* [PATCH] ipv4: Don't increase PMTU with Datagram Too Big message.
From: Li Wei @ 2015-01-29 8:09 UTC (permalink / raw)
To: davem; +Cc: netdev, Li Wei
RFC 1191 said, "a host MUST not increase its estimate of the Path
MTU in response to the contents of a Datagram Too Big message."
Signed-off-by: Li Wei <lw@cn.fujitsu.com>
---
net/ipv4/route.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index d58dd0e..52e1f2b 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -966,6 +966,9 @@ static void __ip_rt_update_pmtu(struct rtable *rt, struct flowi4 *fl4, u32 mtu)
if (dst->dev->mtu < mtu)
return;
+ if (rt->rt_pmtu && rt->rt_pmtu < mtu)
+ return;
+
if (mtu < ip_rt_min_pmtu)
mtu = ip_rt_min_pmtu;
--
2.1.0
^ permalink raw reply related
* Re: [RESEND] ssb: Fix Sparse error in main
From: Kalle Valo @ 2015-01-29 8:18 UTC (permalink / raw)
To: Pramod Gurav
Cc: John Linville, Michael Buesch, netdev, linux-kernel,
linux-wireless, Pramod Gurav
In-Reply-To: <1422424187-19059-1-git-send-email-pramod.gurav@smartplayin.com>
> This change fixes below sparse error:
> drivers/ssb/main.c:94:16: warning: symbol 'ssb_sdio_func_to_bus'
> was not declared. Should it be static?
>
> Acked-by: Michael Buesch <m@bues.ch>
> Signed-off-by: Pramod Gurav <pramod.gurav@smartplayin.com>
Thanks, applied to wireless-drivers-next.git.
Kalle Valo
^ permalink raw reply
* Re: [PATCH v2] lib/checksum.c: fix carry in csum_tcpudp_nofold
From: Karl Beldan @ 2015-01-29 8:55 UTC (permalink / raw)
To: David Miller
Cc: karl.beldan, viro, eric.dumazet, arnd, vapier, netdev,
linux-kernel, stable
In-Reply-To: <20150128.223249.779296549693168395.davem@davemloft.net>
On Wed, Jan 28, 2015 at 10:32:49PM -0800, David Miller wrote:
> From: Karl Beldan <karl.beldan@gmail.com>
> Date: Wed, 28 Jan 2015 10:58:11 +0100
>
> > The carry from the 64->32bits folding was dropped, e.g with:
> > saddr=0xFFFFFFFF daddr=0xFF0000FF len=0xFFFF proto=0 sum=1,
> > csum_tcpudp_nofold returned 0 instead of 1.
> >
> > Signed-off-by: Karl Beldan <karl.beldan@rivierawaves.com>
>
> Applied, thanks.
Fengguang's robot reported I managed to break the build, this occurs
when using non-generic do_csum and generic csum_tcpudp_nofold.
I guess a v3 is irrelevant now, should I send another patch for the
build fix only ? Sorry for that.
Karl
^ permalink raw reply
* Re: [PATCH] net, ethernet, LLVMLinux: Add missing MODULE_DEVICE_TABLE()
From: Arnd Bergmann @ 2015-01-29 9:10 UTC (permalink / raw)
To: David Miller
Cc: behanw, sathya.perla, ajit.khaparde, linux-kernel, netdev,
subbu.seetharaman
In-Reply-To: <20150128.224228.1533046356464752424.davem@davemloft.net>
On Wednesday 28 January 2015 22:42:28 David Miller wrote:
> From: Behan Webster <behanw@converseincode.com>
> Date: Wed, 28 Jan 2015 17:36:14 -0800
>
> > Missing MODULE_DEVICE_TABLE for pci ids from benet driver found by clang.
> >
> > Signed-off-by: Behan Webster <behanw@converseincode.com>
> > Suggested-by: Arnd Bergmann <arnd@arndb.de>
>
> Why are you removing the device table?
Behan took a patch that I did earlier and split it up to add descriptions.
The patch is correct, but he either misunderstood or misexpressed the
intention.
This driver has two identical lines that both say
MODULE_DEVICE_TABLE(pci, be_dev_ids);
I don't remember the exact symptom, but llvm/clang trips over this, while gcc
silently ignores the second one.
Arnd
^ permalink raw reply
* [PATCH] stmmac: DMA threshold mode or SF mode can be different among multiple device instance
From: Sonic Zhang @ 2015-01-29 8:54 UTC (permalink / raw)
To: Giuseppe Cavallaro, David S. Miller
Cc: netdev, adi-buildroot-devel, Sonic Zhang
From: Sonic Zhang <sonic.zhang@analog.com>
- In tx_hard_error_bump_tc interrupt, tc should be bumped only when current
device instance is in DMA threshold mode. Check per device xstats.threshold
other than global tc.
- Set per device xstats.threshold to SF_DMA_MODE when current device
instance is set to SF mode.
Signed-off-by: Sonic Zhang <sonic.zhang@analog.com>
---
drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index cd6ebda..6c87850 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -1287,7 +1287,7 @@ static void stmmac_dma_operation_mode(struct stmmac_priv *priv)
* that needs to not insert csum in the TDES.
*/
priv->hw->dma->dma_mode(priv->ioaddr, SF_DMA_MODE, SF_DMA_MODE);
- tc = SF_DMA_MODE;
+ priv->xstats.threshold = SF_DMA_MODE;
} else
priv->hw->dma->dma_mode(priv->ioaddr, tc, SF_DMA_MODE);
}
@@ -1444,7 +1444,8 @@ static void stmmac_dma_interrupt(struct stmmac_priv *priv)
}
if (unlikely(status & tx_hard_error_bump_tc)) {
/* Try to bump up the dma threshold on this failure */
- if (unlikely(tc != SF_DMA_MODE) && (tc <= 256)) {
+ if (unlikely(priv->xstats.threshold != SF_DMA_MODE) &&
+ (tc <= 256)) {
tc += 64;
if (priv->plat->force_thresh_dma_mode)
priv->hw->dma->dma_mode(priv->ioaddr, tc, tc);
--
1.7.9.5
^ permalink raw reply related
* Re: Bug in netlink_bind
From: Pablo Neira Ayuso @ 2015-01-29 9:40 UTC (permalink / raw)
To: Ivan Delalande; +Cc: netdev
In-Reply-To: <20150128234646.GA23945@ycc.fr>
On Thu, Jan 29, 2015 at 12:46:46AM +0100, Ivan Delalande wrote:
> Hi,
>
> I’ve been trying to debug some of our tests that began failing when
> upgrading to 3.18. Our actual failure is caused by the condition added
> in commit 97840cb to nfnetlink_bind but the bug has probably been
> introduced by 0329274.
Will resend this patch asap:
http://patchwork.ozlabs.org/patch/426205/
It didn't apply cleanly last time I submitted it, sorry.
^ permalink raw reply
* [PATCH net] netlink: fix wrong subscription bitmask to group mapping in
From: Pablo Neira Ayuso @ 2015-01-29 9:51 UTC (permalink / raw)
To: netdev; +Cc: davem, colona, andre
The subscription bitmask passed via struct sockaddr_nl is converted to
the group number when calling the netlink_bind() and netlink_unbind()
callbacks.
The conversion is however incorrect since bitmask (1 << 0) needs to be
mapped to group number 1. Note that you cannot specify the group number 0
(usually known as _NONE) from setsockopt() using NETLINK_ADD_MEMBERSHIP
since this is rejected through -EINVAL.
This problem became noticeable since 97840cb ("netfilter: nfnetlink:
fix insufficient validation in nfnetlink_bind") when binding to bitmask
(1 << 0) in ctnetlink.
Reported-by: Andre Tomt <andre@tomt.net>
Reported-by: Ivan Delalande <colona@arista.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
v2: Rebased upon current net tree. Previous patch:
http://patchwork.ozlabs.org/patch/426205/
did not apply cleanly.
net/netlink/af_netlink.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c
index 02fdde2..75532ef 100644
--- a/net/netlink/af_netlink.c
+++ b/net/netlink/af_netlink.c
@@ -1438,7 +1438,7 @@ static void netlink_undo_bind(int group, long unsigned int groups,
for (undo = 0; undo < group; undo++)
if (test_bit(undo, &groups))
- nlk->netlink_unbind(sock_net(sk), undo);
+ nlk->netlink_unbind(sock_net(sk), undo + 1);
}
static int netlink_bind(struct socket *sock, struct sockaddr *addr,
@@ -1476,7 +1476,7 @@ static int netlink_bind(struct socket *sock, struct sockaddr *addr,
for (group = 0; group < nlk->ngroups; group++) {
if (!test_bit(group, &groups))
continue;
- err = nlk->netlink_bind(net, group);
+ err = nlk->netlink_bind(net, group + 1);
if (!err)
continue;
netlink_undo_bind(group, groups, sk);
--
1.7.10.4
^ permalink raw reply related
* [PATCH v2 3/3] lib/string_helpers.c: Change semantics of string_escape_mem
From: Rasmus Villemoes @ 2015-01-29 10:03 UTC (permalink / raw)
To: Andy Shevchenko, Andrew Morton, Trond Myklebust, J. Bruce Fields,
David S. Miller
Cc: Rasmus Villemoes, linux-kernel, linux-nfs, netdev
In-Reply-To: <1422525801-26560-1-git-send-email-linux@rasmusvillemoes.dk>
The current semantics of string_escape_mem are inadequate for one of
its two current users, vsnprintf(). If that is to honour its contract,
it must know how much space would be needed for the entire escaped
buffer, and string_escape_mem provides no way of obtaining that (short
of allocating a large enough buffer (~4 times input string) to let it
play with, and that's definitely a big no-no inside vsnprintf).
So change the semantics for string_escape_mem to be more
snprintf-like: Return the size of the output that would be generated
if the destination buffer was big enough, but of course still only
write to the part of dst it is allowed to, and don't do
'\0'-termination. It is then up to the caller to detect whether output
was truncated and to append a '\0' if desired.
