From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from down.free-electrons.com ([37.187.137.238] helo=mail.free-electrons.com) by bombadil.infradead.org with esmtp (Exim 4.80.1 #2 (Red Hat Linux)) id 1bEcah-0004bV-CF for linux-mtd@lists.infradead.org; Sun, 19 Jun 2016 13:16:44 +0000 Date: Sun, 19 Jun 2016 15:16:19 +0200 From: Boris Brezillon To: icenowy@aosc.xyz Cc: "maxime.ripard@free-electrons.com" , "wens@csie.org" , "robh+dt@kernel.org" , "richard@nod.at" , "dwmw2@infradead.org" , "computersforpeace@gmail.com" , "devicetree@vger.kernel.org" , "linux-kernel@vger.kernel.org" , "linux-mtd@lists.infradead.org" , "linux-sunxi@googlegroups.com" , Philipp Zabel Subject: Re: [PATCH 2/2] mtd: nand: sunxi: add reset line support Message-ID: <20160619151619.4edc64f6@bbrezillon> In-Reply-To: <2748041466341860@web7m.yandex.ru> References: <20160619113739.30362-1-icenowy@aosc.xyz> <20160619113739.30362-2-icenowy@aosc.xyz> <20160619140652.07ab03c9@bbrezillon> <2686901466340069@web27g.yandex.ru> <20160619145303.0bdf59ea@bbrezillon> <2748041466341860@web7m.yandex.ru> MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable List-Id: Linux MTD discussion mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Sun, 19 Jun 2016 21:11:00 +0800 icenowy@aosc.xyz wrote: > Then I will soon make the v2 patch set, with the error detection part fix= ed. >=20 > But why does sunxi-mmc check only EPROBE_DEFER? I guess someone had a probe-dependency problem and fixed this case but ignored all the possible errors. But there may be other reasons for devm_reset_control_get_optional() to fail. The only one that is really reflecting that the reset line is not defined in the DT is -ENOENT. >=20 > 19.06.2016, 20:53, "Boris Brezillon" : > > On Sun, 19 Jun 2016 20:41:09 +0800 > > icenowy@aosc.xyz wrote: > > =20 > >> =C2=A0To be honest, I copied them from sunxi-mmc.c. > >> > >> =C2=A0What function should be chosen better? =20 > > > > You did the right thing (except for the error detection part). My > > question was addressed to Philipp (the reset subsystem maintainer). > > =20 > >> =C2=A019.06.2016, 20:06, "Boris Brezillon" : =20 > >> =C2=A0> +Philipp > >> =C2=A0> > >> =C2=A0> On Sun, 19 Jun 2016 19:37:39 +0800 > >> =C2=A0> Icenowy Zheng wrote: > >> =C2=A0> =20 > >> =C2=A0>> =C2=A0The NAND controller on some sun8i chips needs its reset= line to be deasserted > >> =C2=A0>> =C2=A0before they can enter working state. This commit added = the reset line process > >> =C2=A0>> =C2=A0to the driver. > >> =C2=A0>> > >> =C2=A0>> =C2=A0Signed-off-by: Icenowy Zheng > >> =C2=A0>> =C2=A0--- > >> =C2=A0>> =C2=A0=C2=A0drivers/mtd/nand/sunxi_nand.c | 14 ++++++++++++++ > >> =C2=A0>> =C2=A0=C2=A01 file changed, 14 insertions(+) > >> =C2=A0>> > >> =C2=A0>> =C2=A0diff --git a/drivers/mtd/nand/sunxi_nand.c b/drivers/mt= d/nand/sunxi_nand.c > >> =C2=A0>> =C2=A0index a83a690..1502748 100644 > >> =C2=A0>> =C2=A0--- a/drivers/mtd/nand/sunxi_nand.c > >> =C2=A0>> =C2=A0+++ b/drivers/mtd/nand/sunxi_nand.c > >> =C2=A0>> =C2=A0@@ -39,6 +39,7 @@ > >> =C2=A0>> =C2=A0=C2=A0#include > >> =C2=A0>> =C2=A0=C2=A0#include > >> =C2=A0>> =C2=A0=C2=A0#include > >> =C2=A0>> =C2=A0+#include > >> =C2=A0>> > >> =C2=A0>> =C2=A0=C2=A0#define NFC_REG_CTL 0x0000 > >> =C2=A0>> =C2=A0=C2=A0#define NFC_REG_ST 0x0004 > >> =C2=A0>> =C2=A0@@ -269,6 +270,7 @@ struct sunxi_nfc { > >> =C2=A0>> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0v= oid __iomem *regs; > >> =C2=A0>> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0s= truct clk *ahb_clk; > >> =C2=A0>> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0s= truct clk *mod_clk; > >> =C2=A0>> =C2=A0+ struct reset_control *reset; > >> =C2=A0>> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0u= nsigned long assigned_cs; > >> =C2=A0>> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0u= nsigned long clk_rate; > >> =C2=A0>> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0s= truct list_head chips; > >> =C2=A0>> =C2=A0@@ -1871,6 +1873,18 @@ static int sunxi_nfc_probe(struc= t platform_device *pdev) > >> =C2=A0>> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0i= f (ret) > >> =C2=A0>> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0= =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0goto out_ahb_clk_unprepare; > >> =C2=A0>> > >> =C2=A0>> =C2=A0+ nfc->reset =3D devm_reset_control_get_optional(dev, "= ahb"); > >> =C2=A0>> =C2=A0+ if (PTR_ERR(nfc->reset) =3D=3D -EPROBE_DEFER) > >> =C2=A0>> =C2=A0+ return PTR_ERR(nfc->reset); =20 > >> =C2=A0> > >> =C2=A0> Actually you should test for !=3D -ENOENT, because all error c= odes except > >> =C2=A0> this one should stop the ->probe(). > >> =C2=A0> > >> =C2=A0> BTW, this devm_reset_control_get_optional() is really weird. W= hile most > >> =C2=A0> _optional() methods return NULL when the element is not define= d in the > >> =C2=A0> DT, this one returns -ENOTENT, which makes it impossible to > >> =C2=A0> differentiate a real error from a undefined reset line (which = is a > >> =C2=A0> valid case for _optional()). > >> =C2=A0> > >> =C2=A0> Philipp, is there a good reason for doing that? > >> =C2=A0> =20 > >> =C2=A0>> =C2=A0+ > >> =C2=A0>> =C2=A0+ if (!IS_ERR(nfc->reset)) { > >> =C2=A0>> =C2=A0+ ret =3D reset_control_deassert(nfc->reset); > >> =C2=A0>> =C2=A0+ if (ret) { > >> =C2=A0>> =C2=A0+ dev_err(dev, "reset err %d\n", ret); > >> =C2=A0>> =C2=A0+ goto out_mod_clk_unprepare; > >> =C2=A0>> =C2=A0+ } > >> =C2=A0>> =C2=A0+ } > >> =C2=A0>> =C2=A0+ =20