- Aug 09, 2021
-
-
Yunsheng Lin authored
Currently page pool only support page recycling when there is only one user of the page, and the split page reusing implemented in the most driver can not use the page pool as bing-pong way of reusing requires the multi user support in page pool. Those reusing or recycling has below limitations: 1. page from page pool can only be used be one user in order for the page recycling to happen. 2. Bing-pong way of reusing in most driver does not support multi desc using different part of the same page in order to save memory. So add multi-users support and frag page recycling in page pool to overcome the above limitation. Signed-off-by:
Yunsheng Lin <linyunsheng@huawei.com> Signed-off-by:
Jakub Kicinski <kuba@kernel.org>
-
Yunsheng Lin authored
For 32 bit systems with 64 bit dma, dma_addr[1] is used to store the upper 32 bit dma addr, those system should be rare those days. For normal system, the dma_addr[1] in 'struct page' is not used, so we can reuse dma_addr[1] for storing frag count, which means how many frags this page might be splited to. In order to simplify the page frag support in the page pool, the PAGE_POOL_DMA_USE_PP_FRAG_COUNT macro is added to indicate the 32 bit systems with 64 bit dma, and the page frag support in page pool is disabled for such system. The newly added page_pool_set_frag_count() is called to reserve the maximum frag count before any page frag is passed to the user. The page_pool_atomic_sub_frag_count_return() is called when user is done with the page frag. Signed-off-by:
-
Yunsheng Lin authored
Currently, page->pp is cleared and set everytime the page is recycled, which is unnecessary. So only set the page->pp when the page is added to the page pool and only clear it when the page is released from the page pool. This is also a preparation to support allocating frag page in page pool. Reviewed-by:
Ilias Apalodimas <ilias.apalodimas@linaro.org> Signed-off-by:
-
- Jun 07, 2021
-
-
Ilias Apalodimas authored
Up to now several high speed NICs have custom mechanisms of recycling the allocated memory they use for their payloads. Our page_pool API already has recycling capabilities that are always used when we are running in 'XDP mode'. So let's tweak the API and the kernel network stack slightly and allow the recycling to happen even during the standard operation. The API doesn't take into account 'split page' policies used by those drivers currently, but can be extended once we have users for that. The idea is to be able to intercept the packet on skb_release_data(). If it's a buffer coming from our page_pool API recycle it back to the pool for further usage or just release the packet entirely. To achieve that we introduce a bit in struct sk_buff (pp_recycle:1) and a field in struct page (page->pp) to store the page_pool pointer. Storing the information in page->pp allows us to recycle both SKBs and their fragments. We could have skipped the skb bit entirely, since identical information can bederived from struct page. However, in an effort to affect the free path as less as possible, reading a single bit in the skb which is already in cache, is better that trying to derive identical information for the page stored data. The driver or page_pool has to take care of the sync operations on it's own during the buffer recycling since the buffer is, after opting-in to the recycling, never unmapped. Since the gain on the drivers depends on the architecture, we are not enabling recycling by default if the page_pool API is used on a driver. In order to enable recycling the driver must call skb_mark_for_recycle() to store the information we need for recycling in page->pp and enabling the recycling bit, or page_pool_store_mem_info() for a fragment. Co-developed-by:
Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by:
Matteo Croce <mcroce@microsoft.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
- May 14, 2021
-
-
Matthew Wilcox (Oracle) authored
32-bit architectures which expect 8-byte alignment for 8-byte integers and need 64-bit DMA addresses (arm, mips, ppc) had their struct page inadvertently expanded in 2019. When the dma_addr_t was added, it forced the alignment of the union to 8 bytes, which inserted a 4 byte gap between 'flags' and the union. Fix this by storing the dma_addr_t in one or two adjacent unsigned longs. This restores the alignment to that of an unsigned long. We always store the low bits in the first word to prevent the PageTail bit from being inadvertently set on a big endian platform. If that happened, get_user_pages_fast() racing against a page which was freed and reallocated to the page_pool could dereference a bogus compound_head(), which would be hard to trace back to this cause. Link: https://lkml.kernel.org/r/20210510153211.1504886-1-willy@infradead.org Fixes: c25fff71 ("mm: add dma_addr_t to struct page") Signed-off-by:
Matthew Wilcox (Oracle) <willy@infradead.org> Acked-by:
Vlastimil Babka <vbabka@suse.cz> Tested-by:
Matteo Croce <mcroce@linux.microsoft.com> Cc: <stable@vger.kernel.org> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
- Apr 30, 2021
-
-
Jesper Dangaard Brouer authored
There are cases where the page_pool need to refill with pages from the page allocator. Some workloads cause the page_pool to release pages instead of recycling these pages. For these workload it can improve performance to bulk alloc pages from the page-allocator to refill the alloc cache. For XDP-redirect workload with 100G mlx5 driver (that use page_pool) redirecting xdp_frame packets into a veth, that does XDP_PASS to create an SKB from the xdp_frame, which then cannot return the page to the page_pool. Performance results under GitHub xdp-project[1]: [1] https://github.com/xdp-project/xdp-project/blob/master/areas/mem/page_pool06_alloc_pages_bulk.org Mel: The patch "net: page_pool: convert to use alloc_pages_bulk_array variant" was squashed with this patch. From the test page, the array variant was superior with one of the test results as follows. Kernel XDP stats CPU pps Delta Baseline XDP-RX CPU total 3,771,0...
