James Morris wrote:
>>An Introduction To Using Generic Netlink
>>===
>
>
> Wow, this is great!
Thanks. I consider it an act of penance for all of the evil things I did
with Netlink on my first few iterations of NetLabel
> An Introduction To Using Generic Netlink
> ===
Wow, this is great!
--
James Morris
<[EMAIL PROTECTED]>
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
A couple of months ago I promised Jamal and Thomas I would post some comments to
Jamal's original genetlink how-to. However, as I started to work on the
document the diff from the original started to get a little ridiculous so
instead of posting a patch against Jamal's original how-to I'm just pos
Hi,
Long time lurker, first time poster. ^_^
I've been backporting the bcm43xx-d80211 driver to whatever the released
2.6 kernel was using the rt2x00 project's d80211 stack (equivalent to
current wireless-dev but with a workaround for not having a ieee80211_dev
pointer and still using the _tfm i
Hi,
I've got an interesting problem to contend with and
need some advice from the great wise ones here.
First of all, is it possible (and/or "reasonable
practice") when developing a network driver to do
zero-copy transfers between main memory and the
network device?
Secondly, the network device
Ar Iau, 2006-11-09 am 16:12 -0800, ysgrifennodd David Miller:
> Really, this driver has a ton of unresolved portability problems.
Agreed - but at least its now 64bit clean. No objection to leaving it !
64BIT at all even with the patch merged.
-
To unsubscribe from this list: send the line "unsubs
Michael Stone <[EMAIL PROTECTED]> wrote:
>On Tue, Nov 07, 2006 at 12:28:26PM -0800, Jay Vosburgh wrote:
>> Can you provide some bonding configuration details? Which mode,
>>options, etc, as well as the relevant bits from dmesg (you can send it
>> to me privately if it's huge)?
>
>I think I s
From: Stephen Hemminger <[EMAIL PROTECTED]>
Date: Tue, 31 Oct 2006 15:01:45 -0800
> Rearrange TCP entries in alpha order.
>
> Signed-off-by: Stephen Hemminger <[EMAIL PROTECTED]>
Also applied, thanks a lot.
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a mess
From: Stephen Hemminger <[EMAIL PROTECTED]>
Date: Tue, 31 Oct 2006 15:01:43 -0800
> Allow normal users to only choose among a restricted set of congestion
> control choices. The default is reno and what ever has been configured
> as default. But the policy can be changed by administrator at any t
From: Stephen Hemminger <[EMAIL PROTECTED]>
Date: Tue, 31 Oct 2006 15:01:44 -0800
> If user has permision to load modules, then autoload then attempt
> autoload of TCP congestion module.
>
> Signed-off-by: Stephen Hemminger <[EMAIL PROTECTED]>
Applied, thanks.
-
To unsubscribe from this list: se
On Tue, Nov 07, 2006 at 12:28:26PM -0800, Jay Vosburgh wrote:
Can you provide some bonding configuration details? Which mode,
options, etc, as well as the relevant bits from dmesg (you can send it
to me privately if it's huge)?
I think I sent another message that I'm just doing
ifen
From: Stephen Hemminger <[EMAIL PROTECTED]>
Date: Tue, 31 Oct 2006 15:01:42 -0800
> Create /proc/sys/net/ipv4/tcp_available_congestion_control
> that reflects currently available TCP choices.
>
> Signed-off-by: Stephen Hemminger <[EMAIL PROTECTED]>
Applied, thanks Stephen.
-
To unsubscribe from
From: Sridhar Samudrala <[EMAIL PROTECTED]>
Date: Thu, 02 Nov 2006 10:29:49 -0800
> On Thu, 2006-11-02 at 11:09 -0500, Vlad Yasevich wrote:
> > Meelis Roos wrote:
> > >> Actually, I'm backing this one out, it creates new warnings because
> > >> callers of this function pass in a "const" pointer.
>
On 11/9/06, John <[EMAIL PROTECTED]> wrote:
> The second thought is that the adapter is in D3, and something about
> your kernel or the driver doesn't successfully wake it up to D0.
On my NICs, the EEPROM ID (Word 0Ah) is set to 0x40a2.
Thus DDPD (bit 6) is set to 0.
DDPD is the "Disable Deep P
From: David Kimdon <[EMAIL PROTECTED]>
Date: Wed, 8 Nov 2006 06:06:18 -0800
> Based on patch by Patrick McHardy.
>
> Add a new option, NET_SCH_FIFO, which provides a simple fifo qdisc
> without requiring CONFIG_NET_SCHED.
>
> The d80211 stack needs a generic fifo qdisc for WME. At present it
>
From: [EMAIL PROTECTED]
Date: Wed, 08 Nov 2006 19:51:04 -0800
> From: Alan Cox <[EMAIL PROTECTED]>
>
> Signed-off-by: Alan Cox <[EMAIL PROTECTED]>
> Signed-off-by: Andrew Morton <[EMAIL PROTECTED]>
This fixes the most obvious 64-bit problems, but it is still very very
broken in other aspects. I
From: [EMAIL PROTECTED]
Date: Wed, 08 Nov 2006 19:51:01 -0800
> The bonding driver nests other drivers, give the bonding driver its own
> lock class.
>
> Signed-off-by: Peter Zijlstra <[EMAIL PROTECTED]>
> Acked-by: Ingo Molnar <[EMAIL PROTECTED]>
> Cc: Stephen Hemminger <[EMAIL PROTECTED]>
> Sig
Basheer, Mansoor Ahamed <[EMAIL PROTECTED]> :
[...]
> My understanding here is, on receive interrupt the ISR should disable
> the receive interrupt irrespective of the polling task's state (active
> or inactive).
Apparently it could happen even with 2.6.19-rc5, yes.
> I changed the code (as shown
From: Thomas Graf <[EMAIL PROTECTED]>
Date: Thu, 09 Nov 2006 12:27:41 +0100
> Introduces a new flag FIB_RULE_INVERT causing rules to apply
> if the specified selector doesn't match.
>
> Signed-off-by: Thomas Graf <[EMAIL PROTECTED]>
Also applied, thanks.
-
To unsubscribe from this list: send the
From: Thomas Graf <[EMAIL PROTECTED]>
Date: Thu, 09 Nov 2006 12:27:38 +0100
> For the sake of consistency.
>
> Signed-off-by: Thomas Graf <[EMAIL PROTECTED]>
Applied, thanks Thomas.
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
From: Thomas Graf <[EMAIL PROTECTED]>
Date: Thu, 09 Nov 2006 12:27:37 +0100
> Now that all protocols have been made aware of the mark
> field it can be moved out of the union thus simplyfing
> its usage.
>
> The config options in the IPv4/IPv6/DECnet subsystems
> to enable respectively disable ma
From: Thomas Graf <[EMAIL PROTECTED]>
Date: Thu, 09 Nov 2006 12:27:40 +0100
> Move the attribute policy for the non-specific attributes into
> net/fib_rules.h and include it in the respective protocols.
>
> Signed-off-by: Thomas Graf <[EMAIL PROTECTED]>
Looks nice, applied, thanks.
-
To unsubscr
From: Thomas Graf <[EMAIL PROTECTED]>
Date: Thu, 09 Nov 2006 12:27:39 +0100
> Move mark selector currently implemented per protocol into
> the protocol independant part.
>
> Signed-off-by: Thomas Graf <[EMAIL PROTECTED]>
Applied, thanks.
-
To unsubscribe from this list: send the line "unsubscrib
From: Thomas Graf <[EMAIL PROTECTED]>
Date: Thu, 09 Nov 2006 12:27:36 +0100
> nfmark is being used in various subsystems and has become
> the defacto mark field for all kinds of packets. Therefore
> it makes sense to rename it to `mark' and remove the
> dependency on CONFIG_NETFILTER.
