| Eric W. Biederman | Re: [PATCH] ARCH: Fix 32-bit x86 MSI-X allocation leakage
"Brandeburg, Jesse" <jesse.brandeburg@intel.com>, <linux-pci@atrey.karlin.mff.cuni.cz>, "Waskiewicz Jr, Peter P" <peter.p.waskiewicz.jr@intel.com>, <michael@ellerman.id.au>, Rusty Russell <rusty@rustcorp.com.au>, Andi Kleen <ak@suse.de>, Ingo Molnar <mingo@elte.hu>, Thomas Gleixner <tglx@linutronix.de>
Date: Mon, 28 Apr 2008 12:09:57 -0700
In-Reply-To: <20080426005850.7098.77479.stgit@scrappy.jf.intel.com> (PJ
Waskiewicz's message of "Fri, 25 Apr 2...
| Apr 28, 3:09 pm 2008 |
| Waskiewicz Jr, Peter P | RE: [PATCH] ARCH: Fix 32-bit x86 MSI-X allocation leakage
> Thanks Jesse for forwarding this to me.
Yes, Jesse took care of that oversight for me. I owe him a tasty
Either way, I'm fine if that gets reverted or not, just as long as we
don't revert back to leaking vectors on device unload. But if things
are somewhat functional today, I'd say leave the code in there until the
proper solution can be worked out. Then go ahead and pull it out and
put the new bits in, so anyone currently using the functionality
provided isn't left out in the cold. That...
| Apr 28, 5:27 pm 2008 |
| Paulius Zaleckas |
| Apr 28, 7:45 pm 2008 |
| Steve Glendinning | [PATCH 2/2] Add support for SMSC LAN8187 and LAN8700 PHYs
Add support for two additional SMSC PHY models with identical interrupt
source and mask registers to the LAN83C185
Signed-off-by: Steve Glendinning <steve.glendinning@smsc.com>
---
drivers/net/phy/Kconfig | 2 +-
drivers/net/phy/smsc.c | 69 ++++++++++++++++++++++++++++++++++++++++++++++-
2 files changed, 69 insertions(+), 2 deletions(-)
diff --git a/drivers/net/phy/Kconfig b/drivers/net/phy/Kconfig
index 3ac8529..6bf9e76 100644
--- a/drivers/net/phy/Kconfig
+++ b/drivers/net/phy/...
| Apr 28, 1:37 pm 2008 |
| Steve Glendinning | [PATCH 1/2] Rename SMSC phy functions to be more generic
Several models of SMSC PHY have the same interrupt status and mask
registers as the LAN83C185, so these functions can service multiple
different PHY drivers.
Signed-off-by: Steve Glendinning <steve.glendinning@smsc.com>
---
drivers/net/phy/smsc.c | 14 +++++++-------
1 files changed, 7 insertions(+), 7 deletions(-)
diff --git a/drivers/net/phy/smsc.c b/drivers/net/phy/smsc.c
index b1d8ed4..6a901f0 100644
--- a/drivers/net/phy/smsc.c
+++ b/drivers/net/phy/smsc.c
@@ -38,7 +38,7 @@
(...
| Apr 28, 1:36 pm 2008 |
| John Heffner | Re: Socket buffer sizes with autotuning
I had not actually known about this. (Just cooked up the patches
after I saw your mail.) This looks to me like a clear bug. After
looking carefully at the code there, it seems there was another small
problem, too, though not likely to have much effect.
Dave, can you apply the patches I just sent?
Thanks,
-John
--
| Apr 28, 7:28 pm 2008 |
| David Miller |
| Apr 28, 7:35 pm 2008 |
| Paulius Zaleckas |
| Apr 28, 7:27 pm 2008 |
| Allan Stephens | [PATCH 0/9 net-next-2.6] [TIPC]: System & debug output m...
This patch set includes a number of fixes and cleanups for the code TIPC
uses to print its system and debug output. Please see each individual patch
for further description.
