Add a INTR_TRIG_INVALID, and use it in the powerpc interrupt code.

[jhibbits]

Summary

Clang throws the following warning in powerpc intr_machdep:

/usr/src/sys/powerpc/powerpc/intr_machdep.c:454:15: warning: comparison of
constant -1 with expression of type 'enum intr_trigger' is always false
[-Wtautological-constant-out-of-range-compare]

if (i->trig == -1)

~~~~~~~ ^ ~~

This may lead to legitimate problems with aggressive optimizations, if not now
then in the future. To avoid this, add a new enum, INTR_TRIG_INVALID, set to
-1, and use this new enumeration in these checks.

Test Plan

Compile test.

[kib]

Did you compiled universe with the patch ? My concern is that now switch() statements which handled three existing INTR_TRIGGER_XXX values become incomplete and potentially trigger corresponding warning.

[jhibbits]

In D9300#192273, @kib wrote:

Did you compiled universe with the patch ? My concern is that now switch() statements which handled three existing INTR_TRIGGER_XXX values become incomplete and potentially trigger corresponding warning.

Nope, but that's a good point, didn't think about it. I'll kick off a universe build soon. If the fallout is too great, I may instead make the struct member an int instead of the enum, but if it's small enough, I can make the changes in the other files.

[jhibbits]

In D9300#192322, @jhibbits wrote:

In D9300#192273, @kib wrote:

Did you compiled universe with the patch ? My concern is that now switch() statements which handled three existing INTR_TRIGGER_XXX values become incomplete and potentially trigger corresponding warning.

Nope, but that's a good point, didn't think about it. I'll kick off a universe build soon. If the fallout is too great, I may instead make the struct member an int instead of the enum, but if it's small enough, I can make the changes in the other files.

Universe completed successfully (3 unrelated failures in netmap for powerpc and sparc LINT), and I see no warnings for this in the logs.