From: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
To: Avi Kivity <avi@redhat.com>
Cc: KVM <kvm@vger.kernel.org>, Jan Kiszka <jan.kiszka@siemens.com>,
Marcelo Tosatti <mtosatti@redhat.com>,
qemu-devel@nongnu.org, Kevin O'Connor <kevin@koconnor.net>,
Anthony Liguori <anthony@codemonkey.ws>,
Liu Sheng <liusheng@linux.vnet.ibm.com>
Subject: [Qemu-devel] [PATCH v2 0/5] Qemu: implement readonly memory
Date: Thu, 25 Oct 2012 17:20:34 +0800 [thread overview]
Message-ID: <50890462.5010307@linux.vnet.ibm.com> (raw)
This patch set make the readonly memory in qemu really readonly by using
readonly memory slots feature in kvm to make qemu-kvm safer. Memory
regions with readonly property would be plug into kvm as readonly memory
slots.
Below module can test this feature:
static int rom_tester_probe(struct pci_dev *dev, const struct pci_device_id *id)
{
struct resource *res = &dev->resource[PCI_ROM_RESOURCE];
char buf[6];
size_t rom_size;
char * __iomem map;
int i;
if (res->flags & (IORESOURCE_ROM_SHADOW | IORESOURCE_ROM_COPY |
IORESOURCE_ROM_BIOS_COPY)) {
dev_printk(KERN_INFO, &dev->dev, "skip ROM COPY.\n");
return 0;
}
if (res->flags &
(IORESOURCE_ROM_COPY | IORESOURCE_ROM_BIOS_COPY))
dev_printk(KERN_INFO, &dev->dev, "rom tester\n");
if (pci_enable_rom(dev)) {
dev_printk(KERN_INFO, &dev->dev, "do not found Rom\n");
goto exit;
}
map = pci_map_rom(dev, &rom_size);
if (!map) {
dev_printk(KERN_INFO, &dev->dev, "map rom fail.\n");
goto disable_exit;
}
dev_printk(KERN_INFO, &dev->dev, "Rom map: %p [size: %lx], phsyc:%llx.\n",
map, rom_size, pci_resource_start(dev, PCI_ROM_RESOURCE));
if (rom_size < 6) {
printk("map size < 6.\n");
goto unmap_exit;
}
printk("The first 6 bytes:\n");
for (i = 0; i < 6; i++) {
buf[i] = map[i];
printk("%x ", buf[i]);
}
printk("\n\n");
memcpy(map, "KVMKVM", 6);
if (!memcmp(map, "KVMKVM", 6)) {
printk("Rom Test: fail.\n");
goto unmap_exit;
}
for (i = 0; i < 6; i++)
if (buf[i] != ((char *)map)[i]) {
printk("The %d byte is changed: %x -> %x.\n",
i, buf[i], map[i]);
printk("Rom Test: fail.\n");
goto unmap_exit;
}
printk("Rom Test: Okay.\n");
unmap_exit:
pci_unmap_rom(dev, map);
disable_exit:
pci_disable_rom(dev);
exit:
return 0;
}
static DEFINE_PCI_DEVICE_TABLE(rom_tester_tbl) = {
{ PCI_DEVICE(PCI_ANY_ID, PCI_ANY_ID)},
{0,} /* 0 terminated list. */
};
MODULE_DEVICE_TABLE(pci, rom_tester_tbl);
static struct pci_driver rom_tester = {
.name = "pci-rom-tester",
.id_table = rom_tester_tbl,
.probe = rom_tester_probe,
};
static int __init pci_rom_test_init(void)
{
int rc;
rc = pci_register_driver(&rom_tester);
if (rc)
return rc;
return 0;
}
static void __exit pci_rom_test_exit(void)
{
pci_unregister_driver(&rom_tester);
}
module_init(pci_rom_test_init);
module_exit(pci_rom_test_exit);
MODULE_LICENSE("GPL");
MODULE_AUTHOR("Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>");
we test it with the rom of Intel 82540EM Gigabit Ethernet Controller.
0. start qemu:
qemu-system-x86_64 -enable-kvm -m 1G -smp 2 -hda fedora16.qcow2 \
-nic nic,model=e1000,vlan=1,macadrr=52:00:12:34:56 \
-net user,vlan=1
1. unbind the device:
echo "0000:00:03.0" > /sys/bus/pci/devices/0000\:00\:03.0/driver/unbind
2. install the test kernel module, here we name it write_rom:
modprobe write_rom
3. print dmesg to verify the result is ok or fail:
dmesg
4. remove the test kernel module.
rmmod write_rom
5. rebind the device to its driver, test if the nic still works:
echo "0000:00:03.0" > /sys/bus/pci/drivers/e1000/bind
open firefox and try some web page.
when I use kvm without readonly memory slot, in step 2 it reports:
Rom Test: fail. this means we can write to the memory region of a rom,
which is quite not safe for the guest and host.
when I use kvm with readonly memory slot, in step 2 it reports:
Rom Test: Okay.
this means write operation is not successful, and return back to qemu from kvm.
thus we can make the rom real rom.
next reply other threads:[~2012-10-25 9:22 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-10-25 9:20 Xiao Guangrong [this message]
2012-10-25 9:21 ` [Qemu-devel] [PATCH v2 1/5] KVM: define KVM_CAP_READONLY_MEM unconditionally Xiao Guangrong
2012-10-25 12:14 ` Jan Kiszka
2012-10-25 9:21 ` [Qemu-devel] [PATCH v2 2/5] Qemu: update header files Xiao Guangrong
2012-10-25 11:03 ` Peter Maydell
2012-10-25 9:22 ` [Qemu-devel] [PATCH v2 3/5] Qemu: do not mark bios readonly Xiao Guangrong
2012-10-26 10:35 ` Jan Kiszka
2012-10-29 7:09 ` Xiao Guangrong
2012-10-29 7:44 ` Jan Kiszka
2012-10-29 8:31 ` Xiao Guangrong
2012-10-31 6:03 ` Jan Kiszka
2012-10-31 6:35 ` Xiao Guangrong
2012-10-31 6:46 ` Jan Kiszka
2012-10-31 7:01 ` Xiao Guangrong
2012-10-31 7:21 ` Jan Kiszka
2012-10-25 9:22 ` [Qemu-devel] [PATCH v2 4/5] Qemu: implement readonly memory Xiao Guangrong
2012-10-25 9:23 ` [Qemu-devel] [PATCH v2 5/5] Qemu: mark pci rom readonly Xiao Guangrong
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=50890462.5010307@linux.vnet.ibm.com \
--to=xiaoguangrong@linux.vnet.ibm.com \
--cc=anthony@codemonkey.ws \
--cc=avi@redhat.com \
--cc=jan.kiszka@siemens.com \
--cc=kevin@koconnor.net \
--cc=kvm@vger.kernel.org \
--cc=liusheng@linux.vnet.ibm.com \
--cc=mtosatti@redhat.com \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).