2018-05-02 18:01:23 +07:00
|
|
|
// SPDX-License-Identifier: GPL-2.0
|
|
|
|
/* XDP user-space packet buffer
|
|
|
|
* Copyright(c) 2018 Intel Corporation.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/sched/mm.h>
|
|
|
|
#include <linux/sched/signal.h>
|
|
|
|
#include <linux/sched/task.h>
|
|
|
|
#include <linux/uaccess.h>
|
|
|
|
#include <linux/slab.h>
|
|
|
|
#include <linux/bpf.h>
|
|
|
|
#include <linux/mm.h>
|
|
|
|
|
|
|
|
#include "xdp_umem.h"
|
2018-06-04 19:05:51 +07:00
|
|
|
#include "xsk_queue.h"
|
2018-05-02 18:01:23 +07:00
|
|
|
|
xsk: new descriptor addressing scheme
Currently, AF_XDP only supports a fixed frame-size memory scheme where
each frame is referenced via an index (idx). A user passes the frame
index to the kernel, and the kernel acts upon the data. Some NICs,
however, do not have a fixed frame-size model, instead they have a
model where a memory window is passed to the hardware and multiple
frames are filled into that window (referred to as the "type-writer"
model).
By changing the descriptor format from the current frame index
addressing scheme, AF_XDP can in the future be extended to support
these kinds of NICs.
In the index-based model, an idx refers to a frame of size
frame_size. Addressing a frame in the UMEM is done by offseting the
UMEM starting address by a global offset, idx * frame_size + offset.
Communicating via the fill- and completion-rings are done by means of
idx.
In this commit, the idx is removed in favor of an address (addr),
which is a relative address ranging over the UMEM. To convert an
idx-based address to the new addr is simply: addr = idx * frame_size +
offset.
We also stop referring to the UMEM "frame" as a frame. Instead it is
simply called a chunk.
To transfer ownership of a chunk to the kernel, the addr of the chunk
is passed in the fill-ring. Note, that the kernel will mask addr to
make it chunk aligned, so there is no need for userspace to do
that. E.g., for a chunk size of 2k, passing an addr of 2048, 2050 or
3000 to the fill-ring will refer to the same chunk.
On the completion-ring, the addr will match that of the Tx descriptor,
passed to the kernel.
Changing the descriptor format to use chunks/addr will allow for
future changes to move to a type-writer based model, where multiple
frames can reside in one chunk. In this model passing one single chunk
into the fill-ring, would potentially result in multiple Rx
descriptors.
This commit changes the uapi of AF_XDP sockets, and updates the
documentation.
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2018-06-04 18:57:13 +07:00
|
|
|
#define XDP_UMEM_MIN_CHUNK_SIZE 2048
|
2018-05-02 18:01:23 +07:00
|
|
|
|
2018-06-04 19:05:57 +07:00
|
|
|
void xdp_add_sk_umem(struct xdp_umem *umem, struct xdp_sock *xs)
|
|
|
|
{
|
|
|
|
unsigned long flags;
|
|
|
|
|
|
|
|
spin_lock_irqsave(&umem->xsk_list_lock, flags);
|
|
|
|
list_add_rcu(&xs->list, &umem->xsk_list);
|
|
|
|
spin_unlock_irqrestore(&umem->xsk_list_lock, flags);
|
|
|
|
}
|
|
|
|
|
|
|
|
void xdp_del_sk_umem(struct xdp_umem *umem, struct xdp_sock *xs)
|
|
|
|
{
|
|
|
|
unsigned long flags;
|
|
|
|
|
|
|
|
if (xs->dev) {
|
|
|
|
spin_lock_irqsave(&umem->xsk_list_lock, flags);
|
|
|
|
list_del_rcu(&xs->list);
|
|
|
|
spin_unlock_irqrestore(&umem->xsk_list_lock, flags);
|
|
|
|
|
|
|
|
if (umem->zc)
|
|
|
|
synchronize_net();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-06-04 19:05:55 +07:00
|
|
|
int xdp_umem_assign_dev(struct xdp_umem *umem, struct net_device *dev,
|
|
|
|
u32 queue_id, u16 flags)
|
|
|
|
{
|
|
|
|
bool force_zc, force_copy;
|
|
|
|
struct netdev_bpf bpf;
|
|
|
|
int err;
|
|
|
|
|
|
|
|
force_zc = flags & XDP_ZEROCOPY;
|
|
|
|
force_copy = flags & XDP_COPY;
|
|
|
|
|
|
|
|
if (force_zc && force_copy)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
if (force_copy)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
dev_hold(dev);
|
|
|
|
|
2018-06-04 19:05:57 +07:00
|
|
|
if (dev->netdev_ops->ndo_bpf && dev->netdev_ops->ndo_xsk_async_xmit) {
|
2018-06-04 19:05:55 +07:00
|
|
|
bpf.command = XDP_QUERY_XSK_UMEM;
|
|
|
|
|
|
|
|
rtnl_lock();
|
|
|
|
err = dev->netdev_ops->ndo_bpf(dev, &bpf);
|
|
|
|
rtnl_unlock();
|
|
|
|
|
|
|
|
if (err) {
|
|
|
|
dev_put(dev);
|
|
|
|
return force_zc ? -ENOTSUPP : 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
bpf.command = XDP_SETUP_XSK_UMEM;
|
|
|
|
bpf.xsk.umem = umem;
|
|
|
|
bpf.xsk.queue_id = queue_id;
|
|
|
|
|
|
|
|
rtnl_lock();
|
|
|
|
err = dev->netdev_ops->ndo_bpf(dev, &bpf);
|
|
|
|
rtnl_unlock();
|
|
|
|
|
|
|
|
if (err) {
|
|
|
|
dev_put(dev);
|
|
|
|
return force_zc ? err : 0; /* fail or fallback */
|
|
|
|
}
|
|
|
|
|
|
|
|
umem->dev = dev;
|
|
|
|
umem->queue_id = queue_id;
|
|
|
|
umem->zc = true;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
dev_put(dev);
|
|
|
|
return force_zc ? -ENOTSUPP : 0; /* fail or fallback */
|
|
|
|
}
|
|
|
|
|
2018-06-04 19:05:57 +07:00
|
|
|
static void xdp_umem_clear_dev(struct xdp_umem *umem)
|
2018-06-04 19:05:55 +07:00
|
|
|
{
|
|
|
|
struct netdev_bpf bpf;
|
|
|
|
int err;
|
|
|
|
|
|
|
|
if (umem->dev) {
|
|
|
|
bpf.command = XDP_SETUP_XSK_UMEM;
|
|
|
|
bpf.xsk.umem = NULL;
|
|
|
|
bpf.xsk.queue_id = umem->queue_id;
|
|
|
|
|
|
|
|
rtnl_lock();
|
|
|
|
err = umem->dev->netdev_ops->ndo_bpf(umem->dev, &bpf);
|
|
|
|
rtnl_unlock();
|
|
|
|
|
|
|
|
if (err)
|
|
|
|
WARN(1, "failed to disable umem!\n");
|
|
|
|
|
|
|
|
dev_put(umem->dev);
|
|
|
|
umem->dev = NULL;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-05-02 18:01:23 +07:00
|
|
|
static void xdp_umem_unpin_pages(struct xdp_umem *umem)
|
|
|
|
{
|
|
|
|
unsigned int i;
|
|
|
|
|
2018-05-22 14:35:02 +07:00
|
|
|
for (i = 0; i < umem->npgs; i++) {
|
|
|
|
struct page *page = umem->pgs[i];
|
2018-05-02 18:01:23 +07:00
|
|
|
|
2018-05-22 14:35:02 +07:00
|
|
|
set_page_dirty_lock(page);
|
|
|
|
put_page(page);
|
2018-05-02 18:01:23 +07:00
|
|
|
}
|
2018-05-22 14:35:02 +07:00
|
|
|
|
|
|
|
kfree(umem->pgs);
|
|
|
|
umem->pgs = NULL;
|
2018-05-02 18:01:23 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
static void xdp_umem_unaccount_pages(struct xdp_umem *umem)
|
|
|
|
{
|
2018-06-08 05:06:01 +07:00
|
|
|
if (umem->user) {
|
|
|
|
atomic_long_sub(umem->npgs, &umem->user->locked_vm);
|
|
|
|
free_uid(umem->user);
|
|
|
|
}
|
2018-05-02 18:01:23 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
static void xdp_umem_release(struct xdp_umem *umem)
|
|
|
|
{
|
|
|
|
struct task_struct *task;
|
|
|
|
struct mm_struct *mm;
|
|
|
|
|
2018-06-04 19:05:55 +07:00
|
|
|
xdp_umem_clear_dev(umem);
|
|
|
|
|
2018-05-02 18:01:24 +07:00
|
|
|
if (umem->fq) {
|
|
|
|
xskq_destroy(umem->fq);
|
|
|
|
umem->fq = NULL;
|
|
|
|
}
|
|
|
|
|
2018-05-02 18:01:31 +07:00
|
|
|
if (umem->cq) {
|
|
|
|
xskq_destroy(umem->cq);
|
|
|
|
umem->cq = NULL;
|
|
|
|
}
|
|
|
|
|
2018-05-22 14:35:02 +07:00
|
|
|
xdp_umem_unpin_pages(umem);
|
2018-05-02 18:01:23 +07:00
|
|
|
|
2018-05-22 14:35:02 +07:00
|
|
|
task = get_pid_task(umem->pid, PIDTYPE_PID);
|
|
|
|
put_pid(umem->pid);
|
|
|
|
if (!task)
|
|
|
|
goto out;
|
|
|
|
mm = get_task_mm(task);
|
|
|
|
put_task_struct(task);
|
|
|
|
if (!mm)
|
|
|
|
goto out;
|
2018-05-02 18:01:23 +07:00
|
|
|
|
2018-05-22 14:35:02 +07:00
|
|
|
mmput(mm);
|
2018-06-04 19:05:52 +07:00
|
|
|
kfree(umem->pages);
|
|
|
|
umem->pages = NULL;
|
|
|
|
|
2018-05-02 18:01:23 +07:00
|
|
|
xdp_umem_unaccount_pages(umem);
|
|
|
|
out:
|
|
|
|
kfree(umem);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void xdp_umem_release_deferred(struct work_struct *work)
|
|
|
|
{
|
|
|
|
struct xdp_umem *umem = container_of(work, struct xdp_umem, work);
|
|
|
|
|
|
|
|
xdp_umem_release(umem);
|
|
|
|
}
|
|
|
|
|
|
|
|
void xdp_get_umem(struct xdp_umem *umem)
|
|
|
|
{
|
2018-05-22 14:35:03 +07:00
|
|
|
refcount_inc(&umem->users);
|
2018-05-02 18:01:23 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
void xdp_put_umem(struct xdp_umem *umem)
|
|
|
|
{
|
|
|
|
if (!umem)
|
|
|
|
return;
|
|
|
|
|
2018-05-22 14:35:03 +07:00
|
|
|
if (refcount_dec_and_test(&umem->users)) {
|
2018-05-02 18:01:23 +07:00
|
|
|
INIT_WORK(&umem->work, xdp_umem_release_deferred);
|
|
|
|
schedule_work(&umem->work);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static int xdp_umem_pin_pages(struct xdp_umem *umem)
|
|
|
|
{
|
|
|
|
unsigned int gup_flags = FOLL_WRITE;
|
|
|
|
long npgs;
|
|
|
|
int err;
|
|
|
|
|
2018-06-11 18:57:12 +07:00
|
|
|
umem->pgs = kcalloc(umem->npgs, sizeof(*umem->pgs),
|
|
|
|
GFP_KERNEL | __GFP_NOWARN);
|
2018-05-02 18:01:23 +07:00
|
|
|
if (!umem->pgs)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
down_write(¤t->mm->mmap_sem);
|
|
|
|
npgs = get_user_pages(umem->address, umem->npgs,
|
|
|
|
gup_flags, &umem->pgs[0], NULL);
|
|
|
|
up_write(¤t->mm->mmap_sem);
|
|
|
|
|
|
|
|
if (npgs != umem->npgs) {
|
|
|
|
if (npgs >= 0) {
|
|
|
|
umem->npgs = npgs;
|
|
|
|
err = -ENOMEM;
|
|
|
|
goto out_pin;
|
|
|
|
}
|
|
|
|
err = npgs;
|
|
|
|
goto out_pgs;
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
out_pin:
|
|
|
|
xdp_umem_unpin_pages(umem);
|
|
|
|
out_pgs:
|
|
|
|
kfree(umem->pgs);
|
|
|
|
umem->pgs = NULL;
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int xdp_umem_account_pages(struct xdp_umem *umem)
|
|
|
|
{
|
|
|
|
unsigned long lock_limit, new_npgs, old_npgs;
|
|
|
|
|
|
|
|
if (capable(CAP_IPC_LOCK))
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
lock_limit = rlimit(RLIMIT_MEMLOCK) >> PAGE_SHIFT;
|
|
|
|
umem->user = get_uid(current_user());
|
|
|
|
|
|
|
|
do {
|
|
|
|
old_npgs = atomic_long_read(&umem->user->locked_vm);
|
|
|
|
new_npgs = old_npgs + umem->npgs;
|
|
|
|
if (new_npgs > lock_limit) {
|
|
|
|
free_uid(umem->user);
|
|
|
|
umem->user = NULL;
|
|
|
|
