A preparation / cleanup patch simplifying the buf ring - mmap
synchronisation. Instead of relying on RCU, which is trickier, do it by
grabbing the mmap_lock when when anyone tries to publish or remove a
registered buffer to / from ->io_bl_xa.
Modifications of the xarray should always be protected by both
->uring_lock and ->mmap_lock, while lookups should hold either of them.
While a struct io_buffer_list is in the xarray, the mmap related fields
like ->flags and ->buf_pages should stay stable.
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/af13bde56ee1a26bcaefaa9aad37a9ea318a590e.1732886067.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
By default, any recv/read operation that uses provided buffers will
consume at least 1 buffer fully (and maybe more, in case of bundles).
This adds support for incremental consumption, meaning that an
application may add large buffers, and each read/recv will just consume
the part of the buffer that it needs.
For example, let's say an application registers 1MB buffers in a
provided buffer ring, for streaming receives. If it gets a short recv,
then the full 1MB buffer will be consumed and passed back to the
application. With incremental consumption, only the part that was
actually used is consumed, and the buffer remains the current one.
This means that both the application and the kernel needs to keep track
of what the current receive point is. Each recv will still pass back a
buffer ID and the size consumed, the only difference is that before the
next receive would always be the next buffer in the ring. Now the same
buffer ID may return multiple receives, each at an offset into that
buffer from where the previous receive left off. Example:
Application registers a provided buffer ring, and adds two 32K buffers
to the ring.
Buffer1 address: 0x1000000 (buffer ID 0)
Buffer2 address: 0x2000000 (buffer ID 1)
A recv completion is received with the following values:
cqe->res 0x1000 (4k bytes received)
cqe->flags 0x11 (CQE_F_BUFFER|CQE_F_BUF_MORE set, buffer ID 0)
and the application now knows that 4096b of data is available at
0x1000000, the start of that buffer, and that more data from this buffer
will be coming. Now the next receive comes in:
cqe->res 0x2010 (8k bytes received)
cqe->flags 0x11 (CQE_F_BUFFER|CQE_F_BUF_MORE set, buffer ID 0)
which tells the application that 8k is available where the last
completion left off, at 0x1001000. Next completion is:
cqe->res 0x5000 (20k bytes received)
cqe->flags 0x1 (CQE_F_BUFFER set, buffer ID 0)
and the application now knows that 20k of data is available at
0x1003000, which is where the previous receive ended. CQE_F_BUF_MORE
isn't set, as no more data is available in this buffer ID. The next
completion is then:
cqe->res 0x1000 (4k bytes received)
cqe->flags 0x10001 (CQE_F_BUFFER|CQE_F_BUF_MORE set, buffer ID 1)
which tells the application that buffer ID 1 is now the current one,
hence there's 4k of valid data at 0x2000000. 0x2001000 will be the next
receive point for this buffer ID.
When a buffer will be reused by future CQE completions,
IORING_CQE_BUF_MORE will be set in cqe->flags. This tells the application
that the kernel isn't done with the buffer yet, and that it should expect
more completions for this buffer ID. Will only be set by provided buffer
rings setup with IOU_PBUF_RING INC, as that's the only type of buffer
that will see multiple consecutive completions for the same buffer ID.
For any other provided buffer type, any completion that passes back
a buffer to the application is final.
Once a buffer has been fully consumed, the buffer ring head is
incremented and the next receive will indicate the next buffer ID in the
CQE cflags.
On the send side, the application can manage how much data is sent from
an existing buffer by setting sqe->len to the desired send length.
An application can request incremental consumption by setting
IOU_PBUF_RING_INC in the provided buffer ring registration. Outside of
that, any provided buffer ring setup and buffer additions is done like
before, no changes there. The only change is in how an application may
see multiple completions for the same buffer ID, hence needing to know
where the next receive will happen.
Note that like existing provided buffer rings, this should not be used
with IOSQE_ASYNC, as both really require the ring to remain locked over
the duration of the buffer selection and the operation completion. It
will consume a buffer otherwise regardless of the size of the IO done.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
In preparation for needing the consumed length, pass in the length being
completed. Unused right now, but will be used when it is possible to
partially consume a buffer.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Committing the selected ring buffer is currently done in three different
spots, combine it into a helper and just call that.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
nr_iovs is capped at 1024, and mode only has a few low values. We can
safely make them u16, in preparation for adding a few more members.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
We could just move these two and save some space, but in preparation
for adding another flag, turn them into flags first.
