On Thu, Mar 04, 2021 at 17:23:05 +0100, Markus Armbruster wrote:
Daniel P. Berrangé <berrange(a)redhat.com> writes:
> On Thu, Mar 04, 2021 at 03:26:55PM +0100, Markus Armbruster wrote:
>> Daniel P. Berrangé <berrange(a)redhat.com> writes:
>>
>> > On Thu, Mar 04, 2021 at 11:00:57AM +0100, Markus Armbruster wrote:
>> >> Drop the crap deprecated in commit 4a27a638e7 "fdc: Deprecate
>> >> configuring floppies with -global isa-fdc" (v5.1.0).
>> >>
>> >> Signed-off-by: Markus Armbruster <armbru(a)redhat.com>
>> >> ---
>> >> docs/system/deprecated.rst | 26 --
>> >> docs/system/removed-features.rst | 26 ++
>> >> hw/block/fdc.c | 54 +--
>> >> tests/qemu-iotests/172 | 31 +-
>> >> tests/qemu-iotests/172.out | 562
+------------------------------
>> >> 5 files changed, 30 insertions(+), 669 deletions(-)
[...]
>>
>> Correct.
>>
>> This was deprecated in commit 4a27a638e7 "fdc: Deprecate configuring
>> floppies with -global isa-fdc" (v5.1.0). Since then, its use triggers a
>> warning:
>>
>> $ qemu-system-x86_64 -nodefaults -M q35 -display none -drive
if=none,id=drive-fdc0-0-0 -device isa-fdc,driveA=drive-fdc0-0-0,bootindexA=1
>> qemu-system-x86_64: -device isa-fdc,driveA=drive-fdc0-0-0,bootindexA=1:
warning: warning: property isa-fdc.driveA is deprecated
>> Use -device floppy,unit=0,drive=... instead.
>>
>> Note the -M q35. Needed because the default machine type has an onboard
>> isa-fdc, which cannot be configured this way.
>>
>> Sadly, the commit's update of docs/system/deprecated.rst neglects to
>> cover this use. Looks the series overtaxed my capacity to juggle
>> details; my apologies.
>>
>> Is libvirt still using these properties?
>
> Unfortunately yes, but it seems like it ought to be fairly easy to
> change the syntax. Just need to figure out what the right way to
> detect the availability of the new syntax is. Presumably just look
> for existance of the 'floppy' device type ?
Yes. The device type was added in merge commit fd209e4a7, v2.8.0.
> Can you confirm that switching from -global to the new -device floppy
> does /not/ have any live migration impact ?
Yes, it must not affect migration.
When Kevin split the floppy device type off the floppy controller, he
had to add some moderately ugly hackery to keep the old qdev properties
working. Think propagate property values to floppy from controller,
which otherwise ignores them.
The way you get the values into the floppy device cannot affect the
migration data. Only different values can.
This patch removes a deprecated way.
Note that when QEMU_CAPS_BLOCKDEV is asserted we format floppies as:
-blockdev
'{"driver":"file","filename":"/tmp/firmware.img",\
"node-name":"libvirt-2-storage","auto-read-only":true,"discard":"unmap"}'
\
-blockdev
'{"node-name":"libvirt-2-format","read-only":false,"driver":"raw",\
"file":"libvirt-2-storage"}' \
-device floppy,unit=0,drive=libvirt-2-format,id=fdc0-0-0 \
-blockdev
'{"driver":"file","filename":"/tmp/data.img",\
"node-name":"libvirt-1-storage","auto-read-only":true,"discard":"unmap"}'
\
-blockdev
'{"node-name":"libvirt-1-format","read-only":false,"driver":"qcow2",\
"file":"libvirt-1-storage"}' \
-device floppy,unit=1,drive=libvirt-1-format,id=fdc0-0-1 \
as visible in the test file:
tests/qemuxml2argvdata/disk-floppy-q35-2_11.x86_64-latest.args
So libvirt should be in the clear. isa-fdc with driveA/driveB is
formatted only when the blockdev capability isn't present.