[PATCH] cxl: fix NULL dereference in cxl_context_init() on PowerVM guests

Commit f67a6722d650 ("cxl: Workaround PE=0 hardware limitation in Mellanox
CX4") added a "min_pe" field to struct cxl_service_layer_ops, to allow us
to work around a Mellanox CX-4 hardware limitation.

When allocating the PE number in cxl_context_init(), we read from
ctx->afu->adapter->native->sl_ops->min_pe to get the minimum PE number.
Unsurprisingly, in a PowerVM guest ctx->afu->adapter->native is NULL, and
guests don't have a cxl_service_layer_ops struct anywhere.

Move min_pe from struct cxl_service_layer_ops to struct cxl so it's
accessible in both native and PowerVM environments. For the Mellanox CX-4,
set the min_pe value in set_sl_ops().

Fixes: f67a6722d650 ("cxl: Workaround PE=0 hardware limitation in Mellanox CX4")
Reported-by: Frederic Barrat <redacted>
Signed-off-by: Andrew Donnellan <redacted>
---
 drivers/misc/cxl/context.c | 3 +--
 drivers/misc/cxl/cxl.h     | 2 +-
 drivers/misc/cxl/pci.c     | 3 ++-
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/misc/cxl/context.c b/drivers/misc/cxl/context.c
index bdee9a0..c466ee2 100644
--- a/drivers/misc/cxl/context.c
+++ b/drivers/misc/cxl/context.c

@@ -90,8 +90,7 @@ int cxl_context_init(struct cxl_context *ctx, struct cxl_afu *afu, bool master,
 	 */
 	mutex_lock(&afu->contexts_lock);
 	idr_preload(GFP_KERNEL);
-	i = idr_alloc(&ctx->afu->contexts_idr, ctx,
-		      ctx->afu->adapter->native->sl_ops->min_pe,
+	i = idr_alloc(&ctx->afu->contexts_idr, ctx, ctx->afu->adapter->min_pe,
 		      ctx->afu->num_procs, GFP_NOWAIT);
 	idr_preload_end();
 	mutex_unlock(&afu->contexts_lock);

diff --git a/drivers/misc/cxl/cxl.h b/drivers/misc/cxl/cxl.h
index 19b132f..76f8b94 100644
--- a/drivers/misc/cxl/cxl.h
+++ b/drivers/misc/cxl/cxl.h

@@ -549,7 +549,6 @@ struct cxl_service_layer_ops {
 	u64 (*timebase_read)(struct cxl *adapter);
 	int capi_mode;
 	bool needs_reset_before_disable;
-	int min_pe;
 };
 
 struct cxl_native {

@@ -591,6 +590,7 @@ struct cxl {
 	struct bin_attribute cxl_attr;
 	int adapter_num;
 	int user_irqs;
+	int min_pe;
 	u64 ps_size;
 	u16 psl_rev;
 	u16 base_image;

diff --git a/drivers/misc/cxl/pci.c b/drivers/misc/cxl/pci.c
index 3c8d0d2..bbf5fc9 100644
--- a/drivers/misc/cxl/pci.c
+++ b/drivers/misc/cxl/pci.c

@@ -1552,14 +1552,15 @@ static const struct cxl_service_layer_ops xsl_ops = {
 	.write_timebase_ctrl = write_timebase_ctrl_xsl,
 	.timebase_read = timebase_read_xsl,
 	.capi_mode = OPAL_PHB_CAPI_MODE_DMA,
-	.min_pe = 1, /* Workaround for Mellanox CX4 HW bug */
 };
 
 static void set_sl_ops(struct cxl *adapter, struct pci_dev *dev)
 {
 	if (dev->vendor == PCI_VENDOR_ID_MELLANOX && dev->device == 0x1013) {
+		/* Mellanox CX-4 */
 		dev_info(&adapter->dev, "Device uses an XSL\n");
 		adapter->native->sl_ops = &xsl_ops;
+		adapter->min_pe = 1; /* Workaround for CX-4 hardware bug */
 	} else {
 		dev_info(&adapter->dev, "Device uses a PSL\n");
 		adapter->native->sl_ops = &psl_ops;

-- 
Andrew Donnellan              OzLabs, ADL Canberra
andrew.donnellan@au1.ibm.com  IBM Australia Limited

Whoops!

Acked-by: Ian Munsie <redacted>

Le 28/07/2016 à 07:39, Andrew Donnellan a écrit :

Commit f67a6722d650 ("cxl: Workaround PE=0 hardware limitation in Mellanox
CX4") added a "min_pe" field to struct cxl_service_layer_ops, to allow us
to work around a Mellanox CX-4 hardware limitation.

When allocating the PE number in cxl_context_init(), we read from
ctx->afu->adapter->native->sl_ops->min_pe to get the minimum PE number.
Unsurprisingly, in a PowerVM guest ctx->afu->adapter->native is NULL, and
guests don't have a cxl_service_layer_ops struct anywhere.

Move min_pe from struct cxl_service_layer_ops to struct cxl so it's
accessible in both native and PowerVM environments. For the Mellanox CX-4,
set the min_pe value in set_sl_ops().

Fixes: f67a6722d650 ("cxl: Workaround PE=0 hardware limitation in Mellanox CX4")
Reported-by: Frederic Barrat <redacted>
Signed-off-by: Andrew Donnellan <redacted>

Reviewed-by: Frederic Barrat <redacted>

On Thu, 2016-28-07 at 05:39:41 UTC, Andrew Donnellan wrote:

Commit f67a6722d650 ("cxl: Workaround PE=0 hardware limitation in Mellanox
CX4") added a "min_pe" field to struct cxl_service_layer_ops, to allow us
to work around a Mellanox CX-4 hardware limitation.

When allocating the PE number in cxl_context_init(), we read from
ctx->afu->adapter->native->sl_ops->min_pe to get the minimum PE number.
Unsurprisingly, in a PowerVM guest ctx->afu->adapter->native is NULL, and
guests don't have a cxl_service_layer_ops struct anywhere.

Move min_pe from struct cxl_service_layer_ops to struct cxl so it's
accessible in both native and PowerVM environments. For the Mellanox CX-4,
set the min_pe value in set_sl_ops().

Fixes: f67a6722d650 ("cxl: Workaround PE=0 hardware limitation in Mellanox CX4")
Reported-by: Frederic Barrat <redacted>
Signed-off-by: Andrew Donnellan <redacted>
Acked-by: Ian Munsie <redacted>
Reviewed-by: Frederic Barrat <redacted>

Applied to powerpc fixes, thanks.

https://git.kernel.org/powerpc/c/164793379ad3b7ef5fc5a28260

cheers