linux-kernel - Re: [PATCH v1] misc: fastrpc: Pass proper scm arguments for static process init

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: <16d255fc-7186-11a9-58b5-2d3787f69d59@linaro.org>
Date:   Mon, 12 Jun 2023 12:10:32 +0100
From:   Srinivas Kandagatla <srinivas.kandagatla@...aro.org>
To:     Ekansh Gupta <quic_ekangupt@...cinc.com>,
        linux-arm-msm@...r.kernel.org
Cc:     ekangupt@....qualcomm.com, gregkh@...uxfoundation.org,
        linux-kernel@...r.kernel.org, bkumar@....qualcomm.com,
        fastrpc.upstream@....qualcomm.com, stable <stable@...nel.org>
Subject: Re: [PATCH v1] misc: fastrpc: Pass proper scm arguments for static
 process init

Thanks Ekansh for the patch.


On 05/06/2023 14:48, Ekansh Gupta wrote:
> Memory is allocated for dynamic loading when audio daemon is trying
> to attach to audioPD on DSP side. This memory is allocated from
> reserved CMA memory region and needs ownership assignment to
> new VMID in order to use it from audioPD.
> 
> In the current implementation, arguments are not correctly passed
> to the scm call which might result in failure of dynamic loading
> on audioPD. Added changes to pass correct arguments during daemon
> attach request.
> 
> Fixes: 	0871561055e6 ("misc: fastrpc: Add support for audiopd")
> Cc: stable <stable@...nel.org>
> Tested-by: Ekansh Gupta <quic_ekangupt@...cinc.com>
> Signed-off-by: Ekansh Gupta <quic_ekangupt@...cinc.com>
> ---
>   drivers/misc/fastrpc.c | 31 +++++++++++++++++++++++++------
>   1 file changed, 25 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/misc/fastrpc.c b/drivers/misc/fastrpc.c
> index 30d4d04..b7335dd 100644
> --- a/drivers/misc/fastrpc.c
> +++ b/drivers/misc/fastrpc.c
> @@ -1278,10 +1278,23 @@ static int fastrpc_init_create_static_process(struct fastrpc_user *fl,
>   
>   		/* Map if we have any heap VMIDs associated with this ADSP Static Process. */
>   		if (fl->cctx->vmcount) {
> +			u64 src_perms = BIT(QCOM_SCM_VMID_HLOS);

This is redundant too, we already have cctx->perms initialized to 
exactly same bit map.

> +			struct qcom_scm_vmperm *dst_perms;
> +			u32 i;
> +
> +			dst_perms = kcalloc(fl->cctx->vmcount,
> +							sizeof(struct qcom_scm_vmperm), GFP_KERNEL);
> +			if (!dst_perms)
> +				return -ENOMEM;
> +			for (i = 0; i < fl->cctx->vmcount; i++) {
> +				dst_perms[i].vmid = fl->cctx->vmperms[i].vmid;
> +				dst_perms[i].perm = fl->cctx->vmperms[i].perm;

why do we need to copy this to another struct when we already have this 
information in fl->cctx->vmperms ?

> +			}
> +
>   			err = qcom_scm_assign_mem(fl->cctx->remote_heap->phys,
>   							(u64)fl->cctx->remote_heap->size,
> -							&fl->cctx->perms,
> -							fl->cctx->vmperms, fl->cctx->vmcount);
> +							&src_perms, dst_perms, fl->cctx->vmcount);


> +			kfree(dst_perms);
>   			if (err) {
>   				dev_err(fl->sctx->dev, "Failed to assign memory with phys 0x%llx size 0x%llx err %d",
>   					fl->cctx->remote_heap->phys, fl->cctx->remote_heap->size, err);
> @@ -1322,13 +1335,19 @@ static int fastrpc_init_create_static_process(struct fastrpc_user *fl,
>   	return 0;
>   err_invoke:
>   	if (fl->cctx->vmcount) {
> -		struct qcom_scm_vmperm perm;
> +		u64 src_perms = 0;
> +		struct qcom_scm_vmperm dst_perms;
> +		u32 i;
>   
> -		perm.vmid = QCOM_SCM_VMID_HLOS;
> -		perm.perm = QCOM_SCM_PERM_RWX;
> +		for (i = 0; i < fl->cctx->vmcount; i++) {
> +			src_perms |= BIT(fl->cctx->vmperms[i].vmid);
> +		}
no need for brackets here.
> +
> +		dst_perms.vmid = QCOM_SCM_VMID_HLOS;
> +		dst_perms.perm = QCOM_SCM_PERM_RWX;
>   		err = qcom_scm_assign_mem(fl->cctx->remote_heap->phys,
>   						(u64)fl->cctx->remote_heap->size,
> -						&fl->cctx->perms, &perm, 1);
> +						&src_perms, &dst_perms, 1);
>   		if (err)
>   			dev_err(fl->sctx->dev, "Failed to assign memory phys 0x%llx size 0x%llx err %d",
>   				fl->cctx->remote_heap->phys, fl->cctx->remote_heap->size, err);