Memory check in tee_mm_alloc about optee_os HOT 2 CLOSED

Hi,

You're right, that code is broken. It doesn't look right when TEE_MM_POOL_HI_ALLOC is set either.

Yes entry->offset is an offset from pool->lo, or rather (entry->offset << pool->shift) is the relative address from pool->lo.

Regards,
Jens

from optee_os.

Hello,

Thank you for reviewing the change in pull request #308.
I believe there is also an issue in this line in tee_mm_init (tee_mm.c:52):

pool->entry->offset = ((hi - lo - 1) >> shift) + 1;

This is using the paramters hi and lo and then rounding up by one block. I believe this should use the rounded pool->hi and pool->lo instead of the parameters, or else you can end up with an offset that is past the end of the segment.

Thanks,
Youssef

from optee_os.