linuxkpi: Fix iteration in __sg_alloc_table_from_pages again

[ashafer_badland.io]

Summary

Commit 3f686532c9b4 tried to fix an issue with not properly starting
at the first page in the sg list to prevent a panic. This worked but
with the side effect of incrementing "s" during the final iteration,
causing it to be NULL since the list had ended.

This also causes a panic with drm-5.15, since s is null when we later
pass it to sg_mark_end. This change decouples the iteration sg from
s so that s is never incremented past the final page in the chain.

[ashafer_badland.io]

This is part of a fix for https://github.com/amshafer/nvidia-driver/issues/20

I don't think I noticed this problem with my fix the first time since the sg_mark_end(s) call is only hit in drm 5.15, I'm guessing I tested with 5.10.

[bz]

sys/compat/linuxkpi/common/include/linux/scatterlist.h

386

Please make this a C-style comment. /* .. */

[ashafer_badland.io]

Fixed styling

[bz]

@dumbbell @manu @wulf unless someone else will review it, I'll likely commit it the next days so it doesn't get lost.

@ashafer_badland.io whe you suggest in your commit message that we panic later passing a NULL, I assume you only test with a non-debug kernel and never hit the KASSERT after the loop which should catch exactly that?

[manu]

In D41574#951132, @bz wrote:

@dumbbell @manu @wulf unless someone else will review it, I'll likely commit it the next days so it doesn't get lost.

@ashafer_badland.io whe you suggest in your commit message that we panic later passing a NULL, I assume you only test with a non-debug kernel and never hit the KASSERT after the loop which should catch exactly that?

Sorry, forgot about this review, please do commit it so it's in 14.0

[ashafer_badland.io]

Off the top of my head I thought I tested with GENERIC but because that KASSERT didn't fire I'm assuming I had GENERIC-NODEBUG installed, which I normally do on half my machines.