[Bug tree-optimization/84011] Optimize switch table with run-time relocation

[peter at cordes dot ca]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84011

--- Comment #12 from Peter Cordes <peter at cordes dot ca> ---
(In reply to Jakub Jelinek from comment #10)
> (In reply to Peter Cordes from comment #9)
> > gcc already totally misses optimizations here where one string is a suffix
> > of another.  "mii" could just be a pointer to the 3rd byte of "sgmii", but
> > we instead duplicate all the characters.  That's where major savings are
> > possible for this function.
> 
> ??  That is the task for the linker SHF_MERGE|SHF_STRINGS handling.
> Why should gcc duplicate that?

Oops, right I was only looking at gcc's asm output, didn't check an actual
linked binary.

Will the linker currently catch a case like this?

.LC_base:
.LC2: .string "mii"
.LC3: .string "gmii"

table:
.byte  .LC2 - .LC_base,  .LC3 - .LC_base

and drop .string "mii" entirely + rewrite the table to
.byte  .LC3+1 - .LC_base,  .LC3 - .LC_base

(This discussion should probably be happening on bug 85585.)

Sorry I don't know the actual mechanism by which gcc signals to the linker that
it can / can't merge.  I guess only in some sections?  Because gcc couldn't
allow it if was emitting an array like this, where dropping a string would
change the offsets for later data and break offset calculations:

const struct { char str[11]; } table[] = { {"mii"}, {"gmii"} };