Hello all, I am learning how to add static trace points via trace_events. I am seeing the trace using 'cat /sys/kernel/debug/tracing/trace' . Naturally the tracing buffer size is limited. My question is, what is the recommended way of reading trace data? Ive looked at perf record/report, but so far, i have only managed to get aggregated statistics, or complete dumps. What i need is just a way to collect all the trace output to one file. Basically i am using the trace events as a souped up printk (since i need to collect large amounts of traces). Do i need to use LTTng/viewer or something for this task? Thanks!
On Sat, Aug 20, 2011 at 3:20 PM, Prateek Sharma <prateek3.14@gmail.com>wrote:
Hello all, I am learning how to add static trace points via trace_events. I am seeing the trace using 'cat /sys/kernel/debug/tracing/trace' . Naturally the tracing buffer size is limited. My question is, what is the recommended way of reading trace data? Ive looked at perf record/report, but so far, i have only managed to get aggregated statistics, or complete dumps. What i need is just a way to collect all the trace output to one file. Basically i am using the trace events as a souped up printk (since i need to collect large amounts of traces). Do i need to use LTTng/viewer or something for this task?
Thanks!
I've already found the solution to this, turns out there is trace_pipe as well. Sorry!
Hi On Sat, Aug 20, 2011 at 12:10 PM, Prateek Sharma <prateek3.14@gmail.com> wrote:
On Sat, Aug 20, 2011 at 3:20 PM, Prateek Sharma <prateek3.14@gmail.com> wrote:
Hello all, I am learning how to add static trace points via trace_events. I am seeing the trace using 'cat /sys/kernel/debug/tracing/trace' .
I am new and I try to learn trace. Can you give me how can use trace?
Naturally the tracing buffer size is limited. My question is, what is the recommended way of reading trace data? Ive looked at perf record/report, but so far, i have only managed to get aggregated statistics, or complete dumps. What i need is just a way to collect all the trace output to one file. Basically i am using the trace events as a souped up printk (since i need to collect large amounts of traces). Do i need to use LTTng/viewer or something for this task? Thanks!
I've already found the solution to this, turns out there is trace_pipe as well. Sorry! _______________________________________________ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
On 8/20/11, esmaeil mirzaee <esmaeil.debian@gmail.com> wrote:
Hi
On Sat, Aug 20, 2011 at 12:10 PM, Prateek Sharma <prateek3.14@gmail.com> wrote:
On Sat, Aug 20, 2011 at 3:20 PM, Prateek Sharma <prateek3.14@gmail.com> wrote:
Hello all, I am learning how to add static trace points via trace_events. I am seeing the trace using 'cat /sys/kernel/debug/tracing/trace' .
I am new and I try to learn trace. Can you give me how can use trace?
Naturally the tracing buffer size is limited. My question is, what is the recommended way of reading trace data? Ive looked at perf record/report, but so far, i have only managed to get aggregated statistics, or complete dumps. What i need is just a way to collect all the trace output to one file. Basically i am using the trace events as a souped up printk (since i need to collect large amounts of traces). Do i need to use LTTng/viewer or something for this task? Thanks!
I've already found the solution to this, turns out there is trace_pipe as well. Sorry! _______________________________________________ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
_______________________________________________ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
Hi Esmaeil! Some good starting points for you: [1] Documentation/ftrace.txt [2] http://lwn.net/Articles/322666/ [3] http://lwn.net/Articles/365835/ [4] http://lwn.net/Articles/366796/ [5] http://lwn.net/Articles/370423/
On 8/20/11, Prateek Sharma <prateek3.14@gmail.com> wrote:
Hello all, I am learning how to add static trace points via trace_events. I am seeing the trace using 'cat /sys/kernel/debug/tracing/trace' . Naturally the tracing buffer size is limited. My question is, what is the recommended way of reading trace data? Ive looked at perf record/report, but so far, i have only managed to get aggregated statistics, or complete dumps. What i need is just a way to collect all the trace output to one file. Basically i am using the trace events as a souped up printk (since i need to collect large amounts of traces). Do i need to use LTTng/viewer or something for this task?
Thanks!
Hi Prateek! Have a look at trace-cmd utility [1]. Hope you.find it useful [1] http://lwn.net/Articles/410200/
participants (3)
-
esmaeil mirzaee -
Prateek Sharma -
Vladimir Murzin