Re: [PATCH -tip v2 0/3] ftrace, kprobes: Introduce IPMODIFY flag for ftrace_ops to detect conflicts

[Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>]

(2014/06/17 20:04), Masami Hiramatsu wrote:
> Hi,
> 
> Here is the version 2 of the series of patches which introduces
> IPMODIFY flag for ftrace_ops to detect conflicts of ftrace users
> who can modify regs->ip in their handler.
> In this version, I fixed some bugs in previous version and
> added a patch which made kprobe itself free from IPMODIFY
> except for jprobe.
> 
> Currently, only kprobes can change the regs->ip in the handler,
> but recently kpatch is also want to change it. Moreover, since
> the ftrace itself exported to modules, it might be considerable
> senario.
> 
> Here we talked on github.
>  https://github.com/dynup/kpatch/issues/47
> 
> To protect modified regs-ip from each other, this series
> introduces FTRACE_OPS_FL_IPMODIFY flag and ftrace now ensures
> the flag can be set on each function entry location. If there
> is someone who already reserve regs->ip on target function
> entry, ftrace_set_filter_ip or register_ftrace_function will
> return -EBUSY. Users must handle that.
> 
> The 3rd patch adds a special reservation of IPMODIFY on the
> jprobed address, since it is the only user who will change
> the regs->ip. Other kprobes do not change it anymore. 
> 
> Testing:
> BTW, I've tested it with samples/kprobes/{k,j}probe_example.ko
> and a small test module which I added to the last of this mail.
> 
> Here is the test script. I think I should put this under
> tools/testing, maybe kprobes? or selftest/kprobes?
> 
> ----
> #!/bin/bash
> 
> echo "IPMODIFY test"
> ADDR=0x`grep do_fork /proc/kallsyms | cut -f 1 -d" "`
> 
> echo "Case1: kprobe->jprobe->ipmodify(fail)"
> modprobe kprobe_example
> modprobe jprobe_example
> insmod ./ipmodify.ko && echo "Case1: [FAIL]" || echo "Case1: [OK]"

Oops, I missed the target=$ADDR for these insmod...

Anyway, without passing target=$ADDR option, ipmodify.ko tries to
register ftrace_ops with an empty filter hash which means
trace all functions. So anyway jprobe and that ftrace_ops collide
with each other on the do_fork and the 2nd one should be failed.

Thus, of course this test doesn't fail with or without target=
param.

Thank you,


-- 
Masami HIRAMATSU
IT Management Research Dept. Linux Technology Center
Hitachi, Ltd., Yokohama Research Laboratory
E-mail: masami.hiramatsu.pt@hitachi.com