From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mail-pf1-f176.google.com (mail-pf1-f176.google.com [209.85.210.176]) (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 CD2C425F994 for ; Sat, 9 May 2026 07:33:13 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.210.176 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1778311996; cv=none; b=W9HSGeTbPAldziZCw4/fIW7/FgkJFAsYdNwHlWMrYhXLZigyXhIxhEzy54Sesf6ehUtHQHAss3fSmKc4N3/J797GoKABhOKlG3OD/oE64WiTWpXV6gOOqvH6igiLhfylc7+ZtNyqQAr3dHq541JLmVARRkcLI7XpEDJtM4mguwE= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1778311996; c=relaxed/simple; bh=WsXgQqttQr6IBvVqKvxBRYRaHQiBAFIC9t5MITEoVeY=; h=Date:From:To:Cc:Subject:Message-ID:References:MIME-Version: Content-Type:Content-Disposition:In-Reply-To; b=vF5k0gVMK9cm3Jko++ARN7rRyWxPG8xSi8EAet1G/8TJspzMimkg9RWy9CV6MhdSgwNwp6l75a8EqH3p0FhkHPM+LQ3jTkdzklzALvxLWWuw5UtP2T20STLUa97nuUCCzjVwVCi2HvpWXfEGdMCJL5/LgYeUvHkYjJvE53kj15Y= 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=kZSShXei; arc=none smtp.client-ip=209.85.210.176 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="kZSShXei" Received: by mail-pf1-f176.google.com with SMTP id d2e1a72fcca58-83659d38e38so1270075b3a.1 for ; Sat, 09 May 2026 00:33:13 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20251104; t=1778311993; x=1778916793; darn=vger.kernel.org; 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=4rrWX8PR+HWUg34CwgZ0qTRfJ+rZargOz41FLZiRD60=; b=kZSShXei4Ke0NduTO91R8eSghN32Rai9VSLu/a7oHud0p71W4+tTDzHlVTX2lNO8uV wOX5K7X5ExG+etRTbSuk1Equky+HtdN9FOG8+MFPQQDlERQvI0cWdKGGZuaIzZxp8nR7 RhxQGIP7W1451PtK/5kNwTInXzYiR/Hvldw1S3Y/tkC3HmZEJsD8Lm/nCKmVtRh2nSB7 el0lDMc0fw8aR6R8nB3DN8zemQET7dJNBcd3UwqgdrBqsodpt945jOpjEhS/cLfDVvr9 UNgqt9hrs5XQ14OsqalOolDYDcMp4NHZoFgXXJV8eMff6q4iAfB5cqjL3+4OZOeuOv6d PkEw== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20251104; t=1778311993; x=1778916793; h=in-reply-to:content-disposition:mime-version:references:message-id :subject:cc:to:from:date:x-gm-gg:x-gm-message-state:from:to:cc :subject:date:message-id:reply-to; bh=4rrWX8PR+HWUg34CwgZ0qTRfJ+rZargOz41FLZiRD60=; b=LoViImYLI2eaDTRTtVgATK/x+9vCorHSuckfDzMq4o0MxxA13HVFVeS+cYCiidmioF ipLXeNXeoLetiCPlYMluS8J4ZSJjkVb1R433MU/wYLYtavUTupr8hO+OC4PPGUluCZMI vG0zJPblb7R1KqVKY3RYwtfTthI7wE20R3sX+krA7HEVVe+8lp5HNyQkhCQYgZZRaNWv 17Pa2OJgbcs3SLad5hAy5laLqWzgeZ+kt8KW5IjdFT6p+IDn9Xf7X7RW3B8QDHPITbVD iLo6JglfVzzqZb1OgC06TNdLcUMuTLB9UR1UnivF4WJe5Je9HPAjT/x4OgWCQh6ON6qS ZX5w== X-Forwarded-Encrypted: i=1; AFNElJ88nTwRP6haOf7nZJ7aezeB/tOLPbOzGveLkcob13UcI/jDeX+SBiEWYxs61B54VOOh1qXewO75moqrS6w=@vger.kernel.org X-Gm-Message-State: AOJu0Yyvg6ejsQOS4fGu9SLh2DLpCufLUlyOujh8K28MZjR01L43PeNL dKceo2RNPURB1qYs4q/SG+gmN2mvZCWmdMrAxiEfA6tzc+ZyNOcZBZg0 X-Gm-Gg: Acq92OEkTqo8tSRCaPamGQIEF4CvKqZNl3FNuY6o5Lae3vQ5MeRVHjHU9egK5V0rDWy J4DeVzhsXXH3M0DSjmg/YOgP6fSEkKl5z4ovCJiXtZCsdqZyfr9C5zQ02dJG83gqnzu60cyGeGZ 2Yx+TEkVPY8FnsCjn5SkAdjwenUY32N1ha/Aet+dBlgMDEdxS3YMGyZ6qAbBbprdYlzDFVbtNmR SO+4dS7a6LF27hif+DTZQwVVjFRidUbGwV5iTTZCp9iCPAQBlGvJxNsae5kv7Qyu2X3Q1CiyUpP rTZEcOn8zWtMjef6ST6eFwhaimZJjvxnO43TH4l6Zsz2Lte8b6iBkxSsyKyKdATrctmbxRcUUSI BLpj68dvccKQ5UIT2Ol61QsMHRTyESI44e8Eg5U7WoukDFDrL+yECTwTQDsTufzWrGvdlsJADNt bf5IYt+lcec3aDnHIpz/hOv8AwL5VVkEBxbA== X-Received: by 2002:a05:6a00:1407:b0:837:eaa9:381f with SMTP id d2e1a72fcca58-83a57c7782dmr15579899b3a.0.1778311993011; Sat, 09 May 2026 00:33:13 -0700 (PDT) Received: from localhost ([2001:19f0:8001:1b2d:5400:5ff:fefa:a95d]) by smtp.gmail.com with ESMTPSA id d2e1a72fcca58-83967dbf67fsm14681697b3a.47.2026.05.09.00.33.12 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Sat, 09 May 2026 00:33:12 -0700 (PDT) Date: Sat, 9 May 2026 15:32:52 +0800 From: Inochi Amaoto To: Bjorn Helgaas , Inochi Amaoto Cc: Jingoo Han , Manivannan Sadhasivam , Bjorn Helgaas , Lorenzo Pieralisi , Krzysztof =?utf-8?Q?Wilczy=C5=84ski?