Would be nice if it used atl_ not at_ so its less likely to cause
namespace clashes.
You have various macros for swaps that are pretty ugly - we have
cpu_to and le/be_to_cpu functions for most swapping cases and these are
generally optimised assembler (eg bswap on x86)
AT_DESC_USED/UNUSED would b
Evgeniy Polyakov wrote:
Possible solution:
a) it would be possible to have a "used" flag in each ring buffer entry.
That's too expensive, I guess.
b) kevent_wait needs another parameter which specifies the which is the
last (i.e., least recently added) entry in the ring buffer.
Everyth
On Sunday 19 November 2006 21:30, Jay Cliburn wrote:
> This patch contains the main C file for the Attansic L1 gigabit ethernet
> adapter driver.
Just a few style comments:
> + /* PCI config space info */
> + hw->vendor_id = pdev->vendor;
> + hw->device_id = pdev->device;
> + hw->
On Fri, 2006-11-17 at 15:51 -0800, David Kimdon wrote:
> Hi,
>
> commit 448bf25bc9e3d70a211fdf235426472089371c43 added
> ieee80211_get_wireless_stats in net/d80211/ieee80211_ioctl.c. At
> present we get a divide by zero (oops) if the low level driver does
> not set the new hw->maxssi field. Perh
On Sat, 2006-11-18 at 13:06 -0600, Larry Finger wrote:
> Chris Wright wrote:
> > -stable review patch. If anyone has any objections, please let us know.
> > --
> >
> > From: Michael Buesch <[EMAIL PROTECTED]>
> >
> > Drain the Microcode TX-status-FIFO before we enable IRQs.
> > T
On September 5, 2006, Johannes Berg changed the specifications for RSSI
processing. Implementing those changes involves specifying the bytes in
the RX header that contain the LNA (low-noise amplifier) bits, adding a
new argument to bcm43xx_rssi_postprocess, and the actual changes in
the rssi proces
On some of our systems (Linux 2.6, 4-way SMP), we have found a race where
occasionally recv() can detect a shutdown before the last bytes written to
the socket, and will exhibit the odd behavior where recv() will return 0
to indicate a shutdown socket, and a subsequent call will return the last
bit
Alright, a bit more thought later :)
Michael noted that letting people rename their wiphys is probably a good
idea to have stable based on the permanent mac address or such. Will
actually require moving the perm_addr field from struct ieee80211_hw
into struct wiphy (as below) too so that it's poss
On Sun, 19 Nov 2006 14:29:15 -0600 Jay Cliburn wrote:
> From: Jay Cliburn <[EMAIL PROTECTED]>
>
> This patch contains the build files for the Attansic L1 gigabit ethernet
> adapter driver.
>
> Signed-off-by: Jay Cliburn <[EMAIL PROTECTED]>
> ---
>
> Kconfig | 12
> Makefil
On Sun, 2006-11-19 at 20:29 +0100, Johannes Berg wrote:
> + memcpy(ndev->dev_addr, local->wiphy.perm_addr, ETH_ALEN);
> + memcpy(local->mdev->dev_addr, local->wiphy.perm_addr, ETH_ALEN);
ugh, somehow I messed up. that should be local->hw.perm_addr of course.
Same in patch 8.
johannes
From: Masahide NAKAMURA <[EMAIL PROTECTED]>
Date: Sat, 18 Nov 2006 18:08:42 +0900
> On Fri, 17 Nov 2006 08:48:31 -0500
> jamal <[EMAIL PROTECTED]> wrote:
>
> > Another one in the same spirit as before. Compiles. I dont have
> > a good test case, but looks right.
> > Nakamura-san please ACK and Da
From: Masahide NAKAMURA <[EMAIL PROTECTED]>
Date: Sat, 18 Nov 2006 17:51:59 +0900
> On Fri, 17 Nov 2006 08:34:44 -0500
> jamal <[EMAIL PROTECTED]> wrote:
>
> >
> > Found the cause of my problems.
> > Dave, this is against Linus tree because it is bug fix.
> >
> > Nakamura-san please ACK.
> >
From: Ville Nuorvala <[EMAIL PROTECTED]>
Date: Fri, 17 Nov 2006 21:27:21 +0200
> YOSHIFUJI Hideaki wrote:
> > In article <[EMAIL PROTECTED]> (at Fri, 17 Nov 2006 15:26:28 +0200), Ville
> > Nuorvala <[EMAIL PROTECTED]> says:
> >
> >
> >> - dst_release(&rt->u.dst);
> >> + if (rt
First batch of patches to convert the DECNet address
code to the new netlink interface.
-
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
Signed-off-by: Thomas Graf <[EMAIL PROTECTED]>
Index: net-2.6.20/net/decnet/dn_dev.c
===
--- net-2.6.20.orig/net/decnet/dn_dev.c 2006-11-19 18:31:06.0 +0100
+++ net-2.6.20/net/decnet/dn_dev.c 2006-11-19 18:33:43.0
The name rtmsg_ifa is heavly overused and confusing.
Signed-off-by: Thomas Graf <[EMAIL PROTECTED]>
Index: net-2.6.20/net/decnet/dn_dev.c
===
--- net-2.6.20.orig/net/decnet/dn_dev.c 2006-11-19 18:33:43.0 +0100
+++ net-2.6.20/
Extends the netlink interface to support the __le16 type and
converts address addition, deletion and, dumping to use the
new netlink interface.
Fixes multiple occasions of possible illegal memory references
due to not validated netlink attributes.
Signed-off-by: Thomas Graf <[EMAIL PROTECTED]>
I
From: "Michael Chan" <[EMAIL PROTECTED]>
Date: Thu, 16 Nov 2006 18:58:53 -0800
> David, A set of BNX2 patches for 2.6.20 to fix bugs, improve
> SerDes, and support a new chip. Please review and apply.
...
> [PATCH 1/17][BNX2]: Fix Xen problem.
> [PATCH 2/17][BNX2]: Improve SerDes handling.
> [P
>+char at_driver_name[] = "atl1";
>+static const char at_driver_string[] = "Attansic(R) L1 Ethernet Network
>Driver";
>+const char at_driver_version[] = DRV_VERSION;
>+static const char at_copyright[] =
>+"Copyright(c) 2005-2006 Attansic Corporation.";
>+
>+extern s32 at_read_mac_addr(struct
On Nov 19 2006 14:30, Jay Cliburn wrote:
>+
>+#define LBYTESWAP( a ) ( ( ( (a) & 0x00ff00ff ) << 8 ) | ( ( (a) &
>0xff00ff00 ) >> 8 ) )
>+#define LONGSWAP( a ) ( ( LBYTESWAP( a ) << 16 ) | ( LBYTESWAP( a )
>>> 16 ) )
>+#define SHORTSWAP( a ) ( ( (a) << 8 ) | ( (a) >> 8 ) )
Based upon feedback from Stephen Hemminger and Francois Romieu, please
consider this resubmitted patchset that provides support for the Attansic
L1 gigabit ethernet adapter. This patchset is built against 2.6.19-rc6.
The original patchset was submitted 20060927.
The monolithic version of this
From: Jay Cliburn <[EMAIL PROTECTED]>
This patch contains auxiliary C files for the Attansic L1 gigabit ethernet
adapter driver.
Signed-off-by: Jay Cliburn <[EMAIL PROTECTED]>
---
atl1_ethtool.c | 530 +++
atl1_hw.c | 840
++
From: Jay Cliburn <[EMAIL PROTECTED]>
This patch contains the header files needed by the Attansic L1 gigabit
ethernet adapter driver.
Signed-off-by: Jay Cliburn <[EMAIL PROTECTED]>
---
atl1.h | 251 ++
atl1_hw.h| 991
++
From: Jay Cliburn <[EMAIL PROTECTED]>
This patch contains the build files for the Attansic L1 gigabit ethernet
adapter driver.
Signed-off-by: Jay Cliburn <[EMAIL PROTECTED]>
---
Kconfig | 12
Makefile |1 +
atl1/Makefile | 30 ++
3 fil
On Sun, 2006-11-19 at 20:43 +0100, Michael Buesch wrote:
> On Sunday 19 November 2006 20:31, Johannes Berg wrote:
>
> > --- wireless-dev.orig/drivers/net/wireless/d80211/bcm43xx/bcm43xx_debugfs.c
> > 2006-11-19 20:12:47.629275208 +0100
> > +++ wireless-dev/drivers/net/wireless/d80211/bcm43xx/bcm4
* jamal <[EMAIL PROTECTED]> 2006-11-19 14:22
> On 11/19/06, jamal <[EMAIL PROTECTED]> wrote:
> >This is probably 2.6.19 material (a little annoyance)
> >I am testing using a new mailer - if this patch doesnt look right let me
> know.
>
> Seems like i failed. Ok, heres the patch again..
HTML mail
On Sunday 19 November 2006 20:31, Johannes Berg wrote:
> --- wireless-dev.orig/drivers/net/wireless/d80211/bcm43xx/bcm43xx_debugfs.c
> 2006-11-19 20:12:47.629275208 +0100
> +++ wireless-dev/drivers/net/wireless/d80211/bcm43xx/bcm43xx_debugfs.c
> 2006-11-19 20:14:01.189275208 +0100
> @@
If hardware shall do scanning, the hw_scan operation must be set. However,
if the driver is for multiple cards that may or may not do hardware
scanning, it'll need a flag.
Similar issues arise with passive_scan().
This patch introduces flags to fix these issues.
Signed-off-by: Johannes Berg <[EM
The calibration interval is far too hardware dependent to be useful as a
generic stack setting and some hardware doesn't even have that parameter.
Signed-off-by: Johannes Berg <[EMAIL PROTECTED]>
--- wireless-dev.orig/include/net/d80211.h 2006-11-19 20:14:02.289275208
+0100
+++ wireless-dev
There's no point in trying to tell a driver globally whether sw or hw
crypto is used, if it's sw then we just don't give it keys...
Besides, these weren't ever used!
Remove IEEE80211_CONF_SW_DECRYPT and IEEE80211_CONF_SW_ENCRYPT.
Acked-by: Michael Buesch <[EMAIL PROTECTED]>
Signed-off-by: Johanne
Updates bcm43xx.
Signed-off-by: Johannes Berg <[EMAIL PROTECTED]>
--- wireless-dev.orig/drivers/net/wireless/d80211/bcm43xx/Makefile
2006-11-19 20:12:47.459275208 +0100
+++ wireless-dev/drivers/net/wireless/d80211/bcm43xx/Makefile 2006-11-19
20:14:01.189275208 +0100
@@ -7,6 +7,6 @@ bcm43
This introduces the promised IEEE80211_HW_FRAG used to determine whether the
hardware can do fragmentation or not.
It actually makes sense to split this capability flag out from the function
pointer, maybe some hardware wants to be notified about the fragmentation
threshold even though it doesn't
struct ieee80211_ops has a driver name field that's never used.
Remove it.
Signed-off-by: Johannes Berg <[EMAIL PROTECTED]>
--- wireless-dev.orig/include/net/d80211.h 2006-11-19 20:14:00.039275208
+0100
+++ wireless-dev/include/net/d80211.h 2006-11-19 20:14:00.639275208 +0100
@@ -531,9 +5
After removing knowledge of the master net_dev from drivers,
they'll still need a way to tell us which MAC address they have.
This is that way, the perm_addr is initially used for all devices.
Signed-off-by: Johannes Berg <[EMAIL PROTECTED]>
--- wireless-dev.orig/include/net/d80211.h 2006-11
Traditionally, drivers were given a struct net_device * in order
to identify the wireless device. This was the master device, but
I'm trying to cut down it's use. Now, there long was a comment
that this might change. That time has come, this patch gives back
a struct ieee80211_hw pointer.
Currentl
This patch changes a bunch of prototypes to have struct ieee80211_local*
instead of struct net_device* where that makes sense.
Signed-off-by: Johannes Berg <[EMAIL PROTECTED]>
--- wireless-dev.orig/net/d80211/ieee80211.c2006-11-19 20:13:56.719275208
+0100
+++ wireless-dev/net/d80211/ieee8021
This patch reduces mdev usage by replacing struct net_device *
arguments that are never used except for dereferencing to get
struct ieee80211_local from ieee80211_ptr by struct
ieee80211_local directly.
Also, this patch changes ieee80211_rx_mgmt to no longer be callable
when local->apdev is NULL.
This patch reduces mdev usage by replacing struct net_device *
arguments that are never used except for dereferencing to get
struct ieee80211_local from ieee80211_ptr by struct
ieee80211_local directly. Also removes ->master from sub_if_data.
Signed-off-by: Johannes Berg <[EMAIL PROTECTED]>
--- w
This patch reduces mdev usage by replacing struct net_device *
arguments that are never used except for dereferencing to get
struct ieee80211_local from ieee80211_ptr by struct
ieee80211_local directly.
Signed-off-by: Johannes Berg <[EMAIL PROTECTED]>
--- wireless-dev.orig/net/d80211/ieee80211.c
"for (; condition ;)"??
Ever heard of while loops?
Also clean up some list handling (still. *sigh*)
Signed-off-by: Johannes Berg <[EMAIL PROTECTED]>
--- wireless-dev.orig/net/d80211/ieee80211_sta.c2006-11-19
20:13:52.269275208 +0100
+++ wireless-dev/net/d80211/ieee80211_sta.c 2006-1
This is a repost of the whole patchset modified according to various
comments.
I've folded patches 6, 7 and 8 into one driver API update patch that is
smaller than the three combined. Unfortunately, it's still very large.
I've also changed the introduction of ieee80211_wiphy, in fact the patch
doe
This is probably 2.6.19 material (a little annoyance)
I am testing using a new mailer - if this patch doesnt look right let me know.
cheers,
jamal
-
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.
From: David Stevens <[EMAIL PROTECTED]>
Date: Thu, 16 Nov 2006 18:49:30 -0700
> The IGMPV3_EXP() macro doesn't correctly shift the normalization bit, so
> time-out values are longer than they should be. Patch below for viewing
> and attached for applying.
>
> Thanks to Dirk Ooms for finding the p
On Sun, 2006-11-19 at 09:25 -0800, David Kimdon wrote:
> Perhaps that is a split that we do not need? I don't see the problem
> that 'd80211: split ieee80211_hw' is solving. I do see what it is
> doing, but maybe I am missing something . . .
Oh, I just figured that on 64-bit systems the _ops
--- Begin Message ---
http://bugzilla.kernel.org/show_bug.cgi?id=7551
Summary: Fluctuating and slow network perfomance
Kernel Version: 2.6.18.x
Status: NEW
Severity: high
Owner: [EMAIL PROTECTED]
Submitter: [EMAIL PROTECTED]
Most recen
On Sun, Nov 19, 2006 at 05:57:39PM +0100, Johannes Berg wrote:
> On Sun, 2006-11-19 at 08:55 -0800, David Kimdon wrote:
>
> > ok. I am concerned that making this split between per driver and per
> > card is difficult to get right. Setting or not setting a function
> > pointer for an operation is
Removes dependency on buggy rta_buf, fixes a memory corruption bug due to
a unvalidated netlink attribute, and simplifies the code.
Signed-off-by: Thomas Graf <[EMAIL PROTECTED]>
Index: net-2.6.20/net/bridge/br_netlink.c
===
--- net-
On Sun, 2006-11-19 at 08:55 -0800, David Kimdon wrote:
> ok. I am concerned that making this split between per driver and per
> card is difficult to get right. Setting or not setting a function
> pointer for an operation is fairly standard practice and I don't see
> the value in introducing yet
On Sun, Nov 19, 2006 at 05:34:49PM +0100, Johannes Berg wrote:
> On Sun, 2006-11-19 at 07:56 -0800, David Kimdon wrote:
>
> > What is wrong with the driver setting the function pointer to NULL for
> > the cards that do not support scanning? Where does this requirment
> > come from that the functi
On Sun, 2006-11-19 at 17:34 +0100, Johannes Berg wrote:
> Well I want to allow drivers to make assign the 33 function pointers in
> a static structure, and differences between cards must then be handled
> in the non-static part.
Uh, that didn't come out too well.
The point is that ieee80211_ops
> > + /* if no one will get it anyway, don't even allocate it.
> > +* unlikely because this is only relevant for APs
> > +* where the device must be open... */
> > + if (unlikely(!local->apdev))
> > + return;
>
> Why not just let it Oops and show the bug? In what cases is i
On Sun, 2006-11-19 at 17:15 +0100, Michael Buesch wrote:
> > - if (local->ops->hw_scan) {
> > + if (local->ops->hw_scan && local->wiphy.flags & IEEE80211_HW_SCAN) {
>
> Please wrap this into ()
Good point :)
johannes
signature.asc
Description: This is a digitally signed message part
On Sun, 2006-11-19 at 07:56 -0800, David Kimdon wrote:
> What is wrong with the driver setting the function pointer to NULL for
> the cards that do not support scanning? Where does this requirment
> come from that the function pointers in struct ieee80211_wiphy be
> identical for all cards?
Well
Reply-To:
In-Reply-To: <[EMAIL PROTECTED]>
>
> --- wireless-dev.orig/net/d80211/ieee80211.c 2006-11-17 20:01:54.999703408
> +0100
> +++ wireless-dev/net/d80211/ieee80211.c 2006-11-17 20:01:55.659703408
> +0100
> @@ -210,9 +210,16 @@ static void ieee80211_key_threshold_noti
>
On Sunday 19 November 2006 00:53, Johannes Berg wrote:
> There's no point in trying to tell a driver globally whether sw or hw
> crypto is used, if it's sw then we just don't give it keys...
>
> Besides, these weren't ever used!
> Remove IEEE80211_CONF_SW_DECRYPT and IEEE80211_CONF_SW_ENCRYPT.
>
On Sunday 19 November 2006 01:21, Johannes Berg wrote:
> If hardware shall do scanning, the hw_scan operation must be set. However,
> if the driver is for multiple cards that may or may not do hardware
> scanning, it'll need a flag.
>
> Similar issues arise with passive_scan().
>
> This patch int
On Saturday 18 November 2006 20:02, Larry Finger wrote:
> Ray Lee wrote:
> > Larry Finger wrote:
> >> Johannes Berg wrote:
> >>> Hah, that's a lot more plausible than bcm43xx's drain patch actually
> >>> causing this. So maybe somehow interrupts for bcm43xx aren't routed
> >>> properly or something
On Sun, Nov 19, 2006 at 01:21:13AM +0100, Johannes Berg wrote:
> If hardware shall do scanning, the hw_scan operation must be set. However,
> if the driver is for multiple cards that may or may not do hardware
> scanning, it'll need a flag.
What is wrong with the driver setting the function pointe
Please consider the patch below. The original consensus was that
since this is documented in the errata it should be done by default
rather than as a config option. I think the original author of the
patch doesn't have time to submit a revised version but now that
there's a new maintainer of the
On Sat, 18 Nov 2006, Adrian Bunk wrote:
> This patch removes net drivers that:
> - had already been marked as BROKEN in 2.6.0 three years ago and
> - are still marked as BROKEN.
>
> These are the following drivers:
> - MAC89x0
I still have a patch lying around for this one:
http://linux-m68k-cvs
60 matches
Mail list logo
