Skip to contentSkip to conversation

nvme: change nvme_passthru_cmd64 to explicitly mark rsvd

PATCH Applied Charles Machalow 5 participants 10 messages Nov 5, 2019 – Nov 6, 2019

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 nvme: change nvme_passthru_cmd64 to explicitly mark rsvd

Changing nvme_passthru_cmd64 to add a field: rsvd2. This field is an explicit marker for the padding space added on certain platforms as a result of the enlargement of the result field from 32 bit to 64 bits in size.

---
 include/uapi/linux/nvme_ioctl.h | 1 +
 1 file changed, 1 insertion(+)
include/uapi/linux/nvme_ioctl.h +1 -0
index e168dc59e..d99b5a772 100644
@@ -63,6 +63,7 @@ struct nvme_passthru_cmd64 {
 	__u32	cdw14;
 	__u32	cdw15;
 	__u32	timeout_ms;
+	__u32   rsvd2;
 	__u64	result;
 };
 
Signature · 1 line
2.17.1
#1 lore raw
Quoted 12 lines · ----- On 5 Nov, 2019, at 07:15, Charles Machalow [email protected] wrote:

Changing nvme_passthru_cmd64 to add a field: rsvd2. This field is an explicit marker for the padding space added on certain platforms as a result of the enlargement of the result field from 32 bit to 64 bits in size.
---
include/uapi/linux/nvme_ioctl.h | 1 +

1 file changed, 1 insertion(+)
include/uapi/linux/nvme_ioctl.h +0 -0
index e168dc59e..d99b5a772 100644

include/uapi/linux/nvme_ioctl.h struct nvme_passthru_cmd64 { · line 69

Quoted 7 lines
include/uapi/linux/nvme_ioctl.h +1 -0
@@ -63,6 +63,7 @@ struct nvme_passthru_cmd64 {
 	__u32	cdw14;
 	__u32	cdw15;
 	__u32	timeout_ms;
+	__u32   rsvd2;
 	__u64	result;
 };
 

Looks good to me. However, please note that the new ioctl made it already to 5.3.8.

Regards,
Marta

#2 lore raw
Quoted 1 line · Marta Rybczynska

Looks good to me. However, please note that the new ioctl made it already to 5.3.8.

It wasn't in 5.3, but it seems like you are right and it somehow got picked for the stable releases.

Sasha, can you please revert 76d609da9ed1cc0dc780e2b539d7b827ce28f182 in 5.3-stable ASAP and make sure crap like backporting new ABIs that haven't seen a release yet is never ever going to happen again?

maintainer
#3 lore raw
Quoted 9 lines · Christoph Hellwig
Quoted 1 line · Marta Rybczynska

Looks good to me. However, please note that the new ioctl made it already to 5.3.8.

It wasn't in 5.3, but it seems like you are right and it somehow got picked for the stable releases.

Sasha, can you please revert 76d609da9ed1cc0dc780e2b539d7b827ce28f182 in 5.3-stable ASAP and make sure crap like backporting new ABIs that haven't seen a release yet is never ever going to happen again?

Sure, I'll revert it. I guess I wasn't expecting to see something like this in a -rc release. How did it make it into one if it's not a fix?

Signature · 2 lines
Thanks,
Sasha
maintainer
#4 lore raw
Quoted 7 lines · Sasha Levin
Quoted 6 lines · Christoph Hellwig
Quoted 1 line · Marta Rybczynska

Looks good to me. However, please note that the new ioctl made it already to 5.3.8.

It wasn't in 5.3, but it seems like you are right and it somehow got picked for the stable releases.

Quoted 6 lines
Quoted 3 lines

Sasha, can you please revert 76d609da9ed1cc0dc780e2b539d7b827ce28f182 in 5.3-stable ASAP and make sure crap like backporting new ABIs that haven't seen a release yet is never ever going to happen again?

Sure, I'll revert it. I guess I wasn't expecting to see something like this in a -rc release. How did it make it into one if it's not a fix?

It is a fix, but it wasn't marked as such in the original changelog. I've adjusted it for the pull request, currently staged here:

https://git.kernel.dk/cgit/linux-block/commit/?h=for-linus&id=0d6eeb1fd625272bd60d25f2d5e116cf582fc7dc

Still, a new ioctl seems pretty aggressive for stable bot, yeah?

#5 lore raw
In reply to Sasha Levin
Quoted 7 lines · Sasha Levin
Quoted 6 lines · Christoph Hellwig
Quoted 1 line · Marta Rybczynska

Looks good to me. However, please note that the new ioctl made it already to 5.3.8.

It wasn't in 5.3, but it seems like you are right and it somehow got picked for the stable releases.

Quoted 6 lines
Quoted 3 lines

Sasha, can you please revert 76d609da9ed1cc0dc780e2b539d7b827ce28f182 in 5.3-stable ASAP and make sure crap like backporting new ABIs that haven't seen a release yet is never ever going to happen again?

Sure, I'll revert it. I guess I wasn't expecting to see something like this in a -rc release. How did it make it into one if it's not a fix?

76d609da9ed1cc0dc780e2b539d7b827ce28f182 is a backport of 65e68edce0db433aa0c2b26d7dc14fbbbeb89fbb, which went into 5.4-rc2 and was not marked for stable. It might kinda bend the normal merge window rules a little, but I don't see how it could be considered something to backport.

maintainer
#7 lore raw
In reply to Charles Machalow
Quoted 3 lines · Charles Machalow

Changing nvme_passthru_cmd64 to add a field: rsvd2. This field is an explicit marker for the padding space added on certain platforms as a result of the enlargement of the result field from 32 bit to 64 bits in size.

Charles,
Could you reply with your "Signed-off-by" so I can apply this patch? Thanks.

#8 lore raw

Sorry, still new to this process.

- Charlie Scott Machalow

Quoted 9 lines · Keith Busch
Quoted 3 lines · Charles Machalow

Changing nvme_passthru_cmd64 to add a field: rsvd2. This field is an explicit marker for the padding space added on certain platforms as a result of the enlargement of the result field from 32 bit to 64 bits in size.

Charles,
Could you reply with your "Signed-off-by" so I can apply this patch? Thanks.

maintainer Applied
#9 lore raw
In reply to Charles Machalow

Thanks, applied for the next 5.4-rc pull with an updated changelog to indicate the "Fixes" commit since not having the padding does indeed break the compat ioctl.