Ok, here is the current version of the patch, and what I intend
to push to Linus:
diff-tree 4980a059ef42741e80e9efa0dabdf520f9ba0c5a (from
6b39374a27eb4be7e9d82145ae270ba02ea90dc8)
Author: David S. Miller <[EMAIL PROTECTED]>
Date: Thu Sep 1 15:06:18 2005 -0700
[TCP]: Keep TSO e
From: Mike Kershaw <[EMAIL PROTECTED]>
Date: Tue, 23 Aug 2005 19:48:37 -0400
> Same patch as before, only following Dave's advice to not change the
> link type of an interface while its up.
Applied, thanks Mike.
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of
From: Geoff Levand <[EMAIL PROTECTED]>
Date: Mon, 29 Aug 2005 15:04:24 -0700
> Sure, your fix works, and seems to be the best way to do it.
>
> Signed-off-by: Geoff Levand <[EMAIL PROTECTED]>
Applied, thanks everyone.
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the bod
From: Harald Welte <[EMAIL PROTECTED]>
Date: Tue, 30 Aug 2005 16:11:19 +0200
> Hi Dave, Please apply.
Applied, thanks.
-
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.ht
From: "Michael Chan" <[EMAIL PROTECTED]>
Date: Mon, 29 Aug 2005 13:29:50 -0700
> On Sat, 2005-08-27 at 22:38 +0200, Eric Dumazet wrote:
>
> >
> > - [TG3] : tx_lock spinlock is taken in tg3_tx() only when really needed.
> >
>
> This is similar to your tx_lock patch for tg3 but takes it one step
From: Adrian Bunk <[EMAIL PROTECTED]>
Date: Wed, 24 Aug 2005 17:58:06 +0200
> "extern inline" doesn't make much sense.
>
> Signed-off-by: Adrian Bunk <[EMAIL PROTECTED]>
Applied, thanks Adrian.
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL
From: Ben Greear <[EMAIL PROTECTED]>
Date: Thu, 01 Sep 2005 15:39:31 -0700
> Also, earlier in the method it does a __dev_get_by_index(parm->ifindex),
> and continues to use the returned value after that. Couldn't this lead
> to a reference-after-free, or does external locking prohibit this?
Prob
From: YOSHIFUJI Hideaki <[EMAIL PROTECTED]>
Date: Thu, 01 Sep 2005 20:51:14 +0900 (JST)
> This patch fixes the issue by using appropriate incoming interface,
> in the sense of scoping architecture.
>
> Signed-off-by: YOSHIFUJI Hideaki <[EMAIL PROTECTED]>
Applied, thanks.
-
To unsubscribe from th
From: Herbert Xu <[EMAIL PROTECTED]>
Date: Fri, 2 Sep 2005 10:43:40 +1000
> I've finally found a potential cause of the sk_forward_alloc underflows
> that people have been reporting sporadically.
>
> When tcp_sendmsg tacks on extra bits to an existing TCP_PAGE we don't
> check sk_forward_alloc ev
From: Herbert Xu <[EMAIL PROTECTED]>
Date: Fri, 2 Sep 2005 08:11:37 +1000
> On Thu, Sep 01, 2005 at 03:06:47PM -0700, David S. Miller wrote:
> >
> > - if (TCP_SKB_CB(buff)->sacked&TCPCB_LOST) {
> > - tp->lost_out += tcp_skb_pcount(buff);
> >
From: Herbert Xu <[EMAIL PROTECTED]>
Date: Fri, 2 Sep 2005 11:08:12 +1000
> Yes, because
>
> diff = pcount(orig_skb) - (pcount(skb) + pcount(buff))
><= pcount(orig_skb)
>
> Now if orig_skb is marked as TCPCB_LOST, then by definition
>
> tp->lost_out >= pcount(orig_skb)
>
From: Herbert Xu <[EMAIL PROTECTED]>
Subject: Re: [PATCHES]: Two TSO refinements
Date: Fri, 2 Sep 2005 14:30:26 +1000
> On Thu, Sep 01, 2005 at 09:28:16PM -0700, David S. Miller wrote:
> >
> > > Therefore,
> > >
> > > tp->lost_out >= diff
>
From: Ben Greear <[EMAIL PROTECTED]>
Date: Mon, 29 Aug 2005 16:01:11 -0700
> Latest netdevice ref-count debugging patch is up. The
> patch is against 2.6.13:
>
> http://www.candelatech.com/oss/rfcnt.patch
I reviewed this, and I think the approach can be refined and made more
robust.
