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
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <343c2bc4-46b5-443b-bb3b-ed4a45ecce76@kernel.org>
Date: Fri, 30 Aug 2024 01:45:41 +0200
From: Konrad Dybcio <konradybcio@...nel.org>
To: Unnathi Chalicheemala <quic_uchalich@...cinc.com>,
 Bjorn Andersson <andersson@...nel.org>,
 Konrad Dybcio <konradybcio@...nel.org>
Cc: linux-arm-msm@...r.kernel.org, linux-kernel@...r.kernel.org,
 kernel@...cinc.com
Subject: Re: [PATCH v2 1/2] firmware: qcom_scm: Add API to get waitqueue IRQ
 info

On 30.08.2024 12:15 AM, Unnathi Chalicheemala wrote:
> Bootloader and firmware for SM8650 and older chipsets expect node
> name as "qcom_scm". However, DeviceTree uses node name "scm" and this
> mismatch prevents firmware from correctly identifying waitqueue IRQ
> information. Waitqueue IRQ is used for signaling between secure and
> non-secure worlds.
> 
> To resolve this, introduce qcom_scm_get_waitq_irq() that'll get the
> hardware irq number to be used from firmware instead of relying on data
> provided by devicetree, thereby bypassing the DeviceTree node name
> mismatch.
> 
> This hardware irq number is converted to a linux irq number using newly
> defined fill_irq_fwspec_params(). This linux irq number is then supplied to
> the threaded_irq call.
> 
> Signed-off-by: Unnathi Chalicheemala <quic_uchalich@...cinc.com>
> ---
>  drivers/firmware/qcom/qcom_scm.c | 59 +++++++++++++++++++++++++++++++-
>  drivers/firmware/qcom/qcom_scm.h |  1 +
>  2 files changed, 59 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/firmware/qcom/qcom_scm.c b/drivers/firmware/qcom/qcom_scm.c
> index 00c379a3cceb..ed51fbb1c065 100644
> --- a/drivers/firmware/qcom/qcom_scm.c
> +++ b/drivers/firmware/qcom/qcom_scm.c
> @@ -32,6 +32,14 @@
>  #include "qcom_scm.h"
>  #include "qcom_tzmem.h"
>  
> +#define GIC_SPI_BASE            32
> +#define GIC_MAX_SPI             987  // 1019 - 32
> +#define GIC_ESPI_BASE           4096
> +#define GIC_MAX_ESPI            1024 // 5120 - 4096

Are these going to remain constant on different implementations of the
interrupt controller across different SoCs that use this? Are these
mandated anywhere in the arm spec and/or present across the tree with
parts touching gicv3?

Also, the subtraction comments take some guesswork.. perhaps something like
0..31 etc. would be easier.

The MAX_(E)SPI macros could also just have the hwirq number to make the
if-conditions below simpler

> +
> +#define GIC_IRQ_TYPE_SPI        0
> +#define GIC_IRQ_TYPE_ESPI       2

We can definitely use dt-bindings for this

> +
>  static bool download_mode = IS_ENABLED(CONFIG_QCOM_SCM_DOWNLOAD_MODE_DEFAULT);
>  module_param(download_mode, bool, 0);
>  
> @@ -1819,6 +1827,55 @@ bool qcom_scm_is_available(void)
>  }
>  EXPORT_SYMBOL_GPL(qcom_scm_is_available);
>  
> +static int qcom_scm_fill_irq_fwspec_params(struct irq_fwspec *fwspec, u32 virq)
> +{
> +	if (WARN(virq < GIC_SPI_BASE, "Unexpected virq: %d\n", virq)) {
> +		return -ENXIO;
> +	} else if (virq <= (GIC_SPI_BASE + GIC_MAX_SPI)) {
> +		fwspec->param_count = 3;
> +		fwspec->param[0] = GIC_IRQ_TYPE_SPI;
> +		fwspec->param[1] = virq - GIC_SPI_BASE;
> +		fwspec->param[2] = IRQ_TYPE_EDGE_RISING;
> +	} else if (WARN(virq < GIC_ESPI_BASE, "Unexpected virq: %d\n", virq)) {
> +		return -ENXIO;
> +	} else if (virq < (GIC_ESPI_BASE + GIC_MAX_ESPI)) {
> +		fwspec->param_count = 3;
> +		fwspec->param[0] = GIC_IRQ_TYPE_ESPI;
> +		fwspec->param[1] = virq - GIC_ESPI_BASE;
> +		fwspec->param[2] = IRQ_TYPE_EDGE_RISING;
> +	} else {
> +		WARN(1, "Unexpected virq: %d\n", virq);
> +		return -ENXIO;
> +	}
> +	return 0;

This could use some prettifying (incl the previous comment):

if (GIC_SPI_BASE <= virq && virq <= GIC_SPI_MAX) {
	fwspec->param[0] = GIC_IRQ_TYPE_SPI;
	fwspec->param[1] = virq - GIC_SPI_BASE;
} else if (GIC_ESPI_BASE <= virq && virq <= GIC_ESPI_MAX) {
	fwspec->param[0] = GIC_IRQ_TYPE_ESPI;
	fwspec->param[1] = virq - GIC_ESPI_BASE;
} else {
	WARN(1, "Unexpected virq"...
	return -ENXIO;
}

fwspec->param[2] = IRQ_TYPE_EDGE_RISING;
fwspec->param_count = 3;

is much easier to follow along in my opinion

> +}
> +
> +static int qcom_scm_get_waitq_irq(void)
> +{
> +	int ret;
> +	u32 hwirq;
> +	struct qcom_scm_desc desc = {
> +		.svc = QCOM_SCM_SVC_WAITQ,
> +		.cmd = QCOM_SCM_WAITQ_GET_INFO,
> +		.owner = ARM_SMCCC_OWNER_SIP
> +	};
> +	struct qcom_scm_res res;
> +	struct irq_fwspec fwspec;
> +
> +	ret = qcom_scm_call_atomic(__scm->dev, &desc, &res);
> +	if (ret)
> +		return ret;
> +
> +	fwspec.fwnode = of_node_to_fwnode(__scm->dev->of_node);
> +	hwirq = res.result[1] & 0xffff;

GENMASK(15, 0)

> +	ret = qcom_scm_fill_irq_fwspec_params(&fwspec, hwirq);
> +	if (ret)
> +		return ret;
> +	ret = irq_create_fwspec_mapping(&fwspec);
> +
> +	return ret;
> +}
> +
>  static int qcom_scm_assert_valid_wq_ctx(u32 wq_ctx)
>  {
>  	/* FW currently only supports a single wq_ctx (zero).
> @@ -1936,7 +1993,7 @@ static int qcom_scm_probe(struct platform_device *pdev)
>  	/* Let all above stores be available after this */
>  	smp_store_release(&__scm, scm);
>  
> -	irq = platform_get_irq_optional(pdev, 0);
> +	irq = qcom_scm_get_waitq_irq();
>  	if (irq < 0) {
>  		if (irq != -ENXIO)

Is this smc call left unimplemented on !auto platforms? If it's not
(or it spits out bogus data), we're going to get a WARN splat in the
log..

Additionally, this mechanism ties the trustzone and hypervisor together..
Why isn't this done in gunyah which abstracts these resources? A hypercall
sounds much saner than tying in a third party into the mix

Konrad

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