[RFC PATCH] mm: z3fold: rename CONFIG_Z3FOLD to CONFIG_Z3FOLD_DEPRECATED
Nhat Pham
nphamcs at gmail.com
Tue Jan 16 07:01:58 AEDT 2024
On Mon, Jan 15, 2024 at 4:27 AM Vitaly Wool <vitaly.wool at konsulko.com> wrote:
>
> On Fri, Jan 12, 2024 at 8:31 PM Yosry Ahmed <yosryahmed at google.com> wrote:
> >
> > The z3fold compressed pages allocator is not widely used, most users use
> > zsmalloc. The only disadvantage of zsmalloc in comparison is the
> > dependency on MMU, and zbud is a more common option for !MMU as it was
> > the default zswap allocator for a long time.
> >
> > In hopes of having a single compressed pages allocator at some point,
> > and following in the footsteps of SLAB, deprecate z3fold. Rename the
> > user-visible option so that users with CONFIG_Z3FOLD=y get a new prompt
> > with explanation during make oldconfig. Remove CONFIG_Z3FOLD=y from
> > defconfigs.
>
> I believe that having a single compressed pages allocator is a false goal.
I think it should be, where possible. Having fewer code will lower the
maintenance burden. Of course, this is not always possible, as one
option might be the best in one aspect/use case, but not so in others.
In the case of zswap (which is the only user for zbud and z3fold), it
seems (to me anyway) zbud and z3fold aren't really serious competitors
for zsmalloc anymore (outside of that aforementioned !MMU case). I'd
love to hear some use cases if I'm mistaken though :)
>
> > Existing users, if any, should voice their objections. Otherwise, we can
> > remove z3fold in a few releases.
>
> At this point I NACK this patch. We're about to submit an allocator
> which is clearly better that z3fold and is faster that zsmalloc in
> most cases and that submission will mark z3fold as deprecated. But for
> now this move is premature.
Ah, this sounds promising. I'd love to hear more about this new
allocator, and once it's available, experiment with it internally too
:)
That said, even at this point, does anyone actually use z3fold and/or
zbud, and cannot use zsmalloc? If yes, then yeah this is quite
premature. If not, then we can mark them as deprecate, no? Introducing
the other allocator and fixing zsmalloc for !MMU can be done in
parallel - we're not removing z3fold and zbud anytime soon even with
this deprecation notice.
>
> Best,
> Vitaly
>
> > Signed-off-by: Yosry Ahmed <yosryahmed at google.com>
> > ---
> > I have limited understanding of Kconfigs. I modelled this after commit
> > eb07c4f39c3e ("mm/slab: rename CONFIG_SLAB to CONFIG_SLAB_DEPRECATED"),
> > but one difference is that CONFIG_Z3FOLD is a tristate. I made
> > CONFIG_Z3FOLD_DEPRECATED a boolean config, and CONFIG_Z3FOLD default y
> > so that it is on by default if CONFIG_Z3FOLD_DEPRECATED is selected. I
> > am not sure if that's the correct way to do this.
> >
> > ---
> > arch/loongarch/configs/loongson3_defconfig | 1 -
> > arch/powerpc/configs/ppc64_defconfig | 1 -
> > mm/Kconfig | 13 +++++++++++--
> > 3 files changed, 11 insertions(+), 4 deletions(-)
> >
> > diff --git a/arch/loongarch/configs/loongson3_defconfig b/arch/loongarch/configs/loongson3_defconfig
> > index 33795e4a5bd63..89b66b6c6a1d5 100644
> > --- a/arch/loongarch/configs/loongson3_defconfig
> > +++ b/arch/loongarch/configs/loongson3_defconfig
> > @@ -85,7 +85,6 @@ CONFIG_ZPOOL=y
> > CONFIG_ZSWAP=y
> > CONFIG_ZSWAP_COMPRESSOR_DEFAULT_ZSTD=y
> > CONFIG_ZBUD=y
> > -CONFIG_Z3FOLD=y
> > CONFIG_ZSMALLOC=m
> > # CONFIG_COMPAT_BRK is not set
> > CONFIG_MEMORY_HOTPLUG=y
> > diff --git a/arch/powerpc/configs/ppc64_defconfig b/arch/powerpc/configs/ppc64_defconfig
> > index 544a65fda77bc..d39284489aa26 100644
> > --- a/arch/powerpc/configs/ppc64_defconfig
> > +++ b/arch/powerpc/configs/ppc64_defconfig
> > @@ -81,7 +81,6 @@ CONFIG_MODULE_SIG_SHA512=y
> > CONFIG_PARTITION_ADVANCED=y
> > CONFIG_BINFMT_MISC=m
> > CONFIG_ZSWAP=y
> > -CONFIG_Z3FOLD=y
> > CONFIG_ZSMALLOC=y
> > # CONFIG_SLAB_MERGE_DEFAULT is not set
> > CONFIG_SLAB_FREELIST_RANDOM=y
> > diff --git a/mm/Kconfig b/mm/Kconfig
> > index 1902cfe4cc4f5..bc6cc97c08349 100644
> > --- a/mm/Kconfig
> > +++ b/mm/Kconfig
> > @@ -193,15 +193,24 @@ config ZBUD
> > deterministic reclaim properties that make it preferable to a higher
> > density approach when reclaim will be used.
> >
> > -config Z3FOLD
> > - tristate "3:1 compression allocator (z3fold)"
> > +config Z3FOLD_DEPRECATED
> > + bool "3:1 compression allocator (z3fold) (DEPRECATED)"
> > depends on ZSWAP
> > help
> > + Deprecated and scheduled for removal in a few cycles. If you have
> > + a good reason for using Z3FOLD rather than ZSMALLOC or ZBUD, please
> > + contact linux-mm at kvack.org and the zswap maintainers.
> > +
> > A special purpose allocator for storing compressed pages.
> > It is designed to store up to three compressed pages per physical
> > page. It is a ZBUD derivative so the simplicity and determinism are
> > still there.
> >
> > +config Z3FOLD
> > + tristate
> > + default y
> > + depends on Z3FOLD_DEPRECATED
> > +
> > config ZSMALLOC
> > tristate
> > prompt "N:1 compression allocator (zsmalloc)" if ZSWAP
> > --
> > 2.43.0.275.g3460e3d667-goog
> >
More information about the Linuxppc-dev
mailing list