Re: [PATCH 2/2] module: add support to avoid duplicates early on load

[Johan Hovold]

On Thu, May 25, 2023 at 06:39:52PM -0700, Linus Torvalds wrote:

> Ok, I decided to just move it from my experimental tree to my main tree.
> 
> I think I used about three times the time and effort (and lines of
> text) on writing the commit message compared to what I did on the
> patch itself.
> 
> I tried to lay out the background and the implications of the change -
> it may be pretty darn simple, but it does have some subtle issues.
> 
> Anyway: I've committed it to my tree. This is not necessarily the best
> time to do that, but let's get this behind us, and in particular,
> let's get it out and into wider testing asap.
> 
> If it causes any problems what-so-ever, I'll just revert it very
> aggressively (unless the problem is trivially and obviously fixable).
> It is, after all, not a fix for a _kernel_ bug per se, and whil eI
> think the patch is very benign, it does change user-visible behavior.
> Very intentionally so, but still..

This change breaks module loading during boot on the Lenovo Thinkpad
X13s (aarch64).

Specifically it results in indefinite probe deferral of the display and
USB (ethernet) which makes it a pain to debug. Typing in the dark to
acquire some logs reveals that other modules are missing as well.

Fortunately commit 9828ed3f695a ("module: error out early on concurrent
load of the same module file") stood out when skimming the changes that
went into -rc4, and reverting it make all the expected modules be loaded
again.

I have not tried to figure out exactly why things break, but it does
seem like this one should be reverted.

Johan