Re: [PATCH] uprobes: get rid of bogus trace_uprobe hit counter

On Wed, Aug 7, 2024 at 2:44 PM Masami Hiramatsu wrote: > > On Tue, 6 Aug 2024 10:26:25 -0700 > Andrii Nakryiko wrote: > > > On Tue, Aug 6, 2024 at 12:37 AM Jiri Olsa wrote: > > > > > > On Mon, Aug 05, 2024 at 01:28:03PM -0700, Andrii Nakryiko wrote: > > > > trace_uprobe->nhit counter is not incr

Re: [PATCH] uprobes: get rid of bogus trace_uprobe hit counter

On Tue, 6 Aug 2024 10:26:25 -0700 Andrii Nakryiko wrote: > On Tue, Aug 6, 2024 at 12:37 AM Jiri Olsa wrote: > > > > On Mon, Aug 05, 2024 at 01:28:03PM -0700, Andrii Nakryiko wrote: > > > trace_uprobe->nhit counter is not incremented atomically, so its value > > > is bogus in practice. On the oth

Re: [PATCH] uprobes: get rid of bogus trace_uprobe hit counter

On Tue, Aug 6, 2024 at 12:37 AM Jiri Olsa wrote: > > On Mon, Aug 05, 2024 at 01:28:03PM -0700, Andrii Nakryiko wrote: > > trace_uprobe->nhit counter is not incremented atomically, so its value > > is bogus in practice. On the other hand, it's actually a pretty big > > uprobe scalability problem du

Re: [PATCH] uprobes: get rid of bogus trace_uprobe hit counter

On Mon, Aug 05, 2024 at 01:28:03PM -0700, Andrii Nakryiko wrote: > trace_uprobe->nhit counter is not incremented atomically, so its value > is bogus in practice. On the other hand, it's actually a pretty big > uprobe scalability problem due to heavy cache line bouncing between CPUs > triggering the

[PATCH] uprobes: get rid of bogus trace_uprobe hit counter

trace_uprobe->nhit counter is not incremented atomically, so its value is bogus in practice. On the other hand, it's actually a pretty big uprobe scalability problem due to heavy cache line bouncing between CPUs triggering the same uprobe. Drop it and emit obviously unrealistic value in its stead