Dear Greg,
Thanks for your reply and sorry for my mistake.
Looks to me like the issue is the commit hash which should be 12 chars.
Does that mean I need to fix and resend the whole thing to everyone?
Best regards,
Laurent
>
-----------------------------
Eaton Industries Manufacturing GmbH ~ Registered place of business: Route de la
Longeraie 7, 1110, Morges, Switzerland
-----------------------------
-----Original Message-----
> From: gre...@linuxfoundation.org <gre...@linuxfoundation.org>
> Sent: Wednesday, April 29, 2020 11:40 AM
> To: Badel, Laurent <laurentba...@eaton.com>
> Cc: fugang.d...@nxp.com; netdev@vger.kernel.org; and...@lunn.ch;
> f.faine...@gmail.com; hkallwe...@gmail.com; li...@armlinux.org.uk;
> richard.leit...@skidata.com; da...@davemloft.net;
> alexander.le...@microsoft.com; Quette, Arnaud
> <arnaudque...@eaton.com>
> Subject: [EXTERNAL] Re: [PATCH 1/2] Revert commit
> 1b0a83ac04e383e3bed21332962b90710fcf2828
>
> On Wed, Apr 29, 2020 at 09:03:32AM +0000, Badel, Laurent wrote:
> > Description: This patch reverts commit 1b0a83ac04e3
> > ("net: fec: add phy_reset_after_clk_enable() support") which produces
> > undesirable behavior when PHY interrupts are enabled.
> >
> > Rationale: the SMSC LAN8720 (and possibly other chips) is known to
> > require a reset after the external clock is enabled. Calls to
> > phy_reset_after_clk_enable() in fec_main.c have been introduced in
> > commit 1b0a83ac04e3 ("net: fec: add phy_reset_after_clk_enable()
> > support") to handle the chip reset after enabling the clock.
> > However, this breaks when interrupts are enabled because the reset
> > reverts the configuration of the PHY interrupt mask to default (in
> > addition it also reverts the "energy detect" mode setting).
> > As a result the driver does not receive the link status change and
> > other notifications resulting in loss of connectivity.
> >
> > Proposed solution: revert commit 1b0a83ac04e3 and bring the reset
> > before the PHY configuration by adding it to phy_init_hw() [phy_device.c].
> >
> > Test results: using an iMX28-EVK-based board, these 2 patches
> > successfully restore network interface functionality when interrupts are
> enabled.
> > Tested using both linux-5.4.23 and latest mainline (5.6.0) kernels.
> >
> > Fixes: 1b0a83ac04e383e3bed21332962b90710fcf2828 ("net: fec: add
> > phy_reset_after_clk_enable() support")
>
> Please read Documentation/process/submitting-patches.rst and the section
> "2) Describe your changes" at the end, it says how to do lines like this
> "properly".
>
> thanks,
>
> greg k-h
