Re: [ewg] status of ofed ipoib changes which are not upstream
Or Gerlitz wrote: What's the status of merging your ipoib related patches? Looking on Roland's git I see that the checksum offload, LSO, and most of the cq moderation patches are merged, but things like the cq split along with all the small packet optimizations aren't. Hi Eli, I made a pass on the ofed 1.3 ipoib patches attempting to sort them to ones which are merged upstream and ones which aren't, I'll be happy if you can validate my sorting. I can assist with reviewing the non merged ones, once you send them to the general list. ones that were already merged to kernel later then the one 1.3 was based on, or were that merged since 1.3 was released --> 1 kernel_patches/fixes/ipoib_0010_Add-high-dma-support-to-ipoib.patch 2 kernel_patches/fixes/ipoib_0020_Add-s-g-support-for-IPOIB.patch 3 kernel_patches/fixes/ipoib_0040_checksum-offload.patch 4 kernel_patches/fixes/ipoib_0050_Add-LSO-support.patch 5 kernel_patches/fixes/ipoib_0060_ethtool-support.patch 6 kernel_patches/fixes/ipoib_0070_modiy_cq_params.patch 7 kernel_patches/fixes/ipoib_0120_check_grat_arp_with_cm.patch 8 kernel_patches/fixes/ipoib_0200_non_srq.patch 9 kernel_patches/fixes/ipoib_0240_4kmtu.patch 10 kernel_patches/fixes/ipoib_0260_pkey_change.patch 11 kernel_patches/fixes/ipoib_0270_remove_alloc.patch 12 kernel_patches/fixes/ipoib_0300_reap.patch non merged and (except for cq def params & cq split) not submitted for review --> 13 kernel_patches/fixes/ipoib_0110_set_default_cq_patams.patch 14 kernel_patches/fixes/ipoib_0180_split_cq.patch 15 kernel_patches/fixes/ipoib_0190_unsig_udqp.patch 16 kernel_patches/fixes/ipoib_0210_draft_wr.patch 17 kernel_patches/fixes/ipoib_0220_ud_post_list.patch 18 kernel_patches/fixes/ipoib_0230_srq_post_n.patch 19 kernel_patches/fixes/ipoib_0250_non_srq_param.patch 20 kernel_patches/fixes/ipoib_0290_reduce_cm_tx.patch 21 kernel_patches/fixes/ipoib_0310_def_ring_sizes.patch 22 kernel_patches/fixes/ipoib_0320_small_skb_copy.patch 23 kernel_patches/fixes/ipoib_0330_child_mtu.patch 24 kernel_patches/fixes/ipoib_selector_updated.patch merged partially and in different form (yours applied to few allocations and used vmap, roland's handled one allocation and uses vmalloc) --> 25 kernel_patches/fixes/ipoib_0280_vmap.patch I am mostly worried from patches 13-24 skipping the review cycle for 2.6.26, some of them are kind of very sensitive on the one hand and not very attractive on the other hand eg: you have decided to hold off with the merge of ipoib_0190_unsig_udqp.patch since you say it does not provide any benefit. I see in the ofed git that eight bugs where found in this patch through the 1.3 cycle and this is before mainline review! other patches. I say that if you don't submit it to mainline, drop it from ofed. Also - - the srq post list and ud post receive list are kind of simple and not involving much states, can they be submitted to review without much effort from your side? - the child mtu patch seems like a bug fix, doesn't it apply also to mainline? - the selector patch is from ofed 1.1 or so, can't it just be removed? - etc Or. ___ ewg mailing list ewg@lists.openfabrics.org http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
Re: [ewg] status of ofed ipoib changes which are not upstream
On Mon, 2008-04-28 at 11:31 +0300, Or Gerlitz wrote: > Or Gerlitz wrote: > > What's the status of merging your ipoib related patches? Looking on > > Roland's git I > > see that the checksum offload, LSO, and most of the cq moderation patches > > are merged, > > but things like the cq split along with all the small packet optimizations > > aren't. > Hi Eli, > > I made a pass on the ofed 1.3 ipoib patches attempting to sort them to > ones which are merged upstream and ones which aren't, I'll be happy if > you can validate my sorting. > > I can assist with reviewing the non merged ones, once you send them to > the general list. > > ones that were already merged to kernel later then the one 1.3 was based > on, or were that merged since 1.3 was released --> > > 1 > kernel_patches/fixes/ipoib_0010_Add-high-dma-support-to-ipoib.patch > 2kernel_patches/fixes/ipoib_0020_Add-s-g-support-for-IPOIB.patch > 3kernel_patches/fixes/ipoib_0040_checksum-offload.patch > 4kernel_patches/fixes/ipoib_0050_Add-LSO-support.patch > 5kernel_patches/fixes/ipoib_0060_ethtool-support.patch > 6kernel_patches/fixes/ipoib_0070_modiy_cq_params.patch > 7kernel_patches/fixes/ipoib_0120_check_grat_arp_with_cm.patch > 8kernel_patches/fixes/ipoib_0200_non_srq.patch > 9kernel_patches/fixes/ipoib_0240_4kmtu.patch > 10kernel_patches/fixes/ipoib_0260_pkey_change.patch > 11kernel_patches/fixes/ipoib_0270_remove_alloc.patch > 12kernel_patches/fixes/ipoib_0300_reap.patch Yes, the above patches are already in upstream kernel. > > non merged and (except for cq def params & cq split) not submitted for review > --> > > 13kernel_patches/fixes/ipoib_0110_set_default_cq_patams.patch > 14kernel_patches/fixes/ipoib_0180_split_cq.patch We have to relate to the version I sent against the "for-2.6.26" branch and use it eventually. > 15kernel_patches/fixes/ipoib_0190_unsig_udqp.patch On older kernels this patch seems to improve throughput of small messages so we should make the effort to include it. I would like to verify this again and if this is correct I will send for review. > 16kernel_patches/fixes/ipoib_0210_draft_wr.patch > 17kernel_patches/fixes/ipoib_0220_ud_post_list.patch > 18kernel_patches/fixes/ipoib_0230_srq_post_n.patch > 19kernel_patches/fixes/ipoib_0250_non_srq_param.patch > 20kernel_patches/fixes/ipoib_0290_reduce_cm_tx.patch > 21kernel_patches/fixes/ipoib_0310_def_ring_sizes.patch > 22kernel_patches/fixes/ipoib_0320_small_skb_copy.patch > 23kernel_patches/fixes/ipoib_0330_child_mtu.patch > 24kernel_patches/fixes/ipoib_selector_updated.patch > > merged partially and in different form (yours applied to few allocations and > used vmap, roland's handled one allocation and uses vmalloc) --> > > 25kernel_patches/fixes/ipoib_0280_vmap.patch > > I am mostly worried from patches 13-24 skipping the review cycle for 2.6.26, > some of them are kind of very sensitive on the one hand and not very > attractive on the other hand eg: you have decided to hold off with the merge > of ipoib_0190_unsig_udqp.patch > since you say it does not provide any benefit. I see in the ofed git that > eight bugs where found in this patch through the 1.3 cycle and this is before > mainline review! other patches. I say that if you don't submit it to > mainline, drop it from ofed. OK, I'll review them and submit what I think is important and drop the others. > > Also - > > - the srq post list and ud post receive list are kind of simple and not > involving much > states, can they be submitted to review without much effort from your side? > > - the child mtu patch seems like a bug fix, doesn't it apply also to mainline? > > - the selector patch is from ofed 1.1 or so, can't it just be removed? Will review them and post to the list. > > - etc > > Or. > > > > > > ___ > ewg mailing list > ewg@lists.openfabrics.org > http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg ___ ewg mailing list ewg@lists.openfabrics.org http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
Re: [ewg] status of ofed ipoib changes which are not upstream
Eli Cohen wrote: 15 kernel_patches/fixes/ipoib_0190_unsig_udqp.patch On older kernels this patch seems to improve throughput of small messages so we should make the effort to include it. I would like to verify this again and if this is correct I will send for review. I don't see why should performance gain only with old kernels justify including such patch, Roland? 16 kernel_patches/fixes/ipoib_0210_draft_wr.patch 17 kernel_patches/fixes/ipoib_0220_ud_post_list.patch 18 kernel_patches/fixes/ipoib_0230_srq_post_n.patch 19 kernel_patches/fixes/ipoib_0250_non_srq_param.patch 20 kernel_patches/fixes/ipoib_0290_reduce_cm_tx.patch 21 kernel_patches/fixes/ipoib_0310_def_ring_sizes.patch 22 kernel_patches/fixes/ipoib_0320_small_skb_copy.patch 23 kernel_patches/fixes/ipoib_0330_child_mtu.patch 24 kernel_patches/fixes/ipoib_selector_updated.patch OK, I'll review them and submit what I think is important and drop the others. Good, the only tiny problem is that the merge window on 2.6.26 opened eleven days ago, which means its going to be closed soon, seems we had somehow bad timing here, I guess it was not intended, correct? Anyway, lets review them all now, such that this way or another, each patch of this list is either dropped (unsig_udqp ?!) or merged through 2.6.26 or the for-2.6.27 branch of Roland's tree, OK? Or. ___ ewg mailing list ewg@lists.openfabrics.org http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
Re: [ewg] status of ofed ipoib changes which are not upstream
On Mon, 2008-04-28 at 13:43 +0300, Or Gerlitz wrote: > Eli Cohen wrote: > >> 15 kernel_patches/fixes/ipoib_0190_unsig_udqp.patch > > On older kernels this patch seems to improve throughput of small > > messages so we should make the effort to include it. I would like to > > verify this again and if this is correct I will send for review. > I don't see why should performance gain only with old kernels justify > including such patch, Roland? I don't intend to push this to upstream kernel but I think we should keep it for ofed since for older kernels it provides better performance. > >> 16 kernel_patches/fixes/ipoib_0210_draft_wr.patch > >> 17 kernel_patches/fixes/ipoib_0220_ud_post_list.patch > >> 18 kernel_patches/fixes/ipoib_0230_srq_post_n.patch > >> 19 kernel_patches/fixes/ipoib_0250_non_srq_param.patch > >> 20 kernel_patches/fixes/ipoib_0290_reduce_cm_tx.patch > >> 21 kernel_patches/fixes/ipoib_0310_def_ring_sizes.patch > >> 22 kernel_patches/fixes/ipoib_0320_small_skb_copy.patch > >> 23 kernel_patches/fixes/ipoib_0330_child_mtu.patch > >> 24 kernel_patches/fixes/ipoib_selector_updated.patch We have a few critical bugs in ofed which I will address first and in the background I'll push the fixes first and then the other patches. I hope to get more of them to 2.6.26 and the rest will wait for 2.6.27. ___ ewg mailing list ewg@lists.openfabrics.org http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
Re: [ewg] status of ofed ipoib changes which are not upstream
Eli Cohen wrote: I don't intend to push this to upstream kernel but I think we should keep it for ofed since for older kernels it provides better performance. Again, the patch is very sensitive by nature, it did not pass a review and numerous bugs were associated with it through the 1.3 cycle, I think isn't it should be enough to see that there is a problem here. If you see this differently, then send it for "review only for old kernels", state on which kernels you see improvement, what is this improvement and then we see. 16 kernel_patches/fixes/ipoib_0210_draft_wr.patch 17 kernel_patches/fixes/ipoib_0220_ud_post_list.patch 18 kernel_patches/fixes/ipoib_0230_srq_post_n.patch 19 kernel_patches/fixes/ipoib_0250_non_srq_param.patch 20 kernel_patches/fixes/ipoib_0290_reduce_cm_tx.patch 21 kernel_patches/fixes/ipoib_0310_def_ring_sizes.patch 22 kernel_patches/fixes/ipoib_0320_small_skb_copy.patch 23 kernel_patches/fixes/ipoib_0330_child_mtu.patch 24 kernel_patches/fixes/ipoib_selector_updated.patch We have a few critical bugs in ofed which I will address first and in the background I'll push the fixes first and then the other patches. I hope to get more of them to 2.6.26 and the rest will wait for 2.6.27. With "critical bugs in ofed (1.3)" are you referring to the IPoIB related cases 985,989,1004,etc or its something else? 2.6.26 feature window is going to be closed any day now, so the way to go is review and merge them into the for-2.6.27 branch, but lets do it now and not 3 months away. Or. ___ ewg mailing list ewg@lists.openfabrics.org http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg