Rik van Riel
2006-Aug-11 06:28 UTC
[Xen-devel] RFC/PATCH: hdparm tunable IDE write cache for HVM
qemu 0.8.2 has a flush callback to the storage backends, so now it is possible to implement hdparm tunable IDE write cache enable/disable for guest domains, allowing people to pick speed or data consistency on a case by case basis. As an added benefit, really large LBA48 IOs will now no longer be broken up into smaller IOs on the host side. What do you think? -- "Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it." - Brian W. Kernighan _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Christian Limpach
2006-Aug-11 11:07 UTC
Re: [Xen-devel] RFC/PATCH: hdparm tunable IDE write cache for HVM
On 8/11/06, Rik van Riel <riel@redhat.com> wrote:> qemu 0.8.2 has a flush callback to the storage backends, so now it is > possible to implement hdparm tunable IDE write cache enable/disable for > guest domains, allowing people to pick speed or data consistency on a > case by case basis. > > As an added benefit, really large LBA48 IOs will now no longer be broken > up into smaller IOs on the host side. > > What do you think?Looks good, are you confident that you have identified all the places where a flush is needed? Doesn''t qemu do flushes in the first place and wouldn''t it then have the same issues? christian _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Rik van Riel
2006-Aug-11 17:44 UTC
Re: [Xen-devel] RFC/PATCH: hdparm tunable IDE write cache for HVM
Christian Limpach wrote:> On 8/11/06, Rik van Riel <riel@redhat.com> wrote: >> qemu 0.8.2 has a flush callback to the storage backends, so now it is >> possible to implement hdparm tunable IDE write cache enable/disable for >> guest domains, allowing people to pick speed or data consistency on a >> case by case basis. >> >> As an added benefit, really large LBA48 IOs will now no longer be broken >> up into smaller IOs on the host side. >> >> What do you think? > > Looks good, are you confident that you have identified all the places > where a flush is needed?Yes, bdrv_write() is only called from 2 places in ide.c, and both of them have a bdrv_flush() call now.> Doesn''t qemu do flushes in the first place > and wouldn''t it then have the same issues?Nope, qemu does not seem to do any flushes by itself... -- "Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it." - Brian W. Kernighan _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Christian Limpach
2006-Aug-20 23:03 UTC
Re: [Xen-devel] RFC/PATCH: hdparm tunable IDE write cache for HVM
On 8/11/06, Rik van Riel <riel@redhat.com> wrote:> qemu 0.8.2 has a flush callback to the storage backends, so now it is > possible to implement hdparm tunable IDE write cache enable/disable for > guest domains, allowing people to pick speed or data consistency on a > case by case basis. > > As an added benefit, really large LBA48 IOs will now no longer be broken > up into smaller IOs on the host side. > > What do you think?Checked in. Please add a Signed-off-by line in the future. christian _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel