Re: [PATCH v2 2/2] mm/vmpressure: split v1 userspace eventfd code into vmpressure-v1.c

From: Johannes Weiner

Date: Mon Jun 29 2026 - 14:16:07 EST


On Mon, Jun 29, 2026 at 06:23:58PM +0100, Usama Arif wrote:
>
>
> On 29/06/2026 17:48, Johannes Weiner wrote:
> > On Mon, Jun 29, 2026 at 05:59:37AM -0700, Usama Arif wrote:
> >> @@ -101,7 +101,7 @@ obj-$(CONFIG_DEVICE_MIGRATION) += migrate_device.o
> >> obj-$(CONFIG_TRANSPARENT_HUGEPAGE) += huge_memory.o khugepaged.o
> >> obj-$(CONFIG_PAGE_COUNTER) += page_counter.o
> >> obj-$(CONFIG_LIVEUPDATE_MEMFD) += memfd_luo.o
> >> -obj-$(CONFIG_MEMCG_V1) += memcontrol-v1.o
> >> +obj-$(CONFIG_MEMCG_V1) += memcontrol-v1.o vmpressure-v1.o
> >
> > Might as well move the interface part to memcontrol-v1.c instead of
> > creating a new file.
> >
>
> I think it would make it easier once we move cgroup v2 off of
> vmpressure. Then we can rename vmpressure-v1.c to just vmpressure.c
> and gate it to CONFIG_MEMCG_V1. The other option would be everything
> living in memcontrol-v1.c?

Hm? I just mean move whatever you move into that new vmpressure-v1.c
into memcontrol-v1.c instead.

> I think its nice to keep it separate as
> memcontrol-v1.c is already 2K+ lines and this is standalone feature
> that can sit in a separate file.

It's all deprecated code with no active development. Containing it and
making it easy to see the footprint of the cgroup1 option is the
higher priority, more than maintaining neat component separation.

No need to work it naturally into the file either - decls/variables up
top, functions somewhere further down etc. Just paste it into one
contiguous vmpressure block under memcg1_oom_finish(). That's how that
file is already structured with the other components in there.