| 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: <20200327215156.GB211617@minitux>
Date: Fri, 27 Mar 2020 14:51:56 -0700
From: Bjorn Andersson <bjorn.andersson@...aro.org>
To: ansuelsmth@...il.com
Cc: 'Andy Gross' <agross@...nel.org>,
'Ajay Kishore' <akisho@...eaurora.org>,
'Linus Walleij' <linus.walleij@...aro.org>,
linux-arm-msm@...r.kernel.org, linux-gpio@...r.kernel.org,
linux-kernel@...r.kernel.org
Subject: Re: R: [PATCH v2] pinctrl: qcom: use scm_call to route GPIO irq to
Apps
On Fri 27 Mar 13:58 PDT 2020, ansuelsmth@...il.com wrote:
> > On Thu 26 Mar 10:35 PDT 2020, Ansuel Smith wrote:
> >
> > > From: Ajay Kishore <akisho@...eaurora.org>
> > >
> > > For IPQ806x targets, TZ protects the registers that are used to
> > > configure the routing of interrupts to a target processor.
> > > To resolve this, this patch uses scm call to route GPIO interrupts
> > > to application processor. Also the scm call interface is changed.
> > >
> > > Signed-off-by: Ajay Kishore <akisho@...eaurora.org>
> > > Signed-off-by: Ansuel Smith <ansuelsmth@...il.com>
> >
> > Thanks for respinning this Ansuel, just some minor things below.
> >
> > > ---
> > > v2:
> > > * Move static varibale in msm_pinctrl struct
> > > * Revert '4b024225c4a8 ("pinctrl: use
> > devm_platform_ioremap_resource() to simplify code")'
> > > to get base_reg addr
> > >
> > > drivers/pinctrl/qcom/pinctrl-msm.c | 37
> > ++++++++++++++++++++++++++----
> > > 1 file changed, 32 insertions(+), 5 deletions(-)
> > >
> > > diff --git a/drivers/pinctrl/qcom/pinctrl-msm.c
> > b/drivers/pinctrl/qcom/pinctrl-msm.c
> > > index 9a8daa256a32..9627ebd41ff9 100644
> > > --- a/drivers/pinctrl/qcom/pinctrl-msm.c
> > > +++ b/drivers/pinctrl/qcom/pinctrl-msm.c
> > > @@ -22,6 +22,8 @@
> > > #include <linux/reboot.h>
> > > #include <linux/pm.h>
> > > #include <linux/log2.h>
> > > +#include <linux/qcom_scm.h>
> > > +#include <linux/io.h>
> > >
> > > #include <linux/soc/qcom/irq.h>
> > >
> > > @@ -60,6 +62,9 @@ struct msm_pinctrl {
> > > struct irq_chip irq_chip;
> > > int irq;
> > >
> > > + int route_to_apps;
> >
> > We "always" route our interrupts to apps, so please rename this to
> > intr_target_use_scm. And using "bool" instead would make it clear that
> > this is a binary flag.
> >
>
> Will fix in v3
>
> > > + u32 base_reg;
> >
> > Even though I think it's fine that you only fill in the first entry,
> > please make this phys_base[MAX_NR_GPIO]; next to regs.
> >
>
> If I understand correctly, I should replace base_reg with
> phys_base[MAX_NR_GPIO]
> I think we should use MAX_NR_TILES instead of MAX_NR_GPIO
Yes of course, didn't pay attention to my autocomplete!
> Wouldn't this be a waste of memory since only the base_reg will be
> needed? I don't think there will be a soc with multiple tiles and the
> need to add code in the loop function. (Also sorry for off-topic...
> since we have soc_data->ntiles why don't we use kmem allocation
> instead of statically use half empty array?)
>
While I hope you're right, I think it looks better if it mimics the
overall design - at the expense of wasting 12 bytes.
> > > +
> > > raw_spinlock_t lock;
> > >
> > > DECLARE_BITMAP(dual_edge_irqs, MAX_NR_GPIO);
> > > @@ -883,10 +888,27 @@ static int msm_gpio_irq_set_type(struct
> > irq_data *d, unsigned int type)
> > > clear_bit(d->hwirq, pctrl->dual_edge_irqs);
> > >
> > > /* Route interrupts to application cpu */
> > > - val = msm_readl_intr_target(pctrl, g);
> > > - val &= ~(7 << g->intr_target_bit);
> > > - val |= g->intr_target_kpss_val << g->intr_target_bit;
> > > - msm_writel_intr_target(val, pctrl, g);
> > > + if (pctrl->route_to_apps && pctrl->base_reg) {
> >
> > I meant that you could fill out "base_reg" only if it's supposed to be
> > used, but looking at your patch I think it's nicer to have a separate
> > flag - so you can omit the check for base_reg != NULL here.
> >
>
> Since we return error if devm_ioremap_resource fails, it is good to only
> check
> the route_to_apps as base_reg will be set for sure?
>
Right.
My initial suggestion was that base != NULL would have the meaning of
intr_target_use_scm. But what I meant here is that after looking at your
patch I think one flag and a separate base is better.
Thanks,
Bjorn
> > > + u32 addr = pctrl->base_reg + g->intr_target_reg;
> > > + int ret;
> > > +
> > > + qcom_scm_io_readl(addr, &val);
> > > +
> > > + val &= ~(7 << g->intr_target_bit);
> > > + val |= g->intr_target_kpss_val << g->intr_target_bit;
> > > +
> > > + ret = qcom_scm_io_writel(addr, val);
> > > + if (ret)
> > > + dev_err(pctrl->dev,
> > > + "Failed routing %lu interrupt to Apps proc",
> > > + d->hwirq);
> > > + }
> > > + } else {
> > > + val = msm_readl_intr_target(pctrl, g);
> > > + val &= ~(7 << g->intr_target_bit);
> > > + val |= g->intr_target_kpss_val << g->intr_target_bit;
> > > + msm_writel_intr_target(val, pctrl, g);
> > > + }
> > >
> > > /* Update configuration for gpio.
> > > * RAW_STATUS_EN is left on for all gpio irqs. Due to the
> > > @@ -1241,6 +1263,8 @@ int msm_pinctrl_probe(struct platform_device
> > *pdev,
> > > pctrl->dev = &pdev->dev;
> > > pctrl->soc = soc_data;
> > > pctrl->chip = msm_gpio_template;
> > > + pctrl->route_to_apps = of_device_is_compatible(pctrl->dev-
> > >of_node,
> > > + "qcom,ipq8064-pinctrl");
> > >
> > > raw_spin_lock_init(&pctrl->lock);
> > >
> > > @@ -1253,9 +1277,12 @@ int msm_pinctrl_probe(struct
> > platform_device *pdev,
> > > return PTR_ERR(pctrl->regs[i]);
> > > }
> > > } else {
> > > - pctrl->regs[0] = devm_platform_ioremap_resource(pdev, 0);
> > > + res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> > > + pctrl->regs[0] = devm_ioremap_resource(&pdev->dev, res);
> > > if (IS_ERR(pctrl->regs[0]))
> > > return PTR_ERR(pctrl->regs[0]);
> > > + else
> >
> > No need for the else, as the "positive" case is a return.
> >
>
> Will fix in v3
>
> > Thank you,
> > Bjorn
> >
> > > + pctrl->base_reg = res->start;
> > > }
> > >
> > > msm_pinctrl_setup_pm_reset(pctrl);
> > > --
> > > 2.25.1
> > >
>
Powered by blists - more mailing lists