staging/rdma/hfi1: Adaptive PIO for short messages
The change requires a new pio_busy field in the iowait structure to
track the number of outstanding pios. The new counter together
with the sdma counter serve as the basis for a packet by packet decision
as to which egress mechanism to use. Since packets given to different
egress mechanisms are not ordered, this scheme will preserve the order.
The iowait drain/wait mechanisms are extended for a pio case. An
additional qp wait flag is added for the PIO drain wait case.
Currently the only pio wait is for buffers, so the no_bufs_available()
routine name is changed to pio_wait() and a third argument is passed
with one of the two pio wait flags to generalize the routine. A module
parameter is added to hold a configurable threshold. For now, the
module parameter is zero.
A heuristic routine is added to return the func pointer of the proper
egress routine to use.
The heuristic is as follows:
- SMI always uses pio
- GSI,UD qps <= threshold use pio
- UD qps > threadhold use sdma
o No coordination with sdma is required because order is not required
and this qp pio count is not maintained for UD
- RC/UC ONLY packets <= threshold chose as follows:
o If sdmas pending, use SDMA
o Otherwise use pio and enable the pio tracking count at
the time the pio buffer is allocated
- RC/UC ONLY packets > threshold use SDMA
o If pio's are pending the pio_wait with the new wait flag is
called to delay for pios to drain
The threshold is potentially reduced by the QP's mtu.
The sc_buffer_alloc() has two additional args (a callback, a void *)
which are exploited by the RC/UC cases to pass a new complete routine
and a qp *.
When the shadow ring completes the credit associated with a packet,
the new complete routine is called. The verbs_pio_complete() will then
decrement the busy count and trigger any drain waiters in qp destroy
or reset.
Reviewed-by: Jubin John <jubin.john@intel.com>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
diff --git a/include/rdma/rdmavt_qp.h b/include/rdma/rdmavt_qp.h
index 5c307ed..f2f4df0 100644
--- a/include/rdma/rdmavt_qp.h
+++ b/include/rdma/rdmavt_qp.h
@@ -82,6 +82,7 @@
* RVT_S_WAIT_DMA - waiting for send DMA queue to drain before generating
* next send completion entry not via send DMA
* RVT_S_WAIT_PIO - waiting for a send buffer to be available
+ * RVT_S_WAIT_PIO_DRAIN - waiting for a qp to drain pio packets
* RVT_S_WAIT_TX - waiting for a struct verbs_txreq to be available
* RVT_S_WAIT_DMA_DESC - waiting for DMA descriptors to be available
* RVT_S_WAIT_KMEM - waiting for kernel memory to be available
@@ -101,16 +102,17 @@
#define RVT_S_WAIT_SSN_CREDIT 0x0100
#define RVT_S_WAIT_DMA 0x0200
#define RVT_S_WAIT_PIO 0x0400
-#define RVT_S_WAIT_TX 0x0800
-#define RVT_S_WAIT_DMA_DESC 0x1000
-#define RVT_S_WAIT_KMEM 0x2000
-#define RVT_S_WAIT_PSN 0x4000
-#define RVT_S_WAIT_ACK 0x8000
-#define RVT_S_SEND_ONE 0x10000
-#define RVT_S_UNLIMITED_CREDIT 0x20000
-#define RVT_S_AHG_VALID 0x40000
-#define RVT_S_AHG_CLEAR 0x80000
-#define RVT_S_ECN 0x100000
+#define RVT_S_WAIT_PIO_DRAIN 0x0800
+#define RVT_S_WAIT_TX 0x1000
+#define RVT_S_WAIT_DMA_DESC 0x2000
+#define RVT_S_WAIT_KMEM 0x4000
+#define RVT_S_WAIT_PSN 0x8000
+#define RVT_S_WAIT_ACK 0x10000
+#define RVT_S_SEND_ONE 0x20000
+#define RVT_S_UNLIMITED_CREDIT 0x40000
+#define RVT_S_AHG_VALID 0x80000
+#define RVT_S_AHG_CLEAR 0x100000
+#define RVT_S_ECN 0x200000
/*
* Wait flags that would prevent any packet type from being sent.