Skip to content

Commit

Permalink
powerpc/64s/idle: avoid POWER9 DD1 and DD2.0 ERAT workaround on DD2.1
Browse files Browse the repository at this point in the history
DD2.1 does not have to flush the ERAT after a state-loss idle.

Performance testing was done on a DD2.1 using only the stop0 idle state
(the shallowest state which supports state loss), using context_switch
selftest configured to ping-poing between two threads on the same core
and two different cores.

Performance improvement for same core is 7.0%, different cores is 14.8%.

Reviewed-by: Vaidyanathan Srinivasan <svaidy@linux.vnet.ibm.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
  • Loading branch information
Nicholas Piggin authored and Michael Ellerman committed Nov 6, 2017
1 parent b6b3755 commit 9d2f510
Showing 1 changed file with 2 additions and 0 deletions.
2 changes: 2 additions & 0 deletions arch/powerpc/kernel/idle_book3s.S
Original file line number Diff line number Diff line change
Expand Up @@ -542,7 +542,9 @@ pnv_restore_hyp_resource_arch300:
* then clear bit 60 in MMCRA to ensure the PMU starts running.
*/
blt cr3,1f
BEGIN_FTR_SECTION
PPC_INVALIDATE_ERAT
END_FTR_SECTION_IFSET(CPU_FTR_POWER9_DD1 | CPU_FTR_POWER9_DD20)
ld r1,PACAR1(r13)
mfspr r4,SPRN_MMCRA
ori r4,r4,(1 << (63-60))
Expand Down

0 comments on commit 9d2f510

Please sign in to comment.