linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH v2 net-next 0/2] net: tipc: fix FB_MTU eat two pages and do some code cleanup
@ 2021-06-09 10:32 menglong8.dong
  2021-06-09 10:32 ` [PATCH v2 net-next 1/2] net: tipc: fix FB_MTU eat two pages menglong8.dong
                   ` (2 more replies)
  0 siblings, 3 replies; 5+ messages in thread
From: menglong8.dong @ 2021-06-09 10:32 UTC (permalink / raw)
  To: jmaloy
  Cc: ying.xue, davem, kuba, netdev, tipc-discussion, linux-kernel,
	Menglong Dong

From: Menglong Dong <dong.menglong@zte.com.cn>

In the first patch, FB_MTU is redefined to make sure data size will not
exceed PAGE_SIZE. Besides, I removed the alignment for buf_size in
tipc_buf_acquire, because skb_alloc_fclone will do the alignment job.

In the second patch, I removed align() in msg.c and replace it with
ALIGN().




Menglong Dong (2):
  net: tipc: fix FB_MTU eat two pages
  net: tipc: replace align() with ALIGN in msg.c

 net/tipc/bcast.c |  2 +-
 net/tipc/msg.c   | 31 ++++++++++++++-----------------
 net/tipc/msg.h   |  3 ++-
 3 files changed, 17 insertions(+), 19 deletions(-)

-- 
2.32.0


^ permalink raw reply	[flat|nested] 5+ messages in thread

* [PATCH v2 net-next 1/2] net: tipc: fix FB_MTU eat two pages
  2021-06-09 10:32 [PATCH v2 net-next 0/2] net: tipc: fix FB_MTU eat two pages and do some code cleanup menglong8.dong
@ 2021-06-09 10:32 ` menglong8.dong
  2021-06-09 10:32 ` [PATCH v2 net-next 2/2] net: tipc: replace align() with ALIGN in msg.c menglong8.dong
  2021-06-09 10:47 ` [PATCH v2 net-next 0/2] net: tipc: fix FB_MTU eat two pages and do some code cleanup Jon Maloy
  2 siblings, 0 replies; 5+ messages in thread
From: menglong8.dong @ 2021-06-09 10:32 UTC (permalink / raw)
  To: jmaloy
  Cc: ying.xue, davem, kuba, netdev, tipc-discussion, linux-kernel,
	Menglong Dong, Zeal Robot

From: Menglong Dong <dong.menglong@zte.com.cn>

FB_MTU is used in 'tipc_msg_build()' to alloc smaller skb when memory
allocation fails, which can avoid unnecessary sending failures.

The value of FB_MTU now is 3744, and the data size will be:

  (3744 + SKB_DATA_ALIGN(sizeof(struct skb_shared_info)) + \
    SKB_DATA_ALIGN(BUF_HEADROOM + BUF_TAILROOM + 3))

which is larger than one page(4096), and two pages will be allocated.

To avoid it, replace '3744' with a calculation:

FB_MTU = (PAGE_SIZE - SKB_DATA_ALIGN(sizeof(struct skb_shared_info))
          - SKB_DATA_ALIGN(BUF_HEADROOM + BUF_TAILROOM))

What's more, alloc_skb_fclone() will call SKB_DATA_ALIGN for data size,
and it's not unnecessary to make alignment for buf_size in
tipc_buf_acquire(). So, just remove it.

Fixes: 4c94cc2d3d57 ("tipc: fall back to smaller MTU if allocation of local send skb fails")

Reported-by: Zeal Robot <zealci@zte.com.cn>
Signed-off-by: Menglong Dong <dong.menglong@zte.com.cn>
---
V2:
- define FB_MTU in msg.c instead of introduce a new file
- remove align for buf_size in tipc_buf_acquire()
---
 net/tipc/bcast.c |  2 +-
 net/tipc/msg.c   | 15 ++++++++-------
 net/tipc/msg.h   |  3 ++-
 3 files changed, 11 insertions(+), 9 deletions(-)