Summary:
net/tipc/config.c | 2
net/tipc/core.c | 6 +
net/tipc/core.h | 124 ++++++++++++++++----------
net/tipc/dbg.c | 231 ++++++++++++++++++++++++-------------------------
net/tipc/dbg.h | 12 +--
net/tipc/link.c | 20 +++-
net/tipc/msg.c | 10 +-
net/tipc/...
| Apr 28, 3:48 pm 2008 |
| Allan Stephens | [PATCH 7/9 net-next-2.6] [TIPC]: Cosmetic cleanup of system ...
This patch contains changes to make TIPC's system & debug
message declarations more readable. Declarations have been
regrouped and recommented to make it easier to understand
what output is generated in both standard and debugging modes.
In addition, oversize lines have been fixed to respect the
80 character upper bound used in the kernel.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
---
net/tipc/core.h | 117 +++++++++++++++++++++++++++++++-----------------------
1 f...
| Apr 28, 3:48 pm 2008 |
| Allan Stephens | [PATCH 8/9 net-next-2.6] [TIPC]: Exclude debug-only print bu...
This patch modifies TIPC to only exclude debug-related print buffer
routines when debugging capabilities are not required. It also
fixes up some related #defines that exceed 80 characters.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
---
net/tipc/core.h | 12 ++++++++----
net/tipc/dbg.c | 8 ++++++--
net/tipc/link.c | 20 ++++++++++++++++----
net/tipc/msg.c | 10 +++++++---
4 files changed, 37 insertions(+), 13 deletions(-)
diff --git a/net/tipc/core.h b/net/...
| Apr 28, 3:48 pm 2008 |
| Allan Stephens | [PATCH 6/9 net-next-2.6] [TIPC]: Cosmetic cleanup of print b...
This patch contains changes to make TIPC's print buffer code
conform more closely to Linux kernel coding guidelines.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
---
net/tipc/core.h | 6 ++++--
net/tipc/dbg.c | 14 +++++++-------
2 files changed, 11 insertions(+), 9 deletions(-)
diff --git a/net/tipc/core.h b/net/tipc/core.h
index 7042ef3..bc63355 100644
--- a/net/tipc/core.h
+++ b/net/tipc/core.h
@@ -2,7 +2,7 @@
* net/tipc/core.h: Include file for TIPC global ...
| Apr 28, 3:48 pm 2008 |
| Allan Stephens | [PATCH 9/9 net-next-2.6] [TIPC]: Update version number to TI...
This patch updates TIPC's version number to 1.6.4.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
---
net/tipc/core.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/net/tipc/core.c b/net/tipc/core.c
index 862d415..6d6aa5a 100644
--- a/net/tipc/core.c
+++ b/net/tipc/core.c
@@ -49,7 +49,7 @@
#include "config.h"
-#define TIPC_MOD_VER "1.6.3"
+#define TIPC_MOD_VER "1.6.4"
#ifndef CONFIG_TIPC_ZONES
#define CONFIG_TIPC_ZONES 3
--
1.5....
| Apr 28, 3:48 pm 2008 |
| Allan Stephens | [PATCH 1/9 net-next-2.6] [TIPC]: Simplify log buffer resizing
This patch simplifies & standardizes the way TIPC's print buffer
log is resized. Code to terminate use of the log buffer is
eliminated by simply setting the log buffer size to 0 bytes.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
---
net/tipc/config.c | 2 +-
net/tipc/core.c | 4 ++--
net/tipc/dbg.c | 26 +++++++-------------------
net/tipc/dbg.h | 6 +++---
4 files changed, 13 insertions(+), 25 deletions(-)
diff --git a/net/tipc/config.c b/net/ti...
| Apr 28, 3:48 pm 2008 |
| Allan Stephens | [PATCH 5/9 net-next-2.6] [TIPC]: Elimination of print buffer...
This patch revamps TIPC's print buffer subsystem to eliminate
support for arbitrary chains of print buffers, which were
rarely needed and difficult to use safely.
In its place, print buffers can now be configured to echo their
output to the system console. This provides an equivalent for
the only chaining currently utilized by TIPC, in a faster and
more compact manner.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
---
net/tipc/core.h | 7 ++-
net/tipc/dbg.c | 163 ++...
| Apr 28, 3:48 pm 2008 |
| Allan Stephens | [PATCH 3/9 net-next-2.6] [TIPC]: Fix recursive spinlock invo...
This patch fixes two routines that allow the global print buffer
spinlock to be taken recursively.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
---
net/tipc/dbg.c | 17 ++++++++++++-----
1 files changed, 12 insertions(+), 5 deletions(-)
diff --git a/net/tipc/dbg.c b/net/tipc/dbg.c
index c680691..834319e 100644
--- a/net/tipc/dbg.c
+++ b/net/tipc/dbg.c
@@ -179,8 +179,10 @@ void tipc_printbuf_move(struct print_buf *pb_to, struct print_buf *pb_from)
}
if (pb_to-&...
| Apr 28, 3:48 pm 2008 |
| Allan Stephens | [PATCH 2/9 net-next-2.6] [TIPC]: Provide feedback when log b...
This patch provides feedback to the user when TIPC is unable
to set its log buffer to the requested size.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
---
net/tipc/dbg.c | 13 ++++++++++---
net/tipc/dbg.h | 4 ++--
2 files changed, 12 insertions(+), 5 deletions(-)
diff --git a/net/tipc/dbg.c b/net/tipc/dbg.c
index 1a8d0a2..c680691 100644
--- a/net/tipc/dbg.c
+++ b/net/tipc/dbg.c
@@ -2,7 +2,7 @@
* net/tipc/dbg.c: TIPC print buffer routines for debugging
*
*...
| Apr 28, 3:48 pm 2008 |
| Allan Stephens | [PATCH 4/9 net-next-2.6] [TIPC]: Fix null pointer dereferenc...
This patch eliminates an obsolete use of the DBG_OUTPUT print
buffer which could lead to a null pointer crash in tipc_printf()
if TIPC's debugging capabilities are configured.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
---
net/tipc/name_distr.c | 3 ---
1 files changed, 0 insertions(+), 3 deletions(-)
diff --git a/net/tipc/name_distr.c b/net/tipc/name_distr.c
index 39fd161..aecba5c 100644
--- a/net/tipc/name_distr.c
+++ b/net/tipc/name_distr.c
@@ -41,9 +41,6 @@
#...
| Apr 28, 3:48 pm 2008 |
| PJ Waskiewicz | [PATCH] ARCH 2.6.25.y: Fix 32-bit x86 MSI-X allocation leakage
Resend now that this patch is in -upstream.
This bug was introduced in the 2.6.24 i386/x86_64 tree merge, where
MSI-X vector allocation will eventually fail. The cause is the new
bit array tracking used vectors is not getting cleared properly on
IRQ destruction on the 32-bit APIC code.
This can be seen easily using the ixgbe 10 GbE driver on multi-core
systems by simply loading and unloading the driver a few times.
Depending on the number of available vectors on the host system, the
MSI-X alloc...
| Apr 28, 2:56 pm 2008 |
| PJ Waskiewicz | [PATCH] ARCH 2.6.24.y: Fix 32-bit x86 MSI-X allocation leakage
Resend now that this patch was merged into -upstream.
This bug was introduced in the 2.6.24 i386/x86_64 tree merge, where
MSI-X vector allocation will eventually fail. The cause is the new
bit array tracking used vectors is not getting cleared properly on
IRQ destruction on the 32-bit APIC code.
This can be seen easily using the ixgbe 10 GbE driver on multi-core
systems by simply loading and unloading the driver a few times.
Depending on the number of available vectors on the host system, the
M...
| Apr 28, 2:56 pm 2008 |
| Patrick McHardy | [RFC XFRM]: esp: fix scatterlist of out bounds access with c...
I ran into occasional BUGs in scatterlist.h, which turned
out the be caused by accessing an uninitialized scatterlist
entry from eseqiv. I'm not sure whether this patch is correct
since I'm seeing invalid packets with and without this patch
(probably related to HIFN though) and I don't understand why
scatterwalk_sg_next() returns either a scatterlist or a
struct page dependant on the length, but at least it fixes
the BUG() for me :)
| Apr 28, 2:55 pm 2008 |
| Arnaud Ebalard | [RESEND][NETFILTER][PATCH] nfqnl_mangle() not requesting eno...