The worst
From: Ben Greear <[EMAIL PROTECTED]>
Date: Thu, 01 Sep 2005 23:11:28 -0700
> A quick optimization is to kmalloc chunks of 1000 or so structs at once
> and then write a cheap foomalloc method to grab and release them. We
> already take a lock (at least in my implementation), so this should be
> sm
From: John Heffner <[EMAIL PROTECTED]>
Date: Thu, 1 Sep 2005 22:51:48 -0400
> I have an idea why this is going on. Packets are pre-allocated by the
> driver to be a max packet size, so when you send small packets, it
> wastes a lot of memory. Currently Linux uses the packets at the
> beginnin
From: Ben Greear <[EMAIL PROTECTED]>
Date: Thu, 01 Sep 2005 23:55:53 -0700
> Ok, so each object that now has a net_device* dev pointer instead
> gets this netdev_pointer object. When a reference is taken,
> this netdev_pointer object is linked into the netdevice object
> in a list. This requires
From: Ben Greear <[EMAIL PROTECTED]>
Date: Fri, 02 Sep 2005 13:53:58 -0700
> I'm trying to write some code to walk all of the places where
> routes might exist and find any that reference a particular
> net-device.
Man, good luck. The IPSEC destination cache entries live in
chains only obtainabl
From: Ben Greear <[EMAIL PROTECTED]>
Date: Fri, 02 Sep 2005 09:35:59 -0700
> Is there any reason we couldn't create a method:
>
> void dev_hold_ref(struct dbg_dev_rfcnt_info* i, netdevice* dev, void* key);
Please strongly consider my "netdev_pointer" idea,
and drop this dynamically allocated db
You can, BTW, log the value of "__builtin_return_address(0)" at
the time of the neighbour table entry creation, so see who asked
for it. That might help track it down.
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo
All 5 patches applied, looks really good.
In particular, patch 4 with the prefetching, was particularly
nice to see :-)
Thanks a lot Michael.
-
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.kern
From: Ben Greear <[EMAIL PROTECTED]>
Date: Mon, 05 Sep 2005 23:50:45 -0700
> In net/core/neighbour.c, method neigh_alloc, the
> ref-count is set to one near the bottom of the method.
>
> I notice in neigh_create, for instance, the ref-count is
> increased again as the neighbour is put into the ta
From: Eric Dumazet <[EMAIL PROTECTED]>
Date: Tue, 06 Sep 2005 01:51:16 +0200
> Avoid touching file->f_dentry on sockets, since file->private_data directly
> gives us the socket pointer.
>
> Signed-off-by: Eric Dumazet <[EMAIL PROTECTED]>
Applied, thanks Eric.
-
To unsubscribe from this list: se
From: Eugene Surovegin <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 00:40:01 -0700
> According to Documentation/networking/netdevices.txt
> dev->hard_start_xmit must be called with interrupts *enabled*.
>
> Unfortunately, current netconsole code always calls netpoll with local
> interrupts disable
From: Eric Dumazet <[EMAIL PROTECTED]>
Date: Tue, 06 Sep 2005 16:11:19 +0200
> I dont know if it's safe to change l_linger to 'unsigned int' in the include
> file (It might be defined as int in ABI specs)
>
> So I believe this patch is needed :
>
> Signed-off-by: Eric Dumazet <[EMAIL PROTECTED]
From: Ingo Oeser <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 16:57:41 +0200
> I'm just asking myself, why is AES-256 not announced by the IPsec framework?
> The kernel crypto-API seems to support a keysize of 256.
> Or is the blocksize (of 256 bits) meant by AES-256?
>
> I'm a bit lost on this one.
From: Eugene Surovegin <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 15:04:17 -0700
> David, correct me if I'm wrong, but I think there is a major problem
> with current netconsole/netpoll approach.