diff --git a/net/tipc/bcast.c b/net/tipc/bcast.c
index d4beca895992..9daace9542f4 100644
--- a/net/tipc/bcast.c
+++ b/net/tipc/bcast.c
@@ -699,7 +699,7 @@ int tipc_bcast_init(struct net *net)
 	spin_lock_init(&tipc_net(net)->bclock);
 
 	if (!tipc_link_bc_create(net, 0, 0, NULL,
-				 FB_MTU,
+				 fb_mtu,
 				 BCLINK_WIN_DEFAULT,
 				 BCLINK_WIN_DEFAULT,
 				 0,
diff --git a/net/tipc/msg.c b/net/tipc/msg.c
index ce6ab54822d8..a5c030ca7065 100644
--- a/net/tipc/msg.c
+++ b/net/tipc/msg.c
@@ -47,8 +47,14 @@
 #define BUF_TAILROOM (TIPC_AES_GCM_TAG_SIZE)
 #else
 #define BUF_HEADROOM (LL_MAX_HEADER + 48)
-#define BUF_TAILROOM 16
+#define BUF_TAILROOM 0
 #endif
+#define FB_MTU (PAGE_SIZE - \
+		SKB_DATA_ALIGN(sizeof(struct skb_shared_info)) - \
+		SKB_DATA_ALIGN(BUF_HEADROOM + BUF_TAILROOM) \
+		)
+
+const int fb_mtu = FB_MTU;
 
 static unsigned int align(unsigned int i)
 {
@@ -69,13 +75,8 @@ static unsigned int align(unsigned int i)
 struct sk_buff *tipc_buf_acquire(u32 size, gfp_t gfp)
 {
 	struct sk_buff *skb;
-#ifdef CONFIG_TIPC_CRYPTO
-	unsigned int buf_size = (BUF_HEADROOM + size + BUF_TAILROOM + 3) & ~3u;
-#else
-	unsigned int buf_size = (BUF_HEADROOM + size + 3) & ~3u;
-#endif
 
-	skb = alloc_skb_fclone(buf_size, gfp);
+	skb = alloc_skb_fclone(BUF_HEADROOM + size + BUF_TAILROOM, gfp);
 	if (skb) {
 		skb_reserve(skb, BUF_HEADROOM);
 		skb_put(skb, size);
diff --git a/net/tipc/msg.h b/net/tipc/msg.h
index 5d64596ba987..2c214691037c 100644
--- a/net/tipc/msg.h
+++ b/net/tipc/msg.h
@@ -99,9 +99,10 @@ struct plist;
 #define MAX_H_SIZE                60	/* Largest possible TIPC header size */
 
 #define MAX_MSG_SIZE (MAX_H_SIZE + TIPC_MAX_USER_MSG_SIZE)
-#define FB_MTU                  3744
 #define TIPC_MEDIA_INFO_OFFSET	5
 
+extern const int fb_mtu;
+
 struct tipc_skb_cb {
 	union {
 		struct {
-- 
2.32.0


^ permalink raw reply related	[flat|nested] 5+ messages in thread

* [PATCH v2 net-next 2/2] net: tipc: replace align() with ALIGN in msg.c
  2021-06-09 10:32 [PATCH v2 net-next 0/2] net: tipc: fix FB_MTU eat two pages and do some code cleanup menglong8.dong
  2021-06-09 10:32 ` [PATCH v2 net-next 1/2] net: tipc: fix FB_MTU eat two pages menglong8.dong
@ 2021-06-09 10:32 ` menglong8.dong
  2021-06-09 10:47 ` [PATCH v2 net-next 0/2] net: tipc: fix FB_MTU eat two pages and do some code cleanup Jon Maloy
  2 siblings, 0 replies; 5+ messages in thread
From: menglong8.dong @ 2021-06-09 10:32 UTC (permalink / raw)
  To: jmaloy
  Cc: ying.xue, davem, kuba, netdev, tipc-discussion, linux-kernel,
	Menglong Dong

From: Menglong Dong <dong.menglong@zte.com.cn>

The function align() which is defined in msg.c is redundant, replace it
with ALIGN() and introduce a BUF_ALIGN().

