Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   Leonardo Bras <leobras.c@gmail.com>
To:     Michael Ellerman <mpe@ellerman.id.au>,
        Benjamin Herrenschmidt <benh@kernel.crashing.org>,
        Paul Mackerras <paulus@samba.org>,
        Leonardo Bras <leobras.c@gmail.com>,
        Alexey Kardashevskiy <aik@ozlabs.ru>,
        Thiago Jung Bauermann <bauerman@linux.ibm.com>,
        Ram Pai <linuxram@us.ibm.com>
Cc:     linuxppc-dev@lists.ozlabs.org, linux-kernel@vger.kernel.org
Subject: [PATCH v2 6/6] powerpc/pseries/iommu: Avoid errors when DDW starts at 0x00
Date:   Wed, 24 Jun 2020 03:24:11 -0300
Message-Id: <20200624062411.367796-7-leobras.c@gmail.com>
In-Reply-To: <20200624062411.367796-1-leobras.c@gmail.com>
References: <20200624062411.367796-1-leobras.c@gmail.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

As of today, enable_ddw() will return a non-null DMA address if the
created DDW maps the whole partition. If the address is valid,
iommu_bypass_supported_pSeriesLP() will consider iommu bypass enabled.

This can cause some trouble if the DDW happens to start at 0x00.

Instead if checking if the address is non-null, check directly if
the DDW maps the whole partition, so it can bypass iommu.

Signed-off-by: Leonardo Bras <leobras.c@gmail.com>
---
 arch/powerpc/platforms/pseries/iommu.c | 17 ++++++++---------
 1 file changed, 8 insertions(+), 9 deletions(-)

diff --git a/arch/powerpc/platforms/pseries/iommu.c b/arch/powerpc/platforms/pseries/iommu.c
index 2d217cda4075..967634a379b0 100644
--- a/arch/powerpc/platforms/pseries/iommu.c
+++ b/arch/powerpc/platforms/pseries/iommu.c
@@ -1078,7 +1078,8 @@ static void reset_dma_window(struct pci_dev *dev, struct device_node *par_dn)
  *
  * returns the dma offset for use by the direct mapped DMA code.
  */
-static u64 enable_ddw(struct pci_dev *dev, struct device_node *pdn)
+static u64 enable_ddw(struct pci_dev *dev, struct device_node *pdn,
+		      bool *maps_partition)
 {
 	int len, ret;
 	struct ddw_query_response query;
@@ -1237,9 +1238,9 @@ static u64 enable_ddw(struct pci_dev *dev, struct device_node *pdn)
 	list_add(&window->list, &direct_window_list);
 	spin_unlock(&direct_window_list_lock);
 
-	/* Only returns the dma_addr if DDW maps the whole partition */
 	if (len == order_base_2(max_addr))
-		dma_addr = be64_to_cpu(ddwprop->dma_base);
+		*maps_partition = true;
+	dma_addr = be64_to_cpu(ddwprop->dma_base);
 	goto out_unlock;
 
 out_free_window:
@@ -1324,6 +1325,7 @@ static bool iommu_bypass_supported_pSeriesLP(struct pci_dev *pdev, u64 dma_mask)
 {
 	struct device_node *dn = pci_device_to_OF_node(pdev), *pdn;
 	const __be32 *dma_window = NULL;
+	bool ret = false;
 
 	/* only attempt to use a new window if 64-bit DMA is requested */
 	if (dma_mask < DMA_BIT_MASK(64))
@@ -1344,13 +1346,10 @@ static bool iommu_bypass_supported_pSeriesLP(struct pci_dev *pdev, u64 dma_mask)
 			break;
 	}
 
-	if (pdn && PCI_DN(pdn)) {
-		pdev->dev.archdata.dma_offset = enable_ddw(pdev, pdn);
-		if (pdev->dev.archdata.dma_offset)
-			return true;
-	}
+	if (pdn && PCI_DN(pdn))
+		pdev->dev.archdata.dma_offset = enable_ddw(pdev, pdn, &ret);
 
-	return false;
+	return ret;
 }
 
 static int iommu_mem_notifier(struct notifier_block *nb, unsigned long action,
-- 
2.25.4