Comment 2 for bug 1035348

Here is a list of 6 critical patches that have been applied to net-next tree. Please apply these patches to be2net source in your tree. I have given a brief description in addition to what is available in commit.

Thanks,
Sarvesh

Patch 1. Performance related, will cause xps to mis-behave.

commit c297977ec18deb36b2c0a5ee57101f7ab736ec00
Author: Sarveshwar Bandi <email address hidden>
Date: Wed Jul 25 21:29:50 2012 +0000

    be2net: Fix to parse RSS hash from Receive completions correctly.

    Wrong pointer variable is being used to parse the rss hash from
    receive completions leading to corrupted rss_hash values filled into skb.

    Signed-off-by: Sarveshwar Bandi <email address hidden>
    Signed-off-by: David S. Miller <email address hidden>

Patch 2. OOPS in PowerPC.

commit ac46a4623815113a0305b2a491c125f473a88c53
Author: Anton Blanchard <email address hidden>
Date: Tue Jul 24 15:05:25 2012 +0000

    be2net: Missing byteswap in be_get_fw_log_level causes oops on PowerPC

    We are seeing an oops in be_get_fw_log_level on ppc64 where we walk
    off the end of memory.

    commit 941a77d582c8 (be2net: Fix to allow get/set of debug levels in
    the firmware.) requires byteswapping of num_modes and num_modules.

    Cc: <email address hidden> # 3.5+
    Signed-off-by: Anton Blanchard <email address hidden>
    Acked-by: Sathya Perla <email address hidden>
    Signed-off-by: David S. Miller <email address hidden>

Patch 3. Performance related.

commit ac1ae5f33fd225f46da0072e2091962410a0431b
Author: Eric Dumazet <email address hidden>
Date: Fri Jul 13 03:19:41 2012 +0000

    be2net: dont pull too much data in skb linear part

    skb_fill_rx_data() pulls 64 byte of data in skb->data

    Its too much for TCP (with no options) on IPv4, as total size of headers
    is 14 + 40 = 54

    This means tcp stack and splice() are suboptimal, since tcp payload
    is in part in tcp->data, and in part in skb frag.

    Signed-off-by: Eric Dumazet <email address hidden>
    Acked-by: Padmanabh Ratnakar <email address hidden>
    Signed-off-by: David S. Miller <email address hidden>

Patch 4. ASIC stall issue (Related to Patch 5)

commit e8efcec5394cbf7ae67ccb137cb1a45ae3e6f6c8
Author: Li RongQing <email address hidden>
Date: Wed Jul 4 16:05:42 2012 +0000

    be2net: Fix Endian

    ETH_P_IP is host Endian, skb->protocol is big Endian, when
    compare them, we should change ETH_P_IP from host endian
    to big endian, htons, not ntohs.

    CC: Somnath Kotur <email address hidden>
    Signed-off-by: Li RongQing <email address hidden>
    Acked-by: Somnath Kotur <email address hidden>
    Signed-off-by: David S. Miller <email address hidden>

Patch 5. Patches avoids ASIC stall.

commit 93040ae5cc8dcc893eca4a4366dc8415af278edf
Author: Somnath Kotur <email address hidden>
Date: Tue Jun 26 22:32:10 2012 +0000

    be2net: Fix to trim skb for padded vlan packets to workaround an ASIC Bug

    Fixed spelling error in a comment as pointed out by DaveM.
    Also refactored existing code a bit to provide placeholders for another ASIC
    Bug workaround that will be checked-in soon after this.

    Signed-off-by: Somnath Kotur <email address hidden>
    Signed-off-by: David S. Miller <email address hidden>

Patch 6. Patch avoids ASIC stall.

commit b7e5887e0e414bde0a2f311ae3fbea5611562e29
Author: Sarveshwar Bandi <email address hidden>
Date: Wed Jun 13 19:51:43 2012 +0000

    be2net: reduce gso_max_size setting to account for ethernet header.

    The maximum size of packet that can be handled by controller including ethernet
    header is 65535. Reducing gso_max_size accordingly.

    Signed-off-by: Sarveshwar Bandi <email address hidden>
    Signed-off-by: David S. Miller <email address hidden>