It is introduced by commit '9837684316 hostmem-file: add "align" option'.

Different assert() being hit, but similar problem as bug 1629720.

I'm lowering the severity and deferring it to 7.7 + 7.6.z, because I don't think this should block RHEL-7.6, just like bug 1629720.

Hi,

align option is for host side address alignment. "align=1.5G" is not the correct value and it asserts on `is_power_of_2(align)'. It acts as an indication for the user to restart the VM with appropriate value. This is important file backed memory and needs to be power of 2. Check is for expected behaviour. Don't think its a bug or regression.

Thanks,
Pankaj

(In reply to pagupta from comment #5)
> Hi,
> 
> align option is for host side address alignment. "align=1.5G" is not the
> correct value and it asserts on `is_power_of_2(align)'. It acts as an
> indication for the user to restart the VM with appropriate value. This is
> important file backed memory and needs to be power of 2. Check is for
> expected behaviour. Don't think its a bug or regression.

It's fine to print the error message, but it causes qemu core dumped, which I think is an issue need to be fixed. Besides, if hotplug the backend to a alive guest, qemu would core dumped too, might cause terrible consequence. IMO, that's definitely a bug. Thanks!

> 
> Thanks,
> Pankaj

(In reply to Yumei Huang from comment #6)
> (In reply to pagupta from comment #5)
> > Hi,
> > 
> > align option is for host side address alignment. "align=1.5G" is not the
> > correct value and it asserts on `is_power_of_2(align)'. It acts as an
> > indication for the user to restart the VM with appropriate value. This is
> > important file backed memory and needs to be power of 2. Check is for
> > expected behaviour. Don't think its a bug or regression.
> 
> It's fine to print the error message, but it causes qemu core dumped, which
> I think is an issue need to be fixed. Besides, if hotplug the backend to a
> alive guest, qemu would core dumped too, might cause terrible consequence.
> IMO, that's definitely a bug. Thanks!

Agree, Just thinking what's the right thing to do:

1] Asset VM start if file backed memory is misalign.

2] Throw an error at VM start and allow VM to start with :
   a) Not properly align file backed memory - might not be the best thing to do.
   b) Should not allow to add file backed memory if it is not properly aligned.

3] If hotplug backend to live guest also has misalign memory than hotplug   
   interface should check and throw an error thus does not effect live VM.

Among these scenarios I feel 1] & 3] should be right thing to do.

Thanks,
Pankaj

Hi,

This is fixed in upstream qemu, version >= v3.0.0. New check is added for align "is_power_of_2" in function "file_ram_alloc". This avoids assert in function "mmap-alloc.c:qemu_ram_mmap".
------------------

commit 61362b71c105ccb151ca16897a7d56534423f390
Author: David Hildenbrand <david>
Date:   Thu Jun 7 17:47:05 2018 +0200

    exec: check that alignment is a power of two
    
    Right now we can crash QEMU using e.g.
    
    qemu-system-x86_64 -m 256M,maxmem=20G,slots=2 \
     -object memory-backend-file,id=mem0,size=12288,mem-path=/dev/zero,align=12288 \
     -device pc-dimm,id=dimm1,memdev=mem0
    
    qemu-system-x86_64: util/mmap-alloc.c:115:
     qemu_ram_mmap: Assertion `is_power_of_2(align)' failed
    
    Fix this by adding a proper check.
    
    Signed-off-by: David Hildenbrand <david>
    Message-Id: <20180607154705.6316-3-david>
    Reviewed-by: Michael S. Tsirkin <mst>
    Reviewed-by: Igor Mammedov <imammedo>
    Signed-off-by: Paolo Bonzini <pbonzini>


Thanks,
Pankaj

Fix included in qemu-kvm-rhev-2.12.0-19.el7

Verify:
qemu-kvm-rhev-2.12.0-19.el7

QEMU quits with error message when boot with align=1.5G:

# /usr/libexec/qemu-kvm -m 4G,slots=20,maxmem=50G -M pc,nvdimm -object memory-backend-file,id=mem1,size=2G,mem-path=/tmp/nvdimm1,share,align=1.5G -device nvdimm,memdev=mem1,id=nv1 
qemu-kvm: -object memory-backend-file,id=mem1,size=2G,mem-path=/tmp/nvdimm1,share,align=1.5G: alignment 0x60000000 must be a power of two


If hotplug the object after guest boot up, hmp prints error message and guest works well.

# /usr/libexec/qemu-kvm -m 4G,slots=20,maxmem=50G -M pc,nvdimm -monitor  stdio  rhel76-64-virtio-scsi.qcow2  -vnc :0
(qemu) object_add  memory-backend-file,id=mem1,size=2G,mem-path=/tmp/nvdimm1,share,align=1.5G
alignment 0x60000000 must be a power of two
(qemu) info memdev

(qemu) info status
VM status: running

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHSA-2019:2553