On 31/12/19 7:47 PM, Andrew F. Davis wrote:
> On 12/31/19 1:20 AM, Lokesh Vutla wrote:
>>
>>
>> On 31/12/19 12:20 AM, Andrew F. Davis wrote:
>>> During suspend CPU context may be lost in both non-secure and secure CPU
>>> states. The kernel can handle saving and restoring the non-secure context
>>> but must call into the secure side to allow it to save any context it may
>>> lose. Add these calls here.
>>>
>>> Note that on systems with OP-TEE available the suspend call is issued to
>>> OP-TEE using the ARM SMCCC, but the resume call is always issued to the
>>> ROM. This is because on waking from suspend the ROM is restored as the
>>> secure monitor. It is this resume call that instructs the ROM to restore
>>> OP-TEE, all subsequent calls will be handled by OP-TEE and should use the
>>> ARM SMCCC.
>>>
>>> Signed-off-by: Andrew F. Davis <afd@xxxxxx>
>>> Acked-by: Dave Gerlach <d-gerlach@xxxxxx>
>>> ---
>>> arch/arm/mach-omap2/omap-secure.h | 3 +++
>>> arch/arm/mach-omap2/pm33xx-core.c | 17 +++++++++++++++++
>>> 2 files changed, 20 insertions(+)
>>>
>>> diff --git a/arch/arm/mach-omap2/omap-secure.h b/arch/arm/mach-omap2/omap-secure.h
>>> index 736e594365f4..ba8c486c0454 100644
>>> --- a/arch/arm/mach-omap2/omap-secure.h
>>> +++ b/arch/arm/mach-omap2/omap-secure.h
>>> @@ -53,6 +53,9 @@
>>> #define OMAP4_PPA_L2_POR_INDEX 0x23
>>> #define OMAP4_PPA_CPU_ACTRL_SMP_INDEX 0x25
>>>
>>> +#define AM43xx_PPA_SVC_PM_SUSPEND 0x71
>>> +#define AM43xx_PPA_SVC_PM_RESUME 0x72
>>> +
>>> /* Secure RX-51 PPA (Primary Protected Application) APIs */
>>> #define RX51_PPA_HWRNG 29
>>> #define RX51_PPA_L2_INVAL 40
>>> diff --git a/arch/arm/mach-omap2/pm33xx-core.c b/arch/arm/mach-omap2/pm33xx-core.c
>>> index f11442ed3eff..4a564f676ff9 100644
>>> --- a/arch/arm/mach-omap2/pm33xx-core.c
>>> +++ b/arch/arm/mach-omap2/pm33xx-core.c
>>> @@ -28,6 +28,7 @@
>>> #include "prm33xx.h"
>>> #include "soc.h"
>>> #include "sram.h"
>>> +#include "omap-secure.h"
>>>
>>> static struct powerdomain *cefuse_pwrdm, *gfx_pwrdm, *per_pwrdm, *mpu_pwrdm;
>>> static struct clockdomain *gfx_l4ls_clkdm;
>>> @@ -166,6 +167,16 @@ static int am43xx_suspend(unsigned int state, int (*fn)(unsigned long),
>>> {
>>> int ret = 0;
>>>
>>> + /* Suspend secure side on HS devices */
>>> + if (omap_type() != OMAP2_DEVICE_TYPE_GP) {
>>> + if (optee_available)
>>> + omap_smccc_smc(AM43xx_PPA_SVC_PM_SUSPEND, 0);
>>> + else
>>> + omap_secure_dispatcher(AM43xx_PPA_SVC_PM_SUSPEND,
>>> + FLAG_START_CRITICAL,
>>> + 0, 0, 0, 0, 0);
>>> + }
>>> +
>>> amx3_pre_suspend_common();
>>> scu_power_mode(scu_base, SCU_PM_POWEROFF);
>>> ret = cpu_suspend(args, fn);
>>> @@ -174,6 +185,12 @@ static int am43xx_suspend(unsigned int state, int (*fn)(unsigned long),
>>> if (!am43xx_check_off_mode_enable())
>>> amx3_post_suspend_common();
>>>
>>> + /* Resume secure side on HS devices */
>>> + if (omap_type() != OMAP2_DEVICE_TYPE_GP)
>>> + omap_secure_dispatcher(AM43xx_PPA_SVC_PM_RESUME,
>>> + FLAG_START_CRITICAL,
>>> + 0, 0, 0, 0, 0);
>>
>> Don't you need to check optee_available here?
>>
>
>
> I address this in the second paragraph of the commit message. I can add
> a comment in code also if you think anyone will find it confusing.
Yeah, a comment would help.
Thanks and regards,
Lokesh
>
> Andrew
>
>
>> Thanks and regards,
>> Lokesh
>>
>>> +
>>> return ret;
>>> }
>>>
>>>