return -ENOBUFS;
|
|
|
|
}
|
|
|
|
} while (atomic_long_cmpxchg(&umem->user->locked_vm, old_npgs,
|
|
|
|
new_npgs) != old_npgs);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2018-05-22 14:35:02 +07:00
|
|
|
static int xdp_umem_reg(struct xdp_umem *umem, struct xdp_umem_reg *mr)
|
2018-05-02 18:01:23 +07:00
|
|
|
{
|
xsk: new descriptor addressing scheme
Currently, AF_XDP only supports a fixed frame-size memory scheme where
each frame is referenced via an index (idx). A user passes the frame
index to the kernel, and the kernel acts upon the data. Some NICs,
however, do not have a fixed frame-size model, instead they have a
model where a memory window is passed to the hardware and multiple
frames are filled into that window (referred to as the "type-writer"
model).
By changing the descriptor format from the current frame index
addressing scheme, AF_XDP can in the future be extended to support
these kinds of NICs.
In the index-based model, an idx refers to a frame of size
frame_size. Addressing a frame in the UMEM is done by offseting the
UMEM starting address by a global offset, idx * frame_size + offset.
Communicating via the fill- and completion-rings are done by means of
idx.
In this commit, the idx is removed in favor of an address (addr),
which is a relative address ranging over the UMEM. To convert an
idx-based address to the new addr is simply: addr = idx * frame_size +
offset.
We also stop referring to the UMEM "frame" as a frame. Instead it is
simply called a chunk.
To transfer ownership of a chunk to the kernel, the addr of the chunk
is passed in the fill-ring. Note, that the kernel will mask addr to
make it chunk aligned, so there is no need for userspace to do
that. E.g., for a chunk size of 2k, passing an addr of 2048, 2050 or
3000 to the fill-ring will refer to the same chunk.
On the completion-ring, the addr will match that of the Tx descriptor,
passed to the kernel.
Changing the descriptor format to use chunks/addr will allow for
future changes to move to a type-writer based model, where multiple
frames can reside in one chunk. In this model passing one single chunk
into the fill-ring, would potentially result in multiple Rx
descriptors.
This commit changes the uapi of AF_XDP sockets, and updates the
documentation.
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2018-06-04 18:57:13 +07:00
|
|
|
u32 chunk_size = mr->chunk_size, headroom = mr->headroom;
|
|
|
|
unsigned int chunks, chunks_per_page;
|
2018-05-02 18:01:23 +07:00
|
|
|
u64 addr = mr->addr, size = mr->len;
|
2018-06-04 19:05:52 +07:00
|
|
|
int size_chk, err, i;
|
2018-05-02 18:01:23 +07:00
|
|
|
|
xsk: new descriptor addressing scheme
Currently, AF_XDP only supports a fixed frame-size memory scheme where
each frame is referenced via an index (idx). A user passes the frame
index to the kernel, and the kernel acts upon the data. Some NICs,
however, do not have a fixed frame-size model, instead they have a
model where a memory window is passed to the hardware and multiple
frames are filled into that window (referred to as the "type-writer"
model).
By changing the descriptor format from the current frame index
addressing scheme, AF_XDP can in the future be extended to support
these kinds of NICs.
In the index-based model, an idx refers to a frame of size
frame_size. Addressing a frame in the UMEM is done by offseting the
UMEM starting address by a global offset, idx * frame_size + offset.
Communicating via the fill- and completion-rings are done by means of
idx.
In this commit, the idx is removed in favor of an address (addr),
which is a relative address ranging over the UMEM. To convert an
idx-based address to the new addr is simply: addr = idx * frame_size +
offset.
We also stop referring to the UMEM "frame" as a frame. Instead it is
simply called a chunk.
To transfer ownership of a chunk to the kernel, the addr of the chunk
is passed in the fill-ring. Note, that the kernel will mask addr to
make it chunk aligned, so there is no need for userspace to do
that. E.g., for a chunk size of 2k, passing an addr of 2048, 2050 or
3000 to the fill-ring will refer to the same chunk.
On the completion-ring, the addr will match that of the Tx descriptor,
passed to the kernel.
Changing the descriptor format to use chunks/addr will allow for
future changes to move to a type-writer based model, where multiple
frames can reside in one chunk. In this model passing one single chunk
into the fill-ring, would potentially result in multiple Rx
descriptors.
This commit changes the uapi of AF_XDP sockets, and updates the
documentation.
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2018-06-04 18:57:13 +07:00
|
|
|
if (chunk_size < XDP_UMEM_MIN_CHUNK_SIZE || chunk_size > PAGE_SIZE) {
|
2018-05-02 18:01:23 +07:00
|
|
|
/* Strictly speaking we could support this, if:
|
|
|
|
* - huge pages, or*
|
|
|
|
* - using an IOMMU, or
|
|
|
|
* - making sure the memory area is consecutive
|
|
|
|
* but for now, we simply say "computer says no".
|
|
|
|
*/
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
xsk: new descriptor addressing scheme
Currently, AF_XDP only supports a fixed frame-size memory scheme where
each frame is referenced via an index (idx). A user passes the frame
index to the kernel, and the kernel acts upon the data. Some NICs,
however, do not have a fixed frame-size model, instead they have a
model where a memory window is passed to the hardware and multiple
frames are filled into that window (referred to as the "type-writer"
model).
By changing the descriptor format from the current frame index
addressing scheme, AF_XDP can in the future be extended to support
these kinds of NICs.
In the index-based model, an idx refers to a frame of size
frame_size. Addressing a frame in the UMEM is done by offseting the
UMEM starting address by a global offset, idx * frame_size + offset.
Communicating via the fill- and completion-rings are done by means of
idx.
In this commit, the idx is removed in favor of an address (addr),
which is a relative address ranging over the UMEM. To convert an
idx-based address to the new addr is simply: addr = idx * frame_size +
offset.
We also stop referring to the UMEM "frame" as a frame. Instead it is
simply called a chunk.
To transfer ownership of a chunk to the kernel, the addr of the chunk
is passed in the fill-ring. Note, that the kernel will mask addr to
make it chunk aligned, so there is no need for userspace to do
that. E.g., for a chunk size of 2k, passing an addr of 2048, 2050 or
3000 to the fill-ring will refer to the same chunk.
On the completion-ring, the addr will match that of the Tx descriptor,
passed to the kernel.
Changing the descriptor format to use chunks/addr will allow for
future changes to move to a type-writer based model, where multiple
frames can reside in one chunk. In this model passing one single chunk
into the fill-ring, would potentially result in multiple Rx
descriptors.
This commit changes the uapi of AF_XDP sockets, and updates the
documentation.
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2018-06-04 18:57:13 +07:00
|
|
|
if (!is_power_of_2(chunk_size))
|
2018-05-02 18:01:23 +07:00
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
if (!PAGE_ALIGNED(addr)) {
|
|
|
|
/* Memory area has to be page size aligned. For
|
|
|
|
* simplicity, this might change.
|
|
|
|
*/
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
if ((addr + size) < addr)
|
|
|
|
return -EINVAL;
|
|
|
|
|
xsk: new descriptor addressing scheme
Currently, AF_XDP only supports a fixed frame-size memory scheme where
each frame is referenced via an index (idx). A user passes the frame
index to the kernel, and the kernel acts upon the data. Some NICs,
however, do not have a fixed frame-size model, instead they have a
model where a memory window is passed to the hardware and multiple
frames are filled into that window (referred to as the "type-writer"
model).
