RE: [patch v11 - re-ordered 03/12] FIXME platform/mellanox: Remove redundant dev_err messages on device_create

From: Vadim Pasternak
Date: Wed Jan 31 2018 - 13:47:35 EST




> -----Original Message-----
> From: Darren Hart [mailto:dvhart@xxxxxxxxxxxxx]
> Sent: Wednesday, January 31, 2018 8:41 PM
> To: Vadim Pasternak <vadimp@xxxxxxxxxxxx>
> Cc: andy.shevchenko@xxxxxxxxx; gregkh@xxxxxxxxxxxxxxxxxxx; linux-
> kernel@xxxxxxxxxxxxxxx; platform-driver-x86@xxxxxxxxxxxxxxx; jiri@xxxxxxxxxxx
> Subject: Re: [patch v11 - re-ordered 03/12] FIXME platform/mellanox: Remove
> redundant dev_err messages on device_create
>
> On Fri, Jan 26, 2018 at 03:41:17PM -0800, Darren Hart wrote:
> > On Fri, Jan 26, 2018 at 04:59:32PM +0000, Vadim Pasternak wrote:
> > >
> > >
> > > > -----Original Message-----
> > > > From: Darren Hart [mailto:dvhart@xxxxxxxxxxxxx]
> > > > Sent: Thursday, January 25, 2018 11:47 PM
> > > > To: Vadim Pasternak <vadimp@xxxxxxxxxxxx>
> > > > Cc: andy.shevchenko@xxxxxxxxx; gregkh@xxxxxxxxxxxxxxxxxxx; linux-
> > > > kernel@xxxxxxxxxxxxxxx; platform-driver-x86@xxxxxxxxxxxxxxx;
> > > > jiri@xxxxxxxxxxx
> > > > Subject: Re: [patch v11 - re-ordered 03/12] FIXME
> > > > platform/mellanox: Remove redundant dev_err messages on
> > > > device_create
> > > >
> > > > On Wed, Jan 24, 2018 at 08:33:37PM +0000, Vadim Pasternak wrote:
> > > > > The upper layer will manage and report errors returned by the
> > > > > create and destroy functions.
> > > > >
> > > > > FIXME: but it doesn't - we don't even check
> > > >
> > > > I think we agreed to just drop this patch entirely, since this
> > > > error is not checked elsewhere? Keeping the dev_err messages?
> > > >
> > >
> > > Yes, we'll keep it.
> > > We just had conflict in patch order and I wanted to fix it, staying
> > > as much as possible aligned with the top of mellanox v10 branch.
> > >
> > > However currently we still have a problem in mellanox v11 after
> > > patch
> > > platform/mellanox: mlxreg-hotplug: Modify to use a regmap interface
> > > It dropped parameter device in mlxreg_hotplug_device_create, which
> > > is Used inside the routine.
> > >
> > > I'll send a separate patch to fix it from top of v11. In other case
> > > it'll be necessary to fix Modify to use a regmap interface and
> > > re-send all from 5-th patch in series.
> > >
> > > Also we should drop the patch:
> > > platform/x86: mlx-platform: Simplify IO access to regmap context It
> > > introduced the issue reported by sparse.
> > >
> > > I am resending two last patches from review-dvhart-mellanox-v11
> > > branch and also patch with fix for mlxreg_hotplug_device_create, as:
> > > 0010-platform-x86-mlx-platform-Extend-register-map-config.patch
> > > 0011-platform-mellanox-mlxreg-hotplug-Add-check-for-negat.patch
> > > 0012-platform-mellanox-mlxreg-hotplug-Fix-issue-of-commit.patch
> > >
> > > Thank you very much for your help,
> >
> > OK, please see review-dvhart-mellanox-v12 on infradead or github. I've
> > verified this passes a clean C=1 (sparse) build at each patch. I
> > believe this is in a state that can be moved to testing. Please confirm.
> >
> > If you still want to move the dev_err messages out of the create
> > function, we can do that separately. Let's get this merged so we can move
> forward.
>
> I'm pushing v12 into testing, shortly to for-next, and then on to Linus. Please
> speak up if there are any objections.
>

Hi Darren,

Thank you.
I'd like to send one small extra patch for this series - I missed
Removing hotplug driver in case of error here:
/* Sync registers with hardware. */
regcache_mark_dirty(mlxplat_hotplug->regmap);
err = regcache_sync(mlxplat_hotplug->regmap);
if (err)
goto fail_platform_mux_register;

I'll do it today.

Thanks,
Vadim.

> --
> Darren Hart
> VMware Open Source Technology Center