nvlink: Enable NPU device BAR before triggering freeze

NPU freeze injection works by performing an invalid MMIO read on the NPU
device BAR.  If the BAR isn't enabled, which is the case when the
appropriate driver isn't loaded, this checkstops the machine.

Work around this by making sure the BAR is enabled before performing the
read.  The idea of an error inject doing anything other than an error
inject isn't great, but it's better than unintentionally crashing your
machine.

Also, fix the comment incorrectly stating the operation was a write
instead of a read.

Signed-off-by: Russell Currey <ruscur@russell.cc>
Signed-off-by: Stewart Smith <stewart@linux.vnet.ibm.com>

1 files changed, 4 insertions, 1 deletions

diff --git a/hw/npu.c b/hw/npu.c
index 67b86c0e..d838cc1c 100644
--- a/hw/npu.c
+++ b/hw/npu.c
@@ -1128,7 +1128,10 @@ static int64_t npu_err_inject(struct phb *phb, uint32_t pe_num,
 		/* Emulate fence mode. */
 		p->fenced = true;
 	} else {
-		/* Cause a freeze with an invalid MMIO write. */
+		/* Cause a freeze with an invalid MMIO read.  If the BAR is not
+		 * enabled, this will checkstop the machine.
+		 */
+		npu_dev_bar_update(p->chip_id, &dev->bar, true);
 		in_be64((void *)dev->bar.base);
 	}