Changes for drivers those have taskqueue_create(M_WAITOK) / taskqueue_create_fast(M_WAITOK) is split into D45853 .

The full diff is quite large. See https://github.com/gmshake/freebsd-src/tree/main_m_waitok or I can upload it here if requested.

This looks good. Likely a side effect of a different auto* version.

I did take a quick look at the Linux code and it
appears to be completely different to me, so I
don't think trying to convert it makes sense, at
least for now.

Used strdup() to copy the options into a malloc'd area,
as suggested by mav@. This makes sense, since a sharenfs
with many semicolons could be much larger than an
option list.

In D45434#1045067, @jhb wrote:

@imp did you get a chance to test this?

With D45781.140474.patch added I got a deadlock after 6h30:
https://people.freebsd.org/~pho/stress/log/log0534.txt
It's not clear to me if this is related to your patch, so I'll repeat the test with a pristine kernel.

Revise according Hiroki advise

Oh, didn't hit submit before, though this doesn't affect the commit

In D45826#1045245, @kib wrote:

In D45826#1044991, @rlibby wrote:

By the way, I thought about also making these strict-aliasing safe with a union. It's straightforward and also gets rid of the casts. However, since we explicitly build with -fno-strict-aliasing, there's not much motivation. But in case you'd prefer that, I can tweak these patches.

Why would a special treat for aliasing needed? There is the "memory" clobber.

update with jhb's suggestions. Hopefully I got them right