Re: [PATCH v2 net-next] ixgbe: Fix endian handling for ACI descriptor registers

From: Simon Horman
Date: Fri Jan 17 2025 - 13:10:06 EST


On Fri, Jan 17, 2025 at 11:01:22AM +0100, Przemek Kitszel wrote:
> On 1/16/25 17:21, Simon Horman wrote:
> > On Wed, Jan 15, 2025 at 09:11:17AM +0530, Dheeraj Reddy Jonnalagadda wrote:
> > > The ixgbe driver was missing proper endian conversion for ACI descriptor
> > > register operations. Add the necessary conversions when reading and
> > > writing to the registers.
> > >
> > > Fixes: 46761fd52a88 ("ixgbe: Add support for E610 FW Admin Command Interface")
> > > Closes: https://scan7.scan.coverity.com/#/project-view/52337/11354?selectedIssue=1602757
> > > Signed-off-by: Dheeraj Reddy Jonnalagadda <dheeraj.linuxdev@xxxxxxxxx>
> >
> > Hi Dheeraj,
> >
> > It seems that Sparse is not very happy about __le32 values appearing
> > where u32 ones are expected. I wonder if something like what is below
> > (compile tested only!) would both address the problem at hand and
> > keep Sparse happy (even if negting much of it's usefulness by using casts).
> >
> > diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_common.h b/drivers/net/ethernet/intel/ixgbe/ixgbe_common.h
> > index 6639069ad528..8b3787837128 100644
> > --- a/drivers/net/ethernet/intel/ixgbe/ixgbe_common.h
> > +++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_common.h
> > @@ -150,6 +150,9 @@ static inline void ixgbe_write_reg(struct ixgbe_hw *hw, u32 reg, u32 value)
> > }
> > #define IXGBE_WRITE_REG(a, reg, value) ixgbe_write_reg((a), (reg), (value))
>
> Simon,
>
> As all ixgbe registers are LE, it would be beneficial to change
> ixgbe_write_reg(), as @value should be __le32, (perhaps @reg too).
> Similar for 64b.

Understood, sounds good to me.

> This clearly would not be a "fix" material, as all call sites should be
> examined to check if they conform.

Sure, that also seems reasonable.
But do you also think a more minimal fix is in order?

>
> > +#define IXGBE_WRITE_REG_LE32(a, reg, value) \
> > + ixgbe_write_reg((a), (reg), (u32 __force)cpu_to_le32(value))
> > +
>