| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <20230622083921.tc3wsl6jyluenmru@blmsp>
Date: Thu, 22 Jun 2023 10:39:21 +0200
From: Markus Schneider-Pargmann <msp@...libre.com>
To: AngeloGioacchino Del Regno
<angelogioacchino.delregno@...labora.com>
Cc: Rob Herring <robh+dt@...nel.org>,
Krzysztof Kozlowski <krzysztof.kozlowski+dt@...aro.org>,
Conor Dooley <conor+dt@...nel.org>,
Matthias Brugger <matthias.bgg@...il.com>,
Chun-Jie Chen <chun-jie.chen@...iatek.com>,
Tinghan Shen <tinghan.shen@...iatek.com>,
Fabien Parent <parent.f@...il.com>, devicetree@...r.kernel.org,
linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
linux-mediatek@...ts.infradead.org,
Alexandre Bailon <abailon@...libre.com>,
Fabien Parent <fparent@...libre.com>
Subject: Re: [PATCH v5 6/8] soc: mediatek: Add support for WAY_EN operations
On Mon, Jun 19, 2023 at 11:29:18AM +0200, AngeloGioacchino Del Regno wrote:
> Il 19/06/23 10:53, Markus Schneider-Pargmann ha scritto:
> > From: Alexandre Bailon <abailon@...libre.com>
> >
> > This updates the power domain to support WAY_EN operations. WAY_EN
> > operations on mt8365 are using a different component to check for the
> > acknowledgment, namely the infracfg-nao component. Also to enable a way
> > it the bit needs to be cleared while disabling a way needs a bit to be
> > set. To support these two operations two flags are added,
> > BUS_PROT_INVERTED and BUS_PROT_STA_COMPONENT_INFRA_NAO. Additionally
> > another regmap is created if the INFRA_NAO capability is set.
> >
> > This operation is required by the mt8365 for the MM power domain.
> >
> > Signed-off-by: Alexandre Bailon <abailon@...libre.com>
> > Signed-off-by: Fabien Parent <fparent@...libre.com>
> > Signed-off-by: Markus Schneider-Pargmann <msp@...libre.com>
> > ---
> > drivers/soc/mediatek/mtk-pm-domains.c | 39 +++++++++++++++++++++++----
> > drivers/soc/mediatek/mtk-pm-domains.h | 7 +++--
> > 2 files changed, 39 insertions(+), 7 deletions(-)
> >
> > diff --git a/drivers/soc/mediatek/mtk-pm-domains.c b/drivers/soc/mediatek/mtk-pm-domains.c
> > index 3cdf62c0b6bd..4659f0a0aa08 100644
> > --- a/drivers/soc/mediatek/mtk-pm-domains.c
> > +++ b/drivers/soc/mediatek/mtk-pm-domains.c
> > @@ -44,6 +44,7 @@ struct scpsys_domain {
> > struct clk_bulk_data *clks;
> > int num_subsys_clks;
> > struct clk_bulk_data *subsys_clks;
> > + struct regmap *infracfg_nao;
> > struct regmap *infracfg;
> > struct regmap *smi;
> > struct regulator *supply;
> > @@ -127,13 +128,26 @@ static struct regmap *scpsys_bus_protect_get_regmap(struct scpsys_domain *pd,
> > return pd->infracfg;
> > }
> > +static struct regmap *scpsys_bus_protect_get_sta_regmap(struct scpsys_domain *pd,
> > + const struct scpsys_bus_prot_data *bpd)
> > +{
> > + if (bpd->flags & BUS_PROT_STA_COMPONENT_INFRA_NAO)
> > + return pd->infracfg_nao;
> > + else
> > + return scpsys_bus_protect_get_regmap(pd, bpd);
> > +}
> > +
> > static int scpsys_bus_protect_clear(struct scpsys_domain *pd,
> > const struct scpsys_bus_prot_data *bpd)
> > {
> > + struct regmap *sta_regmap = scpsys_bus_protect_get_sta_regmap(pd, bpd);
> > struct regmap *regmap = scpsys_bus_protect_get_regmap(pd, bpd);
> > + u32 expected_ack;
> > u32 val;
> > u32 sta_mask = bpd->bus_prot_sta_mask;
> > + expected_ack = (bpd->flags & BUS_PROT_STA_COMPONENT_INFRA_NAO ? sta_mask : 0);
> > +
> > if (bpd->flags & BUS_PROT_REG_UPDATE)
> > regmap_clear_bits(regmap, bpd->bus_prot_clr, bpd->bus_prot_set_clr_mask);
> > else
> > @@ -142,14 +156,15 @@ static int scpsys_bus_protect_clear(struct scpsys_domain *pd,
> > if (bpd->flags & BUS_PROT_IGNORE_CLR_ACK)
> > return 0;
> > - return regmap_read_poll_timeout(regmap, bpd->bus_prot_sta,
> > - val, !(val & sta_mask),
> > + return regmap_read_poll_timeout(sta_regmap, bpd->bus_prot_sta,