This saves 8 bytes in struct io_buffer_list, making it exactly half
a cacheline on 64-bit archs now rather than 40 bytes.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Our provided buffer interface only allows selection of a single buffer.
Add an API that allows getting/peeking multiple buffers at the same time.
This is only implemented for the ring provided buffers. It could be added
for the legacy provided buffers as well, but since it's strongly
encouraged to use the new interface, let's keep it simpler and just
provide it for the new API. The legacy interface will always just select
a single buffer.
There are two new main functions:
io_buffers_select(), which selects up as many buffers as it can. The
caller supplies the iovec array, and io_buffers_select() may allocate a
bigger array if the 'out_len' being passed in is non-zero and bigger
than what fits in the provided iovec. Buffers grabbed with this helper
are permanently assigned.
io_buffers_peek(), which works like io_buffers_select(), except they can
be recycled, if needed. Callers using either of these functions should
call io_put_kbufs() rather than io_put_kbuf() at completion time. The
peek interface must be called with the ctx locked from peek to
completion.
This add a bit state for the request:
- REQ_F_BUFFERS_COMMIT, which means that the the buffers have been
peeked and should be committed to the buffer ring head when they are
put as part of completion. Prior to this, req->buf_list was cleared to
NULL when committed.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The only caller doesn't handle the return value of io_put_kbuf_comp(), so
change its return type into void.
Also follow Jens's suggestion to rename it as io_put_kbuf_drop().
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20240407132759.4056167-1-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Rather than use remap_pfn_range() for this and manually free later,
switch to using vm_insert_page() and have it Just Work.
This requires a bit of effort on the mmap lookup side, as the ctx
uring_lock isn't held, which otherwise protects buffer_lists from being
torn down, and it's not safe to grab from mmap context that would
introduce an ABBA deadlock between the mmap lock and the ctx uring_lock.
Instead, lookup the buffer_list under RCU, as the the list is RCU freed
already. Use the existing reference count to determine whether it's
possible to safely grab a reference to it (eg if it's not zero already),
and drop that reference when done with the mapping. If the mmap
reference is the last one, the buffer_list and the associated memory can
go away, since the vma insertion has references to the inserted pages at
that point.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
If we look up the kbuf, ensure that it doesn't get unregistered until
after we're done with it. Since we're inside mmap, we cannot safely use
the io_uring lock. Rely on the fact that we can lookup the buffer list
under RCU now and grab a reference to it, preventing it from being
unregistered until we're done with it. The lookup returns the
io_buffer_list directly with it referenced.
Cc: stable@vger.kernel.org # v6.4+
Fixes: 5cf4f52e6d ("io_uring: free io_buffer_list entries via RCU")
Signed-off-by: Jens Axboe <axboe@kernel.dk>
No functional changes in this patch, just in preparation for being able
to keep the buffer list alive outside of the ctx->uring_lock.
Cc: stable@vger.kernel.org # v6.4+
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Now that xarray is being exclusively used for the buffer_list lookup,
this check is no longer needed. Get rid of it and the is_ready member.
Cc: stable@vger.kernel.org # v6.4+
Signed-off-by: Jens Axboe <axboe@kernel.dk>
In buffer lists we have ->is_mapped as well as ->is_mmap, it's
pretty hard to stay sane double checking which one means what,
and in the long run there is a high chance of an eventual bug.
Rename ->is_mapped into ->is_buf_ring.
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/c4838f4d8ad506ad6373f1c305aee2d2c1a89786.1710343154.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
We only use the flag for this purpose, so rename it accordingly. This
further prevents various other use cases of it, keeping it clean and
consistent. Then we can also check it in one spot, when it's being
attempted recycled, and remove some dead code in io_kbuf_recycle_ring().
Signed-off-by: Jens Axboe <axboe@kernel.dk>
We have various functions calculating the CQE cflags we need to pass
back, but it's all the same everywhere. Make a number of the putting
functions void, and just have the two main helps for this, io_put_kbuf()
and io_put_kbuf_comp() calculate the actual mask and pass it back.
While at it, cleanup how we put REQ_F_BUFFER_RING buffers. Before
this change, we would call into __io_put_kbuf() only to go right back
in to the header defined functions. As clearing this type of buffer
is just re-assigning the buf_index and incrementing the head, this
is very wasteful.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The tail of the provided ring buffer is shared between the kernel and
the application, but the head is private to the kernel as the
application doesn't need to see it. However, this also prevents the
application from knowing how many buffers the kernel has consumed.
Usually this is fine, as the information is inherently racy in that
the kernel could be consuming buffers continually, but for cleanup
purposes it may be relevant to know how many buffers are still left
in the ring.
Add IORING_REGISTER_PBUF_STATUS which will return status for a given
provided buffer ring. Right now it just returns the head, but space
is reserved for more information later in, if needed.
Link: https://github.com/axboe/liburing/discussions/1020
Signed-off-by: Jens Axboe <axboe@kernel.dk>
mmap_lock nests under uring_lock out of necessity, as we may be doing
user copies with uring_lock held. However, for mmap of provided buffer
rings, we attempt to grab uring_lock with mmap_lock already held from
do_mmap(). This makes lockdep, rightfully, complain:
WARNING: possible circular locking dependency detected
6.7.0-rc1-00009-gff3337ebaf94-dirty #4438 Not tainted
------------------------------------------------------
buf-ring.t/442 is trying to acquire lock:
ffff00020e1480a8 (&ctx->uring_lock){+.+.}-{3:3}, at: io_uring_validate_mmap_request.isra.0+0x4c/0x140
but task is already holding lock:
ffff0000dc226190 (&mm->mmap_lock){++++}-{3:3}, at: vm_mmap_pgoff+0x124/0x264
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #1 (&mm->mmap_lock){++++}-{3:3}:
__might_fault+0x90/0xbc
io_register_pbuf_ring+0x94/0x488
__arm64_sys_io_uring_register+0x8dc/0x1318
invoke_syscall+0x5c/0x17c
el0_svc_common.constprop.0+0x108/0x130
do_el0_svc+0x2c/0x38
el0_svc+0x4c/0x94
el0t_64_sync_handler+0x118/0x124
el0t_64_sync+0x168/0x16c
-> #0 (&ctx->uring_lock){+.+.}-{3:3}:
__lock_acquire+0x19a0/0x2d14
lock_acquire+0x2e0/0x44c
__mutex_lock+0x118/0x564
mutex_lock_nested+0x20/0x28
io_uring_validate_mmap_request.isra.0+0x4c/0x140
io_uring_mmu_get_unmapped_area+0x3c/0x98
get_unmapped_area+0xa4/0x158
do_mmap+0xec/0x5b4
vm_mmap_pgoff+0x158/0x264
ksys_mmap_pgoff+0x1d4/0x254
__arm64_sys_mmap+0x80/0x9c
invoke_syscall+0x5c/0x17c
el0_svc_common.constprop.0+0x108/0x130
do_el0_svc+0x2c/0x38
el0_svc+0x4c/0x94
el0t_64_sync_handler+0x118/0x124
el0t_64_sync+0x168/0x16c
From that mmap(2) path, we really just need to ensure that the buffer
list doesn't go away from underneath us. For the lower indexed entries,
they never go away until the ring is freed and we can always sanely
reference those as long as the caller has a file reference. For the
higher indexed ones in our xarray, we just need to ensure that the
buffer list remains valid while we return the address of it.
Free the higher indexed io_buffer_list entries via RCU. With that we can
avoid needing ->uring_lock inside mmap(2), and simply hold the RCU read
lock around the buffer list lookup and address check.
To ensure that the arrayed lookup either returns a valid fully formulated
entry via RCU lookup, add an 'is_ready' flag that we access with store
and release memory ordering. This isn't needed for the xarray lookups,
but doesn't hurt either. Since this isn't a fast path, retain it across
both types. Similarly, for the allocated array inside the ctx, ensure
we use the proper load/acquire as setup could in theory be running in
parallel with mmap.
