| Message ID | 20240909054318.1809580-1-almasrymina@google.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | Device Memory TCP | expand |
On 2024/9/9 13:43, Mina Almasry wrote: > > Perf - page-pool benchmark: > --------------------------- > > bench_page_pool_simple.ko tests with and without these changes: > https://pastebin.com/raw/ncHDwAbn > > AFAIK the number that really matters in the perf tests is the > 'tasklet_page_pool01_fast_path Per elem'. This one measures at about 8 > cycles without the changes but there is some 1 cycle noise in some > results. > > With the patches this regresses to 9 cycles with the changes but there > is 1 cycle noise occasionally running this test repeatedly. > > Lastly I tried disable the static_branch_unlikely() in > netmem_is_net_iov() check. To my surprise disabling the > static_branch_unlikely() check reduces the fast path back to 8 cycles, > but the 1 cycle noise remains. Sorry for the late report, as I was adding a testing page_pool ko basing on [1] to avoid introducing performance regression when fixing the bug in [2]. I used it to test the performance impact of devmem patchset for page_pool too, it seems there might be some noticable performance impact quite stably for the below testcases, about 5%~16% performance degradation as below in the arm64 system: Before the devmem patchset: Performance counter stats for 'insmod ./page_pool_test.ko test_push_cpu=16 test_pop_cpu=16 nr_test=100000000 test_napi=1' (100 runs): 17.167561 task-clock (msec) # 0.003 CPUs utilized ( +- 0.40% ) 8 context-switches # 0.474 K/sec ( +- 0.65% ) 0 cpu-migrations # 0.001 K/sec ( +-100.00% ) 84 page-faults # 0.005 M/sec ( +- 0.13% ) 44576552 cycles # 2.597 GHz ( +- 0.40% ) 59627412 instructions # 1.34 insn per cycle ( +- 0.03% ) 14370325 branches # 837.063 M/sec ( +- 0.02% ) 21902 branch-misses # 0.15% of all branches ( +- 0.27% ) 6.818873600 seconds time elapsed ( +- 0.02% ) Performance counter stats for 'insmod ./page_pool_test.ko test_push_cpu=16 test_pop_cpu=16 nr_test=100000000 test_napi=1 test_direct=1' (100 runs): 17.595423 task-clock (msec) # 0.004 CPUs utilized ( +- 0.01% ) 8 context-switches # 0.460 K/sec ( +- 0.50% ) 0 cpu-migrations # 0.000 K/sec 84 page-faults # 0.005 M/sec ( +- 0.15% ) 45693020 cycles # 2.597 GHz ( +- 0.01% ) 59676212 instructions # 1.31 insn per cycle ( +- 0.00% ) 14385384 branches # 817.564 M/sec ( +- 0.00% ) 21786 branch-misses # 0.15% of all branches ( +- 0.14% ) 4.098627802 seconds time elapsed ( +- 0.11% ) After the devmem patchset: Performance counter stats for 'insmod ./page_pool_test.ko test_push_cpu=16 test_pop_cpu=16 nr_test=100000000 test_napi=1' (100 runs): 17.047973 task-clock (msec) # 0.002 CPUs utilized ( +- 0.39% ) 8 context-switches # 0.488 K/sec ( +- 0.82% ) 0 cpu-migrations # 0.001 K/sec ( +- 70.35% ) 84 page-faults # 0.005 M/sec ( +- 0.12% ) 44269558 cycles # 2.597 GHz ( +- 0.39% ) 59594383 instructions # 1.35 insn per cycle ( +- 0.02% ) 14362599 branches # 842.481 M/sec ( +- 0.02% ) 21949 branch-misses # 0.15% of all branches ( +- 0.25% ) 7.964890303 seconds time elapsed ( +- 0.16% ) Performance counter stats for 'insmod ./page_pool_test.ko test_push_cpu=16 test_pop_cpu=16 nr_test=100000000 test_napi=1 test_direct=1' (100 runs): 17.660975 task-clock (msec) # 0.004 CPUs utilized ( +- 0.02% ) 8 context-switches # 0.458 K/sec ( +- 0.57% ) 0 cpu-migrations # 0.003 K/sec ( +- 43.81% ) 84 page-faults # 0.005 M/sec ( +- 0.17% ) 45862652 cycles # 2.597 GHz ( +- 0.02% ) 59764866 instructions # 1.30 insn per cycle ( +- 0.01% ) 14404323 branches # 815.602 M/sec ( +- 0.01% ) 21826 branch-misses # 0.15% of all branches ( +- 0.19% ) 4.304644609 seconds time elapsed ( +- 0.75% ) 1. https://lore.kernel.org/all/20240906073646.2930809-2-linyunsheng@huawei.com/ 2. https://lore.kernel.org/lkml/8067f204-1380-4d37-8ffd-007fc6f26738@kernel.org/T/ >
On Mon, Sep 9, 2024 at 4:21 AM Yunsheng Lin <linyunsheng@huawei.com> wrote: > > On 2024/9/9 13:43, Mina Almasry wrote: > > > > > Perf - page-pool benchmark: > > --------------------------- > > > > bench_page_pool_simple.ko tests with and without these changes: > > https://pastebin.com/raw/ncHDwAbn > > > > AFAIK the number that really matters in the perf tests is the > > 'tasklet_page_pool01_fast_path Per elem'. This one measures at about 8 > > cycles without the changes but there is some 1 cycle noise in some > > results. > > > > With the patches this regresses to 9 cycles with the changes but there > > is 1 cycle noise occasionally running this test repeatedly. > > > > Lastly I tried disable the static_branch_unlikely() in > > netmem_is_net_iov() check. To my surprise disabling the > > static_branch_unlikely() check reduces the fast path back to 8 cycles, > > but the 1 cycle noise remains. > > Sorry for the late report, as I was adding a testing page_pool ko basing > on [1] to avoid introducing performance regression when fixing the bug in > [2]. > I used it to test the performance impact of devmem patchset for page_pool > too, it seems there might be some noticable performance impact quite stably > for the below testcases, about 5%~16% performance degradation as below in > the arm64 system: > Correct me if I'm wrong here, but on the surface here it seems that you're re-reporting a known issue. Consensus seems to be that it's a non-issue. In v6 I reported that the bench_page_pool_simple.ko test reports a 1 cycle regression with these patches, from 8->9 cycles. That is roughly consistent with the 5-15% you're reporting. I root caused the reason for the regression to be the netmem_is_net_iov() check in the fast path. I removed this regression in v7 (see the change log) by conditionally compiling the check in that function. In v8, Pavel/Jens/David pushed back on the ifdef check. See this entire thread, but in particular this response from Jens: https://lore.kernel.org/lkml/11f52113-7b67-4b45-ba1d-29b070050cec@kernel.dk/ Seems consensus that it's 'not really worth it in this scenario'.
On 2024/9/10 0:54, Mina Almasry wrote: > On Mon, Sep 9, 2024 at 4:21 AM Yunsheng Lin <linyunsheng@huawei.com> wrote: >> >> On 2024/9/9 13:43, Mina Almasry wrote: >> >>> >>> Perf - page-pool benchmark: >>> --------------------------- >>> >>> bench_page_pool_simple.ko tests with and without these changes: >>> https://pastebin.com/raw/ncHDwAbn >>> >>> AFAIK the number that really matters in the perf tests is the >>> 'tasklet_page_pool01_fast_path Per elem'. This one measures at about 8 >>> cycles without the changes but there is some 1 cycle noise in some >>> results. >>> >>> With the patches this regresses to 9 cycles with the changes but there >>> is 1 cycle noise occasionally running this test repeatedly. >>> >>> Lastly I tried disable the static_branch_unlikely() in >>> netmem_is_net_iov() check. To my surprise disabling the >>> static_branch_unlikely() check reduces the fast path back to 8 cycles, >>> but the 1 cycle noise remains. >> >> Sorry for the late report, as I was adding a testing page_pool ko basing >> on [1] to avoid introducing performance regression when fixing the bug in >> [2]. >> I used it to test the performance impact of devmem patchset for page_pool >> too, it seems there might be some noticable performance impact quite stably >> for the below testcases, about 5%~16% performance degradation as below in >> the arm64 system: >> > > Correct me if I'm wrong here, but on the surface here it seems that > you're re-reporting a known issue. Consensus seems to be that it's a > non-issue. > > In v6 I reported that the bench_page_pool_simple.ko test reports a 1 > cycle regression with these patches, from 8->9 cycles. That is roughly > consistent with the 5-15% you're reporting. From the description above in the cover letter, I thought the performance data using the out of tree testing ko is not stable enough to justify the performance impact. > > I root caused the reason for the regression to be the > netmem_is_net_iov() check in the fast path. I removed this regression > in v7 (see the change log) by conditionally compiling the check in > that function. > > In v8, Pavel/Jens/David pushed back on the ifdef check. See this > entire thread, but in particular this response from Jens: It seemed the main objection is about how to enable this feature for the io_uring? And it seemed that you had added the CONFIG_NET_DEVMEM for this devmem thing, why not use it for that? > > https://lore.kernel.org/lkml/11f52113-7b67-4b45-ba1d-29b070050cec@kernel.dk/ > > Seems consensus that it's 'not really worth it in this scenario'. I was only reading through the above thread, it didn't seemed to reach to consensus as Jesper pointed out the performance impact for the XDP DROP case in the same thread. https://lore.kernel.org/lkml/779b9542-4170-483a-af54-ca0dd471f774@kernel.org/ >
On 9/10/24 11:44, Yunsheng Lin wrote: > On 2024/9/10 0:54, Mina Almasry wrote: >> On Mon, Sep 9, 2024 at 4:21 AM Yunsheng Lin <linyunsheng@huawei.com> wrote: >>> >>> On 2024/9/9 13:43, Mina Almasry wrote: >>> >>>> >>>> Perf - page-pool benchmark: >>>> --------------------------- >>>> >>>> bench_page_pool_simple.ko tests with and without these changes: >>>> https://pastebin.com/raw/ncHDwAbn >>>> >>>> AFAIK the number that really matters in the perf tests is the >>>> 'tasklet_page_pool01_fast_path Per elem'. This one measures at about 8 >>>> cycles without the changes but there is some 1 cycle noise in some >>>> results. >>>> >>>> With the patches this regresses to 9 cycles with the changes but there >>>> is 1 cycle noise occasionally running this test repeatedly. >>>> >>>> Lastly I tried disable the static_branch_unlikely() in >>>> netmem_is_net_iov() check. To my surprise disabling the >>>> static_branch_unlikely() check reduces the fast path back to 8 cycles, >>>> but the 1 cycle noise remains. >>> >>> Sorry for the late report, as I was adding a testing page_pool ko basing >>> on [1] to avoid introducing performance regression when fixing the bug in >>> [2]. >>> I used it to test the performance impact of devmem patchset for page_pool >>> too, it seems there might be some noticable performance impact quite stably >>> for the below testcases, about 5%~16% performance degradation as below in >>> the arm64 system: >>> >> >> Correct me if I'm wrong here, but on the surface here it seems that >> you're re-reporting a known issue. Consensus seems to be that it's a >> non-issue. >> >> In v6 I reported that the bench_page_pool_simple.ko test reports a 1 >> cycle regression with these patches, from 8->9 cycles. That is roughly >> consistent with the 5-15% you're reporting. > > From the description above in the cover letter, I thought the performance > data using the out of tree testing ko is not stable enough to justify the > performance impact. > >> >> I root caused the reason for the regression to be the >> netmem_is_net_iov() check in the fast path. I removed this regression >> in v7 (see the change log) by conditionally compiling the check in >> that function. >> >> In v8, Pavel/Jens/David pushed back on the ifdef check. See this >> entire thread, but in particular this response from Jens: > > It seemed the main objection is about how to enable this feature > for the io_uring? The pushback was that config checks as optimisation don't work in real life, they inevitably get enabled everywhere but some niche cases. io_uring could do another config for memory providers, but even if it's not enabled by default (which is not a great option), distributions will eventually turn it on. So, if you have that "niche use case" that fully controls the kernel and wants to shed this overhead, we can do a config structure, but if it's about overhead for everyone in general, configs hardly help anything, even without any io_uring in the picture. > And it seemed that you had added the CONFIG_NET_DEVMEM for this > devmem thing, why not use it for that?
On Mon, 9 Sep 2024 05:43:05 +0000 Mina Almasry wrote: > Device memory TCP (devmem TCP) is a proposal for transferring data to and/or > from device memory efficiently, without bouncing the data to a host memory > buffer. Mina, if you'd like to see this in v6.12 -- please fix the nits and repost ASAP.
On Tue, Sep 10, 2024 at 8:02 AM Jakub Kicinski <kuba@kernel.org> wrote: > > On Mon, 9 Sep 2024 05:43:05 +0000 Mina Almasry wrote: > > Device memory TCP (devmem TCP) is a proposal for transferring data to and/or > > from device memory efficiently, without bouncing the data to a host memory > > buffer. > > Mina, if you'd like to see this in v6.12 -- please fix the nits and > repost ASAP. Running my presubmits now and will repost in the next 2 hours or so. -- Thanks, Mina
v25: https://patchwork.kernel.org/project/netdevbpf/list/?series=885396&state=* === Major changes: - Moved devmem.h and mp_dmabuf_devmem.h to internal header files. - Changed the page_pool_params to take in a queue_idx rather than a struct netdev_rx_queue. - Added WARN_ON_ONCE around __skb_checksum readability check and added check to skb_checksum_help(). Other more minor feedback addressed as well. v24: https://patchwork.kernel.org/project/netdevbpf/list/?series=884556&state=* ==== No major changes. Mostly addressing issues in the error paths of dmabuf binding, and code cleanups/improvements from reviewers: Changes: - Fix failing ynl regen error. - Error path fixes & extack error messages in dmabuf binding. - Code cleanup in introspection. - gitignore ynl.d generated file. Full devmem TCP changes including the full GVE driver implementation is here: https://github.com/mina/linux/commits/tcpdevmem-v24/ v23: https://patchwork.kernel.org/project/netdevbpf/list/?series=882978&state=* ==== Fixing relatively minor issues called out in v22. (thanks again!) Mostly code cleanups, extack error messages, and minor reworks. Nothing major really changed, so the exact changes per commit is called in the commit messages. Full devmem TCP changes including the full GVE driver implementation is here: https://github.com/mina/linux/commits/tcpdevmem-v23/ v22: https://patchwork.kernel.org/project/netdevbpf/list/?series=881158&state=* ==== v22 aims to resolve the pending issue pointed to in v21, which is the interaction with xdp. In this series I rebase on top of the minor refactor which refactors propagating xdp configuration to slave devices: https://patchwork.kernel.org/project/netdevbpf/list/?series=881994&state=* I then disable setting xdp on devices using memory providers, and propagating xdp configuration to devices using memory providers. Full devmem TCP changes including the full GVE driver implementation is here: https://github.com/mina/linux/commits/tcpdevmem-v22/ v21: https://patchwork.kernel.org/project/netdevbpf/list/?series=880735&state=* ==== v20 addressed some comments and resolved a test failure, but introduced an unfortunate build error with a config edge case I wasn't testing. v21 simply resolves that error. Major Changes: - Resolve build error with CONFIG_PAGE_POOL=n && CONFIG_NET=y Full devmem TCP changes including the full GVE driver implementation is here: https://github.com/mina/linux/commits/tcpdevmem-v21/ v20: https://patchwork.kernel.org/project/netdevbpf/list/?series=879373&state=* ==== v20 aims to resolve a couple of bug reports against v19, and addresses some review comments around the page_pool_check_memory_provider mechanism. Major changes: - Test edge cases such as header split disabled in selftest. - Change `offset = 0` back to `offset = offset - start` to resolve issue found in RX path by Taehee (thanks!) - Address a few comments around page_pool_check_memory_provider() from Pavel & Jakub. - Removed some unnecessary includes across various patches in the series. - Removed unnecessary EXPORT_SYMBOL(page_pool_mem_providers) (Jakub). - Fix regression caused by incorrect dev_get_max_mp_channel check, along with rename (Jakub). Full devmem TCP changes including the full GVE driver implementation is here: https://github.com/mina/linux/commits/tcpdevmem-v20/ v19: https://patchwork.kernel.org/project/netdevbpf/list/?series=876852&state=* ==== v18 got a thorough review (thanks!), and this iteration addresses the feedback. Major changes: - Prevent deactivating mp bound queues. - Prevent installing xdp on mp bound netdevs, or installing mps on xdp installed netdevs. - Fix corner cases in netlink API vis-a-vis missing attributes. - Iron out the unreadable netmem driver support story. To be honest, the conversation with Jakub & Pavel got a bit confusing for me. I've implemented an approach in this set that makes sense to me, and AFAICT, addresses the requirements. It may be good as-is, or it may be a conversation starter/continuer. To be honest IMO there are many ways to skin this cat and I don't see an extremely strong reason to go for one approach over another. Here is one approach you may like. - Don't reset niov dma_addr on allocation & free. - Add some tests to the selftest that catches some of the issues around missing netlink attributes or deactivating mp-bound queues. Full devmem TCP changes including the full GVE driver implementation is here: https://github.com/mina/linux/commits/tcpdevmem-v19/ v18: https://patchwork.kernel.org/project/netdevbpf/list/?series=874848&state=* ==== v17 got minor feedback: (a) to beef up the description on patch 1 and (b) to remove the leading underscores in the header definition. I applied (a). (b) seems to be against current conventions so I did not apply before further discussion. Full devmem TCP changes including the full GVE driver implementation is here: https://github.com/mina/linux/commits/tcpdevmem-v17/ v17: https://patchwork.kernel.org/project/netdevbpf/list/?series=869900&state=* ==== v16 also got a very thorough review and some testing (thanks again!). Thes version addresses all the concerns reported on v15, in terms of feedback and issues reported. Major changes: - Use ASSERT_RTNL. - Moved around some of the page_pool helpers definitions so I can hide some netmem helpers in private files as Jakub suggested. - Don't make every net_iov hold a ref on the binding as Jakub suggested. - Fix issue reported by Taehee where we access queues after they have been freed. Full devmem TCP changes including the full GVE driver implementation is here: https://github.com/mina/linux/commits/tcpdevmem-v17/ v16: https://patchwork.kernel.org/project/netdevbpf/list/?series=866353&state=* ==== v15 got a thorough review and some testing, and this version addresses almost all the feedback. Some more minor comments where the authors said it could be done later, I left out. Major changes: - Addition of dma-buf introspection to page-pool-get and queue-get. - Fixes to selftests suggested by Taehee. - Fixes to documentation suggested by Donald. - A couple of suggestions and fixes to TCP patches by Eric and David. - Fixes to number assignements suggested by Arnd. - Use rtnl_lock()ing to guard against queue reconfiguration while the page_pool initialization is happening. (Jakub). - Fixes to a few warnings reproduced by Taehee. - Fixes to dma-buf binding suggested by Taehee and Jakub. - Fixes to netlink UAPI suggested by Jakub - Applied a number of Reviewed-bys and Acked-bys (including ones I lost from v13+). Full devmem TCP changes including the full GVE driver implementation is here: https://github.com/mina/linux/commits/tcpdevmem-v16/ One caveat: Taehee reproduced a KASAN warning and reported it here: https://lore.kernel.org/netdev/CAMArcTUdCxOBYGF3vpbq=eBvqZfnc44KBaQTN7H-wqdUxZdziw@mail.gmail.com/ I estimate the issue to be minor and easily fixable: https://lore.kernel.org/netdev/CAHS8izNgaqC--GGE2xd85QB=utUnOHmioCsDd1TNxJWKemaD_g@mail.gmail.com/ I hope to be able to follow up with a fix to net tree as net-next closes imminently, but if this iteration doesn't make it in, I will repost with a fix squashed after net-next reopens, no problem. v15: https://patchwork.kernel.org/project/netdevbpf/list/?series=865481&state=* ==== No material changes in this version, only a fix to linking against libynl.a from the last version. Per Jakub's instructions I've pulled one of his patches into this series, and now use the new libynl.a correctly, I hope. As usual, the full devmem TCP changes including the full GVE driver implementation is here: https://github.com/mina/linux/commits/tcpdevmem-v15/ v14: https://patchwork.kernel.org/project/netdevbpf/list/?series=865135&archive=both&state=* ==== No material changes in this version. Only rebase and re-verification on top of net-next. v13, I think, raced with commit ebad6d0334793 ("net/ipv4: Use nested-BH locking for ipv4_tcp_sk.") being merged to net-next that caused a patchwork failure to apply. This series should apply cleanly on commit c4532232fa2a4 ("selftests: net: remove unneeded IP_GRE config"). I did not wait the customary 24hr