[GIT PULL] i2c for v7.4, part 2

[GIT PULL] i2c for v7.4, part 2

[Andi Shyti]

Hi Linus,

as anticipated, here is the second part of the I2C pull request
for v7.4.

This round introduces dynamic transfer timeouts in the I2C core,
already used by qcom-geni. There is also support for additional
transfer modes in imx and a handful of driver fixes.

I've also merged Igor's Rust I2C pull request, which removes an
unsafe conversion.

The branch is rebased on top of your part 1 I2C merge[*].

I'll now start queuing patches for a possible third week of the
merge window. Otherwise, they will go into v7.5.

Thanks and wishing you a good rest of your holidays,
Andi

[*] bb436f2a8568 ("Merge tag 'i2c-7.4-part1' ...")

The following changes since commit bb436f2a8568d14f176475c79b0d5069f0a6f9e3:

  Merge tag 'i2c-7.4-part1' of git://git.kernel.org/pub/scm/linux/kernel/git/andi.shyti/linux (2026-10-22 07:36:22 -1000)

are available in the Git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/andi.shyti/linux.git tags/i2c-7.4-part2

for you to fetch changes up to f7b70efba26963bf725f10574cc071f86cd5ee1d:

  i2c: sh_mobile: request IRQs after mapping registers (2026-10-24 00:05:15 +0200)

----------------------------------------------------------------
i2c for v7.4, part 2

This second part introduces dynamic transfer timeouts in the I2C
core, already used by qcom-geni. Alongside some driver
improvements and fixes, it also brings in the Rust I2C pull
request.

Core:

- core, qcom-geni: add and use a dynamic timeout helper
- acpi: fix adapter reference leak when creating I2C clients

Drivers:

- imx: support reversed address direction and no-start transfers
- designware: queue block data reads once length is known
- mt7621: skip the second data register for chunks up to 4 bytes
- sh_mobile: ensure IRQs are freed before unmapping registers

Rust:

- Remove unsound device-to-client reference conversion

----------------------------------------------------------------
Andi Shyti (1):
      Merge tag 'rust-i2c-for-7.4' into i2c/i2c

Aniket Randive (2):
      i2c: core: Add i2c_update_timeout() helper for dynamic transfer timeouts
      i2c: qcom-geni: Add dynamic transfer timeout based on transfer length and frequency

Christian Marangi (1):
      i2c: mt7621: fix copying uninitialized kernel stack with reduced page_len

Ma Ke (1):
      i2c: acpi: Fix adapter reference leak in i2c_acpi_new_device_by_fwnode()

Myeonghun Pak (1):
      i2c: sh_mobile: request IRQs after mapping registers

Navon John Lukose (1):
      i2c: designware: queue only one read command for block reads

Nicolás Antinori (1):
      rust: i2c: remove TryFrom<&Device> implementation for &I2cClient

Vincent Jardin (1):
      i2c: imx: add I2C_M_REV_DIR_ADDR and I2C_M_NOSTART

 drivers/i2c/Kconfig                        |  13 +++++++++++++
 drivers/i2c/busses/i2c-designware-master.c |   8 ++++++++
 drivers/i2c/busses/i2c-imx.c               | 157 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++----------------------------------------
 drivers/i2c/busses/i2c-mt7621.c            |   9 ++++++---
 drivers/i2c/busses/i2c-qcom-geni.c         |  48 +++++++++++++++++++++++++++++++++++++++---------
 drivers/i2c/busses/i2c-sh_mobile.c         |   8 ++++----
 drivers/i2c/i2c-core-acpi.c                |   6 +++++-
 drivers/i2c/i2c-core-base.c                |  43 +++++++++++++++++++++++++++++++++++++++++++
 drivers/i2c/i2c-dev.c                      |   3 +++
 include/linux/i2c.h                        |  13 +++++++++++++
 rust/kernel/i2c.rs                         |  21 ---------------------
 11 files changed, 251 insertions(+), 78 deletions(-)

Re: [GIT PULL] i2c for v7.4, part 2

[Linus Torvalds]

On Sun, 25 Oct 2026 at 10:20, Andi Shyti <andi.shyti@kernel.org> wrote:
>
> I'll now start queuing patches for a possible third week of the
> merge window. Otherwise, they will go into v7.5.

