mbox series

[v2,00/35] bitops: add atomic find_bit() operations

Message ID 20231203192422.539300-1-yury.norov@gmail.com
Headers show
Series bitops: add atomic find_bit() operations | expand

Message

Yury Norov Dec. 3, 2023, 7:23 p.m. UTC
Add helpers around test_and_{set,clear}_bit() that allow to search for
clear or set bits and flip them atomically.

The target patterns may look like this:

	for (idx = 0; idx < nbits; idx++)
		if (test_and_clear_bit(idx, bitmap))
			do_something(idx);

Or like this:

	do {
		bit = find_first_bit(bitmap, nbits);
		if (bit >= nbits)
			return nbits;
	} while (!test_and_clear_bit(bit, bitmap));
	return bit;

In both cases, the opencoded loop may be converted to a single function
or iterator call. Correspondingly:

	for_each_test_and_clear_bit(idx, bitmap, nbits)
		do_something(idx);

Or:
	return find_and_clear_bit(bitmap, nbits);

Obviously, the less routine code people have to write themself, the
less probability to make a mistake.

Those are not only handy helpers but also resolve a non-trivial
issue of using non-atomic find_bit() together with atomic
test_and_{set,clear)_bit().

The trick is that find_bit() implies that the bitmap is a regular
non-volatile piece of memory, and compiler is allowed to use such
optimization techniques like re-fetching memory instead of caching it.

For example, find_first_bit() is implemented like this:

      for (idx = 0; idx * BITS_PER_LONG < sz; idx++) {
              val = addr[idx];
              if (val) {
                      sz = min(idx * BITS_PER_LONG + __ffs(val), sz);
                      break;
              }
      }

On register-memory architectures, like x86, compiler may decide to
access memory twice - first time to compare against 0, and second time
to fetch its value to pass it to __ffs().

When running find_first_bit() on volatile memory, the memory may get
changed in-between, and for instance, it may lead to passing 0 to
__ffs(), which is undefined. This is a potentially dangerous call.

find_and_clear_bit() as a wrapper around test_and_clear_bit()
naturally treats underlying bitmap as a volatile memory and prevents
compiler from such optimizations.

Now that KCSAN is catching exactly this type of situations and warns on
undercover memory modifications. We can use it to reveal improper usage
of find_bit(), and convert it to atomic find_and_*_bit() as appropriate.

The 1st patch of the series adds the following atomic primitives:

	find_and_set_bit(addr, nbits);
	find_and_set_next_bit(addr, nbits, start);
	...

Here find_and_{set,clear} part refers to the corresponding
test_and_{set,clear}_bit function. Suffixes like _wrap or _lock
derive their semantics from corresponding find() or test() functions.

For brevity, the naming omits the fact that we search for zero bit in
find_and_set, and correspondingly search for set bit in find_and_clear
functions.

The patch also adds iterators with atomic semantics, like
for_each_test_and_set_bit(). Here, the naming rule is to simply prefix
corresponding atomic operation with 'for_each'.

This series is a result of discussion [1]. All find_bit() functions imply
exclusive access to the bitmaps. However, KCSAN reports quite a number
of warnings related to find_bit() API. Some of them are not pointing
to real bugs because in many situations people intentionally allow
concurrent bitmap operations.

If so, find_bit() can be annotated such that KCSAN will ignore it:

        bit = data_race(find_first_bit(bitmap, nbits));

This series addresses the other important case where people really need
atomic find ops. As the following patches show, the resulting code
looks safer and more verbose comparing to opencoded loops followed by
atomic bit flips.

In [1] Mirsad reported 2% slowdown in a single-thread search test when
switching find_bit() function to treat bitmaps as volatile arrays. On
the other hand, kernel robot in the same thread reported +3.7% to the
performance of will-it-scale.per_thread_ops test.

Assuming that our compilers are sane and generate better code against
properly annotated data, the above discrepancy doesn't look weird. When
running on non-volatile bitmaps, plain find_bit() outperforms atomic
find_and_bit(), and vice-versa.

So, all users of find_bit() API, where heavy concurrency is expected,
are encouraged to switch to atomic find_and_bit() as appropriate.

The 1st patch of this series adds atomic find_and_bit() API, 2nd adds
a basic test for new API, and all the following patches spread it over
the kernel.

They can be applied separately from each other on per-subsystems basis,
or I can pull them in bitmap tree, as appropriate.

