Even though GDB is a traditional debugger, it provides support for dynamic fast user-space tracing. Tracing in simple terms is super fast data logging from a target application or the kernel. The data is usually a superset of what a user would normally want from debugging but cannot get because of the debugger overhead. The traditional debugging approach can indeed alter the correctness of the application output or alter its behavior. Thus, the need for tracing arises. GDB in fact is one of the first projects which tried to have an integrated approach of debugging and tracing using the same tool. It has been designed in a manner such that sufficient decoupling is maintained – allowing it to expand and be flexible. An example is the use of In-Process Agent (IPA) which is crucial to fast tracing in GDB but is not necessary for TRAP-based normal tracepoints.
GDB’s Tracing Infrastructure
The tracing is performed by
collect commands. The location where the user wants to collect some data is called a tracepoint. It is just a special type of breakpoint without support of running GDB commands on a tracepoint hit. As the program progresses, and passes the tracepoint, data (such as register values, memory values etc) gets collected based on certain conditions (if desired so). The data collection is done in a trace buffer when the tracepoint is hit. Later, that data can be examined from the collected trace snapshot using
tfind. However, tracing for now is restricted to remote targets (such as
gdbserver). Apart from this type of dynamic tracing, there is also support for static tracepoints in which instrumentation points known as markers are embedded in the target and can be activated or deactivated. The process of installing these static tracepoints is known as probing a marker. Considering that you have started GDB and your binary is loaded, a sample trace session may look something like this :
(gdb) trace foo (gdb) actions Enter actions for tracepoint #1, one per line. > collect $regs,$locals > while-stepping 9 > collect $regs > end > end (gdb) tstart [program executes/continues...] (gdb) tstop
This puts up a tracepoint at foo, collects all register values at tracepoint hit, and then for subsequent 9 instruction executions, collects all register values. We can now analyze the data using
(gdb) tfind 0 Found trace frame 0, tracepoint 1 54 bar = (bar & 0xFF000000) >> 24; (gdb) tdump Data collected at tracepoint 1, trace frame 0: rax 0x2000028 33554472 rbx 0x0 0 rcx 0x33402e76b0 220120118960 rdx 0x1 1 rsi 0x33405bca30 220123089456 rdi 0x2000028 33554472 rbp 0x7fffffffdca0 0x7fffffffdca0 rsp 0x7fffffffdca0 0x7fffffffdca0 . . rip 0x4006f1 0x4006f1 <foo+7> [and so on...] (gdb) tfind 4 Found trace frame 4, tracepoint 1 0x0000000000400700 55 r1 = (bar & 0x00F00000) >> 20; (gdb) p $rip $1 = (void (*)()) 0x400700 <foo+22>
So one can observe data collected from different trace frames in this manner and even output to a separate file if required. Going more in depth to know how tracing works, lets see the GDB’s two tracing mechanisms :
These type of tracepoints are the basic default tracepoints. The idea of their use is similar to breakpoints where GDB replaces the target instruction with a TRAP or any other exception causing instruction. On x86, this can usually be an
int 3 which has a special single byte instruction –
0xCC reserved for it. Replacing a target instruction with this 1 byte ensures that the normal instructions are not corrupted. So, during the execution of the process, the OS hits the
int 3 where it halts and program state is saved. The OS sends a
SIGTRAP signal to the process. As GDB is attached or is running the process, it receives a
SIGCHLD as a notification, that something happened with a child. It does a
wait(), which will tell it that process has received a
SIGTRAP. Thus the
SIGTRAP never reaches the process as GDB intercepts it. The original instruction is first restored, or executed out-of-line for non-stop multi-threaded debugging. GDB transfers the control to the trace collection which does the data collection part upon evaluating any condition set. The data is stored into a trace buffer. Then, the original instruction is replaced again with the tracepoint and normal execution continues. This all fine and good, but there is a catch – the TRAP mechanism alters the flow of the application and the control is passed to the OS which leads to some delay a compromise in speed. But even with that, because of a very restrictive conditional tracing design, and better interaction of interrupt-driven approaches with instruction caches, normal interrupt- based tracing in GDB is a robust technique. A faster solution would indeed be a pure user-space approach, where everything is done at the application level.
Owing to the limitations stated above, a fast tracepoint approach was developed. This special type of tracepoint uses a dynamic tracing approach. Instead of using the breakpoint approach, GDB uses a mix of IPA and remote target (
gdbserver) to replace the target instruction with a 5 byte jump to a special section in memory called a jump-pad. This jump-pad, also known as a trampoline, first pushes all registers to stack (saving the program state). Then, it calls the collector function for trace data collection, it executes the displaced instruction out-of-line, and jumps back to the original instruction stream. I will probably write something more about how trampolines work and some techniques used in dynamic instrumentation frameworks like Dyninst in a subsequent post later on.
Fast tracepoints can be used with command
ftrace, almost exactly like with the
trace command. A special command in the following format is sent to the In-Process Agent by
<tracepoint object> is the object containing bytecode for conditional tracing, address, type, action etc. and
<jump pad> is the 8-byte long head of the jump pad location in the memory. The IPA prepares all that and if all goes well, responds to such a query by,
<target address> is the address where the tracepoint is put in the inferior,
<jump_pad> is the updated address of the jump pad head and the
<fjump_size> are the jump instruction sequence and its size copied to the command buffer, sent back by IPA. The remote target (
gdbserver) then modifies the memory of the target process. Much more fine grained information about fast tracepoint installation is available in the GDB documentation. This is a very pure user-space approach to tracing. However, there is a catch – the target instruction to be replaced should be at least 5 bytes long for this to work as the jump is itself 5 byte long (on Intel x86). This means that fast tracepoints using GDB cannot be put everywhere. How code is modified when patching a 5 bytes instruction is a discussion of another time. This is probably the fastest way yet to perform dynamic tracing and is a good candidate for such work.
The GDB traces can be saved with tsave and with the –ctf switch may be exported to CTF also. I have not tried this but hopefully it should at least open the traces with TraceCompass for further analysis. The GDB’s fast tracepoint mechanism is quite fast I must say – but in terms of usability, LTTng is a far better and advanced option. GDB however allows dynamic insertion of tracepoints and the tracing features are well integrated in your friendly neighborhood debugger. Happy Tracing!