>
> Signed-
On 11/9/06, Luis R. Rodriguez <[EMAIL PROTECTED]> wrote:
On 11/9/06, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote:
> Am Mittwoch, 8. November 2006 01:39 schrieben Sie:
> > On Fri, Nov 03, 2006 at 01:41:46PM -0500, Luis R. Rodriguez wrote:
> > > On 11/3/06, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wro
On 11/9/06, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote:
Am Mittwoch, 8. November 2006 01:39 schrieben Sie:
> On Fri, Nov 03, 2006 at 01:41:46PM -0500, Luis R. Rodriguez wrote:
> > On 11/3/06, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote:
> > >yes, especially mgt_commit_list caused alot headaches,
On Thu, Nov 09, Stephen Hemminger wrote:
> On Thu, 9 Nov 2006 19:40:21 +0100 (MET)
> Olaf Hering <[EMAIL PROTECTED]> wrote:
>
> >
> > Provide drivers for the old toys only on i386
> > isa_bus_to_virt is defined only on i386, mips and arm
> > isa_virt_to_bus is used for floppy.ko
>
> Why not mar
From: Brian Haley <[EMAIL PROTECTED]>
Date: Thu, 09 Nov 2006 12:32:18 -0500
> Al Viro wrote:
> > AFAICS, the rules are:
> >
> > (1) checksum is 16-bit one's complement of the one's complement sum of
> > relevant 16bit words.
> >
> > (2) for v4 UDP all-zeroes has special meaning - no checksum
Jeff V. Merkey wrote:
Jesse Brandeburg wrote:
On 11/9/06, Jeffrey V. Merkey <[EMAIL PROTECTED]> wrote:
In the case I am referring to, the memory is already mapped with a
previous call, which means it may be getting
mapped twice.
I guess maybe I'm not keeping up with you. This is what I
Please use GREG_STAT_* instead of magic constants for the
interrupt mask and ACK register writes. In fact, there are
some questionable values you use, in particular this one:
+static inline void gem_ack_int(struct gem *gp)
+{
+ writel(0x3f, gp->regs + GREG_IACK);
+}
There is no bit define
Jesse Brandeburg wrote:
On 11/9/06, Jeffrey V. Merkey <[EMAIL PROTECTED]> wrote:
In the case I am referring to, the memory is already mapped with a
previous call, which means it may be getting
mapped twice.
I guess maybe I'm not keeping up with you. This is what I see looking
in 2.6.18, i
On 11/9/06, Jeffrey V. Merkey <[EMAIL PROTECTED]> wrote:
In the case I am referring to, the memory is already mapped with a
previous call, which means it may be getting
mapped twice.
I guess maybe I'm not keeping up with you. This is what I see looking
in 2.6.18, i see e1000_clean_rx_irq:
che
The attached patch improves locking in the sungem driver:
- a single lock is used in the driver
- gem_start_xmit, gem_poll, and gem_interrupt are lockless
The new locking design is based on what's in tg3.c.
The patch runs smoothly on my ibook (with CONFIG_SMP set), but it will
need extensive te
On Thu, 9 Nov 2006, Davide Libenzi wrote:
> On Thu, 9 Nov 2006, Evgeniy Polyakov wrote:
>
> > On Thu, Nov 09, 2006 at 10:51:56AM -0800, Davide Libenzi
> > (davidel@xmailserver.org) wrote:
> > > On Thu, 9 Nov 2006, Evgeniy Polyakov wrote:
> > >
> > > > +static int kevent_poll_callback(struct kev
On Thu, 9 Nov 2006 19:40:21 +0100 (MET)
Olaf Hering <[EMAIL PROTECTED]> wrote:
>
> Provide drivers for the old toys only on i386
> isa_bus_to_virt is defined only on i386, mips and arm
> isa_virt_to_bus is used for floppy.ko
Why not mark all of ISA as i386, mips, arm only?
