Closed. This question does not meet Stack Overflow guidelines. It is not currently accepting answers.
This question does not appear to be about programming within the scope defined in the help center.
Closed 5 years ago.
Improve this question
I'd like to verify on any given Linux machine if PCI passthrough is supported. After a bit of googling, I found that I should rather check if IOMMU is supported, and I did so by running:
dmesg | grep IOMMU
If it supports IOMMU (and not IOMMUv2), I would get:
IOMMU
[ 0.000000] DMAR: IOMMU enabled
[ 0.049734] DMAR-IR: IOAPIC id 8 under DRHD base 0xfbffc000 IOMMU 0
[ 0.049735] DMAR-IR: IOAPIC id 9 under DRHD base 0xfbffc000 IOMMU 0
[ 1.286567] AMD IOMMUv2 driver by Joerg Roedel <jroedel#suse.de>
[ 1.286568] AMD IOMMUv2 functionality not available on this system
...where DMAR: IOMMU enabled is what I'm looking for.
Now, if the machine has been running for days without a reboot, that first message [ 0.000000] DMAR: IOMMU enabled might not appear any more in the log with the previous command.
Is there any way to check for IOMMU support when that message disappears from the log?
Since 2014 enabled iommu are registered in /sys (sysfs) special file system as class iommu (documented at ABI/testing/sysfs-class-iommu):
https://patchwork.kernel.org/patch/4345491/ "[2/3] iommu/intel: Make use of IOMMU sysfs support" - June 12, 2014
Register our DRHD IOMMUs, cross link devices, and provide a base set
of attributes for the IOMMU. ...
On a typical desktop system, this provides the following (pruned):
$ find /sys | grep dmar
/sys/devices/virtual/iommu/dmar0
...
/sys/class/iommu/dmar0
/sys/class/iommu/dmar1
The code is iommu_device_create (http://elixir.free-electrons.com/linux/v4.5/ident/iommu_device_create, around 4.5) or iommu_device_sysfs_add (http://elixir.free-electrons.com/linux/v4.11/ident/iommu_device_sysfs_add) in more recent kernels.
/*
* Create an IOMMU device and return a pointer to it. IOMMU specific
* attributes can be provided as an attribute group, allowing a unique
* namespace per IOMMU type.
*/
struct device *iommu_device_create(struct device *parent, void *drvdata,
const struct attribute_group **groups,
const char *fmt, ...)
Registration is done only for enabled IOMMU. DMAR:
if (intel_iommu_enabled) {
iommu->iommu_dev = iommu_device_create(NULL, iommu,
intel_iommu_groups,
"%s", iommu->name);
AMD IOMMU:
static int iommu_init_pci(struct amd_iommu *iommu)
{ ...
if (!iommu->dev)
return -ENODEV;
...
iommu->iommu_dev = iommu_device_create(&iommu->dev->dev, iommu,
amd_iommu_groups, "ivhd%d",
iommu->index);
Intel:
int __init intel_iommu_init(void)
{ ...
pr_info("Intel(R) Virtualization Technology for Directed I/O\n");
...
for_each_active_iommu(iommu, drhd)
iommu->iommu_dev = iommu_device_create(NULL, iommu,
intel_iommu_groups,
"%s", iommu->name);
With 4.11 linux kernel version iommu_device_sysfs_add is referenced in many IOMMU drivers, so checking /sys/class/iommu is better (more universal) way to programmatically detect enabled IOMMU than parsing dmesg output or searching in /var/log/kern.log or /var/log/messages for driver-specific enable messages:
Referenced in 10 files:
drivers/iommu/amd_iommu_init.c, line 1640
drivers/iommu/arm-smmu-v3.c, line 2709
drivers/iommu/arm-smmu.c, line 2163
drivers/iommu/dmar.c, line 1083
drivers/iommu/exynos-iommu.c, line 623
drivers/iommu/intel-iommu.c, line 4878
drivers/iommu/iommu-sysfs.c, line 57
drivers/iommu/msm_iommu.c, line 797
drivers/iommu/mtk_iommu.c, line 581
Related
To check if Intel's VT-X or AMD's AMD-V is enabled in the BIOS/UEFI, I use:
if systool -m kvm_amd -v &> /dev/null || systool -m kvm_intel -v &> /dev/null ; then
echo "AMD-V / VT-X is enabled in the BIOS/UEFI."
else
echo "AMD-V / VT-X is not enabled in the BIOS/UEFI"
fi
I couldn't find a way to check if Intel's VT-D or AMD's IOMMU are enabled in the BIOS/UEFI.
I need a way to detect if it is enabled or not without having the iommu kernel parameters set (iommu=1, amd_iommu=on, intel_iommu=on).
One idea I had was to use rdmsr, but I'm not sure if that would work. Instead of systool I initially wanted to use sudo rdmsr 0x3A, but it didn't work for me. It always reports:
rdmsr: CPU 0 cannot read MSR 0x0000003a
rdmsr is part of msr-tools btw. And to be used requires the msr kenel module to be loaded (sudo modprobe msr) first.
Allegedly sudo rdmsr 0x3A should have returned 3 or 5 to indicate that VT-X/AMD-V is enabled...
If VT-d is enabled, Linux will configure DMA Remapping at boot time. The easiest way to find this is to look in dmesg for DMAR entries. If you don't see errors, then VT-d is enabled.
For example:
[root#localhost ~]# dmesg | grep DMAR
[ 0.000000] ACPI: DMAR 0x00000000BBECB000 0000A8 (v01 LENOVO TP-R0D 00000930 PTEC 00000002)
[ 0.001000] DMAR: Host address width 39
[ 0.001000] DMAR: DRHD base: 0x000000fed90000 flags: 0x0
[ 0.001000] DMAR: dmar0: reg_base_addr fed90000 ver 1:0 cap 1c0000c40660462 ecap 19e2ff0505e
[ 0.001000] DMAR: DRHD base: 0x000000fed91000 flags: 0x1
[ 0.001000] DMAR: dmar1: reg_base_addr fed91000 ver 1:0 cap d2008c40660462 ecap f050da
[ 0.001000] DMAR: RMRR base: 0x000000bbdd8000 end: 0x000000bbdf7fff
[ 0.001000] DMAR: RMRR base: 0x000000bd000000 end: 0x000000bf7fffff
[ 0.001000] DMAR-IR: IOAPIC id 2 under DRHD base 0xfed91000 IOMMU 1
[ 0.001000] DMAR-IR: HPET id 0 under DRHD base 0xfed91000
[ 0.001000] DMAR-IR: Queued invalidation will be enabled to support x2apic and Intr-remapping.
[ 0.002000] DMAR-IR: Enabled IRQ remapping in x2apic mode
Another example with x2apic opt out:
[root#localhost ~]# dmesg | grep DMAR
[ 0.000000] ACPI: DMAR 0000000079a20300 000C4 (v01 SUPERM SMCI--MB 00000001 INTL 20091013)
[ 0.106389] DMAR: Host address width 46
[ 0.106392] DMAR: DRHD base: 0x000000fbffc000 flags: 0x1
[ 0.106400] DMAR: dmar0: reg_base_addr fbffc000 ver 1:0 cap 8d2078c106f0466 ecap f020de
[ 0.106402] DMAR: RMRR base: 0x0000007bb24000 end: 0x0000007bb32fff
[ 0.106404] DMAR: ATSR flags: 0x0
[ 0.106407] DMAR: RHSA base: 0x000000fbffc000 proximity domain: 0x0
[ 0.106409] DMAR-IR: IOAPIC id 8 under DRHD base 0xfbffc000 IOMMU 0
[ 0.106411] DMAR-IR: HPET id 0 under DRHD base 0xfbffc000
[ 0.106413] DMAR-IR: x2apic is disabled because BIOS sets x2apic opt out bit.
[ 0.106414] DMAR-IR: Use 'intremap=no_x2apic_optout' to override the BIOS setting.
[ 0.106591] DMAR-IR: Enabled IRQ remapping in xapic mode
Either way, you're looking for that last line, DMAR-IR: Enabled IRQ remapping in <whichever> mode.
On a system with VT-d disabled, you will either see an error message, or nothing at all.
