[PATCH 0/3] RISC-V: alignment in text sections

[Jan Beulich]

The other day I suggested to a co-worker to uniformly use ".insn 0"
to mark certain "bad" positions in code (Linux'es BUG() and alike).
Trying it out myself, I was surprised that I was unable to re-align
code after such a .insn use (or, equivalently, after .hword). I'm
aiming at fixing issues noticed in this area with this series.

1: process rs_align_code also when relaxing
2: correct alignment directive handling for text sections
3: odd data padding vs mapping symbols

However, this isn't quite enough yet. Tracking is now (I hope; maybe
better to say should now be) consistently per-section. Yet that's
still not fine-grained enough: It needs to be per-subsection. I'm
intending to clone x86'es respective handling to further address this
as well. Yet in the meantime the fixes here could already go in.

Jan