-
To unsubscribe from
On Thu, 9 Nov 2006, Evgeniy Polyakov wrote:
> On Thu, Nov 09, 2006 at 10:51:56AM -0800, Davide Libenzi
> (davidel@xmailserver.org) wrote:
> > On Thu, 9 Nov 2006, Evgeniy Polyakov wrote:
> >
> > > +static int kevent_poll_callback(struct kevent *k)
> > > +{
> > > + if (k->event.req_flags & KEVENT_
On Thu, Nov 09, 2006 at 10:51:56AM -0800, Davide Libenzi
(davidel@xmailserver.org) wrote:
> On Thu, 9 Nov 2006, Evgeniy Polyakov wrote:
>
> > +static int kevent_poll_callback(struct kevent *k)
> > +{
> > + if (k->event.req_flags & KEVENT_REQ_LAST_CHECK) {
> > + return 1;
> > + } els
On Thu, 9 Nov 2006, Evgeniy Polyakov wrote:
> +static int kevent_poll_callback(struct kevent *k)
> +{
> + if (k->event.req_flags & KEVENT_REQ_LAST_CHECK) {
> + return 1;
> + } else {
> + struct file *file = k->st->origin;
> + unsigned int revents = file-
Provide drivers for the old toys only on i386
isa_bus_to_virt is defined only on i386, mips and arm
isa_virt_to_bus is used for floppy.ko
Add missing '&& ISA_DMA_API' to NI52
WARNING: "isa_bus_to_virt" [drivers/net/ni65.ko] undefined!
WARNING: "isa_virt_to_bus" [drivers/net/ni65.ko] undefined!
W
On 11/9/06, Michael Chan <[EMAIL PROTECTED]> wrote:
Eric Lemoine wrote:
> > On 11/9/06, Michael Chan <[EMAIL PROTECTED]> wrote:
> > > So it is not possible for tg3_poll() -> tg3_tx() to run any more
> > > after tg3_close() is called.
> >
> > But, while tg3_close() starts executing, an interrupt
Am Mittwoch, 8. November 2006 01:39 schrieben Sie:
> On Fri, Nov 03, 2006 at 01:41:46PM -0500, Luis R. Rodriguez wrote:
> > On 11/3/06, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote:
> > >yes, especially mgt_commit_list caused alot headaches, until I removed
> > >DOT11_OID_PSM from the cache list.
>
On Thu, 2006-11-09 at 18:22 +0100, Eric Lemoine wrote:
> Actually I don't understand the purpose of having dev_close() wait for
> __LINK_STATE_RX_SCHED to be cleared. An interrupt may arrive at any
> time after it's cleared, and reset __LINK_STATE_RX_SCHED. Can someone
> explain please?
Further
Hi Al,
Al Viro wrote:
AFAICS, the rules are:
(1) checksum is 16-bit one's complement of the one's complement sum of
relevant 16bit words.
(2) for v4 UDP all-zeroes has special meaning - no checksum; if you get
it from (1), send all-ones instead.
(3) for v6 UDP we have the same remappi
Eric Lemoine wrote:
> > On 11/9/06, Michael Chan <[EMAIL PROTECTED]> wrote:
> > > So it is not possible for tg3_poll() -> tg3_tx() to run any more
> > > after tg3_close() is called.
> >
> > But, while tg3_close() starts executing, an interrupt may
> > come in and
> > schedule polling (set __LINK
James Morris wrote:
> On Thu, 9 Nov 2006, Paul Moore wrote:
>
>>It sounds like you have an idea of how you would like to see this implemented,
>>can you give me a rough outline? Is this the partitioned SECMARK field you
>>talked about earlier?
>
> No, just the fact that you are in the same kerne
On 11/9/06, Eric Lemoine <[EMAIL PROTECTED]> wrote:
On 11/9/06, Michael Chan <[EMAIL PROTECTED]> wrote:
> Michael Chan wrote:
>
> > Eric Lemoine wrote:
> >
> > > Instead of tg3_netif_stop() tg3_close() uses netif_stop_queue()
> > > to stop xmit. This doesn't seem right to me. E.g. another CPU
> >
On 11/9/06, Michael Chan <[EMAIL PROTECTED]> wrote:
Michael Chan wrote:
> Eric Lemoine wrote:
>
> > Instead of tg3_netif_stop() tg3_close() uses netif_stop_queue()
> > to stop xmit. This doesn't seem right to me. E.g. another CPU
> > in tg3_tx()
> > could do netif_wake_queue() just after tg3_clo
On Thu, 9 Nov 2006, Eric Dumazet wrote:
> > > Lost forever means? If there are more processes watching some fd (external
> > > events), they all get their own copy of the events in their own private
> > > epoll fd. It's not that we "steal" things out of the kernel, is not a 1:1
> > > producer/cons
John wrote:
Auke Kok wrote:
This is what I was afraid of: even though the code allows you to
bypass the EEPROM checksum, the probe fails on a further check to see
if the MAC address is valid.
Since something with this NIC specifically made the EEPROM return all
0xff's, the MAC address is au
added
-
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
Michael Chan wrote:
> Eric Lemoine wrote:
>
> > Instead of tg3_netif_stop() tg3_close() uses netif_stop_queue()
> > to stop xmit. This doesn't seem right to me. E.g. another CPU
> > in tg3_tx()
> > could do netif_wake_queue() just after tg3_close() did
> > netif_stop_queue(). Isn't a bug?
> >
>
Eric Lemoine wrote:
> Instead of tg3_netif_stop() tg3_close() uses netif_stop_queue() to
> stop xmit. This doesn't seem right to me. E.g. another CPU in tg3_tx()
> could do netif_wake_queue() just after tg3_close() did
> netif_stop_queue(). Isn't a bug?
>
I think you're right. It is more correc
Hi,
On Thu, Nov 09, 2006 at 01:49:11PM +0100, Thomas Graf wrote:
> * Steven Whitehouse <[EMAIL PROTECTED]> 2006-11-09 11:46
> >
> > so far as all the DECnet bits go. One question though... will you
> > be adding later (as your slide #5 and #11 from your netconf presentation
> > appear to imply) a
Hi All
I found an issue with Realtek 8139 driver (2.6.10 branch) at high
bi-directional traffic. On transmit timeout, driver's timeout callback
re-enables the receive interrupt. On the next receive interrupt, the ISR
disables the receive interrupt "only" when the receive poll task is not
active. B
Jesse Brandeburg wrote:
I suspect that one reason Becker's code works is that it uses IO
based access (slower, and different method) to the adapter rather
than memory mapped access.
I've noticed this difference.
The second thought is that the adapter is in D3, and something about
your kernel
> I think this should be aimed at 2.6.20, because we are at the last or
> second-last -rc currently, and I don't think these fixes are
> urgent enough
> to justify the risk at this stage.
That makes sense. Thanks.
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body o
Driver for the Atmel MACB on-chip ethernet module.
Tested on AVR32/AT32AP7000/ATSTK1000. I've heard rumours that it works
with AT91SAM9260 as well, and it may be possible to share some code with
the at91_ether driver for AT91RM9200.
Hardware documentation can be found in the AT32AP7000 data sheet
Hi
Instead of tg3_netif_stop() tg3_close() uses netif_stop_queue() to
stop xmit. This doesn't seem right to me. E.g. another CPU in tg3_tx()
could do netif_wake_queue() just after tg3_close() did
netif_stop_queue(). Isn't a bug?
Thanks,
--
Eric
-
To unsubscribe from this list: send the line "uns
* Eric Dumazet <[EMAIL PROTECTED]> 2006-11-09 14:23
> I give a big NACK to this patch.
>
> By moving fwmark outside of union, you basically touch more cache lines in
> lookups. I have many machines doing XX.XXX of lookups per second, with long
> chains, already using 10% of CPU. I am sure a lot
On Thursday 09 November 2006 12:27, Thomas Graf wrote:
> Now that all protocols have been made aware of the mark
> field it can be moved out of the union thus simplyfing
> its usage.
