This patch was superseded by the refactoring of D50849 and D50850.

@bz: Do you have any remaining objections to this patch?

Also, on Linux, struct eventfd_ctx is defined privately in fs/eventfd.c. The struct fields are never accessed by consumers of the eventfd API.

Thank you for all your feedbacks!

Refresh patch after abandonning D50848.

Refresh patch after abandonning D50848.

Refresh patch after abandonning D50848.

In D51559#1237814, @bz wrote:

According to your comments the open TODOs are done. Did you forget to push the updated version of the change?

Address comments from @bz

I spotted another typo (see the attached comment). I approve it at the same time to not waste time in back and forth comments.

As I understand, HWEIGHT* are always supposed to be run on compile-time constants, so they could just be defined to this const_bitcount macro.

I can create a patch for this if this seems like a good solution to you all.

Handle negative return value from callback.

Rebase on top of main branch.

In D50848#1164156, @emaste wrote:

Hmm, OK. The struct is different and we might end up with a slightly puzzling build failure if things end up cross-threaded, but I suppose it's not really different than the status quo.

Thank you @emaste for the bugfix. I was not sure myself if the + 1 was intended.

Address comment from @bz: default the #if/#else logic to the latest version. A user has to define LINUXKPI_VERSION to request the older variant.

Address comment from @bz: move the definition of div64_ul() just after div64_u64().

Address comments from @wulf and @bz: don’t introduce a <uapi/linux/eventfd.h> header.