This also fixes a bug in the escaped_string() helper function, which
used to unconditionally pass a length of "end-buf" to
string_escape_mem(); since the latter doesn't check osz for being
insanely large, it would happily write to dst. For example,
kasprintf(GFP_KERNEL, "something and then %pE", ...); is an easy way
to trigger an oops.
In test-string_helpers.c, I removed the now meaningless -ENOMEM test,
and replaced it with testing for getting the expected return value
even if the buffer is too small. Also ensure that nothing is written
when osz==0.
In net/sunrpc/cache.c, I think qword_add still has the same
semantics. Someone should definitely double-check this.
Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
---
include/linux/string_helpers.h | 10 +++++-----
lib/string_helpers.c | 25 ++++++++-----------------
lib/test-string_helpers.c | 37 ++++++++++++++++---------------------
lib/vsprintf.c | 2 +-
net/sunrpc/cache.c | 8 +++++---
5 files changed, 35 insertions(+), 47 deletions(-)
diff --git a/include/linux/string_helpers.h b/include/linux/string_helpers.h
index 6eb567ac56bc..7a082aa183a8 100644
--- a/include/linux/string_helpers.h
+++ b/include/linux/string_helpers.h
@@ -47,22 +47,22 @@ static inline int string_unescape_any_inplace(char *buf)
#define ESCAPE_ANY_NP (ESCAPE_ANY | ESCAPE_NP)
#define ESCAPE_HEX 0x20
-int string_escape_mem(const char *src, size_t isz, char **dst, size_t osz,
+size_t string_escape_mem(const char *src, size_t isz, char *dst, size_t osz,
unsigned int flags, const char *esc);
-static inline int string_escape_mem_any_np(const char *src, size_t isz,
- char **dst, size_t osz, const char *esc)
+static inline size_t string_escape_mem_any_np(const char *src, size_t isz,
+ char *dst, size_t osz, const char *esc)
{
return string_escape_mem(src, isz, dst, osz, ESCAPE_ANY_NP, esc);
}
-static inline int string_escape_str(const char *src, char **dst, size_t sz,
+static inline size_t string_escape_str(const char *src, char *dst, size_t sz,
unsigned int flags, const char *esc)
{
return string_escape_mem(src, strlen(src), dst, sz, flags, esc);
}
-static inline int string_escape_str_any_np(const char *src, char **dst,
+static inline size_t string_escape_str_any_np(const char *src, char *dst,
size_t sz, const char *esc)
{
return string_escape_str(src, dst, sz, ESCAPE_ANY_NP, esc);
diff --git a/lib/string_helpers.c b/lib/string_helpers.c
index e14dd8555760..05d4a583d20e 100644
--- a/lib/string_helpers.c
+++ b/lib/string_helpers.c
@@ -414,20 +414,17 @@ static bool escape_hex(unsigned char c, char **dst, char *end)
* it if needs.
*
* Return:
- * The amount of the characters processed to the destination buffer, or
- * %-ENOMEM if the size of buffer is not enough to put an escaped character is
- * returned.
- *
- * Even in the case of error @dst pointer will be updated to point to the byte
- * after the last processed character.
+ * The total size of the escaped output that would be generated for
+ * the given input and flags. To check whether the output was
+ * truncated, compare the return value to osz. There is room left in
+ * dst for a '\0' terminator if and only if ret < osz.
*/
-int string_escape_mem(const char *src, size_t isz, char **dst, size_t osz,
- unsigned int flags, const char *esc)
+size_t string_escape_mem(const char *src, size_t isz, char *dst, size_t osz,
+ unsigned int flags, const char *esc)
{
- char *p = *dst;
+ char *p = dst;
char *end = p + osz;
bool is_dict = esc && *esc;
- int ret;
while (isz--) {
unsigned char c = *src++;
@@ -466,13 +463,7 @@ int string_escape_mem(const char *src, size_t isz, char **dst, size_t osz,
escape_passthrough(c, &p, end);
}
- if (p > end) {
- *dst = end;
- return -ENOMEM;
- }
- ret = p - *dst;
- *dst = p;
- return ret;
+ return p - dst;
}
EXPORT_SYMBOL(string_escape_mem);
diff --git a/lib/test-string_helpers.c b/lib/test-string_helpers.c
index ab0d30e1e18f..5f95114a2f86 100644
--- a/lib/test-string_helpers.c
+++ b/lib/test-string_helpers.c
@@ -264,12 +264,12 @@ static __init void test_string_escape(const char *name,
const struct test_string_2 *s2,
unsigned int flags, const char *esc)
{
- int q_real = 512;
- char *out_test = kmalloc(q_real, GFP_KERNEL);
- char *out_real = kmalloc(q_real, GFP_KERNEL);
+ size_t out_size = 512;
+ char *out_test = kmalloc(out_size, GFP_KERNEL);
+ char *out_real = kmalloc(out_size, GFP_KERNEL);
char *in = kmalloc(256, GFP_KERNEL);
- char *buf = out_real;
- int p = 0, q_test = 0;
+ size_t p = 0, q_test = 0;
+ size_t q_real;
if (!out_test || !out_real || !in)
goto out;
@@ -301,29 +301,26 @@ static __init void test_string_escape(const char *name,
q_test += len;
}
- q_real = string_escape_mem(in, p, &buf, q_real, flags, esc);
+ q_real = string_escape_mem(in, p, out_real, out_size, flags, esc);
test_string_check_buf(name, flags, in, p, out_real, q_real, out_test,
q_test);
+
+ memset(out_real, 'Z', out_size);
+ q_real = string_escape_mem(in, p, out_real, 0, flags, esc);
+ if (q_real != q_test)
+ pr_warn("Test '%s' failed: flags = %u, osz = 0, expected %zu, got %zu\n",
+ name, flags, q_test, q_real);
+ if (memchr_inv(out_real, 'Z', out_size))
+ pr_warn("Test '%s' failed: osz = 0 but string_escape_mem wrote to the buffer\n",
+ name);
+
out:
kfree(in);
kfree(out_real);
kfree(out_test);
}
-static __init void test_string_escape_nomem(void)
-{
- char *in = "\eb \\C\007\"\x90\r]";
- char out[64], *buf = out;
- int rc = -ENOMEM, ret;
-
- ret = string_escape_str_any_np(in, &buf, strlen(in), NULL);
- if (ret == rc)
- return;
-
- pr_err("Test 'escape nomem' failed: got %d instead of %d\n", ret, rc);
-}
-
static int __init test_string_helpers_init(void)
{
unsigned int i;
@@ -342,8 +339,6 @@ static int __init test_string_helpers_init(void)
for (i = 0; i < (ESCAPE_ANY_NP | ESCAPE_HEX) + 1; i++)
test_string_escape("escape 1", escape1, i, TEST_STRING_2_DICT_1);
- test_string_escape_nomem();
-
return -EINVAL;
}
module_init(test_string_helpers_init);
diff --git a/lib/vsprintf.c b/lib/vsprintf.c
index 3568e3906777..d02c394b5b58 100644
--- a/lib/vsprintf.c
+++ b/lib/vsprintf.c
@@ -1160,7 +1160,7 @@ char *escaped_string(char *buf, char *end, u8 *addr, struct printf_spec spec,
len = spec.field_width < 0 ? 1 : spec.field_width;
/* Ignore the error. We print as many characters as we can */
- string_escape_mem(addr, len, &buf, end - buf, flags, NULL);
+ buf += string_escape_mem(addr, len, buf, buf < end ? end - buf : 0, flags, NULL);
return buf;
}
diff --git a/net/sunrpc/cache.c b/net/sunrpc/cache.c
index 33fb105d4352..22c4418057f4 100644
--- a/net/sunrpc/cache.c
+++ b/net/sunrpc/cache.c
@@ -1068,12 +1068,14 @@ void qword_add(char **bpp, int *lp, char *str)
{
char *bp = *bpp;
int len = *lp;
- int ret;
+ int ret, written;
if (len < 0) return;
- ret = string_escape_str(str, &bp, len, ESCAPE_OCTAL, "\\ \n\t");
- if (ret < 0 || ret == len)
+ ret = string_escape_str(str, bp, len, ESCAPE_OCTAL, "\\ \n\t");
+ written = min(ret, len);
+ bp += written;
+ if (ret >= len)
len = -1;
else {
len -= ret;
--
2.1.3
^ permalink raw reply related
* [PATCH v2 0/3] Two printf fixes
From: Rasmus Villemoes @ 2015-01-29 10:03 UTC (permalink / raw)
To: Andy Shevchenko, Andrew Morton, Trond Myklebust, J. Bruce Fields,
David S. Miller
Cc: Rasmus Villemoes, linux-kernel, linux-nfs, netdev
In-Reply-To: <1422451543-12401-1-git-send-email-linux@rasmusvillemoes.dk>
Both %pE and %ph are unusable in kasprintf(), since the occurrence of
either will trigger an oops during the first vsnprintf call where
kasprintf tries to find the correct size to allocate. These oopses
could be papered over with somewhat smaller patches than these, but
then the return value from vsnprintf would still not reflect the
actual size needed. For %pE, this requires a change of semantics of
string_escape_mem and hence an annoyingly large diffstat.
Whether this is 3.20, 3.21 and/or -stable material (or /dev/null
material, for that matter ;-)) I'll leave to others to decide.
v2: Suggestions from Andy Shevchenko:
* Simpler fix of hex_string().
* The string_escape_mem change is split in two, 2/3 updating the
internal helpers and 3/3 then changing the external interface.
Rasmus Villemoes (3):
lib/vsprintf.c: Fix potential NULL deref in hex_string
lib/string_helpers.c: Refactor string_escape_mem
lib/string_helpers.c: Change semantics of string_escape_mem
include/linux/string_helpers.h | 10 +--
lib/string_helpers.c | 191 ++++++++++++++++-------------------------
lib/test-string_helpers.c | 37 ++++----
lib/vsprintf.c | 18 ++--
net/sunrpc/cache.c | 8 +-
5 files changed, 111 insertions(+), 153 deletions(-)
--
2.1.3
^ permalink raw reply
* [PATCH net] net: sctp: fix passing wrong parameter header to param_type2af in sctp_process_param
From: Daniel Borkmann @ 2015-01-29 10:05 UTC (permalink / raw)
To: davem; +Cc: vyasevich, netdev, linux-sctp, Saran Maruti Ramanara
From: Saran Maruti Ramanara <saran.neti@telus.com>
When making use of RFC5061, section 4.2.4. for setting the primary IP
address, we're passing a wrong parameter header to param_type2af(),
resulting always in NULL being returned.
At this point, param.p points to a sctp_addip_param struct, containing
a sctp_paramhdr (type = 0xc004, length = var), and crr_id as a correlation
id. Followed by that, as also presented in RFC5061 section 4.2.4., comes
the actual sctp_addr_param, which also contains a sctp_paramhdr, but
this time with the correct type SCTP_PARAM_IPV{4,6}_ADDRESS that
param_type2af() can make use of. Since we already hold a pointer to
addr_param from previous line, just reuse it for param_type2af().
Fixes: d6de3097592b ("[SCTP]: Add the handling of "Set Primary IP Address" parameter to INIT")
Signed-off-by: Saran Maruti Ramanara <saran.neti@telus.com>
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
---
net/sctp/sm_make_chunk.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/net/sctp/sm_make_chunk.c b/net/sctp/sm_make_chunk.c
index e49e231..06320c8 100644
--- a/net/sctp/sm_make_chunk.c
+++ b/net/sctp/sm_make_chunk.c
@@ -2608,7 +2608,7 @@ do_addr_param:
addr_param = param.v + sizeof(sctp_addip_param_t);
- af = sctp_get_af_specific(param_type2af(param.p->type));
+ af = sctp_get_af_specific(param_type2af(addr_param->p.type));
if (af == NULL)
break;
--
1.7.11.7
^ permalink raw reply related
* [PATCH] lib/checksum.c: fix build for generic csum_tcpudp_nofold
From: Karl Beldan @ 2015-01-29 10:10 UTC (permalink / raw)
To: David S . Miller
Cc: Karl Beldan, Karl Beldan, netdev, linux-kernel, Eric Dumazet
Fixed commit added from64to32 under _#ifndef do_csum_ but used it
under _#ifndef csum_tcpudp_nofold_, breaking some builds (Fengguang's
robot reported TILEGX's). Move from64to32 under the latter.
Fixes: 150ae0e94634 ("lib/checksum.c: fix carry in csum_tcpudp_nofold")
Reported-by: kbuild test robot <fengguang.wu@intel.com>
Signed-off-by: Karl Beldan <karl.beldan@rivierawaves.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: David S. Miller <davem@davemloft.net>
---
lib/checksum.c | 18 +++++++++---------
1 file changed, 9 insertions(+), 9 deletions(-)
diff --git a/lib/checksum.c b/lib/checksum.c
index fcf3894..8b39e86 100644
--- a/lib/checksum.c
+++ b/lib/checksum.c
@@ -47,15 +47,6 @@ static inline unsigned short from32to16(unsigned int x)
return x;
}
-static inline u32 from64to32(u64 x)
-{
- /* add up 32-bit and 32-bit for 32+c bit */
- x = (x & 0xffffffff) + (x >> 32);
- /* add up carry.. */
- x = (x & 0xffffffff) + (x >> 32);
- return (u32)x;
-}
-
static unsigned int do_csum(const unsigned char *buff, int len)
{
int odd;
@@ -190,6 +181,15 @@ csum_partial_copy(const void *src, void *dst, int len, __wsum sum)
EXPORT_SYMBOL(csum_partial_copy);
#ifndef csum_tcpudp_nofold
+static inline u32 from64to32(u64 x)
+{
+ /* add up 32-bit and 32-bit for 32+c bit */
+ x = (x & 0xffffffff) + (x >> 32);
+ /* add up carry.. */
+ x = (x & 0xffffffff) + (x >> 32);
+ return (u32)x;
+}
+
__wsum csum_tcpudp_nofold(__be32 saddr, __be32 daddr,
unsigned short len,
unsigned short proto,
--
2.0.1
^ permalink raw reply related
* Re: [PATCH RFC 1/2] ipv6: Fix after pmtu events dissapearing host routes
From: Hannes Frederic Sowa @ 2015-01-29 10:26 UTC (permalink / raw)
To: Steffen Klassert; +Cc: Yang Yingliang, netdev, David S. Miller
In-Reply-To: <20150128121150.GN13046@secunet.com>
On Mi, 2015-01-28 at 13:11 +0100, Steffen Klassert wrote:
> We currently don't clone host routes before we use them.
> If a pmtu event is received on such a route, it gets
> an expires value. As soon as the expiration time is
> elapsed, the route is deleted. As a result, the host
> is not reachable any more.
>
> We fix this by cloning host routes if they are gatewayed,
> i.e. if pmtu events can happen.
>
> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
> ---
> net/ipv6/route.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/net/ipv6/route.c b/net/ipv6/route.c
> index c910831..3e864e7 100644
> --- a/net/ipv6/route.c
> +++ b/net/ipv6/route.c
> @@ -961,7 +961,7 @@ redo_rt6_select:
>
> if (!(rt->rt6i_flags & (RTF_NONEXTHOP | RTF_GATEWAY)))
> nrt = rt6_alloc_cow(rt, &fl6->daddr, &fl6->saddr);
> - else if (!(rt->dst.flags & DST_HOST))
> + else if (!(rt->dst.flags & DST_HOST) || (rt->rt6i_flags & RTF_GATEWAY))
> nrt = rt6_alloc_clone(rt, &fl6->daddr);
> else
> goto out2;
My approach was to suppress mtu updates on the loopback interface.
Hmm...
Bye,
Hannes
^ permalink raw reply
* Re: [PATCHv3, ipsec-next] xfrm: Do not parse 32bits compiled xfrm netlink msg on 64bits host
From: Nicolas Dichtel @ 2015-01-29 10:29 UTC (permalink / raw)
To: Fan Du, steffen.klassert; +Cc: herbert, davem, netdev, fengyuleidian0615
In-Reply-To: <1422349230-17394-1-git-send-email-fan.du@intel.com>
Le 27/01/2015 10:00, Fan Du a écrit :
> structure like xfrm_usersa_info or xfrm_userpolicy_info
> has different sizeof when compiled as 32bits and 64bits
> due to not appending pack attribute in their definition.
> This will result in broken SA and SP information when user
> trying to configure them through netlink interface.
>
> Inform user land about this situation instead of keeping
> silent, the upper test scripts would behave accordingly.
>
> Quotes from: http://marc.info/?l=linux-netdev&m=142226348715503&w=2
>>
>> Before a clean solution show up, I think it's better to warn user in some way
>> like http://patchwork.ozlabs.org/patch/323842/ did. Otherwise, many people
>> who stuck there will always spend time and try to fix this issue in whatever way.
>
> Yes, this is the first thing we should do. I'm willing to accept a patch
>
> Signed-off-by: Fan Du <fan.du@intel.com>
A way to solve this problem was to provide to userland a xfrm compat header
file, which match the ABI of the kernel. Something like:
#include <linux/xfrm.h>
#define xfrm_usersa_info xfrm_usersa_info_64
#define xfrm_usersa_info_compat xfrm_usersa_info
struct xfrm_usersa_info_compat {
struct xfrm_selector sel;
struct xfrm_id id;
xfrm_address_t saddr;
struct xfrm_lifetime_cfg lft;
struct xfrm_lifetime_cur curlft;
struct xfrm_stats stats;
__u32 seq;
__u32 reqid;
__u16 family;
__u8 mode;
__u8 replay_window;
__u8 flags;
__u8 hole1;
__u32 hole2;
};
The point I try to make is that patching userland apps allows to use xfrm on a
32bits userland / 64bits kernel.
If I understand well your patch, it will not be possible anymore, all messages
will be rejected. And this may break existing apps.
Regards,
Nicolas
^ permalink raw reply
* Re: [PATCH RFC 1/2] ipv6: Fix after pmtu events dissapearing host routes
From: Steffen Klassert @ 2015-01-29 10:44 UTC (permalink / raw)
To: Hannes Frederic Sowa; +Cc: Yang Yingliang, netdev, David S. Miller
In-Reply-To: <1422527211.2861.2.camel@redhat.com>
On Thu, Jan 29, 2015 at 11:26:51AM +0100, Hannes Frederic Sowa wrote:
> On Mi, 2015-01-28 at 13:11 +0100, Steffen Klassert wrote:
> > We currently don't clone host routes before we use them.
> > If a pmtu event is received on such a route, it gets
> > an expires value. As soon as the expiration time is
> > elapsed, the route is deleted. As a result, the host
> > is not reachable any more.
> >
> > We fix this by cloning host routes if they are gatewayed,
> > i.e. if pmtu events can happen.
> >
> > Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
> > ---
> > net/ipv6/route.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/net/ipv6/route.c b/net/ipv6/route.c
> > index c910831..3e864e7 100644
> > --- a/net/ipv6/route.c
> > +++ b/net/ipv6/route.c
> > @@ -961,7 +961,7 @@ redo_rt6_select:
> >
> > if (!(rt->rt6i_flags & (RTF_NONEXTHOP | RTF_GATEWAY)))
> > nrt = rt6_alloc_cow(rt, &fl6->daddr, &fl6->saddr);
> > - else if (!(rt->dst.flags & DST_HOST))
> > + else if (!(rt->dst.flags & DST_HOST) || (rt->rt6i_flags & RTF_GATEWAY))
> > nrt = rt6_alloc_clone(rt, &fl6->daddr);
> > else
> > goto out2;
>
> My approach was to suppress mtu updates on the loopback interface.
> Hmm...
Maybe we need to do this too. My patch fixes just the case
where we get a valid pmtu update from a remote host.
^ permalink raw reply
* Re: [ath9k-devel] [PATCH] Repair soft lockup with monitor mode of ath9k_htc card
From: Oleksij Rempel @ 2015-01-29 10:52 UTC (permalink / raw)
To: zhengyuwei, linux-kernel, ath9k-devel, linux-wireless, kvalo,
ath9k-devel
Cc: netdev
In-Reply-To: <1422504583-17839-1-git-send-email-zhengyuwei@360.cn>
[-- Attachment #1: Type: text/plain, Size: 12559 bytes --]
Am 29.01.2015 um 05:09 schrieb zhengyuwei@360.cn:
> From: Yuwei Zheng <zhengyuwei@360.cn>
>
> In the environment with heavy wifi traffic, set the ar9271 into monitor mode, will
> trigger a deadloop panic.
>
> The ath9k_hif_usb_rx_cb function excute on the interrupt context, and ath9k_rx_tasklet excute
> on the soft irq context. In other words, the ath9k_hif_usb_rx_cb have more chance to excute than
> ath9k_rx_tasklet. So in the worst condition, the rx.rxbuf receive list is always full,
> and the do {}while(true) loop will not be break. The kernel get a soft lockup panic.
>
> [59011.007210] BUG: soft lockup - CPU#0 stuck for 23s!
> [kworker/0:0:30609]
> [59011.030560] BUG: scheduling while atomic: kworker/0:0/30609/0x40010100
> [59013.804486] BUG: scheduling while atomic: kworker/0:0/30609/0x40010100
> [59013.858522] Kernel panic - not syncing: softlockup: hung tasks
>
> [59014.038891] Exception stack(0xdf4bbc38 to 0xdf4bbc80)
> [59014.046834] bc20: de57b950 60000113
> [59014.059579] bc40: 00000000 bb32bb32 60000113 de57b948 de57b500 dc7bb440 df4bbcd0 00000000
> [59014.072337] bc60: de57b950 60000113 df4bbcd0 df4bbc80 c04c259d c04c25a0 60000133 ffffffff
> [59014.085233] [<c04c28db>] (__irq_svc+0x3b/0x5c) from [<c04c25a0>] (_raw_spin_unlock_irqrestore+0xc/0x10)
> [59014.100437] [<c04c25a0>] (_raw_spin_unlock_irqrestore+0xc/0x10) from [<bf9c2089>] (ath9k_rx_tasklet+0x290/0x490 [ath9k_htc])
> [59014.118267] [<bf9c2089>] (ath9k_rx_tasklet+0x290/0x490 [ath9k_htc]) from [<c0036d23>] (tasklet_action+0x3b/0x98)
> [59014.134132] [<c0036d23>] (tasklet_action+0x3b/0x98) from [<c0036709>] (__do_softirq+0x99/0x16c)
> [59014.147784] [<c0036709>] (__do_softirq+0x99/0x16c) from [<c00369f7>] (irq_exit+0x5b/0x5c)
> [59014.160653] [<c00369f7>] (irq_exit+0x5b/0x5c) from [<c000cfc3>] (handle_IRQ+0x37/0x78)
> [59014.173124] [<c000cfc3>] (handle_IRQ+0x37/0x78) from [<c00085df>] (omap3_intc_handle_irq+0x5f/0x68)
> [59014.187225] [<c00085df>] (omap3_intc_handle_irq+0x5f/0x68) from [<c04c28db>](__irq_svc+0x3b/0x5c)
>
> This bug can be see with low performance board, such as uniprocessor beagle bone board.
> Signed-off-by: Yuwei Zheng <zhengyuwei@360.cn>
>
> ---
> drivers/net/wireless/ath/ath9k/hif_usb.c | 53 ++++++++++++++++++++++----
> drivers/net/wireless/ath/ath9k/hif_usb.h | 5 +++
> drivers/net/wireless/ath/ath9k/htc.h | 13 +++++++
> drivers/net/wireless/ath/ath9k/htc_drv_debug.c | 49 ++++++++++++++++++++++++
> drivers/net/wireless/ath/ath9k/htc_drv_txrx.c | 26 +++++++++++++
> 5 files changed, 139 insertions(+), 7 deletions(-)
First of all, thank you for you work! :D
Please run ./scripts/checkpatch.pl yourpatch_path
i get:
total: 139 errors, 12 warnings, 2 checks, 231 lines checked
You use tasklet_hrtimer_start. So far i know, there is no this kind of
hrtimer which is actually hidden behind this word on this SoC.
Especially if requested value is any way in 1 millisecond range you
probably can and should use normal priority tasklet. (correct me if i'm
wrong)
> diff --git a/drivers/net/wireless/ath/ath9k/hif_usb.c b/drivers/net/wireless/ath/ath9k/hif_usb.c
> index 8e7153b..febea5e 100644
> --- a/drivers/net/wireless/ath/ath9k/hif_usb.c
> +++ b/drivers/net/wireless/ath/ath9k/hif_usb.c
> @@ -658,7 +658,6 @@ static void ath9k_hif_usb_rx_cb(struct urb *urb)
> default:
> goto resubmit;
> }
> -
> if (likely(urb->actual_length != 0)) {
> skb_put(skb, urb->actual_length);
> ath9k_hif_usb_rx_stream(hif_dev, skb);
> @@ -667,12 +666,18 @@ static void ath9k_hif_usb_rx_cb(struct urb *urb)
> resubmit:
> skb_reset_tail_pointer(skb);
> skb_trim(skb, 0);
> -
> - usb_anchor_urb(urb, &hif_dev->rx_submitted);
> - ret = usb_submit_urb(urb, GFP_ATOMIC);
> - if (ret) {
> - usb_unanchor_urb(urb);
> - goto free;
> + if (atomic_read(&hif_dev->rx_urb_submit_delay) > 0) {
> + usb_anchor_urb(urb, &hif_dev->rx_delayed_submitted);
> + ret = tasklet_hrtimer_start(&hif_dev->rx_submit_timer,
> + ktime_set(0, atomic_read(&hif_dev->rx_urb_submit_delay)*1000),
> + HRTIMER_MODE_REL);
> + } else {
> + usb_anchor_urb(urb, &hif_dev->rx_submitted);
> + ret = usb_submit_urb(urb, GFP_ATOMIC);
> + if (ret) {
> + usb_unanchor_urb(urb);
> + goto free;
> + }
> }
>
> return;
> @@ -818,9 +823,37 @@ err:
> return -ENOMEM;
> }
>
> +static enum hrtimer_restart rx_urb_submit_timer_handler(struct hrtimer *me)
> +{
> + struct tasklet_hrtimer *thr = container_of(me, struct tasklet_hrtimer, timer);
> + struct hif_device_usb *hif_dev = container_of(thr, struct hif_device_usb, rx_submit_timer);
> + struct urb *urb = NULL;
> + struct sk_buff *skb = NULL;
> + int ret;
> +
> + while (true) {
> + urb = usb_get_from_anchor(&hif_dev->rx_delayed_submitted);
> + if (urb != NULL) {
> + skb = (struct sk_buff *)urb->context;
> + ret = usb_submit_urb(urb, GFP_ATOMIC);
> + if (ret != -EBUSY) {
> + usb_unanchor_urb(urb);
> + dev_kfree_skb_any(skb);
> + urb->context = NULL;
> + }
> + } else {
> + break;
> + }
> + }
> +
> + return HRTIMER_NORESTART;
> +}
> +
> static void ath9k_hif_usb_dealloc_rx_urbs(struct hif_device_usb *hif_dev)
> {
> usb_kill_anchored_urbs(&hif_dev->rx_submitted);
> + usb_kill_anchored_urbs(&hif_dev->rx_delayed_submitted);
> + tasklet_hrtimer_cancel(&hif_dev->rx_submit_timer);
> }
>
> static int ath9k_hif_usb_alloc_rx_urbs(struct hif_device_usb *hif_dev)
> @@ -830,6 +863,8 @@ static int ath9k_hif_usb_alloc_rx_urbs(struct hif_device_usb *hif_dev)
> int i, ret;
>
> init_usb_anchor(&hif_dev->rx_submitted);
> + init_usb_anchor(&hif_dev->rx_delayed_submitted);
> +
> spin_lock_init(&hif_dev->rx_lock);
>
> for (i = 0; i < MAX_RX_URB_NUM; i++) {
> @@ -871,6 +906,10 @@ static int ath9k_hif_usb_alloc_rx_urbs(struct hif_device_usb *hif_dev)
> usb_free_urb(urb);
> }
>
> + /* add for flow control*/
> + atomic_set(&hif_dev->rx_urb_submit_delay, 0);
> + tasklet_hrtimer_init(&hif_dev->rx_submit_timer, rx_urb_submit_timer_handler, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
> +
> return 0;
>
> err_submit:
> diff --git a/drivers/net/wireless/ath/ath9k/hif_usb.h b/drivers/net/wireless/ath/ath9k/hif_usb.h
> index 51496e7..56d6be8 100644
> --- a/drivers/net/wireless/ath/ath9k/hif_usb.h
> +++ b/drivers/net/wireless/ath/ath9k/hif_usb.h
> @@ -98,9 +98,14 @@ struct hif_device_usb {
> struct hif_usb_tx tx;
> struct usb_anchor regout_submitted;
> struct usb_anchor rx_submitted;
> + struct usb_anchor rx_delayed_submitted; /* delayed submit anchor */
> struct usb_anchor reg_in_submitted;
> struct usb_anchor mgmt_submitted;
> struct sk_buff *remain_skb;
> +
> + struct tasklet_hrtimer rx_submit_timer;/* delayed submit hrtimer */
> + atomic_t rx_urb_submit_delay; /*us*/
> +
> const char *fw_name;
> int rx_remain_len;
> int rx_pkt_len;
> diff --git a/drivers/net/wireless/ath/ath9k/htc.h b/drivers/net/wireless/ath/ath9k/htc.h
> index 9dde265..453d0a8 100644
> --- a/drivers/net/wireless/ath/ath9k/htc.h
> +++ b/drivers/net/wireless/ath/ath9k/htc.h
> @@ -331,6 +331,10 @@ static inline struct ath9k_htc_tx_ctl *HTC_SKB_CB(struct sk_buff *skb)
>
> #define TX_QSTAT_INC(q) (priv->debug.tx_stats.queue_stats[q]++)
