bug#29895: Parted 3.2 accessing /dev/mmcblk0rpmb considered a bug?

2018-02-14 Thread Phil Susi 
On 12/29/2017 9:00 AM, Ceriel Jacobs wrote:
> eMMC spec 4.5 has a tiny (2MiB or 4MiB) Replay Protected Memory Block.
> As well as UFS and NVMe have support for RPMB.

This came to my attention a while back elsewhere.. probably the debian
bug tracker.  The question is how to identify such a device so it can be
excluded?  Does it have a unique dev no?  Or some sysfs attribute that
flags it as an RPMB?

bug#29895: Parted 3.2 accessing /dev/mmcblk0rpmb considered a bug?

2018-01-02 Thread Brian C. Lane 
On Fri, Dec 29, 2017 at 03:00:54PM +0100, Ceriel Jacobs wrote:
> eMMC spec 4.5 has a tiny (2MiB or 4MiB) Replay Protected Memory Block.
> As well as UFS and NVMe have support for RPMB.
> 
> That RPMB area is protected from readwrite access.
> 
> ---
> The kernel has been patched to not access this rpmb area:
> 
> $ dmesg | grep rpmb
> [6.712468] mmcblk0rpmb: mmc0:0001 SEM04G partition 3 2.00 MiB
> ---
> Parted (GNU parted) 3.2 still tries to access the RPMB:
> 
> Error: /dev/mmcblk0rpmb: unrecognised disk label
> Warning: Error fsyncing/closing /dev/mmcblk0rpmb: Input/output error
> Retry/Ignore?
> 
> Resulting in dmesg logged errors like:
> 
> [46684.715228] print_req_error: I/O error, dev mmcblk0rpmb, sector 0
> ---
> 
> Should (GNU) parted be patched too, or not?

What was the parted cmdline you were running?

And what are the permissions of the device node?

-- 
Brian C. Lane (PST8PDT)

bug#29895: Parted 3.2 accessing /dev/mmcblk0rpmb considered a bug?

2017-12-29 Thread Ceriel Jacobs 

eMMC spec 4.5 has a tiny (2MiB or 4MiB) Replay Protected Memory Block.
As well as UFS and NVMe have support for RPMB.

That RPMB area is protected from readwrite access.

---
The kernel has been patched to not access this rpmb area:

$ dmesg | grep rpmb
[6.712468] mmcblk0rpmb: mmc0:0001 SEM04G partition 3 2.00 MiB
---
Parted (GNU parted) 3.2 still tries to access the RPMB:

Error: /dev/mmcblk0rpmb: unrecognised disk label
Warning: Error fsyncing/closing /dev/mmcblk0rpmb: Input/output error
Retry/Ignore?

Resulting in dmesg logged errors like:

[46684.715228] print_req_error: I/O error, dev mmcblk0rpmb, sector 0
---

Should (GNU) parted be patched too, or not?


Cheers, Ceriel