On Tue, Oct 27, 2020 at 01:54:46PM +0100, Ulf Hansson wrote: > > For information security purpose, some companies or business users set their notebook SD as "read only". > > Because a lot of "read only" requirements from those companies or business users, notebook vendor controls reader write protect pin to achieve it. > > Notebook BIOS might have option to choose "read only" or not. > > This is why we think write protect is more important than speed. > > I understand that it may be used, in some way or the other to provide > a hint to the operating system to mount it in read-only mode. > > Although, if there were a real security feature involved, the internal > FW of the SD card would also monitor the switch, to support read-only > mode. As I understand it, that's not the common case. Yes. "Security" that relies on the driver to fall back to a different mode doesn't work. > > > If you prefer to consistent behavior, I can ignore the write protect switch for SD express. > > At this point, I prefer if you would ignore the write protect switch > in the SD controller driver. Same here. > According to Christoph, it should be possible to support read-only > mode via PCIe/NVMe. You may need to add some tweaks to support this in > the PCIe controller driver, but I can't advise you how to exactly do > this. The NVMe driver already supports write protected namespaces. I'll ask my contact in the JEDEC SD card working group if there was any consideration of the read-only handling for classic SD vs NVMe.