> > + val, (val & sta_mask) == expected_ack,
> > MTK_POLL_DELAY_US, MTK_POLL_TIMEOUT);
> > }
> > static int scpsys_bus_protect_set(struct scpsys_domain *pd,
> > const struct scpsys_bus_prot_data *bpd)
> > {
> > + struct regmap *sta_regmap = scpsys_bus_protect_get_sta_regmap(pd, bpd);
> > struct regmap *regmap = scpsys_bus_protect_get_regmap(pd, bpd);
> > u32 val;
> > u32 sta_mask = bpd->bus_prot_sta_mask;
> > @@ -159,7 +174,7 @@ static int scpsys_bus_protect_set(struct scpsys_domain *pd,
> > else
> > regmap_write(regmap, bpd->bus_prot_set, bpd->bus_prot_set_clr_mask);
> > - return regmap_read_poll_timeout(regmap, bpd->bus_prot_sta,
> > + return regmap_read_poll_timeout(sta_regmap, bpd->bus_prot_sta,
> > val, (val & sta_mask) == sta_mask,
> > MTK_POLL_DELAY_US, MTK_POLL_TIMEOUT);
> > }
> > @@ -173,7 +188,10 @@ static int scpsys_bus_protect_enable(struct scpsys_domain *pd)
> > if (!bpd->bus_prot_set_clr_mask)
> > break;
> > - ret = scpsys_bus_protect_set(pd, bpd);
> > + if (bpd->flags & BUS_PROT_INVERTED)
> > + ret = scpsys_bus_protect_clear(pd, bpd);
> > + else
> > + ret = scpsys_bus_protect_set(pd, bpd);
> > if (ret)
> > return ret;
> > }
> > @@ -190,7 +208,10 @@ static int scpsys_bus_protect_disable(struct scpsys_domain *pd)
> > if (!bpd->bus_prot_set_clr_mask)
> > continue;
> > - ret = scpsys_bus_protect_clear(pd, bpd);
> > + if (bpd->flags & BUS_PROT_INVERTED)
> > + ret = scpsys_bus_protect_set(pd, bpd);
> > + else
> > + ret = scpsys_bus_protect_clear(pd, bpd);
> > if (ret)
> > return ret;
> > }
> > @@ -377,6 +398,14 @@ generic_pm_domain *scpsys_add_one_domain(struct scpsys *scpsys, struct device_no
> > return ERR_CAST(pd->smi);
> > }
> > + pd->infracfg_nao = syscon_regmap_lookup_by_phandle(node, "mediatek,infracfg-nao");
>
> If we don't expect infracfg-nao to be present, what's the point about trying to
> get a regmap handle and then failing only if we do expect it to be there?
>
> At this point you can just do...
>
> if (MTK_SCPD_CAPS(pd, MTK_SCPD_HAS_INFRA_NAO)) {
> pd->infracfg_nao = syscon_regmap_lookup_by_phandle(...);
> if (IS_ERR(....))
> return ....
> }
Yes! My code looks stupid. Thanks!
>
> > + if (IS_ERR(pd->infracfg_nao)) {
> > + if (MTK_SCPD_CAPS(pd, MTK_SCPD_HAS_INFRA_NAO))
> > + return ERR_CAST(pd->infracfg_nao);
> > +
> > + pd->infracfg_nao = NULL;
> > + }
> > +
> > num_clks = of_clk_get_parent_count(node);
> > if (num_clks > 0) {
> > /* Calculate number of subsys_clks */
> > diff --git a/drivers/soc/mediatek/mtk-pm-domains.h b/drivers/soc/mediatek/mtk-pm-domains.h
> > index 356788263db2..562d4e92ce16 100644
> > --- a/drivers/soc/mediatek/mtk-pm-domains.h
> > +++ b/drivers/soc/mediatek/mtk-pm-domains.h
> > @@ -11,6 +11,7 @@
> > /* can't set MTK_SCPD_KEEP_DEFAULT_OFF at the same time */
> > #define MTK_SCPD_ALWAYS_ON BIT(5)
> > #define MTK_SCPD_EXT_BUCK_ISO BIT(6)
> > +#define MTK_SCPD_HAS_INFRA_NAO BIT(7)
> > #define MTK_SCPD_CAPS(_scpd, _x) ((_scpd)->data->caps & (_x))
> > #define SPM_VDE_PWR_CON 0x0210
> > @@ -45,8 +46,10 @@
> > enum scpsys_bus_prot_flags {
> > BUS_PROT_REG_UPDATE = BIT(1),
> > BUS_PROT_IGNORE_CLR_ACK = BIT(2),
> > - BUS_PROT_COMPONENT_INFRA = BIT(3),
> > - BUS_PROT_COMPONENT_SMI = BIT(4),
> > + BUS_PROT_INVERTED = BIT(3),
>
> I get the reason why you're setting inverted as bit 3, but at that point you can
> just set BUS_PROT_COMPONENT_INFRA to bit 4 from the very beginning, instead of
> using bit 3 for that and then changing them all in a subsequent commit (this one).
Yes, I was torn between making the commits independent and avoiding this
move later on. I decided for the first. If you prefer I can set it to
the correct bits right from the beginning.
Best,
Markus
Powered by blists - more mailing lists