[1] https://lore.kernel.org/lkml/634f5fdf-e236-42cf-be8d-48a581c21660@alu.unizg.hr/T/#m3e7341eb3571753f3acf8fe166f3fb5b2c12e615
---
v1: https://lore.kernel.org/netdev/20231118155105.25678-29-yury.norov@gmail.com/T/
v2:
 - Add a basic test for the new API # Bart Van Assche;
 - Add collected reviewers' tags. Thank you guys!
 - Fix typos where found/pointed by;
 - Drop erroneous patch #v1-31 ("drivers/perf: optimize m1_pmu_get_event_idx()...") @ Marc Zyngier;
 - Drop unneeded patch #v1-12 ("wifi: intel: use atomic find_bit() API...") @ Johannes Berg;
 - Patch #v1-15: split SCSI changes per subsystems @ Bart Van Assche;
 - Patch  #5: keep changes inside __mm_cid_try_get() @ Mathieu Desnoyers;
 - Patch  #8: use find_and_set_next_bit() @ Will Deacon;
 - Patch #13: keep test against stimer->config.enable @ Vitaly Kuznetsov;
 - Patch #15: use find_and_set_next_bit @ Bart Van Assche;
 - Patch #31: edit commit message @ Tony Lu, Alexandra Winter;
 - Patch #35: edit tag @ John Paul Adrian Glaubitz;

Yury Norov (35):
  lib/find: add atomic find_bit() primitives
  lib/find: add test for atomic find_bit() ops
  lib/sbitmap; make __sbitmap_get_word() using find_and_set_bit()
  watch_queue: use atomic find_bit() in post_one_notification()
  sched: add cpumask_find_and_set() and use it in __mm_cid_get()
  mips: sgi-ip30: rework heart_alloc_int()
  sparc: fix opencoded find_and_set_bit() in alloc_msi()
  perf/arm: optimize opencoded atomic find_bit() API
  drivers/perf: optimize ali_drw_get_counter_idx() by using find_bit()
  dmaengine: idxd: optimize perfmon_assign_event()
  ath10k: optimize ath10k_snoc_napi_poll() by using find_bit()
  wifi: rtw88: optimize rtw_pci_tx_kick_off() by using find_bit()
  KVM: x86: hyper-v: optimize and cleanup kvm_hv_process_stimers()
  PCI: hv: switch hv_get_dom_num() to use atomic find_bit()
  scsi: core: use atomic find_bit() API where appropriate
  scsi: mpi3mr: switch to using atomic find_and_set_bit()
  scsi: qedi: rework qedi_get_task_idx()
  powerpc: use atomic find_bit() API where appropriate
  iommu: use atomic find_bit() API where appropriate
  media: radio-shark: use atomic find_bit() API where appropriate
  sfc: switch to using atomic find_bit() API where appropriate
  tty: nozomi: optimize interrupt_handler()
  usb: cdc-acm: optimize acm_softint()
  block: null_blk: fix opencoded find_and_set_bit() in get_tag()
  RDMA/rtrs: fix opencoded find_and_set_bit_lock() in
    __rtrs_get_permit()
  mISDN: optimize get_free_devid()
  media: em28xx: cx231xx: fix opencoded find_and_set_bit()
  ethernet: rocker: optimize ofdpa_port_internal_vlan_id_get()
  serial: sc12is7xx: optimize sc16is7xx_alloc_line()
  bluetooth: optimize cmtp_alloc_block_id()
  net: smc:  use find_and_set_bit() in smc_wr_tx_get_free_slot_index()
  ALSA: use atomic find_bit() functions where applicable
  m68k: rework get_mmu_context()
  microblaze: rework get_mmu_context()
  sh: mach-x3proto: rework ilsel_enable()