as Jakub said it's OK to repost as soon as I build test the rebased version: https://lore.kernel.org/netdev/20240625075926.146d769d@kernel.org/ v13: https://patchwork.kernel.org/project/netdevbpf/list/?series=861406&archive=both&state=* ==== Major changes: -------------- This iteration addresses Pavel's review comments, applies his reviewed-by's, and seeks to fix the patchwork build error (sorry!). As usual, the full devmem TCP changes including the full GVE driver implementation is here: https://github.com/mina/linux/commits/tcpdevmem-v13/ v12: https://patchwork.kernel.org/project/netdevbpf/list/?series=859747&state=* ==== Major changes: -------------- This iteration only addresses one minor comment from Pavel with regards to the trace printing of netmem, and the patchwork build error introduced in v11 because I missed doing an allmodconfig build, sorry. Other than that v11, AFAICT, received no feedback. There is one discussion about how the specifics of plugging io uring memory through the page pool, but not relevant to content in this particular patchset, AFAICT. As usual, the full devmem TCP changes including the full GVE driver implementation is here: https://github.com/mina/linux/commits/tcpdevmem-v12/ v11: https://patchwork.kernel.org/project/netdevbpf/list/?series=857457&state=* ==== Major Changes: -------------- v11 addresses feedback received in v10. The major change is the removal of the memory provider ops as requested by Christoph. We still accomplish the same thing, but utilizing direct function calls with if statements rather than generic ops. Additionally address sparse warnings, bugs and review comments from folks that reviewed. As usual, the full devmem TCP changes including the full GVE driver implementation is here: https://github.com/mina/linux/commits/tcpdevmem-v11/ Detailed changelog: ------------------- - Fixes in netdev_rx_queue_restart() from Pavel & David. - Remove commit e650e8c3a36f5 ("net: page_pool: create hooks for custom page providers") from the series to address Christoph's feedback and rebased other patches on the series on this change. - Fixed build errors with CONFIG_DMA_SHARED_BUFFER && !CONFIG_GENERIC_ALLOCATOR build. - Fixed sparse warnings pointed out by Paolo. - Drop unnecessary gro_pull_from_frag0 checks. - Added Bagas reviewed-by to docs. v10: https://patchwork.kernel.org/project/netdevbpf/list/?series=852422&state=* ==== Major Changes: -------------- v9 was sent right before the merge window closed (sorry!). v10 is almost a re-send of the series now that the merge window re-opened. Only rebased to latest net-next and addressed some minor iterative comments received on v9. As usual, the full devmem TCP changes including the full GVE driver implementation is here: https://github.com/mina/linux/commits/tcpdevmem-v10/ Detailed changelog: ------------------- - Fixed tokens leaking in DONTNEED setsockopt (Nikolay). - Moved net_iov_dma_addr() to devmem.c and made it a devmem specific helpers (David). - Rename hook alloc_pages to alloc_netmems as alloc_pages is now preprocessor macro defined and causes a build error. v9: === Major Changes: -------------- GVE queue API has been merged. Submitting this version as non-RFC after rebasing on top of the merged API, and dropped the out of tree queue API I was carrying on github. Addressed the little feedback v8 has received. Detailed changelog: ------------------ - Added new patch from David Wei to this series for netdev_rx_queue_restart() - Fixed sparse error. - Removed CONFIG_ checks in netmem_is_net_iov() - Flipped skb->readable to skb->unreadable - Minor fixes to selftests & docs. RFC v8: ======= Major Changes: -------------- - Fixed build error generated by patch-by-patch build. - Applied docs suggestions from Randy. RFC v7: ======= Major Changes: -------------- This revision largely rebases on top of net-next and addresses the feedback RFCv6 received from folks, namely Jakub, Yunsheng, Arnd, David, & Pavel. The series remains in RFC because the queue-API ndos defined in this series are not yet implemented. I have a GVE implementation I carry out of tree for my testing. A upstreamable GVE implementation is in the works. Aside from that, in my estimation all the patches are ready for review/merge. Please do take a look. As usual the full devmem TCP changes including the full GVE driver implementation is here: https://github.com/mina/linux/commits/tcpdevmem-v7/ Detailed changelog: - Use admin-perm in netlink API. - Addressed feedback from Jakub with regards to netlink API implementation. - Renamed devmem.c functions to something more appropriate for that file. - Improve the performance seen through the page_pool benchmark. - Fix the value definition of all the SO_DEVMEM_* uapi. - Various fixes to documentation. Perf - page-pool benchmark: --------------------------- Improved performance of bench_page_pool_simple.ko tests compared to v6: https://pastebin.com/raw/v5dYRg8L net-next base: 8 cycle fast path. RFC v6: 10 cycle fast path. RFC v7: 9 cycle fast path. RFC v7 with CONFIG_DMA_SHARED_BUFFER disabled: 8 cycle fast path, same as baseline. Perf - Devmem TCP benchmark: --------------------- Perf is about the same regardless