From: Masami Hiramatsu (Google) <mhiramat@kernel.org>
To: Tom Zanussi <zanussi@kernel.org>
Cc: rostedt@goodmis.org, mhiramat@kernel.org,
linux-kernel@vger.kernel.org, linux-trace-kernel@vger.kernel.org,
akpm@linux-foundation.org, zwisler@google.com,
chinglinyu@google.com
Subject: Re: [PATCH 1/4] tracing/histogram: Don't use strlen to find length of stacktrace variables
Date: Tue, 14 Feb 2023 00:24:18 +0900 [thread overview]
Message-ID: <20230214002418.0103b9e765d3e5c374d2aa7d@kernel.org> (raw)
In-Reply-To: <1ed6906cd9d6477ef2bd8e63c61de20a9ffe64d7.1676063532.git.zanussi@kernel.org>
On Fri, 10 Feb 2023 15:33:03 -0600
Tom Zanussi <zanussi@kernel.org> wrote:
> Because stacktraces are saved in dynamic strings,
> trace_event_raw_event_synth() uses strlen to determine the length of
> the stack. Stacktraces may contain 0-bytes, though, in the saved
> addresses, so the length found and passed to reserve() will be too
> small.
Good catch!
>
> Fix this by using the first unsigned long in the stack variables to
> store the actual number of elements in the stack and have
> trace_event_raw_event_synth() use that to determine the length of the
> stack.
>
> Signed-off-by: Tom Zanussi <zanussi@kernel.org>
> ---
> kernel/trace/trace_events_hist.c | 12 ++++++++----
> kernel/trace/trace_events_synth.c | 7 ++++++-
> 2 files changed, 14 insertions(+), 5 deletions(-)
>
> diff --git a/kernel/trace/trace_events_hist.c b/kernel/trace/trace_events_hist.c
> index 888b7a394ce5..76bd105988c6 100644
> --- a/kernel/trace/trace_events_hist.c
> +++ b/kernel/trace/trace_events_hist.c
> @@ -3135,13 +3135,15 @@ static inline void __update_field_vars(struct tracing_map_elt *elt,
> size = min(val->size, STR_VAR_LEN_MAX);
> strscpy(str, val_str, size);
> } else {
> + char *stack_start = str + sizeof(unsigned long);
> int e;
>
> - e = stack_trace_save((void *)str,
> + e = stack_trace_save((void *)stack_start,
> HIST_STACKTRACE_DEPTH,
> HIST_STACKTRACE_SKIP);
BTW, the size of "str" is enough to store HIST_STACKTRACE_DEPTH?
In string case,
size = min(val->size, STR_VAR_LEN_MAX);
will limit the max size.
Thank you,
> if (e < HIST_STACKTRACE_DEPTH - 1)
> - ((unsigned long *)str)[e] = 0;
> + ((unsigned long *)stack_start)[e] = 0;
> + *((unsigned long *)str) = e;
> }
> var_val = (u64)(uintptr_t)str;
> }
> @@ -5133,13 +5135,15 @@ static void hist_trigger_elt_update(struct hist_trigger_data *hist_data,
> size = min(hist_field->size, STR_VAR_LEN_MAX);
> strscpy(str, val_str, size);
> } else {
> + char *stack_start = str + sizeof(unsigned long);
> int e;
>
> - e = stack_trace_save((void *)str,
> + e = stack_trace_save((void *)stack_start,
> HIST_STACKTRACE_DEPTH,
> HIST_STACKTRACE_SKIP);
> if (e < HIST_STACKTRACE_DEPTH - 1)
> - ((unsigned long *)str)[e] = 0;
> + ((unsigned long *)stack_start)[e] = 0;
> + *((unsigned long *)str) = e;
> }
> hist_val = (u64)(uintptr_t)str;
> }
> diff --git a/kernel/trace/trace_events_synth.c b/kernel/trace/trace_events_synth.c
> index d458d7a0dfd7..6209b23c863f 100644
> --- a/kernel/trace/trace_events_synth.c
> +++ b/kernel/trace/trace_events_synth.c
> @@ -538,7 +538,12 @@ static notrace void trace_event_raw_event_synth(void *__data,
> val_idx = var_ref_idx[field_pos];
> str_val = (char *)(long)var_ref_vals[val_idx];
>
> - len = kern_fetch_store_strlen((unsigned long)str_val);
> + if (event->dynamic_fields[i]->is_stack) {
> + len = *((unsigned long *)str_val);
> + len *= sizeof(unsigned long);
> + } else {
> + len = kern_fetch_store_strlen((unsigned long)str_val);
> + }
>
> fields_size += len;
> }
> --
> 2.34.1
>
--
Masami Hiramatsu (Google) <mhiramat@kernel.org>
next prev parent reply other threads:[~2023-02-13 15:24 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-02-10 21:33 [PATCH 0/4] tracing/histogram: Some fixes for new stacktrace variables Tom Zanussi
2023-02-10 21:33 ` [PATCH 1/4] tracing/histogram: Don't use strlen to find length of " Tom Zanussi
2023-02-13 15:24 ` Masami Hiramatsu [this message]
2023-02-13 15:56 ` Steven Rostedt
2023-02-10 21:33 ` [PATCH 2/4] tracing/histogram: Fix a few problems with stacktrace variable printing Tom Zanussi
2023-02-10 21:33 ` [PATCH 3/4] tracing/histogram: Fix stacktrace key Tom Zanussi
2023-02-10 23:07 ` kernel test robot
2023-02-10 21:33 ` [PATCH 4/4] tracing/histogram: Fix stacktrace histogram Documententation Tom Zanussi
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20230214002418.0103b9e765d3e5c374d2aa7d@kernel.org \
--to=mhiramat@kernel.org \
--cc=akpm@linux-foundation.org \
--cc=chinglinyu@google.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-trace-kernel@vger.kernel.org \
--cc=rostedt@goodmis.org \
--cc=zanussi@kernel.org \
--cc=zwisler@google.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).