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] [day] [month] [year] [list]
Message-ID: <Z5cvkKPF+6WXJZtM@hu-mojha-hyd.qualcomm.com>
Date: Mon, 27 Jan 2025 12:32:40 +0530
From: Mukesh Ojha <mukesh.ojha@....qualcomm.com>
To: Manivannan Sadhasivam <mani@...nel.org>
Cc: manivannan.sadhasivam@...aro.org, jassisinghbrar@...il.com,
        linux-arm-msm@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] mailbox: qcom-ipcc: Reset CLEAR_ON_RECV_RD if set
 from boot firmware

On Fri, Jan 24, 2025 at 01:26:46PM +0530, Manivannan Sadhasivam wrote:
> On Mon, Dec 30, 2024 at 01:24:25PM +0530, Mukesh Ojha wrote:
> > For some SoCs, boot firmware is using the same IPCC instance used
> > by Linux and it has kept CLEAR_ON_RECV_RD set which basically means
> > interrupt pending registers are cleared when RECV_ID is read and the
> > register automatically updates to the next pending interrupt/client
> > status based on priority.
> > 
> > Clear the CLEAR_ON_RECV_RD if it is set from the boot firmware.
> > 
> 
> Any user visible implications due to this?
>

User visible impact is the case of missing interrupt.

We are accessing IPCC_REG_RECV_ID register both from qcom_ipcc_irq_fn()
and from qcom_ipcc_pm_resume() and for some reason if
qcom_ipcc_pm_resume() run just millisecond before qcom_ipcc_irq_fn after
assertion of interrupt from the remote, qcom_ipcc_pm_resume reads the
IPCC_REG_RECV_ID and clears it and when the same register read from the
qcom_ipcc_irq_fn() it does not see the interrupt.

> > Signed-off-by: Mukesh Ojha <mukesh.ojha@....qualcomm.com>
> > ---
> > Change in v2:
> >  - Corrected email id in the Sob.
> > 
> >  drivers/mailbox/qcom-ipcc.c | 16 ++++++++++++++++
> >  1 file changed, 16 insertions(+)
> > 
> > diff --git a/drivers/mailbox/qcom-ipcc.c b/drivers/mailbox/qcom-ipcc.c
> > index 14c7907c6632..0b17a38ea6bf 100644
> > --- a/drivers/mailbox/qcom-ipcc.c
> > +++ b/drivers/mailbox/qcom-ipcc.c
> > @@ -14,6 +14,7 @@
> >  #include <dt-bindings/mailbox/qcom-ipcc.h>
> >  
> >  /* IPCC Register offsets */
> > +#define IPCC_REG_CONFIG			0x08
> >  #define IPCC_REG_SEND_ID		0x0c
> >  #define IPCC_REG_RECV_ID		0x10
> >  #define IPCC_REG_RECV_SIGNAL_ENABLE	0x14
> > @@ -21,6 +22,7 @@
> >  #define IPCC_REG_RECV_SIGNAL_CLEAR	0x1c
> >  #define IPCC_REG_CLIENT_CLEAR		0x38
> >  
> > +#define IPCC_CLEAR_ON_RECV_RD		BIT(0)
> >  #define IPCC_SIGNAL_ID_MASK		GENMASK(15, 0)
> >  #define IPCC_CLIENT_ID_MASK		GENMASK(31, 16)
> >  
> > @@ -274,6 +276,7 @@ static int qcom_ipcc_pm_resume(struct device *dev)
> >  static int qcom_ipcc_probe(struct platform_device *pdev)
> >  {
> >  	struct qcom_ipcc *ipcc;
> > +	u32 config_value;
> >  	static int id;
> >  	char *name;
> >  	int ret;
> > @@ -288,6 +291,19 @@ static int qcom_ipcc_probe(struct platform_device *pdev)
> >  	if (IS_ERR(ipcc->base))
> >  		return PTR_ERR(ipcc->base);
> >  
> > +	/*
> > +	 * It is possible that boot firmware is using the same IPCC instance
> > +	 * as of the HLOS and it has kept CLEAR_ON_RECV_RD set which basically
> > +	 * means Interrupt pending registers are cleared when RECV_ID is read.
> > +	 * The register automatically updates to the next pending interrupt/client
> > +	 * status based on priority.
> > +	 */
> > +	config_value = readl(ipcc->base + IPCC_REG_CONFIG);
> > +	if (config_value & IPCC_CLEAR_ON_RECV_RD) {
> 
> Can't you just clear this bit always?

We can do, but i wanted to be explicit here that something been done
from bootloader which Linux would need to clear.

I see this patch is merged in mainline.

-Mukesh
> 
> - Mani
> 
> -- 
> மணிவண்ணன் சதாசிவம்

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