linux-stable/net/core/xdp.c

513 lines
12 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-only
/* net/core/xdp.c
*
* Copyright (c) 2017 Jesper Dangaard Brouer, Red Hat Inc.
*/
#include <linux/bpf.h>
#include <linux/filter.h>
#include <linux/types.h>
#include <linux/mm.h>
#include <linux/netdevice.h>
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
#include <linux/slab.h>
#include <linux/idr.h>
#include <linux/rhashtable.h>
#include <linux/bug.h>
#include <net/page_pool.h>
#include <net/xdp.h>
#include <net/xdp_priv.h> /* struct xdp_mem_allocator */
#include <trace/events/xdp.h>
xsk: Introduce AF_XDP buffer allocation API In order to simplify AF_XDP zero-copy enablement for NIC driver developers, a new AF_XDP buffer allocation API is added. The implementation is based on a single core (single producer/consumer) buffer pool for the AF_XDP UMEM. A buffer is allocated using the xsk_buff_alloc() function, and returned using xsk_buff_free(). If a buffer is disassociated with the pool, e.g. when a buffer is passed to an AF_XDP socket, a buffer is said to be released. Currently, the release function is only used by the AF_XDP internals and not visible to the driver. Drivers using this API should register the XDP memory model with the new MEM_TYPE_XSK_BUFF_POOL type. The API is defined in net/xdp_sock_drv.h. The buffer type is struct xdp_buff, and follows the lifetime of regular xdp_buffs, i.e. the lifetime of an xdp_buff is restricted to a NAPI context. In other words, the API is not replacing xdp_frames. In addition to introducing the API and implementations, the AF_XDP core is migrated to use the new APIs. rfc->v1: Fixed build errors/warnings for m68k and riscv. (kbuild test robot) Added headroom/chunk size getter. (Maxim/Björn) v1->v2: Swapped SoBs. (Maxim) v2->v3: Initialize struct xdp_buff member frame_sz. (Björn) Add API to query the DMA address of a frame. (Maxim) Do DMA sync for CPU till the end of the frame to handle possible growth (frame_sz). (Maxim) Signed-off-by: Björn Töpel <bjorn.topel@intel.com> Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org> Link: https://lore.kernel.org/bpf/20200520192103.355233-6-bjorn.topel@gmail.com
2020-05-20 19:20:53 +00:00
#include <net/xdp_sock_drv.h>
#define REG_STATE_NEW 0x0
#define REG_STATE_REGISTERED 0x1
#define REG_STATE_UNREGISTERED 0x2
#define REG_STATE_UNUSED 0x3
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
static DEFINE_IDA(mem_id_pool);
static DEFINE_MUTEX(mem_id_lock);
#define MEM_ID_MAX 0xFFFE
#define MEM_ID_MIN 1
static int mem_id_next = MEM_ID_MIN;
static bool mem_id_init; /* false */
static struct rhashtable *mem_id_ht;
static u32 xdp_mem_id_hashfn(const void *data, u32 len, u32 seed)
{
const u32 *k = data;
const u32 key = *k;
BUILD_BUG_ON(sizeof_field(struct xdp_mem_allocator, mem.id)
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
!= sizeof(u32));
/* Use cyclic increasing ID as direct hash key */
return key;
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
}
static int xdp_mem_id_cmp(struct rhashtable_compare_arg *arg,
const void *ptr)
{
const struct xdp_mem_allocator *xa = ptr;
u32 mem_id = *(u32 *)arg->key;
return xa->mem.id != mem_id;
}
static const struct rhashtable_params mem_id_rht_params = {
.nelem_hint = 64,
.head_offset = offsetof(struct xdp_mem_allocator, node),
.key_offset = offsetof(struct xdp_mem_allocator, mem.id),
.key_len = sizeof_field(struct xdp_mem_allocator, mem.id),
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
.max_size = MEM_ID_MAX,
.min_size = 8,
.automatic_shrinking = true,
.hashfn = xdp_mem_id_hashfn,
.obj_cmpfn = xdp_mem_id_cmp,
};
static void __xdp_mem_allocator_rcu_free(struct rcu_head *rcu)
{
struct xdp_mem_allocator *xa;
xa = container_of(rcu, struct xdp_mem_allocator, rcu);
/* Allow this ID to be reused */
ida_simple_remove(&mem_id_pool, xa->mem.id);
kfree(xa);
}
static void mem_xa_remove(struct xdp_mem_allocator *xa)
xdp: tracking page_pool resources and safe removal 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 1ba3aab3033b and 5a581b367b5, 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 <brouer@redhat.com> Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-06-18 13:05:47 +00:00
{
trace_mem_disconnect(xa);
xdp: tracking page_pool resources and safe removal 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 1ba3aab3033b and 5a581b367b5, 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 <brouer@redhat.com> Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-06-18 13:05:47 +00:00
if (!rhashtable_remove_fast(mem_id_ht, &xa->node, mem_id_rht_params))
xdp: tracking page_pool resources and safe removal 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 1ba3aab3033b and 5a581b367b5, 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 <brouer@redhat.com> Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-06-18 13:05:47 +00:00
call_rcu(&xa->rcu, __xdp_mem_allocator_rcu_free);
}
static void mem_allocator_disconnect(void *allocator)
{
struct xdp_mem_allocator *xa;
struct rhashtable_iter iter;
mutex_lock(&mem_id_lock);
rhashtable_walk_enter(mem_id_ht, &iter);
do {
rhashtable_walk_start(&iter);
while ((xa = rhashtable_walk_next(&iter)) && !IS_ERR(xa)) {
if (xa->allocator == allocator)
mem_xa_remove(xa);
}
rhashtable_walk_stop(&iter);
xdp: tracking page_pool resources and safe removal 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 1ba3aab3033b and 5a581b367b5, 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 <brouer@redhat.com> Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-06-18 13:05:47 +00:00
} while (xa == ERR_PTR(-EAGAIN));
rhashtable_walk_exit(&iter);
mutex_unlock(&mem_id_lock);
}
static void mem_id_disconnect(int id)
xdp: tracking page_pool resources and safe removal 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 1ba3aab3033b and 5a581b367b5, 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 <brouer@redhat.com> Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-06-18 13:05:47 +00:00
{
struct xdp_mem_allocator *xa;
mutex_lock(&mem_id_lock);
xdp: tracking page_pool resources and safe removal 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 1ba3aab3033b and 5a581b367b5, 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 <brouer@redhat.com> Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-06-18 13:05:47 +00:00
xa = rhashtable_lookup_fast(mem_id_ht, &id, mem_id_rht_params);
if (!xa) {
mutex_unlock(&mem_id_lock);
WARN(1, "Request remove non-existing id(%d), driver bug?", id);
xdp: tracking page_pool resources and safe removal 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 1ba3aab3033b and 5a581b367b5, 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 <brouer@redhat.com> Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-06-18 13:05:47 +00:00
return;
}
xdp: tracking page_pool resources and safe removal 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 1ba3aab3033b and 5a581b367b5, 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 <brouer@redhat.com> Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-06-18 13:05:47 +00:00
trace_mem_disconnect(xa);
if (!rhashtable_remove_fast(mem_id_ht, &xa->node, mem_id_rht_params))
call_rcu(&xa->rcu, __xdp_mem_allocator_rcu_free);
mutex_unlock(&mem_id_lock);
xdp: tracking page_pool resources and safe removal 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 1ba3aab3033b and 5a581b367b5, 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 <brouer@redhat.com> Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-06-18 13:05:47 +00:00
}
void xdp_rxq_info_unreg_mem_model(struct xdp_rxq_info *xdp_rxq)
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
{
struct xdp_mem_allocator *xa;
int id = xdp_rxq->mem.id;
if (xdp_rxq->reg_state != REG_STATE_REGISTERED) {
WARN(1, "Missing register, driver bug");
return;
}
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
if (id == 0)
return;
