Skip to contentSkip to conversation

s390/virtio_ccw: fix config change notifications

PATCH In review Halil Pasic 4 participants 5 messages Jun 11, 2024 – Jun 13, 2024

Subsystems (4)
lore
Share & tools
lore

Halil Pasic

author maintainer
#0 lore raw

PATCH 1/1 s390/virtio_ccw: fix config change notifications

Commit e3e9bda38e6d ("s390/virtio_ccw: use DMA handle from DMA API") broke configuration change notifications for virtio-ccw by putting the DMA address of *indicatorp directly into ccw->cda disregarding the fact that if !!(vcdev->is_thinint) then the function virtio_ccw_register_adapter_ind() will overwrite that ccw->cda value with the address of the virtio_thinint_area so it can actually set up the adapter interrupts via CCW_CMD_SET_IND_ADAPTER. Thus we end up pointing to the wrong object for both CCW_CMD_SET_IND if setting up the adapter interrupts fails, and for CCW_CMD_SET_CONF_IND regardless whether it succeeds or fails.

To fix this, let us save away the dma address of *indicatorp in a local variable, and copy it to ccw->cda after the "vcdev->is_thinint" branch.

Fixes e3e9bda38e6d ("s390/virtio_ccw: use DMA handle from DMA API")

---
I know that checkpatch.pl complains about a missing 'Closes' tag. Unfortunately I don't have an appropriate URL at hand. @Sebastian, @Boqiao: do you have any suggetions?

---
 drivers/s390/virtio/virtio_ccw.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/s390/virtio/virtio_ccw.c +3 -1
index d7569f395559..d6491fc84e8c 100644
@@ -698,6 +698,7 @@ static int virtio_ccw_find_vqs(struct virtio_device *vdev, unsigned nvqs,
 	dma64_t *indicatorp = NULL;
 	int ret, i, queue_idx = 0;
 	struct ccw1 *ccw;
+	dma32_t indicatorp_dma = 0;
 
 	ccw = ccw_device_dma_zalloc(vcdev->cdev, sizeof(*ccw), NULL);
 	if (!ccw)
@@ -725,7 +726,7 @@ static int virtio_ccw_find_vqs(struct virtio_device *vdev, unsigned nvqs,
 	*/
 	indicatorp = ccw_device_dma_zalloc(vcdev->cdev,
 					   sizeof(*indicatorp),
-					   &ccw->cda);
+					   &indicatorp_dma);
 	if (!indicatorp)
 		goto out;
 	*indicatorp = indicators_dma(vcdev);
@@ -735,6 +736,7 @@ static int virtio_ccw_find_vqs(struct virtio_device *vdev, unsigned nvqs,
 			/* no error, just fall back to legacy interrupts */
 			vcdev->is_thinint = false;
 	}
+	ccw->cda = indicatorp_dma;
 	if (!vcdev->is_thinint) {
 		/* Register queue indicators with host. */
 		*indicators(vcdev) = 0;

base-commit: 83a7eefedc9b56fe7bfeff13b6c7356688ffa670

Signature · 1 line
2.40.1
Tested-by on v1 1/1
#1 lore raw
Quoted 14 lines · Halil Pasic

Commit e3e9bda38e6d ("s390/virtio_ccw: use DMA handle from DMA API") broke configuration change notifications for virtio-ccw by putting the DMA address of *indicatorp directly into ccw->cda disregarding the fact that if !!(vcdev->is_thinint) then the function virtio_ccw_register_adapter_ind() will overwrite that ccw->cda value with the address of the virtio_thinint_area so it can actually set up the adapter interrupts via CCW_CMD_SET_IND_ADAPTER. Thus we end up pointing to the wrong object for both CCW_CMD_SET_IND if setting up the adapter interrupts fails, and for CCW_CMD_SET_CONF_IND regardless whether it succeeds or fails.

To fix this, let us save away the dma address of *indicatorp in a local variable, and copy it to ccw->cda after the "vcdev->is_thinint" branch.

Quoted 8 lines
Fixes e3e9bda38e6d ("s390/virtio_ccw: use DMA handle from DMA API")

---
I know that checkpatch.pl complains about a missing 'Closes' tag. Unfortunately I don't have an appropriate URL at hand. @Sebastian, @Boqiao: do you have any suggetions?

?

