[tip:x86/platform] x86/platform/UV: Remove Obsolete GRU MMR address translation

From: tip-bot for Dimitri Sivanich
Date: Wed May 04 2016 - 03:23:37 EST


Commit-ID: 40bfb8eedf1e7fa0535c685ff1000c05bcf7a637
Gitweb: http://git.kernel.org/tip/40bfb8eedf1e7fa0535c685ff1000c05bcf7a637
Author: Dimitri Sivanich <sivanich@xxxxxxx>
AuthorDate: Fri, 29 Apr 2016 16:54:22 -0500
Committer: Ingo Molnar <mingo@xxxxxxxxxx>
CommitDate: Wed, 4 May 2016 08:48:51 +0200

x86/platform/UV: Remove Obsolete GRU MMR address translation

Use no-op messages in place of cross-partition interrupts when nacking a
put message in the GRU. This allows us to remove MMR's as a destination
from the GRU driver.

Tested-by: John Estabrook <estabrook@xxxxxxx>
Tested-by: Gary Kroening <gfk@xxxxxxx>
Tested-by: Nathan Zimmer <nzimmer@xxxxxxx>
Signed-off-by: Dimitri Sivanich <sivanich@xxxxxxx>
Signed-off-by: Mike Travis <travis@xxxxxxx>
Cc: Andrew Banman <abanman@xxxxxxx>
Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
Cc: Andy Lutomirski <luto@xxxxxxxxxxxxxx>
Cc: Borislav Petkov <bp@xxxxxxxxx>
Cc: Brian Gerst <brgerst@xxxxxxxxx>
Cc: Denys Vlasenko <dvlasenk@xxxxxxxxxx>
Cc: H. Peter Anvin <hpa@xxxxxxxxx>
Cc: Len Brown <len.brown@xxxxxxxxx>
Cc: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Russ Anderson <rja@xxxxxxx>
Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Link: http://lkml.kernel.org/r/20160429215406.012228480@xxxxxxxxxxxxxxxxxxxxxxx
Signed-off-by: Ingo Molnar <mingo@xxxxxxxxxx>
---
arch/x86/include/asm/uv/uv_hub.h | 10 ----------
drivers/misc/sgi-gru/grukservices.c | 38 +++++++++++++++++++++----------------
2 files changed, 22 insertions(+), 26 deletions(-)

diff --git a/arch/x86/include/asm/uv/uv_hub.h b/arch/x86/include/asm/uv/uv_hub.h
index ae979f7..097b80c 100644
--- a/arch/x86/include/asm/uv/uv_hub.h
+++ b/arch/x86/include/asm/uv/uv_hub.h
@@ -647,16 +647,6 @@ static inline unsigned long uv_read_global_mmr64(int pnode, unsigned long offset
return readq(uv_global_mmr64_address(pnode, offset));
}

-/*
- * Global MMR space addresses when referenced by the GRU. (GRU does
- * NOT use socket addressing).
- */
-static inline unsigned long uv_global_gru_mmr_address(int pnode, unsigned long offset)
-{
- return UV_GLOBAL_GRU_MMR_BASE | offset |
- ((unsigned long)pnode << uv_hub_info->m_val);
-}
-
static inline void uv_write_global_mmr8(int pnode, unsigned long offset, unsigned char val)
{
writeb(val, uv_global_mmr64_address(pnode, offset));
diff --git a/drivers/misc/sgi-gru/grukservices.c b/drivers/misc/sgi-gru/grukservices.c
index 967b9dd..0307690 100644
--- a/drivers/misc/sgi-gru/grukservices.c
+++ b/drivers/misc/sgi-gru/grukservices.c
@@ -718,8 +718,8 @@ cberr:
static int send_message_put_nacked(void *cb, struct gru_message_queue_desc *mqd,
void *mesg, int lines)
{
- unsigned long m, *val = mesg, gpa, save;
- int ret;
+ unsigned long m;
+ int ret, loops = 200; /* experimentally determined */

m = mqd->mq_gpa + (gru_get_amo_value_head(cb) << 6);
if (lines == 2) {
@@ -735,22 +735,28 @@ static int send_message_put_nacked(void *cb, struct gru_message_queue_desc *mqd,
return MQE_OK;

/*
- * Send a cross-partition interrupt to the SSI that contains the target
- * message queue. Normally, the interrupt is automatically delivered by
- * hardware but some error conditions require explicit delivery.
- * Use the GRU to deliver the interrupt. Otherwise partition failures
+ * Send a noop message in order to deliver a cross-partition interrupt
+ * to the SSI that contains the target message queue. Normally, the
+ * interrupt is automatically delivered by hardware following mesq
+ * operations, but some error conditions require explicit delivery.
+ * The noop message will trigger delivery. Otherwise partition failures
* could cause unrecovered errors.
*/
- gpa = uv_global_gru_mmr_address(mqd->interrupt_pnode, UVH_IPI_INT);
- save = *val;
- *val = uv_hub_ipi_value(mqd->interrupt_apicid, mqd->interrupt_vector,
- dest_Fixed);
- gru_vstore_phys(cb, gpa, gru_get_tri(mesg), IAA_REGISTER, IMA);
- ret = gru_wait(cb);
- *val = save;
- if (ret != CBS_IDLE)
- return MQE_UNEXPECTED_CB_ERR;
- return MQE_OK;
+ do {
+ ret = send_noop_message(cb, mqd, mesg);
+ } while ((ret == MQIE_AGAIN || ret == MQE_CONGESTION) && (loops-- > 0));
+
+ if (ret == MQIE_AGAIN || ret == MQE_CONGESTION) {
+ /*
+ * Don't indicate to the app to resend the message, as it's
+ * already been successfully sent. We simply send an OK
+ * (rather than fail the send with MQE_UNEXPECTED_CB_ERR),
+ * assuming that the other side is receiving enough
+ * interrupts to get this message processed anyway.
+ */
+ ret = MQE_OK;
+ }
+ return ret;
}

/*