On Fri, Jun 16, 2017 at 12:15:19PM -0000, Karl Palsson wrote: > > Daniel Golle <dan...@makrotopia.org> wrote: > > On Fri, Jun 16, 2017 at 10:30:14AM -0000, Karl Palsson wrote: > > > > > > > > > I fairly strong feel that this change brings no value to the > > > table. > > > > I disagree. For now, the two allocation schemes (hardcoded vs. > > dynamic) are competing for the same address space. This can > > result in a hard-coded UID/GID to be already allocated to a > > package using the dynamic allocation method. Shifting the > > dynamic allocation to the range starting from 65536 solves that > > problem in a convenient way. Hence I support Yousong's change. > > > > This doesn't fix that. There's absolutely nothing here that stops > someone using a hardcoded uid/gid of 65536 or so either. This > just changes one magic number to be a different magic number. => > no gain.
The subtle difference here is that there are for now no hardcoded UIDs/GIDs above 65536 and we can easily establish a convention that all static allocations should be below 65536. Starting from 100 is just obviously created a problem for now, and there is no non-violent solution as not all packages can easily be converted to use dynamic allocations. Cheers Daniel _______________________________________________ Lede-dev mailing list Lede-dev@lists.infradead.org http://lists.infradead.org/mailman/listinfo/lede-dev