= , Rob Herring , Krzysztof Kozlowski , Conor Dooley , Yixun Lan , Paul Walmsley , Palmer Dabbelt , Albert Ou , Alexandre Ghiti , Alex Elder , Gustavo Pimentel , linux-pci@vger.kernel.org, devicetree@vger.kernel.org, linux-kernel@vger.kernel.org, linux-riscv@lists.infradead.org, spacemit@lists.linux.dev, Yixun Lan , Longbin Li Subject: Re: [PATCH 5/5] PCI: spacemit-k1: Add Spacemit K3 PCIe host controller support Message-ID: References: <20260502101319.2364052-6-inochiama@gmail.com> <20260507224217.GA48780@bhelgaas> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20260507224217.GA48780@bhelgaas> On Thu, May 07, 2026 at 05:42:17PM -0500, Bjorn Helgaas wrote: > On Sat, May 02, 2026 at 06:13:18PM +0800, Inochi Amaoto wrote: > > The PCIe controller on Spacemit K3 is almost a standard Synopsys > > Designware PCIe IP with extra link and reset control. Unlike > > the PCIe controller on K1, this controller supports external MSI > > interrupt controller and can use multiple phy at the same time. > > > > Add driver to support PCIe controller on Spacemit K3 PCIe. > > > > Signed-off-by: Inochi Amaoto > > Sashiko had some good questions: > https://sashiko.dev/#/patchset/20260502101319.2364052-1-inochiama%40gmail.com > > Looks like the CONFIG_PCIE_SPACEMIT_K1 menu item and help text > drivers/pci/controller/dwc/Kconfig should be updated to include K3. > > The "CONFIG_PCIE_SPACEMIT_K1" name itself should stay the same. > > s/Designware/DesignWare/, also in 4/5 commit log > s/phy/PHY/ here and other patches and subject lines > s/msi/MSI/ in 3/5 subject and commit log when it's a stand-alone word > s/pci:/PCI:/ in 4/5 subject to match history (and patch 3/5) > See it, thanks, I will take care of that > > +++ b/drivers/pci/controller/dwc/pcie-spacemit-k1.c > > > +#define INTR_STATUS 0x0010 > > + > > #define INTR_ENABLE 0x0014 > > #define MSI_CTRL_INT BIT(11) > > +#define RDLH_LINK_UP_INT BIT(20) > > + > > +#define K3_PHY_AHB_IRQSTATUS_INTX 0x0008 > > + > > +#define K3_PHY_AHB_IRQENABLE_SET_INTX 0x000c > > +#define LEG_EP_INTERRUPTS (BIT(6) | BIT(7) | BIT(8) | BIT(9)) > > Would be nicer to use "INTX" rather than "LEG" here since we use > "INTX" in K3_PHY_AHB_IRQENABLE_SET_INTX, in the comments, etc. > > > +#define K3_PHY_AHB_IRQENABLE_SET_MSI 0x0014 > > +/* MSI defined as BIT(11) in existing INTR_ENABLE, reusing */ > > + > > +#define K3_ADDR_INTR_STATUS1 0x0018 > > + > > +#define K3_ADDR_INTR_ENABLE1 0x001C > > You're using a mix of upper- and lower-case hex here. Be consistent > and match the existing code. > > Seems a little weird to have a mix of "IRQ" names (e.g., > K1_PHY_AHB_IRQ_EN, K3_PHY_AHB_IRQSTATUS_INTX, > K3_PHY_AHB_IRQENABLE_SET_INTX) and "INTR" names (e.g., INTR_STATUS, > INTR_ENABLE, K3_ADDR_INTR_STATUS1, K3_ADDR_INTR_ENABLE1) when I think > they're really talking about the same concept. > > And why do the new K3 names have "ADDR" in the middle when the > existing "INTR_ENABLE" names don't? It's obvious these are addresses > (well, actually I think they're *offsets*, but no need to be that > detailed). > In fact I have no detailed document about these name, but reference to their comments, I think it is a register for some link features. So it could be more accurate to be named with "LINK" > > +static int k3_pcie_init(struct dw_pcie_rp *pp) > > +{ > > ... > > + val = dw_pcie_readl_dbi(pci, GEN3_EQ_CONTROL_OFF); > > + val &= ~(0xffff << 8); > > + val |= ((0x1 << 4) << 8); > > Can you use FIELD_MODIFY and some #defines here? It is fine for me. > > > + dw_pcie_writel_dbi(pci, GEN3_EQ_CONTROL_OFF, val); > > + > > + /* Set the PCI vendor and device ID */ > > Superfluous comment since the code is obvious. > OK, I will remove it > > + dw_pcie_dbi_ro_wr_en(pci); > > + dw_pcie_writew_dbi(pci, PCI_VENDOR_ID, PCI_VENDOR_ID_SPACEMIT); > > + dw_pcie_writew_dbi(pci, PCI_DEVICE_ID, PCI_DEVICE_ID_SPACEMIT_K3); > > + dw_pcie_dbi_ro_wr_dis(pci); > > + > > + /* Finally, as a workaround, disable ASPM L1 */ > > I guess this means a device erratum? It advertises L1 but it doesn't > actually work? > > > + k1_pcie_disable_aspm_l1(k1); > > > +static int k3_pcie_msi_host_init(struct dw_pcie_rp *pp) > > +{ > > ... > > + val = dw_pcie_readl_dbi(pci, COHERENCY_CONTROL_3_OFF); > > + val |= (0xf << 11); > > FIELD_MODIFY and some #defines here? > OK. > > +static int k3_pcie_start_link(struct dw_pcie *pci) > > +{ > > + struct k1_pcie *k1 = to_k1_pcie(pci); > > + u32 val; > > + > > + k1_pcie_start_link(pci); > > + > > + /* Enable INTx */ > > + val = readl_relaxed(k1->link + K3_PHY_AHB_IRQENABLE_SET_INTX); > > + val |= LEG_EP_INTERRUPTS; > > + writel_relaxed(val, k1->link + K3_PHY_AHB_IRQENABLE_SET_INTX); > > + > > + /* Enable MSI/MSIX specific to K3 */ > > s/MSIX/MSI-X/ to match spec usage. > > > + val = readl_relaxed(k1->link + K3_ADDR_INTR_ENABLE1); > > + val |= (MSI_INT | MSIX_INT); > > + writel_relaxed(val, k1->link + K3_ADDR_INTR_ENABLE1); > > Generally speaking I think the interrupt setup belongs somewhere other > than .start_link(). Usually .start_link() only enables LTSSM. > Yes, this logic are not needed any more after I recheck the vendor code. Only thing related to the link will be left. With this, the macro like LEG_EP_INTERRUPTS can be removed. > > + return 0; > > +} > > > +static irqreturn_t k3_pcie_irq_thread(int irq, void *data) > > +{ > > + struct k1_pcie *k1 = data; > > + struct dw_pcie_rp *pp = &k1->pci.pp; > > + struct device *dev = k1->pci.dev; > > + u32 status0, status1, status2; > > + > > + k3_pcie_clear_irq_status(k1, &status0, &status1, &status2); > > + > > + writel_relaxed(status0, k1->link + K3_PHY_AHB_IRQSTATUS_INTX); > > + writel_relaxed(status1, k1->link + INTR_STATUS); > > + writel_relaxed(status2, k1->link + K3_ADDR_INTR_STATUS1); > > + > > + if (FIELD_GET(RDLH_LINK_UP_INT, status1)) { > > + msleep(PCIE_RESET_CONFIG_WAIT_MS); > > + /* Rescan the bus to enumerate endpoint devices */ > > + pci_lock_rescan_remove(); > > + pci_rescan_bus(pp->bridge->bus); > > This is the *only* driver that uses pci_rescan_bus() this way, which > automatically makes it suspicous. Maybe it's the first hardware that > implements or is willing to use RDLH_LINK_UP_INT for this, but somehow > I doubt it. > I am going to remove this. At least I do not think it is very proper to add this in the first version. The vendor explained that they use this interrupt to speed up the device link up check. This depends on a feature that make dwc skip the link up delay. And this feature is removed in v7.0. In commit 142d5869f6ee ("Revert "PCI: dwc: Don't wait for link up if driver can detect Link Up event"") Regards, Inochi > > + pci_unlock_rescan_remove(); > > + } else if (!status0 && !status1 && !status2) > > + dev_WARN_ONCE(dev, true, > > + "Received unknown event. status0=0x%08x status1=0x%08x status2=0x%08x\n", > > + status0, status1, status2); > > + > > + return IRQ_HANDLED; > > +}