Open louisom opened 8 years ago
At this moment, f9-kernel Kprobe approach was using function address in kernel, we can saw it at kernel/sampling-kdb.c for the example.
Compare to Linux kernel Kprobe, it was using symbol name for user space program to insert the kprobe into kernel:
struct kprobe kp = { .symbol_name = "_do_fork", };
After reviewing, f9-kernel already provide the infrastructure ksym for this approach, we can take time and improve Kprobe for real dynamic tracing in user space.
ksym
Hello, Has any work been done in this area - perhaps on one of the forks? Thank you.
At this moment, f9-kernel Kprobe approach was using function address in kernel, we can saw it at kernel/sampling-kdb.c for the example.
Compare to Linux kernel Kprobe, it was using symbol name for user space program to insert the kprobe into kernel:
After reviewing, f9-kernel already provide the infrastructure
ksym
for this approach, we can take time and improve Kprobe for real dynamic tracing in user space.