On Mon, Nov 25, 2019 at 5:02 PM Hans Verkuil firstname.lastname@example.org wrote:
On 11/11/19 9:38 PM, Arnd Bergmann wrote:
I'm in the process of finishing up the last bits on y2038-unsafe code in the kernel, this series is for v4l2, which has no problem with overflow, but has multiple ioctls that break with user space built against a new 32-bit libc.
Thank you for working on this. Much appreciated!
I've reviewed this v4 series and found a few issues (mostly things that ended up in videodev2.h that shouldn't be there).
Once a v5 is posted I'll try to test this more.
Is there an easy(-ish) way to test this with a time64 ABI? Do you have such an environment set up for testing?
If you can build your user space with musl, you can test using a recent snapshot from http://git.musl-libc.org/cgit/musl/.
I'm not entirely happy with the compat-ioctl implementation that adds quite a bit of code duplication, but I hope this is acceptable anyway, as a better implementation would likely require a larger refactoring of the compat-ioctl file, while my approach simply adds support for the additional data structure variants.
A cleanup is indeed much more work. This y2038 code is awful, but that's really because the original structs are aligned in the worst possible way.