Hi Josh
We’ve been working to securely zero secrets from memory when they’re no longer 
required, and I see this bug as being in the same class.

To really have confidence the value is erased from NVRAM is probably going to 
be platform specific because wear levelling on flash filesystems can create 
multiple copies of data (e.g., https://en.wikipedia.org/wiki/F2FS).  The fix 
for ASACORE-2618 is probably create a secure delete function that makes a best 
effort but that can be easily replaced by device manufacturers.

Greg

From: [email protected] 
[mailto:[email protected]] On Behalf Of Josh Spain
Sent: Thursday, February 25, 2016 1:17 PM
To: [email protected]
Subject: [Allseen-core] AJ_NVRAM_SecureDelete

Issue 
ASACORE-2618<https://na01.safelinks.protection.outlook.com/?url=https%3a%2f%2fjira.allseenalliance.org%2fbrowse%2fASACORE-2618&data=01%7c01%7cgregz%40microsoft.com%7ced024d93aaad4f27e6a608d33e290e05%7c72f988bf86f141af91ab2d7cd011db47%7c1&sdata=PSAcWQdafcy8rmufhpePk9CrkpVHAmPE2JkFc5cvojE%3d>

 discusses the fact that AJ_NVRAM_Delete does not clear the memory, making 
sensitive data vulnerable to access after deletion. I believe adding an 
AJ_NVRAM_SecureDelete function would satisfy the requirements without adding 
undue writes to NVRAM every single time delete is called. There are specific 
places in the TCL code where we would make the Secure call, but we could leave 
everything else using the original one, which does nothing but mark the NVRAM 
element id as invalid (much more optimal).

Does anyone have any thoughts on this?

-Josh

[http://i61.tinypic.com/5luc5u.png]

Josh Spain, Director of Engineering, Affinegy

1705 S. Capital of Texas Hwy, Ste. 310, Austin, TX, 78746
512.535.1700
[email protected]<mailto:[email protected]>      
http://affinegy.com<https://na01.safelinks.protection.outlook.com/?url=http%3a%2f%2faffinegy.com%2f&data=01%7c01%7cgregz%40microsoft.com%7ced024d93aaad4f27e6a608d33e290e05%7c72f988bf86f141af91ab2d7cd011db47%7c1&sdata=vKSHo23hNGK5OFnREh5705UdybHoHVOOk2kdsbySc%2bI%3d>

_______________________________________________
Allseen-core mailing list
[email protected]
https://lists.allseenalliance.org/mailman/listinfo/allseen-core

Reply via email to