>
> The config options in the IPv4/IPv6/DECnet subsystems
> to enable respectively disable mark based routing only
>
> The mark is already a bitfield, you may dividide it into separate
> marks with the exception of routes which do not yet support a
> mask.
Just checked, now that we have --and-mask and --or-mask, this is much
better than before.
The bitmask is OK when up to 32 marks are needed (like, for
class
* Steven Whitehouse <[EMAIL PROTECTED]> 2006-11-09 11:46
> On Thu, Nov 09, 2006 at 12:27:35PM +0100, Thomas Graf wrote:
> > Renames nfmark to mark and remove the dependency on netfilter
> > to ease usage by all subsystems. Also removes all the unneeded
> > config options to enable routing by fwmark
* Meelis Roos <[EMAIL PROTECTED]> 2006-11-09 14:32
> Another thought: sometimes a single mark makes rulesets inconvenient.
> What about several independent marks on a packet?
The mark is already a bitfield, you may dividide it into separate
marks with the exception of routes which do not yet suppo
Another thought: sometimes a single mark makes rulesets inconvenient.
What about several independent marks on a packet?
--
Meelis Roos <[EMAIL PROTECTED]>
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http
Auke Kok wrote:
This is what I was afraid of: even though the code allows you to bypass
the EEPROM checksum, the probe fails on a further check to see if the
MAC address is valid.
Since something with this NIC specifically made the EEPROM return all
0xff's, the MAC address is automatically i
Hi,
On Thu, Nov 09, 2006 at 12:27:35PM +0100, Thomas Graf wrote:
> Renames nfmark to mark and remove the dependency on netfilter
> to ease usage by all subsystems. Also removes all the unneeded
> config options to enable routing by fwmark, it can be safely
> enabled by default.
>
> Moves mark sel
Index: iproute2.git/include/linux/fib_rules.h
===
--- /dev/null 1970-01-01 00:00:00.0 +
+++ iproute2.git/include/linux/fib_rules.h 2006-11-09 11:48:07.0
+0100
@@ -0,0 +1,66 @@
+#ifndef __LINUX_FIB_RULES_H
+#
Introduces a new flag FIB_RULE_INVERT causing rules to apply
if the specified selector doesn't match.
Signed-off-by: Thomas Graf <[EMAIL PROTECTED]>
Index: net-2.6.20/include/linux/fib_rules.h
===
--- net-2.6.20.orig/include/linux/fi
Move the attribute policy for the non-specific attributes into
net/fib_rules.h and include it in the respective protocols.
Signed-off-by: Thomas Graf <[EMAIL PROTECTED]>
Index: net-2.6.20/include/net/fib_rules.h
===
--- net-2.6.20.or
Renames nfmark to mark and remove the dependency on netfilter
to ease usage by all subsystems. Also removes all the unneeded
config options to enable routing by fwmark, it can be safely
enabled by default.
Moves mark selector code from per protocol part into the generic
part and adds support for i
Move mark selector currently implemented per protocol into
the protocol independant part.
Signed-off-by: Thomas Graf <[EMAIL PROTECTED]>
Index: net-2.6.20/include/net/fib_rules.h
===
--- net-2.6.20.orig/include/net/fib_rules.h 20
nfmark is being used in various subsystems and has become
the defacto mark field for all kinds of packets. Therefore
it makes sense to rename it to `mark' and remove the
dependency on CONFIG_NETFILTER.
Signed-off-by: Thomas Graf <[EMAIL PROTECTED]>
Index: net-2.6.20/include/linux/skbuff.h
===
For the sake of consistency.
Signed-off-by: Thomas Graf <[EMAIL PROTECTED]>
Index: net-2.6.20/include/net/ip_fib.h
===
--- net-2.6.20.orig/include/net/ip_fib.h2006-11-08 15:34:12.0
+0100
+++ net-2.6.20/include/net/ip
Now that all protocols have been made aware of the mark
field it can be moved out of the union thus simplyfing
its usage.
