linux-kernel - Re: [PATCH v1] misc: fastrpc: Free DMA handles for RPC calls with no arguments

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: <32e00823-35d3-be56-eb5d-b1c7eee77551@quicinc.com>
Date:   Tue, 26 Sep 2023 17:14:41 +0530
From:   Ekansh Gupta <quic_ekangupt@...cinc.com>
To:     Srinivas Kandagatla <srinivas.kandagatla@...aro.org>,
        <linux-arm-msm@...r.kernel.org>
CC:     <ekangupt@....qualcomm.com>, <gregkh@...uxfoundation.org>,
        <linux-kernel@...r.kernel.org>, <fastrpc.upstream@....qualcomm.com>
Subject: Re: [PATCH v1] misc: fastrpc: Free DMA handles for RPC calls with no
 arguments



On 9/26/2023 3:57 PM, Srinivas Kandagatla wrote:
> Thanks Ekansh for this patch.
> 
> few comments below
> 
> On 31/08/2023 07:23, Ekansh Gupta wrote:
>> The FDs for DMA handles to be freed is updated in fdlist by DSP over
> 
> So the dsp is updating the fd list after invoke?
> 
Yes, correct. DSP updates this fd list when the buffer is no longer 
needed by the user PD.
> 
>> a remote call. This holds true even for remote calls with no
>> arguments. To handle this, get_args and put_args are needed to
>> be called for remote calls with no arguments also as fdlist
>> is allocated in get_args and FDs updated in fdlist is freed
>> in put_args.
>>
>> Signed-off-by: Ekansh Gupta <quic_ekangupt@...cinc.com>
>> ---
>>   drivers/misc/fastrpc.c | 22 +++++++++-------------
>>   1 file changed, 9 insertions(+), 13 deletions(-)
>>
>> diff --git a/drivers/misc/fastrpc.c b/drivers/misc/fastrpc.c
>> index 9666d28..e6df66e 100644
>> --- a/drivers/misc/fastrpc.c
>> +++ b/drivers/misc/fastrpc.c
>> @@ -1153,11 +1153,9 @@ static int fastrpc_internal_invoke(struct 
>> fastrpc_user *fl,  u32 kernel,
>>       if (IS_ERR(ctx))
>>           return PTR_ERR(ctx);
>> -    if (ctx->nscalars) {
> 
> Why do we need to remove this check?
> fastrpc_internal_invoke will have nscalars set before calling. and we 
> are not dealing with fdlist in fastrpc_get_args(), so am not sure what 
> this change is helping with.
Memory for fdlist is allocated as part of fastrpc_get_args. The reason 
to add this change is that fdlist can be updated by DSP over a call with 
no remote arguments, for this call, there should be some fdlist 
allocated so the DSP can update the list if needed. Same apply for 
fastrpc_put_args also as when DSP updates fdlist for any remote call 
with no arguments, the maps corresponding to the fdlist should be freed.
> 
> 
>> -        err = fastrpc_get_args(kernel, ctx);
>> -        if (err)
>> -            goto bail;
>> -    }
>> +    err = fastrpc_get_args(kernel, ctx);
>> +    if (err)
>> +        goto bail;
>>       /* make sure that all CPU memory writes are seen by DSP */
>>       dma_wmb();
>> @@ -1181,14 +1179,12 @@ static int fastrpc_internal_invoke(struct 
>> fastrpc_user *fl,  u32 kernel,
>>       if (err)
>>           goto bail;
>> -    if (ctx->nscalars) {
>> -        /* make sure that all memory writes by DSP are seen by CPU */
>> -        dma_rmb();
>> -        /* populate all the output buffers with results */
>> -        err = fastrpc_put_args(ctx, kernel);
>> -        if (err)
>> -            goto bail;
>> -    }
>> +    /* make sure that all memory writes by DSP are seen by CPU */
>> +    dma_rmb();
>> +    /* populate all the output buffers with results */
> 
> A comment about fdlist here would be really useful
Sure, I will add a comment in the next patch. Do you suggest to add 
comment here or inside fastrpc_put_args function:
https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/tree/drivers/misc/fastrpc.c#n1092
> 
>> +    err = fastrpc_put_args(ctx, kernel);
>> +    if (err)
>> +        goto bail;
>>   bail:
>>       if (err != -ERESTARTSYS && err != -ETIMEDOUT) {