By changing the descriptor format from the current frame index
addressing scheme, AF_XDP can in the future be extended to support
these kinds of NICs.
In the index-based model, an idx refers to a frame of size
frame_size. Addressing a frame in the UMEM is done by offseting the
UMEM starting address by a global offset, idx * frame_size + offset.
Communicating via the fill- and completion-rings are done by means of
idx.
In this commit, the idx is removed in favor of an address (addr),
which is a relative address ranging over the UMEM. To convert an
idx-based address to the new addr is simply: addr = idx * frame_size +
offset.
We also stop referring to the UMEM "frame" as a frame. Instead it is
simply called a chunk.
To transfer ownership of a chunk to the kernel, the addr of the chunk
is passed in the fill-ring. Note, that the kernel will mask addr to
make it chunk aligned, so there is no need for userspace to do
that. E.g., for a chunk size of 2k, passing an addr of 2048, 2050 or
3000 to the fill-ring will refer to the same chunk.
On the completion-ring, the addr will match that of the Tx descriptor,
passed to the kernel.
Changing the descriptor format to use chunks/addr will allow for
future changes to move to a type-writer based model, where multiple
frames can reside in one chunk. In this model passing one single chunk
into the fill-ring, would potentially result in multiple Rx
descriptors.
This commit changes the uapi of AF_XDP sockets, and updates the
documentation.
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2018-06-04 18:57:13 +07:00
|
|
|
chunks = (unsigned int)div_u64(size, chunk_size);
|
|
|
|
if (chunks == 0)
|
2018-05-02 18:01:23 +07:00
|
|
|
return -EINVAL;
|
|
|
|
|
xsk: new descriptor addressing scheme
Currently, AF_XDP only supports a fixed frame-size memory scheme where
each frame is referenced via an index (idx). A user passes the frame
index to the kernel, and the kernel acts upon the data. Some NICs,
however, do not have a fixed frame-size model, instead they have a
model where a memory window is passed to the hardware and multiple
frames are filled into that window (referred to as the "type-writer"
model).
By changing the descriptor format from the current frame index
addressing scheme, AF_XDP can in the future be extended to support
these kinds of NICs.
In the index-based model, an idx refers to a frame of size
frame_size. Addressing a frame in the UMEM is done by offseting the
UMEM starting address by a global offset, idx * frame_size + offset.
Communicating via the fill- and completion-rings are done by means of
idx.
In this commit, the idx is removed in favor of an address (addr),
which is a relative address ranging over the UMEM. To convert an
idx-based address to the new addr is simply: addr = idx * frame_size +
offset.
We also stop referring to the UMEM "frame" as a frame. Instead it is
simply called a chunk.
To transfer ownership of a chunk to the kernel, the addr of the chunk
is passed in the fill-ring. Note, that the kernel will mask addr to
make it chunk aligned, so there is no need for userspace to do
that. E.g., for a chunk size of 2k, passing an addr of 2048, 2050 or
3000 to the fill-ring will refer to the same chunk.
On the completion-ring, the addr will match that of the Tx descriptor,
passed to the kernel.
Changing the descriptor format to use chunks/addr will allow for
future changes to move to a type-writer based model, where multiple
frames can reside in one chunk. In this model passing one single chunk
into the fill-ring, would potentially result in multiple Rx
descriptors.
This commit changes the uapi of AF_XDP sockets, and updates the
documentation.
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2018-06-04 18:57:13 +07:00
|
|
|
chunks_per_page = PAGE_SIZE / chunk_size;
|
|
|
|
if (chunks < chunks_per_page || chunks % chunks_per_page)
|
2018-05-02 18:01:23 +07:00
|
|
|
return -EINVAL;
|
|
|
|
|
xsk: new descriptor addressing scheme
Currently, AF_XDP only supports a fixed frame-size memory scheme where
each frame is referenced via an index (idx). A user passes the frame
index to the kernel, and the kernel acts upon the data. Some NICs,
however, do not have a fixed frame-size model, instead they have a
model where a memory window is passed to the hardware and multiple
frames are filled into that window (referred to as the "type-writer"
model).
By changing the descriptor format from the current frame index
addressing scheme, AF_XDP can in the future be extended to support
these kinds of NICs.