Hi,
This is a resend, first post was to netdev only and slipped through.
While reinjecting *bigger* modified versions of IPv6 packets using
libnetfilter_queue, things work fine on a 2.6.24 kernel (2.6.22 too)
but I get the following on recents kernels (2.6.25, trace below is
against today's net-2.6 git tree):
skb_over_panic: text:c04fddb0 len:696 put:632 head:f7592c00 data:f7592c00 tail:0xf7592eb8 end:0xf7592e80 dev:eth0
------------[ cut here ]------------
invalid opcode: 0000 [#1] PREEMPT
...
| Apr 28, 1:49 pm 2008 |
| Patrick McHardy |
| Apr 28, 5:56 pm 2008 |
| Arnaud Ebalard | Re: [RESEND][NETFILTER][PATCH] nfqnl_mangle() not requesting...
Hi,
Ack.
BTW, while looking at the code, I noticed the call to skb_copy_expand()
leads to a complete copy of current sk_buff data to the new sk_buff.
Those data are then completely rewritten a few lines later. Too bad
there is no simple way (considering skbuff.h) to prevent the useless
copy.
Cheers,
a+
--
| Apr 28, 7:07 pm 2008 |
| Randy.Dunlap | [PATCH] phy: fix kernel-doc notation
From: Randy Dunlap <randy.dunlap@oracle.com>
Fix kernel-doc warning:
Warning(linux-2.6.25-git11//drivers/net/phy/phy_device.c:275): No description found for parameter 'bus_id'
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
---
drivers/net/phy/phy_device.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
--- linux-2.6.25-git11.orig/drivers/net/phy/phy_device.c
+++ linux-2.6.25-git11/drivers/net/phy/phy_device.c
@@ -256,7 +256,7 @@ void phy_prepare_link(struct phy_de...
| Apr 28, 1:58 pm 2008 |
| Andy Fleming |
| Apr 28, 2:35 pm 2008 |
| Thomas Gleixner |
| Apr 28, 4:30 pm 2008 |
| Waskiewicz Jr, Peter P | RE: [PATCH] ARCH: Fix 32-bit x86 MSI-X allocation leakage
Eric replied with the actual commit during the 2.6.24 merge window that
introduced this bug. The io_apic.c code from the i386 tree did not stay
completely static when it was merged into the x86 io_apic_32.c code.
Here is the commit that Eric identified that introduced the defect:
In particular commit dbeb2be21d678c49a8d8bbf774903df15dd55474
Author: Rusty Russell <rusty@rustcorp.com.au>
Date: Fri Oct 19 20:35:03 2007 +0200
i386: introduce "used_vectors" bitmap which can be used to r...
| Apr 28, 4:42 pm 2008 |
| Thomas Gleixner | RE: [PATCH] ARCH: Fix 32-bit x86 MSI-X allocation leakage
Ok, that makes sense. I just did not get the context out of your
commit log. Adding the commit which caused the trouble to the commit
log makes it easier to understand. So the problem is unrelated to the
merger, it's caused by a patch which was pending against the i386 tree
around the time of the merger and which was converted to the merge
tree.
Thanks,
tglx
--
| Apr 28, 5:16 pm 2008 |
| Waskiewicz Jr, Peter P | RE: [PATCH] ARCH: Fix 32-bit x86 MSI-X allocation leakage
That's my miss for not including the specific context. I will try to be
better about that in the future. I do appreciate the interest though to
make sure it's specifically understood when this bug was introduced.
Cheers,
-PJ Waskiewicz
--
| Apr 28, 5:21 pm 2008 |
| Greg KH | Re: [stable] [PATCH] ARCH 2.6.25.y: Fix 32-bit x86 MSI-X all...
