RE: linux-next: Tree for December 29 (cxgb3i)

From: Karen Xie
Date: Tue Dec 30 2008 - 00:56:41 EST



Hi, Randy & James,

The cxgb3i driver was using the skb->sp for some internal stuff. I have just submitted a patch to remove the usage of it since it is not really related to the secure path.

The patch is submitted to the list (http://marc.info/?l=linux-scsi&m=123061555414193&w=2) and is also attached here.

Sorry for the late response, I am traveling this week.

Thanks,
Karen

-----Original Message-----
From: Randy Dunlap [mailto:randy.dunlap@xxxxxxxxxx]
Sent: Mon 12/29/2008 2:10 PM
To: James Bottomley
Cc: Stephen Rothwell; scsi; linux-next@xxxxxxxxxxxxxxx; LKML; Karen Xie
Subject: Re: linux-next: Tree for December 29 (cxgb3i)

James Bottomley wrote:
> On Mon, 2008-12-29 at 12:35 -0800, Randy Dunlap wrote:
>> On Tue, 30 Dec 2008 03:16:21 +1100 Stephen Rothwell wrote:
>>
>>> Hi all,
>>>
>>> Changes since 20081219:
>>>
>>> Undropped tree:
>>> scci
>>> mtd
>>>
>>> Dropped trees (temporarily):
>>> nfs (akpm request due to 2.6.30 features)
>>> kvm (build problem)
>>> rr (build poblem)
>>> semaphore-removal (due to unfixed conflicts against Linus' tree)
>>> cpu_alloc (build problem)
>>> audit (difficult conflicts)
>>>
>>> Linus' tree had three build failures requiring patches and one requiring
>>> a revert.
>>
>> linux-next-20081229/drivers/scsi/cxgb3i/cxgb3i_offload.c:499: error: 'struct sk_buff' has no member named 'sp'
>> linux-next-20081229/drivers/scsi/cxgb3i/cxgb3i_offload.c:512: error: 'struct sk_buff' has no member named 'sp'
>> linux-next-20081229/drivers/scsi/cxgb3i/cxgb3i_offload.c:532: error: 'struct sk_buff' has no member named 'sp'
>> linux-next-20081229/drivers/scsi/cxgb3i/cxgb3i_offload.c:533: error: 'struct sk_buff' has no member named 'sp'
>
> In the config 20 questions, my guess for this is CONFIG_XFRM=n

That's correct. config file is now attached.

> I'm not at all sure why this driver is playing with the secure path ...
> I suspect the use needs to be enclosed in #ifdef CONFIG_XFRM pairs, but
> I'd like the maintainers to verify.

--
andy

[PATCH 1/1] cxgb3i - remove use of skb->sp

From: Karen Xie <kxie@xxxxxxxxxxx>

The cxgb3i was using skb->sp pointer for some internal book-keeping which is not related to the secure path. Changed it to use skb->cb[] instead.

Signed-off-by: Karen Xie <kxie@xxxxxxxxxxx>
---

drivers/scsi/cxgb3i/cxgb3i_offload.c | 8 ++++----
drivers/scsi/cxgb3i/cxgb3i_offload.h | 6 +++---
2 files changed, 7 insertions(+), 7 deletions(-)


diff --git a/drivers/scsi/cxgb3i/cxgb3i_offload.c b/drivers/scsi/cxgb3i/cxgb3i_offload.c
index 5f16081..a865f1f 100644
--- a/drivers/scsi/cxgb3i/cxgb3i_offload.c
+++ b/drivers/scsi/cxgb3i/cxgb3i_offload.c
@@ -496,7 +496,7 @@ static inline void reset_wr_list(struct s3_conn *c3cn)
static inline void enqueue_wr(struct s3_conn *c3cn,
struct sk_buff *skb)
{
- skb->sp = NULL;
+ skb_wr_data(skb) = NULL;

/*
* We want to take an extra reference since both us and the driver
@@ -509,7 +509,7 @@ static inline void enqueue_wr(struct s3_conn *c3cn,
if (!c3cn->wr_pending_head)
c3cn->wr_pending_head = skb;
else
- c3cn->wr_pending_tail->sp = (void *)skb;
+ skb_wr_data(skb) = skb;
c3cn->wr_pending_tail = skb;
}

@@ -529,8 +529,8 @@ static inline struct sk_buff *dequeue_wr(struct s3_conn *c3cn)

if (likely(skb)) {
/* Don't bother clearing the tail */
- c3cn->wr_pending_head = (struct sk_buff *)skb->sp;
- skb->sp = NULL;
+ c3cn->wr_pending_head = skb_wr_data(skb);
+ skb_wr_data(skb) = NULL;
}
return skb;
}
diff --git a/drivers/scsi/cxgb3i/cxgb3i_offload.h b/drivers/scsi/cxgb3i/cxgb3i_offload.h
index 5b93d62..d231569 100644
--- a/drivers/scsi/cxgb3i/cxgb3i_offload.h
+++ b/drivers/scsi/cxgb3i/cxgb3i_offload.h
@@ -180,7 +180,7 @@ void cxgb3i_c3cn_release(struct s3_conn *);
* @seq: tcp sequence number
* @ddigest: pdu data digest
* @pdulen: recovered pdu length
- * @ulp_data: scratch area for ULP
+ * @wr_data: scratch area for tx wr
*/
struct cxgb3_skb_cb {
__u8 flags;
@@ -188,7 +188,7 @@ struct cxgb3_skb_cb {
__u32 seq;
__u32 ddigest;
__u32 pdulen;
- __u8 ulp_data[16];
+ struct sk_buff *wr_data;
};

#define CXGB3_SKB_CB(skb) ((struct cxgb3_skb_cb *)&((skb)->cb[0]))
@@ -196,7 +196,7 @@ struct cxgb3_skb_cb {
#define skb_ulp_mode(skb) (CXGB3_SKB_CB(skb)->ulp_mode)
#define skb_ulp_ddigest(skb) (CXGB3_SKB_CB(skb)->ddigest)
#define skb_ulp_pdulen(skb) (CXGB3_SKB_CB(skb)->pdulen)
-#define skb_ulp_data(skb) (CXGB3_SKB_CB(skb)->ulp_data)
+#define skb_wr_data(skb) (CXGB3_SKB_CB(skb)->wr_data)

enum c3cb_flags {
C3CB_FLAG_NEED_HDR = 1 << 0, /* packet needs a TX_DATA_WR header */