From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from bombadil.infradead.org (bombadil.infradead.org [198.137.202.133]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id ACF78E77181 for ; Mon, 9 Dec 2024 15:55:43 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=lists.infradead.org; s=bombadil.20210309; h=Sender:List-Subscribe:List-Help :List-Post:List-Archive:List-Unsubscribe:List-Id:Content-Transfer-Encoding: Content-Type:MIME-Version:Message-ID:Date:References:In-Reply-To:Subject:Cc: To:From:Reply-To:Content-ID:Content-Description:Resent-Date:Resent-From: Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID:List-Owner; bh=Oa/eRmj9sytyxrud4O4mAvjeRx/S9NypgThFLDmDcCE=; b=zFexwU137/sCz2cN7QYtOaxL++ Ibl95Er4FAhnBySx1SPE7hVzG/XvoeuDGMnrsr3M8utnINAQz6dI5ZtDwFdUvr82tDMdyisMIPerR /Zzs3a1iUlIC4JnyifTWG5CPbJRVXGlnWo9I5FUj0V7Y9BT3LCM8tc9I3a09HydfdGoRjuLjwXY4v dnl6uO87GqXRerLyodU8WIRLC+ZVlFkvzbK4TtXYPT5fAfQ6+cHgkLvY3Ozea+zqlgwGdtopByqrO aWB7UxNQBUGjLyWYPXK/4O9b+QgncoMuX8PECeUxOQbtZ4HEHCV9ot8M7Tki/VAvzt+NoJgwApQpc WRA2dc2A==; Received: from localhost ([::1] helo=bombadil.infradead.org) by bombadil.infradead.org with esmtp (Exim 4.98 #2 (Red Hat Linux)) id 1tKg6f-00000008S9S-3WGc; Mon, 09 Dec 2024 15:55:33 +0000 Received: from galois.linutronix.de ([2a0a:51c0:0:12e:550::1]) by bombadil.infradead.org with esmtps (Exim 4.98 #2 (Red Hat Linux)) id 1tKg4i-00000008Ro5-0oYe; Mon, 09 Dec 2024 15:53:34 +0000 From: Thomas Gleixner DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linutronix.de; s=2020; t=1733759610; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=Oa/eRmj9sytyxrud4O4mAvjeRx/S9NypgThFLDmDcCE=; b=Ykr7ZkxqH+rHf85Xrj7aG+Lm+COeUJB7sSvEhIQX7r7g0TwE81qtfgtCxTODRzFZ+NGi7D +SK6ikPzhzfRcle7vTYH8T3Rz11u4xVeLGYboZuoDf/GtrUW0YWdKmcbXyZiuuSPB/i/Gs 0cYTX0XvY3sw9SlgygwKpdp7QkhThC5xVdcppxnbCVOAc6QJ5/o5BckCrtPd1O8PdQ8opn mz2pVwzYBL//F6pNBm4DopJCTDMseN2OsoVWmDyqxoa6tFQ28yshIkhV9mPjTD5hv74ERv HOoCreIJah+0eKDO6mwPcDMEyXHV9TUvoyB6Zi8LEsLa2WRxH8P6hGM7yFHS1w== DKIM-Signature: v=1; a=ed25519-sha256; c=relaxed/relaxed; d=linutronix.de; s=2020e; t=1733759610; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=Oa/eRmj9sytyxrud4O4mAvjeRx/S9NypgThFLDmDcCE=; b=mEGMYqFqXVsV1WU4zD3oERVxJV8xPiOpwHIShLDSIpc44nxTYpe8IO6Y5H/V3ZQV8ipTLT /X781OAzFoM2DGCw== To: Anup Patel Cc: Marc Zyngier , Shawn Guo , Sascha Hauer , Pengutronix Kernel Team , Andrew Lunn , Gregory Clement , Sebastian Hesselbarth , Palmer Dabbelt , Paul Walmsley , Atish Patra , Andrew Jones , Sunil V L , Anup Patel , linux-riscv@lists.infradead.org, linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org, imx@lists.linux.dev Subject: Re: [PATCH 1/4] irqchip/riscv-imsic: Handle non-atomic MSI updates for device In-Reply-To: References: <20241208150711.297624-1-apatel@ventanamicro.com> <20241208150711.297624-2-apatel@ventanamicro.com> <875xnuq6dc.ffs@tglx> Date: Mon, 09 Dec 2024 16:53:29 +0100 Message-ID: <87r06gq2di.ffs@tglx> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable X-CRM114-Version: 20100106-BlameMichelson ( TRE 0.8.0 (BSD) ) MR-646709E3 X-CRM114-CacheID: sfid-20241209_075332_529764_93B2C9AE X-CRM114-Status: GOOD ( 26.45 ) X-BeenThere: linux-arm-kernel@lists.infradead.org X-Mailman-Version: 2.1.34 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: "linux-arm-kernel" Errors-To: linux-arm-kernel-bounces+linux-arm-kernel=archiver.kernel.org@lists.infradead.org Anup! On Mon, Dec 09 2024 at 17:38, Anup Patel wrote: > On Mon, Dec 9, 2024 at 1:44=E2=80=AFAM Thomas Gleixner wrote: >> There is no guarantee that set_affinity() runs on the original target >> CPU (CPU 1). Your scheme only works, when CPU1 vector 0x20 is not used >> by some other device. If it's used, you lost as CPU1 will consume the >> vector and your pending check is not seeing anything. >> >> x86 ensures CPU locality by deferring the affinity move to