test-pmd: Free of address-of expression

test-pmd: Free of address-of expression

[Mrozowicz, SlawomirX]

Hi,

Noticed is that in the file:
app/test-pmd/mempool.c
using of the function munmap() could cause a problem.

Coverity static code analyzer provide error (CID 13184) in line 158:
munmap frees incorrect pointer uv.

I noticed information on the net:
https://www.ibm.com/support/knowledgecenter/SSLTBW_2.1.0/com.ibm.zos.v2r1.bpxbd00/mumap.htm
"If addr is not the address of a mapping established by a prior call to mmap(), the behavior is undefined"

I have analyzed the code and I have done some test with gcc.
It seems that it is possible to free subrange of the mapping memory.
In the mempool.c code the address is calculated independently.
Anyway in my opinion the address variable uv is calculated correctly.

So we should classify this issue as a  False Positive.
Please accept the conclusion.

Slawomir Mrozowicz
Sii Engineer
Delivering outsourced services to Intel
Intel Technology Poland sp. z o.o. - KRS 101882 - ul. Slowackiego 173, 80-298 Gdansk

Re: test-pmd: Free of address-of expression

[Olivier MATZ]

Hi,


On 05/17/2016 09:55 AM, Mrozowicz, SlawomirX wrote:
> Hi,
>
> Noticed is that in the file:
>
> app/test-pmd/mempool.c
>
> using of the function munmap() could cause a problem.
>
> Coverity static code analyzer provide error (CID 13184) in line 158:
>
> munmap frees incorrect pointer uv.
>
> I noticed information on the net:
>
> https://www.ibm.com/support/knowledgecenter/SSLTBW_2.1.0/com.ibm.zos.v2r1.bpxbd00/mumap.htm
>
> “If addr is not the address of a mapping established by a prior call to
> mmap(), the behavior is undefined”
>
> I have analyzed the code and I have done some test with gcc.
>
> It seems that it is possible to free subrange of the mapping memory.
>
> In the mempool.c code the address is calculated independently.
>
> Anyway in my opinion the address variable uv is calculated correctly.
>
> So we should classify this issue as a  False Positive.
>
> Please accept the conclusion.

This file (app/test-pmd/mempool_anon.c) will be removed soon.
Please see http://dpdk.org/dev/patchwork/patch/12067/

So I think we can ignore this alert, as the new code does not
have the same issue.

Regards,
Olivier