You're preaching to the choir. I think the whole netpoll
implementation is fundamentally flawed, and
From: Matt Mackall <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 15:32:50 -0700
> Unfortunately, netpoll fundamentally requires everything to work with
> interrupts off. This can't be changed. It could be called from the
> context of an oops, or worse, by the kgdb stub at a breakpoint. If
> drivers re
From: Ben Greear <[EMAIL PROTECTED]>
Date: Tue, 06 Sep 2005 14:37:26 -0700
> I'm also attaching a patch that just fixes the problem,
> with no debugging info. (Compiled but not tested by
> itself.)
>
> Signed-off-by Ben Greear <[EMAIL PROTECTED]>
Thanks for tracking this down, I'll review it mo
From: Stephen Hemminger <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 14:01:17 -0700
> The new timestamp get/set routines should have const attribute
> on parameters (helps to indicate direction).
>
> Signed-off-by: Stephen Hemminger <[EMAIL PROTECTED]>
Applied, thanks Stephen.
-
To unsubscribe fro
From: Stephen Hemminger <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 13:26:15 -0700
> This was found by inspection while looking for checksum problems
> with the skge driver that sets CHECKSUM_HW. It did not fix the
> problem, but it looks like it is needed.
>
> If IP reassembly is trimming an overl
From: Matt Mackall <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 15:50:25 -0700
> > So we try to spit out netconsole messages in hw IRQ context and stuff
> > like that, as you stated. The tg3 driver is susceptible to the
> > problem you mention, as is bnx2, because they use purely software
> > interr
From: Patrick McHardy <[EMAIL PROTECTED]>
Date: Wed, 07 Sep 2005 01:02:01 +0200
> You're right, good catch. This patch fixes it by moving the lock
> down to the list-operation which it is supposed to protect.
I think we need to unlink from the list first if you're
going to do it this way. Otherw
From: Matt Mackall <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 18:03:38 -0700
> Option c) is obviously a big project but maybe we can get from here to
> there. One possible step in that direction would be exposing a
> standard driver lock that netpoll can see and switching drivers that
> have troubl
> Thu, 07 Jul 2005 03:51:31 -0700
> committer David S. Miller <[EMAIL PROTECTED]> Thu, 07 Jul 2005 03:51:31 -0700
>
> [CRYPTO] Add plumbing for multi-block operations
So, does Herbert's patch posted today fix the bug or not?
-
To unsubscribe from this list: send the line &
From: Patrick McHardy <[EMAIL PROTECTED]>
Date: Wed, 07 Sep 2005 01:42:48 +0200
> The only other user of proto_list besides proto_register, which
> doesn't care, are the seqfs functions. They use the slab pointer,
> but in a harmless way:
>
> proto->slab == NULL ? "no" : "yes"
From: Ben Greear <[EMAIL PROTECTED]>
Date: Tue, 06 Sep 2005 16:03:36 -0700
> At any rate, I'll be happy to see this fix go in unless someone
> finds a problem with it!
As Yoshifuji showed, NUD_INCOMPLETE is a part of the
bitmask NUD_IN_TIMER, as is NUD_DELAY and a few others.
So it is actually a
From: YOSHIFUJI Hideaki <[EMAIL PROTECTED]>
Date: Wed, 07 Sep 2005 08:36:41 +0900 (JST)
> In article <[EMAIL PROTECTED]> (at Tue, 06 Sep 2005 13:05:16 -0700), Ben
> Greear <[EMAIL PROTECTED]> says:
>
> > neigh->nud_state = NUD_INCOMPLETE;
> > *** neigh_hol
From: Matt Mackall <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 20:08:10 -0700
> Think upon the kgdb-over-ethernet case, please. The kernel hits a
> breakpoint, the kgdb stub stops everything, sends a packet to the
> debugging client, waits for a packet back.. This simply can't work if
> we delay sen
From: Ben Greear <[EMAIL PROTECTED]>
Date: Tue, 06 Sep 2005 20:24:38 -0700
> How about this call site? The check is for new & NUD_IN_TIMER,
> but there is no guarantee (that I can see) that neigh->nud_timer
> has any of the NUD_IN_TIMER bits set. The one place earlier
> that sets neigh->nud_time
From: Matt Mackall <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 20:35:32 -0700
> Kgdboe simply wants the asynchronous subsystem to get out of the
> way. For most cards, it works just fine.
