- 22 Dec, 2016 1 commit
-
-
4ast authored
add cpuunclaimed
-
- 21 Dec, 2016 6 commits
-
-
Brendan Gregg authored
-
Brendan Gregg authored
-
4ast authored
Add snapcraft script to package up bcc as a snap
-
4ast authored
Don't close extern table when a module destructs
-
Huapeng Zhou authored
-
4ast authored
add bpf_obj_pin/bpf_obj_get to pin/get bpf objects
-
- 20 Dec, 2016 4 commits
-
-
Huapeng Zhou authored
-
Huapeng Zhou authored
-
Huapeng Zhou authored
-
Brenden Blanco authored
support macro in call arguments
-
- 19 Dec, 2016 29 commits
-
-
shodoco authored
-
4ast authored
trace: -L, -t, -T switches
-
4ast authored
USDT tools for higher level languages
-
Sasha Goldshtein authored
The `-t` switch for specifying a thread id has been renamed to `-L`. The `-T` switch specifies that a time column should be provided, whereas the `-t` switch requests a relative timestamp (from the beginning of the trace). This is more consistent with other tools, and lets the user choose the desired format for the time column, if any.
-
Sasha Goldshtein authored
-
Sasha Goldshtein authored
When C allocations are traced, attach the `malloc` uprobe only to the requested process -- otherwise, we get allocation information from the entire system.
-
Sasha Goldshtein authored
-
Sasha Goldshtein authored
-
Sasha Goldshtein authored
-
Sasha Goldshtein authored
-
Sasha Goldshtein authored
-
Sasha Goldshtein authored
-
Sasha Goldshtein authored
-
Sasha Goldshtein authored
If the process in question doesn't have the USDT probes we expect, handle the failure gracefully. After all, it could be a binary compiled without USDT support, the pid could be recycled to some other application, or -- the process name could match, but by accident, and really belong to some other runtime.
-
Sasha Goldshtein authored
The /proc/PID/cmdline file has \0 for spaces in the command line, and there may be trailing \0 characters as well. Replace them all with spaces.
-
Sasha Goldshtein authored
This tool traces method executions and prints out messages with a nesting depth indicator that allows relatively easy visualization of program flow. To reduce overhead, `-C` and `-M` switches take class and method prefixes to filter the output by. The filtering takes place in the BPF program, so the unnecessary data never ends up in user space. Usage examples: ``` uflow java 185 uflow -M index python 180 uflow -C 'IRB::Irb' ruby 10041 ```
-
Sasha Goldshtein authored
-
Sasha Goldshtein authored
-
Sasha Goldshtein authored
-
Sasha Goldshtein authored
This tool gathers statistics (by pid) on method calls, class loads, garbage collections, thread creations, object allocations, and exceptions raised and prints them in a top-like style. Supported languages currently include Java, Python, Ruby, and Node. Not all languages provide the necessary probes for all event types. Usage examples: ``` ustat -C ustat -l java 1 ustat -l ruby -S objnew ```
-
Sasha Goldshtein authored
Allow callers to obtain the list of active (enabled) USDT probes from the USDT object, if they need to enable the uprobes by hand without using the bcc attach support. This is required when the same uprobe needs to be enabled in more than one process.
-
Sasha Goldshtein authored
-
Sasha Goldshtein authored
Use various USDT probes in Java and Ruby, and `malloc()` uprobe in C, to summarize object allocations by type or by size. Print statistics at the end or at intervals, with optional trimming of top types by number of instances allocated or number of bytes.
-
Sasha Goldshtein authored
-
Sasha Goldshtein authored
-
Sasha Goldshtein authored
Trace `pthread_start` calls and get the symbol for the newly-started thread for printing. Also allow no-Java runs with no language specified for pthread-only tracing.
-
Sasha Goldshtein authored
`uthreads` attaches to thread start and stop USDT probes and traces out thread creation and destruction events. Currently, only Java is supported.
-
Sasha Goldshtein authored
Add `-S` switch for tracing syscalls with or without high- level language method calls. When `-S` is provided, attach to all `sys_` and `SyS_` entry points in the kernel and trace the number of calls and execution latency (if `-L` was also specified). This is now also a useful general-purpose tool for syscall count/latency summarizing when a language is not specified.
-
Sasha Goldshtein authored
Add Node support for gc__start and gc__done probes, which includes GC type flag parsing.
-