the next >> device interrupt on the original target CPU (CPU1 in the above >> example). See CONFIG_GENERIC_IRQ_PENDING. > > I agree with you. > > The IMSIC driver must do the affinity move upon the next device > interrupt on the old CPU. I will update this patch in the next revision. > > BTW, I did not find CONFIG_GENERIC_IRQ_PENDING. Is the > name correct ? CONFIG_GENERIC_PENDING_IRQ is close enough :) >> The interrupt domains which are not affected (remap) set the >> IRQ_MOVE_PCNTXT flag to avoid that dance and don't use that affinity >> setter code path at all. > > Yes, setting the IRQ_MOVE_PCNTXT flag in the remap domain > makes perfect sense. > > I suggest adding IRQ_MOVE_PCNTXT usage as part of Drew's > irqbypass series which adds a remap domain in the IOMMU > driver. Unless you insist on having it as part of this series ? You need to look at the other RISC-V controllers. Those which do not need this should set it. That's historically backwards. I think we can reverse the logic here. As this needs backporting, I can't make a full cleanup of this, but for your problem the patch below should just work. Select GENERIC_PENDING_IRQ and GENERIC_PENDING_IRQ_CHIPFLAGS and set the IRQCHIP_MOVE_DEFERRED flag on your interrrupt chip and the core logic takes care of the PCNTXT bits. I'll convert x86 in a seperate step and remove the PCNTXT leftovers and the new config knob once the dust has settled. Thanks, tglx --- --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -567,6 +567,7 @@ struct irq_chip { * in the suspend path if they are in d= isabled state * IRQCHIP_AFFINITY_PRE_STARTUP: Default affinity update before start= up * IRQCHIP_IMMUTABLE: Don't ever change anything in this chip + * IRQCHIP_MOVE_DEFERRED: Move the interrupt in actual interrupt con= text */ enum { IRQCHIP_SET_TYPE_MASKED =3D (1 << 0), @@ -581,6 +582,7 @@ enum { IRQCHIP_ENABLE_WAKEUP_ON_SUSPEND =3D (1 << 9), IRQCHIP_AFFINITY_PRE_STARTUP =3D (1 << 10), IRQCHIP_IMMUTABLE =3D (1 << 11), + IRQCHIP_MOVE_DEFERRED =3D (1 << 12), }; =20 #include --- a/kernel/irq/Kconfig +++ b/kernel/irq/Kconfig @@ -31,6 +31,10 @@ config GENERIC_IRQ_EFFECTIVE_AFF_MASK config GENERIC_PENDING_IRQ bool =20 +# Deduce delayed migration from top-level interrupt chip flags +config GENERIC_PENDING_IRQ_CHIPFLAGS + bool + # Support for generic irq migrating off cpu before the cpu is offline. config GENERIC_IRQ_MIGRATION bool --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -47,6 +47,13 @@ int irq_set_chip(unsigned int irq, const return -EINVAL; =20 desc->irq_data.chip =3D (struct irq_chip *)(chip ?: &no_irq_chip); + + if (IS_ENABLED(CONFIG_GENERIC_PENDING_IRQ_CHIPFLAGS) && chip) { + if (chip->flags & IRQCHIP_MOVE_DEFERRED) + irqd_clear(&desc->irq_data, IRQD_MOVE_PCNTXT); + else + irqd_set(&desc->irq_data, IRQD_MOVE_PCNTXT); + } irq_put_desc_unlock(desc, flags); /* * For !CONFIG_SPARSE_IRQ make the irq show up in @@ -1114,16 +1121,21 @@ void irq_modify_status(unsigned int irq, trigger =3D irqd_get_trigger_type(&desc->irq_data); =20 irqd_clear(&desc->irq_data, IRQD_NO_BALANCING | IRQD_PER_CPU | - IRQD_TRIGGER_MASK | IRQD_LEVEL | IRQD_MOVE_PCNTXT); + IRQD_TRIGGER_MASK | IRQD_LEVEL); if (irq_settings_has_no_balance_set(desc)) irqd_set(&desc->irq_data, IRQD_NO_BALANCING); if (irq_settings_is_per_cpu(desc)) irqd_set(&desc->irq_data, IRQD_PER_CPU); - if (irq_settings_can_move_pcntxt(desc)) - irqd_set(&desc->irq_data, IRQD_MOVE_PCNTXT); if (irq_settings_is_level(desc)) irqd_set(&desc->irq_data, IRQD_LEVEL); =20 + /* Keep this around until x86 is converted over */ + if (!IS_ENABLED(CONFIG_GENERIC_PENDING_IRQ_CHIPFLAGS)) { + irqd_clear(&desc->irq_data, IRQD_MOVE_PCNTXT); + if (irq_settings_can_move_pcntxt(desc)) + irqd_set(&desc->irq_data, IRQD_MOVE_PCNTXT); + } + tmp =3D irq_settings_get_trigger_mask(desc); if (tmp !=3D IRQ_TYPE_NONE) trigger =3D tmp;