From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mail-pl1-f172.google.com (mail-pl1-f172.google.com [209.85.214.172]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 28ED422D4C3 for ; Tue, 19 Aug 2025 16:37:25 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.214.172 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1755621446; cv=none; b=C1j9ZftAf9EhGe8BNieM6rnXmD065FwjpRrzrOMSsIRObknbcFf3mHAYbJ7NjNJAZtjqK5BYG3oYiVkXwoUMceiFEAeT/rqATDW4VwUOd251h+HK4UutgeAFD74J5ECwi4uTFci2TBelp5WTg8v64MWzCLzXveNObENkNvvv19c= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1755621446; c=relaxed/simple; bh=scbaPHNUfafnIekgS1W/IcAwZjeclgXRiBG9pKjzDeY=; h=Date:From:To:Cc:Subject:Message-ID:References:MIME-Version: Content-Type:Content-Disposition:In-Reply-To; b=pnDSWZKWhcUM8RdV7P1JYQbdf85cIG//w/0xLhzvsXViOt3fG6KEmyXguEpC+S3lxN7T5+Q5Ox4gBtkvDFzJ7E8+hOFHTFhZePJoW1GrStZUOSQNOves4Z2SUYX0YG9i9cT16Oolox1e7ldDJwfHzE9m5ap8P1SVKFigO1wo7jk= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=gmail.com; spf=pass smtp.mailfrom=gmail.com; dkim=pass (2048-bit key) header.d=gmail.com header.i=@gmail.com header.b=dwJpxRHC; arc=none smtp.client-ip=209.85.214.172 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=gmail.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=gmail.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=gmail.com header.i=@gmail.com header.b="dwJpxRHC" Received: by mail-pl1-f172.google.com with SMTP id d9443c01a7336-24458298aedso47170975ad.3 for ; Tue, 19 Aug 2025 09:37:25 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20230601; t=1755621444; x=1756226244; darn=lists.linux.dev; h=in-reply-to:content-disposition:mime-version:references:message-id :subject:cc:to:from:date:from:to:cc:subject:date:message-id:reply-to; bh=pN3oR61/c0uiXLRfyBJRzpXCbawo5PgtjuwTdlJFdps=; b=dwJpxRHCnUMJ6Z1szHWxNDnkYOjYoRDwGMKUxdQb0/udhv/mOWx3diuPAj4PNx4CUQ ySTRTk3sSMjtAzDnlAcruEjiFDgRpLKStlcSxilbiBG0lL46XRtfY/9dprisxPTNitdB drHKwACXPQm5c3Wn1TGQPsDAmn/mqlcraPeUqrWZsY76ijhgICR2rNwKgyqff+nf4lZF s2e7w+ZVSTacie63y6UzgqvSzUSeoGzoDASlp6q21LB/w2sglGUuyev8SX7Fw3JIA4Ws dVEC+vphRW2sWvvLDY/zxHSEzz8A1f//p7jHpisQN3CyJgij0U6pEzATsH69C52BlfD1 foGg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1755621444; x=1756226244; h=in-reply-to:content-disposition:mime-version:references:message-id :subject:cc:to:from:date:x-gm-message-state:from:to:cc:subject:date :message-id:reply-to; bh=pN3oR61/c0uiXLRfyBJRzpXCbawo5PgtjuwTdlJFdps=; b=QWupLPwvqyFUn6QIfKOdVBOyc+h4PlxRExoKOJ13IUylGuZvVxEE+FSCiS27YxGxUG OIo8hs8xpEfZUuXGMEzMSYwnl3NROGfq70Cjx5NphyCFHfgtM3mf92UZFKKkAd1ORqDJ cQppZJ5Ai7up8HYbs6M4IYQMFuGZAiT3btVw1qrZ9pRdX+JeVP4BbF/cDF88pD/Wmudj 5wjFYEUi+i44c08+JfmdibXuMpzFbmSYDc+5GQCjaKVJsSoTErMCUSuVgZMK4AW1iJ0X wV90e6nFZRF3V/uQUtvpYpGTkLFUWPql35gMHAah9iePi653F0+/0oO3ZQNg5vMUUrRW rnFg== X-Forwarded-Encrypted: i=1; AJvYcCUJla0hbW85wjZ2RVCHKSgLmEHvHVYNVXqh6KVReYdILeeH2S5V3ql+9IsTlREwg3NvzBgFIdShOGUvfCtOpeIjELFKRA==@lists.linux.dev X-Gm-Message-State: AOJu0Yz2Mq98W77SIEDl5eH0wBhne7QbmiA9mO5ff1iq+r0pJsZd956a JVqqObSEuoKuDnFu9F7jfvxgYQ+KgbnMwIAe4KAg/0RV5RpuKHokZFGW X-Gm-Gg: ASbGnctAvQEH0+FBLWTiLGZh/VMW4rzzNaXwIisAoYPX53GuQwycyJtIhkTo+dMABDb ve39rGMvmXPl4oWDwbK9eg4Gmpj5QytY51wygadR3JuiEMR9MeO8Mwhf2ohsBG6eUfOeDXn2ED5 QJu1JPQ7kY+6cW9lpUU3z3ZNmc4o886+tXe0McbkFnde+C7SFt/eckMEmVp2WSsOsU0vwLnX4ss dtawXDWjLNdq/KuDSrA3eBcv3lCmb90JZ2qtMA52gv+bAR03uVjFhSYprtz8berHecMWR+ZUrbR NyoAT+IZluUiWiAur0jxIqQ2OAbVj6NF16c0G7uleI+Kqy2x+c7DcKDQ7egQPeDEK3VnqEJPFLJ 8fRPsUeLZ X-Google-Smtp-Source: AGHT+IElLHXy5Z0g7yBhgA4isDUqzATIIqXde4d2BmuwggJSAMpaov5koyT/PbDQ1G2u3jDfv9QvHg== X-Received: by 2002:a17:902:d48f:b0:242:a0b0:3c1f with SMTP id d9443c01a7336-245e02cfc12mr45322725ad.7.1755621444380; Tue, 19 Aug 2025 09:37:24 -0700 (PDT) Received: from gmail.com ([2406:5900:2:f21::2a1]) by smtp.gmail.com with ESMTPSA id d9443c01a7336-245ed51c0aasm1428815ad.144.2025.08.19.09.37.21 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Tue, 19 Aug 2025 09:37:23 -0700 (PDT) Date: Wed, 20 Aug 2025 01:37:18 +0900 From: Ryan Chung To: Masami Hiramatsu Cc: rostedt@goodmis.org, mathieu.desnoyer@efficios.com, linux-kernel@vger.kernel.org, linux-trace-kernel@vger.kernel.org, linux-kernel-mentees@lists.linux.dev Subject: Re: [PATCH] trace/trace_fprobe.c: TODO: handle filter, nofilter or symbol list Message-ID: References: <20250812162101.5981-1-seokwoo.chung130@gmail.com> <20250814121504.2784e740a4e6fd4e0dd563d6@kernel.org> Precedence: bulk X-Mailing-List: linux-kernel-mentees@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20250814121504.2784e740a4e6fd4e0dd563d6@kernel.org> On Thu, Aug 14, 2025 at 12:15:04PM +0900, Masami Hiramatsu wrote: > Hi Ryan, > > On Wed, 13 Aug 2025 01:21:01 +0900 > Ryan Chung wrote: > > > Resolve TODO in `__register_trace_fprobe()`: > > parse `tf->symbol` robustly (support `sym!filter` and comma-separated lists), trim tokens, ignore empties, deduplicate symbols, use bulk registration for lists, return `-EEXIST` if already registered, and preserve lockdown/tracepoint deferral semantics. > > Thanks for the improvement! > And could you add the new syntax in the document too ? > Yes. I will add the syntax in the document. To clarify, by document, you mean Documentation/trace/fprobetrace.rst? > > > > Please note that this was my personal interpretation of what TODO > > required here. Welcoming any feedback. > > > > Signed-off-by: Ryan Chung > > --- > > kernel/trace/trace_fprobe.c | 102 +++++++++++++++++++++++++++++++++++- > > 1 file changed, 100 insertions(+), 2 deletions(-) > > > > diff --git a/kernel/trace/trace_fprobe.c b/kernel/trace/trace_fprobe.c > > index b40fa59159ac..37d4260b9012 100644 > > --- a/kernel/trace/trace_fprobe.c > > +++ b/kernel/trace/trace_fprobe.c > > @@ -12,6 +12,8 @@ > > #include > > #include > > #include > > +#include > > +#include > > Headers should be sorted alphabetically. > I will fix this in v2. > > > > #include "trace_dynevent.h" > > #include "trace_probe.h" > > @@ -762,8 +764,104 @@ static int __register_trace_fprobe(struct trace_fprobe *tf) > > return __regsiter_tracepoint_fprobe(tf); > > } > > > > - /* TODO: handle filter, nofilter or symbol list */ > > - return register_fprobe(&tf->fp, tf->symbol, NULL); > > + /* Parse tf->symbol */ > > Please make this parse and check as a sub-function instead of new > scope. Also, it should be done in parse_symbol_and_return(), so that > we can handle wrong syntax when parsing it. > I will move the parsing into parse_symbol_and_return() so syntax errors are detected at parse time. > > + { > > + char *spec, *bang, *p; > > + int n = 0, w = 0, j, rc; > > + char **syms = NULL; > > + > > + spec = kstrdup(tf->symbol, GFP_KERNEL); > > + if (!spec) > > + return -ENOMEM; > > + > > + /* If a '!' exists, treat it as single symbol + filter */ > > + bang = strchr(spec, '!'); > > + if (bang) { > > + char *sym, *flt; > > + > > + *bang = '\0'; > > + sym = strim(spec); > > + flt = strim(bang + 1); > > You don't need to do strim, since if there is a space, it > should be parsed already. New syntax must be ',' separated. > My basic syntax for this probe event is; > > WORD WORD WORD[:OPTWORD] SUBWORD[,SUBWORD] > > OPTWORD is qualifying the previous WORD, SUBWORDs are not > quarifying, but the same-level words. (Currently using "%return" > for the return of the function, that is a special case.) > Understood. I will drop strim() and treat tokens as you mentioned. I will leave return behavior unchanged. > > + > > + if (!*sym || !*flt) { > > + kfree(spec); > > Please use __free(kfree) instead of repeating kfree(). > I will also include this in v2. > > + return -EINVAL; /* reject empty symbol/filter */ > > Also, before returning an error, use trace_probe_log_err() to > notice the reason and the place of the error to user. > I will log parse failiures with trace_probe_log_err(). > > + } > > + > > + rc = register_fprobe(&tf->fp, sym, flt); > > + kfree(spec); > > + return rc; > > + } > > + > > + /* Comma list (or single symbol without '!') */ > > + /* First pass: count non-empty tokens */ > > + p = spec; > > + while (p) { > > + char *tok = strsep(&p, ","); > > + if (tok && *strim(tok)) > > + n++; > > + } > > + > > + if (n == 0){ > > + kfree(spec); > > + return -EINVAL; > > + } > > + > > + /* Allocate array for pointers into spec (callee copies/consumes) */ > > + syms = kcalloc(n, sizeof(*syms), GFP_KERNEL); > > + if (!syms) { > > + kfree(spec); > > + return -ENOMEM; > > + } > > + > > + /* Second pass: fill, skipping empties */ > > Again, symbol should not have a space. > Understood. I will also fix this in v2. > > + p = spec; > > + while (p) { > > + char *tok = strsep(&p, ","); > > + char *s; > > + > > + if (!tok) > > + break; > > + s = strim(tok); > > + if (!*s) > > + continue; > > + syms[w++] = s; > > + } > > + > > + /* Dedup in-place */ > > + for (i = 0; i < w; i++){ > > + if (!syms[i]) > > + continue; > > + for (j = i + 1; j < w; j++) { > > + if (syms[j] && !strcmp(syms[i], syms[j])) > > + syms[j] = NULL; > > + } > > I think dedup will be done in ftrace, so we don't need to do this > costly operation. > I see. I will remove the dedup here. > > + } > > + > > + /* Compact */ > > + for (i = 0, j = 0; i < w; i++) { > > + if (syms[i]) > > + syms[j++] = syms[i]; > > + } > > + w = j; > > + > > + /* After dedup, ensure we still have at least one symbol */ > > + if (w == 0){ > > + kfree(syms); > > + kfree(spec); > > + return -EINVAL; > > + } > > + > > + /* Register list or single symbol, using the existing bulk API */ > > + if (w == 1) > > + rc = register_fprobe(&tf->fp, syms[0], NULL); > > Hmm, you might misunderstand this. What you need to do is to classify > the list of symbols with '!' as nofilter, and others as "filter", > and pass those as "register_fprobe(&tf->fp, filter, nofilter)". > Thank you for the clarification. I will change as followed: - tokens prefixed with '!' go to the nofileter list - all other tokens go to filter list - pass both to register_fprobe(&tf->fp, filter, nofilter) > Thank you, > > > + else > > + rc = register_fprobe_syms(&tf->fp, (const char **)syms, w); > > + > > + kfree(syms); > > + kfree(spec); > > + return rc; > > + } > > } > > > > /* Internal unregister function - just handle fprobe and flags */ > > -- > > 2.43.0 > > > > > -- > Masami Hiramatsu (Google) Thank you. Please let me know if you have any questions or concerns. Best regards, Ryan Chung