>
> +#define TASKLETRX_STAT_INC(c) (hif_dev->htc_handle->drv_priv->debug.taskletrx_stats.c++)
> +#define TASKLETRX_STAT_ADD(c, a) (hif_dev->htc_handle->drv_priv->debug.taskletrx_stats.c += a)
> +#define TASKLETRX_STAT_SET(c, a) (hif_dev->htc_handle->drv_priv->debug.taskletrx_stats.c = a)
> +
> void ath9k_htc_err_stat_rx(struct ath9k_htc_priv *priv,
> struct ath_rx_status *rs);
>
> @@ -352,11 +356,20 @@ struct ath_skbrx_stats {
> u32 skb_dropped;
> };
>
> +struct ath_taskletrx_stats {
> + u32 taskletrx_looptimes;
> + u32 taskletrx_highwater;
> + u32 taskletrx_lowwater;
> + u32 taskletrx_watermark_triggered;
> + u32 taskletrx_urb_submit_delay;
> +};
> +
> struct ath9k_debug {
> struct dentry *debugfs_phy;
> struct ath_tx_stats tx_stats;
> struct ath_rx_stats rx_stats;
> struct ath_skbrx_stats skbrx_stats;
> + struct ath_taskletrx_stats taskletrx_stats;
> };
>
> void ath9k_htc_get_et_strings(struct ieee80211_hw *hw,
> diff --git a/drivers/net/wireless/ath/ath9k/htc_drv_debug.c b/drivers/net/wireless/ath/ath9k/htc_drv_debug.c
> index 8cef1ed..7c8322e 100644
> --- a/drivers/net/wireless/ath/ath9k/htc_drv_debug.c
> +++ b/drivers/net/wireless/ath/ath9k/htc_drv_debug.c
> @@ -286,6 +286,51 @@ static const struct file_operations fops_skb_rx = {
> .llseek = default_llseek,
> };
>
> +static ssize_t read_file_tasklet_rx(struct file *file, char __user *user_buf,
> + size_t count, loff_t *ppos)
> +{
> + struct ath9k_htc_priv *priv = file->private_data;
> + char *buf;
> + unsigned int len = 0, size = 1500;
> + ssize_t retval = 0;
> +
> + buf = kzalloc(size, GFP_KERNEL);
> + if (buf == NULL)
> + return -ENOMEM;
> +
> + len += scnprintf(buf + len, size - len,
> + "%20s : %10u\n", "Loop times",
> + priv->debug.taskletrx_stats.taskletrx_looptimes);
> + len += scnprintf(buf + len, size - len,
> + "%20s : %10u\n", "High watermark",
> + priv->debug.taskletrx_stats.taskletrx_highwater);
> + len += scnprintf(buf + len, size - len,
> + "%20s : %10u\n", "Low watermark",
> + priv->debug.taskletrx_stats.taskletrx_lowwater);
> +
> + len += scnprintf(buf + len, size - len,
> + "%20s : %10u\n", "WM triggered",
> + priv->debug.taskletrx_stats.taskletrx_watermark_triggered);
> +
> + len += scnprintf(buf + len, size - len,
> + "%20s : %10u\n", "URB delay",
> + priv->debug.taskletrx_stats.taskletrx_urb_submit_delay);
> + if (len > size)
> + len = size;
> +
> + retval = simple_read_from_buffer(user_buf, count, ppos, buf, len);
> + kfree(buf);
> +
> + return retval;
> +}
> +
> +static const struct file_operations fops_tasklet_rx = {
> + .read = read_file_tasklet_rx,
> + .open = simple_open,
> + .owner = THIS_MODULE,
> + .llseek = default_llseek,
> +};
> +
> static ssize_t read_file_slot(struct file *file, char __user *user_buf,
> size_t count, loff_t *ppos)
> {
> @@ -518,7 +563,11 @@ int ath9k_htc_init_debug(struct ath_hw *ah)
> debugfs_create_file("skb_rx", S_IRUSR, priv->debug.debugfs_phy,
> priv, &fops_skb_rx);
>
> + debugfs_create_file("tasklet_rx", S_IRUSR, priv->debug.debugfs_phy,
> + priv, &fops_tasklet_rx);
> +
> ath9k_cmn_debug_recv(priv->debug.debugfs_phy, &priv->debug.rx_stats);
> +
> ath9k_cmn_debug_phy_err(priv->debug.debugfs_phy, &priv->debug.rx_stats);
>
> debugfs_create_file("slot", S_IRUSR, priv->debug.debugfs_phy,
> diff --git a/drivers/net/wireless/ath/ath9k/htc_drv_txrx.c b/drivers/net/wireless/ath/ath9k/htc_drv_txrx.c
> index a0f58e2..f5e6217 100644
> --- a/drivers/net/wireless/ath/ath9k/htc_drv_txrx.c
> +++ b/drivers/net/wireless/ath/ath9k/htc_drv_txrx.c
> @@ -1061,7 +1061,28 @@ void ath9k_rx_tasklet(unsigned long data)
> unsigned long flags;
> struct ieee80211_hdr *hdr;
>
> + /* add for adaptive flow control*/
> + int looptimes = 0;
> + int highwatermark = ATH9K_HTC_RXBUF*3/4;
> + int lowwatermark = ATH9K_HTC_RXBUF/4;
> + unsigned int delay = 0;
> +
> + struct htc_target *htc = priv->htc;
> + struct hif_device_usb *hif_dev = htc->hif_dev;
> +
> + TASKLETRX_STAT_SET(taskletrx_highwater, highwatermark);
> + TASKLETRX_STAT_SET(taskletrx_lowwater, lowwatermark);
> +
> do {
> + looptimes++;
> + TASKLETRX_STAT_SET(taskletrx_looptimes, looptimes);
> + if (looptimes > highwatermark) {
> + delay = looptimes*10;
> + atomic_set(&hif_dev->rx_urb_submit_delay, delay);
> + TASKLETRX_STAT_INC(taskletrx_watermark_triggered);
> + TASKLETRX_STAT_SET(taskletrx_urb_submit_delay, delay);
> + }
> +
> spin_lock_irqsave(&priv->rx.rxbuflock, flags);
> list_for_each_entry(tmp_buf, &priv->rx.rxbuf, list) {
> if (tmp_buf->in_process) {
> @@ -1072,6 +1093,11 @@ void ath9k_rx_tasklet(unsigned long data)
>
> if (rxbuf == NULL) {
> spin_unlock_irqrestore(&priv->rx.rxbuflock, flags);
> + if (looptimes < lowwatermark) {
> + atomic_set(&hif_dev->rx_urb_submit_delay, 0);
> + TASKLETRX_STAT_SET(taskletrx_urb_submit_delay, 0);
> + }
> +
> break;
> }
>
>
--
Regards,
Oleksij
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 213 bytes --]
^ permalink raw reply
* Re: [ath9k-devel] [PATCH] Repair soft lockup with monitor mode of ath9k_htc card
From: Kalle Valo @ 2015-01-29 11:01 UTC (permalink / raw)
To: Oleksij Rempel
Cc: zhengyuwei, linux-kernel, ath9k-devel, linux-wireless,
ath9k-devel, netdev
In-Reply-To: <54CA10DB.3070903@rempel-privat.de>
Oleksij Rempel <linux@rempel-privat.de> writes:
> Am 29.01.2015 um 05:09 schrieb zhengyuwei@360.cn:
>> From: Yuwei Zheng <zhengyuwei@360.cn>
>>
>> In the environment with heavy wifi traffic, set the ar9271 into monitor mode, will
>> trigger a deadloop panic.
>>
>> The ath9k_hif_usb_rx_cb function excute on the interrupt context, and ath9k_rx_tasklet excute
>> on the soft irq context. In other words, the ath9k_hif_usb_rx_cb have more chance to excute than
>> ath9k_rx_tasklet. So in the worst condition, the rx.rxbuf receive list is always full,
>> and the do {}while(true) loop will not be break. The kernel get a soft lockup panic.
>>
>> [59011.007210] BUG: soft lockup - CPU#0 stuck for 23s!
>> [kworker/0:0:30609]
>> [59011.030560] BUG: scheduling while atomic: kworker/0:0/30609/0x40010100
>> [59013.804486] BUG: scheduling while atomic: kworker/0:0/30609/0x40010100
>> [59013.858522] Kernel panic - not syncing: softlockup: hung tasks
>>
>> [59014.038891] Exception stack(0xdf4bbc38 to 0xdf4bbc80)
>> [59014.046834] bc20: de57b950 60000113
>> [59014.059579] bc40: 00000000 bb32bb32 60000113 de57b948 de57b500 dc7bb440 df4bbcd0 00000000
>> [59014.072337] bc60: de57b950 60000113 df4bbcd0 df4bbc80 c04c259d c04c25a0 60000133 ffffffff
>> [59014.085233] [<c04c28db>] (__irq_svc+0x3b/0x5c) from [<c04c25a0>] (_raw_spin_unlock_irqrestore+0xc/0x10)
>> [59014.100437] [<c04c25a0>] (_raw_spin_unlock_irqrestore+0xc/0x10) from [<bf9c2089>] (ath9k_rx_tasklet+0x290/0x490 [ath9k_htc])
>> [59014.118267] [<bf9c2089>] (ath9k_rx_tasklet+0x290/0x490 [ath9k_htc]) from [<c0036d23>] (tasklet_action+0x3b/0x98)
>> [59014.134132] [<c0036d23>] (tasklet_action+0x3b/0x98) from [<c0036709>] (__do_softirq+0x99/0x16c)
>> [59014.147784] [<c0036709>] (__do_softirq+0x99/0x16c) from [<c00369f7>] (irq_exit+0x5b/0x5c)
>> [59014.160653] [<c00369f7>] (irq_exit+0x5b/0x5c) from [<c000cfc3>] (handle_IRQ+0x37/0x78)
>> [59014.173124] [<c000cfc3>] (handle_IRQ+0x37/0x78) from [<c00085df>] (omap3_intc_handle_irq+0x5f/0x68)
>> [59014.187225] [<c00085df>] (omap3_intc_handle_irq+0x5f/0x68) from [<c04c28db>](__irq_svc+0x3b/0x5c)
>>
>> This bug can be see with low performance board, such as uniprocessor beagle bone board.
