On Thu, Jul 17, 2025 at 10:28:44AM +0800, Ling Xu wrote:
> 在 7/17/2025 3:31 AM, Bjorn Andersson 写道:
> > On Mon, Jul 14, 2025 at 11:11:33AM +0530, Ling Xu wrote:
> >> Some platforms (like sa8775p) feature one or more GPDSPs (General
> >> Purpose DSPs). Similar to other kinds of Hexagon DSPs, they provide
> >> a FastRPC implementation, allowing code execution in both signed and
> >> unsigned protection domains. Extend the checks to allow domain names
> >> starting with "gdsp" (possibly followed by an index).
> >>
> > 
> > This was called cdsp1 before patch 3 where you removed it and now the
> > same id is introduced but this time with the name GDSP.
> > 
> > Iirc there was a cdsp1 in SA8295P/SA8540P, are you silently dropping
> > support for that here? Or perhaps just renaming it?
> > 
> Cdsp1 support is still there. It's instance of cdsp domain, so we merged it
> in cdsp logic.
> 

But doesn't that mean that

  #define CDSP1_DOMAIN_ID (4)

was wrong? It should have been using the 3?

Isn't that a bugfix? If so there should be one patch fixing that.

Perhaps I'm misunderstanding the relevance of these numbers though.

Regards,
Bjorn

> else if (!strncmp(domain, "cdsp", 4))
>       return CDSP_DOMAIN_ID;
> 
> In fastrpc_get_domain_id, it return CDSP_DOMAIN_ID for cdsp1 because they use
> same deamon.
> > Regards,
> > Bjorn
> > 
> >> Signed-off-by: Ling Xu <quic_l...@quicinc.com>
> >> ---
> >>  drivers/misc/fastrpc.c | 6 +++++-
> >>  1 file changed, 5 insertions(+), 1 deletion(-)
> >>
> >> diff --git a/drivers/misc/fastrpc.c b/drivers/misc/fastrpc.c
> >> index 85b6eb16b616..d05969de406e 100644
> >> --- a/drivers/misc/fastrpc.c
> >> +++ b/drivers/misc/fastrpc.c
> >> @@ -27,6 +27,7 @@
> >>  #define MDSP_DOMAIN_ID (1)
> >>  #define SDSP_DOMAIN_ID (2)
> >>  #define CDSP_DOMAIN_ID (3)
> >> +#define GDSP_DOMAIN_ID (4)
> >>  #define FASTRPC_MAX_SESSIONS      14
> >>  #define FASTRPC_MAX_VMIDS 16
> >>  #define FASTRPC_ALIGN             128
> >> @@ -2249,6 +2250,8 @@ static int fastrpc_get_domain_id(const char *domain)
> >>            return MDSP_DOMAIN_ID;
> >>    else if (!strncmp(domain, "sdsp", 4))
> >>            return SDSP_DOMAIN_ID;
> >> +  else if (!strncmp(domain, "gdsp", 4))
> >> +          return GDSP_DOMAIN_ID;
> >>  
> >>    return -EINVAL;
> >>  }
> >> @@ -2323,13 +2326,14 @@ static int fastrpc_rpmsg_probe(struct rpmsg_device 
> >> *rpdev)
> >>    case ADSP_DOMAIN_ID:
> >>    case MDSP_DOMAIN_ID:
> >>    case SDSP_DOMAIN_ID:
> >> -          /* Unsigned PD offloading is only supported on CDSP */
> >> +          /* Unsigned PD offloading is only supported on CDSP and GDSP */
> >>            data->unsigned_support = false;
> >>            err = fastrpc_device_register(rdev, data, secure_dsp, domain);
> >>            if (err)
> >>                    goto err_free_data;
> >>            break;
> >>    case CDSP_DOMAIN_ID:
> >> +  case GDSP_DOMAIN_ID:
> >>            data->unsigned_support = true;
> >>            /* Create both device nodes so that we can allow both Signed 
> >> and Unsigned PD */
> >>            err = fastrpc_device_register(rdev, data, true, domain);
> >> -- 
> >> 2.34.1
> >>
> 
> -- 
> Thx and BRs,
> Ling Xu
> 

Reply via email to