Alasdair G Kergon <agk@xxxxxxxxxx> wrote on 08.06.2007 00:57:44: > On Tue, Jun 05, 2007 at 03:05:30PM +0200, Stefan Bader wrote: > > This patch adds the new interface to dm-io. > > And also undoes some formatting, comments, variable name changes etc. > that I made when preparing earlier patches for upstream. Please revert > those changes and also look at what I changed last time, and make > similar changes to this code. > Ok, I will see to modify the patch based on your comments. > > - int bi_rw; /* READ|WRITE - not READA */ > + int bi_rw; /* READ|WRITE + bio flags */ > > Do we now support READA? > If I am not mistaken the only reason READA wasn't working were some places that compared against READ or WRITE. I changed those to test the bit values. > > But before that, let's examine and understand the proposed interface > changes in dm-io.h. The patch is unfortunately not easy to read in its > current form. Adding the new interface in parallel with the old was > appropriate previously (change to nature of interface, and we needed to > retain the original interface in RHEL) but I don't think it works very > well this time. > The main goal of the interface changes was to have more flexibility for the caller. Specifically it should be possible to get more information on the outcome of a region operation (error code). At the same time I tried to add additional notification calls for regions. But that probably is to much for one step. So I rework the whole patch for smaller steps. If that is okay this time I will just modify the interface without retaining the old one for the process. I guess this will make things simpler. Stefan -- dm-devel mailing list dm-devel@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/dm-devel