>> Signed-off-by: Yuwei Zheng <zhengyuwei@360.cn>
>>
>> ---
>> drivers/net/wireless/ath/ath9k/hif_usb.c | 53 ++++++++++++++++++++++----
>> drivers/net/wireless/ath/ath9k/hif_usb.h | 5 +++
>> drivers/net/wireless/ath/ath9k/htc.h | 13 +++++++
>> drivers/net/wireless/ath/ath9k/htc_drv_debug.c | 49 ++++++++++++++++++++++++
>> drivers/net/wireless/ath/ath9k/htc_drv_txrx.c | 26 +++++++++++++
>> 5 files changed, 139 insertions(+), 7 deletions(-)
>
> First of all, thank you for you work! :D
>
> Please run ./scripts/checkpatch.pl yourpatch_path
> i get:
> total: 139 errors, 12 warnings, 2 checks, 231 lines checked
Also please add "ath9k_htc:" prefix to the title.
--
Kalle Valo
^ permalink raw reply
* [PATCH net-next] net: mark some potential candidates __read_mostly
From: Daniel Borkmann @ 2015-01-29 11:15 UTC (permalink / raw)
To: davem; +Cc: netdev
They are all either written once or extremly rarely (e.g. from init
code), so we can move them to the .data..read_mostly section.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
---
drivers/net/ipvlan/ipvlan_core.c | 2 +-
net/bridge/br_netlink.c | 2 +-
net/ipv4/devinet.c | 2 +-
net/ipv6/addrconf.c | 2 +-
net/mpls/mpls_gso.c | 4 ++--
5 files changed, 6 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ipvlan/ipvlan_core.c b/drivers/net/ipvlan/ipvlan_core.c
index 2e19528..2a17500 100644
--- a/drivers/net/ipvlan/ipvlan_core.c
+++ b/drivers/net/ipvlan/ipvlan_core.c
@@ -9,7 +9,7 @@
#include "ipvlan.h"
-static u32 ipvlan_jhash_secret;
+static u32 ipvlan_jhash_secret __read_mostly;
void ipvlan_init_secret(void)
{
diff --git a/net/bridge/br_netlink.c b/net/bridge/br_netlink.c
index 3875ea5..e08b260 100644
--- a/net/bridge/br_netlink.c
+++ b/net/bridge/br_netlink.c
@@ -701,7 +701,7 @@ static size_t br_get_link_af_size(const struct net_device *dev)
return pv->num_vlans * nla_total_size(sizeof(struct bridge_vlan_info));
}
-static struct rtnl_af_ops br_af_ops = {
+static struct rtnl_af_ops br_af_ops __read_mostly = {
.family = AF_BRIDGE,
.get_link_af_size = br_get_link_af_size,
};
diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index 59ebe16..f0b4a31 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -2322,7 +2322,7 @@ static __net_initdata struct pernet_operations devinet_ops = {
.exit = devinet_exit_net,
};
-static struct rtnl_af_ops inet_af_ops = {
+static struct rtnl_af_ops inet_af_ops __read_mostly = {
.family = AF_INET,
.fill_link_af = inet_fill_link_af,
.get_link_af_size = inet_get_link_af_size,
diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
index 7dcc065e..8623118 100644
--- a/net/ipv6/addrconf.c
+++ b/net/ipv6/addrconf.c
@@ -5405,7 +5405,7 @@ static struct pernet_operations addrconf_ops = {
.exit = addrconf_exit_net,
};
-static struct rtnl_af_ops inet6_ops = {
+static struct rtnl_af_ops inet6_ops __read_mostly = {
.family = AF_INET6,
.fill_link_af = inet6_fill_link_af,
.get_link_af_size = inet6_get_link_af_size,
diff --git a/net/mpls/mpls_gso.c b/net/mpls/mpls_gso.c
index 349295d..809df53 100644
--- a/net/mpls/mpls_gso.c
+++ b/net/mpls/mpls_gso.c
@@ -60,14 +60,14 @@ out:
return segs;
}
-static struct packet_offload mpls_mc_offload = {
+static struct packet_offload mpls_mc_offload __read_mostly = {
.type = cpu_to_be16(ETH_P_MPLS_MC),
.callbacks = {
.gso_segment = mpls_gso_segment,
},
};
-static struct packet_offload mpls_uc_offload = {
+static struct packet_offload mpls_uc_offload __read_mostly = {
.type = cpu_to_be16(ETH_P_MPLS_UC),
.callbacks = {
.gso_segment = mpls_gso_segment,
--
1.7.11.7
^ permalink raw reply related
* Re: [PATCH] net, ethernet, LLVMLinux: Add missing MODULE_DEVICE_TABLE()
From: Måns Rullgård @ 2015-01-29 11:29 UTC (permalink / raw)
To: Arnd Bergmann
Cc: David Miller, behanw, sathya.perla, ajit.khaparde, linux-kernel,
netdev, subbu.seetharaman
In-Reply-To: <3392784.h8cmHX9fyr@wuerfel>
Arnd Bergmann <arnd@arndb.de> writes:
> On Wednesday 28 January 2015 22:42:28 David Miller wrote:
>> From: Behan Webster <behanw@converseincode.com>
>> Date: Wed, 28 Jan 2015 17:36:14 -0800
>>
>> > Missing MODULE_DEVICE_TABLE for pci ids from benet driver found by clang.
>> >
>> > Signed-off-by: Behan Webster <behanw@converseincode.com>
>> > Suggested-by: Arnd Bergmann <arnd@arndb.de>
>>
>> Why are you removing the device table?
>
> Behan took a patch that I did earlier and split it up to add descriptions.
> The patch is correct, but he either misunderstood or misexpressed the
> intention.
>
> This driver has two identical lines that both say
>
> MODULE_DEVICE_TABLE(pci, be_dev_ids);
>
> I don't remember the exact symptom, but llvm/clang trips over this, while gcc
> silently ignores the second one.
GCC is probably being "helpful" and allowing multiple identical
definitions even though the C spec does not. Obviously all but one
should go away. No need to mention llvm, IMO.
--
Måns Rullgård
mans@mansr.com
^ permalink raw reply
* [PATCH net 0/4] Fix checksum error when using stacked vlan
From: Toshiaki Makita @ 2015-01-29 11:37 UTC (permalink / raw)
To: David S . Miller, Linux NICS; +Cc: e1000-devel, netdev, Toshiaki Makita
When I was testing 802.1ad, I found several drivers don't take into
account 802.1ad or multiple vlans when retrieving L3 (IP/IPv6) or
L4 (TCP/UDP) protocol for checksum offload.
It is mainly due to vlan_get_protocol(), which extracts ether type only
when it is tagged with single 802.1Q. When 802.1ad is used or there are
multiple vlans, it extracts vlan protocol and drivers cannot determine
which L3/L4 protocol is used.
Those drivers, most of which have IP_CSUM/IPV6_CSUM features, get L3/L4
header-offset by software, so it seems that their checksum offload works
with multiple vlans if we can parse protocols correctly.
(They know mac header length, and probably don't care about what is in it.)
And another thing, some of Intel's drivers seem to use skb->protocol where
vlan_get_protocol() is more suitable.
I tested that at least igb/igbvf on I350 works with this patch set.
Note:
We can hand a double tagged packet with CHECKSUM_PARTIAL to a HW driver
by creating a vlan device on a bridge device and enabling vlan_filtering
of the bridge with 802.1ad protocol.
Toshiaki Makita (4):
net: Fix vlan_get_protocol for stacked vlan
igbvf: Fix checksum error when using stacked vlan
ixgbe: Fix checksum error when using stacked vlan
ixgbevf: Fix checksum error when using stacked vlan
drivers/net/ethernet/intel/igbvf/netdev.c | 19 ++++---
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 2 +-
drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c | 4 +-
include/linux/if_vlan.h | 60 +++++++++++++++++------
net/core/dev.c | 31 +-----------
5 files changed, 61 insertions(+), 55 deletions(-)
--
1.8.1.2
------------------------------------------------------------------------------
Dive into the World of Parallel Programming. The Go Parallel Website,
sponsored by Intel and developed in partnership with Slashdot Media, is your
hub for all things parallel software development, from weekly thought
leadership blogs to news, videos, case studies, tutorials and more. Take a
look and join the conversation now. http://goparallel.sourceforge.net/
_______________________________________________
E1000-devel mailing list
E1000-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/e1000-devel
To learn more about Intel® Ethernet, visit http://communities.intel.com/community/wired
^ permalink raw reply
* [PATCH net 1/4] net: Fix vlan_get_protocol for stacked vlan
From: Toshiaki Makita @ 2015-01-29 11:37 UTC (permalink / raw)
To: David S . Miller, Linux NICS; +Cc: e1000-devel, netdev, Toshiaki Makita
In-Reply-To: <1422531430-32012-1-git-send-email-makita.toshiaki@lab.ntt.co.jp>
vlan_get_protocol() could not get network protocol if a skb has a 802.1ad
vlan tag or multiple vlans, which caused incorrect checksum calculation
in several drivers.
Fix vlan_get_protocol() to retrieve network protocol instead of incorrect
vlan protocol.
As the logic is the same as skb_network_protocol(), create a common helper
function __vlan_get_protocol() and call it from existing functions.
Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
---
include/linux/if_vlan.h | 60 +++++++++++++++++++++++++++++++++++++------------
net/core/dev.c | 31 +------------------------
2 files changed, 47 insertions(+), 44 deletions(-)
diff --git a/include/linux/if_vlan.h b/include/linux/if_vlan.h
index 515a35e..960e666 100644
--- a/include/linux/if_vlan.h
+++ b/include/linux/if_vlan.h
@@ -472,27 +472,59 @@ static inline int vlan_get_tag(const struct sk_buff *skb, u16 *vlan_tci)
/**
* vlan_get_protocol - get protocol EtherType.
