public inbox for passt-dev@passt.top
 help / color / mirror / code / Atom feed
From: Stefano Brivio <sbrivio@redhat.com>
To: Anshu Kumari <anskuma@redhat.com>
Cc: passt-dev@passt.top, lvivier@redhat.com, dgibson@redhat.com
Subject: Re: [PATCH v3] log: Add rate-limiting macros for log messages
Date: Thu, 26 Mar 2026 01:16:01 +0100 (CET)	[thread overview]
Message-ID: <20260326011601.38b9b1f4@elisabeth> (raw)
In-Reply-To: <20260325131120.299924-2-anskuma@redhat.com>

Thanks, this looks almost correct to me. Still two comments from my
side:

On Wed, 25 Mar 2026 18:41:21 +0530
Anshu Kumari <anskuma@redhat.com> wrote:

> Currently, some log messages that would be useful at info or warn level
> are kept at debug level because there is no way to throttle them, and a
> guest could otherwise flood the host logs.
> 
> Add a logmsg_ratelimit() macro that uses per-call-site static variables
> to independently track each call site's rate.  It allows up to
> LOG_RATELIMIT_BURST (5) messages per LOG_RATELIMIT_INTERVAL (1 second)
> window, then prints a suppression notice.  When a new window opens and
> messages were suppressed, the count is reported after the next allowed
> message.
> 

...there should be the Link: tag here, before Signed-off-by:, that
Laurent and myself suggested.

> Signed-off-by: Anshu Kumari <anskuma@redhat.com>
> ---
> v3:
>    - Print suppressed count after the message, not before
>    - Add suppression notice when burst limit is hit
>    - Reverse Christmas tree variable ordering
>    - Fix tab/space alignment in macro
>    - swapped LOG_RATELIMIT_INTERVAL and LOG_RATELIMIT_BURST location
> 
> v2:
>    - Use _ suffix for macro variables instead of prefix
>    - Remove intv parameter from convenience wrappers
> ---
>  log.h | 45 +++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 45 insertions(+)
> 
> diff --git a/log.h b/log.h
> index 6ceb686..8e6d65e 100644
> --- a/log.h
> +++ b/log.h
> @@ -48,6 +48,51 @@ void logmsg_perror(int pri, const char *format, ...)
>  		passt_exit(EXIT_FAILURE);				\
>  	} while (0)
>  
> +#define LOG_RATELIMIT_INTERVAL	1	/* Default rate limit window in seconds */
> +#define LOG_RATELIMIT_BURST	5	/* Max messages per window per call site */
> +
> +/**
> + * logmsg_ratelimit() - Log a message with rate limiting
> + * @fn:		Logging function name (e.g. warn, info, debug)
> + * @now:	Current timestamp
> + */
> +#define logmsg_ratelimit(fn, now, ...)					\
> +	do {								\
> +		static unsigned int rl_suppressed_;			\
> +		static unsigned int rl_printed_;			\
> +		static time_t rl_last_;					\
> +									\
> +		if ((now)->tv_sec - rl_last_ > LOG_RATELIMIT_INTERVAL) {\
> +			rl_last_ = (now)->tv_sec;			\
> +			rl_printed_ = 0;				\
> +		}							\
> +									\
> +		if (rl_printed_ < LOG_RATELIMIT_BURST) {		\
> +			fn(__VA_ARGS__);				\
> +			if (rl_suppressed_) {				\
> +				fn("(suppressed %u similar messages)",	\
> +				   rl_suppressed_);			\
> +				rl_suppressed_ = 0;			\
> +			}						\
> +			rl_printed_++;					\
> +		} else if (rl_printed_ == LOG_RATELIMIT_BURST) {	\
> +			fn("(suppressing further similar messages)");	\

There's one remaining problem with this implementation: this message is
printed only the first time the original logging message is *not*
printed.

There should be an overlap of the two cases instead, so that this
message is printed *just after* the last logged message.
Otherwise you could have the following situation:

  message A (#1)
  message A (#2)
  ...
  message A (#5)
  message B (#1)
  (suppressing further similar messages)

...as message A was being printed for the sixth time. But that seems to
refer to message B that was printed in between. You should make sure
it's either:

  message A (#1)
  message A (#2)
  ...
  message A (#5)
  (suppressing further similar messages)
  message B (#1)

or:

  message A (#1)
  message A (#2)
  ...
  message A (#5)
  message B (#1)
  message A (#6)
  (suppressing further similar messages)

I think the first version is slightly preferable as the limit is 5, as
defined, and not 6.

> +			rl_printed_++;					\
> +			rl_suppressed_++;				\
> +		} else {						\
> +			rl_suppressed_++;				\
> +		}							\
> +	} while (0)
> +
> +#define err_ratelimit(now, ...)						\
> +	logmsg_ratelimit(err, now, __VA_ARGS__)
> +#define warn_ratelimit(now, ...)					\
> +	logmsg_ratelimit(warn, now, __VA_ARGS__)
> +#define info_ratelimit(now, ...)					\
> +	logmsg_ratelimit(info, now, __VA_ARGS__)
> +#define debug_ratelimit(now, ...)					\
> +	logmsg_ratelimit(debug, now, __VA_ARGS__)
> +
>  extern int log_file;
>  extern int log_trace;
>  extern bool log_conf_parsed;

The rest looks good to me.

-- 
Stefano


      parent reply	other threads:[~2026-03-26  0:16 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-03-25 13:11 Anshu Kumari
2026-03-25 23:42 ` David Gibson
2026-03-26  0:16 ` Stefano Brivio [this message]

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=20260326011601.38b9b1f4@elisabeth \
    --to=sbrivio@redhat.com \
    --cc=anskuma@redhat.com \
    --cc=dgibson@redhat.com \
    --cc=lvivier@redhat.com \
    --cc=passt-dev@passt.top \
    /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.
Code repositories for project(s) associated with this public inbox

	https://passt.top/passt

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 IMAP folder(s).