Re: [libvirt] [PATCH 2/3] qemu: Explain mlock limit size more in detail

Based on Alex's explanation [1] in the recent discussion let's update the comment explaining the memory lock limit calculation. [1] http://www.redhat.com/archives/libvir-list/2015-November/msg00329.html

---   src/qemu/qemu_domain.c | 19 ++++++++++++++++---   1 file changed, 16 insertions(+), 3 deletions(-) diff --git a/src/qemu/qemu_domain.c b/src/qemu/qemu_domain.c index 12517a5..11cd39e 100644 --- a/src/qemu/qemu_domain.c +++ b/src/qemu/qemu_domain.c @@ -3639,9 +3639,22 @@ qemuDomainGetMlockLimitBytes(virDomainDefPtr def)           goto done;       } -    /* VFIO requires all of the guest's memory to be locked resident, plus some -     * amount for IO space. Alex Williamson suggested adding 1GiB for IO space -     * just to be safe (some finer tuning might be nice, though). */ +    /* For device pasthrough using VFIO the guest memory and MMIO memory regions

+     * need to be locked persistent in order to allow DMA. +     * +     * Currently the below limit is based on assumptions about the x86 platform. +     * +     * The chosen value of 1GiB below originates from x86 systems where it was +     * the used as space reserved for the MMIO region for the whole system.

+     * +     * On x86_64 systems the MMIO regions of the IOMMU mapped devices don't +     * count towards the locked memory limit since the memory is owned by the +     * device. Emulated devices though do count, but the regions are usually +     * small. Although it's not guaranteed that the limit will be enough for all +     * configurations it didn't pose a problem for now. +     * +     * Note that this may not be valid for all platforms. +     */       memKB = virDomainDefGetMemoryActual(def) + 1024 * 1024;    done: