On Tue, Mar 17, 2020 at 11:34 AM Steven Rostedt <rostedt@xxxxxxxxxxx> wrote: > > When the ring buffer was first written for ftrace, there was two > human readable files to read it. One was a standard "producer/consumer" > file (trace_pipe), which would consume data from the ring buffer as > it read it, and the other was a "static iterator" that would not > consume the events, such that the file could be read multiple times > and return the same output each time. > > The "static iterator" was never meant to be read while there was an > active writer to the ring buffer. If writing was enabled, then it > would disable the writer when the trace file was opened. > > There has been some complaints about this by the BPF folks, that did > not realize this little bit of information and it was requested that > the "trace" file does not stop the writing to the ring buffer. > > This patch series attempts to satisfy that request, by creating a > temporary buffer in each of the per cpu iterators to place the > read event into, such that it can be passed to users without worrying > about a writer to corrupt the event while it was being written out. > It also uses the fact that the ring buffer is broken up into pages, > where each page has its own timestamp that gets updated when a > writer crosses over to it. By copying it to the temp buffer, and > doing a "before and after" test of the time stamp with memory barriers, > can allow the events to be saved. Awesome. Thank you so much for working on it. Looks like it addresses all the issues bpf folks reported. cc-ing bpf list for visibility.
