On Mon, Feb 27, 2017 at 10:50:35AM +0000, Stuart Henderson wrote: > On 2017/02/27 21:37, Jonathan Gray wrote: > > On Mon, Feb 27, 2017 at 09:54:33AM +0000, Stuart Henderson wrote: > > > Using the current example (512-byte blocks): > > > 18874368 bytes transferred in 113.823 secs (165821 bytes/sec) > > > > > > Using bs=1m like we suggest in "Creating a bootable USB key using a > > > Un*x-like system": > > > 18874368 bytes transferred in 2.894 secs (6521502 bytes/sec) > > > > > > OK to change it? > > > > You've missed armv7 here. The concern is that conv=sync would need > > to be added in some cases unless the device is switched to a non-raw one. > > Ah - so in that case should conv=sync be added to m4.common as well? > > I've not run into any problems without conv=sync and I've written quite a > few SD cards/USB this way, but I was probably just lucky.. >
I suspect the miniroots you've used have all been created with a multiple of 1MB. I believe this came up in the context of dd'ing u-boot files where this wasn't the case. kettenis preferred switching to non-raw devices rather than using conv=sync for that.