of the changes in v7, namely the removal of the static_branch_unlikely to improve the page_pool benchmark performance: 189/200gbps bi-directional throughput with RX devmem TCP and regular TCP TX i.e. ~95% line rate. RFC v6: ======= Major Changes: -------------- This revision largely rebases on top of net-next and addresses the little feedback RFCv5 received. The series remains in RFC because the queue-API ndos defined in this series are not yet implemented. I have a GVE implementation I carry out of tree for my testing. A upstreamable GVE implementation is in the works. Aside from that, in my estimation all the patches are ready for review/merge. Please do take a look. As usual the full devmem TCP changes including the full GVE driver implementation is here: https://github.com/mina/linux/commits/tcpdevmem-v6/ This version also comes with some performance data recorded in the cover letter (see below changelog). Detailed changelog: - Rebased on top of the merged netmem_ref changes. - Converted skb->dmabuf to skb->readable (Pavel). Pavel's original suggestion was to remove the skb->dmabuf flag entirely, but when I looked into it closely, I found the issue that if we remove the flag we have to dereference the shinfo(skb) pointer to obtain the first frag to tell whether an skb is readable or not. This can cause a performance regression if it dirties the cache line when the shinfo(skb) was not really needed. Instead, I converted the skb->dmabuf flag into a generic skb->readable flag which can be re-used by io_uring 0-copy RX. - Squashed a few locking optimizations from Eric Dumazet in the RX path and the DEVMEM_DONTNEED setsockopt. - Expanded the tests a bit. Added validation for invalid scenarios and added some more coverage. Perf - page-pool benchmark: --------------------------- bench_page_pool_simple.ko tests with and without these changes: https://pastebin.com/raw/ncHDwAbn AFAIK the number that really matters in the perf tests is the 'tasklet_page_pool01_fast_path Per elem'. This one measures at about 8 cycles without the changes but there is some 1 cycle noise in some results. With the patches this regresses to 9 cycles with the changes but there is 1 cycle noise occasionally running this test repeatedly. Lastly I tried disable the static_branch_unlikely() in netmem_is_net_iov() check. To my surprise disabling the static_branch_unlikely() check reduces the fast path back to 8 cycles, but the 1 cycle noise remains. Perf - Devmem TCP benchmark: --------------------- 189/200gbps bi-directional throughput with RX devmem TCP and regular TCP TX i.e. ~95% line rate. Major changes in RFC v5: ======================== 1. Rebased on top of 'Abstract page from net stack' series and used the new netmem type to refer to LSB set pointers instead of re-using struct page. 2. Downgraded this series back to RFC and called it RFC v5. This is because this series is now dependent on 'Abstract page from net stack'[1] and the queue API. Both are removed from the series to reduce the patch # and those bits are fairly independent or pre-requisite work. 3. Reworked the page_pool devmem support to use netmem and for some more unified handling. 4. Reworked the reference counting of net_iov (renamed from page_pool_iov) to use pp_ref_count for refcounting. The full changes including the dependent series and GVE page pool support is here: https://github.com/mina/linux/commits/tcpdevmem-rfcv5/ [1] https://patchwork.kernel.org/project/netdevbpf/list/?series=810774 Major changes in v1: ==================== 1. Implemented MVP queue API ndos to remove the userspace-visible driver reset. 2. Fixed issues in the napi_pp_put_page() devmem frag unref path. 3. Removed RFC tag. Many smaller addressed comments across all the patches (patches have individual change log). Full tree including the rest of the GVE driver changes: https://github.com/mina/linux/commits/tcpdevmem-v1 Changes in RFC v3: ================== 1. Pulled in the memory-provider dependency from Jakub's RFC[1] to make the series reviewable and mergeable. 2. Implemented multi-rx-queue binding which was a todo in v2. 3. Fix to cmsg handling. The sticking point in RFC v2[2] was the device reset required to refill the device rx-queues after the dmabuf bind/unbind. The solution suggested as I understand is a subset of the per-queue management ops Jakub suggested or similar: https://lore.kernel.org/netdev/20230815171638.4c057dcd@kernel.org/ This is not addressed in this revision, because: 1. This point was discussed at netconf & netdev and there is openness to using the current approach of requiring a device reset. 2. Implementing individual queue resetting seems to be difficult for my test bed with GVE. My prototype to test this ran into issues with the rx-queues not coming back up properly if reset individually. At the moment I'm unsure if it's a mistake in the POC or a genuine issue in the virtualization stack behind GVE, which currently doesn't test individual rx-queue restart. 