Signed-off-by: Menglong Dong <dong.menglong@zte.com.cn>
---
 net/tipc/msg.c | 18 +++++++-----------
 1 file changed, 7 insertions(+), 11 deletions(-)

diff --git a/net/tipc/msg.c b/net/tipc/msg.c
index a5c030ca7065..19a7d8eb8abb 100644
--- a/net/tipc/msg.c
+++ b/net/tipc/msg.c
@@ -49,6 +49,7 @@
 #define BUF_HEADROOM (LL_MAX_HEADER + 48)
 #define BUF_TAILROOM 0
 #endif
+#define BUF_ALIGN(x) ALIGN(x, 4)
 #define FB_MTU (PAGE_SIZE - \
 		SKB_DATA_ALIGN(sizeof(struct skb_shared_info)) - \
 		SKB_DATA_ALIGN(BUF_HEADROOM + BUF_TAILROOM) \
@@ -56,11 +57,6 @@
 
 const int fb_mtu = FB_MTU;
 
-static unsigned int align(unsigned int i)
-{
-	return (i + 3) & ~3u;
-}
-
 /**
  * tipc_buf_acquire - creates a TIPC message buffer
  * @size: message size (including TIPC header)
@@ -489,10 +485,10 @@ static bool tipc_msg_bundle(struct sk_buff *bskb, struct tipc_msg *msg,
 	struct tipc_msg *bmsg = buf_msg(bskb);
 	u32 msz, bsz, offset, pad;
 
-	msz = msg_size(msg);
-	bsz = msg_size(bmsg);
-	offset = align(bsz);
-	pad = offset - bsz;
+	msz	= msg_size(msg);
+	bsz	= msg_size(bmsg);
+	offset	= BUF_ALIGN(bsz);
+	pad	= offset - bsz;
 
 	if (unlikely(skb_tailroom(bskb) < (pad + msz)))
 		return false;
@@ -548,7 +544,7 @@ bool tipc_msg_try_bundle(struct sk_buff *tskb, struct sk_buff **skb, u32 mss,
 
 	/* Make a new bundle of the two messages if possible */
 	tsz = msg_size(buf_msg(tskb));
-	if (unlikely(mss < align(INT_H_SIZE + tsz) + msg_size(msg)))
+	if (unlikely(mss < BUF_ALIGN(INT_H_SIZE + tsz) + msg_size(msg)))
 		return true;
 	if (unlikely(pskb_expand_head(tskb, INT_H_SIZE, mss - tsz - INT_H_SIZE,
 				      GFP_ATOMIC)))
@@ -607,7 +603,7 @@ bool tipc_msg_extract(struct sk_buff *skb, struct sk_buff **iskb, int *pos)
 	if (unlikely(!tipc_msg_validate(iskb)))
 		goto none;
 
-	*pos += align(imsz);
+	*pos += BUF_ALIGN(imsz);
 	return true;
 none:
 	kfree_skb(skb);
-- 
2.32.0


^ permalink raw reply related	[flat|nested] 5+ messages in thread

* Re: [PATCH v2 net-next 0/2] net: tipc: fix FB_MTU eat two pages and do some code cleanup
  2021-06-09 10:32 [PATCH v2 net-next 0/2] net: tipc: fix FB_MTU eat two pages and do some code cleanup menglong8.dong
  2021-06-09 10:32 ` [PATCH v2 net-next 1/2] net: tipc: fix FB_MTU eat two pages menglong8.dong
  2021-06-09 10:32 ` [PATCH v2 net-next 2/2] net: tipc: replace align() with ALIGN in msg.c menglong8.dong
@ 2021-06-09 10:47 ` Jon Maloy
  2021-06-09 12:56   ` Menglong Dong
  2 siblings, 1 reply; 5+ messages in thread
From: Jon Maloy @ 2021-06-09 10:47 UTC (permalink / raw)
  To: menglong8.dong
  Cc: ying.xue, davem, kuba, netdev, tipc-discussion, linux-kernel,
	Menglong Dong



On 6/9/21 6:32 AM, menglong8.dong@gmail.com wrote:
> From: Menglong Dong <dong.menglong@zte.com.cn>
>
> In the first patch, FB_MTU is redefined to make sure data size will not
> exceed PAGE_SIZE. Besides, I removed the alignment for buf_size in
> tipc_buf_acquire, because skb_alloc_fclone will do the alignment job.
>
> In the second patch, I removed align() in msg.c and replace it with
> ALIGN().
>
>
>
>
> Menglong Dong (2):
>    net: tipc: fix FB_MTU eat two pages
>    net: tipc: replace align() with ALIGN in msg.c
>
>   net/tipc/bcast.c |  2 +-
>   net/tipc/msg.c   | 31 ++++++++++++++-----------------
>   net/tipc/msg.h   |  3 ++-
>   3 files changed, 17 insertions(+), 19 deletions(-)
>
NACK.
You must have missed my last mail before you sent out this.  We have to 
define a separate macro for bcast.c, since those buffers sometimes will 
need encryption.
Sorry for the confusion.
///jon


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH v2 net-next 0/2] net: tipc: fix FB_MTU eat two pages and do some code cleanup
  2021-06-09 10:47 ` [PATCH v2 net-next 0/2] net: tipc: fix FB_MTU eat two pages and do some code cleanup Jon Maloy
@ 2021-06-09 12:56   ` Menglong Dong
  0 siblings, 0 replies; 5+ messages in thread
From: Menglong Dong @ 2021-06-09 12:56 UTC (permalink / raw)
  To: Jon Maloy
  Cc: ying.xue, David Miller, Jakub Kicinski, netdev, tipc-discussion,
	LKML, Menglong Dong

On Wed, Jun 9, 2021 at 6:47 PM Jon Maloy <jmaloy@redhat.com> wrote:
>
>
>
> On 6/9/21 6:32 AM, menglong8.dong@gmail.com wrote:
> > From: Menglong Dong <dong.menglong@zte.com.cn>
> >
> > In the first patch, FB_MTU is redefined to make sure data size will not
> > exceed PAGE_SIZE. Besides, I removed the alignment for buf_size in
> > tipc_buf_acquire, because skb_alloc_fclone will do the alignment job.
> >
> > In the second patch, I removed align() in msg.c and replace it with
> > ALIGN().
> >
> >
> >
> >
> > Menglong Dong (2):
> >    net: tipc: fix FB_MTU eat two pages
> >    net: tipc: replace align() with ALIGN in msg.c
> >
> >   net/tipc/bcast.c |  2 +-
> >   net/tipc/msg.c   | 31 ++++++++++++++-----------------
> >   net/tipc/msg.h   |  3 ++-
> >   3 files changed, 17 insertions(+), 19 deletions(-)
> >
> NACK.
> You must have missed my last mail before you sent out this.  We have to
> define a separate macro for bcast.c, since those buffers sometimes will
> need encryption.
> Sorry for the confusion.

No, no, I didn't miss your mail. I think it can make us clear about what and how
to do by sending the V2 patches.

So we can define two versions 'FB_MTU' for bcast.c and msg.c, such as CRYPTO_MTU
and NON_CRYPTO_MTU. And within tipc_buf_acquire(), we decide which version
BUF_HEADROOM to use by the data size? Such as:

int buf_size;
if (IS_ENABLED(CONFIG_TIPC_CRYPTO) && size == NON_CRYPTO_MTU) {
    buf_size = size + BUF_HEADROOM_non-crypto + BUF_TAILROOM_non-crypto;
} else {
    buf_size = size + BUF_HEADROOM_crypto + BUF_TAILROOM_crypto;
}

Is this feeling?
(It's a little weird to check whether the data should be crypto by data size).

Thanks!
Menglong Dong

^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2021-06-09 12:57 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-06-09 10:32 [PATCH v2 net-next 0/2] net: tipc: fix FB_MTU eat two pages and do some code cleanup menglong8.dong
2021-06-09 10:32 ` [PATCH v2 net-next 1/2] net: tipc: fix FB_MTU eat two pages menglong8.dong
2021-06-09 10:32 ` [PATCH v2 net-next 2/2] net: tipc: replace align() with ALIGN in msg.c menglong8.dong
2021-06-09 10:47 ` [PATCH v2 net-next 0/2] net: tipc: fix FB_MTU eat two pages and do some code cleanup Jon Maloy
2021-06-09 12:56   ` Menglong Dong

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).