diff mbox

[V2,for-next,1/6] IB/ipoib: Fix ipoib_neigh hashing to use the correct daddr octets

Message ID 1360079337-8173-2-git-send-email-ogerlitz@mellanox.com (mailing list archive)
State Superseded
Headers show

Commit Message

Or Gerlitz Feb. 5, 2013, 3:48 p.m. UTC 
From: Shlomo Pongratz <shlomop@mellanox.com>

The hash function introduced in commit b63b70d877 "IPoIB: Use a private hash
table for path lookup in xmit path" was designd to use the 3 octets of the
IPoIB HW address that holds the remote QPN. However, this currently isn't
the case under little endian machines as the code there uses the flags part
(octet[0]) and not the last octet of the QPN (octet[3]), fix that.

The fix caused a checkpatch warning on line over 80 characters, to
solve that changed the name of the temp variable that holds the daddr.

Signed-off-by: Shlomo Pongratz <shlomop@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
 drivers/infiniband/ulp/ipoib/ipoib_main.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

Comments

Hefty, Sean Feb. 11, 2013, 7:46 p.m. UTC | #1 
> The hash function introduced in commit b63b70d877 "IPoIB: Use a private hash
> table for path lookup in xmit path" was designd to use the 3 octets of the
> IPoIB HW address that holds the remote QPN. However, this currently isn't
> the case under little endian machines as the code there uses the flags part
> (octet[0]) and not the last octet of the QPN (octet[3]), fix that.
> 
> The fix caused a checkpatch warning on line over 80 characters, to
> solve that changed the name of the temp variable that holds the daddr.
> 
> Signed-off-by: Shlomo Pongratz <shlomop@mellanox.com>
> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
> ---
>  drivers/infiniband/ulp/ipoib/ipoib_main.c |    4 ++--
>  1 files changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/infiniband/ulp/ipoib/ipoib_main.c
> b/drivers/infiniband/ulp/ipoib/ipoib_main.c
> index 6fdc9e7..e459fa7 100644
> --- a/drivers/infiniband/ulp/ipoib/ipoib_main.c
> +++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c
> @@ -844,10 +844,10 @@ static u32 ipoib_addr_hash(struct ipoib_neigh_hash *htbl,
> u8 *daddr)
>  	 * different subnets.
>  	 */
>  	 /* qpn octets[1:4) & port GUID octets[12:20) */
> -	u32 *daddr_32 = (u32 *) daddr;
> +	u32 *d32 = (u32 *)daddr;
>  	u32 hv;
> 
> -	hv = jhash_3words(daddr_32[3], daddr_32[4], 0xFFFFFF & daddr_32[0], 0);
> +	hv = jhash_3words(d32[3], d32[4], cpu_to_be32(0xFFFFFF) & d32[0], 0);

Should d32 be declared as __be32 *?
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Shlomo Pongratz Feb. 12, 2013, 2:47 p.m. UTC | #2 
On 2/11/2013 9:46 PM, Hefty, Sean wrote:
>> --- a/drivers/infiniband/ulp/ipoib/ipoib_main.c
>> >+++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c
>> >@@ -844,10 +844,10 @@ static u32 ipoib_addr_hash(struct ipoib_neigh_hash *htbl,
>> >u8 *daddr)
>> >  	 * different subnets.
>> >  	 */
>> >  	 /* qpn octets[1:4) & port GUID octets[12:20) */
>> >-	u32 *daddr_32 = (u32 *) daddr;
>> >+	u32 *d32 = (u32 *)daddr;
>> >  	u32 hv;
>> >
>> >-	hv = jhash_3words(daddr_32[3], daddr_32[4], 0xFFFFFF & daddr_32[0], 0);
>> >+	hv = jhash_3words(d32[3], d32[4], cpu_to_be32(0xFFFFFF) & d32[0], 0);
> Should d32 be declared as __be32 *?
Hi Sean,

The IPoIB destination address is indeed in big endian format and 
normally the pointer to it should be of type __be32.
However in this case I just want to feed it into the hash function 
without the flags part.
defining d32 as __be32* will make the code a bit ugly as I'll need to 
cast 3 of "jhash_3words" functions arguments.
That is,

__be32 *d32;
....

hv = jhash_3words((__force u32) d32[3], (__force u32) d32[4], (__force 
u32)(cpu_to_be32(0xFFFFFF) & d32[0]), 0);


Best regards,

S.P.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Hefty, Sean Feb. 12, 2013, 4:33 p.m. UTC | #3 
> On 2/11/2013 9:46 PM, Hefty, Sean wrote:
> >> --- a/drivers/infiniband/ulp/ipoib/ipoib_main.c
> >> >+++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c
> >> >@@ -844,10 +844,10 @@ static u32 ipoib_addr_hash(struct ipoib_neigh_hash
> *htbl,
> >> >u8 *daddr)
> >> >  	 * different subnets.
> >> >  	 */
> >> >  	 /* qpn octets[1:4) & port GUID octets[12:20) */
> >> >-	u32 *daddr_32 = (u32 *) daddr;
> >> >+	u32 *d32 = (u32 *)daddr;
> >> >  	u32 hv;
> >> >
> >> >-	hv = jhash_3words(daddr_32[3], daddr_32[4], 0xFFFFFF & daddr_32[0], 0);
> >> >+	hv = jhash_3words(d32[3], d32[4], cpu_to_be32(0xFFFFFF) & d32[0], 0);
> > Should d32 be declared as __be32 *?
> Hi Sean,
> 
> The IPoIB destination address is indeed in big endian format and
> normally the pointer to it should be of type __be32.
> However in this case I just want to feed it into the hash function
> without the flags part.
> defining d32 as __be32* will make the code a bit ugly as I'll need to
> cast 3 of "jhash_3words" functions arguments.
> That is,
> 
> __be32 *d32;
> ....
> 
> hv = jhash_3words((__force u32) d32[3], (__force u32) d32[4], (__force
> u32)(cpu_to_be32(0xFFFFFF) & d32[0]), 0);

Have you run the V2 patch through sparse?
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Or Gerlitz Feb. 12, 2013, 4:53 p.m. UTC | #4 
On 12/02/2013 18:33, Hefty, Sean wrote:
> Have you run the V2 patch through sparse?

oops, I see now that the V2 patches introduced some sparse warnings, 
will fix for V3, thanks for spotting that over.

Or.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Jason Gunthorpe Feb. 12, 2013, 8:35 p.m. UTC | #5 
On Tue, Feb 12, 2013 at 04:47:41PM +0200, Shlomo Pongratz wrote:
> On 2/11/2013 9:46 PM, Hefty, Sean wrote:
> >>>+++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c
> >>>@@ -844,10 +844,10 @@ static u32 ipoib_addr_hash(struct ipoib_neigh_hash *htbl,
> >>>u8 *daddr)
> >>>  	 * different subnets.
> >>>  	 */
> >>>  	 /* qpn octets[1:4) & port GUID octets[12:20) */
> >>>-	u32 *daddr_32 = (u32 *) daddr;
> >>>+	u32 *d32 = (u32 *)daddr;
> >>>  	u32 hv;
> >>>
> >>>-	hv = jhash_3words(daddr_32[3], daddr_32[4], 0xFFFFFF & daddr_32[0], 0);
> >>>+	hv = jhash_3words(d32[3], d32[4], cpu_to_be32(0xFFFFFF) & d32[0], 0);
> >Should d32 be declared as __be32 *?
> Hi Sean,
> 
> The IPoIB destination address is indeed in big endian format and
> normally the pointer to it should be of type __be32.
> However in this case I just want to feed it into the hash function
> without the flags part.
> defining d32 as __be32* will make the code a bit ugly as I'll need
> to cast 3 of "jhash_3words" functions arguments.
> That is,
> 
> __be32 *d32;
> ....
> 
> hv = jhash_3words((__force u32) d32[3], (__force u32) d32[4],
> (__force u32)(cpu_to_be32(0xFFFFFF) & d32[0]), 0);

Not sure what your hv is used for, but be aware that it is going to
have a different value on big and little endian systems..

This is why the (__force u32) is somewhat desirable, because you are
explicitly, and deliberately ignoring the effect of endianness at that
point in the code.

Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/drivers/infiniband/ulp/ipoib/ipoib_main.c b/drivers/infiniband/ulp/ipoib/ipoib_main.c
index 6fdc9e7..e459fa7 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_main.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c
@@ -844,10 +844,10 @@  static u32 ipoib_addr_hash(struct ipoib_neigh_hash *htbl, u8 *daddr)
 	 * different subnets.
 	 */
 	 /* qpn octets[1:4) & port GUID octets[12:20) */
-	u32 *daddr_32 = (u32 *) daddr;
+	u32 *d32 = (u32 *)daddr;
 	u32 hv;
 
-	hv = jhash_3words(daddr_32[3], daddr_32[4], 0xFFFFFF & daddr_32[0], 0);
+	hv = jhash_3words(d32[3], d32[4], cpu_to_be32(0xFFFFFF) & d32[0], 0);
 	return hv & htbl->mask;
 }