回复: [edk2-devel] [PATCH v1 3/3] CryptoPkg/BaseCryptLib: Fix possible uninitialized use

["gaoliming" <gaoliming@byosoft.com.cn>]

Sergei:
  Yes. GCC49 is LTO disable GCC tool chain. GCC5 is LTO enable tool chain.
They both work on the different GCC version, such as gcc5, gcc6..

  https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90844 mentions
-ffat-lto-objects option that can trig the warning with LTO option. Do you
try it?

  If this option works, we can update GCC5 tool chain definition in
tools_def.txt, then this issue can be detected in CI GCC5 build. 

Thanks
Liming
> -----邮件原件-----
> 发件人: devel@edk2.groups.io <devel@edk2.groups.io> 代表 Sergei
> Dmitrouk
> 发送时间: 2021年5月15日 21:01
> 收件人: devel@edk2.groups.io; jiewen.yao@intel.com
> 抄送: Wang, Jian J <jian.j.wang@intel.com>; Lu, XiaoyuX
> <xiaoyux.lu@intel.com>; Jiang, Guomin <guomin.jiang@intel.com>
> 主题: Re: [edk2-devel] [PATCH v1 3/3] CryptoPkg/BaseCryptLib: Fix possible
> uninitialized use
> 
> Hello Jiewen,
> 
> I get the error only for GCC49 and not for GCC5 toolchain.  CI uses GCC5.
> 
> So I compared build commands and this seems to depend on LTO.  Adding
> `-flto`
> impedes compiler's ability to detect such simple issues.
> 
> I've found relevant bug report, there is even fix suggestion from last
month:
> 
>     https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90844
> 
> Regards,
> Sergei
> 
> On Sat, May 15, 2021 at 12:30:44AM +0000, Yao, Jiewen wrote:
> > Hi Sergei
> > Thank you very much for the fix.
> > Reviewed-by: Jiewen Yao <Jiewen.yao@intel.com>
> >
> > I am a little surprised why it is not caught before. It is an obvious
logic issue.
> >
> > Do you think we can do anything on CI, to catch it during pre-check-in
in the
> future?
> > I just feel it is burden to make it post-check-in fix.
> >
> >
> > Thank you
> > Yao Jiewen
> >
> > > -----Original Message-----
> > > From: Sergei Dmitrouk <sergei@posteo.net>
> > > Sent: Friday, May 14, 2021 8:17 PM
> > > To: devel@edk2.groups.io
> > > Cc: Yao, Jiewen <jiewen.yao@intel.com>; Wang, Jian J
> <jian.j.wang@intel.com>;
> > > Lu, XiaoyuX <xiaoyux.lu@intel.com>; Jiang, Guomin
> <guomin.jiang@intel.com>
> > > Subject: [PATCH v1 3/3] CryptoPkg/BaseCryptLib: Fix possible
uninitialized
> use
> > >
> > > `Result` can be used uninitialized in both functions after following
> > > either first or second `goto` statement.
> > >
> > > Cc: Jiewen Yao <jiewen.yao@intel.com>
> > > Cc: Jian J Wang <jian.j.wang@intel.com>
> > > Cc: Xiaoyu Lu <xiaoyux.lu@intel.com>
> > > Cc: Guomin Jiang <guomin.jiang@intel.com>
> > > Signed-off-by: Sergei Dmitrouk <sergei@posteo.net>
> > > ---
> > >  CryptoPkg/Library/BaseCryptLib/Pk/CryptRsaPss.c     | 1 +
> > >  CryptoPkg/Library/BaseCryptLib/Pk/CryptRsaPssSign.c | 1 +
> > >  2 files changed, 2 insertions(+)
> > >
> > > diff --git a/CryptoPkg/Library/BaseCryptLib/Pk/CryptRsaPss.c
> > > b/CryptoPkg/Library/BaseCryptLib/Pk/CryptRsaPss.c
> > > index 4009d37d5f91..0b2960f06c4c 100644
> > > --- a/CryptoPkg/Library/BaseCryptLib/Pk/CryptRsaPss.c
> > > +++ b/CryptoPkg/Library/BaseCryptLib/Pk/CryptRsaPss.c
> > > @@ -82,6 +82,7 @@ RsaPssVerify (
> > >    EVP_PKEY_CTX *KeyCtx;
> > >    CONST EVP_MD  *HashAlg;
> > >
> > > +  Result = FALSE;
> > >    EvpRsaKey = NULL;
> > >    EvpVerifyCtx = NULL;
> > >    KeyCtx = NULL;
> > > diff --git a/CryptoPkg/Library/BaseCryptLib/Pk/CryptRsaPssSign.c
> > > b/CryptoPkg/Library/BaseCryptLib/Pk/CryptRsaPssSign.c
> > > index b66b6f7296ad..ece765f9ae0a 100644
> > > --- a/CryptoPkg/Library/BaseCryptLib/Pk/CryptRsaPssSign.c
> > > +++ b/CryptoPkg/Library/BaseCryptLib/Pk/CryptRsaPssSign.c
> > > @@ -97,6 +97,7 @@ RsaPssSign (
> > >    EVP_PKEY_CTX          *KeyCtx;
> > >    CONST EVP_MD          *HashAlg;
> > >
> > > +  Result = FALSE;
> > >    EvpRsaKey = NULL;
> > >    EvpVerifyCtx = NULL;
> > >    KeyCtx = NULL;
> > > --
> > > 2.17.6
> 
> 
> 
>