3. Our usecases are not bothered by requiring a device reset to refill the buffer queues, and we'd like to support NICs that run into this limitation with resetting individual queues. My thought is that drivers that have trouble with per-queue configs can use the support in this series, while drivers that support new netdev ops to reset individual queues can automatically reset the queue as part of the dma-buf bind/unbind. The same approach with device resets is presented again for consideration with other sticking points addressed. This proposal includes the rx devmem path only proposed for merge. For a snapshot of my entire tree which includes the GVE POC page pool support & device memory support: https://github.com/torvalds/linux/compare/master...mina:linux:tcpdevmem-v3 [1] https://lore.kernel.org/netdev/f8270765-a27b-6ccf-33ea-cda097168d79@redhat.com/T/ [2] https://lore.kernel.org/netdev/CAHS8izOVJGJH5WF68OsRWFKJid1_huzzUK+hpKbLcL4pSOD1Jw@mail.gmail.com/T/ Changes in RFC v2: ================== The sticking point in RFC v1[1] was the dma-buf pages approach we used to deliver the device memory to the TCP stack. RFC v2 is a proof-of-concept that attempts to resolve this by implementing scatterlist support in the networking stack, such that we can import the dma-buf scatterlist directly. This is the approach proposed at a high level here[2]. Detailed changes: 1. Replaced dma-buf pages approach with importing scatterlist into the page pool. 2. Replace the dma-buf pages centric API with a netlink API. 3. Removed the TX path implementation - there is no issue with implementing the TX path with scatterlist approach, but leaving out the TX path makes it easier to review. 4. Functionality is tested with this proposal, but I have not conducted perf testing yet. I'm not sure there are regressions, but I removed perf claims from the cover letter until they can be re-confirmed. 5. Added Signed-off-by: contributors to the implementation. 6. Fixed some bugs with the RX path since RFC v1. Any feedback welcome, but specifically the biggest pending questions needing feedback IMO are: 1. Feedback on the scatterlist-based approach in general. 2. Netlink API (Patch 1 & 2). 3. Approach to handle all the drivers that expect to receive pages from the page pool (Patch 6). [1] https://lore.kernel.org/netdev/dfe4bae7-13a0-3c5d-d671-f61b375cb0b4@gmail.com/T/ [2] https://lore.kernel.org/netdev/CAHS8izPm6XRS54LdCDZVd0C75tA1zHSu6jLVO8nzTLXCc=H7Nw@mail.gmail.com/ ================== * TL;DR: Device memory TCP (devmem TCP) is a proposal for transferring data to and/or from device memory efficiently, without bouncing the data to a host memory buffer. * Problem: A large amount of data transfers have device memory as the source and/or destination. Accelerators drastically increased the volume of such transfers. Some examples include: - ML accelerators transferring large amounts of training data from storage into GPU/TPU memory. In some cases ML training setup time can be as long as 50% of TPU compute time, improving data transfer throughput & efficiency can help improving GPU/TPU utilization. - Distributed training, where ML accelerators, such as GPUs on different hosts, exchange data among them. - Distributed raw block storage applications transfer large amounts of data with remote SSDs, much of this data does not require host processing. Today, the majority of the Device-to-Device data transfers the network are implemented as the following low level operations: Device-to-Host copy, Host-to-Host network transfer, and Host-to-Device copy. The implementation is suboptimal, especially for bulk data transfers, and can put significant strains on system resources, such as host memory bandwidth, PCIe bandwidth, etc. One important reason behind the current state is the kernel’s lack of semantics to express device to network transfers. * Proposal: In this patch series we attempt to optimize this use case by implementing socket APIs that enable the user to: 1. send device memory across the network directly, and 2. receive incoming network packets directly into device memory. Packet _payloads_ go directly from the NIC to device memory for receive and from device memory to NIC for transmit. Packet _headers_ go to/from host memory and are processed by the TCP/IP stack normally. The NIC _must_ support header split to achieve this. Advantages: - Alleviate host memory bandwidth pressure, compared to existing network-transfer + device-copy semantics. - Alleviate PCIe BW pressure, by limiting data transfer to the lowest level of the PCIe tree, compared to traditional path which sends data through the root complex. * Patch overview: ** Part 1: netlink API Gives user ability to bind dma-buf to an RX queue. ** Part 2: scatterlist support Currently the standard for device memory sharing is DMABUF, which doesn't generate struct pages. On the other hand, networking stack (skbs, drivers, and page pool) operate on pages. We have 2 options: 1. Generate struct pages for dmabuf device memory, or, 2. Modify the networking stack to process scatterlist. Approach #1 was attempted in RFC v1. RFC v2 implements approach #2. ** part 3: page pool support We piggy back on page pool memory providers proposal: https://github.com/kuba-moo/linux/tree/pp-providers It allows the page pool to define a memory provider that provides the page allocation and freeing. It helps abstract most of