Thanks for including these.
We are working on patches for some/most of the receive side features,
subject to the interruptions from support/maintenance work.
Andi - let me know if you still have reservations about the LRO
application.
I went ahead and got Company's permission to sign a waiver fo
applied
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
applied patches 1-18
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Aug 31, 2005 at 09:22:29PM +0300, Kalle Valo wrote:
> Jouni Malinen <[EMAIL PROTECTED]> writes:
> > local->hw_priv was initialized only after the interrupt handler was
> > registered. This could trigger a NULL pointer dereference in
> > prism2_pccard_card_present() that assumed that local->
The iseries_veth driver often has multiple netdevices sending packets over
a single connection to another LPAR. If the bandwidth to the other LPAR is
exceeded, all the netdevices must have their queues stopped.
The current code achieves this by queueing one incoming skb on the
per-netdevice port s
iseries_veth.h is only used by iseries_veth.c, so merge the former into
the latter.
Signed-off-by: Michael Ellerman <[EMAIL PROTECTED]>
---
drivers/net/iseries_veth.h | 46 -
drivers/net/iseries_veth.c | 42 +++--
The iseries_veth driver uses the generic TX timeout watchdog, however a better
solution is in the works, so remove this code.
Signed-off-by: Michael Ellerman <[EMAIL PROTECTED]>
---
drivers/net/iseries_veth.c | 48 -
1 files changed, 48 deletions(-)
The iseries_veth driver can attach to multiple vlans, which correspond to
multiple net devices. However there is only 1 connection between each LPAR,
so the connection structure may be shared by multiple net devices.
This makes module removal messy, because we can't deallocate the connections
unti
To aid in field debugging, add sysfs support for iseries_veth's connection
structures. At the moment this is all read-only, however we could think about
adding write support for some attributes in future.
Signed-off-by: Michael Ellerman <[EMAIL PROTECTED]>
---
drivers/net/iseries_veth.c | 94 +
There's a number of problems with the way iseries_veth counts TX errors.
Firstly it counts conditions which aren't really errors as TX errors. This
includes if we don't have a connection struct for the other LPAR, or if the
other LPAR is currently down (or just doesn't want to talk to us). Neither
The iseries_veth driver has a timer which we use to send acks. When the
connection is reset or stopped we need to delete the timer.
Currently we only call del_timer() when resetting a connection, which means
the timer might run again while the connection is being re-setup. As it turns
out that's o
The iseries_veth driver keeps a stack of messages for each connection
and a lock to protect the stack. However there is also a per-connection lock
which makes the message stack lock redundant.
Remove the message stack lock and document the fact that callers of the
stack-manipulation functions must
The iseries_veth driver uses atomic ops to manipulate the in_use field of
one of its per-connection structures. However all references to the
flag occur while the connection's lock is held, so the atomic ops aren't
necessary.
Signed-off-by: Michael Ellerman <[EMAIL PROTECTED]>
---
drivers/net/is
The iseries_veth driver contains a state machine which is used to manage
how connections are setup and neogotiated between LPARs.
If one side of a connection resets for some reason, the two LPARs can get
stuck in a race to re-setup the connection. This can lead to the connection
being declared dea
Currently the iseries_veth driver contravenes the specification in
Documentation/networking/driver.txt, in that if packets are not acked by
the other LPAR they will sit around forever.
This patch adds a per-connection timer which fires if we've had no acks for
five seconds. This is superior to the
Also to aid debugging, add sysfs support for iseries_veth's port structures.
Signed-off-by: Michael Ellerman <[EMAIL PROTECTED]>
---
drivers/net/iseries_veth.c | 67 +
1 files changed, 67 insertions(+)
Index: veth-dev2/drivers/net/iseries_veth.c
===
Having merged iseries_veth.h, let's remove some of the studly caps that came
with it.
Signed-off-by: Michael Ellerman <[EMAIL PROTECTED]>
---
drivers/net/iseries_veth.c | 74 ++---
1 files changed, 37 insertions(+), 37 deletions(-)
Index: veth-dev2/driv
This patch makes veth_init_connection() and veth_destroy_connection()
symmetrical in that they allocate/deallocate the same data.
