On Sat, 23 Nov 2024, Mikołaj Kołek wrote: > findings. In the MMAP layout section of the page, you can find this > sentence: > Before Linux 2.6.39, there is a bug that means you must allocate > an mmap ring buffer when sampling even if you do not plan to > access it. > Unless I'm somehow misunderstanding it, this statement does not seem > to be well worded, or alternatively this bug does not seem to be > fixed. That text was probably written by me. I tried looking at the 2.6.39 code, my perf_tests, and also PAPI which was where the problem was probably noticed but I can't find a firm reference for how the issue was fixed. If I recall, the problem was if you were trying to create a sampling event without mmap (say you want to get a signal every 100,000 retired instructions, but you don't actually want any sample data). I think before 2.6.39 if you tried setting that up you'd get some sort of error (an EINVAL?) when trying to start(?) the event. It is possible this wasn't fixed. I tried to be pretty good about putting relevant git commits as comments in the manpage but there doesn't seem to be one for that part of the text. I'm guessing it was PeterZ doing the work on this so maybe he remembers. Vince Weaver vincent.weaver@xxxxxxxxx
