On Thu, Sep 19, 2024 at 5:08 AM Mykyta Yatsenko <mykyta.yatsenko5@xxxxxxxxx> wrote: > > On 19/09/2024 03:51, Philo Lu wrote: > > > > Hi Mykyta, > > > > On 2024/9/19 04:39, Mykyta Yatsenko wrote: > >> From: Mykyta Yatsenko <yatsenko@xxxxxxxx> > >> > >> Production BPF programs are increasing in number of instructions and > >> states > >> to the point, where optimising verification process for them is > >> necessary > >> to avoid running into instruction limit. Authors of those BPF programs > >> need to analyze verifier output, for example, collecting the most > >> frequent source code lines to understand which part of the program has > >> the biggest verification cost. > >> > >> This patch introduces `--top-src-lines` flag in veristat. > >> `--top-src-lines=N` makes veristat output N the most popular sorce code > >> lines, parsed from verification log. > >> > >> An example: > >> ``` > >> $ sudo ./veristat --log-size=1000000000 --top-src-lines=4 > >> pyperf600.bpf.o > >> Processing 'pyperf600.bpf.o'... > >> Top source lines (on_event): > >> 4697: (pyperf.h:0) > >> 2334: (pyperf.h:326) event->stack[i] = *symbol_id; > >> 2334: (pyperf.h:118) pidData->offsets.String_data); > >> 1176: (pyperf.h:92) bpf_probe_read_user(&frame->f_back, > >> ... > >> ``` > > > > I think this is useful and wonder how can I use it. In particular, is > > it possible to know the corresponding instruction number contributed > > by the source lines? > > > No, as far as I know, we don't have that info, so we just use number of > source lines as a proxy for number of instructions. Eduard suggested to > collect > instruction count per source line in verifier, maybe that actually what > we should do. Let's start simple and then build upon that. I'd rather finalize the minimal first version before going down to assembly instruction recording and counting. > > Assume a prog is rejected due to instruction limit. I can optimize the > > prog with `--top-src-lines`, but have to check the result with another > > "load" to see the total instruction number (because I don't know how > > many instructions reduced with the optimized src lines). > > > > Am I right? or is there any better method? > Yes, you are right. > > > > Thanks. > >
