Skip to contentSkip to conversation

mptcp: Remove request_version field from subflow_request_sock

PATCH In review Peter Krystad 2 participants 5 messages Dec 12, 2019

Subsystems (2)
lore
Share & tools
lore

Peter Krystad

author
#0 lore raw

PATCH 1/1 mptcp: Remove request_version field from subflow_request_sock

Since only a single version is supported this fields presence is misleading, it dates from when it was thought the version would be configurable. The actual requested version is hard-coded in mptcp_write_options().

squashto: Associate MPTCP context with TCP socket

Signed-off-by Peter Krystad <peter.krystad(a)linux.intel.com>
---
 net/mptcp/protocol.c | 1 -
 net/mptcp/protocol.h | 3 +--
 2 files changed, 1 insertion(+), 3 deletions(-)
net/mptcp/protocol.c +0 -1
index 502c464f3d69..bbd6c58154fd 100644
@@ -123,7 +123,6 @@ static struct socket *mptcp_socket_create_get(struct mptcp_sock *msk)
 	msk->subflow = ssock;
 	subflow = mptcp_subflow_ctx(msk->subflow->sk);
 	subflow->request_mptcp = 1; /* @@ if MPTCP enabled */
-	subflow->request_version = 0; /* currently only v0 supported */
 
 	sock_hold(ssock->sk);
 
net/mptcp/protocol.h +1 -2
index ddb3a7581d41..93d6c41b76d5 100644
@@ -48,8 +48,7 @@ static inline struct mptcp_sock *mptcp_sk(const struct sock *sk)
 
 /* MPTCP subflow context */
 struct mptcp_subflow_context {
-	u32	request_mptcp : 1,  /* send MP_CAPABLE */
-		request_version : 4;
+	u32	request_mptcp : 1;  /* send MP_CAPABLE */
 	struct	sock *tcp_sock;	    /* tcp sk backpointer */
 	struct	sock *conn;	    /* parent mptcp_sock */
 	struct	rcu_head rcu;
Signature · 1 line
2.17.2
#1 lore raw

Hi Peter, Paolo,

Quoted 6 lines · Peter Krystad

Since only a single version is supported this fields presence is misleading, it dates from when it was thought the version would be configurable. The actual requested version is hard-coded in mptcp_write_options().

squashto: Associate MPTCP context with TCP socket

Thank you for the patch and the review!

I just have a question below:

[...]

net/mptcp/protocol.h static inline struct mptcp_sock *mptcp_sk(const struct sock *sk) · line 51

Quoted 11 lines
net/mptcp/protocol.h +1 -2
index ddb3a7581d41..93d6c41b76d5 100644
@@ -48,8 +48,7 @@ static inline struct mptcp_sock *mptcp_sk(const struct sock *sk)
 
  /* MPTCP subflow context */
  struct mptcp_subflow_context {
-	u32	request_mptcp : 1,  /* send MP_CAPABLE */
-		request_version : 4;
+	u32	request_mptcp : 1;  /* send MP_CAPABLE */

When removing these 4 bits, at the end of the series, we now only use 13 bits. Should we switch to u16? I didn't check with pahole if it would really reduce the size of the structure, I don't think so and maybe we don't really need to reduce this bit field but I prefer to ask :)

Cheers,
Matt

Signature · 5 lines
Matthieu Baerts | R&D Engineer
matthieu.baerts(a)tessares.net
Tessares SA | Hybrid Access Solutions
www.tessares.net
1 Avenue Jean Monnet, 1348 Louvain-la-Neuve, Belgium
#2 lore raw

Hi Peter, Paolo

Quoted 11 lines · Matthieu Baerts

Hi Peter, Paolo,

Quoted 6 lines · Peter Krystad

Since only a single version is supported this fields presence is misleading, it dates from when it was thought the version would be configurable. The actual requested version is hard-coded in mptcp_write_options().

squashto: Associate MPTCP context with TCP socket

Thank you for the patch and the review!