NO!

I've taken this.

But the whole "merge window is for new development" needs to stop. I
will *NOT* be taking any more patches from you that aren't fixes.

The merge window is for merging stuff that is ready BEFORE THE MERGE
WINDOW STARTED.

It's absolutely not for "let's add more to the pile". Never has been,
and never will be.

The merge window is two weeks NOT as a way to give maintainers two
weeks (and no, it won't be three weeks this time, I've been keeping
up) to create new stuff. It's so that the stuff that *was* ready
before the merge window can be merged in a sane and controlled manner.

So no. No more of this "I'll have a second - or third - pile of
stuff". If it wasn't ready before, then it wasn't ready, and it goes
into the *next* merge window.

                  Linus

Re: [GIT PULL] i2c for v7.4, part 2

[Andi Shyti]

On Mon, Oct 26, 2026 at 07:15:52AM -1000, Linus Torvalds wrote:
> On Sun, 25 Oct 2026 at 10:20, Andi Shyti <andi.shyti@kernel.org> wrote:
> >
> > I'll now start queuing patches for a possible third week of the
> > merge window. Otherwise, they will go into v7.5.
> 
> NO!
> 
> I've taken this.
> 
> But the whole "merge window is for new development" needs to stop. I
> will *NOT* be taking any more patches from you that aren't fixes.
> 
> The merge window is for merging stuff that is ready BEFORE THE MERGE
> WINDOW STARTED.
> 
> It's absolutely not for "let's add more to the pile". Never has been,
> and never will be.
> 
> The merge window is two weeks NOT as a way to give maintainers two
> weeks (and no, it won't be three weeks this time, I've been keeping
> up) to create new stuff. It's so that the stuff that *was* ready
> before the merge window can be merged in a sane and controlled manner.
> 
> So no. No more of this "I'll have a second - or third - pile of
> stuff". If it wasn't ready before, then it wasn't ready, and it goes
> into the *next* merge window.

OK, got it!

Just to clarify, most of the patches in this pull request are
fixes or patches that were taken well before the merge window but
had to wait for some dependencies to be resolved. They have also
been tested in my -next branch for several -rc cycles now.

Since the merge window opened, I've added only two patches that
are not strictly fixes[*]. One is a trivial fix that doesn't
deserve a Fixes: tag, while the other has been properly reviewed
and in my tree for almost a week.

So, apart from this last patch, which I could drop to be fully
compliant, I believe this pull request meets your requirements.

Am I still missing something? Is there anything I can still do?

As for my mention of a possible third week and a part 3 (not this
part 2 pull request!), I was simply trying to check whether that
would be acceptable for trivial patches. I understand now that it
definitely isn't.

Thanks,
Andi

[*] f7b70efba269 ("i2c: sh_mobile: request IRQs after mapping registers")
    c2867bb0fa1d ("i2c: imx: add I2C_M_REV_DIR_ADDR and I2C_M_NOSTART")

Re: [GIT PULL] i2c for v7.4, part 2

[Andi Shyti]

Hi Linus,

> Am I still missing something? Is there anything I can still do?
> 
> As for my mention of a possible third week and a part 3 (not this
> part 2 pull request!), I was simply trying to check whether that
> would be acceptable for trivial patches. I understand now that it
> definitely isn't.

sorry, I understood you were rejecting the pull request. We are
fully aligned.

Thanks,
Andi

Re: [GIT PULL] i2c for v7.4, part 2

[pr-tracker-bot]

The pull request you sent on Sun, 25 Oct 2026 21:20:26 +0100:

> git://git.kernel.org/pub/scm/linux/kernel/git/andi.shyti/linux.git tags/i2c-7.4-part2

has been merged into torvalds/linux.git:
https://git.kernel.org/torvalds/c/25e7ab89f4774b0f842cee0e8a20c11f755924d5

Thank you!

-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/prtracker.html