problem with q->make_request_fn() in upstream lk 5.9+.

[Sergei Shtepa]

Hello.
I would like to draw your attention to the innovations of the upstream kernel and discuss the problem of intercepting the make_request_fn () function. I am concerned about a kernel 5.9 issue where intercepting the q->make_request_fn() function became impossible.

The problem is that several out-of-tree kernel modules can't run on it (only with a trick with x64_CR0_WP). First of all, this applies to backup tools. Many of them use their kernel modules to create snapshots of block devices. ASMFD will also not be able to connect to the block layer.

I have studied your kernel and see that these block layer changes are not present in your kernel. Not yet. But after some time, you will probably make a synchronization with the upstream kernel and the problem will appear to you.

I am trying to develop and offer to upstream kernel a solution to this problem https://github.com/SergeiShtepa/linux/tree/blk-filter-v0 . I believe that you may be interested in this issue and want to discuss it. I am sure that the Generic Block Device Filter module can be useful for Oracle.