On 30.04.2025 18:00, Andrew Cooper wrote: > On 30/04/2025 2:13 pm, Jan Beulich wrote: >> In a number of cases we use ALTERNATIVE_2 with both replacement insns / >> insn sequences being identical. Avoid emitting the same code twice, and >> instead alias the necessary helper labels to the existing ones. >> >> Signed-off-by: Jan Beulich <jbeul...@suse.com> > > On a random build, the size of .altinstr_replacement drops from 0xe47 to > 0xdf8, so not too bad. > > While the patch is fine, if we're adjusting the assembly ALTERNATIVE_2, > we should make the same adjustment to the C version, even if there's > nothing to benefit from it immediately.
Can do, but I expect that to end up more clumsy for, as you say, no real gain. > P.S. it would be even nicer if we would put these in mergeable sections, > but I haven't figured out way to set the mergable unit size, which needs > to be an absolute expression. Well, have you missed https://lists.xen.org/archives/html/xen-devel/2021-11/msg01009.html then? Yet it might still make sense to use the approach here as well, as there are limitations to what can really be merged (by the toolchain). Jan
