Chr
Provides for pinning user space pages in memory, copying to iovecs,
and copying from sk_buffs including fragmented and chained sk_buffs.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/Makefile |3
drivers/dma/iovlock.c
ceive offload changes
--
Chris Leech <[EMAIL PROTECTED]>
I/O Acceleration Technology Software Development
LAN Access Division / Digital Enterprise Group
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More ma
Adds an async_wait_queue and some additional fields to tcp_sock, and a
dma_cookie_t to sk_buff.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
include/linux/skbuff.h |4
include/linux/tcp.h|8
include/net/sock.h |2 ++
include/net/tcp.h
Needed to be able to call tcp_cleanup_rbuf in tcp_input.c for I/OAT
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
include/net/tcp.h |2 ++
net/ipv4/tcp.c| 10 +-
2 files changed, 7 insertions(+), 5 deletions(-)
diff --git a/include/net/tcp.h b/include/net/tcp.h
Provides an API for offloading memory copies to DMA devices
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/Kconfig |2
drivers/Makefile |1
drivers/dma/Kconfig | 13 +
drivers/dma/Makefile |1
drivers/dma/dmaengine.c
Attempts to allocate per-CPU DMA channels
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/Kconfig | 12 +
include/linux/netdevice.h |4 ++
include/net/netdma.h | 38
net/core/dev.c
Locks down user pages and sets up for DMA in tcp_recvmsg, then calls
dma_async_try_early_copy in tcp_v4_do_rcv
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
net/ipv4/tcp.c | 101 --
net/ipv4/tcp_input.c
Any socket recv of less than this ammount will not be offloaded
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
include/linux/sysctl.h |1 +
include/net/tcp.h |1 +
net/core/user_dma.c|4
net/ipv4/sysctl_net_ipv4.c | 10 ++
4 files chang
Add an extra argument to sk_eat_skb, and make it move early copied packets
to the async_wait_queue instead of freeing them.
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
include/net/sock.h | 13 -
net/dccp/proto.c |4 ++--
net/ipv4/tcp.c |8
n
[I/OAT] Driver for the Intel(R) I/OAT DMA engine
From: Chris Leech <[EMAIL PROTECTED]>
Adds a new ioatdma driver
Signed-off-by: Chris Leech <[EMAIL PROTECTED]>
---
drivers/dma/Kconfig |9
drivers/dma/Makefile|1
drivers/dma/ioatdma.c
On 3/16/06, Scott Feldman <[EMAIL PROTECTED]> wrote:
> Do you have any data to share on header split? Also, can other non-
> Intel nics use I/OAT copy, and if so, is header-split a requirement
> for the copy?
I don't have any header-split data. The I/OAT copy offload will work
for any TCP traffi
> Thanks, that clarifies things. So, if I've understood correctly, the
> benefit kicks in when:
>
> 1) I/OAT is enabled :)
> 2) The user posts a recv() (or the like) of >= 2K
> 3) There is >= 2K of data available to give them
>
> yes?
Yes
-
To unsubscribe from this list: send the line "unsubscrib
should have kept this on list
-- Forwarded message --
From: Chris Leech <[EMAIL PROTECTED]>
Date: Mar 16, 2006 11:13 AM
Subject: Re: I/OAT performance data
To: Rick Jones <[EMAIL PROTECTED]>
> I must be missing something - if the MTU was 1500 bytes, how did t
1 - 100 of 132 matches
Mail list logo
