Message ID | 55754C19.8030100@de.ibm.com |
---|---|
State | New |
Headers | show |
Am 08.06.2015 um 10:02 schrieb Christian Borntraeger: > So I would prefer to not have this workaround and doing > > index c480f64..7627d57 100644 > --- a/blockdev.c > +++ b/blockdev.c > @@ -976,17 +976,6 @@ DriveInfo *drive_new(QemuOpts *all_opts, BlockInterfaceType block_default_type) > goto fail; > } > > - if (type == IF_VIRTIO && !done_orphan_check > - && arch_type == QEMU_ARCH_S390X) { > - /* Virtio drives created on the command line get an implicit device > - * created for them. For non-s390x command line drives, the creation > - * of the implicit device is deferred to drive_check_orphaned. (S390x > - * is special-cased purely for backwards compatibility.) > - * Drives created via the monitor (hotplugged) do not get the > - * magic implicit device created for them. > - */ > - create_implicit_virtio_device(qdict_get_str(bs_opts, "id"), devaddr); > - } > > filename = qemu_opt_get(legacy_opts, "file"); > > actually enables the first command line to work as well. > So lets get rid of the s390 special handling (but I really appreciate that > you cared about s390) As a side note, I cannot test this with libvirt right now, as current qemu broke libvirts capability query see https://lists.gnu.org/archive/html/qemu-devel/2015-06/msg01806.html https://lists.gnu.org/archive/html/qemu-devel/2015-06/msg01488.html Christian
On 8 June 2015 at 09:18, Christian Borntraeger <borntraeger@de.ibm.com> wrote: > Am 08.06.2015 um 10:02 schrieb Christian Borntraeger: >> So I would prefer to not have this workaround and doing >> >> index c480f64..7627d57 100644 >> --- a/blockdev.c >> +++ b/blockdev.c >> @@ -976,17 +976,6 @@ DriveInfo *drive_new(QemuOpts *all_opts, BlockInterfaceType block_default_type) >> goto fail; >> } >> >> - if (type == IF_VIRTIO && !done_orphan_check >> - && arch_type == QEMU_ARCH_S390X) { >> - /* Virtio drives created on the command line get an implicit device >> - * created for them. For non-s390x command line drives, the creation >> - * of the implicit device is deferred to drive_check_orphaned. (S390x >> - * is special-cased purely for backwards compatibility.) >> - * Drives created via the monitor (hotplugged) do not get the >> - * magic implicit device created for them. >> - */ >> - create_implicit_virtio_device(qdict_get_str(bs_opts, "id"), devaddr); >> - } >> >> filename = qemu_opt_get(legacy_opts, "file"); >> >> actually enables the first command line to work as well. >> So lets get rid of the s390 special handling Cool, that certainly makes the code nicer not to have to keep the special case. I'll respin with that change folded in. >> (but I really appreciate that you cared about s390) Non-x86 architectures have to stick together, you know :-) > As a side note, I cannot test this with libvirt right now, as current qemu broke > libvirts capability query > see > https://lists.gnu.org/archive/html/qemu-devel/2015-06/msg01806.html > https://lists.gnu.org/archive/html/qemu-devel/2015-06/msg01488.html There's a fix for that that I've just now committed to master. thanks -- PMM
--- a/blockdev.c +++ b/blockdev.c @@ -976,17 +976,6 @@ DriveInfo *drive_new(QemuOpts *all_opts, BlockInterfaceType block_default_type) goto fail; } - if (type == IF_VIRTIO && !done_orphan_check - && arch_type == QEMU_ARCH_S390X) { - /* Virtio drives created on the command line get an implicit device - * created for them. For non-s390x command line drives, the creation - * of the implicit device is deferred to drive_check_orphaned. (S390x - * is special-cased purely for backwards compatibility.) - * Drives created via the monitor (hotplugged) do not get the - * magic implicit device created for them. - */ - create_implicit_virtio_device(qdict_get_str(bs_opts, "id"), devaddr); - } filename = qemu_opt_get(legacy_opts, "file");