linux-kernel - Re: [PATCH 3/3] tracing/probes: Fix return value when "(fault)" is injected

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 7 Jul 2023 11:51:28 +0900
From:   Masami Hiramatsu (Google) <mhiramat@...nel.org>
To:     Steven Rostedt <rostedt@...dmis.org>
Cc:     Dan Carpenter <dan.carpenter@...aro.org>,
        linux-trace-kernel@...r.kernel.org,
        LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 3/3] tracing/probes: Fix return value when "(fault)" is
 injected

On Thu, 6 Jul 2023 22:20:20 -0400
Steven Rostedt <rostedt@...dmis.org> wrote:

> On Fri, 7 Jul 2023 11:02:10 +0900
> Masami Hiramatsu (Google) <mhiramat@...nel.org> wrote:
> 
> > /* Print type function for string type */
> > int PRINT_TYPE_FUNC_NAME(string)(struct trace_seq *s, void *data, void *ent)
> > {
> >         int len = *(u32 *)data >> 16;
> > 
> >         if (!len)
> >                 trace_seq_puts(s, "(fault)");
> >         else
> > ----
> > 
> > In this case, what we need is to set data_loc length = 0 if ret < 0.
> > 
> > Do you really need to get '"(fault)"' (with double quotation) or
> > just '(fault)' (no double quotation) is OK?
> 
>  ># echo 'e:myopen syscalls/sys_enter_openat file=+0($filename):ustring' >> /sys/kernel/tracing/dynamic_events
>  ># trace-cmd start -e myopen
>  ># trace-cmd show
> # tracer: nop
> #
> # entries-in-buffer/entries-written: 19/19   #P:4
> #
> #                                _-----=> irqs-off/BH-disabled
> #                               / _----=> need-resched
> #                              | / _---=> hardirq/softirq
> #                              || / _--=> preempt-depth
> #                              ||| / _-=> migrate-disable
> #                              |||| /     delay
> #           TASK-PID     CPU#  |||||  TIMESTAMP  FUNCTION
> #              | |         |   |||||     |         |
>        trace-cmd-4688    [000] ...1. 466968.015784: myopen: (syscalls.sys_enter_openat) file=(fault)
>        trace-cmd-4688    [000] ...1. 466968.015816: myopen: (syscalls.sys_enter_openat) file=(fault)
>        trace-cmd-4688    [000] ...1. 466968.015833: myopen: (syscalls.sys_enter_openat) file=(fault)
>        trace-cmd-4688    [000] ...1. 466968.015849: myopen: (syscalls.sys_enter_openat) file=(fault)
>        trace-cmd-4688    [000] ...1. 466968.015864: myopen: (syscalls.sys_enter_openat) file=(fault)
>        trace-cmd-4688    [000] ...1. 466968.015879: myopen: (syscalls.sys_enter_openat) file=(fault)
> 
> 
> Does that answer your question? ;-)

Ah, I meant that your commit 2e9906f84fc7 ("tracing: Add "(fault)" name injection
 to kernel probes") tries to make it '"(fault)"', So it makes 

       trace-cmd-4688    [000] ...1. 466968.015879: myopen: (syscalls.sys_enter_openat) file="(fault)"

Keeping it current '(fault)' makes easy to identify which one is failed to fetch,
but it may require user to parse both "some string" and (fault). I thought that
was the reason why you added that commit.

Thank you,

> 
> -- Steve


-- 
Masami Hiramatsu (Google) <mhiramat@...nel.org>