[root#localhost ~]# dmesg | grep DMAR
[root#localhost ~]#
I just found another way that seems to work even if the iommu kernel parameters have not been set:
if compgen -G "/sys/kernel/iommu_groups/*/devices/*" > /dev/null; then
echo "AMD's IOMMU / Intel's VT-D is enabled in the BIOS/UEFI."
else
echo "AMD's IOMMU / Intel's VT-D is not enabled in the BIOS/UEFI"
fi
Building on Jo-Erlend Schinstad's answer:
Install cpu-checker
$ sudo apt-get update
$ sudo apt-get install cpu-checker
Then check:
$ kvm-ok
If the CPU is enabled, you should see something like:
INFO: /dev/kvm exists
KVM acceleration can be used
Otherwise, you might see something like:
INFO: /dev/kvm does not exist
HINT: sudo modprobe kvm_intel
INFO: Your CPU supports KVM extensions
INFO: KVM (vmx) is disabled by your BIOS
HINT: Enter your BIOS setup and enable Virtualization Technology (VT),
and then hard poweroff/poweron your system
KVM acceleration can NOT be used
I have a custom i.MX6 board, and I want to turn on a particular GPIO.
From the schematic, the GPIO pin is connected to KEY_COL2 pad, and the KEY_COL2 has the following options.
So, I have to export the following GPIO as per the calculation:
linux gpio number = (gpio_bank - 1) * 32 + gpio_bit
gpio number = ( 4 - 1 ) *32 +10 = 106
When I run the following command, i get the error:
# echo 106 > /sys/class/gpio/export
sh: write error: Device or resource busy
What can be the issue, am i missing anything...
After looking at the device tree, this particular GPIO was used by some other device, hence the error.
You can find the GPIO's in use with the following commands:
mount -t debugfs none /sys/kernel/debug
cat /sys/kernel/debug/gpio
Good day,
I am currently working on a project where PCIe SSDs are constantly being swapped out and tested through benchmark programs such as VDBench and Iometer. The problem I face right now, which is only on the Linux side (got it working fine on windows), is that if the drives were not on at initial boot-up, they never appear under GParted or Disks. Here's what I have done:
Cold boot, PCIe Add-in-card SSD is off. It is then powered on through a pass through card that is logically controlled to make sure power and shorts are not an issue.
I turn the device on, then run:
sudo sh -c "echo 1 > /sys/bus/pci/rescan"
Performing a
lspci -tv
The device shows with no issues in the tree. When I check under Disks however, it is not there.
I have tried a bunch of different commands with none of them seeming to help me. I have tried
partprobe
Which did not do anything. and:
sudo sh -c "echo 1 > /sys/bus/pci/devices/0000:82:00.0/remove"
Followed up another rescan:
sudo sh -c "echo 1 > /sys/bus/pci/rescan"
As well as:
sudo sh -c "echo 1 > /sys/bus/pci/devices/0000:82:00.0/enable"
Still nothing. Also ran:
dmesg
Which shows, amongst other things:
[ 68.128778] pci 0000:82:00.0: [8086:0953] type 00 class 0x010802
[ 68.128797] pci 0000:82:00.0: reg 0x10: [mem 0x00000000-0x00003fff 64bit]
[ 68.128820] pci 0000:82:00.0: reg 0x30: [mem 0x00000000-0x0000ffff pref]
[ 68.133398] pci 0000:84:00.0: [1c58:0003] type 00 class 0x010802
..............................
[ 68.141751] nvme 0000:82:00.0: enabling device (0100 -> 0102)
..............................
I do see a lot of fails in dmesg for other addresses, such as:
[ 1264.718446] pcieport 0000:03:04.0: BAR 14: no space for [mem size 0x00400000]
[ 1264.718448] pcieport 0000:03:04.0: BAR 14: failed to assign [mem size 0x00400000]
[ 1264.718451] pcieport 0000:03:04.0: BAR 13: no space for [io size 0x1000]
[ 1264.718453] pcieport 0000:03:04.0: BAR 13: failed to assign [io size 0x1000]
Although I have a feeling that those are unrelated to what I am doing, although I'd be happy for someone to prove me wrong.
So, after all of these attempts, does anyone know if there is a way (or if it is even possible) to scan for this PCIe Add-in NVMe SSD and be able to use it without rebooting? I also took a look at some of the threads for other HDDs that reference a rescan for sata based drives, but this is NOT that, so referencing that won't help either.
Thanks in advance.
I ran into the same issue benchmarking nvme PCIE passthrough with QEMU / Proxmox.