 arch/m68k/include/asm/mmu_context.h          |  11 +-
 arch/microblaze/include/asm/mmu_context_mm.h |  11 +-
 arch/mips/sgi-ip30/ip30-irq.c                |  12 +-
 arch/powerpc/mm/book3s32/mmu_context.c       |  10 +-
 arch/powerpc/platforms/pasemi/dma_lib.c      |  45 +--
 arch/powerpc/platforms/powernv/pci-sriov.c   |  12 +-
 arch/sh/boards/mach-x3proto/ilsel.c          |   4 +-
 arch/sparc/kernel/pci_msi.c                  |   9 +-
 arch/x86/kvm/hyperv.c                        |  39 ++-
 drivers/block/null_blk/main.c                |  41 +--
 drivers/dma/idxd/perfmon.c                   |   8 +-
 drivers/infiniband/ulp/rtrs/rtrs-clt.c       |  15 +-
 drivers/iommu/arm/arm-smmu/arm-smmu.h        |  10 +-
 drivers/iommu/msm_iommu.c                    |  18 +-
 drivers/isdn/mISDN/core.c                    |   9 +-
 drivers/media/radio/radio-shark.c            |   5 +-
 drivers/media/radio/radio-shark2.c           |   5 +-
 drivers/media/usb/cx231xx/cx231xx-cards.c    |  16 +-
 drivers/media/usb/em28xx/em28xx-cards.c      |  37 +--
 drivers/net/ethernet/rocker/rocker_ofdpa.c   |  11 +-
 drivers/net/ethernet/sfc/rx_common.c         |   4 +-
 drivers/net/ethernet/sfc/siena/rx_common.c   |   4 +-
 drivers/net/ethernet/sfc/siena/siena_sriov.c |  14 +-
 drivers/net/wireless/ath/ath10k/snoc.c       |   9 +-
 drivers/net/wireless/realtek/rtw88/pci.c     |   5 +-
 drivers/net/wireless/realtek/rtw89/pci.c     |   5 +-
 drivers/pci/controller/pci-hyperv.c          |   7 +-
 drivers/perf/alibaba_uncore_drw_pmu.c        |  10 +-
 drivers/perf/arm-cci.c                       |  24 +-
 drivers/perf/arm-ccn.c                       |  10 +-
 drivers/perf/arm_dmc620_pmu.c                |   9 +-
 drivers/perf/arm_pmuv3.c                     |   8 +-
 drivers/scsi/mpi3mr/mpi3mr_os.c              |  21 +-
 drivers/scsi/qedi/qedi_main.c                |   9 +-
 drivers/scsi/scsi_lib.c                      |   7 +-
 drivers/tty/nozomi.c                         |   5 +-
 drivers/tty/serial/sc16is7xx.c               |   8 +-
 drivers/usb/class/cdc-acm.c                  |   5 +-
 include/linux/cpumask.h                      |  12 +
 include/linux/find.h                         | 293 +++++++++++++++++++
 kernel/sched/sched.h                         |  14 +-
 kernel/watch_queue.c                         |   6 +-
 lib/find_bit.c                               |  85 ++++++
 lib/sbitmap.c                                |  46 +--
 lib/test_bitmap.c                            |  61 ++++
 net/bluetooth/cmtp/core.c                    |  10 +-
 net/smc/smc_wr.c                             |  10 +-
 sound/pci/hda/hda_codec.c                    |   7 +-
 sound/usb/caiaq/audio.c                      |  13 +-
 49 files changed, 629 insertions(+), 420 deletions(-)

