lkml.org 
[lkml]   [2009]   [Jan]   [28]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
SubjectRe: [PATCH] tracer for sys_open() - sreadahead
From
2009/1/28 Ananth N Mavinakayanahalli <ananth@in.ibm.com>:
> On Tue, Jan 27, 2009 at 11:43:05PM +0100, Frederic Weisbecker wrote:
>> On Tue, Jan 27, 2009 at 12:08:04PM -0800, Kok, Auke wrote:
>> >
>> > This tracer monitors regular file open() syscalls. This is a fast
>> > and low-overhead alternative to strace, and does not allow or
>> > require to be attached to every process.
>> >
>> > The tracer only logs succesfull calls, as those are the only ones we
>> > are currently interested in, and we can determine the absolute path
>> > of these files as we log.
>> >
>> > Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
>>
>>
>> Hi Auke,
>>
>> Speaking about a global syscall tracer, I made a patch to trace only the syscalls
>> with the function-graph-tracer.
>>
>> http://lkml.org/lkml/2008/12/30/267
>>
>> Its approach and purpose is different than a tracer dedicated only to syscalls.
>> The function graph tracer traces execution graph of the functions and is more about
>> execution time spent and code flow whereas a syscall tracer can provide more specific
>> informations about syscalls.
>>
>> So both are not overlaping.
>>
>> But the low level part of my patch creates a thread flag _TIF_SYSCALL_TRACE which triggers
>> a ptrace hook when set.
>> This low-level part can easily be used by all tracers that would like to inspect syscalls.
>>
>> Just a change is needed: Steven requested that the part inside syscall_trace_enter become
>> a tracepoint, making it totally shareable between tracers and easy to turn on and off.
>>
>> And perhaps the parts that set/clear the flag on all tasks can be shared too.
>>
>> So we can start with this low-level syscall tracing facility. If you want, I can adapt
>> this low-level part and submit a patch this week or the next one to give you this base
>> infrastructure.
>>
>>
>> Once we have it, I think a syscall tracer can be fed with new syscalls events through
>> several patch iterations, starting with the open and close one :-)
>
> Here is something I did sometime ago that uses utrace. It is per task
> doesn't use ftrace and is just intended as a prototype. It traces both
> syscalls and returns.
>
> ---
>
> Here is the beginnings of a simple utrace based strace. Right now, one
> needs to invoke this program with a 'insmod <modname> tid=<tid>'.
>
> The output looks something like this:
> [267352.641112] Attached to 32604 => 0xd8f60090
> [267353.981046] 2
> [267353.981085] 197 0x1 0xbff8cd84 0x86cff4 0x86d4c0 0x86d4c0 0xbff8cd50a = 0
> [267353.981097] 192 0x0 0x1000 0x3 0x22 0xffffffff 0x0 = b7f8d000
> [267353.981124] 4 0x1 0xb7f8d000 0xd 0xd 0xb7f8d000 0xbff8cda8 = d
> [267353.981174] 4 0x1 0xb7f8d000 0x3 0x3 0xb7f8d000 0xbff8c7dc = 3
> [267353.981209] 252 0x0 0x0 0x86e0d0 0x0 0x86b274 0xbff8cee8 =
> [267353.981215] Task 32604 exited
> [267355.460180] Cannot find PID 32604
>
> I know strace does a pretty print, but this is a quick and dirty
> prototype.
> ---
>
> #include <linux/module.h>
> #include <linux/utrace.h>
> #include <linux/err.h>
> #include <asm/syscall.h>
>
> MODULE_DESCRIPTION("syscall trace");
> MODULE_LICENSE("GPL");
>
> static int target_tid;
>
> module_param_named(tid, target_tid, int, 0);
>
> static u32 task_syscall_entry(u32 action, struct utrace_attached_engine *engine,
> struct task_struct *task, struct pt_regs *regs)
> {
> long callno;
> unsigned long args[6];
>
> callno = syscall_get_nr(task, regs);
> syscall_get_arguments(task, regs, 0, 6, args);
>
> printk(KERN_INFO "%ld 0x%lx 0x%lx 0x%lx 0x%lx 0x%lx 0x%lx = ",
> callno, args[0], args[1], args[2], args[3], args[4], args[5]);
> return UTRACE_RESUME;
> }
>
> static u32 task_syscall_exit(u32 action, struct utrace_attached_engine *engine,
> struct task_struct *task, struct pt_regs *regs)
> {
> long retval = syscall_get_return_value(task, regs);
> printk("%lx\n", retval);
> return UTRACE_RESUME;
> }
>
> static u32 task_exit(enum utrace_resume_action action,
> struct utrace_attached_engine *engine, struct task_struct *task,
> long orig_code, long *code)
> {
> printk("\n");
> printk(KERN_INFO "Task %d exited\n", target_tid);
> return UTRACE_DETACH;
> }
>
> static const struct utrace_engine_ops syscall_ops =
> {
> .report_syscall_entry = task_syscall_entry,
> .report_syscall_exit = task_syscall_exit,
> .report_exit = task_exit,
> };
>
> static int __init strace_init(void)
> {
> struct pid *pid;
> int ret = 0;
> struct task_struct *target;
> struct utrace_attached_engine *engine;
>
> pid = find_get_pid(target_tid);
> if (pid == NULL) {
> printk(KERN_ERR "Cannot find PID %d\n", target_tid);
> ret = -ESRCH;
> goto out;
> }
>
> engine = utrace_attach_pid(pid, UTRACE_ATTACH_CREATE, &syscall_ops, 0);
> if (IS_ERR(engine)) {
> printk(KERN_ERR "utrace_attach_pid: %ld\n", PTR_ERR(engine));
> ret = -EINVAL;
> goto out;
> } else if (engine == NULL) {
> printk(KERN_ERR "utrace_attach_pid => NULL\n");
> ret = -EINVAL;
> goto out;
> } else
> printk(KERN_INFO "Attached to %d => 0x%p\n",
> target_tid, engine);
>
> /*
> * If utrace_attach_pid() succeeded above, we are sure the target
> * is valid here
> */
> target = pid_task(pid, PIDTYPE_PID);
> put_pid(pid);
>
> ret = utrace_set_events(target, engine, UTRACE_EVENT_SYSCALL |
> UTRACE_EVENT(EXIT));
> if (ret)
> printk(KERN_ERR "utrace_set_events returned %d\n", ret);
>
> out:
> return ret;
> }
>
> static void __exit strace_exit(void)
> {
> int ret = 0;
> struct pid *pid;
> struct utrace_attached_engine *engine;
> struct task_struct *target;
>
> pid = find_get_pid(target_tid);
> if (pid == NULL) {
> printk(KERN_ERR "Cannot find PID %d\n", target_tid);
> return;
> }
>
> target = pid_task(pid, PIDTYPE_PID);
> put_pid(pid);
> engine = utrace_attach_task(target, UTRACE_ATTACH_MATCH_OPS,
> &syscall_ops, 0);
> if (IS_ERR(engine))
> printk(KERN_ERR "Can't find self: %ld\n", PTR_ERR(engine));
> else if (engine == NULL)
> printk(KERN_ERR "Can't find self: no match\n");
> else {
> printk(KERN_INFO "Trying detach 0x%p from %d\n",
> engine, target_tid);
> ret = utrace_control(target, engine, UTRACE_DETACH);
> if (ret)
> printk(KERN_ERR "utrace_control returned %d\n",
> ret);
> }
> }
>
> module_init(strace_init);
> module_exit(strace_exit);
>


This is a very convenient Api. I would be glad to use it with ftrace.
But, what is it's status currently? Some projects to merge it?


\
 
 \ /
  Last update: 2009-01-28 15:23    [W:0.142 / U:2.980 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site