if (xdp_rxq->mem.type == MEM_TYPE_ZERO_COPY)
return mem_id_disconnect(id);
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
if (xdp_rxq->mem.type == MEM_TYPE_PAGE_POOL) {
rcu_read_lock();
xa = rhashtable_lookup(mem_id_ht, &id, mem_id_rht_params);
page_pool_destroy(xa->page_pool);
rcu_read_unlock();
xdp: tracking page_pool resources and safe removal 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 1ba3aab3033b and 5a581b367b5, 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 <brouer@redhat.com> Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-06-18 13:05:47 +00:00
}
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
}
EXPORT_SYMBOL_GPL(xdp_rxq_info_unreg_mem_model);
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
void xdp_rxq_info_unreg(struct xdp_rxq_info *xdp_rxq)
{
/* Simplify driver cleanup code paths, allow unreg "unused" */
if (xdp_rxq->reg_state == REG_STATE_UNUSED)
return;
WARN(!(xdp_rxq->reg_state == REG_STATE_REGISTERED), "Driver BUG");
xdp_rxq_info_unreg_mem_model(xdp_rxq);
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
xdp_rxq->reg_state = REG_STATE_UNREGISTERED;
xdp_rxq->dev = NULL;
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
/* Reset mem info to defaults */
xdp_rxq->mem.id = 0;
xdp_rxq->mem.type = 0;
}
EXPORT_SYMBOL_GPL(xdp_rxq_info_unreg);
static void xdp_rxq_info_init(struct xdp_rxq_info *xdp_rxq)
{
memset(xdp_rxq, 0, sizeof(*xdp_rxq));
}
/* Returns 0 on success, negative on failure */
int xdp_rxq_info_reg(struct xdp_rxq_info *xdp_rxq,
struct net_device *dev, u32 queue_index)
{
if (xdp_rxq->reg_state == REG_STATE_UNUSED) {
WARN(1, "Driver promised not to register this");
return -EINVAL;
}
if (xdp_rxq->reg_state == REG_STATE_REGISTERED) {
WARN(1, "Missing unregister, handled but fix driver");
xdp_rxq_info_unreg(xdp_rxq);
}
if (!dev) {
WARN(1, "Missing net_device from driver");
return -ENODEV;
}
/* State either UNREGISTERED or NEW */
xdp_rxq_info_init(xdp_rxq);
xdp_rxq->dev = dev;
xdp_rxq->queue_index = queue_index;
xdp_rxq->reg_state = REG_STATE_REGISTERED;
return 0;
}
EXPORT_SYMBOL_GPL(xdp_rxq_info_reg);
void xdp_rxq_info_unused(struct xdp_rxq_info *xdp_rxq)
{
xdp_rxq->reg_state = REG_STATE_UNUSED;
}
EXPORT_SYMBOL_GPL(xdp_rxq_info_unused);
bool xdp_rxq_info_is_reg(struct xdp_rxq_info *xdp_rxq)
{
return (xdp_rxq->reg_state == REG_STATE_REGISTERED);
}
EXPORT_SYMBOL_GPL(xdp_rxq_info_is_reg);
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
static int __mem_id_init_hash_table(void)
{
struct rhashtable *rht;
int ret;
if (unlikely(mem_id_init))
return 0;
rht = kzalloc(sizeof(*rht), GFP_KERNEL);
if (!rht)
return -ENOMEM;
ret = rhashtable_init(rht, &mem_id_rht_params);
if (ret < 0) {
kfree(rht);
return ret;
}
mem_id_ht = rht;
smp_mb(); /* mutex lock should provide enough pairing */
mem_id_init = true;
return 0;
}
/* Allocate a cyclic ID that maps to allocator pointer.
* See: https://www.kernel.org/doc/html/latest/core-api/idr.html
*
* Caller must lock mem_id_lock.
*/
static int __mem_id_cyclic_get(gfp_t gfp)
{
int retries = 1;
int id;
again:
id = ida_simple_get(&mem_id_pool, mem_id_next, MEM_ID_MAX, gfp);
if (id < 0) {
if (id == -ENOSPC) {
/* Cyclic allocator, reset next id */
if (retries--) {
mem_id_next = MEM_ID_MIN;
goto again;
}
}
return id; /* errno */
}
mem_id_next = id + 1;
return id;
}
static bool __is_supported_mem_type(enum xdp_mem_type type)
{
if (type == MEM_TYPE_PAGE_POOL)
return is_page_pool_compiled_in();
if (type >= MEM_TYPE_MAX)
return false;
return true;
}
int xdp_rxq_info_reg_mem_model(struct xdp_rxq_info *xdp_rxq,
enum xdp_mem_type type, void *allocator)
{
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
struct xdp_mem_allocator *xdp_alloc;
gfp_t gfp = GFP_KERNEL;
int id, errno, ret;
void *ptr;
if (xdp_rxq->reg_state != REG_STATE_REGISTERED) {
WARN(1, "Missing register, driver bug");
return -EFAULT;
}
if (!__is_supported_mem_type(type))
return -EOPNOTSUPP;
xdp_rxq->mem.type = type;
if (!allocator) {
