linux-kernel - [PATCH 4.9 167/206] s390/pci: fix dma address calculation in map

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20170110131510.520586146@linuxfoundation.org>
Date:   Tue, 10 Jan 2017 14:37:30 +0100
From:   Greg Kroah-Hartman <gregkh@...uxfoundation.org>
To:     linux-kernel@...r.kernel.org
Cc:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        stable@...r.kernel.org, Sebastian Ott <sebott@...ux.vnet.ibm.com>,
        Gerald Schaefer <gerald.schaefer@...ibm.com>,
        Martin Schwidefsky <schwidefsky@...ibm.com>
Subject: [PATCH 4.9 167/206] s390/pci: fix dma address calculation in map_sg

4.9-stable review patch.  If anyone has any objections, please let me know.

------------------

From: Sebastian Ott <sebott@...ux.vnet.ibm.com>

commit 6b7df3ce92ac82ec3f4a2953b6fed77da7b38aaa upstream.

__s390_dma_map_sg maps a dma-contiguous area. Although we only map
whole pages we have to take into account that the area doesn't start
or stop at a page boundary because we use the dma address to loop
over the individual sg entries. Failing to do that might lead to an
access of the wrong sg entry.

Fixes: ee877b81c6b9 ("s390/pci_dma: improve map_sg")
Reported-and-tested-by: Christoph Raisch <raisch@...ibm.com>
Signed-off-by: Sebastian Ott <sebott@...ux.vnet.ibm.com>
Reviewed-by: Gerald Schaefer <gerald.schaefer@...ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@...ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@...uxfoundation.org>

---
 arch/s390/pci/pci_dma.c |   15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

--- a/arch/s390/pci/pci_dma.c
+++ b/arch/s390/pci/pci_dma.c
@@ -419,6 +419,7 @@ static int __s390_dma_map_sg(struct devi
 			     size_t size, dma_addr_t *handle,
 			     enum dma_data_direction dir)
 {
+	unsigned long nr_pages = PAGE_ALIGN(size) >> PAGE_SHIFT;
 	struct zpci_dev *zdev = to_zpci(to_pci_dev(dev));
 	dma_addr_t dma_addr_base, dma_addr;
 	int flags = ZPCI_PTE_VALID;
@@ -426,8 +427,7 @@ static int __s390_dma_map_sg(struct devi
 	unsigned long pa = 0;
 	int ret;
 
-	size = PAGE_ALIGN(size);
-	dma_addr_base = dma_alloc_address(dev, size >> PAGE_SHIFT);
+	dma_addr_base = dma_alloc_address(dev, nr_pages);
 	if (dma_addr_base == DMA_ERROR_CODE)
 		return -ENOMEM;
 
@@ -436,26 +436,27 @@ static int __s390_dma_map_sg(struct devi
 		flags |= ZPCI_TABLE_PROTECTED;
 
 	for (s = sg; dma_addr < dma_addr_base + size; s = sg_next(s)) {
-		pa = page_to_phys(sg_page(s)) + s->offset;
-		ret = __dma_update_trans(zdev, pa, dma_addr, s->length, flags);
+		pa = page_to_phys(sg_page(s));
+		ret = __dma_update_trans(zdev, pa, dma_addr,
+					 s->offset + s->length, flags);
 		if (ret)
 			goto unmap;
 
-		dma_addr += s->length;
+		dma_addr += s->offset + s->length;
 	}
 	ret = __dma_purge_tlb(zdev, dma_addr_base, size, flags);
 	if (ret)
 		goto unmap;
 
 	*handle = dma_addr_base;
-	atomic64_add(size >> PAGE_SHIFT, &zdev->mapped_pages);
+	atomic64_add(nr_pages, &zdev->mapped_pages);
 
 	return ret;
 
 unmap:
 	dma_update_trans(zdev, 0, dma_addr_base, dma_addr - dma_addr_base,
 			 ZPCI_PTE_INVALID);
-	dma_free_address(dev, dma_addr_base, size >> PAGE_SHIFT);
+	dma_free_address(dev, dma_addr_base, nr_pages);
 	zpci_err("map error:\n");
 	zpci_err_dma(ret, pa);
 	return ret;