The config options in the IPv4/IPv6/DECnet subsystems
to enable respectively disable mark based routing only
obfuscate the code with ifdefs, the cost for the
additional compari
On Thu, Nov 09, 2006 at 10:08:44AM +0100, Eric Dumazet ([EMAIL PROTECTED])
wrote:
> Here you test both KEVENT_SOCKET and KEVENT_PIPE
>
> > +#if defined CONFIG_KEVENT_SOCKET || defined CONFIG_KEVENT_PIPE
> > + kevent_storage_init(inode, &inode->st);
> > +#endif
> > }
> > return i
On Wed, Nov 08, 2006 at 02:07:32PM -0800, Tim Chen wrote:
> In my testing, the CPU utilization is at 100%. So
> increase in ACKs will cost CPU to devote more
> time to process those ACKs and reduce throughput.
Oh, I see. I would test on a real network with real clients. I doubt
you would observe
On Thursday 09 November 2006 09:23, Evgeniy Polyakov wrote:
> poll/select() notifications.
>
> This patch includes generic poll/select notifications.
> kevent_poll works simialr to epoll and has the same issues (callback
> is invoked not from internal state machine of the caller, but through
> proc
Hi All
I found an issue with Realtek 8139 driver (2.6.10 branch) at high
bi-directional traffic. On transmit timeout, driver's timeout callback
re-enables the receive interrupt. On the next receive interrupt, the ISR
disables the receive interrupt "only" when the receive poll task is not
active. B
Generic event handling mechanism.
Kevent is a generic subsytem which allows to handle event notifications.
It supports both level and edge triggered events. It is similar to
poll/epoll in some cases, but it is more scalable, it is faster and
allows to work with essentially eny kind of events.
Ev
Pipe notifications.
diff --git a/fs/pipe.c b/fs/pipe.c
index f3b6f71..aeaee9c 100644
--- a/fs/pipe.c
+++ b/fs/pipe.c
@@ -16,6 +16,7 @@ #include
#include
#include
#include
+#include
#include
#include
@@ -312,6 +313,7 @@ redo:
break;
}
Socket notifications.
This patch includes socket send/recv/accept notifications.
Using trivial web server based on kevent and this features
instead of epoll it's performance increased more than noticebly.
More details about various benchmarks and server itself
(evserver_kevent.c) can be found on
John wrote :
Yeah, looks like I was a bit overzealous on the warning squelch...
I'll cook-up a new patch that doesn't error-out.
I hope you do not bloat the kernel with meaningless warning messages.
Something simple like the following will do.
/* enable MWI */
/* Shut up the m
Core files.
This patch includes core kevent files:
* userspace controlling
* kernelspace interfaces
* initialization
* notification state machines
Some bits of documentation can be found on project's homepage (and links from
there):
http://tservice.net.ru/~s0mbre/old/?section=projects&item=
Description.
diff --git a/Documentation/kevent.txt b/Documentation/kevent.txt
new file mode 100644
index 000..ca49e4b
--- /dev/null
+++ b/Documentation/kevent.txt
@@ -0,0 +1,186 @@
+Description.
+
+int kevent_ctl(int fd, unsigned int cmd, unsigned int num, struct ukevent
*arg);
+
+fd - is t
poll/select() notifications.
This patch includes generic poll/select notifications.
kevent_poll works simialr to epoll and has the same issues (callback
is invoked not from internal state machine of the caller, but through
process awake, a lot of allocations and so on).
Signed-off-by: Evgeniy Po
Timer notifications.
Timer notifications can be used for fine grained per-process time
management, since interval timers are very inconvenient to use,
and they are limited.
This subsystem uses high-resolution timers.
id.raw[0] is used as number of seconds
id.raw[1] is used as number of nanosec
86 matches
Mail list logo