It wants to impose a locking model restriction which never ever
existed in the past.
As I said, I myself even
From: Evgeniy Polyakov <[EMAIL PROTECTED]>
Date: Fri, 2 Sep 2005 18:00:55 +0400
> sock_sendfile() and generic_file_sendpage() were implemented
> and presented in the attached patch.
> Such methods allows to use sendfile() for any file descriptor <-> file
> descriptor usage, especially usefull it i
From: Ingo Oeser <[EMAIL PROTECTED]>
Date: Wed, 7 Sep 2005 15:39:08 +0200
> Jeff Garzik wrote:
> > I find them useful in my own drivers; they are definitely not pure noise.
>
> gcc -finstrument-functions
I was going to mention this as well, and also the idea to
enable CONFIG_MCOUNT on a per-file
From: Matt Mackall <[EMAIL PROTECTED]>
Date: Wed, 7 Sep 2005 10:04:29 -0700
> Netpoll is intended to be a replacement for the requirement of serial
> cables for doing things like tricky oops capture and kgdb.
>
> If we move to softirq-only processing, it will not be a serial cable
> replacement.
From: Francois Romieu <[EMAIL PROTECTED]>
Date: Wed, 7 Sep 2005 20:59:49 +0200
> Stephen Hemminger <[EMAIL PROTECTED]> :
> [...]
> > It really sounds like netconsole needs to have a different device hook
> > instead of start_xmit. It also probably doesn't want to have allocate
> > an skb. What y
From: Stephen Hemminger <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 21:40:46 -0700
> Since packets almost never contain extra garbage at the end, it is
> worthwhile to optimize for that case.
>
> Signed-off-by: Stephen Hemminger <[EMAIL PROTECTED]>
Applied, thanks Stephen.
-
To unsubscribe from th
From: Ben Greear <[EMAIL PROTECTED]>
Date: Wed, 07 Sep 2005 12:36:20 -0700
> Thanks for catching that. A new patch is attached. I didn't see any
> printouts for this message even with the fix, so this case must not be
> hit (at least not often).
>
> Please let me know if you see any more areas
From: Stephen Hemminger <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 21:45:55 -0700
> A UDP packet may contain extra data that needs to be trimmed off.
> But when doing so, UDP forgets to fixup the skb checksum if CHECKSUM_HW
> is being used.
>
> I think this explains the case of a NFS receive using
From: Stephen Hemminger <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 21:47:47 -0700
> Several places in IPV6 need to use pskb_trim_rcsum to handle
> the case of skb's received on devices that set CHECKSUM_HW
>
> Signed-off-by: Stephen Hemminger <[EMAIL PROTECTED]>
Everything looks fine except this
From: Ben Greear <[EMAIL PROTECTED]>
Date: Wed, 07 Sep 2005 14:14:42 -0700
> David S. Miller wrote:
> > Doing an add_timer() for something already scheduled it fine.
>
> Does it actually cause a second timer to fire? If not, this
> would keep us from correctly decremen
From: Julian Anastasov <[EMAIL PROTECTED]>
Date: Thu, 8 Sep 2005 10:01:39 +0300 (EEST)
> One such place that can damage the dst refcnts is route.c
> with CONFIG_IP_ROUTE_MULTIPATH_CACHED enabled, i don't see the user's
> .config. In this new code i see that rt_intern_hash is called before
>
From: Stephen Hemminger <[EMAIL PROTECTED]>
Date: Wed, 7 Sep 2005 14:42:44 -0700
> Fix pskb_trim usage in ipv6. Only the udp one is really
> a bug, other places are just doing equivalent code.
>
> Signed-off-by: Stephen Hemminger <[EMAIL PROTECTED]>
Applied, thanks for fixing up this patch.
