What follows the pad site's extra word counts without it #650
Loading…
Reference in a new issue
No description provided.
Delete branch "rpg-9-ap-pad-alignment-20261028"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
What follows the pad site's extra word counts without it
A compile whose pad byte is 0x80 or above stores it as
li,rldicl,sth:one word more than the 0x00 to 0x7f form. v0.1.361 counted the site as one
pattern, but things that stand after the extra word still moved with it. The
saved r6e-num compile and its recompile, with bytes 0x2a and 0x8c, differed in
instructions, control flow and the constant area for that reason.
In a range that holds a pad site, these now count without the extra word:
8-byte aligned and so grow by 4 when the code grows by 4 (only the zero bytes
that the alignment explains are left out);
procedure row, and the size in the control flow;
A range without a pad site hashes as before: the saved comparison goldens of
the PF and LF corpus (
rpg-corpus-captures) are byte-identical. A compile witha pad byte of 0x80 and above still differs from one below it in one small
descriptor value of the constant area (a count that moves by one with the
word); that stays open.
Gates