In the index-based model, an idx refers to a frame of size
frame_size. Addressing a frame in the UMEM is done by offseting the
UMEM starting address by a global offset, idx * frame_size + offset.
Communicating via the fill- and completion-rings are done by means of
idx.
In this commit, the idx is removed in favor of an address (addr),
which is a relative address ranging over the UMEM. To convert an
idx-based address to the new addr is simply: addr = idx * frame_size +
offset.
We also stop referring to the UMEM "frame" as a frame. Instead it is
simply called a chunk.
To transfer ownership of a chunk to the kernel, the addr of the chunk
is passed in the fill-ring. Note, that the kernel will mask addr to
make it chunk aligned, so there is no need for userspace to do
that. E.g., for a chunk size of 2k, passing an addr of 2048, 2050 or
3000 to the fill-ring will refer to the same chunk.
On the completion-ring, the addr will match that of the Tx descriptor,
passed to the kernel.
Changing the descriptor format to use chunks/addr will allow for
future changes to move to a type-writer based model, where multiple
frames can reside in one chunk. In this model passing one single chunk
into the fill-ring, would potentially result in multiple Rx
descriptors.
This commit changes the uapi of AF_XDP sockets, and updates the
documentation.
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2018-06-04 18:57:13 +07:00
|
|
|
headroom = ALIGN(headroom, 64);
|
2018-05-02 18:01:23 +07:00
|
|
|
|
xsk: new descriptor addressing scheme
Currently, AF_XDP only supports a fixed frame-size memory scheme where
each frame is referenced via an index (idx). A user passes the frame
index to the kernel, and the kernel acts upon the data. Some NICs,
however, do not have a fixed frame-size model, instead they have a
model where a memory window is passed to the hardware and multiple
frames are filled into that window (referred to as the "type-writer"
model).
By changing the descriptor format from the current frame index
addressing scheme, AF_XDP can in the future be extended to support
these kinds of NICs.
In the index-based model, an idx refers to a frame of size
frame_size. Addressing a frame in the UMEM is done by offseting the
UMEM starting address by a global offset, idx * frame_size + offset.
Communicating via the fill- and completion-rings are done by means of
idx.
In this commit, the idx is removed in favor of an address (addr),
which is a relative address ranging over the UMEM. To convert an
idx-based address to the new addr is simply: addr = idx * frame_size +
offset.
We also stop referring to the UMEM "frame" as a frame. Instead it is
simply called a chunk.
To transfer ownership of a chunk to the kernel, the addr of the chunk
is passed in the fill-ring. Note, that the kernel will mask addr to
make it chunk aligned, so there is no need for userspace to do
that. E.g., for a chunk size of 2k, passing an addr of 2048, 2050 or
3000 to the fill-ring will refer to the same chunk.
On the completion-ring, the addr will match that of the Tx descriptor,
passed to the kernel.
Changing the descriptor format to use chunks/addr will allow for
future changes to move to a type-writer based model, where multiple
frames can reside in one chunk. In this model passing one single chunk
into the fill-ring, would potentially result in multiple Rx
descriptors.
This commit changes the uapi of AF_XDP sockets, and updates the
documentation.
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2018-06-04 18:57:13 +07:00
|
|
|
size_chk = chunk_size - headroom - XDP_PACKET_HEADROOM;
|
2018-05-02 18:01:23 +07:00
|
|
|
if (size_chk < 0)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
umem->pid = get_task_pid(current, PIDTYPE_PID);
|
|
|
|
umem->address = (unsigned long)addr;
|
xsk: new descriptor addressing scheme
Currently, AF_XDP only supports a fixed frame-size memory scheme where
each frame is referenced via an index (idx). A user passes the frame
index to the kernel, and the kernel acts upon the data. Some NICs,
however, do not have a fixed frame-size model, instead they have a
model where a memory window is passed to the hardware and multiple
frames are filled into that window (referred to as the "type-writer"
model).
By changing the descriptor format from the current frame index
addressing scheme, AF_XDP can in the future be extended to support
these kinds of NICs.
In the index-based model, an idx refers to a frame of size
frame_size. Addressing a frame in the UMEM is done by offseting the
UMEM starting address by a global offset, idx * frame_size + offset.
