On Fri, Jan 12, 2024 at 01:41:41PM -0700, Daniel Xu wrote: > Hi Jiri, > > Thanks for reviewing! > > On Fri, Jan 12, 2024 at 05:16:07PM +0100, Jiri Olsa wrote: > > On Wed, Jan 10, 2024 at 06:14:25PM -0700, Daniel Xu wrote: > > > > are .BTF_ids records guaranteed to be sorted by address? so we are > > sure that the set will be followed by its records? > > I think so, unless I am misunderstanding something. The asm directives > are basically laying out the btf_id_set8 datastructures by hand. So if > the entries don't adhere to the exact layout then lots more will break > right? > > > > > I thought we'd need to find size for each set and then check each > > .BTF_ids record if it belongs to the set > > Not sure I'm following. How would the check be done differently than in > the patch? Ah I got it. Current code iterates through all symbols. Indeed a good question whether or not order is guaranteed. I think it should be possible to process each set8 more reliably. Lemme think on it.