if (type == MEM_TYPE_PAGE_POOL || type == MEM_TYPE_ZERO_COPY)
return -EINVAL; /* Setup time check page_pool req */
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
return 0;
}
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
/* Delay init of rhashtable to save memory if feature isn't used */
if (!mem_id_init) {
mutex_lock(&mem_id_lock);
ret = __mem_id_init_hash_table();
mutex_unlock(&mem_id_lock);
if (ret < 0) {
WARN_ON(1);
return ret;
}
}
xdp_alloc = kzalloc(sizeof(*xdp_alloc), gfp);
if (!xdp_alloc)
return -ENOMEM;
mutex_lock(&mem_id_lock);
id = __mem_id_cyclic_get(gfp);
if (id < 0) {
errno = id;
goto err;
}
xdp_rxq->mem.id = id;
xdp_alloc->mem = xdp_rxq->mem;
xdp_alloc->allocator = allocator;
/* Insert allocator into ID lookup table */
ptr = rhashtable_insert_slow(mem_id_ht, &id, &xdp_alloc->node);
if (IS_ERR(ptr)) {
ida_simple_remove(&mem_id_pool, xdp_rxq->mem.id);
xdp_rxq->mem.id = 0;
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
errno = PTR_ERR(ptr);
goto err;
}
2019-07-08 21:34:28 +00:00
if (type == MEM_TYPE_PAGE_POOL)
page_pool_use_xdp_mem(allocator, mem_allocator_disconnect);
2019-07-08 21:34:28 +00:00
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
mutex_unlock(&mem_id_lock);
trace_mem_connect(xdp_alloc, xdp_rxq);
return 0;
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
err:
mutex_unlock(&mem_id_lock);
kfree(xdp_alloc);
return errno;
}
EXPORT_SYMBOL_GPL(xdp_rxq_info_reg_mem_model);
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
/* XDP RX runs under NAPI protection, and in different delivery error
* scenarios (e.g. queue full), it is possible to return the xdp_frame
* while still leveraging this protection. The @napi_direct boolean
* is used for those calls sites. Thus, allowing for faster recycling
* of xdp_frames/pages in those cases.
*/
static void __xdp_return(void *data, struct xdp_mem_info *mem, bool napi_direct,
xsk: Introduce AF_XDP buffer allocation API In order to simplify AF_XDP zero-copy enablement for NIC driver developers, a new AF_XDP buffer allocation API is added. The implementation is based on a single core (single producer/consumer) buffer pool for the AF_XDP UMEM. A buffer is allocated using the xsk_buff_alloc() function, and returned using xsk_buff_free(). If a buffer is disassociated with the pool, e.g. when a buffer is passed to an AF_XDP socket, a buffer is said to be released. Currently, the release function is only used by the AF_XDP internals and not visible to the driver. Drivers using this API should register the XDP memory model with the new MEM_TYPE_XSK_BUFF_POOL type. The API is defined in net/xdp_sock_drv.h. The buffer type is struct xdp_buff, and follows the lifetime of regular xdp_buffs, i.e. the lifetime of an xdp_buff is restricted to a NAPI context. In other words, the API is not replacing xdp_frames. In addition to introducing the API and implementations, the AF_XDP core is migrated to use the new APIs. rfc->v1: Fixed build errors/warnings for m68k and riscv. (kbuild test robot) Added headroom/chunk size getter. (Maxim/Björn) v1->v2: Swapped SoBs. (Maxim) v2->v3: Initialize struct xdp_buff member frame_sz. (Björn) Add API to query the DMA address of a frame. (Maxim) Do DMA sync for CPU till the end of the frame to handle possible growth (frame_sz). (Maxim) Signed-off-by: Björn Töpel <bjorn.topel@intel.com> Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org> Link: https://lore.kernel.org/bpf/20200520192103.355233-6-bjorn.topel@gmail.com
2020-05-20 19:20:53 +00:00
unsigned long handle, struct xdp_buff *xdp)
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
{
struct xdp_mem_allocator *xa;
struct page *page;
switch (mem->type) {
case MEM_TYPE_PAGE_POOL:
rcu_read_lock();
/* mem->id is valid, checked in xdp_rxq_info_reg_mem_model() */
xa = rhashtable_lookup(mem_id_ht, &mem->id, mem_id_rht_params);
page = virt_to_head_page(data);
napi_direct &= !xdp_return_frame_no_direct();
page_pool_put_full_page(xa->page_pool, page, napi_direct);
rcu_read_unlock();
break;
case MEM_TYPE_PAGE_SHARED:
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
page_frag_free(data);
break;
case MEM_TYPE_PAGE_ORDER0:
page = virt_to_page(data); /* Assumes order0 page*/
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
put_page(page);
break;
case MEM_TYPE_ZERO_COPY:
/* NB! Only valid from an xdp_buff! */
rcu_read_lock();
/* mem->id is valid, checked in xdp_rxq_info_reg_mem_model() */
xa = rhashtable_lookup(mem_id_ht, &mem->id, mem_id_rht_params);
xa->zc_alloc->free(xa->zc_alloc, handle);
rcu_read_unlock();
xsk: Introduce AF_XDP buffer allocation API In order to simplify AF_XDP zero-copy enablement for NIC driver developers, a new AF_XDP buffer allocation API is added. The implementation is based on a single core (single producer/consumer) buffer pool for the AF_XDP UMEM. A buffer is allocated using the xsk_buff_alloc() function, and returned using xsk_buff_free(). If a buffer is disassociated with the pool, e.g. when a buffer is passed to an AF_XDP socket, a buffer is said to be released. Currently, the release function is only used by the AF_XDP internals and not visible to the driver. Drivers using this API should register the XDP memory model with the new MEM_TYPE_XSK_BUFF_POOL type. The API is defined in net/xdp_sock_drv.h. The buffer type is struct xdp_buff, and follows the lifetime of regular xdp_buffs, i.e. the lifetime of an xdp_buff is restricted to a NAPI context. In other words, the API is not replacing xdp_frames. In addition to introducing the API and implementations, the AF_XDP core is migrated to use the new APIs. rfc->v1: Fixed build errors/warnings for m68k and riscv. (kbuild test robot) Added headroom/chunk size getter. (Maxim/Björn) v1->v2: Swapped SoBs. (Maxim) v2->v3: Initialize struct xdp_buff member frame_sz. (Björn) Add API to query the DMA address of a frame. (Maxim) Do DMA sync for CPU till the end of the frame to handle possible growth (frame_sz). (Maxim) Signed-off-by: Björn Töpel <bjorn.topel@intel.com> Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org> Link: https://lore.kernel.org/bpf/20200520192103.355233-6-bjorn.topel@gmail.com
2020-05-20 19:20:53 +00:00
break;
case MEM_TYPE_XSK_BUFF_POOL:
/* NB! Only valid from an xdp_buff! */
xsk_buff_free(xdp);
break;
default:
/* Not possible, checked in xdp_rxq_info_reg_mem_model() */
break;
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
}
}
void xdp_return_frame(struct xdp_frame *xdpf)
{
xsk: Introduce AF_XDP buffer allocation API In order to simplify AF_XDP zero-copy enablement for NIC driver developers, a new AF_XDP buffer allocation API is added. The implementation is based on a single core (single producer/consumer) buffer pool for the AF_XDP UMEM. A buffer is allocated using the xsk_buff_alloc() function, and returned using xsk_buff_free(). If a buffer is disassociated with the pool, e.g. when a buffer is passed to an AF_XDP socket, a buffer is said to be released. Currently, the release function is only used by the AF_XDP internals and not visible to the driver. Drivers using this API should register the XDP memory model with the new MEM_TYPE_XSK_BUFF_POOL type. The API is defined in net/xdp_sock_drv.h. The buffer type is struct xdp_buff, and follows the lifetime of regular xdp_buffs, i.e. the lifetime of an xdp_buff is restricted to a NAPI context. In other words, the API is not replacing xdp_frames. In addition to introducing the API and implementations, the AF_XDP core is migrated to use the new APIs. rfc->v1: Fixed build errors/warnings for m68k and riscv. (kbuild test robot) Added headroom/chunk size getter. (Maxim/Björn) v1->v2: Swapped SoBs. (Maxim) v2->v3: Initialize struct xdp_buff member frame_sz. (Björn) Add API to query the DMA address of a frame. (Maxim) Do DMA sync for CPU till the end of the frame to handle possible growth (frame_sz). (Maxim) Signed-off-by: Björn Töpel <bjorn.topel@intel.com> Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org> Link: https://lore.kernel.org/bpf/20200520192103.355233-6-bjorn.topel@gmail.com
2020-05-20 19:20:53 +00:00
__xdp_return(xdpf->data, &xdpf->mem, false, 0, NULL);
}
xdp: rhashtable with allocator ID to pointer mapping Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. Instead of using the IDR infrastructure, which uses a radix tree, use a dynamic rhashtable, for creating ID to pointer lookup table, because this is faster. The problem that is being solved here is that, the xdp_rxq_info pointer (stored in xdp_buff) cannot be used directly, as the guaranteed lifetime is too short. The info is needed on a (potentially) remote CPU during DMA-TX completion time . In an xdp_frame the xdp_mem_info is stored, when it got converted from an xdp_buff, which is sufficient for the simple page refcnt based recycle schemes. For more advanced allocators there is a need to store a pointer to the registered allocator. Thus, there is a need to guard the lifetime or validity of the allocator pointer, which is done through this rhashtable ID map to pointer. The removal and validity of of the allocator and helper struct xdp_mem_allocator is guarded by RCU. The allocator will be created by the driver, and registered with xdp_rxq_info_reg_mem_model(). It is up-to debate who is responsible for freeing the allocator pointer or invoking the allocator destructor function. In any case, this must happen via RCU freeing. Use the IDA infrastructure for getting a cyclic increasing ID number, that is used for keeping track of each registered allocator per RX-queue xdp_rxq_info. V4: Per req of Jason Wang - Use xdp_rxq_info_reg_mem_model() in all drivers implementing XDP_REDIRECT, even-though it's not strictly necessary when allocator==NULL for type MEM_TYPE_PAGE_SHARED (given it's zero). V6: Per req of Alex Duyck - Introduce rhashtable_lookup() call in later patch V8: Address sparse should be static warnings (from kbuild test robot) Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-17 14:46:12 +00:00
EXPORT_SYMBOL_GPL(xdp_return_frame);
void xdp_return_frame_rx_napi(struct xdp_frame *xdpf)
{
xsk: Introduce AF_XDP buffer allocation API In order to simplify AF_XDP zero-copy enablement for NIC driver developers, a new AF_XDP buffer allocation API is added. The implementation is based on a single core (single producer/consumer) buffer pool for the AF_XDP UMEM. A buffer is allocated using the xsk_buff_alloc() function, and returned using xsk_buff_free(). If a buffer is disassociated with the pool, e.g. when a buffer is passed to an AF_XDP socket, a buffer is said to be released. Currently, the release function is only used by the AF_XDP internals and not visible to the driver. Drivers using this API should register the XDP memory model with the new MEM_TYPE_XSK_BUFF_POOL type. The API is defined in net/xdp_sock_drv.h. The buffer type is struct xdp_buff, and follows the lifetime of regular xdp_buffs, i.e. the lifetime of an xdp_buff is restricted to a NAPI context. In other words, the API is not replacing xdp_frames. In addition to introducing the API and implementations, the AF_XDP core is migrated to use the new APIs. rfc->v1: Fixed build errors/warnings for m68k and riscv. (kbuild test robot) Added headroom/chunk size getter. (Maxim/Björn) v1->v2: Swapped SoBs. (Maxim) v2->v3: Initialize struct xdp_buff member frame_sz. (Björn) Add API to query the DMA address of a frame. (Maxim) Do DMA sync for CPU till the end of the frame to handle possible growth (frame_sz). (Maxim) Signed-off-by: Björn Töpel <bjorn.topel@intel.com> Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org> Link: https://lore.kernel.org/bpf/20200520192103.355233-6-bjorn.topel@gmail.com
2020-05-20 19:20:53 +00:00
__xdp_return(xdpf->data, &xdpf->mem, true, 0, NULL);
}
EXPORT_SYMBOL_GPL(xdp_return_frame_rx_napi);
void xdp_return_buff(struct xdp_buff *xdp)
{
xsk: Introduce AF_XDP buffer allocation API In order to simplify AF_XDP zero-copy enablement for NIC driver developers, a new AF_XDP buffer allocation API is added. The implementation is based on a single core (single producer/consumer) buffer pool for the AF_XDP UMEM. A buffer is allocated using the xsk_buff_alloc() function, and returned using xsk_buff_free(). If a buffer is disassociated with the pool, e.g. when a buffer is passed to an AF_XDP socket, a buffer is said to be released. Currently, the release function is only used by the AF_XDP internals and not visible to the driver. Drivers using this API should register the XDP memory model with the new MEM_TYPE_XSK_BUFF_POOL type. The API is defined in net/xdp_sock_drv.h. The buffer type is struct xdp_buff, and follows the lifetime of regular xdp_buffs, i.e. the lifetime of an xdp_buff is restricted to a NAPI context. In other words, the API is not replacing xdp_frames. In addition to introducing the API and implementations, the AF_XDP core is migrated to use the new APIs. rfc->v1: Fixed build errors/warnings for m68k and riscv. (kbuild test robot) Added headroom/chunk size getter. (Maxim/Björn) v1->v2: Swapped SoBs. (Maxim) v2->v3: Initialize struct xdp_buff member frame_sz. (Björn) Add API to query the DMA address of a frame. (Maxim) Do DMA sync for CPU till the end of the frame to handle possible growth (frame_sz). (Maxim) Signed-off-by: Björn Töpel <bjorn.topel@intel.com> Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org> Link: https://lore.kernel.org/bpf/20200520192103.355233-6-bjorn.topel@gmail.com
2020-05-20 19:20:53 +00:00
__xdp_return(xdp->data, &xdp->rxq->mem, true, xdp->handle, xdp);
}
EXPORT_SYMBOL_GPL(xdp_return_buff);
/* Only called for MEM_TYPE_PAGE_POOL see xdp.h */
void __xdp_release_frame(void *data, struct xdp_mem_info *mem)
{
struct xdp_mem_allocator *xa;
struct page *page;
rcu_read_lock();
xa = rhashtable_lookup(mem_id_ht, &mem->id, mem_id_rht_params);
page = virt_to_head_page(data);
if (xa)
page_pool_release_page(xa->page_pool, page);
rcu_read_unlock();
}
EXPORT_SYMBOL_GPL(__xdp_release_frame);
int xdp_attachment_query(struct xdp_attachment_info *info,
struct netdev_bpf *bpf)
{
bpf->prog_id = info->prog ? info->prog->aux->id : 0;
bpf->prog_flags = info->prog ? info->flags : 0;
return 0;
}
EXPORT_SYMBOL_GPL(xdp_attachment_query);
bool xdp_attachment_flags_ok(struct xdp_attachment_info *info,
struct netdev_bpf *bpf)
{
if (info->prog && (bpf->flags ^ info->flags) & XDP_FLAGS_MODES) {
NL_SET_ERR_MSG(bpf->extack,
"program loaded with different flags");
return false;
}
return true;
}
EXPORT_SYMBOL_GPL(xdp_attachment_flags_ok);
void xdp_attachment_setup(struct xdp_attachment_info *info,
struct netdev_bpf *bpf)
{
if (info->prog)
bpf_prog_put(info->prog);
info->prog = bpf->prog;
info->flags = bpf->flags;
}
EXPORT_SYMBOL_GPL(xdp_attachment_setup);
struct xdp_frame *xdp_convert_zc_to_xdp_frame(struct xdp_buff *xdp)
{
unsigned int metasize, totsize;
void *addr, *data_to_copy;
struct xdp_frame *xdpf;
struct page *page;
/* Clone into a MEM_TYPE_PAGE_ORDER0 xdp_frame. */
metasize = xdp_data_meta_unsupported(xdp) ? 0 :
xdp->data - xdp->data_meta;
totsize = xdp->data_end - xdp->data + metasize;
if (sizeof(*xdpf) + totsize > PAGE_SIZE)
return NULL;
page = dev_alloc_page();
if (!page)
return NULL;
addr = page_to_virt(page);
xdpf = addr;
memset(xdpf, 0, sizeof(*xdpf));
addr += sizeof(*xdpf);
data_to_copy = metasize ? xdp->data_meta : xdp->data;
memcpy(addr, data_to_copy, totsize);
xdpf->data = addr + metasize;
xdpf->len = totsize - metasize;
xdpf->headroom = 0;
xdpf->metasize = metasize;
xdpf->mem.type = MEM_TYPE_PAGE_ORDER0;
xdp_return_buff(xdp);
return xdpf;
}
EXPORT_SYMBOL_GPL(xdp_convert_zc_to_xdp_frame);
/* Used by XDP_WARN macro, to avoid inlining WARN() in fast-path */
void xdp_warn(const char *msg, const char *func, const int line)
{
WARN(1, "XDP_WARN: %s(line:%d): %s\n", func, line, msg);
};
EXPORT_SYMBOL_GPL(xdp_warn);