From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-alma10-1.taild15c8.ts.net [100.103.45.18]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id E26B023C4F3 for ; Fri, 29 May 2026 07:51:34 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=100.103.45.18 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1780041096; cv=none; b=JyDH3nv3HVVmsdSMda/Fs4rqi10uXNN2BZ2svApNPZRKhicerOVIsNNtvq7fQJsmgNdpAB2pK4dRfbXmvoCDaw/Iml3V34mnWCYwr4zn335pHcSYmRlBtnF1zAoXrN+j9sodNuW9EQ0pgXOFYE5250c+YVX/QUtNVFlwjGGDCPQ= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1780041096; c=relaxed/simple; bh=/4tziB1BaFdsDBGhNtQVQksP9Fy+s3iGgIksMEqj/fA=; h=Message-ID:Date:MIME-Version:Subject:To:Cc:References:From: In-Reply-To:Content-Type; b=NnMwf6vL79g/FFdLyIDckzlyLnv7Prrg1YgX7jN2FtTi0qy8V2ySQkBL6ChwQ94RHUkLiWy8/IACiiAZOVsLepGdP3vOP6A+iJceVJbCaLeXdO0mpVd5U3f2a//PfCbPuyBMS3DA6PFItMs4L7z771jTzHB7SoLYQDoLa/l1xcM= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b=eiNDu4Lc; arc=none smtp.client-ip=100.103.45.18 Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b="eiNDu4Lc" Received: by smtp.kernel.org (Postfix) with ESMTPSA id 205B31F00893; Fri, 29 May 2026 07:51:32 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=kernel.org; s=k20260515; t=1780041094; bh=BqItn3go96ajO++fnJeMn+tQO+ikanoSwI/BdjIOUgM=; h=Date:Subject:To:Cc:References:From:In-Reply-To; b=eiNDu4LcaIxM6TpQddEpxQm1z3hE+eHn/jj2Tu2RIkT2p3s630bnEWeaoPh0VcUwq CC60TAEADz8lxWXpDsALAvcRFKOsZxV+zC9DRzFmopKqUbzJK5SDfrL+XINdv8DIjc SBM48AsCEhnGOggMlb+3bJLYvb93bA2l4Ftd8zsMGrWQf06K5DvFpnk79mhrvdIWdz HVVbaucBDpeTpRvmk2cjf7sTuNlsKgke0CqYAuDobV3eaJ/qdQHFG/aeKntbqfCyf1 zZ3Qg2znBu/RHWrJZz4K+PiJb4lj6XzI7e8S1441B6JqSRHKb6rRNpgJOQw+gssRJ1 5O+P6fsCrrNmQ== Message-ID: <584f7d44-4486-4fad-a925-abb69c362f06@kernel.org> Date: Fri, 29 May 2026 09:51:30 +0200 Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 User-Agent: Mozilla Thunderbird Subject: Re: [PATCH 3/3] powerpc/irq: Suppress unlikely interrupt stats by default To: Shrikanth Hegde , maddy@linux.ibm.com, linuxppc-dev@lists.ozlabs.org, tglx@kernel.org Cc: linux-kernel@vger.kernel.org References: <20260523174016.999456-1-sshegde@linux.ibm.com> <20260523174016.999456-4-sshegde@linux.ibm.com> Content-Language: fr-FR From: "Christophe Leroy (CS GROUP)" In-Reply-To: <20260523174016.999456-4-sshegde@linux.ibm.com> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit Le 23/05/2026 à 19:40, Shrikanth Hegde a écrit : > Some interrupts are always zero and that is expected since they occur > very rarely and are mostly error indications. Don't print them by > default. > > "MCE" - "Machine check exceptions" > "NMI" - "System Reset interrupts" > > Print them if they occur once. Maintain a bitmap to know which > interrupts are to be printed. Is that bitmap needed at all ? Can't we just print them as soon as they are not zero ? > > Time taken to read /proc/interrupts 1000 times. > Base and v6 details can be found in cover-letter. > Base : 103us > v6 : 63us > v6+patch 1+2 : 57us > v6+patch 1+2+3 : 54us > > Patch 3 shows an additional 5% gain compared to patch 1+2. So it does > make sense to print them only if they are ever set. > > Note: Since /proc/interrupts depend on kconfig and arch dependent, > userspace tools don't make explicit assumptions. > > Signed-off-by: Shrikanth Hegde > --- > arch/powerpc/include/asm/hardirq.h | 1 + > arch/powerpc/kernel/irq.c | 37 +++++++++++++++++++++++++++--- > arch/powerpc/kernel/traps.c | 4 ++-- > 3 files changed, 37 insertions(+), 5 deletions(-) > > diff --git a/arch/powerpc/include/asm/hardirq.h b/arch/powerpc/include/asm/hardirq.h > index 38098e35b241..be6cd5aab016 100644 > --- a/arch/powerpc/include/asm/hardirq.h > +++ b/arch/powerpc/include/asm/hardirq.h > @@ -31,6 +31,7 @@ DECLARE_PER_CPU(unsigned int, __softirq_pending); > #define local_softirq_pending_ref __softirq_pending > > #define inc_irq_stat(index) __this_cpu_inc(irq_stat.counts[IRQ_COUNT_##index]) > +void inc_irq_stat_and_enable(enum irq_stat_counts which); > > #define __ARCH_IRQ_STAT > #define __ARCH_IRQ_EXIT_IRQS_DISABLED > diff --git a/arch/powerpc/kernel/irq.c b/arch/powerpc/kernel/irq.c > index e67a18f62142..048ddfa66fc4 100644 > --- a/arch/powerpc/kernel/irq.c > +++ b/arch/powerpc/kernel/irq.c > @@ -87,9 +87,13 @@ u32 tau_interrupts(unsigned long cpu); > struct irq_stat_info { > const char *symbol; > const char *text; > + int skip; I'd call it 'optional' instead, and then during the print, if value 0 and optional then don't print. > }; > > -#define ISE(idx, sym, txt)[IRQ_COUNT_##idx] = { .symbol = sym, .text = txt} > +/* ISE - IRQ STAT ENABLED, ISC - IRQ STAT CONDITIONAL */ > +#define ISE(idx, sym, txt)[IRQ_COUNT_##idx] = { .symbol = sym, .text = txt, .skip = 0} > +#define ISC(idx, sym, txt)[IRQ_COUNT_##idx] = { .symbol = sym, .text = txt, .skip = 1} > + > > static struct irq_stat_info irq_stat_info[IRQ_COUNT_MAX] __ro_after_init = { > ISE(LOC_TIMER, "LOC", " Local timer interrupts for timer event device\n"), > @@ -97,8 +101,8 @@ static struct irq_stat_info irq_stat_info[IRQ_COUNT_MAX] __ro_after_init = { > ISE(OTHER_TIMER, "LOC", " Local timer interrupts for others\n"), > ISE(SPURIOUS, "SPU", " Spurious interrupts\n"), > ISE(PMI, "PMI", " Performance monitoring interrupts\n"), > - ISE(MCE, "MCE", " Machine check exceptions\n"), > - ISE(NMI_SRESET, "NMI", " System Reset interrupts\n"), > + ISC(MCE, "MCE", " Machine check exceptions\n"), > + ISC(NMI_SRESET, "NMI", " System Reset interrupts\n"), > #ifdef CONFIG_PPC_WATCHDOG > ISE(WATCHDOG, "WDG", " Watchdog soft-NMI interrupts\n"), > #endif > @@ -107,11 +111,25 @@ static struct irq_stat_info irq_stat_info[IRQ_COUNT_MAX] __ro_after_init = { > #endif > }; > > +/* > + * Used for default disabled counters to increment the stats and to enable the > + * entry for /proc/interrupts output. > + */ > +static DECLARE_BITMAP(irq_stat_count_show, IRQ_COUNT_MAX) __read_mostly; > +void inc_irq_stat_and_enable(enum irq_stat_counts which) > +{ > + __this_cpu_inc(irq_stat.counts[which]); > + set_bit(which, irq_stat_count_show); > +} > + > int arch_show_interrupts(struct seq_file *p, int prec) > { > const struct irq_stat_info *info = irq_stat_info; > > for (unsigned int i = 0; i < ARRAY_SIZE(irq_stat_info); i++, info++) { > + if (!test_bit(i, irq_stat_count_show)) > + continue; > + > seq_printf(p, "%*s:", prec, info->symbol); > irq_proc_emit_counts(p, &irq_stat.counts[i]); > seq_puts(p, info->text); > @@ -138,6 +156,19 @@ int arch_show_interrupts(struct seq_file *p, int prec) > return 0; > } > > +static int __init irq_init_stats(void) > +{ > + struct irq_stat_info *info = irq_stat_info; > + > + for (unsigned int i = 0; i < ARRAY_SIZE(irq_stat_info); i++, info++) { > + if (info->skip == 0) > + set_bit(i, irq_stat_count_show); > + } > + > + return 0; > +} > +late_initcall(irq_init_stats); > + > /* > * /proc/stat helpers > */ > diff --git a/arch/powerpc/kernel/traps.c b/arch/powerpc/kernel/traps.c > index a8f15154bd9a..3eacbd20fc80 100644 > --- a/arch/powerpc/kernel/traps.c > +++ b/arch/powerpc/kernel/traps.c > @@ -459,7 +459,7 @@ DEFINE_INTERRUPT_HANDLER_NMI(system_reset_exception) > } > > hv_nmi_check_nonrecoverable(regs); > - inc_irq_stat(NMI_SRESET); > + inc_irq_stat_and_enable(IRQ_COUNT_NMI_SRESET); > > /* See if any machine dependent calls */ > if (ppc_md.system_reset_exception) { > @@ -816,7 +816,7 @@ static void __machine_check_exception(struct pt_regs *regs) > { > int recover = 0; > > - inc_irq_stat(MCE); > + inc_irq_stat_and_enable(IRQ_COUNT_MCE); > > add_taint(TAINT_MACHINE_CHECK, LOCKDEP_NOW_UNRELIABLE); >