When sending patches to -stable, please add "CC: stable
<stable@kernel.org> to the patch itself in the signed-off-by area, that
way we get notified automatically when it goes into Linus's tree so that
we can then know to apply it to the tree. Otherwise it's a tough job to
try to track all of these knowing when to watch out for them to be
committed.
Can you please resend them to stable@kernel.org when they do go into
Linus's tree?
thanks,
greg k-h
--
| Apr 28, 1:38 pm 2008 |
| Waskiewicz Jr, Peter P |
| Apr 28, 1:39 pm 2008 |
| Andy Gospodarek | [PATCH] r8169: fix panic when using unlisted MAC
When using the r8169 driver with an unlisted MAC, the system will
panic in rtl8169_get_mac_version since it will try to access tp->pci_dev
when it's unset. Moving the call to set pci_dev earlier in
rtnl8169_init_one will resolve this and should not introduce any other
problems.
Signed-off-by: Andy Gospodarek <andy@greyhouse.net>
---
drivers/net/r8169.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/net/r8169.c b/drivers/net/r8169.c
index 3acfeea...
| Apr 28, 1:22 pm 2008 |
| Paulius Zaleckas |
| Apr 28, 11:57 am 2008 |
| Guennadi Liakhovetski |
| Apr 28, 11:40 am 2008 |
| Adrian Bunk | [2.6 patch] net/ipv4/ip_output.c:ip_send_check() mustn't be ...
This patch fixes the following build error with UML and gcc 4.3:
<-- snip -->
...
CC net/ipv4/ip_output.o
/home/bunk/linux/kernel-2.6/git/linux-2.6/net/ipv4/ip_output.c: In function ‘__ip_local_out’:
/home/bunk/linux/kernel-2.6/git/linux-2.6/net/ipv4/ip_output.c:89: sorry, unimplemented: inlining failed in call to ‘ip_send_check’: function body not available
/home/bunk/linux/kernel-2.6/git/linux-2.6/net/ipv4/ip_output.c:99: sorry, unimplemented: called from here
make[3]: **...
| Apr 28, 11:38 am 2008 |
| Rusty Russell | [PATCH 1/4] virtio: wean net driver off NETDEV_TX_BUSY
Herbert tells me that returning NETDEV_TX_BUSY from hard_start_xmit is
seen as a poor thing to do; we should cache the packet and stop the queue.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
---
drivers/net/virtio_net.c | 58 ++++++++++++++++++++++++++++++++---------------
1 file changed, 40 insertions(+), 18 deletions(-)
diff -r 26ba2dd67f46 drivers/net/virtio_net.c
--- a/drivers/net/virtio_net.c Tue Apr 22 13:03:07 2008 +1000
+++ b/drivers/net/virtio_net.c Thu Apr 24 12:05:24 ...
| Apr 28, 11:16 am 2008 |
| Rusty Russell | [PATCH 2/4] virtio: finer-grained features for virtio_net
So, we previously had a 'VIRTIO_NET_F_GSO' bit which meant that 'the
host can handle csum offload, and any TSO (v4&v6 incl ECN) or UFO
packets you might want to send. I thought this was good enough for
Linux, but it actually isn't, since we don't do UFO in software.
So, add separate feature bits for what the host can handle. Add
equivalent ones for the guest to say what it can handle, because LRO
is coming too (thanks Herbert!).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
--...
| Apr 28, 11:17 am 2008 |
| Rusty Russell | [PATCH 3/4] virtio: net: Add ethtool ops for SG/GSO
Date: Fri, 18 Apr 2008 11:21:42 +0800
From: Herbert Xu <herbert@gondor.apana.org.au>
This patch adds some basic ethtool operations to virtio_net so
I could test SG without GSO (which was really useful because TSO
turned out to be buggy :)
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> (remove MTU setting)
---
drivers/net/virtio_net.c | 18 ++++++++++++++++++
1 file changed, 18 insertions(+)
diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
--- a/drivers/ne...
| Apr 28, 11:19 am 2008 |
| Rusty Russell | [PATCH 4/4] virtio: net: Allow receiving SG packets
Date: Fri, 18 Apr 2008 11:24:27 +0800
From: Herbert Xu <herbert@gondor.apana.org.au>
Finally this patch lets virtio_net receive GSO packets in addition
to sending them. This can definitely be optimised for the non-GSO
case. For comparison the Xen approach stores one page in each skb
and uses subsequent skb's pages to construct an SG skb instead of
preallocating the maximum amount of pages per skb.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> (added feature bits)
---
drive...
| Apr 28, 11:21 am 2008 |
| Ben Dooks |
| Apr 28, 11:13 am 2008 |
| Jeff Garzik | Re: DM9000 patches
these two are not. It sounded like your comments may have impacted your
ack, and also, when there are multiple patches in the air from multiple
people, it's generally up to the maintainer to sort them all out and
send them in a batch.
Lastly, a CC to me would be quite helpful. I tend to scan subject lines
on netdev rather than read every message, and sometimes patches can get
missed.
Jeff
--
| Apr 28, 11:21 am 2008 |
| Gerrit Renker | Re: [PATCH 1/1] [DCCP][QPOLICY]: External interface changes
| Separate internal (in-kernel) storage of per packet data from external
| interface (which is now an easily extensible structure).
|
I can see your point, there is an alternative to realise what you are
aiming to do, suggestions/comments are below.
| --- a/include/linux/dccp.h
| +++ b/include/linux/dccp.h
| @@ -98,6 +98,15 @@ struct dccp_hdr_reset {
| dccph_reset_data[3];
| };
|
| +/**
| + * struct dccp_qpolicy_prio_info - information used by prio queuing policy
| + *
| + * @pri...
| Apr 28, 11:08 am 2008 |
| Tomasz Grobelny |
| Apr 28, 5:29 pm 2008 |
| Kostya B | [PATCH][IPv4] UFO: prevent generation of chained skb
[IPv4] UFO: prevent generation of chained skb destined to UFO device
Problem:
ip_append_data() could wrongly generate a chained skb for devices which support UFO.
When sk_write_queue is not empty (e.g. MSG_MORE), __instead__ of appending data
into the next nr_frag of the queued skb, a new chained skb is created.
I would normally assume UFO device should get data in nr_frags and not in frag_list.
Later the udp4_hwcsum_outgoing() resets csum to NONE and skb_gso_segment() has oops.
Proposal:
1...
| Apr 28, 8:13 am 2008 |
| Joakim Tjernlund | TCP slowdown in 2.6.25?
After upgrading from 2.6.23 to 2.6.25 I am seeing a big slow down on
our old www server. Stracing shows lots of EAGAIN but i not sure
these are relevant. I got no idea where to start looking so
I would appreciate some pointers. Perhaps this is a know issue?
CPU is a PowerPC, mpc8321, network driver ucc_geth
Jocke
--
| Apr 28, 8:12 am 2008 |
| Joakim Tjernlund | Re: TCP slowdown in 2.6.25?
Some info:
Even ftp transfers takes forever, but only in one direction: from the
node with kernel 2.6.25 to some other host. Strace show me that
write(2) hangs for long periods before succeeding to transmit
the data.
--
| Apr 28, 9:33 am 2008 |
| Ilpo Järvinen | Re: TCP slowdown in 2.6.25?
Since it's easy, can you quickly try to if setting tcp_sack sysctl to 0
makes any difference (I made a surgery there, perhaps some bug in there
could cause TCP to recover by RTO multiple times in a row, that should
be visible in tcpdump as well)?
Do you have network namespaces enabled in .config?
Netfilter/iptables involved?
To some specific or "any" other host? How long is "long"?
--
i.
--
| Apr 28, 10:42 am 2008 |
| Joakim Tjernlund | Re: TCP slowdown in 2.6.25?
echo 0 > /proc/sys/net/ipv4/tcp_sack
No,
Any host. A 1304632 bytes file takes about 110 seconds to transfer
(12 Kbytes/sec)
Thanks for helping.
A quick capture with wireshark indicates that some sent packages from
the node never reaches the host on the network.
--
| Apr 28, 11:19 am 2008 |
| Ilpo Järvinen |
| Apr 28, 11:48 am 2008 |
