On Mon, Feb 26, 2018 at 03:02:21PM +0200, Leon Romanovsky wrote:
> From: Yishai Hadas <yishaih@xxxxxxxxxxxx>
>
> This patch explicitly aligns 'mlx4_uverbs_ex_query_device_resp' layout
> to u64 and use the correct response's length when copying the data to
> the user area.
>
> The uAPI should be u64 aligned, currently on 64 bit systems this is done
> implicitly as of the next u64 field in the rss_caps structure.
>
> Legacy user space 32 bit system application in any case is broken with
> 64 bit system kernel before this change.
>
> A matching change will be done in the user area for future use also on
> 32 bit systems.
>
> In addition, we move the structure to the uAPI file where it's expected
> to be from day one.
>
> Fixes: 09d208b258a2 ("IB/mlx4: Add report for RSS capabilities by vendor channel")
> Reviewed-by: Mark Bloch <markb@xxxxxxxxxxxx>
> Signed-off-by: Yishai Hadas <yishaih@xxxxxxxxxxxx>
> Signed-off-by: Leon Romanovsky <leon@xxxxxxxxxx>
> drivers/infiniband/hw/mlx4/main.c | 5 +++--
> drivers/infiniband/hw/mlx4/mlx4_ib.h | 14 --------------
> include/uapi/rdma/mlx4-abi.h | 16 ++++++++++++++++
> 3 files changed, 19 insertions(+), 16 deletions(-)
So I am unhappy that we keep finding these 32/64 compat bugs. I sent a
RFC patch for rxe as well.
Any change like this breaks any previously working 32/32
configuration.
So what is the position here?
The 32/32 API is 'not used' but the 32/64 is being used, so we break
one to fix the other?
How do we justify this to Linus?
Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
