On Thu, Feb 27, 2020 at 10:59:57AM -0800, Song Liu wrote: > On Thu, Feb 27, 2020 at 12:50 AM Jiri Olsa <jolsa@xxxxxxxxxx> wrote: > > > > On Wed, Feb 26, 2020 at 01:14:43PM -0800, Song Liu wrote: > > > On Wed, Feb 26, 2020 at 5:04 AM Jiri Olsa <jolsa@xxxxxxxxxx> wrote: > > > > > > > > Adding name to 'struct bpf_ksym' object to carry the name > > > > of the symbol for bpf_prog, bpf_trampoline, bpf_dispatcher. > > > > > > > > The current benefit is that name is now generated only when > > > > the symbol is added to the list, so we don't need to generate > > > > it every time it's accessed. > > > > > > > > Signed-off-by: Jiri Olsa <jolsa@xxxxxxxxxx> > > > > > > The patch looks good. But I wonder whether we want pay the cost of > > > extra 128 bytes per bpf program. Maybe make it a pointer and only > > > generate the string when it is first used? > > > > I thought 128 would not be that bad, also the code is quite > > simple because of that.. if that's really a concern I could > > make the changes, but that would probably mean changing the > > design > > I guess this is OK. We can further optimize it if needed. > > Acked-by: Song Liu <songliubraving@xxxxxx> > ok, thanks for the review, I still have to make some changes, so I'll keep your acked-by on patches that won't be changed, please scream otherwise ;-) thanks, jirka