While in there, add a few lockdep checks for documentation purposes.
Cc: stable@vger.kernel.org
Fixes: c56e022c0a ("io_uring: add support for user mapped provided buffer ring")
Signed-off-by: Jens Axboe <axboe@kernel.dk>
If a provided buffer ring is setup with IOU_PBUF_RING_MMAP, then the
kernel allocates the memory for it and the application is expected to
mmap(2) this memory. However, io_uring uses remap_pfn_range() for this
operation, so we cannot rely on normal munmap/release on freeing them
for us.
Stash an io_buf_free entry away for each of these, if any, and provide
a helper to free them post ->release().
Cc: stable@vger.kernel.org
Fixes: c56e022c0a ("io_uring: add support for user mapped provided buffer ring")
Reported-by: Jann Horn <jannh@google.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
It can be useful to know if io_kbuf_recycle did actually recycle the
buffer on the request, or if it left the request alone.
Signed-off-by: Dylan Yudaken <dyudaken@gmail.com>
Link: https://lore.kernel.org/r/20231106203909.197089-2-dyudaken@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The ring mapped provided buffer rings rely on the application allocating
the memory for the ring, and then the kernel will map it. This generally
works fine, but runs into issues on some architectures where we need
to be able to ensure that the kernel and application virtual address for
the ring play nicely together. This at least impacts architectures that
set SHM_COLOUR, but potentially also anyone setting SHMLBA.
To use this variant of ring provided buffers, the application need not
allocate any memory for the ring. Instead the kernel will do so, and
the allocation must subsequently call mmap(2) on the ring with the
offset set to:
IORING_OFF_PBUF_RING | (bgid << IORING_OFF_PBUF_SHIFT)
to get a virtual address for the buffer ring. Normally the application
would allocate a suitable piece of memory (and correctly aligned) and
simply pass that in via io_uring_buf_reg.ring_addr and the kernel would
map it.
Outside of the setup differences, the kernel allocate + user mapped
provided buffer ring works exactly the same.
Acked-by: Helge Deller <deller@gmx.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Rather than rely on checking buffer_list->buf_pages or ->buf_nr_pages,
add a separate member that tracks if this is a ring mapped provided
buffer list or not.
Acked-by: Helge Deller <deller@gmx.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
In commit 934447a603 ("io_uring: do not recycle buffer in READV") a
temporary fix was put in io_kbuf_recycle to simply never recycle READV
buffers.
Instead of that, rather treat READV with REQ_F_BUFFER_SELECTED the same as
a READ with REQ_F_BUFFER_SELECTED. Since READV requires iov_len of 1 they
are essentially the same.
In order to do this inside io_prep_rw() add some validation to check that
it is in fact only length 1, and also extract the length of the buffer at
prep time.
This allows removal of the io_iov_buffer_select codepaths as they are only
used from the READV op.
Signed-off-by: Dylan Yudaken <dylany@fb.com>
Link: https://lore.kernel.org/r/20220907165152.994979-1-dylany@fb.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
When we don't recycle a selected ring buffer we should advance the head
of the ring, so don't just skip io_kbuf_recycle() for IORING_OP_READV
but adjust the ring.
Fixes: 934447a603 ("io_uring: do not recycle buffer in READV")
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Reviewed-by: Dylan Yudaken <dylany@fb.com>
Link: https://lore.kernel.org/r/a6d85e2611471bcb5d5dcd63a8342077ddc2d73d.1662480490.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Attempt to restore bgid. This is needed when recycling unused buffers as
the next time around it will want the correct bgid.
Signed-off-by: Dylan Yudaken <dylany@fb.com>
Link: https://lore.kernel.org/r/20220630091231.1456789-3-dylany@fb.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
__io_kbuf_recycle() is only called in io_kbuf_recycle(). Kill it and
tweak the code so that the legacy pbuf and ring pbuf code become clear
Signed-off-by: Hao Xu <howeyxu@tencent.com>
Link: https://lore.kernel.org/r/20220622055551.642370-1-hao.xu@linux.dev
Signed-off-by: Jens Axboe <axboe@kernel.dk>