Comments

Andy Shevchenko Dec. 4, 2023, 1:07 p.m. UTC | #1
On Sun, Dec 03, 2023 at 11:23:47AM -0800, Yury Norov wrote:
> Add helpers around test_and_{set,clear}_bit() that allow to search for
> clear or set bits and flip them atomically.
> 
> The target patterns may look like this:
> 
> 	for (idx = 0; idx < nbits; idx++)
> 		if (test_and_clear_bit(idx, bitmap))
> 			do_something(idx);
> 
> Or like this:
> 
> 	do {
> 		bit = find_first_bit(bitmap, nbits);
> 		if (bit >= nbits)
> 			return nbits;
> 	} while (!test_and_clear_bit(bit, bitmap));
> 	return bit;
> 
> In both cases, the opencoded loop may be converted to a single function
> or iterator call. Correspondingly:
> 
> 	for_each_test_and_clear_bit(idx, bitmap, nbits)
> 		do_something(idx);
> 
> Or:
> 	return find_and_clear_bit(bitmap, nbits);
> 
> Obviously, the less routine code people have to write themself, the
> less probability to make a mistake.
> 
> Those are not only handy helpers but also resolve a non-trivial
> issue of using non-atomic find_bit() together with atomic
> test_and_{set,clear)_bit().
> 
> The trick is that find_bit() implies that the bitmap is a regular
> non-volatile piece of memory, and compiler is allowed to use such
> optimization techniques like re-fetching memory instead of caching it.
> 
> For example, find_first_bit() is implemented like this:
> 
>       for (idx = 0; idx * BITS_PER_LONG < sz; idx++) {
>               val = addr[idx];
>               if (val) {
>                       sz = min(idx * BITS_PER_LONG + __ffs(val), sz);
>                       break;
>               }
>       }
> 
> On register-memory architectures, like x86, compiler may decide to
> access memory twice - first time to compare against 0, and second time
> to fetch its value to pass it to __ffs().
> 
> When running find_first_bit() on volatile memory, the memory may get
> changed in-between, and for instance, it may lead to passing 0 to
> __ffs(), which is undefined. This is a potentially dangerous call.
> 
> find_and_clear_bit() as a wrapper around test_and_clear_bit()
> naturally treats underlying bitmap as a volatile memory and prevents
> compiler from such optimizations.
> 
> Now that KCSAN is catching exactly this type of situations and warns on
> undercover memory modifications. We can use it to reveal improper usage
> of find_bit(), and convert it to atomic find_and_*_bit() as appropriate.
> 
> The 1st patch of the series adds the following atomic primitives:
> 
> 	find_and_set_bit(addr, nbits);
> 	find_and_set_next_bit(addr, nbits, start);
> 	...
> 
> Here find_and_{set,clear} part refers to the corresponding
> test_and_{set,clear}_bit function. Suffixes like _wrap or _lock
> derive their semantics from corresponding find() or test() functions.
> 
> For brevity, the naming omits the fact that we search for zero bit in
> find_and_set, and correspondingly search for set bit in find_and_clear
> functions.
> 
> The patch also adds iterators with atomic semantics, like
> for_each_test_and_set_bit(). Here, the naming rule is to simply prefix
> corresponding atomic operation with 'for_each'.
> 
> This series is a result of discussion [1]. All find_bit() functions imply
> exclusive access to the bitmaps. However, KCSAN reports quite a number
> of warnings related to find_bit() API. Some of them are not pointing
> to real bugs because in many situations people intentionally allow
> concurrent bitmap operations.
> 
> If so, find_bit() can be annotated such that KCSAN will ignore it:
> 
>         bit = data_race(find_first_bit(bitmap, nbits));
> 
> This series addresses the other important case where people really need
> atomic find ops. As the following patches show, the resulting code
> looks safer and more verbose comparing to opencoded loops followed by
> atomic bit flips.
> 
> In [1] Mirsad reported 2% slowdown in a single-thread search test when
> switching find_bit() function to treat bitmaps as volatile arrays. On
> the other hand, kernel robot in the same thread reported +3.7% to the
> performance of will-it-scale.per_thread_ops test.
> 
> Assuming that our compilers are sane and generate better code against
> properly annotated data, the above discrepancy doesn't look weird. When
> running on non-volatile bitmaps, plain find_bit() outperforms atomic
> find_and_bit(), and vice-versa.

...

In some cases the better improvements can be achieved by switching
the (very) old code to utilise IDA framework.
Yury Norov Dec. 6, 2023, 5:22 a.m. UTC | #2
On Mon, Dec 04, 2023 at 07:51:01PM +0100, Jan Kara wrote:
> Hello Yury!
> 
> On Sun 03-12-23 11:23:47, Yury Norov wrote:
> > Add helpers around test_and_{set,clear}_bit() that allow to search for
> > clear or set bits and flip them atomically.
> > 
> > The target patterns may look like this:
> > 
> > 	for (idx = 0; idx < nbits; idx++)
> > 		if (test_and_clear_bit(idx, bitmap))
> > 			do_something(idx);
> > 
> > Or like this:
> > 
> > 	do {
> > 		bit = find_first_bit(bitmap, nbits);
> > 		if (bit >= nbits)
> > 			return nbits;
> > 	} while (!test_and_clear_bit(bit, bitmap));
> > 	return bit;
> > 
> > In both cases, the opencoded loop may be converted to a single function
> > or iterator call. Correspondingly:
> > 
> > 	for_each_test_and_clear_bit(idx, bitmap, nbits)
> > 		do_something(idx);
> > 
> > Or:
> > 	return find_and_clear_bit(bitmap, nbits);
> 
> These are fine cleanups but they actually don't address the case that has
> triggered all these changes - namely the xarray use of find_next_bit() in
> xas_find_chunk().
> 
> ...
> > This series is a result of discussion [1]. All find_bit() functions imply
> > exclusive access to the bitmaps. However, KCSAN reports quite a number
> > of warnings related to find_bit() API. Some of them are not pointing
> > to real bugs because in many situations people intentionally allow
> > concurrent bitmap operations.
> > 
> > If so, find_bit() can be annotated such that KCSAN will ignore it:
> > 
> >         bit = data_race(find_first_bit(bitmap, nbits));
> 
> No, this is not a correct thing to do. If concurrent bitmap changes can
> happen, find_first_bit() as it is currently implemented isn't ever a safe
> choice because it can call __ffs(0) which is dangerous as you properly note
> above. I proposed adding READ_ONCE() into find_first_bit() / find_next_bit()
> implementation to fix this issue but you disliked that. So other option we
> have is adding find_first_bit() and find_next_bit() variants that take
> volatile 'addr' and we have to use these in code like xas_find_chunk()
> which cannot be converted to your new helpers.

