On Tue, Mar 17, 2026 at 08:39:16AM +0100, Greg Kroah-Hartman wrote:
Wait, I removed them all, as that should have fixed the issue (I removed the offending commit that was originally causing the build problem here.)
Nathan, what errors are you seeing now? None of these changes are in a release, and I don't see any i3c patches in the current queue.
Right, you did drop those changes back in January. However, Sasha added some of those i3c patches back in 5.15.202.
I see
$ git show -s --format='%h ("%s")' HEAD 91d48252ad4b ("Linux 5.15.202")
$ git log --format=oneline v5.15.201..v5.15.202 drivers/i3c cde3d9035e8afcb3b54aa233c51a0babb5013c67 i3c: master: svc: Initialize 'dev' to NULL in svc_i3c_master_ibi_isr() 6cae2af141da60a68f9e733f69821c3ea457b0c0 i3c: Move device name assignment after i3c_bus_init acbd89719cb195143b5fc610c82e18f887631eaf i3c: remove i2c board info from i2c_dev_desc
$ make -skj"$(nproc)" ARCH=x86_64 LLVM=1 mrproper allmodconfig drivers/i3c/master.o drivers/i3c/master.c:2203:3: error: variable 'i2cdev' is uninitialized when used here [-Werror,-Wuninitialized] 2203 | i2cdev->dev = i2c_new_client_device(adap, &i2cboardinfo->base); | ^~~~~~ ...
which is resolved by commit 6cbf8b38dfe3 ("i3c: fix uninitialized variable use in i2c setup") upstream, a fix of commit 31b9887c7258 ("i3c: remove i2c board info from i2c_dev_desc"), which was backported in 5.15.202, as you can see above.
I merely brought up what happened in January because that patch was already flagged as excessive but I guess it was a stable-dep for a different i3c backport back then.
Hopefully that makes sense, let me know if not.
Cheers, Nathan