linux-kernel - Re: [RFC PATCH 2/2] trace: allocate temparary buffer for trace output usage

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:   Thu, 15 Dec 2022 09:53:27 +0800
From:   Linyu Yuan <quic_linyyuan@...cinc.com>
To:     Steven Rostedt <rostedt@...dmis.org>
CC:     Masami Hiramatsu <mhiramat@...nel.org>,
        <linux-kernel@...r.kernel.org>, <bpf@...r.kernel.org>
Subject: Re: [RFC PATCH 2/2] trace: allocate temparary buffer for trace output
 usage

is it possible add a generic macro like #define __get_buf(offset, len)  
(p->buffer + (offset))  ?

it only use buffer filed of struct trace_seq .

On 12/15/2022 9:22 AM, Linyu Yuan wrote:
> thanks for remind, will check and use it.
>
>
> thanks
>
> On 12/14/2022 10:25 PM, Steven Rostedt wrote:
>> On Wed, 14 Dec 2022 22:11:42 +0800
>> Linyu Yuan <quic_linyyuan@...cinc.com> wrote:
>>
>>> there is one dwc3 trace event declare as below,
>>> DECLARE_EVENT_CLASS(dwc3_log_event,
>>>     TP_PROTO(u32 event, struct dwc3 *dwc),
>>>     TP_ARGS(event, dwc),
>>>     TP_STRUCT__entry(
>>>         __field(u32, event)
>>>         __field(u32, ep0state)
>>>         __dynamic_array(char, str, DWC3_MSG_MAX)
>>>     ),
>>>     TP_fast_assign(
>>>         __entry->event = event;
>>>         __entry->ep0state = dwc->ep0state;
>>>     ),
>>>     TP_printk("event (%08x): %s", __entry->event,
>>>             dwc3_decode_event(__get_str(str), DWC3_MSG_MAX,
>>>                 __entry->event, __entry->ep0state))
>>> );
>>> the problem is when trace function called, it will allocate up to
>>> DWC3_MSG_MAX bytes from trace event buffer, but never fill the buffer
>>> during fast assignment, it only fill the buffer when output function 
>>> are
>>> called, so this means if output function are not called, the buffer 
>>> will
>>> never used.
>>>
>>> add __alloc_buf() and __get_buf() which will not allocate event buffer
>>> when trace function called, but when trace output function called, 
>>> it will
>>> kmalloc buffer with size DWC3_MSG_MAX for temprary usage and free it
>>> before trace output function return.
>> This looks exactly like what the trace_seq *p is to be used for.
>>
>> static notrace enum print_line_t                    \
>> trace_raw_output_##call(struct trace_iterator *iter, int flags,        \
>>             struct trace_event *trace_event)        \
>> {                                    \
>>     struct trace_seq *s = &iter->seq;                \
>>     struct trace_seq __maybe_unused *p = &iter->tmp_seq;        \
>>                                          ^^^^^^^^^^^^^^^^^^^^
>>
>>     struct trace_event_raw_##call *field;                \
>>     int ret;                            \
>>                                     \
>>     field = (typeof(field))iter->ent;                \
>>                                     \
>>     ret = trace_raw_output_prep(iter, trace_event);            \
>>     if (ret != TRACE_TYPE_HANDLED)                    \
>>         return ret;                        \
>>                                     \
>>     trace_event_printf(iter, print);                \
>>                                     \
>>     return trace_handle_return(s);                    \
>> }                                    \
>>
>> That is a trace_seq buffer that is for temporary usage during the 
>> output.
>>
>> See:
>>    include/trace/events/libata.h
>>    include/trace/events/scsi.h
>>
>> As well as the macros trace_print_bitmask_seq(), 
>> trace_print_flags_seq(),
>> trace_print_symbols_seq(), etc.
>>
>> -- Steve
>>