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: <20160916200207.21439-5-yinghai@kernel.org>
Date:   Fri, 16 Sep 2016 13:01:54 -0700
From:   Yinghai Lu <yinghai@...nel.org>
To:     Bjorn Helgaas <bhelgaas@...gle.com>,
        David Miller <davem@...emloft.net>,
        Benjamin Herrenschmidt <benh@...nel.crashing.org>
Cc:     Wei Yang <weiyang@...ux.vnet.ibm.com>,
        Khalid Aziz <khalid.aziz@...cle.com>,
        linux-pci@...r.kernel.org, linux-kernel@...r.kernel.org,
        Yinghai Lu <yinghai@...nel.org>
Subject: [PATCH v14 04/17] PCI: Check resource alignment for /sys pci_mmap_resource path

When user access /sys/.../resourceX  with pci_mmap_resource(),
pci_mmap_resource():
  ...
  pci_resource_to_user(pdev, i, res, &start, &end);
  vma->vm_pgoff += start >> PAGE_SHIFT;
  mmap_type = res->flags & IORESOURCE_MEM ? pci_mmap_mem : pci_mmap_io;
  return pci_mmap_page_range(pdev, vma, mmap_type, write_combine);
so it will return virtual address for round_down of start.

user code should pass offset with PAGE_SIZE offset.
  fd = open(argv[1], O_RDONLY);
  ...
  sscanf(argv[2], "0x%lx", &offset);
  left = offset & (PAGE_SIZE - 1);
  offset &= PAGE_MASK;
  addr = mmap(NULL, PAGE_SIZE, PROT_READ, MAP_SHARED, fd, offset);
  for (i = 0; i < 8; i++)
    printf("%x ", addr[i + left]);
  munmap(addr, PAGE_SIZE);
  close(fd);

When the resource start is not PAGE_SIZE aligned, it should
be io port, pci_mmap_resource could return round_down address of
resource start.
As the whole point for pci_mmap_resource is passing offset in
[0, resource_size), user may assume virtual add is corresponding
to unaligned resource_size. Later they could get wrong value
with offset to resource start.

Block the path for now, and need to use pci_read_resource_io
/pci_write_resource_io path instead.
user code should be like:
  fd = open(argv[1], O_RDONLY);
  ...
  sscanf(argv[2], "0x%lx", &offset);
  for (i = 0; i < 8; i++) {
    pread(fd, &buf, 1, i + offset);
  }
  close(fd);

Signed-off-by: Yinghai Lu <yinghai@...nel.org>
---
 drivers/pci/pci-sysfs.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/drivers/pci/pci-sysfs.c b/drivers/pci/pci-sysfs.c
index d55d93d..e2eb79f 100644
--- a/drivers/pci/pci-sysfs.c
+++ b/drivers/pci/pci-sysfs.c
@@ -1023,6 +1023,16 @@ static int pci_mmap_resource(struct kobject *kobj, struct bin_attribute *attr,
 	if (i >= PCI_ROM_RESOURCE)
 		return -ENODEV;
 
+	/*
+	 * resource start have to be PAGE_SIZE aligned, as we pass
+	 * back virt address include round down of resource_start,
+	 * that caller can not figure out directly.
+	 * when it is not aligned, that mean it is io port, should go
+	 * pci_read_resource_io()/pci_write_resource_io() path.
+	 */
+	if (res->start & ~PAGE_MASK)
+		return -EINVAL;
+
 	if (res->flags & IORESOURCE_MEM && iomem_is_exclusive(res->start))
 		return -EINVAL;
 
-- 
2.8.3

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