First take note of the driver in use:
lspci -nnk -s '0000:82:00.0'
It should say
Kernel driver in use: vfio-pci
Now unbind the driver, then reprobe:
echo '0000:82:00.0' > /sys/bus/pci/drivers/vfio-pci/unbind
echo '0000:82:00.0' > /sys/bus/pci/drivers_probe
Check the driver again with:
lspci -nnk -s '0000:82:00.0'
Kernel driver in use: nvme
lsblk should now show the drive. Found the procedure here
I tried doing that to save time which is used by rebooting. The PCI device driver at that time was dodgy at best about successfully rescanning and getting all its ducks in a row. The device was an FPGA presenting a proprietary interface device for a device driver I was developing. That was with kernel 2.6.30-something tried around March 2014. My (substandard, but acceptable) solution was to reboot the system.
I'm developing a program which need to access a special USB device. This USB device acts as a regular file in filesystem, so I have to open this file with O_DIRECT flag. As follow:
open(pathname, O_CREAT | O_RDWR | O_DIRECT | O_SYNC, S_IRWXU)
The program works well on PC environment. But when I port it to embedded board with openwrt, the "open" function returns EINVAL 22 /* Invalid argument */.
O_DIRECT support is selected in kernel configuration.
The filesystem of openwrt is squashfs and jffs2.
The filesystem of USB device is fat, and mounted on /media/aegis directory.
The ARCH of board is mips.
It seems that error is returned from following function in kernel:
int open_check_o_direct(struct file *f)
{
/* NB: we're sure to have correct a_ops only after f_op->open */
if (f->f_flags & O_DIRECT) {
if (!f->f_mapping->a_ops ||
((!f->f_mapping->a_ops->direct_IO) &&
(!f->f_mapping->a_ops->get_xip_mem))) {
return -EINVAL;
}
}
return 0;
}
Is it known that O_DIRECT isn't supported on jffs2 and supported on fat. When operating on file in /media/aegis I guess the a_ops of fat is used, but program doesn't run in my expectation.
I want to be able to do so from both Windows and from Linux. I know that there are ways by getting sysinfo and using thumb rules related to hardware identifiers.
I want to know if there is a more fundamental method, like looking at a memory address / issuing an interrupt etc.
BTW I am trying to do this on Intel hardware and the virtualization software I use are Vmware Workstation and Windows HyperV.
Here is one more useful command:
$ lscpu | grep -E 'Hypervisor vendor|Virtualization type'
Hypervisor vendor: KVM
Virtualization type: full
Example output of other commands:
$ sudo virt-what
kvm
$ dmesg | grep -i virtual
[ 0.000000] Booting paravirtualized kernel on KVM
[ 0.029160] CPU0: Intel QEMU Virtual CPU version 1.0 stepping 03
$ sudo dmidecode | egrep -i 'manufacturer|product|vendor|domU'
Vendor: Bochs
Manufacturer: Bochs
Product Name: Bochs
Manufacturer: Bochs
Manufacturer: Bochs
Manufacturer: Bochs
Manufacturer: Bochs
Manufacturer: Bochs
Manufacturer: Bochs
Manufacturer: Bochs
Manufacturer: Bochs
Manufacturer: Bochs
At least one of these should work to detect if you are running under VMware (or some other common virtual environment) on Linux:
Check for virtual devices detected by kernel when system boots.
dmesg | grep -i virtual
Another way to detect virtualized hardware devices, if dmesg doesn't say anything useful.
dmidecode | egrep -i 'manufacturer|product|vendor|domU'
You can also check for virtual disks:
cat /proc/ide/hd*/model
Virtuozzo can usually be detected by looking for /proc/vz or /dev/vzfs.
Most software check the hypervisor CPUID leaf -
Leaf 0x40000000, Hypervisor CPUID information
EAX: The maximum input value for hypervisor CPUID info (0x40000010).
EBX, ECX, EDX: Hypervisor vendor ID signature. E.g. "KVMKVMKVM"
Leaf 0x40000010, Timing information.
EAX: (Virtual) TSC frequency in kHz.
EBX: (Virtual) Bus (local apic timer) frequency in kHz.
ECX, EDX: RESERVED
Ofcourse, you are still relying on the hypervisor to give you this information. It may very well decide to not report 0x40000000 at all, in turn leading the guest to believe that it's actually running on real hardware