Firstly, thanks for providing a simple, easy to use test case. Makes our lives so much easier.

Using that test case reveals something interesting:

SQL> drop table t1 purge;

Table dropped.

SQL> drop table t2 purge;

Table dropped.

SQL> create table t1
  2     partition by list(id)
  3    ( partition p1 values (1) )
  4    as
  5   select a.*, 1 id
  6   from all_objects a;

Table created.

SQL> alter table t1 inmemory;

Table altered.

SQL> select count(*) from t1;

  COUNT(*)
----------
     90160

1 row selected.

SQL> select segment_name,partition_name,segment_type,
  2  inmemory_size,bytes_not_populated,populate_status
  3  from v$im_segments ;

SEGMENT_NAME                             PARTITION_NAME                 SEGMENT_TYPE       INMEMORY_SIZE BYTES_NOT_POPULATED POPULATE_
---------------------------------------- ------------------------------ ------------------ ------------- ------------------- ---------
T1                                       P1                             TABLE PARTITION          4325376                   0 COMPLETED

SQL> create table t2
  2  INMEMORY
  3   as
  4  select a.*,2 id
  5  from all_objects a;

Table created.

SQL> select segment_name,partition_name,segment_type,
  2    inmemory_size,bytes_not_populated,populate_status
  3  from v$im_segments ;

SEGMENT_NAME                             PARTITION_NAME                 SEGMENT_TYPE       INMEMORY_SIZE BYTES_NOT_POPULATED POPULATE_
---------------------------------------- ------------------------------ ------------------ ------------- ------------------- ---------
T1                                       P1                             TABLE PARTITION          4325376                   0 COMPLETED

1 row selected.

SQL> select count(*) from t2;

  COUNT(*)
----------
     90161

1 row selected.

SQL> select segment_name,partition_name,segment_type,
  2    inmemory_size,bytes_not_populated,populate_status
  3  from v$im_segments ;

SEGMENT_NAME                             PARTITION_NAME                 SEGMENT_TYPE       INMEMORY_SIZE BYTES_NOT_POPULATED POPULATE_
---------------------------------------- ------------------------------ ------------------ ------------- ------------------- ---------
T1                                       P1                             TABLE PARTITION          4325376                   0 COMPLETED
T2                                                                      TABLE                    4325376                   0 COMPLETED

2 rows selected.

SQL> alter table t1 add partition p2 values(2);

Table altered.

SQL> alter table t1  exchange partition p2 with table t2 ;

Table altered.

SQL> select count(*) from t1;

  COUNT(*)
----------
    180321

1 row selected.

SQL> select count(*) from t1 partition (p2);

  COUNT(*)
----------
     90161

1 row selected.

SQL> select count(*) from t2;

  COUNT(*)
----------
         0

1 row selected.

SQL> select segment_name,partition_name,segment_type,
  2   inmemory_size,bytes_not_populated,populate_status
  3  from v$im_segments ;

SEGMENT_NAME                             PARTITION_NAME                 SEGMENT_TYPE       INMEMORY_SIZE BYTES_NOT_POPULATED POPULATE_
---------------------------------------- ------------------------------ ------------------ ------------- ------------------- ---------
T1                                       P1                             TABLE PARTITION          4325376                   0 COMPLETED
T2                                                                      TABLE                    4325376                   0 COMPLETED

2 rows selected.

So as per your experience, it would appear the freshly exchanged partition is no longer in-memory. However, if we look both

1) the execution plan for accessing partition p2

SQL> set autotrace traceonly explain
SQL> select count(*) from t1 partition (p2);

Execution Plan
----------------------------------------------------------
Plan hash value: 938476605

---------------------------------------------------------------------------------------------
| Id  | Operation                    | Name | Rows  | Cost (%CPU)| Time     | Pstart| Pstop |
---------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT             |      |     1 |    27   (0)| 00:00:01 |       |       |
|   1 |  SORT AGGREGATE              |      |     1 |            |          |       |       |
|   2 |   PARTITION LIST SINGLE      |      | 90161 |    27   (0)| 00:00:01 |     2 |     2 |
|   3 |    TABLE ACCESS INMEMORY FULL| T1   | 90161 |    27   (0)| 00:00:01 |     2 |     2 |
---------------------------------------------------------------------------------------------

and also, the runtime IO when we access the partition

SQL> set autotrace on stat
SQL> select count(*) from t1 partition (p2);

  COUNT(*)
----------
     90161

1 row selected.


Statistics
----------------------------------------------------------
          0  recursive calls
          0  db block gets
          3  consistent gets
          0  physical reads
          0  redo size
        544  bytes sent via SQL*Net to client
        551  bytes received via SQL*Net from client
          2  SQL*Net roundtrips to/from client
          0  sorts (memory)
          0  sorts (disk)
          1  rows processed

SQL>

I think we can safely say that the *functionality* is correct, ie, partition P2 is indeed a in-memory object.

Which might suggest a possible idiosyncracy in the way V$IM_SEGMENTS is reporting its results after a partition exchange. You might to pursue that via 'My Oracle Support'. I'm also asking some internals people about it as well.

Once again, thanks for providing a simple, easy to use test case.