[ofa-general] Re: [ewg] OFED 1.2.c-11 is available

Or Gerlitz ogerlitz at voltaire.com
Thu Aug 9 01:02:10 PDT 2007


Tziporet Koren wrote:

> Main changes from OFED 1.2.c-10:
> - Merged all 1.2 branch patches to 1.2.c

> - mlx4
>         - Added support for interrupt moderation for create cq to support ipoib.
>         - Display the following device information via sysfs: board_id, fw_ver,
>           hw_rev and hca_type. As a result ibstat presenting all entries.
>         - Fixed destroy special QPs.
>         - Fixed the work completion opcode for RDMA_READ operation

> - IPoIB:
>         - Add interrupt moderation support for IPoIB UD
>         - NAPI is available using a module parameter
>         - Fixed a leak in ipoib_transport_dev_init
>         - Fixed kernel oops in IPoIB download

Tziporet, Michael,

Looking in your release announcing email and also on the quilt listing 
of patches applied to the kernel code, it seems that you are actually 
adding some experimental and unreviewed code placed into non signed and 
not documented patches into OFED.

Taking into account that the 'E' and 'D' letters are there for some 
reason, what's the story here? what's the reason to have non-reviewed 
code in OFED one month before the next release feature freeze?

Michael, OFED 1.2 is based on 2.6.21 and OFED 1.2.1 on 2.6.22, did you 
do some 1.2 wrap up of returning fixes to the upstream code? from the 
length of the patch list (specifically, the amount of ipoib and mthca 
ones) it does not seem so, please correct me if I am wrong.

Vlad, I see that development is going in RDS, nice. Is there a plan to 
submit the code for review towards submission upstream? at the long run 
I think that distributing non reviewed code with OFED would create more 
damage then benefit.

Or.

> $ quilt series 

> patches/0001-cxgb3-sge-page-management.patch
> patches/0002-cxgb3-Firmware-update.patch
> patches/0003-cxgb3-TP-SRAM-update.patch
> patches/0004-RDMA-cxgb3-Streaming-RDMA-mode-transition-fixes.patch
> patches/0005-RDMA-cxgb3-TERMINATE-WRs-can-hang-the-tx-ofld-queue.patch
> patches/0006-RDMA-cxgb3-Don-t-count-neg_adv-abort_req_rss-messag.patch
> patches/0007-RDMA-cxgb3-ctrl-qp-init-clear-shouldn-t-set-the-gen.patch
> patches/0008-RDMA-cxgb3-Don-t-post-TID_RELEASE-message.patch
> patches/0009-mlx4_add_wc.patch
> patches/0009-RDMA-cxgb3-Don-t-abort-after-failures-sending-the-m.patch
> patches/0010-mlx4_h_mutex.patch
> patches/0010-RDMA-cxgb3-Check-return-of-kmalloc-in-iwch_regist.patch
> patches/0011-mlx4_ib_h_mutex.patch
> patches/0012-mlx4-query-qp.patch
> patches/0013-query-srq.patch
> patches/0013-RDMA-cxgb3-Remove-cm_id-reference-on-listen-failure.patch
> patches/0014-mlx4_read_max_msg_sz_from_dev_cap.patch
> patches/0015_mlx4_set_cacheline_sz.patch
> patches/0017_mlx4_internal_err_reset.patch
> patches/0018_mlx4_qp_per_mcg.patch
> patches/0019_max_rdma_reads_per_qp.patch
> patches/0019_mlx4_double_free_mr.patch
> patches/0020_mlx4_alloc_coherent.patch
> patches/0022-mlx4-mr-direct-mtt.patch
> patches/0023_mlx4_sg_stamp.patch
> patches/0024_mlx4_fmr.patch
> patches/0025_iw_cxgb3_writable-params.patch
> patches/0025-mlx4-sysfs-dev-info.patch
> patches/0026_iw_cxgb3_deadlock.patch
> patches/0026-mlx4-check-usr-sq-sz.patch
> patches/0027-mlx4-sqp-no-bitmap.patch
> patches/attr_contaner_add_mutex_h.patch
> patches/cma_established1.patch
> patches/cma_response_timeout.patch
> patches/cma_tavor_quirk.patch
> patches/cmd_tout.patch
> patches/dma_map_sg.patch
> patches/ib_mad_001_rmpp_leak.patch
> patches/ipath-22-memcpy_cachebypass.patch
> patches/ipoib_crash_wa.patch
> patches/ipoib_dev_in_ipoib_neigh.patch
> patches/ipoib_napi_optional.patch
> patches/ipoib_selector_updated.patch
> patches/ipoib_t_transport_init_leak.patch
> patches/iwcm_ordird.patch
> patches/mthca_7_change_token_on_timeout.patch
> patches/mthca_catas_wqueue_namelen.patch
> patches/mthca_wrid_swap.patch
> patches/mthca_x_qp_per_mcg.patch
> patches/qos_0_mthca.patch
> patches/sdp_cq_param.patch
> patches/sdp_post_credits.patch
> patches/sean_cm_limit_mra_timeout.patch
> patches/sean_local_sa_1_notifications.patch
> patches/sean_local_sa_2_cache.patch
> patches/sean_local_sa_3_disable.patch
> patches/sean_local_sa_4_fix_hang.patch
> patches/srp_1_recreate_at_reconnect.patch
> patches/wc_opcode.patch
> patches/x_0020_int_mod.patch
> patches/x_0030_ipoib_int_mod.patch





> - RDS:
>         - rds_poll() always returns POLLOUT.  If you try to send and get
>           EWOULDBLOCK because the remote receiver is congested then you 
> don't get
>           to wait for POLLOUT to be raised before sending again -- it's 
> always
>           raised.  This is worked around by implementing some kind of 
> exponential
>           back-off while retrying the send.  This can be done by 
> increasing the
>           timeout given to poll().  At each poll() timeout expiry the 
> send is
>           tried again.
>         - Set max send and receive scatter/gather list size to 2 in the 
> QP attributes.
>           The second sge used for RDS header.
>         - Added rs_poll_flag to rds_socket structure to indicate the 
> back pressure
>           status and send space availability.
>         - Added MODULE_VERSION.




More information about the general mailing list