- a63d1cdb5903: "squashed" in "mptcp: Associate MPTCP context with TCP socket"
- 053bccb0ca21: conflict in t/mptcp-Handle-MP_CAPABLE-options-for-outgoing-connections
- b9a183bde93e: conflict in t/mptcp-parse-and-emit-MP_CAPABLE-option-according-to-v1-spec
- e081b82c8850: conflict in t/mptcp-Add-handling-of-outgoing-MP_JOIN-requests
- d2dba972de89..55f1d621d32b: result

Cheers,
Matt

Signature · 5 lines
Matthieu Baerts | R&D Engineer
matthieu.baerts(a)tessares.net
Tessares SA | Hybrid Access Solutions
www.tessares.net
1 Avenue Jean Monnet, 1348 Louvain-la-Neuve, Belgium
author
#3 lore raw
In reply to Matthieu Baerts

Hi Matt -

Quoted 22 lines · Matthieu Baerts

Hi Peter, Paolo,

Quoted 6 lines · Peter Krystad

Since only a single version is supported this fields presence is misleading, it dates from when it was thought the version would be configurable. The actual requested version is hard-coded in mptcp_write_options().

squashto: Associate MPTCP context with TCP socket

Thank you for the patch and the review!

I just have a question below:

[...]

Quoted 6 lines
net/mptcp/protocol.h +0 -0
index ddb3a7581d41..93d6c41b76d5 100644
@@ -48,8 +48,7 @@ static inline struct mptcp_sock *mptcp_sk(const struct sock *sk)
 
Quoted 10 lines

net/mptcp/protocol.h static inline struct mptcp_sock *mptcp_sk(const struct sock *sk) · line 51

Quoted 5 lines
net/mptcp/protocol.h +1 -2
  /* MPTCP subflow context */
  struct mptcp_subflow_context {
-	u32	request_mptcp : 1,  /* send MP_CAPABLE */
-		request_version : 4;
+	u32	request_mptcp : 1;  /* send MP_CAPABLE */

When removing these 4 bits, at the end of the series, we now only use 13 bits. Should we switch to u16? I didn't check with pahole if it would really reduce the size of the structure, I don't think so and maybe we don't really need to reduce this bit field but I prefer to ask :)

We could go back to u16, but then I would want to move it to the end of the struct (or just before local_id), but that would create a lot of work resolving conflicts as the fields are introduced one by one. Size isn't as critical for subflow_context because it doesn't affect every tcp_sock, only MPTCP sockets.

Peter.

Quoted 2 lines

Cheers,
Matt

#4 lore raw

Hi Peter,

Quoted 7 lines · Peter Krystad

Hi Matt -

Quoted 3 lines · Matthieu Baerts

Hi Peter, Paolo,

On 12/12/2019 05:54, Peter Krystad wrote:

[...]

Quoted 18 lines
Quoted 7 lines
Quoted 7 lines
net/mptcp/protocol.h +0 -0
index ddb3a7581d41..93d6c41b76d5 100644
@@ -48,8 +48,7 @@ static inline struct mptcp_sock *mptcp_sk(const struct sock *sk)
 
Quoted 10 lines

net/mptcp/protocol.h static inline struct mptcp_sock *mptcp_sk(const struct sock *sk) · line 51

Quoted 5 lines
net/mptcp/protocol.h +1 -2
   /* MPTCP subflow context */
   struct mptcp_subflow_context {
-	u32	request_mptcp : 1,  /* send MP_CAPABLE */
-		request_version : 4;
+	u32	request_mptcp : 1;  /* send MP_CAPABLE */

When removing these 4 bits, at the end of the series, we now only use 13 bits. Should we switch to u16? I didn't check with pahole if it would really reduce the size of the structure, I don't think so and maybe we don't really need to reduce this bit field but I prefer to ask :)

Quoted 5 lines

We could go back to u16, but then I would want to move it to the end of the struct (or just before local_id), but that would create a lot of work resolving conflicts as the fields are introduced one by one. Size isn't as critical for subflow_context because it doesn't affect every tcp_sock, only MPTCP sockets.

If we think it is important to do that, I can do it, conflicts would be easy to resolve :)

But if in any cases, we don't reduce the full size of the structure, no reason to change!

Cheers,
Matt

Signature · 5 lines
Matthieu Baerts | R&D Engineer
matthieu.baerts(a)tessares.net
Tessares SA | Hybrid Access Solutions
www.tessares.net
1 Avenue Jean Monnet, 1348 Louvain-la-Neuve, Belgium