linux-trace-devel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Steven Rostedt <rostedt@goodmis.org>
To: "Tzvetomir Stoyanov (VMware)" <tz.stoyanov@gmail.com>
Cc: linux-trace-devel@vger.kernel.org
Subject: Re: [PATCH 1/2] trace-cmd library: Use the real trace buffer page size
Date: Thu, 11 Nov 2021 10:44:58 -0500	[thread overview]
Message-ID: <20211111104458.5ee5449d@gandalf.local.home> (raw)
In-Reply-To: <20211111151326.87227-2-tz.stoyanov@gmail.com>

On Thu, 11 Nov 2021 17:13:25 +0200
"Tzvetomir Stoyanov (VMware)" <tz.stoyanov@gmail.com> wrote:

> When new output handler is created, it assumes that the trace buffer page
> size is equal to the system memory page size. This assumption is valid
> for the current ftrace implementation, but it may change in the future.
> The newly introduced traceevent library API should be used to get the
> real trace buffer page size, bases on the information from the
> "events/header_page" ftrace file.
> 
> This commit depends on:
>  [PATCH] libtraceevent: A new API for trace page size
>  https://lore.kernel.org/linux-trace-devel/20211001062338.2389024-1-tz.stoyanov@gmail.com/
> 
> Signed-off-by: Tzvetomir Stoyanov (VMware) <tz.stoyanov@gmail.com>
> ---
>  lib/trace-cmd/trace-output.c | 52 +++++++++++++++++++++++++++++++++++-
>  1 file changed, 51 insertions(+), 1 deletion(-)
> 
> diff --git a/lib/trace-cmd/trace-output.c b/lib/trace-cmd/trace-output.c
> index a029678b..08fa485d 100644
> --- a/lib/trace-cmd/trace-output.c
> +++ b/lib/trace-cmd/trace-output.c
> @@ -1189,6 +1189,56 @@ static int write_compression_header(struct tracecmd_output *handle)
>  	return 0;
>  }
>  
> +static int get_trace_page_size(struct tracecmd_output *handle)
> +{
> +	struct tep_handle *tep = NULL;
> +	tsize_t psize, size;
> +	char *path = NULL;
> +	char *buff = NULL;
> +	int fd = -1;
> +	int r, s;
> +
> +	/* In case of an error, return user space page size */
> +	psize = getpagesize();
> +
> +	path = get_tracing_file(handle, "events/header_page");
> +	if (!path)
> +		goto out;
> +	fd = open(path, O_RDONLY);
> +	if (fd < 0) /* old style did not show this info */

Note, as I stated in the other patch, calling tep_parse_trace_page_size()
with size zero should get the updates needed for the old format as well.

> +		goto out;
> +
> +	tep = tep_alloc();
> +	if (!tep)
> +		goto out;
> +	size = get_size_fd(fd);
> +	if (!size)
> +		goto out;
> +	buff = malloc(size);
> +	if (!buff)
> +		goto out;
> +
> +	s = size;
> +	do {
> +		r = read(fd, buff, s);
> +		if (r > 0)
> +			s -= r;
> +	} while (r > 0 && s);
> +	if (s)
> +		goto out;

I know that a lot of his is copied from other parts of this file, but this
was written before libtracefs existed. We should start using libtracefs
APIs for new code, and start also moving the old code to it. The above
could be replaced with:

	buf = tracefs_instance_file_read(NULL, "events/header_page", &size);

-- Steve


> +	if (tep_parse_header_page(tep, buff, size, sizeof(long long)))
> +		goto out;
> +	psize = tep_get_trace_page_size(tep);
> +out:
> +	tep_free(tep);
> +	free(buff);
> +	if (path)
> +		put_tracing_file(path);
> +	if (fd >= 0)
> +		close(fd);
> +	return psize;
> +}
> +
>  /**
>   * tracecmd_output_allocate - allocate new output handler to a trace file
>   * @handle: file descriptor to an empty file, it can be -1 if the handler
> @@ -1213,7 +1263,7 @@ struct tracecmd_output *tracecmd_output_allocate(int fd)
>  
>  	handle->file_version = FILE_VERSION_DEFAULT;
>  
> -	handle->page_size = getpagesize();
> +	handle->page_size = get_trace_page_size(handle);
>  	handle->big_endian = tracecmd_host_bigendian();
>  
>  	list_head_init(&handle->options);


  reply	other threads:[~2021-11-11 15:45 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-11-11 15:13 [PATCH 0/2] Get trace buffer page size from kernel Tzvetomir Stoyanov (VMware)
2021-11-11 15:13 ` [PATCH 1/2] trace-cmd library: Use the real trace buffer page size Tzvetomir Stoyanov (VMware)
2021-11-11 15:44   ` Steven Rostedt [this message]
2021-11-11 15:13 ` [PATCH 2/2] trace-cmd library: Introduce buffer page size per instance Tzvetomir Stoyanov (VMware)

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=20211111104458.5ee5449d@gandalf.local.home \
    --to=rostedt@goodmis.org \
    --cc=linux-trace-devel@vger.kernel.org \
    --cc=tz.stoyanov@gmail.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).