* @skb: skbuff to query
+ * @type: first vlan protocol
+ * @depth: buffer to store length of eth and vlan tags in bytes
*
* Returns the EtherType of the packet, regardless of whether it is
* vlan encapsulated (normal or hardware accelerated) or not.
*/
-static inline __be16 vlan_get_protocol(const struct sk_buff *skb)
+static inline __be16 __vlan_get_protocol(struct sk_buff *skb, __be16 type,
+ int *depth)
{
- __be16 protocol = 0;
-
- if (vlan_tx_tag_present(skb) ||
- skb->protocol != cpu_to_be16(ETH_P_8021Q))
- protocol = skb->protocol;
- else {
- __be16 proto, *protop;
- protop = skb_header_pointer(skb, offsetof(struct vlan_ethhdr,
- h_vlan_encapsulated_proto),
- sizeof(proto), &proto);
- if (likely(protop))
- protocol = *protop;
+ unsigned int vlan_depth = skb->mac_len;
+
+ /* if type is 802.1Q/AD then the header should already be
+ * present at mac_len - VLAN_HLEN (if mac_len > 0), or at
+ * ETH_HLEN otherwise
+ */
+ if (type == htons(ETH_P_8021Q) || type == htons(ETH_P_8021AD)) {
+ if (vlan_depth) {
+ if (WARN_ON(vlan_depth < VLAN_HLEN))
+ return 0;
+ vlan_depth -= VLAN_HLEN;
+ } else {
+ vlan_depth = ETH_HLEN;
+ }
+ do {
+ struct vlan_hdr *vh;
+
+ if (unlikely(!pskb_may_pull(skb,
+ vlan_depth + VLAN_HLEN)))
+ return 0;
+
+ vh = (struct vlan_hdr *)(skb->data + vlan_depth);
+ type = vh->h_vlan_encapsulated_proto;
+ vlan_depth += VLAN_HLEN;
+ } while (type == htons(ETH_P_8021Q) ||
+ type == htons(ETH_P_8021AD));
}
- return protocol;
+ if (depth)
+ *depth = vlan_depth;
+
+ return type;
+}
+
+/**
+ * vlan_get_protocol - get protocol EtherType.
+ * @skb: skbuff to query
+ *
+ * Returns the EtherType of the packet, regardless of whether it is
+ * vlan encapsulated (normal or hardware accelerated) or not.
+ */
+static inline __be16 vlan_get_protocol(struct sk_buff *skb)
+{
+ return __vlan_get_protocol(skb, skb->protocol, NULL);
}
static inline void vlan_set_encap_proto(struct sk_buff *skb,
diff --git a/net/core/dev.c b/net/core/dev.c
index 171420e..c87a226 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -2352,7 +2352,6 @@ EXPORT_SYMBOL(skb_checksum_help);
__be16 skb_network_protocol(struct sk_buff *skb, int *depth)
{
- unsigned int vlan_depth = skb->mac_len;
__be16 type = skb->protocol;
/* Tunnel gso handlers can set protocol to ethernet. */
@@ -2366,35 +2365,7 @@ __be16 skb_network_protocol(struct sk_buff *skb, int *depth)
type = eth->h_proto;
}
- /* if skb->protocol is 802.1Q/AD then the header should already be
- * present at mac_len - VLAN_HLEN (if mac_len > 0), or at
- * ETH_HLEN otherwise
- */
- if (type == htons(ETH_P_8021Q) || type == htons(ETH_P_8021AD)) {
- if (vlan_depth) {
- if (WARN_ON(vlan_depth < VLAN_HLEN))
- return 0;
- vlan_depth -= VLAN_HLEN;
- } else {
- vlan_depth = ETH_HLEN;
- }
- do {
- struct vlan_hdr *vh;
-
- if (unlikely(!pskb_may_pull(skb,
- vlan_depth + VLAN_HLEN)))
- return 0;
-
- vh = (struct vlan_hdr *)(skb->data + vlan_depth);
- type = vh->h_vlan_encapsulated_proto;
- vlan_depth += VLAN_HLEN;
- } while (type == htons(ETH_P_8021Q) ||
- type == htons(ETH_P_8021AD));
- }
-
- *depth = vlan_depth;
-
- return type;
+ return __vlan_get_protocol(skb, type, depth);
}
/**
--
1.8.1.2
------------------------------------------------------------------------------
Dive into the World of Parallel Programming. The Go Parallel Website,
sponsored by Intel and developed in partnership with Slashdot Media, is your
hub for all things parallel software development, from weekly thought
leadership blogs to news, videos, case studies, tutorials and more. Take a
look and join the conversation now. http://goparallel.sourceforge.net/
_______________________________________________
E1000-devel mailing list
E1000-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/e1000-devel
To learn more about Intel® Ethernet, visit http://communities.intel.com/community/wired
^ permalink raw reply related
* [PATCH net 3/4] ixgbe: Fix checksum error when using stacked vlan
From: Toshiaki Makita @ 2015-01-29 11:37 UTC (permalink / raw)
To: David S . Miller, Linux NICS; +Cc: e1000-devel, netdev, Toshiaki Makita
In-Reply-To: <1422531430-32012-1-git-send-email-makita.toshiaki@lab.ntt.co.jp>
When a skb has multiple vlans and it is CHECKSUM_PARTIAL,
ixgbe_tx_csum() fails to get the network protocol and checksum related
descriptor fields are not configured correctly because skb->protocol
doesn't show the L3 protocol in this case.
Use vlan_get_protocol() to get the proper network protocol.
Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
---
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index 2ed2c7d..67b02bd 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -7227,11 +7227,11 @@ netdev_tx_t ixgbe_xmit_frame_ring(struct sk_buff *skb,
if (!vhdr)
goto out_drop;
- protocol = vhdr->h_vlan_encapsulated_proto;
tx_flags |= ntohs(vhdr->h_vlan_TCI) <<
IXGBE_TX_FLAGS_VLAN_SHIFT;
tx_flags |= IXGBE_TX_FLAGS_SW_VLAN;
}
+ protocol = vlan_get_protocol(skb);
if (unlikely(skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP) &&
adapter->ptp_clock &&
--
1.8.1.2
------------------------------------------------------------------------------
Dive into the World of Parallel Programming. The Go Parallel Website,
sponsored by Intel and developed in partnership with Slashdot Media, is your
hub for all things parallel software development, from weekly thought
leadership blogs to news, videos, case studies, tutorials and more. Take a
look and join the conversation now. http://goparallel.sourceforge.net/
_______________________________________________
E1000-devel mailing list
E1000-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/e1000-devel
To learn more about Intel® Ethernet, visit http://communities.intel.com/community/wired
^ permalink raw reply related
* [PATCH net 4/4] ixgbevf: Fix checksum error when using stacked vlan
From: Toshiaki Makita @ 2015-01-29 11:37 UTC (permalink / raw)
To: David S . Miller, Linux NICS; +Cc: e1000-devel, netdev, Toshiaki Makita
In-Reply-To: <1422531430-32012-1-git-send-email-makita.toshiaki@lab.ntt.co.jp>
When a skb has multiple vlans and it is CHECKSUM_PARTIAL,
ixgbevf_tx_csum() fails to get the network protocol and checksum related
descriptor fields are not configured correctly because skb->protocol
doesn't show the L3 protocol in this case.
Use first->protocol instead of skb->protocol to get the proper network
protocol.
Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
---
drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c b/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c
index 62a0d8e..38c7a0b 100644
--- a/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c
+++ b/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c
@@ -3099,7 +3099,7 @@ static int ixgbevf_tso(struct ixgbevf_ring *tx_ring,
/* ADV DTYP TUCMD MKRLOC/ISCSIHEDLEN */
type_tucmd = IXGBE_ADVTXD_TUCMD_L4T_TCP;
- if (skb->protocol == htons(ETH_P_IP)) {
+ if (first->protocol == htons(ETH_P_IP)) {
struct iphdr *iph = ip_hdr(skb);
iph->tot_len = 0;
iph->check = 0;
@@ -3156,7 +3156,7 @@ static void ixgbevf_tx_csum(struct ixgbevf_ring *tx_ring,
if (skb->ip_summed == CHECKSUM_PARTIAL) {
u8 l4_hdr = 0;
- switch (skb->protocol) {
+ switch (first->protocol) {
case htons(ETH_P_IP):
vlan_macip_lens |= skb_network_header_len(skb);
type_tucmd |= IXGBE_ADVTXD_TUCMD_IPV4;
--
1.8.1.2
------------------------------------------------------------------------------
Dive into the World of Parallel Programming. The Go Parallel Website,
sponsored by Intel and developed in partnership with Slashdot Media, is your
hub for all things parallel software development, from weekly thought
leadership blogs to news, videos, case studies, tutorials and more. Take a
look and join the conversation now. http://goparallel.sourceforge.net/
_______________________________________________
E1000-devel mailing list
E1000-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/e1000-devel
To learn more about Intel® Ethernet, visit http://communities.intel.com/community/wired
^ permalink raw reply related
* [PATCH net 2/4] igbvf: Fix checksum error when using stacked vlan
From: Toshiaki Makita @ 2015-01-29 11:37 UTC (permalink / raw)
To: David S . Miller, Linux NICS; +Cc: Toshiaki Makita, e1000-devel, netdev
In-Reply-To: <1422531430-32012-1-git-send-email-makita.toshiaki@lab.ntt.co.jp>
When a skb has multiple vlans and it is CHECKSUM_PARTIAL,
igbvf_tx_csum() fails to get the network protocol and checksum related
descriptor fields are not configured correctly because skb->protocol
doesn't show the L3 protocol in this case.
Use vlan_get_protocol() to get the proper network protocol.
Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
---
drivers/net/ethernet/intel/igbvf/netdev.c | 19 +++++++++++--------
1 file changed, 11 insertions(+), 8 deletions(-)
diff --git a/drivers/net/ethernet/intel/igbvf/netdev.c b/drivers/net/ethernet/intel/igbvf/netdev.c
index 63c807c..edea13b 100644
--- a/drivers/net/ethernet/intel/igbvf/netdev.c
+++ b/drivers/net/ethernet/intel/igbvf/netdev.c
@@ -1907,7 +1907,8 @@ static void igbvf_watchdog_task(struct work_struct *work)
static int igbvf_tso(struct igbvf_adapter *adapter,
struct igbvf_ring *tx_ring,
- struct sk_buff *skb, u32 tx_flags, u8 *hdr_len)
+ struct sk_buff *skb, u32 tx_flags, u8 *hdr_len,
+ __be16 protocol)
{
struct e1000_adv_tx_context_desc *context_desc;
struct igbvf_buffer *buffer_info;
@@ -1927,7 +1928,7 @@ static int igbvf_tso(struct igbvf_adapter *adapter,
l4len = tcp_hdrlen(skb);
*hdr_len += l4len;
- if (skb->protocol == htons(ETH_P_IP)) {
+ if (protocol == htons(ETH_P_IP)) {
struct iphdr *iph = ip_hdr(skb);
iph->tot_len = 0;
iph->check = 0;
@@ -1958,7 +1959,7 @@ static int igbvf_tso(struct igbvf_adapter *adapter,
/* ADV DTYP TUCMD MKRLOC/ISCSIHEDLEN */
tu_cmd |= (E1000_TXD_CMD_DEXT | E1000_ADVTXD_DTYP_CTXT);
- if (skb->protocol == htons(ETH_P_IP))
+ if (protocol == htons(ETH_P_IP))
tu_cmd |= E1000_ADVTXD_TUCMD_IPV4;
tu_cmd |= E1000_ADVTXD_TUCMD_L4T_TCP;
@@ -1984,7 +1985,8 @@ static int igbvf_tso(struct igbvf_adapter *adapter,
static inline bool igbvf_tx_csum(struct igbvf_adapter *adapter,
struct igbvf_ring *tx_ring,
- struct sk_buff *skb, u32 tx_flags)
+ struct sk_buff *skb, u32 tx_flags,
+ __be16 protocol)
{
struct e1000_adv_tx_context_desc *context_desc;
unsigned int i;
@@ -2011,7 +2013,7 @@ static inline bool igbvf_tx_csum(struct igbvf_adapter *adapter,
tu_cmd |= (E1000_TXD_CMD_DEXT | E1000_ADVTXD_DTYP_CTXT);
if (skb->ip_summed == CHECKSUM_PARTIAL) {
- switch (skb->protocol) {
+ switch (protocol) {
case htons(ETH_P_IP):
tu_cmd |= E1000_ADVTXD_TUCMD_IPV4;
if (ip_hdr(skb)->protocol == IPPROTO_TCP)
@@ -2211,6 +2213,7 @@ static netdev_tx_t igbvf_xmit_frame_ring_adv(struct sk_buff *skb,
u8 hdr_len = 0;
int count = 0;
int tso = 0;
+ __be16 protocol = vlan_get_protocol(skb);
if (test_bit(__IGBVF_DOWN, &adapter->state)) {
dev_kfree_skb_any(skb);
@@ -2239,13 +2242,13 @@ static netdev_tx_t igbvf_xmit_frame_ring_adv(struct sk_buff *skb,
tx_flags |= (vlan_tx_tag_get(skb) << IGBVF_TX_FLAGS_VLAN_SHIFT);
}
- if (skb->protocol == htons(ETH_P_IP))
+ if (protocol == htons(ETH_P_IP))
tx_flags |= IGBVF_TX_FLAGS_IPV4;
first = tx_ring->next_to_use;
tso = skb_is_gso(skb) ?
- igbvf_tso(adapter, tx_ring, skb, tx_flags, &hdr_len) : 0;
+ igbvf_tso(adapter, tx_ring, skb, tx_flags, &hdr_len, protocol) : 0;
if (unlikely(tso < 0)) {
dev_kfree_skb_any(skb);
return NETDEV_TX_OK;
@@ -2253,7 +2256,7 @@ static netdev_tx_t igbvf_xmit_frame_ring_adv(struct sk_buff *skb,
if (tso)
tx_flags |= IGBVF_TX_FLAGS_TSO;
- else if (igbvf_tx_csum(adapter, tx_ring, skb, tx_flags) &&
+ else if (igbvf_tx_csum(adapter, tx_ring, skb, tx_flags, protocol) &&
(skb->ip_summed == CHECKSUM_PARTIAL))
tx_flags |= IGBVF_TX_FLAGS_CSUM;
--
1.8.1.2
^ permalink raw reply related
* Throughput regression with `tcp: refine TSO autosizing`
From: Michal Kazior @ 2015-01-29 11:48 UTC (permalink / raw)
To: eric.dumazet; +Cc: linux-wireless, Network Development, eyalpe
Hi,
I'm not subscribed to netdev list and I can't find the message-id so I
can't reply directly to the original thread `BW regression after "tcp:
refine TSO autosizing"`.
I've noticed a big TCP performance drop with ath10k
(drivers/net/wireless/ath/ath10k) on 3.19-rc5. Instead of 500mbps I
get 250mbps in my testbed.
After bisecting I ended up at `tcp: refine TSO autosizing`. Reverting
`tcp: refine TSO autosizing` and `tcp: Do not apply TSO segment limit
to non-TSO packets` (for conflict free reverts) fixes the problem.
My testing setup is as follows:
a) ath10k AP, github.com/kvalo/ath/tree/master 3.19-rc5, w/ reverts
b) ath10k STA connected to (a), github.com/kvalo/ath/tree/master
3.19-rc5, w/ reverts
c) (b) w/o reverts
Devices are 3x3 (AP) and 2x2 (Client) and are RF cabled. 11ac@80MHz
2x2 has 866mbps modulation rate. In practice this should deliver
~700mbps of real UDP traffic.
Here are some numbers:
UDP: (b) -> (a): 672mbps
UDP: (a) -> (b): 687mbps
TCP: (b) -> (a): 526mbps
TCP: (a) -> (b): 500mbps
UDP: (c) -> (a): 669mbps*
UDP: (a) -> (c): 689mbps*
TCP: (c) -> (a): 240mbps**
TCP: (a) -> (c): 490mbps*
* no changes/within error margin
** the performance drop
I'm using iperf:
UDP: iperf -i1 -s -u vs iperf -i1 -c XX -u -B 200M -P5 -t 20
TCP: iperf -i1 -s vs iperf -i1 -c XX -P5 -t 20
Result values were obtained at the receiver side.
Iperf reports a few frames lost and out-of-order at each UDP test
start (during first second) but later has no packet loss and no
out-of-order. This shouldn't have any effect on a TCP session, right?
The device delivers batched up tx/rx completions (no way to change
that). I suppose this could be an issue for timing sensitive
algorithms. Also keep in mind 802.11n and 802.11ac devices have frame
aggregation windows so there's an inherent extra (and non-uniform)
latency when compared to, e.g. ethernet devices.
The driver doesn't have GRO. I have an internal patch which implements
it. It improves overall TCP traffic (more stable, up to 600mbps TCP
which is ~100mbps more than without GRO) but the TCP: (c) -> (a)
performance drop remains unaffected regardless.
I've tried applying stretch ACK patchset (v2) on both machines and
re-run the above tests. I got no measurable difference in performance.
I've also run these tests with iwlwifi 7260 (also a 2x2) as (b) and
(c). It didn't seem to be affected by the TSO patch at all (it runs at
~360mbps of TCP regardless of the TSO patch).
Any hints/ideas?
Michał
^ permalink raw reply
* RE: [PATCH] net: remove sock_iocb
From: Jon Maloy @ 2015-01-29 12:13 UTC (permalink / raw)
To: David Miller, viro@ZenIV.linux.org.uk; +Cc: hch@lst.de, netdev@vger.kernel.org
In-Reply-To: <20150129.000125.1532086382063848714.davem@davemloft.net>
> -----Original Message-----
> From: netdev-owner@vger.kernel.org [mailto:netdev-
> owner@vger.kernel.org] On Behalf Of David Miller
> Sent: January-29-15 3:01 AM
> To: viro@ZenIV.linux.org.uk
> Cc: hch@lst.de; netdev@vger.kernel.org
> Subject: Re: [PATCH] net: remove sock_iocb
>
> From: Al Viro <viro@ZenIV.linux.org.uk>
> Date: Thu, 29 Jan 2015 07:57:21 +0000
>
> > You know, that's getting _really_ interesting. The thing is, now
> > there's only one ->sendmsg() instance using iocb argument at all, and
> > it's a really weird one. TIPC. Which only compares it with NULL, and
> > that - to tell the normal calls (== done by sock_sendmsg()
> > et.al.) from tipc_{accept,connect}()-generated ones. And the way it's
> > used is
> > if (iocb)
> > lock_sock(sk);
> > in tipc_send_stream(). IOW, "tipc_accept() and tipc_connect() would
> > like to use the guts of tipc_send_stream(), but they are already
> > holding the socket locked; let's just pass NULL iocb (which
> > net/socket.c never does) to tell it to leave the fucking lock alone, thank you
> very much".
> >
> > And no ->recvmsg() are using iocb at all now. How about we take the
> > guts of tipc_send_stream() into a helper function and have
> > tipc_accept/connect use _that_?
Sounds reasonable. We'll look into this.
///jon
Then we could drop iocb argument
> > completely and for ->sendmsg() it would be the difference between 4
> > and 3 arguments, which has interesting effects on certain register-starved
> architectures...
> >
> > While we are at it, size (both for sendmsg and recvmsg) is always
> > equal to iov_iter_count(&msg->msg_iter), so that's not the only
> > redundant argument there...
>
> No objections from me.
> --
> To unsubscribe from this list: send the line "unsubscribe netdev" in the body
> of a message to majordomo@vger.kernel.org More majordomo info at
> http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox