Re: [PATCH] crypto: qat - remove redundant condition accel_dev->is_vf

From: Giovanni Cabiddu
Date: Fri Oct 25 2019 - 05:50:44 EST


On Fri, Oct 25, 2019 at 10:45:31AM +0100, Yunfeng Ye wrote:
> On 2019/10/25 17:33, Ard Biesheuvel wrote:
> > On Fri, 25 Oct 2019 at 09:24, Yunfeng Ye <yeyunfeng@xxxxxxxxxx> wrote:
> >>
> >> Warning is found by the code analysis tool:
> >> "Redundant condition: accel_dev->is_vf"
> >>
> >> So remove the redundant condition accel_dev->is_vf.
> >>
> >> Signed-off-by: Yunfeng Ye <yeyunfeng@xxxxxxxxxx>
> >> ---
> >> drivers/crypto/qat/qat_common/adf_dev_mgr.c | 4 ++--
> >> 1 file changed, 2 insertions(+), 2 deletions(-)
> >>
> >> diff --git a/drivers/crypto/qat/qat_common/adf_dev_mgr.c b/drivers/crypto/qat/qat_common/adf_dev_mgr.c
> >> index 2d06409bd3c4..b54b8850fe20 100644
> >> --- a/drivers/crypto/qat/qat_common/adf_dev_mgr.c
> >> +++ b/drivers/crypto/qat/qat_common/adf_dev_mgr.c
> >> @@ -196,7 +196,7 @@ int adf_devmgr_add_dev(struct adf_accel_dev *accel_dev,
> >> atomic_set(&accel_dev->ref_count, 0);
> >>
> >> /* PF on host or VF on guest */
> >> - if (!accel_dev->is_vf || (accel_dev->is_vf && !pf)) {
> >> + if (!accel_dev->is_vf || !pf) {
> >
> > I disagree with this change. There is no bug here, and the way the
> > condition is formulated self-documents the code, i.e.,
> >
> > IF NOT is_vf
> > OR (is_vf BUT NOT pf)
> >
> > Using an automated tool to reduce every boolean expression to its
> > minimal representation doesn't seem that useful to me, since the
> > compiler is perfectly capable of doing that when generating the object
> > code.
> >
> ok, thanks, this modify just fix warning, and make code simple.
This change simplifies the code but it makes it less readable.
I'd prefer to leave it as it was.

Regards,

--
Giovanni