-
To
From: YOSHIFUJI Hideaki <[EMAIL PROTECTED]>
Date: Thu, 08 Sep 2005 14:27:20 +0900 (JST)
> git://git.skbuff.net/gitroot/yoshfuji/linux-2.6-git-rfc3542.git/
Pulled, thanks.
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordo
From: "Michael Chan" <[EMAIL PROTECTED]>
Date: Thu, 08 Sep 2005 08:32:25 -0700
> Fix bug in bnx2_interrupt() that caused an unnecessary register read.
> The BNX2_PCICFG_MISC_STATUS should only be read when the status tag
> has not changed.
>
> Add prefetch of the status block in bnx2_msi() simila
From: Dave Hansen <[EMAIL PROTECTED]>
Date: Thu, 08 Sep 2005 07:02:17 -0700
> KERNEL: assertion ((int)tp->lost_out >= 0) failed at net/ipv4/tcp_input.c
> (2148)
So tp->lost_out went negative, didn't we discuss to death last week
how this was impossible? :-)
-
To unsubscribe from this list: send
From: Julian Anastasov <[EMAIL PROTECTED]>
Date: Thu, 8 Sep 2005 10:01:39 +0300 (EEST)
> One such place that can damage the dst refcnts is route.c
> with CONFIG_IP_ROUTE_MULTIPATH_CACHED enabled, i don't see the user's
> .config. In this new code i see that rt_intern_hash is called before
>
From: Ralf Baechle DL5RB <[EMAIL PROTECTED]>
Date: Wed, 7 Sep 2005 15:34:01 +0100
> Asc2ax was still using a static buffer for all invocations which isn't
> exactly SMP-safe. Change asc2ax to take an additional result buffer as
> the argument. Change all callers to provide such a buffer.
>
> Th
From: Patrick McHardy <[EMAIL PROTECTED]>
Date: Thu, 08 Sep 2005 23:00:39 +0200
> When an error is returned in some cases xfrm_lookup releases
> the dst_entry and sets the pointer to NULL, in some it doesn't.
> There are multiple dst leaks in users of xfrm_lookup because
> not all take care to rel
From: Herbert Xu <[EMAIL PROTECTED]>
Date: Fri, 9 Sep 2005 07:46:24 +1000
> On Thu, Sep 08, 2005 at 01:31:26PM -0700, David S. Miller wrote:
> > From: Dave Hansen <[EMAIL PROTECTED]>
> > Date: Thu, 08 Sep 2005 07:02:17 -0700
> >
> > > KERNEL: assertion (
From: Adhiraj Joshi <[EMAIL PROTECTED]>
Date: Fri, 9 Sep 2005 11:35:27 +0530
> Hi,
>
> There was a bug some time back in a Must-Fix list regarding UDP
> applications going in dead lock.
> (http://kernel.org/pub/linux/kernel/people/akpm/must-fix/must-fix-2.txt)
> Here is an extract:
>
From: Stephen Hemminger <[EMAIL PROTECTED]>
Date: Fri, 9 Sep 2005 12:53:19 -0700
> Create one iterator for walking over FIB trie, and use it
> for all the /proc functions. Add a /proc/net/route
> output for backwards compatibility with old applications.
>
> Make initialization of fib_trie same as
From: John Heffner <[EMAIL PROTECTED]>
Date: Fri, 28 Oct 2005 20:39:24 -0400
> On Oct 28, 2005, at 6:10 PM, Stephen Hemminger wrote:
>
> > The following update cwnd for every ACK, but it ends up making Linux
> > more aggressive
> > (but still in compliance with existing RFC's).
>
> Since Linux
From: Krzysztof Halasa <[EMAIL PROTECTED]>
Date: Mon, 07 Nov 2005 17:37:40 +0100
> David?
I'm still 20,000+ emails deep in backlog after my 2 week
vaction, so I won't be able to look into this for at least
a few more days, at best.
Also, please use "[EMAIL PROTECTED]" for emails to me.
-
To unsu
From: Tom Young <[EMAIL PROTECTED]>
Date: Mon, 07 Nov 2005 15:59:49 +1100
> The socket ip id currently gets incremented by 1 + the number of
> segments leading to an increment of 2 in the standard non-TSO case.