the device memory TCP changes from the driver. ** part 4: support for unreadable skb frags Page pool iovs are not accessible by the host; we implement changes throughput the networking stack to correctly handle skbs with unreadable frags. ** Part 5: recvmsg() APIs We define user APIs for the user to send and receive device memory. Not included with this series is the GVE devmem TCP support, just to simplify the review. Code available here if desired: https://github.com/mina/linux/tree/tcpdevmem This series is built on top of net-next with Jakub's pp-providers changes cherry-picked. * NIC dependencies: 1. (strict) Devmem TCP require the NIC to support header split, i.e. the capability to split incoming packets into a header + payload and to put each into a separate buffer. Devmem TCP works by using device memory for the packet payload, and host memory for the packet headers. 2. (optional) Devmem TCP works better with flow steering support & RSS support, i.e. the NIC's ability to steer flows into certain rx queues. This allows the sysadmin to enable devmem TCP on a subset of the rx queues, and steer devmem TCP traffic onto these queues and non devmem TCP elsewhere. The NIC I have access to with these properties is the GVE with DQO support running in Google Cloud, but any NIC that supports these features would suffice. I may be able to help reviewers bring up devmem TCP on their NICs. * Testing: The series includes a udmabuf kselftest that show a simple use case of devmem TCP and validates the entire data path end to end without a dependency on a specific dmabuf provider. ** Test Setup Kernel: net-next with this series and memory provider API cherry-picked locally. Hardware: Google Cloud A3 VMs. NIC: GVE with header split & RSS & flow steering support. Cc: Pavel Begunkov <asml.silence@gmail.com> Cc: David Wei <dw@davidwei.uk> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Yunsheng Lin <linyunsheng@huawei.com> Cc: Shailend Chand <shailend@google.com> Cc: Harshitha Ramamurthy <hramamurthy@google.com> Cc: Shakeel Butt <shakeel.butt@linux.dev> Cc: Jeroen de Borst <jeroendb@google.com> Cc: Praveen Kaligineedi <pkaligineedi@google.com> Cc: Bagas Sanjaya <bagasdotme@gmail.com> Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Christoph Hellwig <hch@infradead.org> Cc: Nikolay Aleksandrov <razor@blackwall.org> Cc: Taehee Yoo <ap420073@gmail.com> Cc: Donald Hunter <donald.hunter@gmail.com> Mina Almasry (13): netdev: add netdev_rx_queue_restart() net: netdev netlink api to bind dma-buf to a net device netdev: support binding dma-buf to netdevice netdev: netdevice devmem allocator page_pool: devmem support memory-provider: dmabuf devmem memory provider net: support non paged skb frags net: add support for skbs with unreadable frags tcp: RX path for devmem TCP net: add SO_DEVMEM_DONTNEED setsockopt to release RX frags net: add devmem TCP documentation selftests: add ncdevmem, netcat for devmem TCP netdev: add dmabuf introspection Documentation/netlink/specs/netdev.yaml | 61 +++ Documentation/networking/devmem.rst | 269 +++++++++++ Documentation/networking/index.rst | 1 + arch/alpha/include/uapi/asm/socket.h | 6 + arch/mips/include/uapi/asm/socket.h | 6 + arch/parisc/include/uapi/asm/socket.h | 6 + arch/sparc/include/uapi/asm/socket.h | 6 + include/linux/netdevice.h | 2 + include/linux/skbuff.h | 61 ++- include/linux/skbuff_ref.h | 9 +- include/linux/socket.h | 1 + include/net/netdev_rx_queue.h | 5 + include/net/netmem.h | 132 +++++- include/net/page_pool/helpers.h | 39 +- include/net/page_pool/types.h | 23 +- include/net/sock.h | 2 + include/net/tcp.h | 3 +- include/trace/events/page_pool.h | 12 +- include/uapi/asm-generic/socket.h | 6 + include/uapi/linux/netdev.h | 13 + include/uapi/linux/uio.h | 17 + net/Kconfig | 5 + net/core/Makefile | 2 + net/core/datagram.c | 6 + net/core/dev.c | 33 +- net/core/devmem.c | 389 ++++++++++++++++ net/core/devmem.h | 180 ++++++++ net/core/gro.c | 3 +- net/core/mp_dmabuf_devmem.h | 44 ++ net/core/netdev-genl-gen.c | 23 + net/core/netdev-genl-gen.h | 6 + net/core/netdev-genl.c | 139 +++++- net/core/netdev_rx_queue.c | 81 ++++ net/core/netmem_priv.h | 31 ++ net/core/page_pool.c | 120 +++-- net/core/page_pool_priv.h | 46 ++ net/core/page_pool_user.c | 32 +- net/core/skbuff.c | 77 +++- net/core/sock.c | 68 +++ net/ethtool/common.c | 8 + net/ipv4/esp4.c | 3 +- net/ipv4/tcp.c | 263 ++++++++++- net/ipv4/tcp_input.c | 13 +- net/ipv4/tcp_ipv4.c | 16 + net/ipv4/tcp_minisocks.c | 2 + net/ipv4/tcp_output.c | 5 +- net/ipv6/esp6.c | 3 +- net/packet/af_packet.c | 4 +- net/xdp/xsk_buff_pool.c | 5 + tools/include/uapi/linux/netdev.h | 13 + tools/net/ynl/lib/.gitignore | 1 + tools/testing/selftests/net/.gitignore | 1 + tools/testing/selftests/net/Makefile | 9 + tools/testing/selftests/net/ncdevmem.c | 570 ++++++++++++++++++++++++ 54 files changed, 2757 insertions(+), 124 deletions(-) create mode 100644 Documentation/networking/devmem.rst create mode 100644 net/core/devmem.c create mode 100644 net/core/devmem.h create mode 100644 net/core/mp_dmabuf_devmem.h create mode 100644 net/core/netdev_rx_queue.c create mode 100644 net/core/netmem_priv.h create mode 100644 tools/testing/selftests/net/ncdevmem.c