Communicating via the fill- and completion-rings are done by means of
idx.
In this commit, the idx is removed in favor of an address (addr),
which is a relative address ranging over the UMEM. To convert an
idx-based address to the new addr is simply: addr = idx * frame_size +
offset.
We also stop referring to the UMEM "frame" as a frame. Instead it is
simply called a chunk.
To transfer ownership of a chunk to the kernel, the addr of the chunk
is passed in the fill-ring. Note, that the kernel will mask addr to
make it chunk aligned, so there is no need for userspace to do
that. E.g., for a chunk size of 2k, passing an addr of 2048, 2050 or
3000 to the fill-ring will refer to the same chunk.
On the completion-ring, the addr will match that of the Tx descriptor,
passed to the kernel.
Changing the descriptor format to use chunks/addr will allow for
future changes to move to a type-writer based model, where multiple
frames can reside in one chunk. In this model passing one single chunk
into the fill-ring, would potentially result in multiple Rx
descriptors.
This commit changes the uapi of AF_XDP sockets, and updates the
documentation.
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2018-06-04 18:57:13 +07:00
|
|
|
umem->props.chunk_mask = ~((u64)chunk_size - 1);
|
|
|
|
umem->props.size = size;
|
|
|
|
umem->headroom = headroom;
|
|
|
|
umem->chunk_size_nohr = chunk_size - headroom;
|
2018-05-02 18:01:23 +07:00
|
|
|
umem->npgs = size / PAGE_SIZE;
|
|
|
|
umem->pgs = NULL;
|
|
|
|
umem->user = NULL;
|
2018-06-04 19:05:57 +07:00
|
|
|
INIT_LIST_HEAD(&umem->xsk_list);
|
|
|
|
spin_lock_init(&umem->xsk_list_lock);
|
2018-05-02 18:01:23 +07:00
|
|
|
|
2018-05-22 14:35:03 +07:00
|
|
|
refcount_set(&umem->users, 1);
|
2018-05-02 18:01:23 +07:00
|
|
|
|
|
|
|
err = xdp_umem_account_pages(umem);
|
|
|
|
if (err)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
err = xdp_umem_pin_pages(umem);
|
|
|
|
if (err)
|
|
|
|
goto out_account;
|
2018-06-04 19:05:52 +07:00
|
|
|
|
|
|
|
umem->pages = kcalloc(umem->npgs, sizeof(*umem->pages), GFP_KERNEL);
|
|
|
|
if (!umem->pages) {
|
|
|
|
err = -ENOMEM;
|
|
|
|
goto out_account;
|
|
|
|
}
|
|
|
|
|
|
|
|
for (i = 0; i < umem->npgs; i++)
|
|
|
|
umem->pages[i].addr = page_address(umem->pgs[i]);
|
|
|
|
|
2018-05-02 18:01:23 +07:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
out_account:
|
|
|
|
xdp_umem_unaccount_pages(umem);
|
|
|
|
out:
|
|
|
|
put_pid(umem->pid);
|
|
|
|
return err;
|
|
|
|
}
|
2018-05-02 18:01:26 +07:00
|
|
|
|
2018-05-22 14:35:02 +07:00
|
|
|
struct xdp_umem *xdp_umem_create(struct xdp_umem_reg *mr)
|
|
|
|
{
|
|
|
|
struct xdp_umem *umem;
|
|
|
|
int err;
|
|
|
|
|
|
|
|
umem = kzalloc(sizeof(*umem), GFP_KERNEL);
|
|
|
|
if (!umem)
|
|
|
|
return ERR_PTR(-ENOMEM);
|
|
|
|
|
|
|
|
err = xdp_umem_reg(umem, mr);
|
|
|
|
if (err) {
|
|
|
|
kfree(umem);
|
|
|
|
return ERR_PTR(err);
|
|
|
|
}
|
|
|
|
|
|
|
|
return umem;
|
|
|
|
}
|
|
|
|
|
2018-05-02 18:01:26 +07:00
|
|
|
bool xdp_umem_validate_queues(struct xdp_umem *umem)
|
|
|
|
{
|
2018-05-18 19:00:23 +07:00
|
|
|
return umem->fq && umem->cq;
|
2018-05-02 18:01:26 +07:00
|
|
|
}
|