Currently if there's an error while initialising connections (ie. ENOMEM)
we call veth_module_cleanup(), however this will oops because we call
driver_unregister() befo
The iseries_veth driver tells sysfs that it's called 'iseries_veth', but if
you ask it via ethtool it thinks it's called 'veth'. I think this comes from
2.4 when the driver was called 'veth', but it's definitely called
'iseries_veth' now, so fix it.
To make sure we don't do it again define DRV_NAM
The iseries_veth driver unconditionally calls dma_unmap_single() even
when the corresponding dma_map_single() may have failed.
Rework the code a bit to keep the return value from dma_unmap_single()
around, and then check if it's a dma_mapping_error() before we do
the dma_unmap_single().
Signed-of
Due to a logic bug, once promiscuous mode is enabled in the iseries_veth
driver it is never disabled.
The driver keeps two flags, promiscuous and all_mcast which have exactly the
same effect. This is because we only ever receive packets destined for us,
or multicast packets. So consolidate them in
Currently the iseries_veth driver prints the file name and line number in its
error messages. This isn't very useful for most users, so just print
"iseries_veth: message" instead.
- convert uses of veth_printk() to veth_debug()/veth_error()/veth_info()
- make terminology consistent, ie. always r
Hi,
This is a series of patches for the iseries_veth driver. Most of these are
pretty much unchanged since I posted them earlier:
http://www.uwsg.iu.edu/hypermail/linux/kernel/0506.3/1837.html
I've added patches to add sysfs support, and do some further code cleanups.
Please merge if they look o
* Alexey Kuznetsov ([EMAIL PROTECTED]) wrote:
> Hello!
>
> > Thanks for that explanation - it helps somewhat - one thing I was
> > confused by was why the timer mechanism for the garbage collection
> > was so elaborate; why does it do all that back off stuff and
> > adjusting itself? Why not just
On Wed, Aug 31, 2005 at 05:43:32PM -0400, Peter Jones wrote:
> On Wed, 2005-08-31 at 10:52 -0700, Jean Tourrilhes wrote:
>
> > Peter Jones <[EMAIL PROTECTED]> wrote :
> > >
> > > I don't think that's really right either. For one thing, things like
> > > DHCP's timeout start counting at about the
Driver version, white space, comments, device id & other
Signed-off-by: Mallikarjuna R Chilakala <[EMAIL PROTECTED]>
Signed-off-by: Ganesh Venkatesan <[EMAIL PROTECTED]>
Signed-off-by: John Ronciak <[EMAIL PROTECTED]>
diff -up netdev-2.6/drivers/net/e1000/e1000_main.c
netdev-2.6.new/drivers/net/
Enable custom configuration bits for 82571/2 controllers. The bits are
required for correct functionality of these controllers.
Signed-off-by: Mallikarjuna R Chilakala <[EMAIL PROTECTED]>
Signed-off-by: Ganesh Venkatesan <[EMAIL PROTECTED]>
Signed-off-by: John Ronciak <[EMAIL PROTECTED]>
diff -u
Implementation of the multi-queue feature
Note: The following set of patches p2[ab] enable support for multiple
transmit/receive queues (aka multi-queue patch).
Note: patch2a, patch2b, patch2c need to be applied as a set. The driver will
NOT compile if only patch2b is applied alone.
Note: A kern
defines/modifies data structures, function prototypes and changes to the
driver rendering it capable of handling n tx/rx queues
Note: The following set of patches p2[ab] enable support for multiple
transmit/receive queues (aka multi-queue patch).
Note: patch2a, patch2b, patch2c need to be applie
Fixes for packet split related issues
* On platforms where PAGE_SIZE > 4K, driver will use only required number of
pages compared to always using 3 pages.
* Packet split won't be used if the PAGE_SIZE is > 16K
* Adds a statistics counter to splits.
* Setting the non Null ptr to zero si
Arnaud Patard <[EMAIL PROTECTED]> :
[...]
> I've also tested the two others patches as suggested but I didn't
> notice any difference (I got only "eth0: GMII mode"). Unfortunately,
> I'm not sure that the test is reliable as the PHY
> transceiver of my card is unknow to the driver...
The test is
On Wed, 2005-08-31 at 10:52 -0700, Jean Tourrilhes wrote:
> Peter Jones <[EMAIL PROTECTED]> wrote :
> >
> > I don't think that's really right either. For one thing, things like
> > DHCP's timeout start counting at about the same time as "ifconfig up",
> > and association can take some time.
>
>
From: "Leonid Grossman" <[EMAIL PROTECTED]>
Date: Sat, 30 Jul 2005 02:41:38 -0400
> Some suggestions on how to proceed on receive side:
I've added your suggestions to:
http://vger.kernel.org/~davem/net_todo.html
sorry for taking so long to integrate...
-
To unsubscribe from this list: send
On Wed, 2005-08-31 at 19:08 +0200, Jiri Benc wrote:
> But is it really needed? Imagine the situation when computer is started
> with unplugged ethernet cable which is plugged in later. I know, this is
> rare, but - shouldn't be the right approach that DHCP is started by
> hotplug when the carrier
From: Qasim Javed <[EMAIL PROTECTED]>
Date: Wed, 31 Aug 2005 21:04:26 +0500
> I have encountered a bug in the pskb_expand_head procedure in
> "net/core/skbuff.h". The following chunk is wrong i think.
You mean "net/core/skbuff.c"
> skb_release_data(skb);
>
> off = (data+nhead) - skb->head;
>
>
From: Vlad Yasevich <[EMAIL PROTECTED]>
Date: Wed, 31 Aug 2005 10:10:48 -0400
> I am trying to teach SCTP about the TOS settings on a per-association
> basis. While trying to do that, I was using rt_tos2priority call.
> Everything worked great while sctp was built into the kernel, but
> builds fa
From: Vlad Yasevich <[EMAIL PROTECTED]>
Date: Wed, 31 Aug 2005 09:16:32 -0400
> Declares ip_tos2prio as an exported symbol, so that modules
> wishing to use rt_tos2priority() call, may do so.
>
> Signed-off-by: Vlad Yasevich <[EMAIL PROTECTED]>
When you add an in-tree legitimate use, we'll expor
Hi!
> > What other changes are required in userspace after your patches are
> > applied?
>
> AFAIK none.
Really?
AFAICS, with your patches ifconfig shows counts of wifi packets. How
do I get ethernet packet counts? Will tcpdump wlan0 work on ethernet
or wifi level?
> > > I agree. Don't know wh
Jouni Malinen <[EMAIL PROTECTED]> writes:
>> applied, but let us know when the root cause is found...
>
> local->hw_priv was initialized only after the interrupt handler was
> registered. This could trigger a NULL pointer dereference in
> prism2_pccard_card_present() that assumed that local->hw_pr
On Wed, Aug 31, 2005 at 07:08:02PM +0200, Jiri Benc wrote:
> But is it really needed? Imagine the situation when computer is started
> with unplugged ethernet cable which is plugged in later. I know, this is
> rare, but - shouldn't be the right approach that DHCP is started by
> hotplug when the ca
James Ketrenos wrote :
>
> When last we looked at this, step 4 was being handled by ifplugd only
> after IFF_RUNNING or carrier_detect indicated link, which with wireless
> is potentially not possible until after step 7.
So, basically you want to change all wireless driver and all
distro
On Wed, 31 Aug 2005 11:57:15 -0400, Peter Jones wrote:
> I don't think that's really right either. For one thing, things like
> DHCP's timeout start counting at about the same time as "ifconfig up",
> and association can take some time.
You're right, thanks for pointing this out.
> But why don't
Qasim Javed wrote:
> Hi,
>
> I have encountered a bug in the pskb_expand_head procedure in
> "net/core/skbuff.h". The following chunk is wrong i think.
>
> skb_release_data(skb);
>
> off = (data+nhead) - skb->head;
>
> The data of skb is first released and then skb->head is used in the
> very n
Daniel Drake wrote:
> Hi,
>
> A gentoo user reported these messages on boot:
>
> eth0: resetting device...
> eth0: uploading firmware...
> eth0: firmware version: 1.0.4.3
> eth0: firmware upload complete
> eth0: interface reset complete
> spurious 8259A interrupt: IRQ7.
> eth0: islpci_close ()
>
Hi,
I have encountered a bug in the pskb_expand_head procedure in
"net/core/skbuff.h". The following chunk is wrong i think.
skb_release_data(skb);
off = (data+nhead) - skb->head;
The data of skb is first released and then skb->head is used in the
very next statement.
regards,
Qasim Javed
-
To
On Wed, 2005-08-31 at 14:39 +0200, Jiri Benc wrote:
> I don't agree with this scheme. Association should be started on
> explicit userspace request (*). As we definitely don't want to add a new
> WE (or some other) call to perform this, the only call we can use for
> telling the driver "ok, now it
Ok. I'll hold off on this till the full patch is complete.
Thanks
-vlad
Patrick McHardy wrote:
> Vlad Yasevich wrote:
>
>>I am trying to teach SCTP about the TOS settings on a per-association
>>basis. While trying to do that, I was using rt_tos2priority call.
>>Everything worked great while sc
Francois Romieu <[EMAIL PROTECTED]> writes:
> Arnaud Patard <[EMAIL PROTECTED]> :
> [...]
>
> Btw it would be nice if you could give a quick test to the three
> attached patches against 2.6.13-git: #0, #0 + #1, #0 + #1 + #2.
The first patch 160 works as expected.
I've also tested the two others
Vlad Yasevich wrote:
> I am trying to teach SCTP about the TOS settings on a per-association
> basis. While trying to do that, I was using rt_tos2priority call.
> Everything worked great while sctp was built into the kernel, but
> builds failed when building as a module.
>
> I tracked it down to
I am trying to teach SCTP about the TOS settings on a per-association
basis. While trying to do that, I was using rt_tos2priority call.
Everything worked great while sctp was built into the kernel, but
builds failed when building as a module.
I tracked it down to this code. rt_tos2prority is a s
Vlad Yasevich wrote:
Declares ip_tos2prio as an exported symbol, so that modules
wishing to use rt_tos2priority() call, may do so.
And which ones would that be? We usually don't export symbols
unless there is an in-kernel user that needs it.
-
To unsubscribe from this list: send the line "unsub
Declares ip_tos2prio as an exported symbol, so that modules
wishing to use rt_tos2priority() call, may do so.
Signed-off-by: Vlad Yasevich <[EMAIL PROTECTED]>
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -3204,3 +3204,4 @@ int __init ip_rt_init
On Sat, 27 Aug 2005 11:21:37 -0500, James Ketrenos wrote:
> The order required of user space is:
>
>kernel hotplug hotplug script
>--------
> 1. module load
> 2. netdev device registered
> 3.
On Tue, Aug 30, 2005 at 10:45:54PM +0200, Jesper Juhl wrote:
>
> I've posted similar patches in the past, but was asked to first until the
> short-circuit patch moved from -mm to mainline - and since it is now
> firmly there in 2.6.13 I assume there's no problem there anymore.
> I was also asked
Hi,
A gentoo user reported these messages on boot:
eth0: resetting device...
eth0: uploading firmware...
eth0: firmware version: 1.0.4.3
eth0: firmware upload complete
eth0: interface reset complete
spurious 8259A interrupt: IRQ7.
eth0: islpci_close ()
eth0: resetting device...
eth0: uploading f
On Sat, 27 Aug 2005 12:25:43 +0200, Pavel Machek wrote:
> > Sequence number is stored in bits 4 to 15 of the Sequence Control field
> > in 802.11 header. Lower 4 bits are used for fragment number.
>
> Comment would be nice.
Added.
> What other changes are required in userspace after your patches
Stephen Hemminger wrote:
You have a version of the Marvell Yukon that was affected
by a fix in 2.6.13.
skge addr 0xfeaf8000 irq 19 chip Yukon-Lite rev 9
Both the skge and sk98lin driver were fixed to check for this.
Without the fix, the chip will be in the wrong power mode.
The version
57 matches
Mail list logo