Here is some examples when concurrent operations with plain find_bit()
are acceptable:

 - two threads running find_*_bit(): safe wrt ffs(0) and returns correct
   value, because underlying bitmap is unchanged;
 - find_next_bit() in parallel with set or clear_bit(), when modifying
   a bit prior to the start bit to search: safe and correct;
 - find_first_bit() in parallel with set_bit(): safe, but may return wrong
   bit number;
 - find_first_zero_bit() in parallel with clear_bit(): same as above.

In last 2 cases find_bit() may not return a correct bit number, but
it may be OK if caller requires any (not exactly first) set or clear
bit, correspondingly.

In such cases, KCSAN may be safely silenced.
 
> > This series addresses the other important case where people really need
> > atomic find ops. As the following patches show, the resulting code
> > looks safer and more verbose comparing to opencoded loops followed by
> > atomic bit flips.
> > 
> > In [1] Mirsad reported 2% slowdown in a single-thread search test when
> > switching find_bit() function to treat bitmaps as volatile arrays. On
> > the other hand, kernel robot in the same thread reported +3.7% to the
> > performance of will-it-scale.per_thread_ops test.
> 
> It was actually me who reported the regression here [2] but whatever :)
> 
> [2] https://lore.kernel.org/all/20231011150252.32737-1-jack@suse.cz

My apologize.

> > Assuming that our compilers are sane and generate better code against
> > properly annotated data, the above discrepancy doesn't look weird. When
> > running on non-volatile bitmaps, plain find_bit() outperforms atomic
> > find_and_bit(), and vice-versa.
> > 
> > So, all users of find_bit() API, where heavy concurrency is expected,
> > are encouraged to switch to atomic find_and_bit() as appropriate.
> 
> Well, all users where any concurrency can happen should switch. Otherwise
> they are prone to the (admittedly mostly theoretical) data race issue.
> 
> 								Honza
> -- 
> Jan Kara <jack@suse.com>
> SUSE Labs, CR
Jan Kara Dec. 7, 2023, 9:10 a.m. UTC | #3
On Tue 05-12-23 21:22:59, Yury Norov wrote:
> On Mon, Dec 04, 2023 at 07:51:01PM +0100, Jan Kara wrote:
> > > This series is a result of discussion [1]. All find_bit() functions imply
> > > exclusive access to the bitmaps. However, KCSAN reports quite a number
> > > of warnings related to find_bit() API. Some of them are not pointing
> > > to real bugs because in many situations people intentionally allow
> > > concurrent bitmap operations.
> > > 
> > > If so, find_bit() can be annotated such that KCSAN will ignore it:
> > > 
> > >         bit = data_race(find_first_bit(bitmap, nbits));
> > 
> > No, this is not a correct thing to do. If concurrent bitmap changes can
> > happen, find_first_bit() as it is currently implemented isn't ever a safe
> > choice because it can call __ffs(0) which is dangerous as you properly note
> > above. I proposed adding READ_ONCE() into find_first_bit() / find_next_bit()
> > implementation to fix this issue but you disliked that. So other option we
> > have is adding find_first_bit() and find_next_bit() variants that take
> > volatile 'addr' and we have to use these in code like xas_find_chunk()
> > which cannot be converted to your new helpers.
> 
> Here is some examples when concurrent operations with plain find_bit()
> are acceptable:
> 
>  - two threads running find_*_bit(): safe wrt ffs(0) and returns correct
>    value, because underlying bitmap is unchanged;
>  - find_next_bit() in parallel with set or clear_bit(), when modifying
>    a bit prior to the start bit to search: safe and correct;
>  - find_first_bit() in parallel with set_bit(): safe, but may return wrong
>    bit number;
>  - find_first_zero_bit() in parallel with clear_bit(): same as above.
> 
> In last 2 cases find_bit() may not return a correct bit number, but
> it may be OK if caller requires any (not exactly first) set or clear
> bit, correspondingly.
> 
> In such cases, KCSAN may be safely silenced.

True - but these are special cases. In particular the case in xas_find_chunk()
is not any of these special cases. It is using find_next_bit() which is can
be racing with clear_bit(). So what are your plans for such usecase?

								Honza