Cool, the document should be the design of the patch. The patch is a lustre in intel. Thanks,-:)
On 17 Sep, 2014, at 5:01 pm, Stefan Hajnoczi <stefa...@gmail.com> wrote: > On Tue, Sep 16, 2014 at 11:03:04PM +0800, Wu Libin wrote: >> Yeah, i know dm-cache, it’s in the layer of block. >> >> Maybe my description is not clear enough.Just because i see a patch, it will >> set flags which can classify IO size in bio->bi_flags, and in the function >> sd_prep_fn, it will set the SCSI command depends on the bio->bi_flags, like: >> SCpnt->cmnd[6] = DSS_BIO_CLASS(SCpnt->request->bio); >> >> And i see the data classification function will be enabled when the device >> model is “QEMU”, so i guess the QEMU must can recognise the SCSI command. >> That’s why i will search the data classification about the QEMU. >> >> It’s true QEMU not care about the filesystem, but we have flags, right, -:). > > After a web search I guess you are are talking about this paper: > http://sigops.org/sosp/sosp11/current/2011-Cascais/printable/05-mesnier.pdf > > The code isn't upstream and I've never heard of it. > > I think you should get in touch with the authors instead. > > Stefan
