On Fri, Jul 08, 2005 at 09:42:21PM +1000, Paul Mackerras wrote:
Some comments on the MPPE kernel patch (sorry it's taken me so long):
Likewise, my apologies for the delay in implementing your
suggestions. I do really appreciate your feedback.
+static inline struct sk_buff *
This patch implements the Microsoft Point-to-Point Tunnelling Protocol
kernel module, with minor changes to ppp_generic to have it drop
packets if they are not compressed/decompressed properly. This adds a
new ppp flag SC_MUST_COMPRESS which will be set by pppd upon CCP UP
with MPPE enabled.
PPP 2.4.3 patch creates the new flag SC_MUST_COMPRESS. This is set on
CCP UP with MPPE enabled, and never cleared. The kernel uses this
flag to drop packets which aren't compressed/decompressed properly
when MPPE is enabled.
In the existing code, on SIGUSR2, CCP is renegotiated, and with MPPE
Matt Domsch wrote:
include/linux/if_ppp.h | 62
+
- * 2. Redistributions in binary form must reproduce the above copyright
- *notice, this list of conditions and the following disclaimer in
- *the documentation and/or other materials
Patrick McHardy wrote:
Matt Domsch wrote:
include/linux/if_ppp.h | 62
+
- * 2. Redistributions in binary form must reproduce the above copyright
- *notice, this list of conditions and the following disclaimer in
- *the
David S. Miller wrote:
I just double-checked, and I didn't lose it during the
recent rebasing, as the tree I started with didn't have
the skb-stamp patch either.
Are you sure I did apply it at some point?
Yes, I know it was in your tree before the first rebasing.
Either way, please resend
On Sat, Aug 13, 2005 at 02:35:53AM +0200, Patrick McHardy ([EMAIL PROTECTED])
wrote:
Hi,
besides a small bugfix, this patchset adds support for dynamic number
of groups to netlink. To support an arbitary number of groups a couple
of changes had to me made, I'll explain them below. The
gcc-4.0.1 (new debian default) emits a couple of warnings in
the dccp code, this patch silences them.
[DCCP]: Fix compiler warnings
may be a false warning if there always is something on ccid3hcrx_hist:
net/dccp/ccids/ccid3.c: In function 'ccid3_hc_rx_packet_recv':
net/dccp/ccids/ccid3.c:1634:
Hello,
I have this error report on ifconfig (also a problem with ip link) that a
down interface receives packets. I remeber this was discussed before but
cannot find a result of the outcome. Is this a driver problem?
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=322654
Bernd
--
(OO)
On Sun, Aug 14, 2005 at 04:31:47PM +0200, Patrick McHardy wrote:
Patrick McHardy wrote:
Matt Domsch wrote:
include/linux/if_ppp.h | 62
+
- * 2. Redistributions in binary form must reproduce the above copyright
- *notice,
(rediff with license on include/linux/if_ppp.h untouched, and I had
named the new flag SC_MUST_COMP not SC_MUST_COMPRESS)
PPP 2.4.3 patch creates the new flag SC_MUST_COMP. This is set on
CCP UP with MPPE enabled, and never cleared. The kernel uses this
flag to drop packets which aren't
Patrick McHardy wrote:
-static inline const int before48(const u64 seq1, const u64 seq2)
+static inline int before48(const u64 seq1, const u64 seq2)
Perhaps __attribute__ ((const)) is what was meant here?
-Mitch
-
To unsubscribe from this list: send the line unsubscribe netdev in
the body of
Ben Greear wrote:
Patrick McHardy wrote:
+static inline void skb_get_timestamp(struct sk_buff *skb, struct
timeval *stamp)
+{
+stamp-tv_sec = skb-tstamp.off_sec;
+stamp-tv_usec = skb-tstamp.off_usec;
+if (skb-tstamp.off_sec) {
+stamp-tv_sec += skb_tv_base.tv_sec;
Evgeniy Polyakov wrote:
You've done a good job, thanks Patrick.
One small comment after diagonal patchset review -
last patch will fail to apply on top of w1 changes :)
since it already has a group changes to 1.
The w1 changes don't seem to be in mainline yet. I'm going to fix
up your
Evgeniy's patch fixed to apply on top of my netlink changes.
[NETLINK]: w1_int.c: fix default netlink group
w1 does not need to multicast its state to several groups at once,
and upcoming netlink changes will not allow bitmask for groups anyway.
Signed-off-by: Evgeniy Polyakov [EMAIL PROTECTED]
On Sat, Aug 13, 2005 at 02:10:55PM -0700, David S. Miller wrote:
Yes, you need an atomic in the fast path, but it's only done
for fast clones and their parent SKBs, and such an atomic is
_MUCH_ cheaper than a kmem_cache_alloc especially since it
touches more cache local state (the atomic
Em Sun, Aug 14, 2005 at 07:11:46PM +0200, Patrick McHardy escreveu:
gcc-4.0.1 (new debian default) emits a couple of warnings in
the dccp code, this patch silences them.
[DCCP]: Fix compiler warnings
Thanks Patrick, I'm applying this to the dccp-2.6.14.git tree.
- Arnaldo
-
To unsubscribe
On Sun, Aug 14, 2005 at 05:02:37PM -0700, David S. Miller wrote:
Bringing an interface down does not disassosciate the IP addresses
confiured to that interface from the system.
Yes, but I am not sure if that is the case here, we need more input from the
reporter. It looked more like the down
From: Thomas Graf [EMAIL PROTECTED]
Date: Sun, 14 Aug 2005 02:31:30 +0200
The disadvantage is that we cannot do clone resycling easly
anymore, that is we allow a fast-clone'able skb to be cloned
up again if the clone portion was already freed.
I do not see much value in that capability
From: Herbert Xu [EMAIL PROTECTED]
Date: Mon, 15 Aug 2005 09:52:51 +1000
1) When you allocate for fast clones you do two kmem_cache_alloc's
in a row. You store the second pointer in the shared skb data area.
The fast path within kmem_cache_alloc is good enough that the second
one should be
From: Patrick McHardy [EMAIL PROTECTED]
Date: Sun, 14 Aug 2005 23:23:32 +0200
[NET]: Store skb-timestamp as offset to a base timestamp
Reduces skb size by 8 bytes.
Signed-off-by: Patrick McHardy [EMAIL PROTECTED]
Updated patch applied, thanks Patrick.
-
To unsubscribe from this list: send
[NETLINK]: w1_int.c: fix default netlink group
w1 does not need to multicast its state to several groups at once,
and upcoming netlink changes will not allow bitmask for groups anyway.
Signed-off-by: Evgeniy Polyakov [EMAIL PROTECTED]
Signed-off-by: Patrick McHardy [EMAIL PROTECTED]
---
commit
[NETLINK]: Fix missing dst_groups initializations in netlink_broadcast users
netlink_broadcast users must initialize NETLINK_CB(skb).dst_groups to the
destination group mask for netlink_recvmsg.
Signed-off-by: Patrick McHardy [EMAIL PROTECTED]
---
commit
[NETLINK]: Fix module refcounting problems
Use-after-free: the struct proto_ops containing the module pointer
is freed when a socket with pid=0 is released, which besides for kernel
sockets is true for all unbound sockets.
Module refcount leak: when the kernel socket is closed before all user
[NETLINK]: Use group numbers instead of bitmasks internally
Using the group number allows increasing the number of groups without
beeing limited by the size of the bitmask. It introduces one limitation
for netlink users: messages can't be broadcasted to multiple groups anymore,
however this
On Sun, Aug 14, 2005 at 05:16:58PM -0700, David S. Miller wrote:
I think the atomic operation will still be cheaper than these
two function calls. Even though the second one will be cheap
we could be completely done with just one call and not even pay
the reduced price of the second
From: Herbert Xu [EMAIL PROTECTED]
Date: Mon, 15 Aug 2005 10:42:44 +1000
Well I don't think I understand the new skb-user solution yet :)
For a start, skb-users will prevent the skb-data area from being
freed as well as the skb itself. We'll also need to audit the TCP
code paths to make
Replace hardware model specific #ifdef's in struct local_info with
void *hw_priv that is pointing to cs/pci/plx specific data
structure. This removes unneeded #ifdef's and as such, is a step
towards making it possible to share objects for hostap_hw.c and
hostap_download.c with cs/pci/plx drivers
PCI version of Prism2.5/3 has undocumented DMA support for TX/RX data,
but this seems to have some hardware bugs that prevent it from being
used properly for TX. RX side could possibly be made to work reliably.
Even though DMA support would be very useful for saving host CPU (from
about 40% to
This set of patches merges a bug fix for the Host AP driver (patch 1),
does some cleanup for different hardware models (patches 2-3), continues
conversion to generic ieee80211 code (patches 4-6), and tests how we can
handle changes that have a dependency to modify ieee80211 and hostap
at the same
IEEE 802.11 frame control has two bits reserved for protocol
version. IEEE80211_FCTL_VERS was not used anywhere, but I would assume
it was supposed to be a mask for the protocol field and as such, it
should be 0x0003, not 0x0002. This matches with WLAN_FC_PVER
definition in hostap.
Signed-off-by:
Replace hostap-specific struct hostap_ieee80211_hdr with struct
ieee80211_hdr from net/ieee80211.h.
Signed-off-by: Jouni Malinen [EMAIL PROTECTED]
Index: netdev-2.6/drivers/net/wireless/hostap/hostap_80211.h
===
---
Remove unused defines that are already available from generic kernel
header files.
Signed-off-by: Jouni Malinen [EMAIL PROTECTED]
Index: netdev-2.6/drivers/net/wireless/hostap/hostap_wlan.h
===
---
Remove backwards compatibility define for WLAN_CAPABILITY_ESS now that
net/ieee80211.h defines this.
Signed-off-by: Jouni Malinen [EMAIL PROTECTED]
Index: netdev-2.6/drivers/net/wireless/hostap/hostap_common.h
===
---
34 matches
Mail list logo