linux-kernel - Re: [PATCH] accel/qaic: initialize ret variable to 0

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <3f27c84f-f8b0-bddf-cdd5-952c9576c6c3@quicinc.com>
Date:   Tue, 18 Apr 2023 15:51:23 -0600
From:   Jeffrey Hugo <quic_jhugo@...cinc.com>
To:     Nick Desaulniers <ndesaulniers@...gle.com>
CC:     Tom Rix <trix@...hat.com>, <ogabbay@...nel.org>,
        <nathan@...nel.org>, <jacek.lawrynowicz@...ux.intel.com>,
        <quic_carlv@...cinc.com>, <stanislaw.gruszka@...ux.intel.com>,
        <quic_pkanojiy@...cinc.com>, <linux-arm-msm@...r.kernel.org>,
        <dri-devel@...ts.freedesktop.org>, <linux-kernel@...r.kernel.org>,
        <llvm@...ts.linux.dev>
Subject: Re: [PATCH] accel/qaic: initialize ret variable to 0

On 4/18/2023 2:48 PM, Nick Desaulniers wrote:
> On Tue, Apr 18, 2023 at 1:46 PM Jeffrey Hugo <quic_jhugo@...cinc.com> wrote:
>>
>> On 4/18/2023 1:20 PM, Tom Rix wrote:
>>> clang static analysis reports
>>> drivers/accel/qaic/qaic_data.c:610:2: warning: Undefined or garbage
>>>     value returned to caller [core.uninitialized.UndefReturn]
>>>           return ret;
>>>           ^~~~~~~~~~
>>>
>>> The ret variable is only set some of the time but is always returned.
>>> So initialize ret to 0.
>>
>> This does not appear to be entirely accurate to me.
>>
>> Do you know what analysis Clang is doing?  Is it limited to the function
>> itself?
>>
>> remap_pfn_range, which initializes ret, will always run at-least once.
> 
> What happens if the loop body is never executed, say if `bo->sgt->sgl` is NULL?

qaic_gem_object_mmap() doesn't get called unless a valid GEM object is 
provided by userspace.  For userspace to get a valid GEM object, it has 
to go through qaic_create_bo_ioctl().  qaic_create_bo_ioctl() will not 
return a valid GEM object unless sgt is allocated and initialized.

The loop body will execute at-least once.  The if body will execute 
at-least once.  remap_pfn_range() will run at-least once.  Therefore, 
ret is always initialized.

This is how the code is intended to operate, and how it appears to be 
implemented from what I see.  Unless I'm missing something.

I can see how Clang might not be able to infer these things, but I don't 
believe the code is broken.  I feel that the commit text is unclear and 
suggests that the code is infact, broken.

Userspace should not call mmap() in a critical path thus I don't see a 
true performance concern here.  So while my understanding of the coding 
style is that redundant initialization of variables are to be avoided, I 
think we can say that this is not redundant because it silences a 
warning (because Clang is limited).

>> Feels more accurate to say that Clang cannot detect that ret will be
>> initialized, and we want to quiet the warning from the false positive.
>>
>>> Fixes: ff13be830333 ("accel/qaic: Add datapath")
>>> Signed-off-by: Tom Rix <trix@...hat.com>
>>> ---
>>>    drivers/accel/qaic/qaic_data.c | 2 +-
>>>    1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/accel/qaic/qaic_data.c b/drivers/accel/qaic/qaic_data.c
>>> index c0a574cd1b35..b46a16fb3080 100644
>>> --- a/drivers/accel/qaic/qaic_data.c
>>> +++ b/drivers/accel/qaic/qaic_data.c
>>> @@ -591,7 +591,7 @@ static int qaic_gem_object_mmap(struct drm_gem_object *obj, struct vm_area_struc
>>>        struct qaic_bo *bo = to_qaic_bo(obj);
>>>        unsigned long offset = 0;
>>>        struct scatterlist *sg;
>>> -     int ret;
>>> +     int ret = 0;
>>>
>>>        if (obj->import_attach)
>>>                return -EINVAL;
>>
> 
>