[Intel-gfx] [patch RFC 00/15] mm/highmem: Provide a preemptible variant of kmap_atomic & friends

Daniel Vetter daniel at ffwll.ch
Sat Sep 19 10:37:53 UTC 2020


On Sat, Sep 19, 2020 at 12:35 PM Daniel Vetter <daniel at ffwll.ch> wrote:
>
> On Sat, Sep 19, 2020 at 11:50 AM Thomas Gleixner <tglx at linutronix.de> wrote:
> >
> > First of all, sorry for the horribly big Cc list!
> >
> > Following up to the discussion in:
> >
> >   https://lore.kernel.org/r/20200914204209.256266093@linutronix.de
> >
> > this provides a preemptible variant of kmap_atomic & related
> > interfaces. This is achieved by:
> >
> >  - Consolidating all kmap atomic implementations in generic code
> >
> >  - Switching from per CPU storage of the kmap index to a per task storage
> >
> >  - Adding a pteval array to the per task storage which contains the ptevals
> >    of the currently active temporary kmaps
> >
> >  - Adding context switch code which checks whether the outgoing or the
> >    incoming task has active temporary kmaps. If so, the outgoing task's
> >    kmaps are removed and the incoming task's kmaps are restored.
> >
> >  - Adding new interfaces k[un]map_temporary*() which are not disabling
> >    preemption and can be called from any context (except NMI).
> >
> >    Contrary to kmap() which provides preemptible and "persistant" mappings,
> >    these interfaces are meant to replace the temporary mappings provided by
> >    kmap_atomic*() today.
> >
> > This allows to get rid of conditional mapping choices and allows to have
> > preemptible short term mappings on 64bit which are today enforced to be
> > non-preemptible due to the highmem constraints. It clearly puts overhead on
> > the highmem users, but highmem is slow anyway.
> >
> > This is not a wholesale conversion which makes kmap_atomic magically
> > preemptible because there might be usage sites which rely on the implicit
> > preempt disable. So this needs to be done on a case by case basis and the
> > call sites converted to kmap_temporary.
> >
> > Note, that this is only lightly tested on X86 and completely untested on
> > all other architectures.
> >
> > The lot is also available from
> >
> >    git://git.kernel.org/pub/scm/linux/kernel/git/tglx/devel.git highmem
>
> I think it should be the case, but I want to double check: Will
> copy_*_user be allowed within a kmap_temporary section? This would
> allow us to ditch an absolute pile of slowpaths.

(coffee just kicked in) copy_*_user is ofc allowed, but if you hit a
page fault you get a short read/write. This looks like it would remove
the need to handle these in a slowpath, since page faults can now be
served in this new kmap_temporary sections. But this sounds too good
to be true, so I'm wondering what I'm missing.
-Daniel
>
> >
> > Thanks,
> >
> >         tglx
> > ---
> >  a/arch/arm/mm/highmem.c               |  121 ---------------------
> >  a/arch/microblaze/mm/highmem.c        |   78 -------------
> >  a/arch/nds32/mm/highmem.c             |   48 --------
> >  a/arch/powerpc/mm/highmem.c           |   67 -----------
> >  a/arch/sparc/mm/highmem.c             |  115 --------------------
> >  arch/arc/Kconfig                      |    1
> >  arch/arc/include/asm/highmem.h        |    8 +
> >  arch/arc/mm/highmem.c                 |   44 -------
> >  arch/arm/Kconfig                      |    1
> >  arch/arm/include/asm/highmem.h        |   30 +++--
> >  arch/arm/mm/Makefile                  |    1
> >  arch/csky/Kconfig                     |    1
> >  arch/csky/include/asm/highmem.h       |    4
> >  arch/csky/mm/highmem.c                |   75 -------------
> >  arch/microblaze/Kconfig               |    1
> >  arch/microblaze/include/asm/highmem.h |    6 -
> >  arch/microblaze/mm/Makefile           |    1
> >  arch/microblaze/mm/init.c             |    6 -
> >  arch/mips/Kconfig                     |    1
> >  arch/mips/include/asm/highmem.h       |    4
> >  arch/mips/mm/highmem.c                |   77 -------------
> >  arch/mips/mm/init.c                   |    3
> >  arch/nds32/Kconfig.cpu                |    1
> >  arch/nds32/include/asm/highmem.h      |   21 ++-
> >  arch/nds32/mm/Makefile                |    1
> >  arch/powerpc/Kconfig                  |    1
> >  arch/powerpc/include/asm/highmem.h    |    6 -
> >  arch/powerpc/mm/Makefile              |    1
> >  arch/powerpc/mm/mem.c                 |    7 -
> >  arch/sparc/Kconfig                    |    1
> >  arch/sparc/include/asm/highmem.h      |    7 -
> >  arch/sparc/mm/Makefile                |    3
> >  arch/sparc/mm/srmmu.c                 |    2
> >  arch/x86/include/asm/fixmap.h         |    1
> >  arch/x86/include/asm/highmem.h        |   12 +-
> >  arch/x86/include/asm/iomap.h          |   29 +++--
> >  arch/x86/mm/highmem_32.c              |   59 ----------
> >  arch/x86/mm/init_32.c                 |   15 --
> >  arch/x86/mm/iomap_32.c                |   57 ----------
> >  arch/xtensa/Kconfig                   |    1
> >  arch/xtensa/include/asm/highmem.h     |    9 +
> >  arch/xtensa/mm/highmem.c              |   44 -------
> >  b/arch/x86/Kconfig                    |    3
> >  include/linux/highmem.h               |  141 +++++++++++++++---------
> >  include/linux/io-mapping.h            |    2
> >  include/linux/sched.h                 |    9 +
> >  kernel/sched/core.c                   |   10 +
> >  mm/Kconfig                            |    3
> >  mm/highmem.c                          |  192 ++++++++++++++++++++++++++++++++--
> >  49 files changed, 422 insertions(+), 909 deletions(-)
>
>
>
> --
> Daniel Vetter
> Software Engineer, Intel Corporation
> http://blog.ffwll.ch



-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch


More information about the Intel-gfx mailing list