> This patch fixes 'more' to be a count of extra segments.
>
> Signed-off-by: Thomas Y
From: Stephen Hemminger <[EMAIL PROTECTED]>
Date: Mon, 7 Nov 2005 11:22:15 -0800
> I am looking into creating a function tcp_is_application_limited()
> to replace the old in flight check. It is kind of the logical inverse
> of the nagle check code on output.
That's a great idea, I was going to s
From: Stephen Hemminger <[EMAIL PROTECTED]>
Date: Mon, 7 Nov 2005 14:14:20 -0800
> The core of what I am now testing is:
Looks good but I'm starting to hate that divide, and now we'll
have to eat it for every ACK too :-/
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the b
From: Tom Young <[EMAIL PROTECTED]>
Date: Tue, 08 Nov 2005 10:33:50 +1100
> I'm not convinced that is true, I don't have a TSO supported card and I
> get tso_segs to be 1 inside the ip_queue_xmit function.
> Looking through tcp_output.c, all references I could find expected
> tso_segs to be a coun
From: Evgeniy Polyakov <[EMAIL PROTECTED]>
Date: Tue, 1 Nov 2005 22:28:30 +0300
> Patch from Carlo Comin
>
> Signed-off-by: Evgeniy Polyakov <[EMAIL PROTECTED]>
Applied, thanks.
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
Mor
From: Thomas Graf <[EMAIL PROTECTED]>
Date: Wed, 2 Nov 2005 00:41:10 +0100
> Fixes an invalid memory reference when the basic classifier
> is used without any ematches but just actions.
>
> Signed-off-by: Thomas Graf <[EMAIL PROTECTED]>
Applied, thanks Thomas.
-
To unsubscribe from this list: se
From: Stephen Hemminger <[EMAIL PROTECTED]>
Date: Thu, 3 Nov 2005 15:44:35 -0800
> There is no reason for sk_add_backlog to be a macro. It can
> just be an inline function and get type checking.
>
> Signed-off-by: Stephen Hemminger <[EMAIL PROTECTED]>
Applied, thanks Stephen.
I actually remembe
From: David Stevens <[EMAIL PROTECTED]>
Date: Thu, 3 Nov 2005 15:58:16 -0700
> The following patch fixes some locking issues with multicast source
> filters:
> 1) Acquire a read lock for the socket mc list in ip6_mc_msfilter().
> 2) Add a read/write lock for the source list for individual socket
>
From: Julian Anastasov <[EMAIL PROTECTED]>
Date: Sun, 6 Nov 2005 13:11:49 +0200 (EET)
> There was a fix in 2.6.13 that changed the behaviour of
> ip_vs_conn_expire_now function not to put reference to connection,
> its callers should hold write lock or connection refcnt. But we
> forgot to c
From: Herbert Xu <[EMAIL PROTECTED]>
Date: Tue, 08 Nov 2005 12:21:38 +1100
> [IPV4] Fix ip_queue_xmit identity increment for TSO packets
>
> When ip_queue_xmit calls ip_select_ident_more for IP identity selection
> it gives it the wrong packet count for TSO packets. The ip_select_*
> functions e
From: Stephen Hemminger <[EMAIL PROTECTED]>
Date: Mon, 7 Nov 2005 16:25:32 -0800
> it could be limited to power of 2 and a shift?
That might not be a bad idea.
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at
From: Anton Blanchard <[EMAIL PROTECTED]>
Date: Tue, 8 Nov 2005 17:56:47 +1100
> Any thoughts on this? The other option was to override dev_alloc_skb,
> but I would prefer not to if possible.
There is still a lot of code that depends upon there being
a scratch of ~15 bytes there at the beginning
From: Pavel Roskin <[EMAIL PROTECTED]>
Date: Tue, 08 Nov 2005 01:40:57 -0500
> The protocol field in ethernet headers is big-endian and should be
> annotated as such. This patch allows detection of missing ntohs() calls
> on the ethernet protocol field when sparse is run with __CHECK_ENDIAN__
> d
From: Anton Blanchard <[EMAIL PROTECTED]>
Date: Wed, 9 Nov 2005 03:12:20 +1100
> Do you mean we expect exactly 16 bytes, or at least 16 bytes. Im happy
> to ammend it so we always require the minimum 16 bytes, and can expand
> it on architectures where it makes sense.
I meant "at least 16 bytes".
From: Rick Jones <[EMAIL PROTECTED]>
Date: Tue, 08 Nov 2005 11:11:17 -0800
> What are the two lines xoring h with shifts of itself meant to be doing?
So that all the upper bits get mixed into the hash somehow, regardless
of the size of the hash table.
-
To unsubscribe from this list: send the lin
From: Rick Jones <[EMAIL PROTECTED]>
Date: Tue, 08 Nov 2005 11:33:16 -0800
> Is there a test case I may have missed - otherwise it looks like it
> was adding much randomness - particularly in the case of the IP's
> being "close" to one another - the xor tends to "knock-out" the
> upper bits of the
From: Rick Jones <[EMAIL PROTECTED]>
Date: Tue, 08 Nov 2005 12:12:19 -0800
> Still seems to be doing as well - I'm assuming that when the
> "random" IPs are generated that they do not have to be put into
> network byte order, since they are randomly generated bits. When
> the IPs aren't random, t
From: Rick Jones <[EMAIL PROTECTED]>
Date: Tue, 08 Nov 2005 13:01:52 -0800
> Endianness often makes my head spin, so please be kind - are the
> addresses being converted to little endian before the hash (ntohl as
> it were) or is it just happening as a consequence? If the former,
> why the conver
From: Harald Welte <[EMAIL PROTECTED]>
Date: Wed, 9 Nov 2005 01:24:19 +0100
> (seems like i can't even count, thus 16/15)
hehe... all 16 patches applied, and yes I did use the
updated version of patch 8/15 :-)
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a me
From: Ken-ichirou MATSUZAWA <[EMAIL PROTECTED]>
Date: Wed, 09 Nov 2005 22:58:42 +0900 (JST)
> ip6ip6_lock is not unlocked in error path.
Patch applied, thanks.
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at
From: Peter Chubb <[EMAIL PROTECTED]>
Date: Wed, 9 Nov 2005 11:16:58 +1100
> Trying to build today's 2.6.14+git snapshot gives undefined references
> to use_tempaddr
>
> Looks like an ifdef got left out.
>
> Signed-off-by: Peter Chubb <[EMAIL PROTECTED]>
Applied, thanks Peter.
-
To unsubscribe
From: Ben Greear <[EMAIL PROTECTED]>
Date: Wed, 09 Nov 2005 13:41:02 -0800
> So, I am asking the TSO folks: Will you pay any attention to
> a bug report against 2.6.13.2, or would I just be wasting my
> time?
>
> If the answer is yes, then I'll get on it..otherwise, I'll wait untill
> I move to
From: Ben Greear <[EMAIL PROTECTED]>
Date: Wed, 09 Nov 2005 15:39:38 -0800
> I will give it a try..as soon as the next stable point release comes
> out..seems there are some networking bug fixes coming soon if I recall
> correctly.
Just a UDP zero-length transfer bug fix from Herbert, nothing
TCP
From: Stephen Hemminger <[EMAIL PROTECTED]>
Date: Wed, 09 Nov 2005 15:06:57 -0800
> TCP peformance with TSO over networks with delay is awful.
> On a 100Mbit link with 150ms delay, we get 4Mbits/sec with TSO and
> 50Mbits/sec without TSO.
>
> The problem is with TSO, we intentionally do not keep
From: Jeff Garzik <[EMAIL PROTECTED]>
Date: Thu, 10 Nov 2005 02:42:19 -0500
> Seems OK to me (though not pulled, since you didn't request such; also,
> I dunno if DaveM has started being bnx2 merge master again)
I can take care of these bnx2 changes tomorrow.
-
To unsubscribe from this list: sen
From: Andrew Morton <[EMAIL PROTECTED]>
Date: Sat, 28 Jan 2006 17:22:40 -0800
> David, your tree
> git+ssh://master.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6.17.git
> is oopsing all over the place. Reverting "[NET]: Do not lose accepted
> socket when -ENFILE/-EMFILE." makes it stop:
Ok,
ILE." makes it stop:
I just checked in the following fix, thanks for the report.
diff-tree e6b303b1b4b890772f9c45f790deb1cfb49e295c (from
dc326c4936f41911046b2dc72cbe04053e9680d6)
Author: David S. Miller <[EMAIL PROTECTED]>
Date: Sun Jan 29 21:08:25 2006 -0800
[NET]
From: "Wei Yongjun" <[EMAIL PROTECTED]>
Date: Mon, 30 Jan 2006 16:29:18 -0800
> I think this problem should not occurred because the patch make the
> ipv4 header of a 32 bit alignment.
I still think the complicated new code is not justified.
Show a real life example where this results in breakag
From: "Wei Yongjun" <[EMAIL PROTECTED]>
Date: Fri, 27 Jan 2006 09:35:02 -0800
> [1]Summary of the problem:
> ip_options_fragment() has no effect on fragmentation
>
> [2]Full description of the problem:
> When I send IPv4 packet(contain Record Route Option) which need to be
> fragmented to the rou
From: Harald Welte <[EMAIL PROTECTED]>
Date: Tue, 31 Jan 2006 00:22:09 +0100
> [NETFILTER] NAT sequence adjustment: Save eight bytes per conntrack
>
> This patch reduces the size of 'struct ip_conntrack' on systems with NAT
> by eight bytes. The sequence number delta values can be int16_t, since
From: Harald Welte <[EMAIL PROTECTED]>
Date: Tue, 31 Jan 2006 00:23:48 +0100
> [NETFILTER] nfnetlink_log: add sequence numbers for log events
>
> By using a sequence number for every logged netfilter event, we can
> determine from userspace whether logging information was lots somewhere
> downstr
From: Chase Douglas <[EMAIL PROTECTED]>
Date: Mon, 30 Jan 2006 13:55:21 -0500
> I have a question about the implementation of sendfile. In my current
> configuration of a server, when requests are made for a file, a new
> thread is spawned and sendfile is called to complete the request (We
> re
From: Sridhar Samudrala <[EMAIL PROTECTED]>
Date: Mon, 30 Jan 2006 11:53:58 -0800
> The number of HEARTBEAT chunks that an association may transmit is
> limited by Association.Max.Retrans count; however, the code allows
> us to send one extra heartbeat.
>
> This patch limits the number of heartbe
From: Sridhar Samudrala <[EMAIL PROTECTED]>
Date: Mon, 30 Jan 2006 11:52:03 -0800
> We currently count the initial INIT/COOKIE_ECHO chunk toward the
> retransmit count and thus sends a total of sctp_max_retrans_init chunks.
> The correct behavior is to retransmit the chunk sctp_max_retrans_init in
From: John Heffner <[EMAIL PROTECTED]>
Date: Tue, 06 Dec 2005 14:42:53 -0500
> I'd like to get a few people at least to look this over, and maybe give
> it a try. One remaining item to consider is how best to cache the state
> between connections. Are there any major concerns or reservations a
From: dean gaudet <[EMAIL PROTECTED]>
Date: Sat, 17 Dec 2005 18:22:40 -0800 (PST)
> let me know what you think... i'd like to get something like this patch
> included upstream so i can eliminate a patch from several of my kernels.
The RTA length check is a little hackish. Maybe use a new
attribu
From: Dave Jones <[EMAIL PROTECTED]>
Date: Mon, 30 Jan 2006 23:22:29 -0500
> This is easily triggerable by sending bogus packets,
> allowing a malicious user to flood remote logs.
>
> Signed-off-by: Dave Jones <[EMAIL PROTECTED]>
Applied, thanks Dave.
I'll push this to -stable once Linus eats i
From: Baruch Even <[EMAIL PROTECTED]>
Date: Thu, 26 Jan 2006 17:17:56 +0200
> This fixes the accounting in H-TCP, the ccount variable is also adjusted
> a few lines above this one.
>
> This line was not supposed to be there and wasn't there in the patches
> originally submitted, the four patches
701 - 800 of 1576 matches
Mail list logo