[dpdk-dev] rte_fbarray with Legacy mem

[kumaraparameshwaran rathinavel]

Hi All,

I am facing an issue with 20.05 when legacy mem is being used.

These are the logs added to the remap_segment function during
eal_legacy_hugepage_init and found below

EAL: remap_segment memseg_len:2097152 ms_idx:1095
msl->base_va:0x600000200000 post-addr:0x600089000000 physaddr:216000000
EAL: remap_segment memseg_len:2097152 ms_idx:1096
msl->base_va:0x600000200000 post-addr:0x600089200000 physaddr:216200000
EAL: remap_segment memseg_len:2097152 ms_idx:1097
msl->base_va:0x600000200000 post-addr:0x600089400000 physaddr:216400000
EAL: remap_segment memseg_len:2097152 ms_idx:1098
msl->base_va:0x600000200000 post-addr:0x600089600000 physaddr:216600000
EAL: remap_segment memseg_len:2097152 ms_idx:1099
msl->base_va:0x600000200000 post-addr:0x600089800000 physaddr:216800000
*EAL: remap_segment memseg_len:2097152 ms_idx:1155
msl->base_va:0x600000200000 post-addr:0x600090800000 physaddr:216e00000*
EAL: remap_segment memseg_len:2097152 ms_idx:1156
msl->base_va:0x600000200000 post-addr:0x600090a00000 physaddr:217000000
EAL: remap_segment memseg_len:2097152 ms_idx:1158
msl->base_va:0x600000200000 post-addr:0x600090e00000 physaddr:217600000
EAL: remap_segment memseg_len:2097152 ms_idx:1159
msl->base_va:0x600000200000 post-addr:0x600091000000 physaddr:217800000
*EAL: remap_segment memseg_len:2097152 ms_idx:1154
msl->base_va:0x600000200000 post-addr:0x600090600000 physaddr:218c00000*
EAL: remap_segment memseg_len:2097152 ms_idx:1161
msl->base_va:0x600000200000 post-addr:0x600091400000 physaddr:21a000000
EAL: remap_segment memseg_len:2097152 ms_idx:1162
msl->base_va:0x600000200000 post-addr:0x600091600000 physaddr:21a200000
EAL: remap_segment memseg_len:2097152 ms_idx:1164
msl->base_va:0x600000200000 post-addr:0x600091a00000 physaddr:21ac00000


   1. Can the above happen? (index returned increases and then decreases)
   2. In legacy mode if the physical address is not contiguous then the
   virtual address should not be contiguous(Please correct if I am wrong). But
   in the above we see that even though they are not physically contiguous
   because of the above, during the memory allocation it is assumed to be
   physically contiguous and facing other issues.
   3. Was there a bug in this part of the code which was fixed.

Thanks,
Param.