Anyway, I've tested the patch and it indeed fixes the problem with virtio-balloon and the link state for me:

author maintainer
#2 lore raw
Quoted 24 lines · On Wed, 12 Jun 2024 16:04:15 +0200 Thomas Huth <[email protected]> wrote:
Quoted 14 lines · Halil Pasic

Commit e3e9bda38e6d ("s390/virtio_ccw: use DMA handle from DMA API") broke configuration change notifications for virtio-ccw by putting the DMA address of *indicatorp directly into ccw->cda disregarding the fact that if !!(vcdev->is_thinint) then the function virtio_ccw_register_adapter_ind() will overwrite that ccw->cda value with the address of the virtio_thinint_area so it can actually set up the adapter interrupts via CCW_CMD_SET_IND_ADAPTER. Thus we end up pointing to the wrong object for both CCW_CMD_SET_IND if setting up the adapter interrupts fails, and for CCW_CMD_SET_CONF_IND regardless whether it succeeds or fails.

To fix this, let us save away the dma address of *indicatorp in a local variable, and copy it to ccw->cda after the "vcdev->is_thinint" branch.

Quoted 8 lines
Fixes e3e9bda38e6d ("s390/virtio_ccw: use DMA handle from DMA API")

---
I know that checkpatch.pl complains about a missing 'Closes' tag. Unfortunately I don't have an appropriate URL at hand. @Sebastian, @Boqiao: do you have any suggetions?

Quoted 2 lines

?

Yep! That is a public bug tracker bug. Qualifies! @Vasily: Can you guys pick hat one up when picking the patch?

Quoted 6 lines

Anyway, I've tested the patch and it indeed fixes the problem with virtio-balloon and the link state for me:

Thanks!

maintainer
#3 lore raw
Quoted 27 lines · Halil Pasic
Quoted 24 lines · On Wed, 12 Jun 2024 16:04:15 +0200 Thomas Huth <[email protected]> wrote:
Quoted 22 lines · Halil Pasic

Commit e3e9bda38e6d ("s390/virtio_ccw: use DMA handle from DMA API") broke configuration change notifications for virtio-ccw by putting the DMA address of *indicatorp directly into ccw->cda disregarding the fact that if !!(vcdev->is_thinint) then the function virtio_ccw_register_adapter_ind() will overwrite that ccw->cda value with the address of the virtio_thinint_area so it can actually set up the adapter interrupts via CCW_CMD_SET_IND_ADAPTER. Thus we end up pointing to the wrong object for both CCW_CMD_SET_IND if setting up the adapter interrupts fails, and for CCW_CMD_SET_CONF_IND regardless whether it succeeds or fails.

To fix this, let us save away the dma address of *indicatorp in a local variable, and copy it to ccw->cda after the "vcdev->is_thinint" branch.

Fixes e3e9bda38e6d ("s390/virtio_ccw: use DMA handle from DMA API")

---
I know that checkpatch.pl complains about a missing 'Closes' tag. Unfortunately I don't have an appropriate URL at hand. @Sebastian, @Boqiao: do you have any suggetions?

Quoted 5 lines
Quoted 2 lines

?

Yep! That is a public bug tracker bug. Qualifies! @Vasily: Can you guys pick hat one up when picking the patch?

Sure, applied. Thanks!

maintainer Reviewed-by on v1 1/1
#4 lore raw
In reply to Halil Pasic
Quoted 23 lines · Halil Pasic

Commit e3e9bda38e6d ("s390/virtio_ccw: use DMA handle from DMA API") broke configuration change notifications for virtio-ccw by putting the
DMA address of *indicatorp directly into ccw->cda disregarding the fact
that if !!(vcdev->is_thinint) then the function virtio_ccw_register_adapter_ind() will overwrite that ccw->cda value with the address of the virtio_thinint_area so it can actually set up the adapter interrupts via CCW_CMD_SET_IND_ADAPTER.  Thus we end up pointing to the wrong object for both CCW_CMD_SET_IND if setting up the
adapter interrupts fails, and for CCW_CMD_SET_CONF_IND regardless whether it succeeds or fails.

To fix this, let us save away the dma address of *indicatorp in a local
variable, and copy it to ccw->cda after the "vcdev->is_thinint" branch.

Fixes e3e9bda38e6d ("s390/virtio_ccw: use DMA handle from DMA API")
Quoted 7 lines

---
I know that checkpatch.pl complains about a missing 'Closes' tag. Unfortunately I don't have an appropriate URL at hand. @Sebastian, @Boqiao: do you have any suggetions?
---
 drivers/s390/virtio/virtio_ccw.c | 4 +++-

 1 file changed, 3 insertions(+), 1 deletion(-)