On 04/26/2012 02:38 PM, Stefan Berger wrote: > On 04/26/2012 04:15 PM, Eric Blake wrote: >> Similar to the recent race fix for 'block-stream', it is possible >> to set the speed of a block copy job up front thanks to an optional >> 'speed' parameter to 'drive-mirror'. >> >> * src/qemu/qemu_monitor_json.c (qemuMonitorJSONDriveMirror): Set >> speed at job start. >> * src/qemu/qemu_monitor_json.h (qemuMonitorJSONDriveMirror): Add >> parameter. >> * src/qemu/qemu_monitor.h (qemuMonitorDriveMirror): Likewise. >> * src/qemu/qemu_driver.c (qemuDomainBlockCopy): Adjust caller. >> * src/qemu/qemu_monitor.c (qemuMonitorDriveMirror): Likewise. >> --- >> >> v6: new patch. I haven't actually seen a qemu patch implementing this >> yet, but it is a logical extension of the recent block-stream patch. >> > > Hm, not sure what this means compared to what you write on the top... It means: Paolo proposed 'drive-mirror' to upstream qemu prior to Stefan Hajnoczi's patch to 'block-stream'. qemu 1.1 has included Stefan's patch, but Paolo's is still in limbo, and needs to be rebased if it will even make qemu 1.1. But it is a logical extension to assume that when Paolo rebases his drive-mirror patches, that he will also add an optional 'speed' parameter; so as to avoid the same data race as was just fixed in 'block-stream'. And until I can point to an actual qemu commit with 'drive-mirror' (at all, not just 'drive-mirror' with speed), this patch is stalled just like the other 8 patches in this series. -- Eric Blake eblake@xxxxxxxxxx +1-919-301-3266 Libvirt virtualization library http://libvirt.org
Attachment:
signature.asc
Description: OpenPGP digital signature
-- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list