-
- Nov 13, 2020
-
-
Lorenzo Bianconi authored
Introduce the capability to batch page_pool ptr_ring refill since it is usually run inside the driver NAPI tx completion loop. Suggested-by:
Lorenzo Bianconi <lorenzo@kernel.org> Signed-off-by:
Daniel Borkmann <daniel@iogearbox.net> Acked-by:
John Fastabend <john.fastabend@gmail.com> Acked-by:
-
- Feb 20, 2020
-
-
Ilias Apalodimas authored
Functions starting with __ usually indicate those which are exported, but should not be called directly. Update some of those declared in the API and make it more readable. page_pool_unmap_page() and page_pool_release_page() were doing exactly the same thing calling __page_pool_clean_page(). Let's rename __page_pool_clean_page() to page_pool_release_page() and export it in order to show up on perf logs and get rid of page_pool_unmap_page(). Finally rename __page_pool_put_page() to page_pool_put_page() since we can now directly call it from drivers and rename the existing page_pool_put_page() to page_pool_put_full_page() since they do the same thing but the latter is trying to sync the full DMA area. This patch also updates netsec, mvneta and stmmac drivers which use those functions. Suggested-by:
Jonathan Lemon <jonathan.lemon@gmail.com> Acked-by:
Toke Høiland-Jørgensen <toke@redhat.com> Acked-by:
-
- Nov 20, 2019
-
-
Lorenzo Bianconi authored
Introduce the following parameters in order to add the possibility to sync DMA memory for device before putting allocated pages in the page_pool caches: - PP_FLAG_DMA_SYNC_DEV: if set in page_pool_params flags, all pages that the driver gets from page_pool will be DMA-synced-for-device according to the length provided by the device driver. Please note DMA-sync-for-CPU is still device driver responsibility - offset: DMA address offset where the DMA engine starts copying rx data - max_len: maximum DMA memory size page_pool is allowed to flush. This is currently used in __page_pool_alloc_pages_slow routine when pages are allocated from page allocator These parameters are supposed to be set by device drivers. This optimization reduces the length of the DMA-sync-for-device. The optimization is valid because pages are initially DMA-synced-for-device as defined via max_len. At RX time, the driver will perform a DMA-sync-for-CPU on the memory for the packet length. What is important is the memory occupied by packet payload, because this is the area CPU is allowed to read and modify. As we don't track cache-lines written into by the CPU, simply use the packet payload length as dma_sync_size at page_pool recycle time. This also take into account any tail-extend. Tested-by:
Matteo Croce <mcroce@redhat.com> Signed-off-by:
-
Saeed Mahameed authored
Add page_pool_update_nid() to be called by page pool consumers when they detect numa node changes. It will update the page pool nid value to start allocating from the new effective numa node. This is to mitigate page pool allocating pages from a wrong numa node, where the pool was originally allocated, and holding on to pages that belong to a different numa node, which causes performance degradation. For pages that are already being consumed and could be returned to the pool by the consumer, in next patch we will add a check per page to avoid recycling them back to the pool and return them to the page allocator. Signed-off-by:
Saeed Mahameed <saeedm@mellanox.com> Acked-by:
-
- Nov 18, 2019
-
-
Jesper Dangaard Brouer authored
When Jonathan change the page_pool to become responsible to its own shutdown via deferred work queue, then the disconnect_cnt counter was removed from xdp memory model tracepoint. This patch change the page_pool_inflight tracepoint name to page_pool_release, because it reflects the new responsability better. And it reintroduces a counter that reflect the number of times page_pool_release have been tried. The counter is also used by the code, to only empty the alloc cache once. With a stuck work queue running every second and counter being 64-bit, it will overrun in approx 584 billion years. For comparison, Earth lifetime expectancy is 7.5 billion years, before the Sun will engulf, and destroy, the Earth. Signed-off-by:
-
- Nov 16, 2019
-
-
Jonathan Lemon authored
The page pool keeps track of the number of pages in flight, and it isn't safe to remove the pool until all pages are returned. Disallow removing the pool until all pages are back, so the pool is always available for page producers. Make the page pool responsible for its own delayed destruction instead of relying on XDP, so the page pool can be used without the xdp memory model. When all pages are returned, free the pool and notify xdp if the pool is registered with the xdp memory system. Have the callback perform a table walk since some drivers (cpsw) may share the pool among multiple xdp_rxq_info. Note that the increment of pages_state_release_cnt may result in inflight == 0, resulting in the pool being released. Fixes: d956a048 ("xdp: force mem allocator removal and periodic warning") Signed-off-by:
-
- Jul 08, 2019
-
-
Ivan Khoronzhuk authored
Jesper recently removed page_pool_destroy() (from driver invocation) and moved shutdown and free of page_pool into xdp_rxq_info_unreg(), in-order to handle in-flight packets/pages. This created an asymmetry in drivers create/destroy pairs. This patch reintroduce page_pool_destroy and add page_pool user refcnt. This serves the purpose to simplify drivers error handling as driver now drivers always calls page_pool_destroy() and don't need to track if xdp_rxq_info_reg_mem_model() was unsuccessful. This could be used for a special cases where a single RX-queue (with a single page_pool) provides packets for two net_device'es, and thus needs to register the same page_pool twice with two xdp_rxq_info structures. This patch is primarily to ease API usage for drivers. The recently merged netsec driver, actually have a bug in this area, which is solved by this API change. This patch is a modified version of Ivan Khoronzhuk's original patch. Link: https://lore.kernel.org/netdev/20190625175948.24771-2-ivan.khoronzhuk@linaro.org/ Fixes: 5c67bf0e ("net: netsec: Use page_pool API") Signed-off-by:
Ivan Khoronzhuk <ivan.khoronzhuk@linaro.org> Signed-off-by:
-
- Jul 01, 2019
-
-
Ilias Apalodimas authored
Since the dma direction is stored in page pool params, offer an API helper for driver that choose not to keep track of it locally Signed-off-by:
-
- Jun 19, 2019
-
-
Jesper Dangaard Brouer authored
Kbuild test robot reported compile warning: warning: no return statement in function returning non-void in function page_pool_request_shutdown, when CONFIG_PAGE_POOL is disabled. The fix makes the code a little more verbose, with a descriptive variable. Fixes: 99c07c43 ("xdp: tracking page_pool resources and safe removal") Signed-off-by:
kbuild test robot <lkp@intel.com> Signed-off-by:
-
Jesper Dangaard Brouer authored
This patch is needed before we can allow drivers to use page_pool for DMA-mappings. Today with page_pool and XDP return API, it is possible to remove the page_pool object (from rhashtable), while there are still in-flight packet-pages. This is safely handled via RCU and failed lookups in __xdp_return() fallback to call put_page(), when page_pool object is gone. In-case page is still DMA mapped, this will result in page note getting correctly DMA unmapped. To solve this, the page_pool is extended with tracking in-flight pages. And XDP disconnect system queries page_pool and waits, via workqueue, for all in-flight pages to be returned. To avoid killing performance when tracking in-flight pages, the implement use two (unsigned) counters, that in placed on different cache-lines, and can be used to deduct in-flight packets. This is done by mapping the unsigned "sequence" counters onto signed Two's complement arithmetic operations. This is e.g. used by kernel's time_after macros, described in kernel commit 1ba3aab3 and 5a581b36 , and also explained in RFC1982. The trick is these two incrementing counters only need to be read and compared, when checking if it's safe to free the page_pool structure. Which will only happen when driver have disconnected RX/alloc side. Thus, on a non-fast-path. It is chosen that page_pool tracking is also enabled for the non-DMA use-case, as this can be used for statistics later. After this patch, using page_pool requires more strict resource "release", e.g. via page_pool_release_page() that was introduced in this patchset, and previous patches implement/fix this more strict requirement. Drivers no-longer call page_pool_destroy(). Drivers already call xdp_rxq_info_unreg() which call xdp_rxq_info_unreg_mem_model(), which will attempt to disconnect the mem id, and if attempt fails schedule the disconnect for later via delayed workqueue. Signed-off-by:
-
Jesper Dangaard Brouer authored
In case driver fails to register the page_pool with XDP return API (via xdp_rxq_info_reg_mem_model()), then the driver can free the page_pool resources more directly than calling page_pool_destroy(), which does a unnecessarily RCU free procedure. This patch is preparing for removing page_pool_destroy(), from driver invocation. Signed-off-by:
Tariq Toukan <tariqt@mellanox.com> Signed-off-by:
-
Jesper Dangaard Brouer authored
When converting an xdp_frame into an SKB, and sending this into the network stack, then the underlying XDP memory model need to release associated resources, because the network stack don't have callbacks for XDP memory models. The only memory model that needs this is page_pool, when a driver use the DMA-mapping feature. Introduce page_pool_release_page(), which basically does the same as page_pool_unmap_page(). Add xdp_release_frame() as the XDP memory model interface for calling it, if the memory model match MEM_TYPE_PAGE_POOL, to save the function call overhead for others. Have cpumap call xdp_release_frame() before xdp_scrub_frame(). Signed-off-by:
-
Ilias Apalodimas authored
On a previous patch dma addr was stored in 'struct page'. Use that to unmap DMA addresses used by network drivers Signed-off-by:
-
Ilias Apalodimas authored
On a previous patch dma addr was stored in 'struct page'. Use that to retrieve DMA addresses used by network drivers Signed-off-by:
-
- May 24, 2018
-
-
Jesper Dangaard Brouer authored
When sending an xdp_frame through xdp_do_redirect call, then error cases can happen where the xdp_frame needs to be dropped, and returning an -errno code isn't sufficient/possible any-longer (e.g. for cpumap case). This is already fully supported, by simply calling xdp_return_frame. This patch is an optimization, which provides xdp_return_frame_rx_napi, which is a faster variant for these error cases. It take advantage of the protection provided by XDP RX running under NAPI protection. This change is mostly relevant for drivers using the page_pool allocator as it can take advantage of this. (Tested with mlx5). Signed-off-by:
Alexei Starovoitov <ast@kernel.org>
-
- Apr 17, 2018
-
-
Jesper Dangaard Brouer authored
New allocator type MEM_TYPE_PAGE_POOL for page_pool usage. The registered allocator page_pool pointer is not available directly from xdp_rxq_info, but it could be (if needed). For now, the driver should keep separate track of the page_pool pointer, which it should use for RX-ring page allocation. As suggested by Saeed, to maintain a symmetric API it is the drivers responsibility to allocate/create and free/destroy the page_pool. Thus, after the driver have called xdp_rxq_info_unreg(), it is drivers responsibility to free the page_pool, but with a RCU free call. This is done easily via the page_pool helper page_pool_destroy() (which avoids touching any driver code during the RCU callback, which could happen after the driver have been unloaded). V8: address issues found by kbuild test robot - Address sparse should be static warnings - Allow xdp.o to be compiled without page_pool.o V9: Remove inline from .c file, compiler knows best Signed-off-by:
-
Jesper Dangaard Brouer authored
Need a fast page recycle mechanism for ndo_xdp_xmit API for returning pages on DMA-TX completion time, which have good cross CPU performance, given DMA-TX completion time can happen on a remote CPU. Refurbish my page_pool code, that was presented[1] at MM-summit 2016. Adapted page_pool code to not depend the page allocator and integration into struct page. The DMA mapping feature is kept, even-though it will not be activated/used in this patchset. [1] http://people.netfilter.org/hawk/presentations/MM-summit2016/generic_page_pool_mm_summit2016.pdf V2: Adjustments requested by Tariq - Changed page_pool_create return codes, don't return NULL, only ERR_PTR, as this simplifies err handling in drivers. V4: many small improvements and cleanups - Add DOC comment section, that can be used by kernel-doc - Improve fallback mode, to work better with refcnt based recycling e.g. remove a WARN as pointed out by Tariq e.g. quicker fallback if ptr_ring is empty. V5: Fixed SPDX license as pointed out by Alexei V6: Adjustments requested by Eric Dumazet - Adjust ____cacheline_aligned_in_smp usage/placement - Move rcu_head in struct page_pool - Free pages quicker on destroy, minimize resources delayed an RCU period - Remove code for forward/backward compat ABI interface V8: Issues found by kbuild test robot - Address sparse should be static warnings - Only compile+link when a driver use/select page_pool, mlx5 selects CONFIG_PAGE_POOL, although its first used in two patches Signed-off-by:
-
