[GIT PULL] io_uring fixes for 5.19-rc3

[Jens Axboe <axboe@kernel.dk>]

Hi Linus,

Bigger than usual at this time, both because we missed -rc2, but also
because of some reverts that we chose to do. In detail:

- Adjust mapped buffer API while we still can (Dylan)

- Mapped buffer fixes (Dylan, Hao, Pavel, me)

- Fix for uring_cmd wrong API usage for task_work (Dylan)

- Fix for bug introduced in fixed file closing (Hao)

- Fix race in buffer/file resource handling (Pavel)

- Revert the NOP support for CQE32 and buffer selection that was brought
  up during the merge window (Pavel)

- Remove IORING_CLOSE_FD_AND_FILE_SLOT introduced in this merge window.
  The API needs further refining, so just yank it for now and we'll
  revisit for a later kernel.

- Series cleaning up the CQE32 support added in this merge window,
  making it more integrated rather than sitting on the side (Pavel)

Please pull!


The following changes since commit b13baccc3850ca8b8cccbf8ed9912dbaa0fdf7f3:

  Linux 5.19-rc2 (2022-06-12 16:11:37 -0700)

are available in the Git repository at:

  git://git.kernel.dk/linux-block.git tags/io_uring-5.19-2022-06-16

for you to fetch changes up to 6436c770f120a9ffeb4e791650467f30f1d062d1:

  io_uring: recycle provided buffer if we punt to io-wq (2022-06-17 06:24:26 -0600)

----------------------------------------------------------------
io_uring-5.19-2022-06-16

----------------------------------------------------------------
Dylan Yudaken (4):
      io_uring: fix index calculation
      io_uring: fix types in provided buffer ring
      io_uring: limit size of provided buffer ring
      io_uring: do not use prio task_work_add in uring_cmd

Hao Xu (2):
      io_uring: openclose: fix bug of closing wrong fixed file
      io_uring: kbuf: fix bug of not consuming ring buffer in partial io case

Jens Axboe (3):
      Merge branch 'io_uring/io_uring-5.19' of https://github.com/isilence/linux into io_uring-5.19
      io_uring: commit non-pollable provided mapped buffers upfront
      io_uring: recycle provided buffer if we punt to io-wq

Pavel Begunkov (13):
      io_uring: fix races with file table unregister
      io_uring: fix races with buffer table unregister
      io_uring: fix not locked access to fixed buf table
      io_uring: fix double unlock for pbuf select
      Revert "io_uring: support CQE32 for nop operation"
      Revert "io_uring: add buffer selection support to IORING_OP_NOP"
      io_uring: remove IORING_CLOSE_FD_AND_FILE_SLOT
      io_uring: get rid of __io_fill_cqe{32}_req()
      io_uring: unite fill_cqe and the 32B version
      io_uring: fill extra big cqe fields from req
      io_uring: fix ->extra{1,2} misuse
      io_uring: remove __io_fill_cqe() helper
      io_uring: make io_fill_cqe_aux honour CQE32

 fs/io_uring.c                 | 367 +++++++++++++++++-------------------------
 include/uapi/linux/io_uring.h |   6 -
 2 files changed, 152 insertions(+), 221 deletions(-)

-- 
Jens Axboe